Sequence Determinants of Pri-miRNA Processing
by
Vincent C. Auyeung
B.S., Biology
California Institute of Technology, 2005
SUBMITTED TO THE DEPARTMENT OF BIOLOGY IN PARTIAL FULFILLMENT OF
THE REQUIREMENTS FOR THE DEGREE OF
DOCTOR OF PHILOSOPHY
AT THE
MASSACHUSETTS INSTITUTE OF TECHNOLOGY
JUNE 2012
© 2012 Massachusetts Institute of Technology
All rights reserved
Signature of Author: ____________________________________________________________
Vincent C. Auyeung
Department of Biology
May 21, 2012
Certified by: ___________________________________________________________________
David P. Bartel
Professor of Biology
Thesis Supervisor
Accepted by: __________________________________________________________________
Robert T. Sauer
Professor of Biology
Chair, Biology Graduate Committee
1
Sequence determinants of pri-miRNA processing
by
Vincent C. Auyeung
Submitted to the Department of Biology on May 21, 2012
in partial fulfillment of the requirements for the degree of Doctor of Philosophy
MicroRNAs (miRNAs) are short RNAs that regulate many processes in physiology and
pathology by guiding the repression of target messenger RNAs. For classification purposes,
miRNAs are defined as ~22 nt RNAs that are produced by the cleavage of endogenously
transcribed hairpins. From a cellular perspective, however, miRNAs are the functional products
of a multistep maturation pathway, and are thus defined by the ability of their precursors to enter
this pathway. The cellular distinction between miRNA precursors and other hairpins is made in
the first step of maturation, when the primary miRNA transcript (pri-miRNA) is cleaved by the
Microprocessor, a complex containing Drosha, an RNase III enzyme, and an RNA-binding
partner DGCR8. However, it is unclear how the Microprocessor distinguishes between these
hairpins and authentic pri-miRNAs. In fact, C. elegans pri-miRNAs are not processed in human
cells, illustrating the complexity of pri-miRNA recognition and processing. To systematically
explore sequence determinants of pri-miRNA recognition, hundreds of billions of variants of
human pri-miRNAs were generated, and millions of variants that were functional
Microprocessor substrates were selected in vitro and sequenced. Analysis of the successful
sequences revealed multiple determinants of pri-miRNA binding and cleavage, including hairpin
secondary structure and primary sequence preferences in the terminal loop and flanking the
hairpin. One of these determinants, a CNNC motif downstream of the Drosha cleavage site, is
enriched in pri-miRNAs throughout bilaterian animals. Addition of the primary sequence motifs
to C. elegans pri-miRNAs promoted their efficient processing in human cells, underscoring the
importance of these determinants. The identification and characterization of specific motifs
greatly expands the understanding of the features that cells use to recognize pri-miRNAs, and
opens the door to future studies of pri-miRNA recognition in humans and other bilaterian
animals. In addition, the approach is applicable to the exploration of a variety of functional RNA
elements that have so far resisted functional dissection, including long noncoding RNAs and
messenger RNA localization signals.
Thesis Advisor: David P. Bartel
Title: Professor
3
4
Acknowledgements
Many individuals have contributed to the work described here, and to my professional
and personal development. Of course, none of this would have been possible without the support
and guidance of Dave Bartel. I have admired and benefited from his willingness to pursue any
approach and master any technique, as long as it moves us closer to answering interesting
scientific questions. Beyond that, Dave is an excellent personal and professional role model.
One of the greatest things about Dave has been his ability to recruit a group of fantastic
scientists to work in his lab. The atmosphere of the lab is incredibly open, with people regularly
speaking to each other throughout the day to exchange ideas and advice. I have also benefited
from members’ backgrounds in in diverse disciplines, including developmental biology, cancer
biology, computational biology, plant biology, biochemistry, and genetics. I am inspired by the
lab members’ ability to creatively combine different experimental approaches and ways of
thinking to address a variety of scientific problems.
The work described in Chapter 2 relied on the critical contributions of David Shechner
and Igor Ulitsky. The circularized-substrate cleavage selection was born of a nighttime
brainstorming session with David Shechner, and was just one of his ideas among many good
ones. Igor Ulitsky performed most of the conservation analysis described in Chapter 2. I am
always amazed by his ability to quickly grasp the biological question and apply his vast
computational expertise to finding the answer, all the while maintaining a great sense of humor.
I am grateful to the people who listened to my constant stream of stupid ideas and
considered them critically: Calvin Jan, J. Graham Ruby, Olivia Rissland, David Weinberg, and
Igor Ulitsky. Our conversations have taught me to take idle musings, strengthen their intellectual
foundations, and operationalize them into productive experiments. I have particularly benefited
from the wisdom of Calvin Jan; without Calvin, I would have wandered around the wildernesses
of science much longer than I did.
Over the years I have also gained from mentors and role models my research and medical
careers. My thesis committee, Phillip Sharp and Uttam RajBhandary, have been with me nearly
every step of my research training. On the medical side, I have benefited from the advice and
guidance of Richard Mitchell.
On a personal level, many people have helped keep me sane over the past few years. My
baymates Ines Anna Drinnenberg and David Garcia have helped maintain an “atmosphere” in the
bay to make spending hours at the bench that much more palatable. I’ve shared good times in
lab and out of lab with Laura Resteghini, Sue-Jean Hong, Huili Guo, Stephen Eichhorn, Igor
Ulitsky, Alena Shkumatava, Christine Mayr, Andrew Grimson, Calvin Jan, Olivia Rissland,
David Shechner, J. Graham Ruby, and Noah Spies. My other Boston friends have kept things in
perspective for me while I was immersed in research; Evgeniy Kreydin, Xavier Rios, and
Takahiro Soda deserve special thanks. Others have kept me physically active, like the many
graduate students and postdocs who played for the Biograds intramural tennis team, which I had
the privilege of organizing. Our consistent losing record never stopped us from loving the game.
David Garcia, Calvin Jan, James Patridge, Dave Kenezevic, and Eveline Stein have cycled
thousands of miles with me, giving me a chance to see a slice of Massachusetts up close and to
get some vitamin D in the process.
No set of acknowledgements would be complete without mentioning the support and
patience of my parents, Marianna and Michael, my brother William, and my sister Wendy.
And, most importantly, Joanne. Thank you for everything.
5
6
Table of Contents
Chapter 1.
What defines a miRNA?...................................................................................... 9
Chapter 2.
Beyond secondary structure: primary-sequence determinants license
pri-miRNA hairpins for processing..................................................................... 59
Chapter 3.
Future directions................................................................................................. 121
Appendix 1.
Experimental protocols........................................................................................ 161
Appendix 2.
Statistical methods.............................................................................................. 225
Appendix 3.
Mammalian microRNAs: experimental evaluation of novel and
previously annotated genes.................................................................................. 235
7
8
Chapter 1.
What defines a miRNA?
Contents
Introduction ................................................................................................................................... 10
Understanding the cellular definition of an animal miRNA ......................................................... 13
The biogenesis of miRNAs ........................................................................................................... 15
Known determinants of pri-miRNA processing ........................................................................... 16
General preferences of the Microprocessor .......................................................................... 17
Regulation of cleavage in subsets of animal pri-miRNAs .................................................... 20
Plant pri-miRNA processing: DCL1..................................................................................... 23
Determinants of canonical biogenesis downstream of the Microprocessor.................................. 24
Specificity of nuclear export mediated by exportin-5........................................................... 24
Specificity of cleavage by Dicer ........................................................................................... 25
Specificity of loading into Argonaute ................................................................................... 28
Regulation of biogenesis in subsets of animal pre-miRNAs ................................................ 30
Finding additional biogenesis determinants in pri-miRNAs ........................................................ 31
Substrate specificity in RNase III family proteins ........................................................................ 33
Eubacterial RNase III ............................................................................................................ 33
Yeast RNase III: Rnt1p and Pac1 ......................................................................................... 35
An exhaustive, quantitative approach to defining pri-miRNAs.................................................... 39
9
Introduction
The microRNA (miRNA) field began with the cloning of the nematode gene lin-4 and the
realization that it formed Watson–Crick base pairs with the 3′ untranslated region of the lin-14
messenger RNA (mRNA). Both genes had previously been identified as key regulators of
developmental timing in the nematode Caenorhabditis elegans, although the molecular
mechanism linking the two genes was unknown. Extensive effort to clone and sequence these
two genes revealed that lin-4 was a tiny RNA that did not encode a protein (Lee et al., 1993).
The realization that lin-4 was complementary to portions of the lin-14 mRNA led to the
hypothesis that the genetic relationship between lin-4 and lin-14 was mediated by the physical
relationship between regulator RNA and target mRNA (Lee et al., 1993; Wightman et al., 1993).
This type of interaction had never been described in animals, and even among prokaryotic
regulatory RNAs lin-4 was exceptional: four times smaller than any other noncoding regulatory
RNA known at that time (Ruvkun et al., 2004). Still, this regulatory scheme seemed likely to be
idiosyncratic (Ruvkun et al., 2004); both examples were from one pathway from one rapidlyevolving nematode.
As it turns out, lin-4 is the founding member of a much larger class of regulatory RNAs
in animals. Several years later, a second small regulatory RNA, let-7, was identified and shown
to regulate lin-14 and other developmental timing genes by binding to their 3′ untranslated
regions (Reinhart et al., 2000), and was conserved across bilaterian animals (Pasquinelli et al.,
2000). The initial trickle of small RNA discovery became a torrent when three groups described
the existence of a large number of tiny RNAs, ranging from 21 to 24 nucleotides (nt) long
(Lagos-Quintana et al., 2001; Lau et al., 2001; Lee and Ambros, 2001). These RNAs were found
in multiple bilaterian animal species (Drosophila melanogaster, Caenorhabditis elegans, and
humans); were both diverse and individually abundant; and were often conserved between all
three species, spanning hundreds of millions of years of evolution. In recognition of their small
size, these RNAs were called “microRNAs” (miRNAs).
In the past decade, both the number of miRNAs and the catalog of their biological
functions have blossomed.
Extensive miRNA discovery efforts in animals have identified
hundreds of miRNA families in animals and plants, and each family can have many individual
miRNA members in each species (Bartel, 2004). For perspective, miRNAs account for over 2%
of predicted mammalian genes, and the number of annotated human miRNA families in
10
miRBase (Griffiths-Jones et al., 2006) is comparable to the number of human protein tyrosine
kinases annotated by the Gene Ontology Consortium (Ashburner et al., 2000). Like lin-4,
miRNAs recognize their targets by base pairing to sites in the mRNA; in plants, the target
pairing occurs throughout the length of the miRNA, while in animals target pairing to ~6–8 nt at
the 5′ end of the miRNA, termed “seed” pairing, is nearly always necessary and often sufficient
for repression (Bartel, 2009). Supplemental elements also contribute to target site efficacy in
animals, including the local nucleotide content, position of the target site in the mRNA 3′
untranslated region (3′ UTR), thermodynamic stability of seed pairing, the abundance of target
sites in the cell, and proximity of the target site to other miRNA sites (Doench and Sharp, 2004;
Grimson et al., 2007; Saetrom et al., 2007; Ui-Tei et al., 2008; Arvey et al., 2010; Garcia et al.,
2011).
The mechanism by which miRNAs repress target mRNAs varies between plants and
animals, in accordance to the targeting mechanism. In plants, this extensive pairing guides
cleavage of the target mRNA (Llave et al., 2002; Tang et al., 2003; Jones-Rhoades et al., 2006).
Although animal miRNAs also guide cleavage of some targets, their limited pairing with most
miRNA targets is insufficient to support target-site cleavage (Yekta et al., 2004; Shin et al.,
2010).
Instead, miRNA targeting results in mRNA destabilization and/or inhibition of
translation.
Which mode of target gene repression predominates depends on time; recent
transcriptome- and proteome-wide studies indicate that steady-state repressive effects are mostly
explained by mRNA destabilization (Baek et al., 2008; Hendrickson et al., 2009; Guo et al.,
2010), while similar studies in the fish embryo show that translational repression can dominate
for a brief period immediately after induction of miRNA expression (Bazzini et al., 2012).
Regardless of mechanism, the magnitude of effects is generally modest but nevertheless
significant.
Despite the subtlety of their effect, the animal miRNAs are critical regulators of the
transcriptome. After all, each miRNA can have hundreds of conserved targets, and >60% of
mammalian mRNAs have been under selective pressure to maintain at least one target site
(Friedman et al., 2009), while other mRNAs have been under selective pressure to avoid
targeting by coexpressed miRNAs (Farh et al., 2005; Stark et al., 2005). As a class, the miRNAs
are essential for normal mammalian development, since mutations that ablate miRNA biogenesis
are lethal in mammals (Bernstein et al., 2003; Babiarz et al., 2008). Individual miRNAs have
11
been implicated in a spectrum of biological processes.
A particular theme has been
spatiotemporal control in development, consistent with the roles of lin-4 and let-7 in
developmental timing in C. elegans. For example, the mammalian miRNA miR-155 regulates
the differentiation of helper T cells in the immune system, and loss of miR-155 impairs the
formation of germinal centers, which are important for effective antibody responses (Thai et al.,
2007). By contrast, sustained overexpression of this miRNA perturbs the early differentiation of
hematopoietic cells, and ultimately causes the inappropriate proliferation of the myeloid cell
compartment in mice (O'Connell et al., 2008). In human cancer, miRNAs are often located at or
near sites of genomic damage and have reduced levels, suggesting that the disruption of miRNA
regulation is a common feature of cancer (Calin et al., 2004b; Lu et al., 2005). Consistent with
this view, a general reduction in miRNA levels by inhibition of biogenesis promotes oncogenic
transformation in mice (Kumar et al., 2007). In fact, the disruption of a single miRNA–target
relationship between let-7 and the oncogene HMGA2 is enough to promote oncogenic
transformation, and disruptions of this relationship occur frequently in human cancers (Mayr et
al., 2007).
More broadly, miRNAs comprise just one class of molecule in a larger paradigm of
biological regulation by small noncoding RNAs. This paradigm emerged from a collection of
mysterious observations in plants, animals, and fungi. In plants and fungi, separate efforts to
overexpress genes unexpectedly caused silencing of both the exogenously introduced gene and
endogenous genes with the same sequence (Napoli et al., 1990; van der Krol et al., 1990;
Romano and Macino, 1992).
In animals, antisense nucleic acids had been used to inhibit
endogenous gene expression, presumably by forming duplexes with the target mRNA (Izant and
Weintraub, 1984), although the method was curiously successful when either sense or antisense
RNAs were injected (Fire et al., 1991; Guo and Kemphues, 1995). It was later discovered that
the efficacy of inhibition could be enhanced over 100-fold in C. elegans when both sense and
antisense RNAs were injected (Fire et al., 1998). These observations, disparate in method, goals,
and even phylogenetic kingdom, likely had a single commonality: the intentional or
unintentional introduction of double-stranded RNA (dsRNA) (Montgomery and Fire, 1998). In
plants, induction of post-transcriptional gene silencing by dsRNA caused the accumulation of
smaller, ~25 nt RNA fragments (Hamilton and Baulcombe, 1999). In animals, the dsRNA was
later shown to be processed into small, 21-22 nt fragments (Hammond et al., 2000; Zamore et al.,
12
2000), and these fragments are the active species that mediate silencing (Elbashir et al., 2001a;
Elbashir et al., 2001b).
Since then, thousands of studies have used artificial dsRNA or the small active species to
silence genes of interest, a technique called RNA interference (RNAi). Yet the use of RNAi as a
tool belies the importance of the many forms of endogenous silencing, each mediated by distinct
small RNAs (Ketting, 2011). With the exception of PIWI-interacting RNAs, the various small
RNAs are derived from paired RNA, including transcribed hairpins with long stems, dual sense
and antisense transcripts from a genomic locus, and duplexes synthesized by RNA-dependent
RNA polymerases.
They have a variety of evolutionarily-conserved biological functions,
including the regulation of gene expression by degrading messages, repressing translation, or
modifying chromatin; and the defense against viruses and other invasive genetic elements by
cleaving gene products or the genomes themselves.
These small RNAs and their diverse
biological functions are interconnected by a web of related biogenesis and effector mechanisms,
including those mediated by the RNase III and Argonaute protein families, leading to the view
that RNAi is ancient and pervasive. It is ironic that the herald of this paradigm, lin-4, was once
thought to be an oddity of nematode development.
Understanding the cellular definition of an animal miRNA
To study the common properties of miRNAs, it is crucial to distinguish those RNAs that
belong to the miRNA class from others.
Accordingly, a set of criteria was adopted for
classifying small RNAs as miRNAs (Ambros et al., 2003). One set of criteria relates to size and
expression: miRNAs should be ~22 nt RNAs, and thus detectable in cellular RNA by methods
such as small RNA blotting or cDNA sequencing (Ambros et al., 2003). The second relates to
origin: miRNAs should be derived from the stem region of relatively regular hairpins, without
large internal loops or bulges; ideally, the pairing in the hairpin should be conserved, and the
hairpin should be cleaved by an RNase III enzyme called Dicer (discussed below) (Ambros et
al., 2003).
Although these criteria are useful for human minds to classify certain small RNAs as
miRNAs, they do not answer an important question: what is a miRNA to the cell? Since
miRNAs are derived from precursor RNAs much longer than the mature miRNA, cells must
somehow recognize certain RNA species as miRNA precursors, as distinct from precursors of
13
pri-miRNA
pri-miRNA
Unstructured terminal loop
(? >10 nt)
Recognition and binding
(Microprocessor)
~1 helical turn
Stem Watson–Crick pairing
(~ 3 helical turns)
Downstream
unstructured sequence
(> 20 nt)
Upstream
unstructured sequence
(>20 nt)
Basal stem junction
Cleavage
(Microprocessor)
pre-miRNA
Nuclear-cytosolic export
(Exportin 5)
Stem Watson–Crick pairing
3′ overhang
(2-8 nt)
P
OH
Unstructured terminal loop
(? >14 nt)
3′ overhang (2 nt)
5′ phosphate
3′ hydroxyl
3′ hydroxyl
OH
Cleavage
(Dicer)
p
Central stem mismatches
OH
5′ phosphate
Argonaute loading
(RISC loading complex)
P
P
Weaker
pairing stability
OH
~2 helical turns
Stem Watson–Crick pairing
(~ 2 helical turns)
P
OH
other RNA species, and process the miRNA precursors accordingly. Thus, from the cellular
perspective, miRNAs are defined by the ability of their precursors to enter a specific biogenesis
pathway; the miRNAs themselves are simply the functional products of this pathway. To
understand the cellular definition of an animal miRNA, we must therefore consider their
biogenesis and the specificity of each step in the pathway for particular RNAs (Figure 1).
The biogenesis of miRNAs
In the canonical pathway of miRNA biogenesis, primary miRNA transcripts (primiRNAs) are synthesized by RNA polymerase II (Lee et al., 2004a) as noncoding transcripts, or
as embedded sequences within introns of protein-coding “host” genes. While still in the nucleus,
the pri-miRNA is cleaved (Lee et al., 2002).
This cleavage is carried out by the
“Microprocessor,” a large protein complex composed of Drosha, an RNase III enzyme, and a
protein cofactor DGCR8, called Pasha and Psh-1 in Drosophila melanogaster and C. elegans,
respectively (Lee et al., 2003; Denli et al., 2004; Gregory et al., 2004; Han et al., 2004;
Landthaler et al., 2004). DGCR8 is thought to recognize the junction between the miRNA
hairpin and flanking single strand RNA, positioning Drosha to cleave approximately one helical
turn above the junction (Han et al., 2006; Yeom et al., 2006). The resulting hairpin is termed the
precursor miRNA (pre-miRNA), and consists of a ~2-turn stem with a characteristic 2 nt 3′overhang. This distinctive hairpin is exported from the nucleus to the cytosol by exportin-5 (Yi
et al., 2003; Bohnsack et al., 2004; Lund et al., 2004); in species where there is no exportin-5
ortholog, the pre-miRNA presumably makes use of exportin-t instead (Murphy et al., 2008). In
the cytosol, the pre-miRNA is cleaved by a complex of proteins containing another RNase III
enzyme called Dicer (Grishok et al., 2001; Hutvagner et al., 2001; Ketting et al., 2001). For
most miRNAs, one strand is preferentially loaded into an Argonaute family member based on the
thermodynamic stability of the Dicer product (Khvorova et al., 2003; Schwarz et al., 2003). The
mature miRNA strand and its Argonaute protein partner form the core of the silencing complex
(Liu et al., 2004; Meister et al., 2004).
Figure 1. Summary of the biogenesis of miRNAs and determinants in intermediate RNA species. For
each intermediate along the path to maturity, determinants are shown that promote the processing of
that intermediate.
15
The vast majority of annotated miRNAs mature through the canonical pathway, based on
their dependencies on DGCR8/Pasha and Dicer (Calabrese et al., 2007; Wang et al., 2007;
Babiarz et al., 2008). However, several miRNAs make use of alternative pathways which bypass
various steps of the canonical pathway. For the miRNA introns, or “mirtrons,” one or both ends
of the pre-miRNA are established by the spliceosome during intron excision (Okamura et al.,
2007; Ruby et al., 2007a). For many mitrons, the debranched introns have all the features of a
canonical pre-miRNA, including a 2 nt 3′-overhang, and they are exported, diced, and loaded
like canonical pre-miRNAs (Okamura et al., 2007; Ruby et al., 2007a). In other mirtrons, the 3′
splice sites are downstream of the pre-miRNA ends; in these cases, the intron 3′ end is trimmed
by the exosome before dicing (Flynt et al., 2010). Endogenous short hairpin RNAs also bypass
Microprocessor cleavage. Although not well-studied as a class, these noncanonical miRNAs are
probably derived from short transcription units that intrinsically produce a hairpin with the
features of a pre-miRNA (Babiarz et al., 2008). A third alternative pathway bypasses Dicer.
Like canonical miRNAs, the primary transcript of mir-451 is cleaved by the Microprocessor and
the pre-miRNA is exported to the cytosol; however, unlike canonical pre-miRNAs, pre-mir-451
is cleaved by Argonaute 2 (Ago2) (Cheloufi et al., 2010; Cifuentes et al., 2010).
Known determinants of pri-miRNA processing
Two parallel lines of investigation converged on the early identification of Drosha in
animals. One was incidental; the drosha locus in Drosophila melantogaster was encountered
during genomic analysis of the rnh1 locus encoding RNase H1 amid questions about the role of
RNase H proteins in animal biology.
Microdeletions in a region adjacent to rnh1 caused
lethality, and sequencing of this region revealed an open reading frame predicted to produce a
153 kDa protein with homology to the endonuclease domain of bacterial and yeast RNase III
proteins (Filippov et al., 2000). Named Drosha, the the novel protein contained two tandem
endonuclease domains instead of just one, and gene database searching revealed highly related
homologues with tandem RNase III domains in the genomes of both C. elegans and humans
(Filippov et al., 2000). Thanks to the growing power of large sequence databases, it was already
known that two classes of RNase III enzymes were present in animals (Mian, 1997); Drosha was
one type, while the other helicase-like (Rotondo and Frendewey, 1996) and would later be
named Dicer (discussed below). At the same time, a separate group interested in mammalian
16
RNase III proteins used a phage cDNA clone library to build the full-length cDNA of a human
protein containing RNase III domains (Wu et al., 2000). This protein degraded long dsRNA,
albeit much more poorly than the E. coli RNase III. Importantly, the protein was nuclearlocalized, and it was believed to mediate rRNA processing, based on the functions of RNase III
in yeast and bacteria (Wu et al., 2000). Because of its localization, the protein was named
RNASEN, the human nuclear RNase III, but is now called Drosha like its ecdysozoan orthologs.
The proteins remained unassociated with RNA interference and the biogenesis of
miRNAs until two observations were made: first, that the initial post-transcriptional event of
miRNA biogenesis was the cleavage of pri-miRNAs in the basal hairpin stem, an activity that
was localized to the nucleus (Lee et al., 2002); and second, that the pre-miRNA product of this
first cleavage had 2 nt 3′-overhangs, precisely the expected product of staggered cleavage by
RNase III enzymes (Lee et al., 2003).
Indeed, immunoprecipitated human Drosha (i.e.,
RNASEN) accurately excised pre-miRNAs from longer primary transcripts (Lee et al., 2003). In
human and Drosophila lysates, Drosha and pri-miRNA cleavage activity fractionates with a
~600 kDa complex which has been called the Microprocessor (Denli et al., 2004; Gregory et al.,
2004). In this complex, Drosha is tightly associated with a binding partner called Pasha in
Drosophila and C. elegans, and DGCR8 in humans, which are homologues of each other (Denli
et al., 2004; Gregory et al., 2004). DGCR8/Pasha is required for Microprocessor cleavage
activity both in vitro and in vivo (Denli et al., 2004; Gregory et al., 2004), and recombinant
human Drosha and DGCR8 are together sufficient to reconstitute the pri-miRNA cleavage in
vitro (Gregory et al., 2004). DGCR8/Pasha contains two double-strand RNA binding domains
(dsRBDs), and the presence of at least one is required to support pri-miRNA cleavage by the
Microprocessor (Yeom et al., 2006). Thus, DGCR8/Pasha probably contributes significantly to
pri-miRNA binding and recognition, since Drosha itself contains just one dsRBD (Lee et al.,
2003). The functions of Drosha and Pasha/DGCR8 are so entwined that the Drosha protein is
unstable in the absence of DGCR8, and Drosha regulates DGCR8 levels by cleaving a hairpin in
the 5′ untranslated region of the DGCR8 mRNA (Han et al., 2009).
General preferences of the Microprocessor
Given its role as the gateway to the canonical miRNA biogenesis pathway, the
Microprocessor and its substrate preferences have been subjected to intense scrutiny. Minimal
17
substrates for the Microprocessor in vitro are composed of the pre-miRNA hairpin flanked by at
least 20-50 nt of genomic sequence; the determinants in these segments are necessary and
sufficient to support at least minimal cleavage by the Microprocessor in vitro (Lee et al., 2003)
and expression of the mature miRNA in vivo (Chen et al., 2004). This region is important in part
because of Watson–Crick base pairing that extends basally to the pre-miRNA hairpin. Mutations
that abolish base pairing impair cleavage, while mutations that preserve base pairing preserve
cleavage, albeit at reduced efficiency (Lee et al., 2003). This pairing is consistent with the
observation that C. elegans, Drosophila, and human pri-miRNAs strongly tend to have base
pairing that extends beyond the pre-miRNA hairpin (Lim et al., 2003b; Han et al., 2006).
Beyond the stem, a length of unstructured RNA flanking the stem is required for pri-miRNA
processing, since the Microprocessor does not cleave substrates lacking flanking RNA (Zeng and
Cullen, 2005; Han et al., 2006). The Microprocessor is thought to recognize the flank-stem
junction, and uses the junction to guide cleavage approximately one helical turn above the base
(Han et al., 2006). Consistent with this model, mutations that shorten or lengthen the base of the
stem shift Microprocessor cleavage site accordingly, at least in vitro (Han et al., 2006).
The structural basis of Microprocessor binding to the stem-flank junction is poorly
understood. It was initially suggested that DGCR8 was responsible for junction recognition
(Han et al., 2006), but substrate affinity studies with the DGCR8 dsRBDs have not consistently
demonstrated that the dsRBDs can distinguish between hairpins and hairpins with flanking RNA
(Sohn et al., 2007). It is worth noting that binding affinities measured in this study were in the
2–4 µM range, far higher than expected, suggesting that physiological binding of the dsRBDs to
pri-miRNAs may require the presence of other domains in DGCR8, or a functional complex
between DGCR8, Drosha, and perhaps other proteins in the Microprocessor complex. Others
have suggested that DGCR8 binding to the pri-miRNA is cooperative, and that DGCR8
monomers can bind to multiple regions of the pri-miRNA (Faller et al., 2010). Whether this
multimerization contributes to substrate specificity is unclear.
The apical stem and terminal loop also contribute to recognition of pri-miRNAs by the
Microprocessor, although their importance has been debated. Earlier in the characterization of
the Microprocessor, it was suggested that the Drosha cleavage site was established by a
molecular ruler two turns away from the loop (Zeng et al., 2005). This model was unsatisfying
because it failed to accurately predict the Drosha cleavage site based on the thermodynamically
18
predicted stem terminus; by contrast, basal stem length robustly predicts cleavage site, even in
artificial substrates that have no loop at all (Han et al., 2006). As a consequence, loop-based
measurement is not widely accepted as the mechanism of cleavage site selection. Nevertheless,
shortening of either the terminal loop or the apical stem impairs binding to DGCR8 and cleavage
by the Microprocessor (Zeng et al., 2005; Han et al., 2006; Zhang and Zeng, 2010). The optimal
length of the loop has not been determined, but the optimal length of the apical stem appears to
be ~2 helical turns above the Microprocessor cleavage site. Thus the optimal structure of a primiRNA is a ~3-turn hairpin (one turn between the base and the cleavage site, and two turns
between the cleavage site and the loop) flanked on each side by some length of unstructured
RNA (Figure 1).
Existing studies of Microprocessor preferences also hint at the existence of other, less
well-defined sequence or structural determinants. First, sequence analysis of miRNA hairpins
demonstrates a propensity for these hairpins to contain internal loops that are reasonably
symmetric (Lim et al., 2003b; Han et al., 2006; Warf et al., 2011). These could contribute
somehow to enhance binding and cleavage by the Microprocessor, although an early study of the
hsa-mir-30a pri-miRNA suggested that its internal loops were dispensable (Lee et al., 2003).
Alternatively, central internal loops could inhibit inappropriate or non-productive binding or
cleavage in the apical stem, one helical turn from the stem-loop junction, thus biasing the
Microprocessor to cleave at the appropriate location (Han et al., 2006). Central loops or bulges
could also facilitate miRNA biogenesis downstream of the Microprocessor, particularly at the
step of loading into Argonaute (discussed below).
Second, a number of stem and loop mutations that impair Drosha processing in vitro have
been described (Zeng et al., 2005; Gottwein et al., 2006), along with some single nucleotide
polymorphisms (SNPs) thought to impair in vivo processing (Duan et al., 2007; Sun et al.,
2009). The significance of these mutations in the hairpin is unclear. One view is that these
mutations could be altering critical sequence motifs that are recognized by the Microprocessor or
auxiliary recognition proteins, although such motifs were not delineated in the studies. Another
view is that such mutations could substantially change the pri-miRNA folding landscape, biasing
the ensemble of folding isoforms away from the optimal structure and thus preventing proper
cleavage (P. Dallaire, personal communication).
differentiate between these two models.
19
Additional investigation is needed to
Finally, the lengths of upstream and downstream RNA required for pri-miRNA
processing in vivo (Chen et al., 2004) are not fully explained by the stem-flank junction model
for pri-miRNA cleavage, since just a few flanking nucleotides are sufficient for cleavage site
determination in artificial substrates (Han et al., 2006). It is possible that additional primary
sequence or structural determinants could reside in the RNA sequence flanking the stem. Indeed,
a SNP downstream of the hsa-mir-16-1 pri-miRNA impairs its processing in cell lines, and is
associated with B-cell chronic lymphocytic leukemia in humans (Calin et al., 2002; Calin et al.,
2005). It is tempting to speculate that this SNP has affected recognition of a motif downstream
of the pri-miRNA hairpin, although no specific motifs were identified in those studies.
Regulation of cleavage in subsets of animal pri-miRNAs
Several mechanisms for the dynamic regulation of pri-miRNA cleavage have been
described. These mechanisms are typically mediated by individual proteins, and may only affect
a subset of miRNAs at specific times. Most described regulatory mechanisms are thought to
mildly enhance pri-miRNA cleavage by the Microprocessor, while the remaining ones sequester
the pri-miRNA or induce its active degradation without necessarily inhibiting the activity of the
Microprocessor per se.
Some regulators of pri-miRNA cleavage appear to depend on binding to the pri-miRNA
terminal loop. One example is the heterogeneous nuclear ribonucleoprotein A1 (hnRNPA1), a
highly abundant protein thought to function in general mRNA metabolism, including splicing,
mRNA export, and regulation of stability (Dreyfuss et al., 1993). A transcriptome-wide study of
hnRNPA1 binding sites by crosslinking and immunoprecipitation incidentally found binding
sites in the terminal loop of hsa-mir-18a; of note, there were no observed binding events to other
terminal loops in the mir-17~92 cluster of miRNAs, of which mir-18a is a member (Guil and
Caceres, 2007). Binding of hnRNPA1 is thought to promote Microprocessor cleavage of the
mir-18a pri-miRNA by altering the conformation of the mid-stem (Michlewski et al., 2008),
based on changes in RNase V1 accessibility, although it is not clear whether this is the
consequence of stem melting or is actually due to occlusion of the RNase V1 cleavage sites by
hnRNPA1 binding. This study noted that miRNA loops are generally more conserved than
nearby sequence (although considerably less conserved than the mature miRNA and the
miRNA* strand), suggesting that other proteins may also have conserved binding sites in many
20
pri-miRNAs (Michlewski et al., 2008).
Nonetheless, the role of hnRNPA1 in pri-miRNA
cleavage may not be straightforward; its binding to the let-7a pri-miRNA appears to antagonize
cleavage by the Microprocessor (Michlewski and Caceres, 2010). How hnRNPA1 enhances
processing in some pri-miRNAs but represses processing in others is not clear.
Similarly, an RNA-binding protein that plays multiple roles in RNA metabolism, the KHtype splicing regulatory protein (KSRP), appears to bind the terminal loop of the let-7a primiRNA and enhance its cleavage by both the Microprocessor and Dicer (Trabucchi et al., 2010).
KSRP may also regulate other miRNAs, based on reductions in the mature miRNA levels after
KSRP knockdown. It has also been suggested that KSRP may compete with hnRNPA1 for
binding to the let-7a terminal loop, and that the relative expression levels of these antagonistic
proteins may dynamically establish the processing efficiency for let-7a (Michlewski and
Caceres, 2010).
The cleavage of a subset of pri-miRNAs is thought to be regulated by extracellular
signals. Signaling through the transforming growth factor β (TGFβ) pathway upregulates mature
miR-21 by increasing processing of the mir-21 pri-miRNA (Davis et al., 2008). The effector
proteins downstream of TGFβ are the Smad proteins, which trimerize and act as transcription
factors. TGFβ signaling appears to induce the association of both SMAD proteins and the
helicase p68 with the hsa-mir-21 pri-miRNA, which results in enhanced cleavage by the
Microprocessor (Davis et al., 2008). This effect of TGFβ signaling does not depend on Smad4,
which is usually a necessary cofactor in trimeric Smad complexes that regulate transcription
(Davis et al., 2008). Many miRNAs appear to be regulated by TGFβ signaling, and appear to
have the common sequence motif
5′−CAGAC−3′
3′−GUCUG−5′
in the mid-stem adjacent to the Microprocessor
cleavage site (Davis et al., 2010). This motif, when grafted onto non-regulated pri-miRNAs, is
sufficient to confer TGFβ regulation (Davis et al., 2010). Oddly, this is nearly exactly the
5′−AGAC−3′
canonical Smad binding element 3′−
TCTG−5′ in DNA which the Smad MH1 domain recognizes by
inserting a β-sheet “hairpin” into the major groove (Shi et al., 1998). The MH1 domain is also
thought to mediate binding to pri-miRNAs (Davis et al., 2010), but it is not clear whether the
domain is capable of inserting into the deeper and narrower major groove of the presumably Aform pri-miRNA stem.
The Smad proteins are not the only transcription factor proteins proposed to bind primiRNAs and regulate their cleavage.
The transcription factor All1 has been suggested to
21
enhance Microprocessor cleavage, but its association with the Microprocessor is DNAdependent (Nakamura et al., 2007), raising the possibility that the Microprocessor could be
regulated by recruitment to sites of active transcription. Two other examples are the DEAD-box
helicases p68 and p72, which are multifunctional DEAD-box helicases that, among other things,
activate transcription in collaboration with other transcriptional regulators like p53 and steroid
receptors. The mouse knockouts of these proteins resulted in reductions in the mature levels of a
handful of mature miRNAs (Fukuda et al., 2007). The ability of Drosha to crosslink to these primiRNAs is impaired in the absence of these helicases, and antibodies against the helicases
inhibit processing in vitro (Fukuda et al., 2007). It is not clear how these proteins promote
cleavage by the Microprocessor, and how the effect is restricted to a subset of pri-miRNAs.
Nevertheless, the association of these helicases with the critical DNA damage response regulator
p53 instigated an analysis of the effect of p53 activation and loss-of-function on miRNA levels.
Activation of p53 by doxorubicin increased the expression of a subset of mature miRNAs
without affecting pri-miRNA levels, and this effect was dependent on p68 and p72 (Suzuki et al.,
2009). Addition of p53 also enhanced pri-miRNA cleavage by the Microprocessor for some of
these miRNAs (Suzuki et al., 2009). As with p68 and p72, the mechanism of p53-mediated
enhancement of pri-miRNA cleavage and how the effect is restricted to a few pri-miRNAs
remain to be elucidated. Further study could also shed light on whether different p53 mutations
affect the levels of subsets of mature miRNAs, and whether these effects contribute to the
pathophysiology of cancer.
Cleavage of pri-miRNAs can also be regulated by “anti-determinants” that inhibit
cleavage or stimulate degradation of the pri-miRNA. One of the earliest examples was the
regulation of let-7 family members by binding of Lin28A to the pri-miRNA terminal loop; this
binding was mediated by a specific motif which may be present in other pri-miRNAs (Newman
et al., 2008; Piskounova et al., 2008; Nam et al., 2011) . Binding was thought to inhibit cleavage
of let-7 pri-miRNAs, and this negative regulation helped maintain the pluripotent state in
embryonic stem cells (Viswanathan et al., 2008). This mechanism was controversial because
Lin-28A is largely localized to the cytosol in embryonic stem cells, and later work demonstrated
that Lin-28A could induce the 3′ terminal polyuridylation of the let-7 pre-miRNA and
subsequent degradation (discussed below). Recently, however, it has been shown that a closely
related paralog, Lin-28B, also inhibits biogenesis of let-7 family members, but is localized to the
22
nucleus and does not appear to interact with terminyl uridyl transferases (Piskounova et al.,
2011).
In particular Lin-28B localizes to nucleoli, where there is little Microprocessor
localization, leading to the model that Lin-28B sequesters let-7 pri-miRNAs to a location where
it is inaccessible to the Microprocessor.
Another example of negative regulation is ADAR1-mediated RNA editing, which causes
the active degradation of some pri-miRNAs. ADAR1, an adenosine deaminase, catalyzes the
conversion of adenosine to inosine in dsRNA. Editing can be highly specific, with editing of
some adenosines but not others within the same substrate, although the contextual determinants
that specify the edited adenosines are not well understood (Nishikura, 2010). A-to-I editing is
detectable in human and mouse precursor and mature miRNAs, and is confined to brainexpressed miRNAs, consistent with the tissue expression pattern of ADAR1 (Blow et al., 2006;
Landgraf et al., 2007; Kawahara et al., 2008; Chiang et al., 2010). The significance of mature
miRNA editing is unclear, although the editing has a propensity to occur in the miRNA seed,
opening the possibility that editing alters the target profile of the mature miRNA (Chiang et al.,
2010).
By contrast, the editing of two pri-miRNAs, mir-142 and mir-151, induces the
degradation of the pri-miRNAs by TudorSN (Yang et al., 2006). Thus A-to-I editing may inhibit
the processing of a subset of miRNAs in a tissue-specific manner.
In summary, several regulatory paradigms have been described that influence the
cleavage of subsets of pri-miRNAs. Since these regulatory schemes affect only subsets of primiRNAs, it seems unlikely that they, either individually or in aggregate, could explain how the
Microprocessor recognizes pri-miRNAs.
Nevertheless, these studies add nuance to our
understanding of how processing can be sensitive to cell type, gene expression state, and
extracellular signals, and could provide insight into how the dysregulation of pri-miRNA
processing could contribute to human disease.
Plant pri-miRNA processing: DCL1
Like the animal miRNAs, plant miRNA biogenesis depends on an RNase III protein.
CARPEL FACTORY (CAF) was identified in a genetic screen for abnormal flower development
(Jacobsen et al., 1999), and encodes an RNase III protein. Its developmental phenotypes and
homology homology to the animal Dicer proteins inspired experiments that demonstrated its
importance in miRNA biogenesis (Park et al., 2002; Reinhart et al., 2002). CAF was later
23
renamed DICER-LIKE1 (DCL1) in recognition of its sequence and functional homology
(Schauer et al., 2002). DCL1 has functions in plants equivalent to those of both Drosha and
Dicer in animals (Park et al., 2002; Reinhart et al., 2002; Kurihara and Watanabe, 2004). Like
the animal Microprocessor, DCL1 and its partners SERRATE and HYPONASTIC LEAVES
(HYL1) appear to recognize junctions between unstructured RNA in internal loops and base
paired stems; the cleavage site is typically 15 bp above the junction (Dong et al., 2008; Mateos et
al., 2010; Song et al., 2010). However, no additional determinants have been identified that
might shed light on how the DCL1 complex distinguishes the appropriate loop-stem junction
corresponding to the pri-miRNA cleavage site from other loop-stem junctions in the pri-miRNA,
much less how DCL1 distinguishes pri-miRNAs from other structured RNAs.
One study
partially-randomized pri-miRNA sequences, expressed the sequences in plants, and selected
functional molecules based on the miRNA overexpression phenotype (Mateos et al., 2010). In
principle, this approach could explore the DCL1 cleavage determinants in great detail, but, in
practice, the low numbers of variants that could be tested limited the study’s ability to find
determinants other than the 15 bp basal stem.
Determinants of canonical biogenesis downstream of the
Microprocessor
Specificity of nuclear export mediated by exportin-5
The product of Microprocessor cleavage, called the pre-miRNA, is exported from the
nucleus by exportin-5 in animals (Yi et al., 2003; Bohnsack et al., 2004; Lund et al., 2004), and
by the homologous protein HASTY in Arabidopsis (Park et al., 2005). Initially characterized
due to its sequence homology to other karyopherin β proteins, exportin-5 was at first thought to
recognize and export dsRBD containing proteins (Brownawell and Macara, 2002). It was soon
shown to export tRNAs and the adenovirus VA1 RNA, both RNAs that contain helices with 3′
overhangs (Bohnsack et al., 2002; Calado et al., 2002; Gwizdek et al., 2003). In fact, short
artificial helices with single-stranded 3′-overhangs are sufficient for exportin-5 recognition and
RanGTP-mediated nuclear export (Gwizdek et al., 2003). A crystal structure of exportin-5 in
complex with a pre-miRNA stem explains this preference: the protein is shaped like a mitt with a
positively charged “palm” that partially wraps the helix, with a positively-charged tunnel at the
24
base of the mitt that accommodates the 3′-overhang (Okada et al., 2009). This tunnel is oriented
in such a way that threading a 2 nt 5′-overhang through it results in a steric clash between the 3′
end and the protein (Okada et al., 2009), explaining the specificity for 3′-overhangs.
Importantly, all protein–RNA contacts are mediated through the phosphate backbone (Okada et
al., 2009), consistent with the view that exportin-5 substrates are defined by their end structure
rather than by sequence. Since RNase III family proteins produce 2 nt 3′-overhangs, exportin-5
is theoretically capable of exporting any product of Drosha cleavage, and seems unlikely to
impose additional constraints on miRNA maturation.
Specificity of cleavage by Dicer
The identification of Dicer emerged from studies of animal RNA interference (RNAi), a
phenomenon where the introduction of exogenous dsRNA derived from the sequences of a
protein-coding gene induced post-transcriptional silencing of that gene (Fire et al., 1998).
Although the dsRNA used to induce RNAi was hundreds of nucleotides long, studies of RNAi
using a Drosophila in vitro lysate system revealed that the long dsRNA was cleaved at regular,
21–22 nt intervals (Zamore et al., 2000). Fractionation of the lysate showed that these short
fragments were associated with target mRNA cleavage activity (Hammond et al., 2000). These
observations led to the view that the long dsRNA is actually a precursor molecule, and the small
fragments derived from it are the active species that guide and induce target mRNA cleavage.
This view was later strengthened by the demonstration that synthetic 21–22 nt fragments were
sufficient to induce mRNA cleavage in the Drosophila lysate and post-transcriptional gene
silencing in mammalian cells (Elbashir et al., 2001a; Elbashir et al., 2001b; Nykanen et al.,
2001).
Given that the RNase III family of enzymes was known to cleave dsRNA into discretelysized products, it seemed likely that the dsRNA-cleaving enzyme would contain an RNase III
domain (Bass, 2000). At that time, the only animal RNase III enzyme that had been described
was Drosha (discussed previously), but analysis of the then-newly available Drosophila
melanogaster and C. elegans genomes picked up three additional, unnamed proteins containing
tandem RNase III domains, one in C. elegans and two in Drosophila (Bernstein et al., 2001).
The Drosophila and human enzymes were sufficient to produce the ~22 nt active fragments from
long dsRNA, and were named “Dicer” accordingly (Bernstein et al., 2001). Consistent with its
25
role in generating the active RNAi-inducing species, loss of Dicer in Drosophila cells and C.
elegans abolished RNAi (Bernstein et al., 2001; Knight and Bass, 2001). This characterization
of the RNAi phenomenon dovetailed with studies of miRNA biogenesis when Dicer was shown
to be necessary for the maturation of lin-4 and let-7 (Grishok et al., 2001; Hutvagner et al., 2001;
Ketting et al., 2001). In Drosophila, Dicer proteins are functionally specialized; Dicer-1 (Dcr-1)
cleaves pre-miRNA hairpins, while Dicer-2 (Dcr-2) processes long dsRNA into siRNAs (Lee et
al., 2004b).
Like Drosha, Dicer requires the association of dsRBD-containing partners for full
activity. Drosophila Dicer-1 (Dcr-1) is associated with Loquacious (Loqs) (Forstemann et al.,
2005; Jiang et al., 2005; Saito et al., 2005), while Dicer-2 (Dcr-2) is associated with R2D2 (Liu
et al., 2003) and additionally depends on isoforms of Loqs for some substrates (Czech et al.,
2008; Okamura et al., 2008; Hartig et al., 2009; Zhou et al., 2009). In mammals, Dicer is
associated with the TAR-element binding protein (TRBP) (Chendrimada et al., 2005; Haase et
al., 2005) and another related protein called PACT (Lee et al., 2006). In humans, TRBP does not
appear to be required for Dicer cleavage of pre-miRNAs (Chendrimada et al., 2005), although its
presence enhances Dicer catalysis of pre-miRNAs and long dsRNA (Chakravarthy et al., 2010).
Similarly, Drosophila Dicer2 processes dsRNA efficiently without R2D2 (Liu et al., 2003). By
contrast, the association of Drosophila Dcr-1 with Loqs is required for processing of the premiRNAs in flies (Forstemann et al., 2005; Jiang et al., 2005; Saito et al., 2005), although some
pre-miRNAs may not depend on Loqs (Liu et al., 2007). No canonical set of substrate binding
preferences have been ascribed to the individual Dicer binding partners, but the Dicer binding
partners can assist in restricting the specificity of Dicer paralogs to specific substrates (Cenik et
al., 2011).
The principal determinant of Dicer cleavage is the structure of the dsRNA ends. Human
Dicer preferentially cleaves from the ends of a dsRNA, suggesting that the phase of dsRNA
cleavage products is set by successive cleavage from the ends (Zhang et al., 2002). Based on
systematic mutagenesis of Dicer, a model for substrate recognition was developed where the
Dicer PAZ domain binds the duplex ends and positions the RNase III domains to cleave the
dsRNA helix (Zhang et al., 2004); once the PAZ domain is positioned, the three-dimensional
organization of domains in Dicer proteins of different organisms sets the length of Dicer
products (Lau et al., 2012). The PAZ domain has a specific preferences for 3′ overhangs at the
26
end of the duplex; a 2 nt overhang with a free 3′-OH is the optimal structure, consistent with
crystal structures of the PAZ domain in complex with a duplex RNA (Ma et al., 2004). In fact,
the recognition of this end structure resides entirely in the PAZ domain: swapping the PAZ
domain for the RNA binding domain of the spliceosomal protein U1A converts Dicer’s
preference for 2 nt 3′-overhangs to a preference for the U1 RNA loop (MacRae et al., 2007).
Consistent with the binding of its PAZ domain, Dicer cleavage of hairpins with 3′-overhangs is
more efficient than cleavage of substrates with blunt ends (Vermeulen et al., 2005).
The
nucleotide identities of the overhanging bases influence binding affinity to the PAZ domain and
Dicer cleavage efficiency, but their contribution is small relative to that of the 2 nt 3′-overhang
(Ma et al., 2004; Vermeulen et al., 2005).
Whether the 5′-phosphate or the 3′-OH in the overhang is more important for defining the
cleavage site is currently debated. Based on the crystal structure of Giardia Dicer, it was
believed that Dicer measured its cleavage site from the 3′-OH of the overhang (Macrae et al.,
2006; MacRae et al., 2007). However, it has been recently proposed that the human Dicer
primarily measures from the 5′-phosphate, and uses measurement from the 3′-OH as a backup
mechanism, explaining why Dicer cleavage sites are preserved in 3′ uridylated pre-miRNAs
(Park et al., 2011). Either way, pre-miRNAs with 2 nt 3′-overhangs are optimal substrates, since
5′ or 3′ single-stranded RNA extensions reduce cleavage efficiency, even when the cleavage site
per se is unaffected (Park et al., 2011). Thus Dicer and its PAZ domain have evolved to
recognize substrates produced by other RNase III cleavage events, consistent with its role in
biogenesis downstream of Drosha and exportin-5.
In addition to the overhang structure, the Drosophila Dcr1-Loqs complex appears to
prefer pre-miRNA-like hairpins with ~22 bp stems capped by a 14 nt loop over substrates with
longer stems and/or shorter loops (Miyoshi et al., 2010; Tsutsumi et al., 2011). Recognition of
pre-miRNA loops depends on the Dcr-1 helicase domain, consistent with a structural model
based on electron microscopy that localizes the Dicer helicase domains to the pre-miRNA apical
stem and loop (Lau et al., 2012). Similarly, human Dicer cleavage is moderately impaired when
the pre-miRNA contains an unusually small loop or short stem (Zhang and Zeng, 2010). This
preference may contribute to the propensity of pri-miRNA hairpins to be ~3 helical turns long.
One possible model is that the Microprocessor has an intrinsic preference for 3-turn helices with
appropriately-sized terminal loops (Zeng et al., 2005; Gottwein et al., 2006; Zhang and Zeng,
27
2010). Dicer then reinforces this structural requirement by preferentially cleaving pre-miRNAs
with ~2 helical turns and the same terminal loop, corresponding to the Microprocessor cleavage
product.
To the extent that Dicer has primary sequence preferences in the pre-miRNA stem, they
are likely to pale beside the structural determinants. Short hairpin RNAs (shRNAs) have been
used extensively to repress the expression of target genes.
In most experimental systems,
shRNAs are transcribed by RNA polymerase III to produce hairpins similar in structure to premiRNAs, which are cleaved by Dicer and ultimately induce repression of target mRNAs
(Brummelkamp et al., 2002; Paddison et al., 2002).
Libraries consisting of hundreds of
thousands of artificial shRNAs have been generated for the purpose of performing loss-offunction genetic screens in mammalian cells; the sequences of the shRNA stems are very
diverse, since it is these sequences which specify targets for cleavage by Argonaute2 within the
RNA-induced silencing complex. Despite the sequence diversity of the library, most examined
shRNAs are cleaved by Dicer and repress target gene expression (Moffat et al., 2006),
demonstrating that Dicer cleaves many hairpins with different sequences but identical structures.
Specificity of loading into Argonaute
Argonaute family proteins form the core of the effector complex that represses targets in
a variety of RNA-induced silencing pathways, including the miRNA pathways in both animals
and plants. The founding member of the family, Arabidopsis AGO1, was identified genetically
in a screen for mutants that cause altered leaf morphology (Bohmert et al., 1998). At the same
time, genetic studies of C. elegans RNAi identified an Argonaute homolog, rde-1, which was
required for RNAi. RNAi seemed similar to a plant phenomenon called post-transcriptional gene
silencing (PTGS), in which the introduction of exogenous transgenes appeared to silence
endogenous genes with the same sequence. This led to the hypothesis that Arabidopsis AGO1
might be related to PTGS; indeed, AGO1 mutants were defective in PTGS (Fagard et al., 2000).
Meanwhile, studies of RNAi in extracts had shown the existence of a nuclease activity that
specifically cleaved target mRNAs, and the protein complex that contained the activity was
called the RNA Induced Silencing Complex (RISC) (Tuschl et al., 1999; Hammond et al., 2000;
Zamore et al., 2000). When RISC was fractionated, a human Argonaute homolog, Argonaute2
28
(Ago2), copurified with the nuclease activity (Hammond et al., 2001), and purified Ago2 carried
out small RNA guided target cleavage (Liu et al., 2004; Meister et al., 2004; Rivas et al., 2005).
Within a given animal or plant species, many Argonaute homologs can be present, which
may be associated with different small RNAs derived from various biogenesis pathways
(Ketting, 2011). For example, the 27 Argonaute superfamily members in C. elegans associate
with Piwi-associated RNAs, endogenously derived siRNAs, exogenously derived siRNAs, and
other small RNAs; of the 27, only ALG-1 and ALG-2 associate with miRNAs (Grishok et al.,
2001). In Drosophila, miRNAs primarily associate with Ago1 (Caudy et al., 2002; Miyoshi et
al., 2009), while in mammals miRNAs are associated with Ago1, Ago2, Ago3, and Ago4 (Liu et
al., 2004; Meister et al., 2004).
Given the plethora of small RNAs and their sorting into different Argonautes, it is
understandable that the Argonaute-loading process inspects Dicer cleavage products for specific
features. Among miRNA-associated Argonautes in animals, Drosophila Ago1 and its loading
are best understood. One principal determinant is stem secondary structure: Ago1 prefers RNA
duplexes with central mismatches, particularly at positions 9-10 of the loaded strand
(Forstemann et al., 2007; Tomari et al., 2007; Kawamata et al., 2009).
Likewise, central
mismatches drive small RNAs into C. elegans ALG-1 (Steiner et al., 2007).
Two other
determinants relate to strand selection: of the two strands in the Dicer cleavage product, only one
strand is preferentially loaded into Ago1. The 5′ end of the loaded strand is usually derived from
the less thermodynamically stable end of the duplex (Schwarz et al., 2003), and the 5′ nucleotide
of the loaded strand is usually U (Czech et al., 2009; Okamura et al., 2009; Ghildiyal et al., 2010;
Seitz et al., 2011). Finally, loading requires a 5′ phosphate, consistent with the products of Dicer
cleavage (Kawamata et al., 2011).
The sorting of small RNAs between Ago1 and Ago2 blurs the distinction between
miRNAs and other small RNAs in flies. On one hand, only Drosophila Ago1 is capable of
mediating targeting using the miRNA seed, which is the basis of most miRNA targeting in
animals; Ago2 efficiently cleaves its targets but requires nearly perfect matches, which is
atypical of metazoan miRNA target sites (Forstemann et al., 2007; Bartel, 2009). Thus RNA
species that are predominantly loaded into Ago2 might not be considered proper miRNAs. On
the other hand, there are hairpins (such as dme-mir-277) encoded in the Drosophila genome that
are processed by the Microprocessor and Dcr-1/Loqs, but the Dcr-1 products are sorted primarily
29
into Ago2 due to extensive pairing (Tomari et al., 2007). Even when the miRNA strand is
selectively loaded into Ago1, the miRNA* strand is often loaded into Ago2 (Czech et al., 2009;
Okamura et al., 2009; Ghildiyal et al., 2010). Thus, different small RNA species can be derived
from the same precursor molecules but come to rest in different maturation endpoints. From a
biogenesis standpoint, it seems that sorting preferences are not really requirements for miRNA
authenticity; instead, one might view them as different ways for the cell to utilize the RNA
precursors that enter the miRNA biogenesis pathway.
Ago loading preferences in vertebrates are less well studied, but are likely to be similar to
those of Drosophila Ago1. Like Drosophila Ago1, loading of mammalian Ago proteins is more
efficient when the duplex contains central mismatches (Yoda et al., 2010). Mammalian Ago
proteins prefer 5′ U or A nucleotides, since mammalian miRNAs and functional shRNAs tend to
start with U or A (Bartel, 2004; Fellmann et al., 2011); structurally, this preference is mediated
by contacts between the Ago MID domain and A and U bases at the 5′ end (Frank et al., 2010).
The 5′ end of the loaded strand is usually derived from the less thermodynamically stable end of
the duplex (Khvorova et al., 2003). No evidence of significant miRNA sorting between Ago
proteins has been found to date (Liu et al., 2004; Meister et al., 2004; Wang et al., 2012).
Regulation of biogenesis in subsets of animal pre-miRNAs
In addition to positive determinants that promote miRNA biogenesis, it is possible in
principle to evolve “anti-determinants” that induce the active elimination of RNA species that
are not authentic pre-miRNAs. There are no convincing examples of this paradigm in the
literature, but individual miRNAs can be dynamically regulated by the active degradation of the
pre-miRNA. This regulation does not help the cell define miRNAs as a class, but does help
control the levels of specific miRNAs in response to internal or external cues. One example is
the ADAR1-mediated editing of pre-mir-151, which inhibits its cleavage by Dicer (Kawahara et
al., 2007). Another, well-studied example is the regulation of let-7 by Lin-28A. Lin-28A binds
two specific motifs in the pre-miRNA loops of let-7 family members (Newman et al., 2008;
Piskounova et al., 2008; Nam et al., 2011) and recruits a terminal uridyl transferase to add
uridines to the 3′ end of the pre-miRNA (Heo et al., 2008; Hagan et al., 2009; Heo et al., 2009;
Lehrbach et al., 2009). Uridylation inhibits Dicer processing of the pre-miRNA and recruits an
unknown nuclease to degrade the pre-miRNA (Heo et al., 2008; Hagan et al., 2009; Heo et al.,
30
2009; Lehrbach et al., 2009). Although Lin-28A regulation was thought to be exclusive to let-7
family members, several other pre-miRNAs have part of the Lin-28 binding motif in their loops,
and the presence of the partial motif correlates with evidence of uridylation, albeit less than that
of pre-let-7 (Heo et al., 2009). Analysis of mature miRNA sequences has shown that terminal
uridylation of mature miRNAs is 3-fold more common in miRNAs derived from the pre-miRNA
3′ arm, suggesting that regulation by uridylation may occur surprisingly frequently (Chiang et al.,
2010). Indeed, this is likely to be an underestimate of regulation by polyuridylation, since
polyuridylated pre-miRNAs are both less likely to be Dicer processed and more likely to be
degraded. The proteins that mediate this putative regulation have not been identified to date, and
it is possible that proteins other than Lin-28 can recruit terminal uridyl transferases to premiRNAs.
Finding additional biogenesis determinants in pri-miRNAs
In summary, the substrate specificity of each successive step in miRNA biogenesis is
largely dictated by the biochemistry of the previous step. This observation is not surprising,
since these steps are joined in a contiguous maturation pathway, but it does reinforce the notion
that authentic miRNA precursors are primarily defined at the first step of biogenesis, when primiRNAs are recognized and cleaved by the Microprocessor.
Given the broad substrate specificity of the Microprocessor, it is not understood how the
complex differentiates between miRNA structures and other hairpins. On an intellectual level,
hairpins are common motifs in structured RNAs, and other RNAs may stochastically assemble
into secondary structures that contain hairpins. Indeed, a genome wide-search found some 11
million hairpins in the human genome (Bentwich et al., 2005). To the extent that they are
transcribed and functionally important, inappropriate cleavage of many of these structures by
Drosha is surely detrimental to the cell. On a practical level, attempts to predict pri-miRNAs
based on canonical secondary structure produce many false-positives, which must be eliminated
using additional criteria, such as evolutionary conservation or experimental evaluation (Lim et
al., 2003a; Lim et al., 2003b; Bentwich et al., 2005; Berezikov et al., 2006; Chiang et al., 2010).
Of course, the Microprocessor has no direct way of assessing the conservation of a pri-miRNA
substrate, so our inability to predict pri-miRNAs from the sequence of a single genome illustrates
our poor understanding of how the Microprocessor recognizes its substrates.
31
The mystery of how Drosha can distinguish pri-miRNAs from other hairpins is part of a
recurrent mystery of how an enzyme with minimal apparent preferences can distinguish its
authentic substrates from other, superficially similar substrates. For example, questions about
the substrate specificity of E. coli RNase III emerged early in the investigation of this enzyme.
Polyoma virus dsRNA could be cleaved exhaustively to produce 11-13 nt fragments (Robertson
and Dunn, 1975), suggesting that RNase III lacked strong nucleotide preferences. It seemed that
RNase III was a general dsRNA endonuclease, at least in vitro, yet it did not seem possible that
an enzyme with few if any discernible substrate preferences (other than secondary structure)
could function in vivo. Indeed, Hugh D. Robertson and John J. Dunn concluded their 1975 paper
on this note:
“In conclusion, we can expect the specific sites in cellular RNAs which are
processed by RNase III to have substantial double helical structure; to be greater
than 20 base pairs in length; to contain 5′-phosphate and 3′-hydroxyl endgroups
after cleavage; and to contain, in all probability, at least one further
characteristic feature, either a common sequence or an additional structural
element, to differentiate them from the many regions of potential secondary
structure now thought to reside at frequent intervals in biological RNA
sequence.”
(Robertson and Dunn, 1975)
Just as investigators studying RNase III in the 1970s recognized that the enzyme had to
have additional determinants for substrate recognition, it is virtually certain that the
Microprocessor recognizes pri-miRNA features beyond the common hairpin structure. I will
describe the known specificity determinants in two classes of RNase III enzymes: the eubacterial
RNase III enzymes, including the eponymous E. coli RNase III; and the yeast RNase III enzymes
Rnt1p and Pac1. Considering these enzymes will provide inspiration about the location and
nature of additional determinants that might define pri-miRNAs, and review experimental
approaches that have been successful for defining substrate specificity in other recognition
paradigms.
32
Substrate specificity in RNase III family proteins
RNase III family proteins have evolved to have divergent cellular roles and substrates, so
the specific preferences of individual RNase III family members may not translate well to the
Microprocessor. However, it is reasonable to believe that the locations and types of preferences
may overlap despite evolutionary divergence.
Indeed, Cα superposition analysis shows
considerable overlap of the structures of Mycobacterium tuberculosis RNase III, Aquifex
aeolicus RNase III, the two endonuclease domains of Giardia intestinalis Dicer, and
Sacchromyces castellii Dcr-1 (Akey and Berger, 2005; Gan et al., 2006; Macrae et al., 2006;
Weinberg et al., 2011). Thus, even though the different classes of RNase III proteins may have
their own idiosyncratic modes of recognition, the preferences of the endonuclease domains
themselves could be relatively well-preserved.
Eubacterial RNase III
The founding member of the RNase III family, the Escherichia coli RNase III, was first
identified as an enzyme which specifically caused dsRNA to become soluble in trichloroacetic
acid (Robertson et al., 1967) in a Mg2+-dependent manner (Robertson et al., 1968), although it
did not have a known biological function at that time. Several years later, another group
determined that the conversion of the T7 phage early transcript from a ~7 kb primary transcript
into five distinct mRNAs depended on a post-transcriptional “sizing factor” which had
chromatographic qualities comparable to that of RNase III (Dunn and Studier, 1973b). Indeed,
processing of the T7 early RNA was defective in an RNase III deficient E. coli strain, and
purified RNase III was sufficient to generate the five messenger RNA products from in vitro
transcribed T7 early RNA (Dunn and Studier, 1973a). Furthermore, the RNase III deficient
strain had delayed production of 16S and 23S ribosomal RNA (rRNA) from a larger RNA
species; as with the T7 early transcript, treatment of this larger RNA with purified RNase III
generated products the same size as the 16S and 23S rRNAs (Dunn and Studier, 1973a).
Consistent with the dsRNA cleavage activity of RNase III, these cleavage sites were later shown
to reside in regions of contiguous pairing between RNA separated by thousands of nucleotides
(Young and Steitz, 1978; Bram et al., 1980). This work was followed by the characterization of
other RNase III cleavage sites, including ones in other phage RNAs (Hughes et al., 1987; Daniels
et al., 1988) and E. coli mRNAs (Barry et al., 1980; Regnier and Portier, 1986; Portier et al.,
33
1987; Regnier and Grunberg-Manago, 1989), including the mRNA encoding RNase III itself
(Bardwell et al., 1989). RNase III also regulates other RNAs through binding without cleavage
(Altuvia et al., 1987).
Efforts to study the RNase III substrate specificity centered on comparative analysis of
the known RNase III cleavage sites.
The observation that cleavage sites in the T7 early
transcript were at least superficially related to each other led to the view that RNase III could be
a restriction endonuclease for RNA: an enzyme which preferentially cleaved dsRNA with a
specific consensus sequence at or near the cleavage site (Robertson, 1982). However, this model
eroded as more RNase III cleavage sites were characterized; in particular, lack of significant
homology between the 16S and 23S rRNA cleavage sites and the T7 early transcript sites
demonstrated that any consensus sequence was, at best, degenerate, if it existed at all.
Nevertheless, aggregation of sequences flanking characterized cleavage sites resulted in
the identification of a common motif:
5′−CUUN NN|−3′
3′−GAAN|NN −5′
where N denotes any nucleotide and “|”
marks the RNase III cleavage sites (Daniels et al., 1988), along with additional preferences
further from the cleavage site (Krinke and Wulff, 1990). However, studies of T7 R1.1 variants
that had altered nucleotide identities but retained Watson–Crick base-pairing showed nearly
identical cleavage rates compared to wildtype (Chelladurai et al., 1991). Instead, it has been
argued that RNase III preferences in this region (“proximal box”) and a second region 5 bp
further away from the cleavage site (“distal box”) are driven by disfavored base pairs or
“antideterminants.” Substitution of the T7 R1.1 base pairs with these disfavored base pairs
results in considerable inhibition of cleavage (Zhang and Nicholson, 1997). On a practical front,
it is difficult to tell whether the observed inhibition by antideterminants is due to truly inhibitory
base pairs, or if the inhibition simply represents the difference between substrates with the most
optimal base pairs and those with the least optimal base pairs. Indeed, a more recent analysis of
determinants of RNase III cleavage has pushed the pendulum back towards the concept of a
consensus sequence. In this study, RNase III preferred
preferred
5′−AG−3′
3′−UU−5′
5′−CWUW NN|−3′
3′−GWAW|NN −5′
in the proximal box, and
in the distal box; shifting these motifs in a dsRNA context was sufficient to
correspondingly shift the cleavage site (Pertzev and Nicholson, 2006).
X-ray crystal structures of Aquifex aeolicus RNase III in complex with model RNA have
revealed RNA-protein contacts in the proximal and distal boxes, and additional contacts between
the two in a region termed the “middle box” (Figure 2A and 2B) (Gan et al., 2006; Gan et al.,
34
2008). The study identified four RNA binding motifs (RBMs). RBMs 1 and 2 occur in the
double strand RNA binding domain (dsRBD) and contact the proximal and middle boxes,
respectively. RBM1 forms 8 contacts to the ribose 2′-OH or the backbone phosphates along the
proximal box and at the cleavage site, explaining the specificity for A-form helical RNA, but
only makes one base contact in the proximal box (Gan et al., 2006; Gan et al., 2008). Likewise,
RBM2 forms a contact to a 2′-OH and a base in the middle box (Gan et al., 2006; Gan et al.,
2008). RBMs 2 and 3 occur in the RNase III domain, and contact the cleavage site and proximal
box, and the distal box, respectively. RBM3 extends into the contacts the two bases immediately
adjacent to the cleavage site, which could translate into base identity preferences at the cleavage
site (Gan et al., 2006; Gan et al., 2008). Interestingly, although RBM4 protrudes into the minor
groove, it did not appear to make any contacts to the bases in the distal box (Gan et al., 2006;
Gan et al., 2008). Overall, the structures explain the specificity of RNase III for dsRNA, but do
not really address the weak preferences for (or against) individual base pairs.
The substrate regions where RNase III has base pair preferences could be important in the
recognition of pri-miRNAs (Figure 2C). Most RNase III contacts with RNA bases occur in the
dsRBDs; since the dsRBDs in different RNase III family proteins contact different parts of the
substrate RNA relative to the cleavage site, it is difficult to know where the dsRBDs of Drosha
and DGCR8 bind the pri-miRNA.
Thus, in drawing parallels between RNase III and the
Microprocessor, the most relevant analogy is between the endonuclease domains in RNase III
and Drosha, corresponding to RBM3 and RBM4. The RBM3 of Drosha may contact the two
bases adjacent to the cleavage site (Gan et al., 2006; Gan et al., 2008), which may translate into a
specific nucleotide preference near its cleavage site. Likewise, RBM4 would be situated on the
3p side of the basal stem, and extends into the minor groove on that side, corresponding to the
distal box of RNase III substrates. Although RBM4 did not make base contacts in the distal box,
it is possible that the RBM4 of Drosha could do so.
Yeast RNase III: Rnt1p and Pac1
The RNase III family member with the most defined substrate preferences is Rnt1p. The
gene encoding this protein was sequenced in the process of exploring Saccharomyces cerevisiae
genome adjacent to a spliceosome factor CUS1. Consistent with its homology with E. coli
RNase III and the Schizosaccharomyces pombe gene pac1, Rnt1p cleaved dsRNA in vitro and
35
A
RNase III domain
RNase III domain
dsRBD domain
dsRBD domain
dsRBD domain
RNase III domain
B
dsRBD domain
C
Bacterial RNAse III substrate
RNase III domain
Microprocessor pri-miRNA substrate
pre-miRNA
RBM2
P9
P7
“Proximal” box
P6
P4
RBM1
P3
P2
RBM3
P1
P10
RBM4
(dsRBD)
(RNase III)
P9
P3
P2
RBM3
-1
-4
+2
-5
+3
-6
+4
-8
+6
-9
+7
-11
+9
-12
+10
Cleavage site
RBM3
“Proximal” box
+1
RBM2
(dsRBD)
RBM4
(RNase III)
“Distal” box
+1
-4
+2
-5
+3
-6
+4
-8
+6
-9
+7
-11
+9
-12
+10
Cleavage site
RBM4
-13
4
-1
+1
6
3’
7
8
+1
5’
+1
“Mid” box
RBM1
“Distal” box
(dsRBD)
RBM3
P4
P1
-1
(RNase III)
“Distal” box
(dsRBD)
P10
“Proximal” box
RBM4
“Proximal” box
(RNase III)
5’
“Mid” box “Distal” box
3’
5’
3’
disruption of RNT1 impaired the removal of both 5′ and 3′ external transcribed spacers (ETS) in
pre-rRNA (Elela et al., 1996). Correspondingly, purified Rnt1p cleaved the 5′ ETS at the A0
cleavage site, and cleaved in the 3′ ETS at a site 21 nt downstream of the 3′ end of mature 28S
rRNA (Elela et al., 1996). Intriguingly, selection of the A0 site had been previously shown to be
dependent on U3 snoRNA binding upstream (Beltrame and Tollervey, 1995), suggesting some
interaction between snoRNAs and Rnt1p, although U3 binding is dispensable for actual cleavage
(Elela et al., 1996).
Despite the superficial similarity to eubacterial rRNA processing, the actual substrate
preferences of Rnt1p are considerably more defined. Further characterization of the RNT1
disrupted strain demonstrated accumulation of many snoRNA precursors (of both H/ACA and
C/D snoRNAs), all of which contained an internal or terminal tetraloop with the degenerate
motif AGNN (Chanfreau et al., 1998). Analysis of an expanded catalog of Rnt1p substrate
sequences, including a panel of snoRNAs, the U1, U2, U4, and U5 small nucleolar RNAs
(snRNAs), and the 3′ ETS cleavage site invariably showed the presence of an (U/A)GNN
tetraloop (nearly always AGNN), situated 13-16 base pairs from the Rnt1p cleavage site (Figure
2) (Chanfreau et al., 2000). Mutation of the A and G nucleotides either severely impaired or
completely abolished cleavage, and changing the tetraloop position shifted the Rnt1p cleavage
site accordingly, demonstrating that the AGNN tetraloop is required for both substrate
recognition and cleavage site selection (Chanfreau et al., 2000).
NMR structure analysis of AGNN tetraloops shows that the tetraloop has a distinct
conformation notable for a syn conformation in the G nucleotide, allowing the base to stack with
the first base of the tetraloop and hydrogen bond to the phosphate in ApG (Figure 2A) (Wu et al.,
2001). The syn conformation causes a backbone turn at the GpN junction (Wu et al., 2001). The
structure is stabilized by non-Watson–Crick base pairs between the first and last bases in the
Figure 2. Specificity determinants in substrates of eubacterial RNase III.
(A) Structural basis of RNase III recognition of substrates. Structure information was taken from PDB
2EZ6.
(B) Locations of published specificity determinants, correlated with protein motifs in the RNase III dsRBD
and RNase III endonuclease domain. The proximal, middle, and distal boxes are shaded in gray,
outlined with the color of the corresponding region in (A).
(C) Inference of potential specificity determinants in pri-miRNAs. The location of determinants is aligned
based on the cleavage site of RNase III and Drosha.
37
tetraloop (Wu et al., 2001) and the adjacent Watson–Crick pairs of the hairpin helix, consistent
with the necessity of these nearby pairs in what has been termed the “binding and stability box”
(Lamontagne et al., 2003). The conformation was conserved between the AGAA, AGUU, and
UGAA tetraloops, but was lost when G was mutated to C to form an ACAA tetraloop (Butcher et
al., 1997; Wu et al., 2001). This opens the possibility that the apparent nucleotide preferences of
Rnt1p actually translate into an RNA conformation preference, which is most easily (or
exclusively) adopted in this specific nucleotide context. Indeed, an NMR structure of the Rnt1p
dsRBD in complex with AGAA tetraloop (Figure 3A) is remarkable for its lack of base-specific
contacts; instead, the α1 helix of the dsRBD contacts tetraloop backbone and the minor groove
formed by its conformation (Wu et al., 2004). In order to distinguish the AGNN tetraloop from
dsRNA, the Rnt1p α1 helix is oriented differently from α1 helices in other dsRBDs (Wu et al.,
2004). Other contacts between the hairpin-tetraloop structure include the α1 helix extending into
the minor groove of the apical stem region, the β3α2 loop resting superficially along stem major
groove, and the β1β2 loop reaching into the minor groove one turn away from the loop (Wu et
al., 2004). These contacts may explain why Rnt1p prefers certain Watson–Crick pairs in the
binding and stability box (Lamontagne et al., 2003). The lack of base-specific contacts suggests
that these preferences may also be conformational in nature.
Even though a structure of the full-length Rnt1p protein is not available at present, it is
very likely that the position of the dsRBD relative to the endonuclease domain in Rnt1p is quite
different from that of other RNase III family members. The uniqueness of the Rnt1p dsRBD
position and its unique conformation compared to other dsRBDs makes it difficult to draw
significant parallels between the substrate recognition of Rnt1p and that of the Microprocessor.
In fact, the other well-characterized yeast RNase III, Pac1, has no discernible preference for
AGNN tetraloops, and instead resembles E. coli RNase III in its relatively relaxed substrate
specificity (Figure 3B) (Rotondo and Frendewey, 1996). However, it is worth noting that, in C.
elegans, trans-splicing between SL1 and cel-let-7 has been reported to be required for the
processing of the pri-miRNA; it was argued that SL1 trans-splicing could alter the predicted
structure of the let-7 pri-miRNA (Bracht et al., 2004). Intriguingly, NMR studies of the donated
portion of the SL1 RNA have demonstrated the presence of an AGUU tetraloop structure above
a buckled A:U pair (Greenbaum et al., 1996). Although the reported G is in a syn conformation,
the base is oriented in the opposite direction as the G in the yeast AGNN tetraloops. It has been
38
argued that the difference is due to a misassigned resonance peak in the SL1 structure (Wu et al.,
2001).
Thus, it is tempting to speculate that AGNN tetraloop recognition plays a role in
nematode pri-miRNA processing, and that SL1 trans-splicing is important to pri-let-7a
processing because it brings an AGNN tetraloop into proximity with the Drosha cleavage site.
Regions recognized by the endonuclease domains of yeast RNase III proteins are more
likely to be relevant to pri-miRNA processing. For Rnt1p, changes to the base pairs immediately
flanking the cleavage site alter the cleavage rate without affecting binding affinity (Lamontagne
et al., 2003). This suggests that Rnt1p substrate affinity is driven primarily by the dsRBD. It
may be more useful to consider Pac1, which has a preference for an internal loop near the
cleavage site in a region analogous to the E. coli proximal box; the position of this loop may
guide the Pac1 cleavage site (Figure 3B) (Lamontagne and Elela, 2004). Although it is not
known whether this internal loop preference is read by the Pac1 dsRBD or the endonuclease
domain, its positioning relative to the cleavage site suggests that the internal loop might be an
endonuclease domain preference. If so, this observation reinforces the idea that RNase III
domains may have substrate preferences in a common region relative to the cleavage site,
although the specific preferences have diverged (Figure 3C).
An exhaustive, quantitative approach to defining pri-miRNAs
It is likely that the Microprocessor has substrate preferences beyond secondary structure,
but no such recognition paradigm has emerged despite many published studies with individual
pri-miRNA variants. When one compares studies of the Microprocessor and studies of bacterial
RNase III, it is apparent that Microprocessor studies are missing two things that were critical to
elucidating RNase III substrate determinants.
First, pri-miRNA cleavage must be measured quantitatively. Nearly all published primiRNA cleavage experiments thus far have been non-quantitative; typical read-outs are
“cleaved” or “uncleaved.” But these qualitative experiments will not suffice to elucidate the
determinants that distinguish pri-miRNA hairpins from other hairpins. For bacterial RNase III,
several determinants contribute relatively subtly to differences in binding affinity and cleavage
rate; together, the presence of these determinants (or the absence of anti-determinants, if they
indeed exist) define the RNase III substrate. The distinguishing features of pri-miRNA hairpins
39
A
Minor groove
Major groove
dsRBD domain
Minor groove
dsRBD domain
A
A
B
G
A
dsRBD domain
S. cerevisiae Rnt1p substrate
3’
C
S. pombe Pac1p substrate
5’
3’
Microprocessor pri-miRNA substrate
5’
pre-miRNA
P1
P1
P1
-1
-1
-1
Cleavage site
+2
dsRBD
G
-11
+9
-12
+10
RNase III
(?)
-13
8
A
N
+3
6
+14
N
+2
+1
+13
-16
-5
-6
+1
-15
RNase III
(?)
Cleavage site
7
+12
+3
AGNN
tetraloop
+11
-14
+2
+1
4
-1
dsRBD
-13
-5
-6
Cleavage site
“Binding and
stability” box
+3
+1
+1
+1
5’
3’
may individually contribute just the difference between moderate and maximal cleavage
efficiency.
Indeed, many mutations introduced into pri-miRNA substrates alter cleavage
efficiency without completely abolishing cleavage, but these substrates are often lumped into the
“cleaved” category, masking the potential contribution of additional determinants. For example,
mutations in the basal stem that shift the Microprocessor cleavage site also reduce the overall
efficiency of cleavage (Han et al., 2006), suggesting either that the mutations perturb an
important sequence or that the new cleavage site is somehow in a suboptimal context. Either
way, it is clear that our understanding of recognition and cleavage is incomplete.
Second, studies must use large sets of variants derived from individual pri-miRNAs,
enabling a more precise understanding of what specific sequence motifs or structural features are
important to recognition by the Microprocessor. For example, conclusions about the optimal
apical stem length for cleavage were primarily based on deletions or mutations that abolished
base pairing (Zeng et al., 2005; Han et al., 2006; Zhang and Zeng, 2010). Based on these
experiments, it is extremely difficult to tell whether recognition by the Microprocessor was
impaired because stem length per se was important, or because a specific recognition
determinant resides in this region and had been deleted or mutated. The approach of mutating
away base pairs is even more prone to misinterpretation, since at least three mechanisms could
be in play: stem length, loop or unstructured RNA length, and primary sequence. Similarly,
studies of the flanking RNA sequence were largely based on deletions (Lee et al., 2003; Zeng
and Cullen, 2005), and the results are compatible with the idea that additional, subtle
determinants reside in the flanking RNA. By contrast, studies of the proximal, mid, and distal
boxes in the substrates of bacterial RNase III systematically tested most or all possible Watson–
Crick base pairs at the interrogated positions (Pertzev and Nicholson, 2006). Similarly detailed
studies will be needed to make significant progress in understanding the cleavage of pri-
Figure 3. Specificity determinants in substrates of yeast RNase III enzymes Rnt1p and Pac1.
(A) Structural basis of Rnt1p recognition of the AGNN tetraloop. Structure information was taken from
PDB 1T4L.
(B) Locations of published specificity determinants, correlated with protein motifs where information is
available. Recognition boxes for Rnt1p and Pac1 are shown.
(C) Inference of potential specificity determinants in pri-miRNAs. The location of determinants is aligned
based on the cleavage sites of Rnt1p, Pac1, and Drosha.
41
miRNAs.
In seeking large sets of pri-miRNA variants, it will probably not be sufficient to gather
collections of the pri-miRNAs encoded in animal genomes and derive determinants from their
common sequences. Many groups have attempted to use sequence analysis and computational
learning models to analyze pri-miRNAs for predictive determinants, largely for the purpose of
computationally predicting novel pri-miRNAs (Grad et al., 2003; Lim et al., 2003a; Lim et al.,
2003b; Bentwich et al., 2005; Nam et al., 2005; Berezikov et al., 2006). The resulting algorithms
perform surprisingly poorly unless conservation or experimental data are taken into account (Lim
et al., 2003a; Lim et al., 2003b; Ruby et al., 2006; Ruby et al., 2007b; Stark et al., 2007; Chiang
et al., 2010). Others have attempted to quantify the Microprocessor cleavage of 250 pri-miRNAs,
in the hope that measured cleavage efficiencies would offer an extra dimension of information
that straight sequence analysis of pri-miRNAs may have lacked. This study was not much more
successful than comparative sequence analysis: it found that conserved pri-miRNAs were
processed more efficiently than less conserved pri-miRNAs, and that the general structural
features defined previously were correlated with cleavage efficiency (Feng et al., 2011). The
success of these studies was likely limited by the small number of “true positive” hairpins in the
training sets, and their wide sequence and structural divergence. In particular, sequence and
structural divergence severely hampers alignment, which in turn limits the power of
computational analysis to discover short or degenerate motifs.
What is needed is an experimental system that systematically generates a large number of
related hairpin substrates, and quantifies the cleavage of these hairpins. These hairpins must be
sufficiently divergent to sample suboptimal pri-miRNA sequences and structures, but sufficiently
similar to enable computational sequence analysis. In the following chapters I will describe such
an experimental and computational approach. Hundreds of billions of pri-miRNA variants were
generated, each related to one of four human pri-miRNAs.
Of these variants, millions of
functional variants cleaved by the Microprocessor were sequenced. Computational analysis of
the successful variants revealed a panel of important determinants of pri-miRNA recognition,
and quantified the relative contribution of these determinants. Together, these evolutionarilyconserved features define the majority of authentic human pri-miRNAs. The elucidation of these
features greatly expands the understanding of what pri-miRNAs are, and how the cell recognizes
the correct hairpins to process into mature miRNAs.
42
Bibliography and References Cited
Akey, D.L., and Berger, J.M. (2005). Structure of the nuclease domain of ribonuclease III from
M. tuberculosis at 2.1 A. Protein Sci 14, 2744-2750.
Altuvia, S., Locker-Giladi, H., Koby, S., Ben-Nun, O., and Oppenheim, A.B. (1987). RNase III
stimulates the translation of the cIII gene of bacteriophage lambda. Proc Natl Acad Sci U S A
84, 6511-6515.
Ambros, V., Bartel, B., Bartel, D.P., Burge, C.B., Carrington, J.C., Chen, X., Dreyfuss, G., Eddy,
S.R., Griffiths-Jones, S., Marshall, M., et al. (2003). A uniform system for microRNA
annotation. RNA 9, 277-279.
Arvey, A., Larsson, E., Sander, C., Leslie, C.S., and Marks, D.S. (2010). Target mRNA
abundance dilutes microRNA and siRNA activity. Mol Syst Biol 6, 363.
Ashburner, M., Ball, C.A., Blake, J.A., Botstein, D., Butler, H., Cherry, J.M., Davis, A.P.,
Dolinski, K., Dwight, S.S., Eppig, J.T., et al. (2000). Gene ontology: tool for the unification
of biology. The Gene Ontology Consortium. Nat Genet 25, 25-29.
Babiarz, J.E., Ruby, J.G., Wang, Y., Bartel, D.P., and Blelloch, R. (2008). Mouse ES cells
express endogenous shRNAs, siRNAs, and other Microprocessor-independent, Dicerdependent small RNAs. Genes Dev 22, 2773-2785.
Baek, D., Villen, J., Shin, C., Camargo, F.D., Gygi, S.P., and Bartel, D.P. (2008). The impact of
microRNAs on protein output. Nature 455, 64-71.
Bardwell, J.C., Regnier, P., Chen, S.M., Nakamura, Y., Grunberg-Manago, M., and Court, D.L.
(1989). Autoregulation of RNase III operon by mRNA processing. EMBO J 8, 3401-3407.
Barry, G., Squires, C., and Squires, C.L. (1980). Attenuation and processing of RNA from the
rplJL--rpoBC transcription unit of Escherichia coli. Proc Natl Acad Sci U S A 77, 33313335.
Bartel, D.P. (2004). MicroRNAs: genomics, biogenesis, mechanism, and function. Cell 116, 281297.
Bartel, D.P. (2009). MicroRNAs: target recognition and regulatory functions. Cell 136, 215-233.
Bass, B.L. (2000). Double-stranded RNA as a template for gene silencing. Cell 101, 235-238.
Bazzini, A.A., Lee, M.T., and Giraldez, A.J. (2012). Ribosome profiling shows that miR-430
reduces translation before causing mRNA decay in zebrafish. Science 336, 233-237.
Beltrame, M., and Tollervey, D. (1995). Base pairing between U3 and the pre-ribosomal RNA is
required for 18S rRNA synthesis. EMBO J 14, 4350-4356.
Bentwich, I., Avniel, A., Karov, Y., Aharonov, R., Gilad, S., Barad, O., Barzilai, A., Einat, P.,
Einav, U., Meiri, E., et al. (2005). Identification of hundreds of conserved and nonconserved
human microRNAs. Nat Genet 37, 766-770.
Berezikov, E., van Tetering, G., Verheul, M., van de Belt, J., van Laake, L., Vos, J., Verloop, R.,
van de Wetering, M., Guryev, V., Takada, S., et al. (2006). Many novel mammalian
microRNA candidates identified by extensive cloning and RAKE analysis. Genome Res 16,
1289-1298.
Bernstein, E., Caudy, A.A., Hammond, S.M., and Hannon, G.J. (2001). Role for a bidentate
ribonuclease in the initiation step of RNA interference. Nature 409, 363-366.
Bernstein, E., Kim, S.Y., Carmell, M.A., Murchison, E.P., Alcorn, H., Li, M.Z., Mills, A.A.,
Elledge, S.J., Anderson, K.V., and Hannon, G.J. (2003). Dicer is essential for mouse
development. Nat Genet 35, 215-217.
43
Blow, M.J., Grocock, R.J., van Dongen, S., Enright, A.J., Dicks, E., Futreal, P.A., Wooster, R.,
and Stratton, M.R. (2006). RNA editing of human microRNAs. Genome Biol 7, R27.
Bohmert, K., Camus, I., Bellini, C., Bouchez, D., Caboche, M., and Benning, C. (1998). AGO1
defines a novel locus of Arabidopsis controlling leaf development. EMBO J 17, 170-180.
Bohnsack, M.T., Czaplinski, K., and Gorlich, D. (2004). Exportin 5 is a RanGTP-dependent
dsRNA-binding protein that mediates nuclear export of pre-miRNAs. RNA 10, 185-191.
Bohnsack, M.T., Regener, K., Schwappach, B., Saffrich, R., Paraskeva, E., Hartmann, E., and
Gorlich, D. (2002). Exp5 exports eEF1A via tRNA from nuclei and synergizes with other
transport pathways to confine translation to the cytoplasm. EMBO J 21, 6205-6215.
Bracht, J., Hunter, S., Eachus, R., Weeks, P., and Pasquinelli, A.E. (2004). Trans-splicing and
polyadenylation of let-7 microRNA primary transcripts. RNA 10, 1586-1594.
Bram, R.J., Young, R.A., and Steitz, J.A. (1980). The ribonuclease III site flanking 23S
sequences in the 30S ribosomal precursor RNA of E. coli. Cell 19, 393-401.
Brownawell, A.M., and Macara, I.G. (2002). Exportin-5, a novel karyopherin, mediates nuclear
export of double-stranded RNA binding proteins. J Cell Biol 156, 53-64.
Brummelkamp, T.R., Bernards, R., and Agami, R. (2002). A system for stable expression of
short interfering RNAs in mammalian cells. Science 296, 550-553.
Butcher, S.E., Dieckmann, T., and Feigon, J. (1997). Solution structure of the conserved 16 Slike ribosomal RNA UGAA tetraloop. J Mol Biol 268, 348-358.
Calabrese, J.M., Seila, A.C., Yeo, G.W., and Sharp, P.A. (2007). RNA sequence analysis defines
Dicer's role in mouse embryonic stem cells. Proc Natl Acad Sci U S A 104, 18097-18102.
Calado, A., Treichel, N., Muller, E.C., Otto, A., and Kutay, U. (2002). Exportin-5-mediated
nuclear export of eukaryotic elongation factor 1A and tRNA. EMBO J 21, 6216-6224.
Calin, G.A., Dumitru, C.D., Shimizu, M., Bichi, R., Zupo, S., Noch, E., Aldler, H., Rattan, S.,
Keating, M., Rai, K., et al. (2002). Frequent deletions and down-regulation of micro- RNA
genes miR15 and miR16 at 13q14 in chronic lymphocytic leukemia. Proc Natl Acad Sci U S
A 99, 15524-15529.
Calin, G.A., Ferracin, M., Cimmino, A., Di Leva, G., Shimizu, M., Wojcik, S.E., Iorio, M.V.,
Visone, R., Sever, N.I., Fabbri, M., et al. (2005). A MicroRNA signature associated with
prognosis and progression in chronic lymphocytic leukemia. N Engl J Med 353, 1793-1801.
Calin, G.A., Sevignani, C., Dumitru, C.D., Hyslop, T., Noch, E., Yendamuri, S., Shimizu, M.,
Rattan, S., Bullrich, F., Negrini, M., et al. (2004). Human microRNA genes are frequently
located at fragile sites and genomic regions involved in cancers. Proc Natl Acad Sci U S A
101, 2999-3004.
Caudy, A.A., Myers, M., Hannon, G.J., and Hammond, S.M. (2002). Fragile X-related protein
and VIG associate with the RNA interference machinery. Genes Dev 16, 2491-2496.
Cenik, E.S., Fukunaga, R., Lu, G., Dutcher, R., Wang, Y., Tanaka Hall, T.M., and Zamore, P.D.
(2011). Phosphate and R2D2 restrict the substrate specificity of Dicer-2, an ATP-driven
ribonuclease. Mol Cell 42, 172-184.
Chakravarthy, S., Sternberg, S.H., Kellenberger, C.A., and Doudna, J.A. (2010). Substratespecific kinetics of Dicer-catalyzed RNA processing. J Mol Biol 404, 392-402.
Chanfreau, G., Buckle, M., and Jacquier, A. (2000). Recognition of a conserved class of RNA
tetraloops by Saccharomyces cerevisiae RNase III. Proc Natl Acad Sci U S A 97, 3142-3147.
Chanfreau, G., Legrain, P., and Jacquier, A. (1998). Yeast RNase III as a key processing enzyme
in small nucleolar RNAs metabolism. J Mol Biol 284, 975-988.
44
Chelladurai, B.S., Li, H., and Nicholson, A.W. (1991). A conserved sequence element in
ribonuclease III processing signals is not required for accurate in vitro enzymatic cleavage.
Nucleic Acids Res 19, 1759-1766.
Cheloufi, S., Dos Santos, C.O., Chong, M.M., and Hannon, G.J. (2010). A dicer-independent
miRNA biogenesis pathway that requires Ago catalysis. Nature 465, 584-589.
Chen, C.Z., Li, L., Lodish, H.F., and Bartel, D.P. (2004). MicroRNAs modulate hematopoietic
lineage differentiation. Science 303, 83-86.
Chendrimada, T.P., Gregory, R.I., Kumaraswamy, E., Norman, J., Cooch, N., Nishikura, K., and
Shiekhattar, R. (2005). TRBP recruits the Dicer complex to Ago2 for microRNA processing
and gene silencing. Nature 436, 740-744.
Chiang, H.R., Schoenfeld, L.W., Ruby, J.G., Auyeung, V.C., Spies, N., Baek, D., Johnston,
W.K., Russ, C., Luo, S., Babiarz, J.E., et al. (2010). Mammalian microRNAs: experimental
evaluation of novel and previously annotated genes. Genes Dev 24, 992-1009.
Cifuentes, D., Xue, H., Taylor, D.W., Patnode, H., Mishima, Y., Cheloufi, S., Ma, E., Mane, S.,
Hannon, G.J., Lawson, N.D., et al. (2010). A novel miRNA processing pathway independent
of Dicer requires Argonaute2 catalytic activity. Science 328, 1694-1698.
Czech, B., Malone, C.D., Zhou, R., Stark, A., Schlingeheyde, C., Dus, M., Perrimon, N., Kellis,
M., Wohlschlegel, J.A., Sachidanandam, R., et al. (2008). An endogenous small interfering
RNA pathway in Drosophila. Nature 453, 798-802.
Czech, B., Zhou, R., Erlich, Y., Brennecke, J., Binari, R., Villalta, C., Gordon, A., Perrimon, N.,
and Hannon, G.J. (2009). Hierarchical rules for Argonaute loading in Drosophila. Mol Cell
36, 445-456.
Daniels, D.L., Subbarao, M.N., Blattner, F.R., and Lozeron, H.A. (1988). Q-mediated late gene
transcription of bacteriophage lambda: RNA start point and RNase III processing sites in
vivo. Virology 167, 568-577.
Davis, B.N., Hilyard, A.C., Lagna, G., and Hata, A. (2008). SMAD proteins control DROSHAmediated microRNA maturation. Nature 454, 56-61.
Davis, B.N., Hilyard, A.C., Nguyen, P.H., Lagna, G., and Hata, A. (2010). Smad proteins bind a
conserved RNA sequence to promote microRNA maturation by Drosha. Mol Cell 39, 373384.
Denli, A.M., Tops, B.B., Plasterk, R.H., Ketting, R.F., and Hannon, G.J. (2004). Processing of
primary microRNAs by the Microprocessor complex. Nature 432, 231-235.
Doench, J.G., and Sharp, P.A. (2004). Specificity of microRNA target selection in translational
repression. Genes Dev 18, 504-511.
Dong, Z., Han, M.H., and Fedoroff, N. (2008). The RNA-binding proteins HYL1 and SE
promote accurate in vitro processing of pri-miRNA by DCL1. Proc Natl Acad Sci U S A 105,
9970-9975.
Dreyfuss, G., Matunis, M.J., Pinol-Roma, S., and Burd, C.G. (1993). hnRNP proteins and the
biogenesis of mRNA. Annu Rev Biochem 62, 289-321.
Duan, R., Pak, C., and Jin, P. (2007). Single nucleotide polymorphism associated with mature
miR-125a alters the processing of pri-miRNA. Hum Mol Genet 16, 1124-1131.
Dunn, J.J., and Studier, F.W. (1973a). T7 early RNAs and Escherichia coli ribosomal RNAs are
cut from large precursor RNAs in vivo by ribonuclease 3. Proc Natl Acad Sci U S A 70,
3296-3300.
Dunn, J.J., and Studier, F.W. (1973b). T7 early RNAs are generated by site-specific cleavages.
Proc Natl Acad Sci U S A 70, 1559-1563.
45
Elbashir, S.M., Harborth, J., Lendeckel, W., Yalcin, A., Weber, K., and Tuschl, T. (2001a).
Duplexes of 21-nucleotide RNAs mediate RNA interference in cultured mammalian cells.
Nature 411, 494-498.
Elbashir, S.M., Lendeckel, W., and Tuschl, T. (2001b). RNA interference is mediated by 21- and
22-nucleotide RNAs. Genes Dev 15, 188-200.
Elela, S.A., Igel, H., and Ares, M., Jr. (1996). RNase III cleaves eukaryotic preribosomal RNA at
a U3 snoRNP-dependent site. Cell 85, 115-124.
Fagard, M., Boutet, S., Morel, J.B., Bellini, C., and Vaucheret, H. (2000). AGO1, QDE-2, and
RDE-1 are related proteins required for post-transcriptional gene silencing in plants, quelling
in fungi, and RNA interference in animals. Proc Natl Acad Sci U S A 97, 11650-11654.
Faller, M., Toso, D., Matsunaga, M., Atanasov, I., Senturia, R., Chen, Y., Zhou, Z.H., and Guo,
F. (2010). DGCR8 recognizes primary transcripts of microRNAs through highly cooperative
binding and formation of higher-order structures. RNA 16, 1570-1583.
Farh, K.K., Grimson, A., Jan, C., Lewis, B.P., Johnston, W.K., Lim, L.P., Burge, C.B., and
Bartel, D.P. (2005). The widespread impact of mammalian MicroRNAs on mRNA repression
and evolution. Science 310, 1817-1821.
Fellmann, C., Zuber, J., McJunkin, K., Chang, K., Malone, C.D., Dickins, R.A., Xu, Q.,
Hengartner, M.O., Elledge, S.J., Hannon, G.J., et al. (2011). Functional identification of
optimized RNAi triggers using a massively parallel sensor assay. Mol Cell 41, 733-746.
Feng, Y., Zhang, X., Song, Q., Li, T., and Zeng, Y. (2011). Drosha processing controls the
specificity and efficiency of global microRNA expression. Biochim Biophys Acta 1809, 700707.
Filippov, V., Solovyev, V., Filippova, M., and Gill, S.S. (2000). A novel type of RNase III
family proteins in eukaryotes. Gene 245, 213-221.
Fire, A., Albertson, D., Harrison, S.W., and Moerman, D.G. (1991). Production of antisense
RNA leads to effective and specific inhibition of gene expression in C. elegans muscle.
Development 113, 503-514.
Fire, A., Xu, S., Montgomery, M.K., Kostas, S.A., Driver, S.E., and Mello, C.C. (1998). Potent
and specific genetic interference by double-stranded RNA in Caenorhabditis elegans. Nature
391, 806-811.
Flynt, A.S., Greimann, J.C., Chung, W.J., Lima, C.D., and Lai, E.C. (2010). MicroRNA
biogenesis via splicing and exosome-mediated trimming in Drosophila. Mol Cell 38, 900907.
Forstemann, K., Horwich, M.D., Wee, L., Tomari, Y., and Zamore, P.D. (2007). Drosophila
microRNAs are sorted into functionally distinct argonaute complexes after production by
dicer-1. Cell 130, 287-297.
Forstemann, K., Tomari, Y., Du, T., Vagin, V.V., Denli, A.M., Bratu, D.P., Klattenhoff, C.,
Theurkauf, W.E., and Zamore, P.D. (2005). Normal microRNA maturation and germ-line
stem cell maintenance requires Loquacious, a double-stranded RNA-binding domain protein.
PLoS Biol 3, e236.
Frank, F., Sonenberg, N., and Nagar, B. (2010). Structural basis for 5'-nucleotide base-specific
recognition of guide RNA by human AGO2. Nature 465, 818-822.
Friedman, R.C., Farh, K.K., Burge, C.B., and Bartel, D.P. (2009). Most mammalian mRNAs are
conserved targets of microRNAs. Genome Res 19, 92-105.
Fukuda, T., Yamagata, K., Fujiyama, S., Matsumoto, T., Koshida, I., Yoshimura, K., Mihara, M.,
Naitou, M., Endoh, H., Nakamura, T., et al. (2007). DEAD-box RNA helicase subunits of the
46
Drosha complex are required for processing of rRNA and a subset of microRNAs. Nat Cell
Biol 9, 604-611.
Gan, J., Shaw, G., Tropea, J.E., Waugh, D.S., Court, D.L., and Ji, X. (2008). A stepwise model
for double-stranded RNA processing by ribonuclease III. Mol Microbiol 67, 143-154.
Gan, J., Tropea, J.E., Austin, B.P., Court, D.L., Waugh, D.S., and Ji, X. (2006). Structural insight
into the mechanism of double-stranded RNA processing by ribonuclease III. Cell 124, 355366.
Garcia, D.M., Baek, D., Shin, C., Bell, G.W., Grimson, A., and Bartel, D.P. (2011). Weak seedpairing stability and high target-site abundance decrease the proficiency of lsy-6 and other
microRNAs. Nat Struct Mol Biol 18, 1139-1146.
Ghildiyal, M., Xu, J., Seitz, H., Weng, Z., and Zamore, P.D. (2010). Sorting of Drosophila small
silencing RNAs partitions microRNA* strands into the RNA interference pathway. RNA 16,
43-56.
Gottwein, E., Cai, X., and Cullen, B.R. (2006). A novel assay for viral microRNA function
identifies a single nucleotide polymorphism that affects Drosha processing. J Virol 80, 53215326.
Grad, Y., Aach, J., Hayes, G.D., Reinhart, B.J., Church, G.M., Ruvkun, G., and Kim, J. (2003).
Computational and experimental identification of C. elegans microRNAs. Mol Cell 11, 12531263.
Greenbaum, N.L., Radhakrishnan, I., Patel, D.J., and Hirsh, D. (1996). Solution structure of the
donor site of a trans-splicing RNA. Structure 4, 725-733.
Gregory, R.I., Yan, K.P., Amuthan, G., Chendrimada, T., Doratotaj, B., Cooch, N., and
Shiekhattar, R. (2004). The Microprocessor complex mediates the genesis of microRNAs.
Nature 432, 235-240.
Griffiths-Jones, S., Grocock, R.J., van Dongen, S., Bateman, A., and Enright, A.J. (2006).
miRBase: microRNA sequences, targets and gene nomenclature. Nucleic Acids Res 34,
D140-144.
Grimson, A., Farh, K.K., Johnston, W.K., Garrett-Engele, P., Lim, L.P., and Bartel, D.P. (2007).
MicroRNA targeting specificity in mammals: determinants beyond seed pairing. Mol Cell
27, 91-105.
Grishok, A., Pasquinelli, A.E., Conte, D., Li, N., Parrish, S., Ha, I., Baillie, D.L., Fire, A.,
Ruvkun, G., and Mello, C.C. (2001). Genes and mechanisms related to RNA interference
regulate expression of the small temporal RNAs that control C. elegans developmental
timing. Cell 106, 23-34.
Guil, S., and Caceres, J.F. (2007). The multifunctional RNA-binding protein hnRNP A1 is
required for processing of miR-18a. Nat Struct Mol Biol 14, 591-596.
Guo, H., Ingolia, N.T., Weissman, J.S., and Bartel, D.P. (2010). Mammalian microRNAs
predominantly act to decrease target mRNA levels. Nature 466, 835-840.
Guo, S., and Kemphues, K.J. (1995). par-1, a gene required for establishing polarity in C.
elegans embryos, encodes a putative Ser/Thr kinase that is asymmetrically distributed. Cell
81, 611-620.
Gwizdek, C., Ossareh-Nazari, B., Brownawell, A.M., Doglio, A., Bertrand, E., Macara, I.G., and
Dargemont, C. (2003). Exportin-5 mediates nuclear export of minihelix-containing RNAs. J
Biol Chem 278, 5505-5508.
47
Haase, A.D., Jaskiewicz, L., Zhang, H., Laine, S., Sack, R., Gatignol, A., and Filipowicz, W.
(2005). TRBP, a regulator of cellular PKR and HIV-1 virus expression, interacts with Dicer
and functions in RNA silencing. EMBO Rep 6, 961-967.
Hagan, J.P., Piskounova, E., and Gregory, R.I. (2009). Lin28 recruits the TUTase Zcchc11 to
inhibit let-7 maturation in mouse embryonic stem cells. Nat Struct Mol Biol 16, 1021-1025.
Hamilton, A.J., and Baulcombe, D.C. (1999). A species of small antisense RNA in
posttranscriptional gene silencing in plants. Science 286, 950-952.
Hammond, S.M., Bernstein, E., Beach, D., and Hannon, G.J. (2000). An RNA-directed nuclease
mediates post-transcriptional gene silencing in Drosophila cells. Nature 404, 293-296.
Hammond, S.M., Boettcher, S., Caudy, A.A., Kobayashi, R., and Hannon, G.J. (2001).
Argonaute2, a link between genetic and biochemical analyses of RNAi. Science 293, 11461150.
Han, J., Lee, Y., Yeom, K.H., Kim, Y.K., Jin, H., and Kim, V.N. (2004). The Drosha-DGCR8
complex in primary microRNA processing. Genes Dev 18, 3016-3027.
Han, J., Lee, Y., Yeom, K.H., Nam, J.W., Heo, I., Rhee, J.K., Sohn, S.Y., Cho, Y., Zhang, B.T.,
and Kim, V.N. (2006). Molecular basis for the recognition of primary microRNAs by the
Drosha-DGCR8 complex. Cell 125, 887-901.
Han, J., Pedersen, J.S., Kwon, S.C., Belair, C.D., Kim, Y.K., Yeom, K.H., Yang, W.Y.,
Haussler, D., Blelloch, R., and Kim, V.N. (2009). Posttranscriptional crossregulation
between Drosha and DGCR8. Cell 136, 75-84.
Hartig, J.V., Esslinger, S., Bottcher, R., Saito, K., and Forstemann, K. (2009). Endo-siRNAs
depend on a new isoform of loquacious and target artificially introduced, high-copy
sequences. EMBO J 28, 2932-2944.
Hendrickson, D.G., Hogan, D.J., McCullough, H.L., Myers, J.W., Herschlag, D., Ferrell, J.E.,
and Brown, P.O. (2009). Concordant regulation of translation and mRNA abundance for
hundreds of targets of a human microRNA. PLoS Biol 7, e1000238.
Heo, I., Joo, C., Cho, J., Ha, M., Han, J., and Kim, V.N. (2008). Lin28 mediates the terminal
uridylation of let-7 precursor MicroRNA. Mol Cell 32, 276-284.
Heo, I., Joo, C., Kim, Y.K., Ha, M., Yoon, M.J., Cho, J., Yeom, K.H., Han, J., and Kim, V.N.
(2009). TUT4 in concert with Lin28 suppresses microRNA biogenesis through premicroRNA uridylation. Cell 138, 696-708.
Hughes, J.A., Brown, L.R., and Ferro, A.J. (1987). Nucleotide sequence and analysis of the
coliphage T3 S-adenosylmethionine hydrolase gene and its surrounding ribonuclease III
processing sites. Nucleic Acids Res 15, 717-729.
Hutvagner, G., McLachlan, J., Pasquinelli, A.E., Balint, E., Tuschl, T., and Zamore, P.D. (2001).
A cellular function for the RNA-interference enzyme Dicer in the maturation of the let-7
small temporal RNA. Science 293, 834-838.
Izant, J.G., and Weintraub, H. (1984). Inhibition of thymidine kinase gene expression by antisense RNA: a molecular approach to genetic analysis. Cell 36, 1007-1015.
Jacobsen, S.E., Running, M.P., and Meyerowitz, E.M. (1999). Disruption of an RNA
helicase/RNAse III gene in Arabidopsis causes unregulated cell division in floral meristems.
Development 126, 5231-5243.
Jiang, F., Ye, X., Liu, X., Fincher, L., McKearin, D., and Liu, Q. (2005). Dicer-1 and R3D1-L
catalyze microRNA maturation in Drosophila. Genes Dev 19, 1674-1679.
Jones-Rhoades, M.W., Bartel, D.P., and Bartel, B. (2006). MicroRNAS and their regulatory roles
in plants. Annu Rev Plant Biol 57, 19-53.
48
Kawahara, Y., Megraw, M., Kreider, E., Iizasa, H., Valente, L., Hatzigeorgiou, A.G., and
Nishikura, K. (2008). Frequency and fate of microRNA editing in human brain. Nucleic
Acids Res 36, 5270-5280.
Kawahara, Y., Zinshteyn, B., Chendrimada, T.P., Shiekhattar, R., and Nishikura, K. (2007).
RNA editing of the microRNA-151 precursor blocks cleavage by the Dicer-TRBP complex.
EMBO Rep 8, 763-769.
Kawamata, T., Seitz, H., and Tomari, Y. (2009). Structural determinants of miRNAs for RISC
loading and slicer-independent unwinding. Nat Struct Mol Biol 16, 953-960.
Kawamata, T., Yoda, M., and Tomari, Y. (2011). Multilayer checkpoints for microRNA
authenticity during RISC assembly. EMBO Rep 12, 944-949.
Ketting, R.F. (2011). The many faces of RNAi. Dev Cell 20, 148-161.
Ketting, R.F., Fischer, S.E., Bernstein, E., Sijen, T., Hannon, G.J., and Plasterk, R.H. (2001).
Dicer functions in RNA interference and in synthesis of small RNA involved in
developmental timing in C. elegans. Genes Dev 15, 2654-2659.
Khvorova, A., Reynolds, A., and Jayasena, S.D. (2003). Functional siRNAs and miRNAs exhibit
strand bias. Cell 115, 209-216.
Knight, S.W., and Bass, B.L. (2001). A role for the RNase III enzyme DCR-1 in RNA
interference and germ line development in Caenorhabditis elegans. Science 293, 2269-2271.
Krinke, L., and Wulff, D.L. (1990). The cleavage specificity of RNase III. Nucleic Acids Res 18,
4809-4815.
Kumar, M.S., Lu, J., Mercer, K.L., Golub, T.R., and Jacks, T. (2007). Impaired microRNA
processing enhances cellular transformation and tumorigenesis. Nat Genet 39, 673-677.
Kurihara, Y., and Watanabe, Y. (2004). Arabidopsis micro-RNA biogenesis through Dicer-like 1
protein functions. Proc Natl Acad Sci U S A 101, 12753-12758.
Lagos-Quintana, M., Rauhut, R., Lendeckel, W., and Tuschl, T. (2001). Identification of novel
genes coding for small expressed RNAs. Science 294, 853-858.
Lamontagne, B., and Elela, S.A. (2004). Evaluation of the RNA determinants for bacterial and
yeast RNase III binding and cleavage. J Biol Chem 279, 2231-2241.
Lamontagne, B., Ghazal, G., Lebars, I., Yoshizawa, S., Fourmy, D., and Elela, S.A. (2003).
Sequence dependence of substrate recognition and cleavage by yeast RNase III. J Mol Biol
327, 985-1000.
Landgraf, P., Rusu, M., Sheridan, R., Sewer, A., Iovino, N., Aravin, A., Pfeffer, S., Rice, A.,
Kamphorst, A.O., Landthaler, M., et al. (2007). A mammalian microRNA expression atlas
based on small RNA library sequencing. Cell 129, 1401-1414.
Landthaler, M., Yalcin, A., and Tuschl, T. (2004). The human DiGeorge syndrome critical
region gene 8 and Its D. melanogaster homolog are required for miRNA biogenesis. Curr
Biol 14, 2162-2167.
Lau, N.C., Lim, L.P., Weinstein, E.G., and Bartel, D.P. (2001). An abundant class of tiny RNAs
with probable regulatory roles in Caenorhabditis elegans. Science 294, 858-862.
Lau, P.W., Guiley, K.Z., De, N., Potter, C.S., Carragher, B., and MacRae, I.J. (2012). The
molecular architecture of human Dicer. Nat Struct Mol Biol 19, 436-440.
Lee, R.C., and Ambros, V. (2001). An extensive class of small RNAs in Caenorhabditis elegans.
Science 294, 862-864.
Lee, R.C., Feinbaum, R.L., and Ambros, V. (1993). The C. elegans heterochronic gene lin-4
encodes small RNAs with antisense complementarity to lin-14. Cell 75, 843-854.
49
Lee, Y., Ahn, C., Han, J., Choi, H., Kim, J., Yim, J., Lee, J., Provost, P., Radmark, O., Kim, S.,
et al. (2003). The nuclear RNase III Drosha initiates microRNA processing. Nature 425, 415419.
Lee, Y., Hur, I., Park, S.Y., Kim, Y.K., Suh, M.R., and Kim, V.N. (2006). The role of PACT in
the RNA silencing pathway. EMBO J 25, 522-532.
Lee, Y., Jeon, K., Lee, J.T., Kim, S., and Kim, V.N. (2002). MicroRNA maturation: stepwise
processing and subcellular localization. EMBO J 21, 4663-4670.
Lee, Y., Kim, M., Han, J., Yeom, K.H., Lee, S., Baek, S.H., and Kim, V.N. (2004a). MicroRNA
genes are transcribed by RNA polymerase II. EMBO J 23, 4051-4060.
Lee, Y.S., Nakahara, K., Pham, J.W., Kim, K., He, Z., Sontheimer, E.J., and Carthew, R.W.
(2004b). Distinct roles for Drosophila Dicer-1 and Dicer-2 in the siRNA/miRNA silencing
pathways. Cell 117, 69-81.
Lehrbach, N.J., Armisen, J., Lightfoot, H.L., Murfitt, K.J., Bugaut, A., Balasubramanian, S., and
Miska, E.A. (2009). LIN-28 and the poly(U) polymerase PUP-2 regulate let-7 microRNA
processing in Caenorhabditis elegans. Nat Struct Mol Biol 16, 1016-1020.
Lim, L.P., Glasner, M.E., Yekta, S., Burge, C.B., and Bartel, D.P. (2003a). Vertebrate
microRNA genes. Science 299, 1540.
Lim, L.P., Lau, N.C., Weinstein, E.G., Abdelhakim, A., Yekta, S., Rhoades, M.W., Burge, C.B.,
and Bartel, D.P. (2003b). The microRNAs of Caenorhabditis elegans. Genes Dev 17, 9911008.
Liu, J., Carmell, M.A., Rivas, F.V., Marsden, C.G., Thomson, J.M., Song, J.J., Hammond, S.M.,
Joshua-Tor, L., and Hannon, G.J. (2004). Argonaute2 is the catalytic engine of mammalian
RNAi. Science 305, 1437-1441.
Liu, Q., Rand, T.A., Kalidas, S., Du, F., Kim, H.E., Smith, D.P., and Wang, X. (2003). R2D2, a
bridge between the initiation and effector steps of the Drosophila RNAi pathway. Science
301, 1921-1925.
Liu, X., Park, J.K., Jiang, F., Liu, Y., McKearin, D., and Liu, Q. (2007). Dicer-1, but not
Loquacious, is critical for assembly of miRNA-induced silencing complexes. RNA 13, 23242329.
Llave, C., Xie, Z., Kasschau, K.D., and Carrington, J.C. (2002). Cleavage of Scarecrow-like
mRNA targets directed by a class of Arabidopsis miRNA. Science 297, 2053-2056.
Lu, J., Getz, G., Miska, E.A., Alvarez-Saavedra, E., Lamb, J., Peck, D., Sweet-Cordero, A.,
Ebert, B.L., Mak, R.H., Ferrando, A.A., et al. (2005). MicroRNA expression profiles classify
human cancers. Nature 435, 834-838.
Lund, E., Guttinger, S., Calado, A., Dahlberg, J.E., and Kutay, U. (2004). Nuclear export of
microRNA precursors. Science 303, 95-98.
Ma, J.B., Ye, K., and Patel, D.J. (2004). Structural basis for overhang-specific small interfering
RNA recognition by the PAZ domain. Nature 429, 318-322.
MacRae, I.J., Zhou, K., and Doudna, J.A. (2007). Structural determinants of RNA recognition
and cleavage by Dicer. Nat Struct Mol Biol 14, 934-940.
Macrae, I.J., Zhou, K., Li, F., Repic, A., Brooks, A.N., Cande, W.Z., Adams, P.D., and Doudna,
J.A. (2006). Structural basis for double-stranded RNA processing by Dicer. Science 311,
195-198.
Mateos, J.L., Bologna, N.G., Chorostecki, U., and Palatnik, J.F. (2010). Identification of
microRNA processing determinants by random mutagenesis of Arabidopsis MIR172a
precursor. Curr Biol 20, 49-54.
50
Mayr, C., Hemann, M.T., and Bartel, D.P. (2007). Disrupting the pairing between let-7 and
Hmga2 enhances oncogenic transformation. Science 315, 1576-1579.
Meister, G., Landthaler, M., Patkaniowska, A., Dorsett, Y., Teng, G., and Tuschl, T. (2004).
Human Argonaute2 mediates RNA cleavage targeted by miRNAs and siRNAs. Mol Cell 15,
185-197.
Mian, I.S. (1997). Comparative sequence analysis of ribonucleases HII, III, II PH and D. Nucleic
Acids Res 25, 3187-3195.
Michlewski, G., and Caceres, J.F. (2010). Antagonistic role of hnRNP A1 and KSRP in the
regulation of let-7a biogenesis. Nat Struct Mol Biol 17, 1011-1018.
Michlewski, G., Guil, S., Semple, C.A., and Caceres, J.F. (2008). Posttranscriptional regulation
of miRNAs harboring conserved terminal loops. Mol Cell 32, 383-393.
Miyoshi, K., Miyoshi, T., and Siomi, H. (2010). Many ways to generate microRNA-like small
RNAs: non-canonical pathways for microRNA production. Mol Genet Genomics 284, 95103.
Miyoshi, K., Okada, T.N., Siomi, H., and Siomi, M.C. (2009). Characterization of the miRNARISC loading complex and miRNA-RISC formed in the Drosophila miRNA pathway. RNA
15, 1282-1291.
Moffat, J., Grueneberg, D.A., Yang, X., Kim, S.Y., Kloepfer, A.M., Hinkle, G., Piqani, B.,
Eisenhaure, T.M., Luo, B., Grenier, J.K., et al. (2006). A lentiviral RNAi library for human
and mouse genes applied to an arrayed viral high-content screen. Cell 124, 1283-1298.
Montgomery, M.K., and Fire, A. (1998). Double-stranded RNA as a mediator in sequencespecific genetic silencing and co-suppression. Trends Genet 14, 255-258.
Murphy, D., Dancis, B., and Brown, J.R. (2008). The evolution of core proteins involved in
microRNA biogenesis. BMC Evol Biol 8, 92.
Nakamura, T., Canaani, E., and Croce, C.M. (2007). Oncogenic All1 fusion proteins target
Drosha-mediated microRNA processing. Proc Natl Acad Sci U S A 104, 10980-10985.
Nam, J.W., Shin, K.R., Han, J., Lee, Y., Kim, V.N., and Zhang, B.T. (2005). Human microRNA
prediction through a probabilistic co-learning model of sequence and structure. Nucleic
Acids Res 33, 3570-3581.
Nam, Y., Chen, C., Gregory, R.I., Chou, J.J., and Sliz, P. (2011). Molecular Basis for Interaction
of let-7 MicroRNAs with Lin28. Cell.
Napoli, C., Lemieux, C., and Jorgensen, R. (1990). Introduction of a Chimeric Chalcone
Synthase Gene into Petunia Results in Reversible Co-Suppression of Homologous Genes in
trans. Plant Cell 2, 279-289.
Newman, M.A., Thomson, J.M., and Hammond, S.M. (2008). Lin-28 interaction with the Let-7
precursor loop mediates regulated microRNA processing. RNA 14, 1539-1549.
Nishikura, K. (2010). Functions and regulation of RNA editing by ADAR deaminases. Annu
Rev Biochem 79, 321-349.
Nykanen, A., Haley, B., and Zamore, P.D. (2001). ATP requirements and small interfering RNA
structure in the RNA interference pathway. Cell 107, 309-321.
O'Connell, R.M., Rao, D.S., Chaudhuri, A.A., Boldin, M.P., Taganov, K.D., Nicoll, J., Paquette,
R.L., and Baltimore, D. (2008). Sustained expression of microRNA-155 in hematopoietic
stem cells causes a myeloproliferative disorder. J Exp Med 205, 585-594.
Okada, C., Yamashita, E., Lee, S.J., Shibata, S., Katahira, J., Nakagawa, A., Yoneda, Y., and
Tsukihara, T. (2009). A high-resolution structure of the pre-microRNA nuclear export
machinery. Science 326, 1275-1279.
51
Okamura, K., Chung, W.J., Ruby, J.G., Guo, H., Bartel, D.P., and Lai, E.C. (2008). The
Drosophila hairpin RNA pathway generates endogenous short interfering RNAs. Nature 453,
803-806.
Okamura, K., Hagen, J.W., Duan, H., Tyler, D.M., and Lai, E.C. (2007). The mirtron pathway
generates microRNA-class regulatory RNAs in Drosophila. Cell 130, 89-100.
Okamura, K., Liu, N., and Lai, E.C. (2009). Distinct mechanisms for microRNA strand selection
by Drosophila Argonautes. Mol Cell 36, 431-444.
Paddison, P.J., Caudy, A.A., Bernstein, E., Hannon, G.J., and Conklin, D.S. (2002). Short hairpin
RNAs (shRNAs) induce sequence-specific silencing in mammalian cells. Genes Dev 16,
948-958.
Park, J.E., Heo, I., Tian, Y., Simanshu, D.K., Chang, H., Jee, D., Patel, D.J., and Kim, V.N.
(2011). Dicer recognizes the 5' end of RNA for efficient and accurate processing. Nature 475,
201-205.
Park, M.Y., Wu, G., Gonzalez-Sulser, A., Vaucheret, H., and Poethig, R.S. (2005). Nuclear
processing and export of microRNAs in Arabidopsis. Proc Natl Acad Sci U S A 102, 36913696.
Park, W., Li, J., Song, R., Messing, J., and Chen, X. (2002). CARPEL FACTORY, a Dicer
homolog, and HEN1, a novel protein, act in microRNA metabolism in Arabidopsis thaliana.
Curr Biol 12, 1484-1495.
Pasquinelli, A.E., Reinhart, B.J., Slack, F., Martindale, M.Q., Kuroda, M.I., Maller, B.,
Hayward, D.C., Ball, E.E., Degnan, B., Muller, P., et al. (2000). Conservation of the
sequence and temporal expression of let-7 heterochronic regulatory RNA. Nature 408, 86-89.
Pertzev, A.V., and Nicholson, A.W. (2006). Characterization of RNA sequence determinants and
antideterminants of processing reactivity for a minimal substrate of Escherichia coli
ribonuclease III. Nucleic Acids Res 34, 3708-3721.
Piskounova, E., Polytarchou, C., Thornton, J.E., LaPierre, R.J., Pothoulakis, C., Hagan, J.P.,
Iliopoulos, D., and Gregory, R.I. (2011). Lin28A and Lin28B inhibit let-7 microRNA
biogenesis by distinct mechanisms. Cell 147, 1066-1079.
Piskounova, E., Viswanathan, S.R., Janas, M., LaPierre, R.J., Daley, G.Q., Sliz, P., and Gregory,
R.I. (2008). Determinants of microRNA processing inhibition by the developmentally
regulated RNA-binding protein Lin28. J Biol Chem 283, 21310-21314.
Portier, C., Dondon, L., Grunberg-Manago, M., and Regnier, P. (1987). The first step in the
functional inactivation of the Escherichia coli polynucleotide phosphorylase messenger is a
ribonuclease III processing at the 5' end. EMBO J 6, 2165-2170.
Regnier, P., and Grunberg-Manago, M. (1989). Cleavage by RNase III in the transcripts of the
met Y-nus-A-infB operon of Escherichia coli releases the tRNA and initiates the decay of the
downstream mRNA. J Mol Biol 210, 293-302.
Regnier, P., and Portier, C. (1986). Initiation, attenuation and RNase III processing of transcripts
from the Escherichia coli operon encoding ribosomal protein S15 and polynucleotide
phosphorylase. J Mol Biol 187, 23-32.
Reinhart, B.J., Slack, F.J., Basson, M., Pasquinelli, A.E., Bettinger, J.C., Rougvie, A.E., Horvitz,
H.R., and Ruvkun, G. (2000). The 21-nucleotide let-7 RNA regulates developmental timing
in Caenorhabditis elegans. Nature 403, 901-906.
Reinhart, B.J., Weinstein, E.G., Rhoades, M.W., Bartel, B., and Bartel, D.P. (2002). MicroRNAs
in plants. Genes Dev 16, 1616-1626.
52
Rivas, F.V., Tolia, N.H., Song, J.J., Aragon, J.P., Liu, J., Hannon, G.J., and Joshua-Tor, L.
(2005). Purified Argonaute2 and an siRNA form recombinant human RISC. Nat Struct Mol
Biol 12, 340-349.
Robertson, H.D. (1982). Escherichia coli ribonuclease III cleavage sites. Cell 30, 669-672.
Robertson, H.D., and Dunn, J.J. (1975). Ribonucleic acid processing activity of Escherichia coli
ribonuclease III. J Biol Chem 250, 3050-3056.
Robertson, H.D., Webster, R.E., and Zinder, N.D. (1967). A nuclease specific for doublestranded RNA. Virology 32, 718-719.
Robertson, H.D., Webster, R.E., and Zinder, N.D. (1968). Purification and properties of
ribonuclease III from Escherichia coli. J Biol Chem 243, 82-91.
Romano, N., and Macino, G. (1992). Quelling: transient inactivation of gene expression in
Neurospora crassa by transformation with homologous sequences. Mol Microbiol 6, 33433353.
Rotondo, G., and Frendewey, D. (1996). Purification and characterization of the Pac1
ribonuclease of Schizosaccharomyces pombe. Nucleic Acids Res 24, 2377-2386.
Ruby, J.G., Jan, C., Player, C., Axtell, M.J., Lee, W., Nusbaum, C., Ge, H., and Bartel, D.P.
(2006). Large-scale sequencing reveals 21U-RNAs and additional microRNAs and
endogenous siRNAs in C. elegans. Cell 127, 1193-1207.
Ruby, J.G., Jan, C.H., and Bartel, D.P. (2007a). Intronic microRNA precursors that bypass
Drosha processing. Nature 448, 83-86.
Ruby, J.G., Stark, A., Johnston, W.K., Kellis, M., Bartel, D.P., and Lai, E.C. (2007b). Evolution,
biogenesis, expression, and target predictions of a substantially expanded set of Drosophila
microRNAs. Genome Res 17, 1850-1864.
Ruvkun, G., Wightman, B., and Ha, I. (2004). The 20 years it took to recognize the importance
of tiny RNAs. Cell 116, S93-96, 92 p following S96.
Saetrom, P., Heale, B.S., Snove, O., Jr., Aagaard, L., Alluin, J., and Rossi, J.J. (2007). Distance
constraints between microRNA target sites dictate efficacy and cooperativity. Nucleic Acids
Res 35, 2333-2342.
Saito, K., Ishizuka, A., Siomi, H., and Siomi, M.C. (2005). Processing of pre-microRNAs by the
Dicer-1-Loquacious complex in Drosophila cells. PLoS Biol 3, e235.
Schauer, S.E., Jacobsen, S.E., Meinke, D.W., and Ray, A. (2002). DICER-LIKE1: blind men and
elephants in Arabidopsis development. Trends Plant Sci 7, 487-491.
Schwarz, D.S., Hutvagner, G., Du, T., Xu, Z., Aronin, N., and Zamore, P.D. (2003). Asymmetry
in the assembly of the RNAi enzyme complex. Cell 115, 199-208.
Seitz, H., Tushir, J.S., and Zamore, P.D. (2011). A 5'-uridine amplifies miRNA/miRNA*
asymmetry in Drosophila by promoting RNA-induced silencing complex formation. Silence
2, 4.
Shi, Y., Wang, Y.F., Jayaraman, L., Yang, H., Massague, J., and Pavletich, N.P. (1998). Crystal
structure of a Smad MH1 domain bound to DNA: insights on DNA binding in TGF-beta
signaling. Cell 94, 585-594.
Shin, C., Nam, J.W., Farh, K.K., Chiang, H.R., Shkumatava, A., and Bartel, D.P. (2010).
Expanding the microRNA targeting code: functional sites with centered pairing. Mol Cell 38,
789-802.
Sohn, S.Y., Bae, W.J., Kim, J.J., Yeom, K.H., Kim, V.N., and Cho, Y. (2007). Crystal structure
of human DGCR8 core. Nat Struct Mol Biol 14, 847-853.
53
Song, L., Axtell, M.J., and Fedoroff, N.V. (2010). RNA secondary structural determinants of
miRNA precursor processing in Arabidopsis. Curr Biol 20, 37-41.
Stark, A., Brennecke, J., Bushati, N., Russell, R.B., and Cohen, S.M. (2005). Animal
MicroRNAs confer robustness to gene expression and have a significant impact on 3'UTR
evolution. Cell 123, 1133-1146.
Stark, A., Kheradpour, P., Parts, L., Brennecke, J., Hodges, E., Hannon, G.J., and Kellis, M.
(2007). Systematic discovery and characterization of fly microRNAs using 12 Drosophila
genomes. Genome Res 17, 1865-1879.
Steiner, F.A., Hoogstrate, S.W., Okihara, K.L., Thijssen, K.L., Ketting, R.F., Plasterk, R.H., and
Sijen, T. (2007). Structural features of small RNA precursors determine Argonaute loading in
Caenorhabditis elegans. Nat Struct Mol Biol 14, 927-933.
Sun, G., Yan, J., Noltner, K., Feng, J., Li, H., Sarkis, D.A., Sommer, S.S., and Rossi, J.J. (2009).
SNPs in human miRNA genes affect biogenesis and function. RNA 15, 1640-1651.
Suzuki, H.I., Yamagata, K., Sugimoto, K., Iwamoto, T., Kato, S., and Miyazono, K. (2009).
Modulation of microRNA processing by p53. Nature 460, 529-533.
Tang, G., Reinhart, B.J., Bartel, D.P., and Zamore, P.D. (2003). A biochemical framework for
RNA silencing in plants. Genes Dev 17, 49-63.
Thai, T.H., Calado, D.P., Casola, S., Ansel, K.M., Xiao, C., Xue, Y., Murphy, A., Frendewey,
D., Valenzuela, D., Kutok, J.L., et al. (2007). Regulation of the germinal center response by
microRNA-155. Science 316, 604-608.
Tomari, Y., Du, T., and Zamore, P.D. (2007). Sorting of Drosophila small silencing RNAs. Cell
130, 299-308.
Trabucchi, M., Briata, P., Filipowicz, W., Ramos, A., Gherzi, R., and Rosenfeld, M.G. (2010).
KSRP promotes the maturation of a group of miRNA precursors. Adv Exp Med Biol 700,
36-42.
Tsutsumi, A., Kawamata, T., Izumi, N., Seitz, H., and Tomari, Y. (2011). Recognition of the premiRNA structure by Drosophila Dicer-1. Nat Struct Mol Biol 18, 1153-1158.
Tuschl, T., Zamore, P.D., Lehmann, R., Bartel, D.P., and Sharp, P.A. (1999). Targeted mRNA
degradation by double-stranded RNA in vitro. Genes Dev 13, 3191-3197.
Ui-Tei, K., Naito, Y., Nishi, K., Juni, A., and Saigo, K. (2008). Thermodynamic stability and
Watson-Crick base pairing in the seed duplex are major determinants of the efficiency of the
siRNA-based off-target effect. Nucleic Acids Res 36, 7100-7109.
van der Krol, A.R., Mur, L.A., Beld, M., Mol, J.N., and Stuitje, A.R. (1990). Flavonoid genes in
petunia: addition of a limited number of gene copies may lead to a suppression of gene
expression. Plant Cell 2, 291-299.
Vermeulen, A., Behlen, L., Reynolds, A., Wolfson, A., Marshall, W.S., Karpilow, J., and
Khvorova, A. (2005). The contributions of dsRNA structure to Dicer specificity and
efficiency. RNA 11, 674-682.
Viswanathan, S.R., Daley, G.Q., and Gregory, R.I. (2008). Selective blockade of microRNA
processing by Lin28. Science 320, 97-100.
Wang, D., Zhang, Z., O'Loughlin, E., Lee, T., Houel, S., O'Carroll, D., Tarakhovsky, A., Ahn,
N.G., and Yi, R. (2012). Quantitative functions of Argonaute proteins in mammalian
development. Genes Dev 26, 693-704.
Wang, Y., Medvid, R., Melton, C., Jaenisch, R., and Blelloch, R. (2007). DGCR8 is essential for
microRNA biogenesis and silencing of embryonic stem cell self-renewal. Nat Genet 39, 380385.
54
Warf, M.B., Johnson, W.E., and Bass, B.L. (2011). Improved annotation of C. elegans
microRNAs by deep sequencing reveals structures associated with processing by Drosha and
Dicer. RNA 17, 563-577.
Weinberg, D.E., Nakanishi, K., Patel, D.J., and Bartel, D.P. (2011). The inside-out mechanism of
Dicers from budding yeasts. Cell 146, 262-276.
Wightman, B., Ha, I., and Ruvkun, G. (1993). Posttranscriptional regulation of the heterochronic
gene lin-14 by lin-4 mediates temporal pattern formation in C. elegans. Cell 75, 855-862.
Wu, H., Henras, A., Chanfreau, G., and Feigon, J. (2004). Structural basis for recognition of the
AGNN tetraloop RNA fold by the double-stranded RNA-binding domain of Rnt1p RNase
III. Proc Natl Acad Sci U S A 101, 8307-8312.
Wu, H., Xu, H., Miraglia, L.J., and Crooke, S.T. (2000). Human RNase III is a 160-kDa protein
involved in preribosomal RNA processing. J Biol Chem 275, 36957-36965.
Wu, H., Yang, P.K., Butcher, S.E., Kang, S., Chanfreau, G., and Feigon, J. (2001). A novel
family of RNA tetraloop structure forms the recognition site for Saccharomyces cerevisiae
RNase III. EMBO J 20, 7240-7249.
Yang, W., Chendrimada, T.P., Wang, Q., Higuchi, M., Seeburg, P.H., Shiekhattar, R., and
Nishikura, K. (2006). Modulation of microRNA processing and expression through RNA
editing by ADAR deaminases. Nat Struct Mol Biol 13, 13-21.
Yekta, S., Shih, I.H., and Bartel, D.P. (2004). MicroRNA-directed cleavage of HOXB8 mRNA.
Science 304, 594-596.
Yeom, K.H., Lee, Y., Han, J., Suh, M.R., and Kim, V.N. (2006). Characterization of
DGCR8/Pasha, the essential cofactor for Drosha in primary miRNA processing. Nucleic
Acids Res 34, 4622-4629.
Yi, R., Qin, Y., Macara, I.G., and Cullen, B.R. (2003). Exportin-5 mediates the nuclear export of
pre-microRNAs and short hairpin RNAs. Genes Dev 17, 3011-3016.
Yoda, M., Kawamata, T., Paroo, Z., Ye, X., Iwasaki, S., Liu, Q., and Tomari, Y. (2010). ATPdependent human RISC assembly pathways. Nat Struct Mol Biol 17, 17-23.
Young, R.A., and Steitz, J.A. (1978). Complementary sequences 1700 nucleotides apart form a
ribonuclease III cleavage site in Escherichia coli ribosomal precursor RNA. Proc Natl Acad
Sci U S A 75, 3593-3597.
Zamore, P.D., Tuschl, T., Sharp, P.A., and Bartel, D.P. (2000). RNAi: double-stranded RNA
directs the ATP-dependent cleavage of mRNA at 21 to 23 nucleotide intervals. Cell 101, 2533.
Zeng, Y., and Cullen, B.R. (2005). Efficient processing of primary microRNA hairpins by
Drosha requires flanking nonstructured RNA sequences. J Biol Chem 280, 27595-27603.
Zeng, Y., Yi, R., and Cullen, B.R. (2005). Recognition and cleavage of primary microRNA
precursors by the nuclear processing enzyme Drosha. EMBO J 24, 138-148.
Zhang, H., Kolb, F.A., Brondani, V., Billy, E., and Filipowicz, W. (2002). Human Dicer
preferentially cleaves dsRNAs at their termini without a requirement for ATP. EMBO J 21,
5875-5885.
Zhang, H., Kolb, F.A., Jaskiewicz, L., Westhof, E., and Filipowicz, W. (2004). Single processing
center models for human Dicer and bacterial RNase III. Cell 118, 57-68.
Zhang, K., and Nicholson, A.W. (1997). Regulation of ribonuclease III processing by doublehelical sequence antideterminants. Proc Natl Acad Sci U S A 94, 13437-13441.
Zhang, X., and Zeng, Y. (2010). The terminal loop region controls microRNA processing by
Drosha and Dicer. Nucleic Acids Res 38, 7689-7697.
55
Zhou, R., Czech, B., Brennecke, J., Sachidanandam, R., Wohlschlegel, J.A., Perrimon, N., and
Hannon, G.J. (2009). Processing of Drosophila endo-siRNAs depends on a specific
Loquacious isoform. RNA 15, 1886-1895.
56
57
58
Chapter 2.
Beyond secondary structure: primary-sequence
determinants license pri-miRNA hairpins for processing
Contents
Summary ........................................................................................................................................61
Introduction ....................................................................................................................................61
Results ............................................................................................................................................63
Existence of auxiliary elements for efficient pri-miRNA processing ................................... 63
Functional substrates from large libraries of pri-miRNA variants. ...................................... 66
Importance of an 11 bp basal stem flanked by ≥ 5 unstructured nucleotides ....................... 70
A basal UG motif enhances processing ................................................................................ 73
The broadly conserved CNNC motif enhances processing .................................................. 75
Loop and apical stem elements can enhance processing ...................................................... 79
Rescue of C. elegans miRNA expression in human cells ..................................................... 81
Discussion ......................................................................................................................................83
Experimental Procedures ...............................................................................................................85
Ectopic pri-miRNA expression in HEK293 cells and S2 cells ............................................. 85
Whole-cell lysate with overexpressed Microprocessor complex.......................................... 86
Competitive binding and cleavage assays............................................................................. 87
Synthesis of pools of pri-miRNA variants ............................................................................ 87
In vitro selection and high-throughput sequencing ............................................................... 88
Sequence analysis ................................................................................................................. 89
Positional enrichments of sequence motifs ........................................................................... 91
59
pri-miRNA collections .......................................................................................................... 91
Accession Numbers .............................................................................................................. 92
Acknowledgements ............................................................................................................... 92
Supplemental Materials .................................................................................................................99
Supplemental Figures............................................................................................................ 99
Supplemental Table S1: Oligonucleotides used in the in vitro selections .......................... 112
Supplemental Table S2. Pri-miRNA collections ............................................................... 116
60
Summary
To use microRNAs to down-regulate mRNA targets, cells must first process these ~22 nt
RNAs from primary transcripts (pri-miRNAs). These transcripts form RNA hairpins important
for processing, but additional unknown determinants must distinguish pri-miRNAs from the
many other hairpin-containing transcripts expressed in each cell. Illustrating the complexity of
this recognition, we show that most Caenorhabditis elegans pri-miRNAs lack determinants
required for processing in human cells. To find these determinants, we generated >1011 variants
of four human pri-miRNAs, sequenced millions that retained function and compared them with
the starting variants. Our results confirmed the importance of pairing in the stem and revealed
three primary-sequence determinants, including a CNNC motif found downstream of most primiRNA hairpins in bilaterian animals but not in nematodes. Adding this and other determinants
to C. elegans pri-miRNAs imparted efficient processing in human cells, further illustrating the
importance of primary-sequence determinants for distinguishing pri-miRNAs from other hairpincontaining transcripts.
Introduction
MicroRNAs (miRNAs) are ~22 nt RNAs that pair to the mRNAs of protein-coding genes
to direct the post-transcriptional repression of these mRNAs (Bartel, 2004, 2009). In animals,
miRNAs are processed from hairpin-containing primary transcripts (pri-miRNAs) that undergo
successive cleavage steps before yielding the functional small RNA. In the canonical pathway,
pri-miRNAs are first cleaved in the nucleus by the Microprocessor, a protein complex containing
an RNase III enzyme Drosha and its cofactor DGCR8 (called Pasha and Psh-1 in Drosophila
melanogaster and C. elegans, respectively) (Lee et al., 2003; Denli et al., 2004; Gregory et al.,
2004; Han et al., 2004; Landthaler et al., 2004). The liberated portion of the hairpin, termed the
pre-miRNA, is exported to the cytosol (Lee et al., 2002; Yi et al., 2003; Bohnsack et al., 2004;
Lund et al., 2004), where it is cleaved by the RNase III enzyme Dicer (Grishok et al., 2001;
Hutvagner et al., 2001; Ketting et al., 2001) approximately two helical turns from the base of the
hairpin to remove the loop (Lee et al., 2004a) and generate two ~22 nt strands that pair to each
other with 2 nt 3′ overhangs (Lim et al., 2003b). One strand of each duplex is loaded into an
Argonaute protein to form the core of the silencing complex, whereas the other strand is
discarded (Khvorova et al., 2003; Schwarz et al., 2003; Liu et al., 2004; Meister et al., 2004).
61
Noncanonical pathways also contribute to the miRNA repertoire of animal cells. For
example, mirtron miRNA precursors are excised from the primary transcript by the spliceosome
rather than Drosha; after debranching, the excised mirtron folds into a hairpin that enters the
canonical pathway at the step of export to the cytoplasm (Okamura et al., 2007; Ruby et al.,
2007a). Tailed mirtrons and endogenous small-hairpin RNAs (shRNAs) are other types of
noncanonical precursors that bypass Microprocessor cleavage (Babiarz et al., 2008), and miR451 is unusual in bypassing Dicer cleavage (Cheloufi et al., 2010; Cifuentes et al., 2010).
A long-standing mystery of canonical miRNA biogenesis has been how the animal cell
determines which of its many hairpin-containing transcripts are recognized by the
Microprocessor to enter the miRNA pathway. Determinants of subsequent Dicer cleavage are
better understood (Zhang et al., 2004; Macrae et al., 2006; Park et al., 2011), as illustrated by
both the success in designing artificial Dicer substrates that bypass Drosha processing
(Brummelkamp et al., 2002; Paddison et al., 2002) and the success in accurately predicting
mirtrons from a single genomic sequence, without considering evolutionary conservation (Chung
et al., 2011). With regard to Microprocessor recognition, sequences within 40 nt upstream and
40 nt downstream of the pre-miRNA hairpin are required for ectopic miRNA expression (Chen
et al., 2004), which is consistent with the observation that these flanking sequences tend to pair
to each other to extend the stem another turn of the helix beyond the site of cleavage (Lim et al.,
2003b). The pairing within this helical extension and a lack of pairing immediately following
the last pair of the basal stem is required for productive Microprocessor recognition, as
illustrated by in vitro studies demonstrating that the human Microprocessor complex can cleave
artificial sequences that form perfectly paired stems flanked by single-strand RNA (ssRNA)
(Han et al., 2006). However, these known structural determinants cannot fully explain the
specificity of Microprocessor cleavage. Many cellular transcripts have paired regions flanked by
ssRNA, and most of these are not endogenous substrates of the Microprocessor.
Indeed,
attempts to predict canonical miRNA hairpins from genomic sequence yield thousands or
millions of false-positive predictions, which must be eliminated using additional criteria, such as
analysis of conservation or experimental evaluation (Lim et al., 2003a; Lim et al., 2003b;
Bentwich et al., 2005; Berezikov et al., 2006; Chiang et al., 2010), illustrating a large gap in our
understanding of how the Microprocessor distinguishes between bona fide pri-miRNA substrates
and other transcribed hairpins.
62
In this study, we found that transcripts that enter the miRNA pathway in C. elegans failed
to do so in human cells. Thus, the definition of a pri-miRNA in one species differs from that in
another, which adds a new dimension to the mystery of pri-miRNA recognition. To elucidate
sequence and structural features of human pri-miRNAs, we generated >1011 variants of four
human pri-miRNA sequences and sequenced millions that were cleaved by the human
Microprocessor complex. Comparison of the cleaved variants with the initial pool of variants
revealed sequence and structural features important for Microprocessor recognition and
cleavage. These features were evolutionarily conserved in non-nematode lineages and sufficient
to increase the processing efficiency of C. elegans hairpins in human cells.
Results
Existence of auxiliary elements for efficient pri-miRNA processing
To examine whether miRNA processing features are shared across animals, we
ectopically expressed C. elegans, D. melanogaster and human pri-miRNAs in human cells and
compared the yields of mature miRNA. For each miRNA investigated, the hairpin and ~100 nt
of flanking genomic sequence were expressed upstream of the human mir-1-1 pri-miRNA on a
bicistronic transcript under control of the CMV promoter (Figure S1A). Cells transfected with
each vector were harvested and pooled, and small RNAs were sequenced.
As reported
previously, most human miRNAs were efficiently expressed (Chiang et al., 2010), as were four
of nine tested Drosophila miRNAs (Figure 1A). However, the tested C. elegans miRNAs were
less efficiently expressed in HEK293 cells (Figure 1A, p = 1.4×10–5, Wilcoxon rank-sum test).
Likewise, in Drosophila S2 cells, C. elegans miRNAs were expressed less efficiently than were
human miRNAs (p = 0.024). These results indicated that most nematode pri-miRNAs are
missing determinants required for proper processing by human or insect cells.
To isolate the processing defect, we probed for processing intermediates. For each
inefficiently expressed miRNA examined, the primary transcript was present (Figure S1A and
S1B), but no pre-miRNA or mature miRNA was detected (Figure 1B). These results suggested
that C. elegans pri-miRNAs are not productively recognized as substrates of the Microprocessor
in the first step of miRNA maturation.
To assay directly for binding to the human
Microprocessor
63
–
+
–
+
–
+
–
+
D
P22
P1
–1
–51
–41
1.00
0.09
1.00
–31 –21 –11
5p position
0.08
1.00
+1
–13 +11
–1
0.09
1.00
0.8
0.7
0.6
0.5
0.33
P1
cel-mir-235
cel-mir-60
cel-mir-59
cel-mir-50
cel-mir-46
600
400
200
0
0
Fly
Worm
C
Nitrocellulose
filtration
Reference
Basal stem
1
0.9
P11 P21 P31 P41 P51
pre-miRNA position
0.5
1
11
hsa-mir-1-1
hsa-mir-128-1
hsa-mir-205
cel-lin-4
cel-lsy-6
cel-mir-40
cel-mir-50
cel-mir-230
cel-mir-240
293 cells
21
31
41
3p position
cel-mir-50
+
0.07
1151
6752
2777
4919
10362
3037
3271
5024
3705
4171
499
1000
cel-mir-40
–
1.00
800
cel-lsy-6
+
cel-mir-44
Human
cel-lin-4
hsa-mir-1-1
–
cel-lsy-6
Microprocessor
Average BLS
+
0.13
Query miRNA:
hsa-mir-122
–
1.00
Query
+
0.21
Reference
–
1.00
DroshaTN
DGCR8
1.65
Query
1.00
hsa-mir-1-1
hsa-mir-17
hsa-mir-18a
hsa-mir-19a
hsa-mir-20a
hsa-mir-19b-1
hsa-mir-92a-1
hsa-mir-122
hsa-mir-125a
hsa-mir-128-1
hsa-mir-133a-1
hsa-mir-138-2
hsa-mir-142
hsa-mir-205
dme-mir-2a-1
dme-mir-4
dme-mir-5
dme-mir-34
dme-mir-92a
dme-mir-125
dme-mir-286
dme-mir-279
dme-mir-281-1
cel-mir-2
cel-lin-4
cel-lsy-6
cel-mir-34
cel-mir-40
cel-mir-43
cel-mir-44
cel-mir-46
cel-mir-50
cel-mir-59
cel-mir-60
cel-mir-124
cel-mir-235
cel-mir-240
Normalized hairpin reads
A
30
25
20
S2 cells
15
10
5
Human
1
Basal stem
0.9
0.8
0.7
0.6
51
Worm
B
pre-miRNA
70
mature
miRNA
20
15 fmol
cognate
control
750
300
Microprocessor, we established a competitive-binding assay that compared the ability of
different pri-miRNAs to bind catalytically-deficient Drosha and DGCR8 (Figure 1C). Whereas
human pri-mir-122 bound the Microprocessor somewhat better than did human pri-mir-125a, all
seven tested C. elegans pri-miRNAs bound worse (Figure 1C). Thus, most C. elegans primiRNAs are missing some of the determinants needed for efficient recognition and processing
by the Microprocessor complex. As a result, many transcripts recognized as pri-miRNAs in C.
elegans cells are not recognized as pri-miRNAs by human cells.
Known features of C. elegans and human pri-miRNAs appear largely similar, as
illustrated by the accuracy of an algorithm trained on C. elegans pri-miRNA features in
predicting most miRNA genes conserved in human, mouse and fish (Lim et al., 2003a).
Nonetheless, the poor specificity of this algorithm when predicting non-conserved miRNAs
supports the idea that unknown features also exist and help the cell define authentic pri-miRNAs.
To look for clues regarding previously unknown features that might be required for human primiRNA recognition, we analyzed the sequence immediately flanking human pre-miRNAs for
conservation in other vertebrates. In a meta-analysis of human pri-miRNAs conserved in other
Figure 1. The existence of auxiliary elements that specify human pri-miRNA transcripts.
(A) Processing of human, fly, and worm pri-miRNAs in human HEK293T cells and Drosophila S2 cells.
Cells were transfected with plasmids expressing the indicated pri-miRNA hairpins with ~100 flanking
genomic nucleotides on each side of each hairpin, and total RNA was pooled for small-RNA
sequencing. Graphs plot the small RNA reads derived from the indicated pri-miRNAs.
(B) Attempted detection of pre-miRNA and mature miRNA production in HEK293T cells. RNA blots of
total RNA isolated from cells transfected with the indicated pri-miRNA were probed for the cognate
miRNA. Blots also included lanes with 15 fmol in vitro transcribed standards derived from the
corresponding pri-miRNAs (pri-RNA controls).
(C) Relative binding of C. elegans and human pri-miRNAs to the Microprocessor. In the competitive
binding assay (left, schematic), radiolabeled query pri-miRNA was mixed with the radiolabeled
shorter reference pri-miRNA (human mir-125a) and incubated with immunopurified, catalytically
impaired Drosha (Drosha-TN) and DGCR8 in conditions of RNA excess. Bound RNA was isolated
on nitrocellulose filters and eluted for analysis on a denaturing gel. Phosphorimaging (right)
indicated the relative amounts of query and reference RNA in the input (–) and bound to the
Microprocessor (+). Numbers below each lane indicate the ratio of bound query miRNA relative to
bound reference RNA, normalized to the input ratio.
(D) Nucleotide conservation of human pri-miRNAs conserved to mouse. At each position, the average
branch-length score (BLS) is plotted, in which each BLS indicates the phylogenetic branch lengths of
all the aligned mammalian species in which the ancestral identity was preserved, divided by the
branch lengths of all the species in which the miRNA is preserved. Positions are numbered based
on the inferred Drosha cleavage site (inset); negative indices are upstream of the 5p Drosha
cleavage site, indices with “P” count from the 5′ end of the pre-miRNA, and positive indices are
downstream of the 3p Drosha cleavage site.
65
mammals, residues extending 13 nt upstream of the 5p Drosha cleavage site (i.e., the site
corresponding to the 5′ end of the pre-miRNA) and 11 nt downstream of the 3p Drosha cleavage
site were conserved above background, consistent with the importance of the ~11 bp basal stem
for pri-miRNA processing (Figure 1D). Upstream of the hairpin, conservation dropped rapidly
with distance from the cleavage site, with just a few nucleotides immediately flanking the basal
stem conserved above background. Conservation also dropped on the 3p side of the pre-miRNA,
but not quite as rapidly in the region 15–25 nt from the cleavage site (Figure 1D). This
asymmetry in the conservation drop-off hinted at potential determinants downstream of the
hairpin.
However, the overall weakness of the conservation signal beyond the basal stem
suggested that any determinants in these flanking regions might either be at variable distances
from the hairpin or present in some subsets of miRNAs but not others, making them difficult to
identify using only comparative sequence analyses.
Functional substrates from large libraries of pri-miRNA variants.
To identify sequence and structural features important for Microprocessor recognition
and cleavage, we generated >1011 variants of a pri-miRNA, sequenced millions that retained
function and compared these sequences to those of the initial pool of variants (Figure 2A). At
each variable nucleotide position, most molecules had the wild-type residue, and a minority had
the other three alternatives. This approach resembled classical in vitro selection approaches,
particularly those that started with degenerate libraries with the goal of characterizing nucleic
acids known to function as ligands and substrates (Ellington and Szostak, 1990; Bartel et al.,
1991; Breaker et al., 1994), except we did not perform multiple rounds of selection. Instead, we
collected the variants that were cleaved by the Microprocessor and then directly prepared them
for high-throughput sequencing.
Similar strategies have been used for DNA-binding and
ribozyme experiments (Zykovich et al., 2009; Jolma et al., 2010; Pitt and Ferre-D'Amare, 2010;
Slattery et al., 2011). Because the differences from the starting pool were from a single round of
cleavage, and because both the starting pool and the selected pool were subject to the same
number of transcription, reverse-transcription and amplification steps, any differences observed
between the two pools were subject to neither the compounding effects of multiple rounds nor
the confounding effects of amplification biases. Moreover, because in each sample millions of
molecules were sequenced, the differences were not influenced by stochastic sampling of small
66
numbers. Thus, compared to the results of classical approaches, enrichment or depletion of a
residue was a more direct reflection of its contribution to biochemical specificity.
To query sequence and structural determinants at the base of the hairpin and flanking the
hairpin, pools of variants were constructed in which residues >8 nt upstream of the 5p Drosha
cleavage site or >8 nt downstream of the 3p cleavage site were varied while the remaining
hairpin residues were not varied. Suspecting that different pri-miRNAs might use different
determinants, four different pools were constructed, based on the pri-miRNAs of human mir125a, mir-16-1, mir-30a, and mir-223, respectively.
Each pool was produced by in vitro
transcription of a DNA template constructed using degenerate oligonucleotides in which variable
positions had non-wild-type residues introduced at a frequency of 21%. For example, at a
variable position in which the wild-type residue was an A, 79% of the pool molecules would
have an A, whereas 7% would have a C, 7% would have a G, and 7% would have a U. The other
key design element was that, borrowing from a strategy used to identify variants of RNAcleaving ribozymes (Pan and Uhlenbeck, 1992), each variant was circularized (Figure 2A).
Without circularization, some variable nucleotides would have resided in the upstream cleavage
product, whereas others would have resided in a separate downstream product, making it
impossible to reconstruct the starting variant from the sequenced products. With circularization,
all the variable nucleotides resided in a single product, the sequence of which revealed the
starting variant, thereby enabling a full analysis of sequence interdependencies and covariation.
In vitro cleavage of circularized pri-miRNA variants was carried out in whole-cell lysate
of HEK293T cells overexpressing Drosha and DGCR8.
Very little pri-miRNA cleavage
occurred in lysate from cells not overexpressing the proteins, which indicated that the dominant
pri-miRNA cleavage activity depended on overexpressed Drosha or DGCR8 (Figure 2B). At a
time in which the lysate cleaved linear and circularized pri-mir-125a nearly to completion, much
of the pool of pri-mir-125a variants remained uncleaved, which indicated that substitutions in the
basal stem and flanking regions can attenuate Microprocessor cleavage in vitro (Figure 2C).
Analogous results were obtained with pools of variants based on the other three pri-miRNAs
(data not shown).
Variants that were cleaved by the Microprocessor were recovered by gel purification,
ligated to sequencing adaptors, and prepared for high-throughput, paired-end sequencing (Figure
2A). Sequence analyses were restricted to products cleaved at the wild-type processing sites,
67
–0.4
5p Position
25
27
29
31
33
35
37
39
41
43
45
25
27
29
31
33
35
37
39
41
43
45
25
27
29
31
33
3p Position
45
C
U
47
43
A
G
45
41
+1
43
39
37
Stem-loop
41
39
0.4
35
–9
33
–1
31
0.6
29
0.8
27
25
23
21
19
17
15
13
293T Transfection
37
13
9
11
Drosha
DGCR8
B
35
23
23
23
9
11
21
0
21
–0.2
21
0.2
19
0.4
19
0.6
19
0.8
17
hsa-mir-223
17
0
17
–0.2
15
0.2
15
0.4
13
0.6
15
0.8
13
hsa-mir-30a
9
0
11
0.2
9
hsa-mir-125a
11
Drosha
DGCR8
Mock
Circular pri-miRNA substrate
(pool of variants)
–9
–11
–13
–15
–17
–19
–21
–23
–25
–27
Nonfunctional variants
–29
1
–31
Information content (bits)
Drosha
DGCR8
–33
1.2
–35
–0.4
–37
1
–39
1.2
–41
–0.4
–43
1
–45
1.2
–47
–51
–49
–47
–45
–43
–41
–39
–37
–35
–33
–31
–29
–27
–25
–23
–21
–19
–17
–15
–13
–11
–9
–0.4
–49
–47
–45
–43
–41
–39
–37
–35
–33
–31
–29
–27
–25
–23
–21
–19
–17
–15
–13
–11
–9
Information content (bits)
1
–49
–47
–45
–43
–41
–39
–37
–35
–33
–31
–29
–27
–25
–23
–21
–19
–17
–15
–13
–11
–9
Information content (bits)
1.2
–49
Information content (bits)
A
C
pri-mir-125a WT WT Pool
Topology Linear Circ. Circ.
Functional variants
Splint-ligated product
Library for
paired-end sequencing
D
+9
Invariant
residues
–14 +12
–0.2
hsa-mir-16-1
0.8
0.6
0.4
0.2
–0.2
0
which were inferred from the dominant reads in small-RNA sequencing data (Landgraf et al.,
2007; Bar et al., 2008; Chiang et al., 2010; Witten et al., 2010), except for miR-16-1* and miR223, which appear to undergo post-cleavage 3′-end trimming (Han et al., 2011).
Because
product ligation and computational analysis both selected for cleavage at the wild-type site,
nucleotide changes that altered the site of Microprocessor cleavage were not distinguished from
those that abolished cleavage.
At each variant position, we compared the odds of each nucleotide in the properly
cleaved pool to the odds of that nucleotide in the starting pool. These odds ratios were used to
calculate the information content of each nucleotide possibility at each variant position—the
greater the information content, the more favorable the influence on Microprocessor recognition
and cleavage, with positive values indicating a favorable influence and negative values
indicating a disruptive influence. Information content was chosen as the metric for displaying
enrichment or depletion in the cleaved pool because it effectively indicated the relative influence
of the nucleotide, regardless of whether it was the wild-type possibility or one of the other three
possibilities.
Some positions had substantial enrichment of one or more nucleotide possibilities, with
corresponding depletion of the others (Figure 2D). To validate the influence of representative
positions on recognition and cleavage, pri-mir-125a mutants were tested in the competitivecleavage assay, comparing cleavage to that of wild-type pri-mir-125a (Figure S2A and B). The
Figure 2. In vitro selection for functional pri-miRNA variants.
(A) Schematic of the selection in which variable sequences (red) flanked the Drosha cleavage site. PrimiRNA variants were circularized by ligation and incubated in whole-cell lysate from HEK293T cells
overexpressing Drosha and DGCR8. Cleaved variants were gel-purified, ligated to adaptors, reverse
transcribed and amplified for high-throughput sequencing.
(B) Cleavage of linear hsa-let-7a in whole-cell lysate from HEK293T cells (mock) and whole-cell lysate
from HEK293T cells transfected with plasmids expressing Drosha and DGCR8. Incubations were for
90 minutes. Body-labeled reactants and products were resolved on a denaturing polyacrylamide gel
and visualized by phosphorimaging.
(C) Cleavage of linear and circular pri-mir-125a (WT linear and WT circ., respectively) and a pool of
circular hsa-mir-125a variants (pool). RNAs were incubated for 5 minutes in the extracts
supplemented with Drosha and DGCR8 and analyzed as in (B). The WT linear RNA was 5′ endlabeled; the other RNAs were body-labeled.
(D) Enrichment and depletion at variable residues in functional pri-miRNA variants. At each varied
position (inset, red inner line), information content was calculated for each residue (green, cyan,
black, and red for A, C, G, and U, respectively).
69
results of changing specific residues closely matched those predicted from analysis of sequenced
variants, thereby confirming that the calculated relative cleavage faithfully reflected the
influence on Microprocessor recognition and cleavage in vitro. The effects of these changes were
also confirmed in HEK293T cells (Figure S2C).
Importance of an 11 bp basal stem flanked by ≥ 5 unstructured
nucleotides
For all four miRNAs, some of the variable residues with the greatest influence fell within
the basal stem (Figure 2D). The high information content at these paired positions could be due
to either the importance of primary sequence or the need to pair to the nucleotide on the other
arm of the hairpin, or both. To distinguish between these possibilities, we examined covariation
matrices, generated by calculating the odds of each pair of nucleotide identities at each predicted
base pair, relative to the odds of those identities in the initial pool. These matrices showed
overall preference for Watson–Crick geometry at each of these basal pairs, with the G:U wobble
being the most frequently preferred non-Watson–Crick alternative (Figure 3A, S3A).
For
example, the most favored alternatives to the C:G pair at positions –11 and +9 of mir-125a are
the G:C and U:A pairs, and to a lesser extent the A:U, G:U and U:G pairs (Figure 3A). In fact,
Watson–Crick pairing was strongly preferred even if it did not occur in the wild-type sequence.
For example, the wild-type A:C pair at positions –12 and +10 of mir-30a was disfavored,
whereas the four Watson–Crick pairs were most strongly favored (Figure 3A). Similarly, the
bulged A at position +10 of mir-223 was preferentially incorporated into an alternative
continuous helix (Fig. S3A–B).
Layered on top of the overall preference for Watson–Crick pairing were primarysequence preferences specific to each basal pair. For example, at positions –11 and +9 the C:G
pair was strongly favored over the A:U alternative. The primary-sequence preference was most
acute at position –13. At this position the preference for a G was often stronger than the
preference for Watson–Crick pairing, in that the G:U wobble and sometimes the G:A or G:G
mismatches were less disruptive than were the other three Watson–Crick alternatives (Figure
3A).
We conclude that primary-sequence features supplement and sometimes supersede
structural features important for basal-stem recognition.
70
Using the same covariation analysis, we screened for evidence of Watson–Crick pairing
between all possible pairs of varied positions. For each of these >3000 possible pairs, the degree
of Watson–Crick preference was evaluated using a scoring metric that compared the average
odds of Watson–Crick pairs to that of non-Watson–Crick alternatives. In each case, the highestscoring pairs were those of the basal stem (Fig. S3C). In the case of mir-223, the highest scoring
pairs also included the alternative pairs that incorporated the bulged A at +10 into a contiguous
helix. For each pri-miRNA, we inspected the next four highest-scoring pairs, and in each case,
the covariation matrix did not appear consistent with Watson–Crick pairing (data not shown).
These results indicate that in the sequence flanking the pre-miRNA, Watson–Crick pairing
important for Microprocessor recognition and cleavage is restricted to the basal stem.
The Microprocessor recognizes the junction between the miRNA hairpin and flanking
ssRNA, and thereby positions the active site to cleave approximately one helical turn (11 bp of
A-form RNA) from the base of the duplex (Han et al., 2006; Yeom et al., 2006). To examine
whether a specific number of pairs was preferred in the basal stem, we calculated the relative
cleavage of different stem-length variants, normalizing to that of an 8 bp stem. Invariant
mismatches within symmetric internal loops (e.g., the A:C mismatch at positions –6 and +4 of
mir-30a) were assumed to be non-canonical pairs that stacked within the stem to contribute to its
length, whereas mismatches at varied positions were assumed to disrupt further pairing and
thereby terminate the inferred basal stem. For all four pri-miRNAs, an 11 bp basal stem was
optimal (Figure 3B), consistent with the single-turn measurement for cleavage-site selection.
Indeed, an 11 bp basal stem was preferred for mir-223, even though the wild-type sequence was
predicted to form a 12 bp stem (Figures 3A and S3A). For most pri-miRNAs, however, the
efficiency of the 12-pair stem approached that of the 11-pair stem (Figure 3B). This tolerance of
a twelfth pair hinted at the influence of other features, such as the G at position –13, in
overriding the single-turn measurement to specify the precise site of cleavage.
The model for single-turn measurement also posits that the nucleotides immediately
flanking the basal stem are unstructured (Han et al., 2006; Yeom et al., 2006). To test this part of
the model, we used RNAfold (Hofacker and Stadler, 2006) to predict the minimum free-energy
structure of all sequenced miRNA variants in the selected pools and the initial pools. For each
sequence with wild-type predicted pairing in the stem, the number of nucleotides between the
base of the stem and the most proximal two consecutive structured bases was recorded.
71
−0.2
−0.4
−0.6
−0.8
−1
Relative cleavage
Timepoint 2
U
0.34 –1.02 –0.04
Position 10
C
G
U
A
–1.48 –1.18 –1.20 –0.23
C
–1.11 –1.60
G
–0.54
U
–0.15 –1.50 –0.40 –0.76
1.17 –1.14
0.17 –0.41 –0.16
Position 11
C
G
U
A
–0.50 –0.93 –0.58 –0.22
C
–0.98 –1.61 –0.40 –1.02
G
–0.35
U
–1 U
C
C-G +1
G-U
U-A
G
A
A
G
C-G
G-U
A-U
–11 C-G +9
–12 U-A +10
–13 G-C +11
U
C
A
A
A
U
C
A
0.36 –0.85 –0.30 –0.48
0.81 –0.27
hsa-mir-16-1
Wildtype basal stem
0.04
–0.04 –0.74 –0.28 –0.30
hsa-mir-125a
A
–4.17 –3.5 –3.45 –0.39
C
–3.29 –3.72
2.5
–2.45
G
–2.85 0.41
–2.6
0.14
U
0.09
A
–2.04 –0.67 –1.06 0.35
C
–1.25 –1.83 1.03
–0.95
G
–0.22 0.7
0.39
U
0.74
0.15
–1.44 –0.03 –0.04
Position 11
C
G
U
Pair 11 A
A
–1.12 –2.05 –1.23 –0.93
C
–1.51 –3.12 –0.69 –1.65
G
0.32
U
–0.57 –1.7 –0.71 –1.22
0.74
hsa-mir-30a
Wildtype basal stem
–0.16 0.37
hsa-mir-30a
64
32
32
32
16
16
16
16
8
8
8
8
4
4
4
4
2
2
2
2
1
1
1
9 10 11 12 13
Basal stem pairs
8
9 10 11 12 13
Basal stem pairs
8
Position 9
C
G
U
A
–3.34 –3.01 –3.07 –0.98
C
–2.18 –2.93 2.64
G
–0.98 0.13
G
–1.60
U
–1.66 –3.63 –1.98 –1.88
–2.18
–1.29 –0.08
1
9 10 11 12 13
Basal stem pairs
2.01 –0.51 –0.21
hsa-mir-223
64
32
8
A
–1 C
U –0.27 –2.72 –1.74 –1.82
A
G-U +1
C-G
Position 10
G-C
Pair 10 A
C
G
U
A
C
A –0.92 –0.65 –0.22 1.63
G-U
U-A
C 0.18 –1.45 2.39 –0.14
G-C
G 0.00 1.57 0.58 0.99
A-U
U 1.63 –1.00 1.38 0.48
–11 C-G +9
C +10
–12 A
–13 G-C +11
U
U
Position 11
Pair 11 A
U
C
C
G
U
G
G
A –1.80 –2.40 –1.15 –0.96
U
G
C –2.37 –2.92 –0.95 –2.17
–2.16 –0.66 –1.29
Position 10
C
G
U
Pair 10 A
hsa-mir-16-1
64
Position –11
–0.90
1.14 –0.90
Position –11
G
Pair 11 A
64
Timepoint 1
–0.95 –1.27
Pair 10 A
hsa-mir-125a
Wildtype basal stem
B
–1.37 –1.22 –1.35 –0.03
Pair 9
pre-miRNA
Position –12
0
A
C
Position 9
C
G
U
A
Position –13
0.2
–1 G
G
A-U +1
U-G
C-G
U
C
C
G
U
U
G-C
A-U
–11 C-G +9
–12 C-G +10
–13 G-C +11
U
C
U
C
G
C
U
A
Pair 9
pre-miRNA
Position –12
0.4
Odds ratio (log2)
0.6
Position –11
0.8
Position –12
1
Position 9
C
G
U
A
Position –13
Pair 9
pre-miRNA
Position –13
A
8
9 10 11 12 13
Basal stem pairs
C
1
hsa-mir-125a
0.8
0.2
0
−0.2
−0.4
3p unstructured nucleotides
0.4
Odds ratio (log2)
0.6
−0.6
−0.8
5
10
10
8
8
6
6
5
4
0
0
0
2
4
6
8
10
12
5
10
5p unstructured nucleotides
0
10
6
5
4
0
10
8
5
4
2
2
0
12
10
10
10
hsa-mir-223
hsa-mir-30a
hsa-mir-16-1
12
12
2
0
0
2
4
6
8
10
12
5
10
5p unstructured nucleotides
0
0
0
0
2
4
6
8
10
12
5
10
5p unstructured nucleotides
0
0
5
10
5p unstructured nucleotides
−1
Timepoint 1
Timepoint 2
16
Relative cleavage
D
hsa-mir-125a
16
hsa-mir-16-1
16
hsa-mir-30a
16
8
8
8
8
4
4
4
4
2
2
2
2
1
0 2 4 6 8 10 12 14 16 18 20
Flanking unstructured nucleotides
1
0 2 4 6 8 10 12 14 16 18 20
Flanking unstructured nucleotides
1
0 2 4 6 8 10 12 14 16 18 20
Flanking unstructured nucleotides
1
hsa-mir-223
0 2 4 6 8 10 12 14 16 18 20
Flanking unstructured nucleotides
Although at best this metric was a rough estimate of the size of the unstructured region
flanking the base of the helix, we observed a clear correlation between the number of flanking
unstructured nucleotides and enrichment in the selection (Figure 3C). Pairing was tolerated in
one flank, provided that the other flank contained at least 5–7 unstructured bases, a result
consistent with the observation that pri-miRNAs are cleaved with partial efficiency when only
one flanking segment is present (Zeng and Cullen, 2005; Han et al., 2006). When summing the
flanking unstructured bases from both sides, the optimum plateaued at ~9–18 nt, depending on
the pre-miRNA (Figure 3D).
A basal UG motif enhances processing
Among the nucleotides upstream of the stem-loop, the most striking enrichment was for a
U at position –14 (Figure 2D). This U immediately preceded the position that, as mentioned
above, displayed a strong primary-sequence preference for a G, either when paired with a C at
position +11 to form the most basal Watson–Crick pair of the helix or when partnered with a
wobble or mismatch. The U and G at positions –14 and –13 both appeared to contribute
independently to recognition; variants with either a U or a G were enriched over variants with
neither, and variants containing both were even more enriched (Figure 4A). For mir-223, the UG
at positions –14 and –13 was preferred (Figure 2D), even though wild-type mir-223 has a UG at
Figure 3. Basal stem secondary structure in functional pri-miRNA variants.
(A) Predicted basal secondary structure and covariation matrices for mir-125a, mir-16-1, and mir-30a.
For each pair of positions, joint nucleotide distributions were tabulated from sequencing data and the
odds ratio calculated. Favored pairs have positive odds ratios and are colored red, whereas
disfavored pairs have negative odds ratios and are colored blue, with color intensity indicating
magnitudes, according to the key (left).
(B) Relative cleavage of variants with different stem lengths. The number of contiguous Watson–Crick
pairs was counted, and the relative cleavage calculated, normalized to the 8 bp stem.
(C) Enrichment for unstructured nucleotides flanking the basal stem. Predicted folds of variant
sequences were generated, and the subset of sequences with wild-type basal stem pairing were
classified based on the distance to the nearest structured consecutive nucleotides upstream of
position –13 and the nearest structured consecutive nucleotides downstream of position +11.
Enrichment (red) and depletion (blue) of different unstructured lengths for the selected variants are
colored according the key (left). Black indicates that the sequencing data were insufficient to
calculate enrichment values.
(D) Relative cleavage of variants with numbers of total unstructured nucleotides flanking the basal stem.
Unstructured lengths upstream and downstream calculated in (C) were summed, and the relative
cleavage calculated, normalized to zero unstructured nucleotides.
73
4
4
2
2
2
2
1
1
1
1
0.5
0.5
0.5
0.5
0
4
–15
–4
–13
–11
5p Position
–17
0.6
0
T G T T G A C A
–4
–15
–13
–11
5p Position
C
hsa-mir-223
chrX
65,238,719–65,238,726
(+)
4
0
C A A T G T C A
A
–17
C
–15
–13
–11
5p Position
G
C C T G C A G T
–4
–17
–15
–13
–11
5p Position
U
0.5
0.4
Frequency
–17
G(–13) only
hsa-mir-30a
chr6
72,113,329–72,113,336
(–)
0
T G T T G C C A
–4
No (–14) motif
hsa-mir-16-1
chr13
50,623,097–50,623,104
(–)
4
0.3
0.2
0.1
0.0
–19 –17 –15 –13 –11
–9
Position
D
Human UG Position
24%
miRNAs with positioned UG
PhyloP
vertebrate
conservation
4
UG(–14)
UG(–14)
G(–13) only
U(–14) only
hsa-mir-125a
chr19
52,196,503–52,196,510
(+)
16%
8%
0%
H. sapiens
*
D. rerio
*
C. intestinalis *
A. gambiae
D. melanogaster
D. pulex
C. elegans
C. briggsae
P. pacificus
C. teleta
*
L. gigantea
S. mediterranea
N. vectensis
Chordates
Arthropods
Nematodes
Ecdysozoans
B
UG(–14)
4
U(–14) only
4
G(–13) only
8
U(–14) only
8
Timepoint 2
No (–14) motif
8
Timepoint 1
8
UG(–14)
hsa-mir-223
16
G(–13) only
hsa-mir-30a
16
No (–14) motif
Relative cleavage
hsa-mir-16-1
16
U(–14) only
hsa-mir-125a
16
No (–14) motif
A
Lophotrochozoans
–22 –20 –18 –16 –14 –12 –10 –8
Position
–6
–4 –2
Drosha cleavage
site
positions –15 and –14, respectively. The –14 preference was also observed among variants of
mir-125a selected for Microprocessor binding rather than for Microprocessor cleavage (Figure
S4B), which indicated that this preference was due at least in part to increased binding affinity to
the Microprocessor. We refer to this dinucleotide motif as the basal UG.
The basal UG was conserved in vertebrate orthologs of mir-16-1 and mir-30a (Figure
4B).
The motif was also enriched in other mammalian pri-miRNAs, as illustrated by the
sequence composition of human pri-miRNAs conserved to mouse, which show clear preferences
for U at position –14 and G at position –13 (Figure 4C). Enrichment was also observed in primiRNAs of zebrafish (D. rerio) and sea squirt (C. intestinalis) but only sporadically in more
distantly related lineages, suggesting that recognition of the UG motif emerged in a common
ancestor of the chordates (Figure 4C).
The broadly conserved CNNC motif enhances processing
Examination of nucleotides preferred in the 3′ flanking sequence revealed a strong
preference for a pair of C residues, separated by two intervening nucleotides, located 17–18 nt
downstream of the Drosha cleavage site in mir-16-1, mir-30a, and mir-223 (Figure 2D). The two
C residues of this CNNC motif (in which N signifies any nucleotide) seemed to act
synergistically, in that variants that retained neither C residue were not disfavored much more
than those that retained one (Figure 5A). As expected, the C residues enriched in the active
variants were also conserved in vertebrate orthologs of these three pri-miRNAs (Figure 5B).
Figure 4. Enrichment and conservation of the basal UG motif.
(A) Relative cleavage of variants with a full UG motif, a partial motif, and no motif. Relative cleavage
values were normalized to that of variants with no motif.
(B) PhyloP conservation across 30 vertebrate species in the region of the basal UG motif (red letters) for
the four selected miRNAs. Bars extending beyond the scale of the graph are truncated (red).
Nucleotides predicted to be paired in the wildtype basal stem are shaded.
(C) Frequencies of A, C, G, and U (green, cyan, black, and red, respectively) at the indicated positions of
human pri-miRNAs conserved to mouse. Analysis was of 202 pri-miRNAs, each representing a
unique miRNA paralog family (Table S2).
(D) Enrichment for the UG dinucleotide in the pri-miRNAs of representative animals with sequenced
genomes (Table S2). For each species, pri-miRNA sequences were aligned according to the
predicted Drosha cleavage site, and upstream UG occurrences tabulated. Species with a statistically
–5
significant enrichment at position –14 are indicated (asterisks, empirical p-value < 10 ).
75
2
1
1
1
1
0.5
0.5
0.5
0.5
C
Signal / background
0
15
17
19
21
3p Position
-4
23
Human miRNAs
(conserved to mouse)
NNNNN
NNNN
NNN
NN
15
17
19
21
3p Position
6
5
5
4
3
2
1
4
7 10 13 16 19 22 25 28
Position
-4
23
4
1
0
A C T C T A C A G
D. melanogaster miRNAs
CNNC
2
0
4
4
A C C A C A C A C
15
17
19
21
3p Position
-4
23
C. elegans miRNAs
7 10 13 16 19 22 25 28
Position
15
17
19
21
3p Position
23
S. mediterranea miRNAs
12
10
6
0
T A C C A G C T C
14
8
0
D
0
G A C T T C A A G
2
1
4
4
3
1
1
chrX
65,238,816–65,238,824
(+)
4
2
1
4
7 10 13 16 19 22 25 28
Position
0
1
4
7 10 13 16 19 22 25 28
Position
Human CNNC window
H. sapiens
D. rerio
C. intestinalis
A. gambiae
D. melanogaster
D. pulex
C. elegans
C. briggsae
P. pacificus
C. teleta
L. gigantea
S. mediterranea
N. vectensis
*
*
*
*
*
*
Chordates
Arthropods
Nematodes
*
*
*
1
3
Drosha cleavage
site
Lophotrochozoans
5
7
9
11
13
15
Position
17
19
21
23
25
27
29
miRNAs with positioned CNNC
PhyloP
Vertebrate
Conservation
-4
chr6
72,113,234–72,113,242
(–)
chr13
50,623,097–50,623,105
(–)
4
0
C(18) only
CNNC
C(19) only
C(16) only
Neither C
Relative cleavage
chr19
52,196,597–52,196,605
(+)
CNNC
2
C(21) Only
2
CNNC
2
C(20) Only
4
C(17) Only
4
Neither C
4
CNNC
4
C(21) only
8
Neither C
8
Timepoint 2
B
3
hsa-mir-223
8
8
4
hsa-mir-30a
C(18) Only
hsa-mir-16-1
Timepoint 1
Neither C
hsa-mir-125a
Ecdysozoans
A
30%
20%
10%
0%
The mir-125a pri-miRNA also had four C residues in the vicinity (positions 16–21), which gave
rise to a CNNC at position 16 and the possibility of creating a CNNC at positions 17 or 18 (by
changing A20 to a C or changing A18 to a C, respectively). However, neither C of the CNNC at
position 16 was preferred in the selection, nor were either of the single-nucleotide changes that
could create a CNNC, and the position 16 CNNC was not conserved in vertebrate orthologs
(Figure 2D, Figure 5A and 5B). These results indicate that unidentified sequence features
present in mir-16-1, mir-30a, and mir-223 but absent in the mir-125a pri-miRNA are required for
the CNNC to exert its effect in increasing Microprocessor cleavage efficiency.
For the three pri-miRNAs in which the CNNC motif was effective, its position fell in a
small window 17–18 nt downstream of the Drosha cleavage site. In variants in which neither
wild-type C was present, an alternative CNNC was strongly preferred one or two nucleotide
registers downstream, which further indicated that a CNNC motif within a small range of
positions can contribute to pri-miRNA recognition (Figure S5A).
Analyses of the 3′ regions of conserved human pri-miRNAs revealed that of the 64
possible dinucleotide motifs with 0–3 intervening nucleotides, CNNC was most highly enriched
(Figure 5C). Moreover, enrichment was limited a small range of positions 16–18 nt downstream
of the Drosha cleavage site, peaking at positions 17 and 18, which matched the positions of the
motifs originally found within mir-16-1, mir-30a, and mir-223. These results suggest that the
CNNC motif enhances recognition and cleavage of many human pri-miRNAs.
Analyses of miRNAs in non-mammalian species indicated strong, position-specific
Figure 5. Enrichment and conservation of the downstream CNNC motif.
(A) Relative cleavage of variants with a full CNNC motif, a partial motif, and no motif. Relative cleavage
values were normalized to that of variants with no motif.
(B) PhyloP conservation across 30 vertebrate species in the region of the downstream CNNC motif (blue
letters) for the four selected miRNAs. Bars extending beyond the scale of the graph are truncated
(red).
(C) Comparison of CNNC enrichment to that of 63 other spaced dinucleotide motifs. Pri-miRNAs
sequences from each species (Table S2) were aligned according to the predicted Drosha cleavage
site, and the occurrences of each spaced dinucleotide motif tabulated. Occurrences were
normalized to those expected by chance, based on the nucleotide composition downstream of primiRNAs in each species.
(D) Enrichment of the CNNC motif in the pri-miRNAs of representative bilaterian animals (Table S2). For
each species, pri-miRNA sequences were aligned as in (C) and downstream CNNC occurrences
tabulated. Species with a statistically significant enrichment at positions 16, 17, or 18 are marked
–4
with an asterisk (empirical p-value < 10 ).
77
A
Linear pri-miRNA substrate
(pool of variants)
ppp
Drosha
DGCR8
Functional variants
ppp
Nonfunctional variants
Library for
single-end sequencing
0.4
0.2
0
−0.2
A
–0.69 –0.79 –1.03 0.52
C
–0.96 –0.97 0.62
G
–0.14 0.63 –0.76 0.28
U
−0.6
−0.8
A
Pair 18
0.49
0.19 –0.76 –0.64
Pair 17
−0.4
Position P43
C
G
U
0.30
Position P42
C
G
U
A
–0.93 –1.29 –0.72 0.22
C
–0.98 –1.58 1.03 –1.00
G
–0.7
U
0.33 –1.02 –0.2 –0.62
0.55 –0.4 –0.25
A
Position P41
C
G
U
A
–0.88 –0.54 –0.49 0.51
C
–1.03 –1.09 0.45 –1.05
G
–0.62 0.88 –0.03 –0.28
U
0.87 –0.67 –0.29 –0.82
Pair 19
A
A
–1.3 –1.38 –1.77 0.26
–0.9 –1.33 0.31 –0.91
G
–1.27
19 21 23 25
Apical stem pairs
Pair 20
Position P40
C
G
U
C
U
17
19 21 23 25
Apical stem pairs
Position P22
0.6
17
hsa-mir-223
64
32
16
8
4
2
1
0.5
0.25
0.125
1.4 –1.16 –0.37
Position P39
C
G
U
A
–1.17 –0.93 –1.23
C
–0.63 –0.88 0.81 –1.02
G
–0.96 0.39 –0.73 –0.23
U
0.98 –0.93 –0.13 –1.04
Pair 21
0.27 –1.32 –0.91 –0.83
A
A
0.10
Position P38
C
G
U
A
–1.24 –1.35 –1.06 0.12
C
–1.11 –1.23 0.47 –0.94
G
–0.77
U
1.1
–0.94 –0.08
0.16 –1.01 –0.16 –0.37
−1
D
hsa-mir-30a
2
1
Timepoint 2
Mature
miRNA
P20 P22 P24 P26 P28 P30
0.5
E
4
0
A AG CT G T GAAG
chr6
–4
72,113,290–72,113,300
(–)
F
H. sapiens
D. rerio
C. intestinalis
A. gambii
D. melanogaster
D. pulex
C. elegans
C. briggsae
P. pacificus
C. teleta
L. gigantea
S. mediterranea
N. vectensis
Human UGU/GUG window
*
*
*
*
*
*
*
Position
miRNAs with positioned UGU/GUG
Timepoint 1
P20
P21
P22
P23
P24
P25
P26
P27
P28
P29
P30
P31
G
A-U
P23 G-C P38
P22 U-A P39
P21 G-C P40
P20 U-A P41
P19 C-G P42
P18 C-G P43
15th Pair
A-U
A |
U |
U-G
U-A
hsa-mir-125a C-G
C-G
pre-miRNA
C
U
A-U
G-C
A-U
G-U
U-G
C-G
C-G
C
A
P1 U-G
C
C P62
0.8
A
Odds ratio
(WT base)
A
G
G
Position P18
G
Pair 16
1
C
19 21 23 25
Apical stem pairs
Position P20
UCC
A
C
A
G
17 19 21 23
Apical stem pairs
hsa-mir-30a
64
32
16
8
4
2
1
0.5
0.25
0.125
Position P21
15
hsa-mir-16-1
64
32
16
8
4
2
1
0.5
0.25
0.125
Position P23
hsa-mir-125a
Position P19
C
64
32
16
8
4
2
1
0.5
0.25
0.125
Timepoint 2
Odds ratio (log2)
Relative cleavage
Timepoint 1
PhyloP
vertebrate
conservation
B
20%
15%
10%
5%
0%
enrichment of the CNNC motif in chordates, arthropods and lophotrochozoans (Figure 5D).
Indeed the CNNC motif was the most enriched dinucleotide motif in the downstream region of
both Drosophila and planarian pri-miRNAs (Figure 5C). Positional enrichment of CNNC was
not observed in sea anemone (Nematostella vectensis), suggesting that usage of motif for primiRNA recognition emerged after the divergence of bilaterians, around the time of the evolution
of the core bilaterian miRNAs (Sempere et al., 2006; Grimson et al., 2008). Interestingly,
enrichment was also absent in nematodes (Figure 5C and D), suggesting an isolated loss of this
mode of recognition in the nematode but not the arthropod branch of the ecdysozoans.
The contributions of basal sequence and structure motifs were confirmed in HEK293T
cells (Figure S5C). Mutation of the basal UG and CNNC motifs each reduced accumulation of
mature miR-30a; mutation of both together reduced accumulation ~8-fold relative to wild type.
Loop and apical stem elements can enhance processing
In addition to the basal stem and flanking regions, another potential location for features
required for processing is in the pri-miRNA loop and apical stem. Indeed, this part of the premiRNA has been the region most intensively studied as potentially harboring determinants of
pri-miRNA recognition. Point mutations in this region abolish cleavage (Zeng et al., 2005;
Gottwein et al., 2006) and this region contains binding sites for proteins reported to modulate
Figure 6. Identification of the apical pairing features and the UGUG motif.
(A) Schematic of the in vitro selection for functional pri-miRNA variants with partially-randomized
sequences in the apical stem and terminal loop. Linear pri-miRNA variants substrates were
incubated in whole-cell lysate from HEK293T cells overexpressing Drosha and DGCR8. Cleaved
pre-miRNA variants were gel-purified, reverse transcribed, and amplified for high-throughput
sequencing.
(B) Relative cleavage of variants with different apical stem lengths. The number of contiguous Watson–
Crick pairs was counted and the relative cleavage calculated, normalized to that of the 15 bp stem.
(C) Predicted secondary structure and covariation matrices for the apical stem of mir-125a. Otherwise,
as in Figure 3A.
(D) Relative cleavage of variants with the apical UGUG motif beginning at the indicated positions,
normalized to variants without the motif. Nucleotides of the mature miRNA are shaded in yellow.
(E) Conservation of the region centered on the apical UGUG of mir-30a. Otherwise, as in Figure 4B.
(F) Enrichment for UGU or GUG trinucleotides in the terminal loops of metazoan pri-miRNAs (Table
S2). For each species, pri-miRNA sequences were aligned according to the predicted Drosha
cleavage site and occurrences of loop UGU or GUG trinucleotides tabulated. Species with
–5
statistically significant enrichment are marked with an asterisk (empirical p-value < 10 ).
79
pri-miRNA processing (Guil and Caceres, 2007; Michlewski et al., 2008; Viswanathan et al.,
2008; Trabucchi et al., 2009). Moreover, the distance from the junction of the terminal loop and
apical stem was reported to determine the Microprocessor cleavage site of hsa-mir-30a (Zeng et
al., 2005), although the terminal loop is dispensable for processing of hsa-mir-16-1 (Han et al.,
2006).
To find processing determinants in this region, we partially randomized the loop and
apical stem sequences of each of the four pri-miRNAs, incubated each pool of variants with
Microprocessor-containing extract, gel-purified the pre-miRNA cleavage products, and prepared
them for high-throughput sequencing (Figure 6A). Comparison to sequences of the starting
pools showed that pairing at the apical portion of the stem contributed to pri-miRNA recognition
and processing for some miRNAs, although the preferred structures differed for different primiRNAs, as might have been suspected based on the different conclusions drawn previously
from studies of different pri-miRNAs (Zeng et al., 2005; Han et al., 2006). For mir-125a, 22 bp
above the 5p Drosha cleavage site was strongly preferred; longer stems were tolerated, whereas
shorter stems were disfavored (Figure 5B). Watson–Crick pairing throughout the apical stem
was supported by analysis of covariation (Figure 6C). A 22-pair apical stem was also preferred,
albeit more weakly, in mir-30a (Figure 6B, Figure S6B). By contrast, no preference for apical
pairing was observed in the stems of mir-16-1 and mir-223 (Figure 6B, Figure S6C). Indeed,
lengthening of the mir-16-1 apical stem at the expense of loop size was detrimental (Figure 5B),
which was consistent with a previous report (Zhang and Zeng, 2010).
Because several loop-binding protein regulators of miRNA processing have been
reported, we looked for evidence of primary-sequence motifs in the loops. Overall, enrichment
was weaker than that observed for flanking residues, particularly for mir-16-1, which showed
almost no primary-sequence enrichment throughout the variable region (Figure S6A). The best
candidate for a loop-binding motif was observed only in mir-30a, in which the wild-type UGUG
at positions P24–27 was preferred (Figure S6A). This motif overlapped a region of vertebrate
conservation that included the last base of the most commonly sequenced isoform of mature
miR-30a (Figure 6D). Human and zebrafish miRNAs were enriched in UGU or GUG in this
region of the loop (empirical p < 10-5 for each species), as were the arthropods and one of three
lophotrochozoans examined (empirical p < 10-5 for each) (Figure 6E). However, the lack of
80
enrichment in several other representative species raises the question of whether the usage of this
motif arose independently in multiple lineages or was ancestral and lost multiple times.
Rescue of C. elegans miRNA expression in human cells
The primary-sequence motifs found in this study are absent in nematode clade, either
because an ancestral mode of recognition was lost (e.g., downstream CNNC), or because the use
of a particular motif is an innovation more recent than the divergence of the vertebrate lineage
from nematodes (e.g., basal UG).
Using our newly acquired knowledge of pri-miRNA
recognition, we tested whether missing primary-sequence motifs might account for the failure of
C. elegans pri-miRNAs to be processed in human cells. We focused on the basal UG and the
flanking CNNC motifs because these were implicated in three of the four human pri-miRNAs
analyzed in detail and thus seemed most likely to function in a variety of pri-miRNA contexts.
These motifs were systematically added to cel-mir-44 in the context of the bicistronic vector,
after first disrupting the predicted pairing between positions –14 and +12 and substituting the
G:C pair at positions –13 and +11 (Figure 7A, construct mir44.1). These changes, which were
expected to simultaneously enhance processing by shortening the basal stem to its optimal length
and inhibit processing by replacing the fortuitous G at position –13, had a marginal net effect on
production of mature miR-44 in human HEK293T cells (Figure 7A). Adding a basal UG
(construct mir44.3) enhanced production of mature miR-44 by 5-fold (8-fold over the wildtype),
primarily from restoring the G at –13 (Figure 7A). Adding a CNNC 17 nt downstream of the
cleavage site (mir44.4) enhanced production another 8-fold, thereby yielding a 64-fold increase
over wildtype (Figure 7A). Likewise, converting the wild-type, asymmetrically bulged stem of
cel-mir-50 to a regular, 11-pair stem and adding the basal UG and CNNC motifs enhanced
expression of mature miR-50 by 31-fold (Figure S7A), while adding the basal UG and CNNC
motifs to cel-mir-40 enhanced expression of mature miR-40 by 5-fold (Figure S7B).
We
conclude that primary-sequence motifs discovered in this study enable human cells to distinguish
pri-miRNAs hairpins from other hairpins and that the absence of these motifs in C. elegans primiRNAs helps to explain why human cells do not regard these transcripts as pri-miRNAs.
81
A
mir44.wt
UGAAA-
Query
pri-miRNA
?
?
Gppp
U
GU
--- AA
GGCCAA CUGGAUGUG CUC UGGUCAUA GACG UC \
CCGGUU GACUUACAC GAG AUCAGUAU UUGU AG C
CGU
C
A
-AC
CA CA
UUUUGA
hsa-mir-1-1
TK pA
mir44.1
UG(–14)
CNNC
WT
-14 Mismatch
-14 Mismatch
-14 Mismatch
-14 Mismatch
AG
AC
CG
UG
UG
None
None
None
None
CNNC(+17)
4
cel-miR-44-3p
hsa-miR-1
128
mir44.2
UGAAA-
C
AAU
U
-
GU
--
--
AA
AG GAA
GGCCAA CUGGAUGUG CUC UGGUCAUA GACG UC |
UC CUU
CCGGUU GACUUACAC GAG AUCAGUAU UUGU AG C
UUUUGA U
CGU
C
A
-AC
CA CA
mir44.3
UGAAA-
U
AAU
U
-
GU
--
--
64
miR-44-3p expression
Basal Stem
3
20-
A
AAU
U
GU
--- AA
AG CAA
GGCCAA CUGGAUGUG CUC UGGUCAUA GACG UC |
UC GUU
CCGGUU GACUUACAC GAG AUCAGUAU UUGU AG C
UUUUGA C
CGU
C
A
-AC
CA CA
mir44.wt
mir44.1
mir44.2
mir44.3
mir44.4
2
30-
UGAAA-
Construct
wt 1
70605040-
cel-miR-44-3p probe binding site
AAAAA
mir-44 mutant
CMV Promoter
AAU
AGAGAA
UCUCUU
Control
pri-miRNA
AA
AG GAA
GGCCAA CUGGAUGUG CUC UGGUCAUA GACG UC |
UC CUU
CCGGUU GACUUACAC GAG AUCAGUAU UUGU AG C
UUUUGA U
CGU
C
A
-AC
CA CA
32
16
8
4
2
1
0.5
wt 1
2
3
4
B
G
UG
Apical stem
structure
–1
5p Arm
+1
3p Arm
Basal stem
structure
CNNC
Other loop
positions
Other loop
positions
Apical stem
0
Basal stem
0
Loop GUG
0
Apical stem
0.1
0
CNNC
0.1
Other basal
positions
0.1
UG
0.2
0.1
Basal stem
0.2
UG
0.2
CNNC
Other basal
positions
apical stem
Other loop
positions
0.3
0.2
Basal stem
0.4
0.3
Other loop
positions
0.4
0.3
Apical stem
0.4
0.3
UG
0.4
Other basal
positions
hsa-mir-223
0.5
CNNC
hsa-mir-30a
0.5
Other basal
positions
hsa-mir-16-1
0.5
UG
hsa-mir-125a
0.5
Basal stem
C
Average information
(bits/nt)
UG
D
UG(–14)
1.3%
17%
7.4%
9.9%
33%
4.5%
2.5%
5.0%
0.6%
0.1%
Loop GUG(P22–P24)
CNNC(16–17)
No motif
E
2.2%
76%
79%
8.2%
0.7% 6.6% 4.7%
12%
9.3%
9.9%
21%
Human pri-miRNAs
65%
Chance
5.6% 1.9%
5.6%
C. elegans pri-miRNAs
7.9%
0.5% 4.7%
Chance
Discussion
We find that secondary structure is generally inadequate on its own to specify pri-miRNA
hairpins: Primary-sequence features, including the basal UG, the CNNC and the apical GUG
motifs, also contribute to efficient processing in human cells (Figure 7B). Complicating the
story (and perhaps explaining why these primary-sequence features had not been observed
earlier), different pri-miRNAs differentially benefit from the different motifs (Figure 7C).
Among human pri-miRNAs, these motifs were nonetheless highly enriched over chance
expectation, with 79% of the conserved human miRNAs containing at least one of the three
motifs (Figure 7D).
The motifs were not enriched in C. elegans pri-miRNAs (Figure 7E), and when added to
the C. elegans pri-miRNAs, the motifs conferred more efficient processing in mammalian cells
(Figure 7A and Figure S7). These experiments that added mammalian features to C. elegans
miRNAs also showed the benefit of disrupting pairing normally present at positions –14 and +12
of the C. elegans miRNAs (Figure 7A and Figure S7). The presence of pairing that is inhibitory
to mammalian processing suggests that measurement from the base of the helix might also differ
in nematodes. We conclude that despite the many broadly conserved features of miRNAs, some
Figure 7. Structural and primary-sequence features important for human pri-miRNA processing.
(A) Processing enhancement from addition of human pri-miRNA motifs to C. elegans mir-44. Changes
that sequentially introduce the listed features were incorporated into mir-44 within the bicistronic
expression vector (left). Secondary structures are shown for mutations predicted to affect the wildtype basal stem (middle) with the annotated Drosha cleavage sites (purple arrowheads). After
transfection into HEK293T cells, accumulation of miR-44-3p was assessed on an RNA blot,
normalizing to the expression of the hsa-miR-1 control, and increased miR-44-3p expression is
plotted (geometric mean ± standard error, n = 3, right).
(B) Summary of human pri-miRNA recognition determinants.
(C) Contributions of individual motifs to in vitro processing. For each pri-miRNA, average information
content per nucleotide is plotted for the indicated features and positions.
(D) Enrichment of primary-sequence motifs in human pri-miRNAs conserved to mouse (Table S2).
Human pri-miRNAs were classified based on whether they had the basal UG, an apical GUG or
UGU, or the flanking CNNC motif (left). Expectations by chance (right) were estimated based on
the nucleotide composition of human pri-miRNAs upstream of the Drosha cleavage site, in the premiRNA, and downstream of the cleavage site for the basal UG, apical GUG or UGU, and flanking
CNNC motifs, respectively.
(E) A search for human primary-sequence motifs in C. elegans pri-miRNAs conserved in other
nematodes (Table S2). Pri-miRNA sequences were analyzed as in (D); the smaller Venn diagrams
reflect the smaller number of analyzed miRNAs.
83
primary-sequence features and some secondary-structure features differ in mammals and
nematodes, which implies that important aspects of pri-miRNA biogenesis differ in different
metazoan lineages.
About a fifth of human pri-miRNAs lack all three newly identified primary-sequence
determinants (Figure 7D).
These are attractive subjects for further study, in that the
combinatorial approach implemented here presumably would identify additional unique
determinants used by these pri-miRNAs. Sequence and structural determinants probably also
exist at the Microprocessor cleavage site and the middle of pri-miRNA stem, regions that were
inaccessible to our approach as implemented. Indeed, point mutations that disrupt pairing in the
middle of the stem dramatically impair processing (Gottwein et al., 2006; Duan et al., 2007;
Jazdzewski et al., 2008; Sun et al., 2009), and although the cleavage site has not been directly
implicated in processing in human miRNAs, the Drosha cleavage sites of C. elegans miRNAs
are enriched for symmetric internal loops, which presumably reflect preferences at the level of
pri-miRNA processing, nuclear export, dicing, or loading (Warf et al., 2011). Also hinting at the
possibility of additional primary-sequence preferences within the stem are results from bacterial
RNase III, which avoids specific base-pair identities in the “proximal box” and the “distal box”
(Zhang and Nicholson, 1997), and fungal Rnt1 and Pac1, which are also influenced by similarly
positioned motifs (Lamontagne and Elela, 2004). The proximal box is adjacent to the cleavage
site, and the distal box is 8 bp away from the cleavage site, but nonetheless in a region not
interrogated in our experiments.
Although more needs to be learned about the recognition of pri-miRNAs for processing,
the emerging picture is that of a modular phenomenon in which each module contributes
modestly to overall discrimination, and different pri-miRNAs depend on any individual module
to varying degrees. Our results quantify the relative importance of each module for each primiRNA (Figure 7C). Pairing within the basal stem was crucial, as expected from previous
analyses (Lim et al., 2003b; Han et al., 2006). In addition, all four miRNAs made use of the
basal UG motif, which provided as much or more information content per nucleotide as the basal
stem nucleotides. For the three miRNAs that used a CNNC motif, the motif information content
per nucleotide was comparable to that of the basal stem nucleotides. Compared to these motifs,
other flanking nucleotides contributed very little to the selection information content.
84
Apical and terminal loop elements were less important than the basal motifs (Figure 7C).
We detected significant contributions only in pri-mir-125a, in which the apical stem nucleotides
were as important as the basal stem nucleotides, and in pri-mir-30a, in which the loop UGUG
motif contributed some information, albeit less than any of the three basal motifs. Together, both
basal and apical motifs described here explained 61–78% of the nucleotide enrichment observed
in the selected sequences. The remaining information content was diffusely distributed among
the other partially-randomized positions. Although some of this remaining enrichment could
have reflected small beneficial contributions of flanking bases, we suspect that most reflected
avoidance of detrimental alternative structures.
A better understanding of features important for miRNA biogenesis will aid in
interpreting human mutations that affect mature miRNA levels. For example, loss of mir-16-1
expression associated with chronic lymphocytic leukemia (CLL) is typically due to deletions
spanning the intron that contains both hsa-mir-15a and hsa-mir-16-1 (Calin et al., 2002).
However, in a study of 75 CLL patients, two had tumors that retain the pri-miRNA hairpins and
instead carry a germline C>T single-nucleotide polymorphism (SNP) downstream of the mir-161 hairpin (Calin et al., 2005). This SNP lowers overexpression of miR-16-1 in HEK293 cells,
and in both patients heterozygosity for the SNP was lost in the tumor, which suggests that it was
a driver mutation (Calin et al., 2005). This SNP corresponds to the C at +18, the first C in the
mir-16-1 CNNC motif, which explains why this mutation flanking the hairpin lowers miR-16
accumulation and leads to CLL: it affects pri-miRNA processing by disrupting the mir-16-1
CNNC motif. Discovery of additional motifs for pri-miRNA recognition and processing and
identification of proteins that recognize these motifs may lead to improved diagnostic and
therapeutic tools in cancer and other diseases in which miRNAs are dysregulated.
Experimental Procedures
Ectopic pri-miRNA expression in HEK293 cells and S2 cells
A genomic fragment corresponding to the human mir-1-1 hairpin and flanking sequences
was amplified and cloned into both pcDNA3.2/V5-DEST (Invitrogen) and pMT-DEST
(Invitrogen) expression plasmids. Query pri-miRNA sequences were cloned into these plasmids
using the Gateway system (Invitrogen), such that they were transcriptionally fused upstream of
85
mir-1-1. Expression plasmids and pMAX-GFP were co-transfected into HEK293 cells using
Lipofectamine 2000 (Invitrogen) and co-transfected into S2 cells using Cellfectin (Invitrogen)
according to manufacturer’s instructions. After 36–48 h, total RNA was collected by addition of
Tri-Reagent (Ambion) according to manufacturer’s instructions. RNA blots for detecting mature
and pre-miRNAs were as described. Ribonuclease protection assays were performed with the
RPA III kit (Invitrogen) according to manufacturer’s instructions.
For detection of expression by sequencing, total RNA from individual transfections was
combined and libraries for small-RNA sequencing prepared as described (Chiang et al., 2010).
Sequencing reads were mapped to a miRNA hairpin collection composed of the miRBaseannotated hairpins of miRNAs endogenously expressed in the cell line and the miRBaseannotated hairpins of the transfected miRNAs. Reads were included if they perfectly matched a
hairpin in this library and excluded if they matched more than one hairpin corresponding to a
transfected miRNA.
Read counts were normalized to the total reads matching a set of
endogenous hairpins that had no transfected counterparts.
For each expressed pri-miRNA
hairpin, number of reads reported is the number obtained after subtracting the number observed
in a normalized, mock-transfected control library.
Whole-cell lysate with overexpressed Microprocessor complex
Microprocessor lysates were prepared as described (Lee and Kim, 2007), with minor
modifications. HEK293T cells were transfected with a mixture of pCK-Drosha-FLAG (Lee and
Kim, 2007) pFLAG-HA-DGCR8 (Landthaler et al., 2004), and a transfection-control plasmid
pMAX-GFP (Amaxa) using Lipofectamine 2000 (Invitrogen) according to the manufacturer’s
instructions. For catalytically-inactive Microprocessor lysates, pCK-DroshaTN-FLAG replaced
the wild-type Drosha plasmid (Han et al., 2009). After 72 h, cells were harvested by rinsing the
monolayer in phosphate buffered saline (PBS, 137 mM NaCl, 2.7 mM KCl, 1.5 mM KH2PO4,
8 mM Na2HP04, [pH 7.4]). Cells were pelleted, resuspended in sonication buffer (100 mM KCl,
0.2 mM EDTA, 20 mM Tris-Cl pH 8.0, and 0.7 µl/ml 2-mercaptoethanol) supplemented with
mini-EDTA Free Protease Inhibitor tablets (Roche), and sonicated. The sonication lysate was
cleared by centrifugation and cell lysis was confirmed by the liberation of GFP into the
supernatant. The supernatant was distributed into single-use aliquots, and stored in liquidnitrogen vapor phase.
86
Competitive binding and cleavage assays
The competitive binding assay was based on that of Bartel et al. (1991). T7-transcribed
~200 nt pri-miRNA substrates were gel-purified, treated with calf intestinal phosphatase (NEB),
extracted in Tri-Reagent (Invitrogen), and 5′ end-labeled by phosphorylation using T4
Polynucleotide Kinase (NEB) and γ-[32P]-ATP. Reference substrates were the same, except for a
10–25 nt difference in length, which enabled separation on denaturing gels.
Complexes
containing Drosha-TN and DGCR8 were immunopurified as described (Lee and Kim, 2007; Han
et al., 2009).
Competitor and reference RNAs were mixed and incubated with
immunoprecipitated Drosha-TN and DGCR8 for 15-30 min [final concentrations, 250 nM each
RNA, 100 mM KCl, 1 mM MgCl2, 0.2 mM EDTA, 20 mM Tris-Cl (pH 8.0), 0.7 µl/ml 2mercaptoethanol and 300 ng/µl yeast total RNA (Ambion)]. RNA-protein complexes were
filtered on Immobilon-NC nitrocellulose filter discs (Millipore), washed with at least 10 reaction
volumes of sonication buffer. RNA was eluted from the membrane by incubating in elution
buffer (300 mM NaCl, 8M urea, and 25 mM EDTA) for 10 min at 85ºC, ethanol precipitated and
resolved on a denaturing 5% polyacrylamide gel.
For competitive cleavage, 5′ end-labeled query and reference pri-miRNA substrates were
mixed and incubated in whole-cell lysate from HEK293T cells overexpressing Drosha and
DGCR8 [final concentrations, 50 nM each RNA, 100 mM KCl, 1 mM MgCl2, 0.2 mM EDTA,
20 mM Tris-Cl (pH 8.0), 0.7 µl/ml 2-mercaptoethanol, 300 ng/µl yeast total RNA, 10 nM
Microprocessor complex (concentration estimated exploiting the single-turnover behavior of the
Microprocessor when cleaving linear pri-miR-125a)]. After incubation for 30 seconds at 37ºC
the reaction was stopped by addition of Tri-Reagent (Ambion) with mixing. Extracted RNA was
precipitated with isopropanol, then resuspended and resolved on a denaturing 5% polyacrylamide
gel.
Synthesis of pools of pri-miRNA variants
Linear pri-miRNA variants for the apical stem and loop selections were transcribed by T7
RNA polymerase from oligonucleotide templates (Table S1) that were synthesized using
nucleoside phosphoramidite mixtures such that they varied at specified positions (IDT). The
transcription reaction included α-[32P]-UTP to body-label the product. Pri-miRNA pools were
gel-purified on urea-polyacrylamide gels before use.
87
For circular pri-miRNA variants, body-labeled linear precursors were transcribed by T7
RNA polymerase from synthetic oligonucleotide templates (Table S1). Each transcript ended
with a minimal HDV ribozyme (Schurer et al., 2002) that co-transcriptionally self-cleaved at a
defined position to produce a homogenous 3′ end. After treatment with TurboDNAse (Ambion),
transcripts were gel-purified, treated with calf intestinal phosphatase (NEB) to remove the 5′
triphosphate, extracted with Tri-Reagent, precipitated with isopropanol, and treated with T4
polynucleotide kinase (NEB) to remove the 2′-3′ cyclic phosphate as described (Guo et al.,
2010). After ethanol precipitation, the transcripts were 5′ phosphorylated with T4 polynucleotide
kinase, diluted, and ligated using T4 RNA ligase 1 (NEB). Circularized pri-miRNAs were
purified from linear species on denaturing polyacrylamide gels.
In vitro selection and high-throughput sequencing
Pools of variants were incubated in HEK293T whole-cell lysate overexpressing FLAGtagged Drosha and FLAG-HA-tagged DGCR8 (Lee and Kim, 2007). At one or two time points
(for circularized pri-miRNA variants, 1 minute for mir-125a, 1 and 4 minutes for mir-16-1, 1 and
5 minutes for mir-30a, and 3 and 15 minutes for mir-223; for apical stem and loop variants, 5
seconds and 15 seconds for mir-125a, 15 seconds and 2 minutes for mir-16-1, 30 seconds and 2
minutes for mir-30a, and 30 seconds and 2 minutes for mir-223) reactions were stopped by
addition of Tri-Reagent (Ambion) with mixing, and cleaved products were purified from
denaturing gels.
Cleavage products of circularized pri-miRNA variants were splint-ligated
(Moore, 1999) to oligonucleotide adaptors containing barcode sequences, reverse transcribed,
and amplified. To sequence the initial pools, a sample of phosphorylated, uncircularized RNA
was reverse transcribed and amplified.
Amplicons from the initial pools and the cleaved
products were pooled for Illumina paired-end sequencing (75 nt reads per end). Pre-miRNA
cleavage products of linear pri-miRNA variants were reverse transcribed, amplified. To sequence
the initial pools, a sample of the pool was taken before cleavage, reverse transcribed, and
amplified. Amplicons from the initial pools and cleaved pre-miRNA products were pooled for
Illumina single-read sequencing (54 nt reads)
88
Sequence analysis
High-throughput sequencing reads were divided into individual experimental groups
according to constant sequences specific to each pri-miRNA, and further divided based on
barcode. After filtering for sequencing quality, discarding any sequences that had an error rate
≥0.1 (phred score ≤10) at any variant position, the sequencing error averaged <0.001 per variant
position (average phred score >30). We also discarded sequences in which the length of a
partially randomized region differed from that of the wildtype, thereby eliminating many
sequences with insertions or deletions. Libraries were collapsed so that sequences that appeared
multiple times with the same bar code were considered just once in the analysis (although in
retrospect this precaution was not required because there was no group of dominant, multi-copy
sequences that would have biased the analysis).
To calculate the information content at each position, we used the data from the initial
pool sequences and the product sequences to calculate the relative cleavage of each base versus
the other three bases. For example, for the A residue, the three relative cleavage values are given
below, where P(N) is estimated by the frequency of a base in the initial pool, and P(N|cleavage)
is estimated by the frequency of that base in the product sequences.
𝑃(cleavage|𝐶) 𝑃(𝐶|cleavage) 𝑃(𝐶)
=
�
𝑃(cleavage|𝐴) 𝑃(𝐴|cleavage) 𝑃(𝐴)
𝑃(cleavage|𝐺) 𝑃(𝐺|cleavage) 𝑃(𝐺)
=
�
𝑃(cleavage|𝐴) 𝑃(𝐴|cleavage) 𝑃(𝐴)
𝑃(cleavage|𝑈) 𝑃(𝑈|cleavage) 𝑃(𝑈)
=
�
𝑃(cleavage|𝐴) 𝑃(𝐴|cleavage) 𝑃(𝐴)
We then used Bayes’ Theorem (Pitman, 1993) to infer the nucleotide composition that
would have resulted after selection from a pool of variants in which there was an equal
probability of an A, C, G, or U at this position. For example, the formula to infer the frequency
of A at a particular position after selection from such a pool was
89
𝑃(cleavage|𝐶)
𝑃inferred A = 𝑃(𝐴|cleavage) = �1 + 𝑃(cleavage|𝐴)
+ 𝑃(cleavage|𝐺)
+ 𝑃(cleavage|𝑈)
�
𝑃(cleavage|𝐴)
𝑃(cleavage|𝐴)
−1
The inferred post-selection distribution was then used to calculate information content
scores for each nucleotide at each position. For example, the information content for A at a
particular position was calculated as
𝐼𝐴 = 𝑃inferred A × [𝑙𝑜𝑔2 (𝑃inferred A ) + 2]
If results from two time points were available, information content values were averaged.
For evaluating motifs, we calculated a relative cleavage value based on the frequencies of
the motif in the reference and selected pools [P(motifi) and P(motifi)|cleavage), respectively], and
the frequencies of a reference motif in the reference and selected pools [P(motifref) and
P(motifref)|cleavage), respectively].
Relative cleavage =
𝑃(motif𝑖 |cleavage)
𝑃(motif𝑖 )
�
𝑃(motifref |cleavage) 𝑃(motifref )
We also used an odds ratio score to calculate the enrichment for particular motifs by
using the frequency of the motif in the reference and selected pools [P(motifi) and
P(motifi)|cleavage), respectively].
Odds ratio =
𝑃(motif𝑖 |cleavage)
𝑃(motif𝑖 )
�
1 − 𝑃(motif𝑖 |cleavage) 1 − 𝑃(motif𝑖 )
If two timepoints were available, the geometric mean of the ratios was reported, unless
noted otherwise.
To screen for specifically for Watson–Crick pairing between all possible combinations of
randomized positions, we used a scoring metric to compare the geometric average of odds ratios
for Watson–Crick pairing to that of odds ratios for non-Watson–Crick pairs.
90
Pairing score = �
�
Watson–Crick
1/4
Odds ratio�
−�
�
non−Watson–Crick
1/12
Odds ratio�
Positional enrichments of sequence motifs
Enrichment of a motif at a set of positions relative to the cleavage site was computed by
generating 100,000 cohorts of miRNAs in which the upstream, downstream and pre-miRNA
sequences were independently shuffled. An empirical P-value was computed by comparing the
number of miRNAs that contained at least one match to the motif in the window to the number
of matches in each of the random cohorts.
pri-miRNA collections
A list of representative pri-miRNAs used for analyses is provided (Table S2).
Coordinates of miRNA loci in miRBase version 17 (Kozomara and Griffiths-Jones, 2011) were
used to extract the sequences of the annotated hairpin and 200 genomic bases flanking each side.
miRBase hairpin sequences and flanking genomic sequences (20 nt on each side) were folded
using RNAFold (Hofacker and Stadler, 2006). The Microprocessor cleavage site was inferred
using the mature sequences annotated in miRBase. Only hairpins for which the predicted folding
and the annotated mature sequences could be reconciled as a 2 nt 3′ overhang were carried
forward for analysis.
For hairpins in miRBase-annotated miRNA families, a single
representative was chosen to represent the family in each species. For human, D. melanogaster,
and C. elegans, the family member with the most conserved pre-miRNA sequence was chosen.
For other species, the representative was chosen at random. Whole-genome alignments and
phylogenetic trees were obtained from the UCSC genome browser (Fujita et al., 2011).
Conservation of a pre-miRNA was defined as the average conservation across the pre-miRNA,
and the conservation of a base was defined as the ratio between the total branch length of the
species that contained the same base as the reference sequence and the total branch length of the
species that had an aligned base at that position.
91
Accession Numbers
Sequencing data have been deposited into the Short Read Archive (SRA, accession
number SRA051323).
Acknowledgements
This work was done in collaboration with Igor Ulitsky, who carried out much of the
evolutionary conservation analysis. I thank D. Shechner for help with circularized-substrate
selections; C. Jan, O. Rissland, D. Weinberg, J. Ruby, J. Nam, and V.N. Kim for valuable
discussions; and L. Schoenfeld and J. Lassar for technical assistance This work was supported
by NIH grant GM067031 (to D.P.B.).
92
Bibliography and References Cited
Babiarz, J.E., Ruby, J.G., Wang, Y., Bartel, D.P., and Blelloch, R. (2008). Mouse ES cells
express endogenous shRNAs, siRNAs, and other Microprocessor-independent, Dicerdependent small RNAs. Genes Dev 22, 2773-2785.
Bar, M., Wyman, S.K., Fritz, B.R., Qi, J., Garg, K.S., Parkin, R.K., Kroh, E.M., Bendoraite, A.,
Mitchell, P.S., Nelson, A.M., et al. (2008). MicroRNA discovery and profiling in human
embryonic stem cells by deep sequencing of small RNA libraries. Stem Cells 26, 2496-2505.
Bartel, D.P. (2004). MicroRNAs: genomics, biogenesis, mechanism, and function. Cell 116, 281297.
Bartel, D.P. (2009). MicroRNAs: target recognition and regulatory functions. Cell 136, 215-233.
Bartel, D.P., Zapp, M.L., Green, M.R., and Szostak, J.W. (1991). HIV-1 Rev regulation involves
recognition of non-Watson-Crick base pairs in viral RNA. Cell 67, 529-536.
Bentwich, I., Avniel, A., Karov, Y., Aharonov, R., Gilad, S., Barad, O., Barzilai, A., Einat, P.,
Einav, U., Meiri, E., et al. (2005). Identification of hundreds of conserved and nonconserved
human microRNAs. Nat Genet 37, 766-770.
Berezikov, E., van Tetering, G., Verheul, M., van de Belt, J., van Laake, L., Vos, J., Verloop, R.,
van de Wetering, M., Guryev, V., Takada, S., et al. (2006). Many novel mammalian
microRNA candidates identified by extensive cloning and RAKE analysis. Genome Res 16,
1289-1298.
Bohnsack, M.T., Czaplinski, K., and Gorlich, D. (2004). Exportin 5 is a RanGTP-dependent
dsRNA-binding protein that mediates nuclear export of pre-miRNAs. RNA 10, 185-191.
Breaker, R.R., Banerji, A., and Joyce, G.F. (1994). Continuous in vitro evolution of
bacteriophage RNA polymerase promoters. Biochemistry 33, 11980-11986.
Brummelkamp, T.R., Bernards, R., and Agami, R. (2002). A system for stable expression of
short interfering RNAs in mammalian cells. Science 296, 550-553.
Calin, G.A., Dumitru, C.D., Shimizu, M., Bichi, R., Zupo, S., Noch, E., Aldler, H., Rattan, S.,
Keating, M., Rai, K., et al. (2002). Frequent deletions and down-regulation of micro- RNA
genes miR15 and miR16 at 13q14 in chronic lymphocytic leukemia. Proc Natl Acad Sci U S
A 99, 15524-15529.
Calin, G.A., Ferracin, M., Cimmino, A., Di Leva, G., Shimizu, M., Wojcik, S.E., Iorio, M.V.,
Visone, R., Sever, N.I., Fabbri, M., et al. (2005). A MicroRNA signature associated with
prognosis and progression in chronic lymphocytic leukemia. N Engl J Med 353, 1793-1801.
Cheloufi, S., Dos Santos, C.O., Chong, M.M., and Hannon, G.J. (2010). A dicer-independent
miRNA biogenesis pathway that requires Ago catalysis. Nature 465, 584-589.
Chen, C.Z., Li, L., Lodish, H.F., and Bartel, D.P. (2004). MicroRNAs modulate hematopoietic
lineage differentiation. Science 303, 83-86.
Chiang, H.R., Schoenfeld, L.W., Ruby, J.G., Auyeung, V.C., Spies, N., Baek, D., Johnston,
W.K., Russ, C., Luo, S., Babiarz, J.E., et al. (2010). Mammalian microRNAs: experimental
evaluation of novel and previously annotated genes. Genes Dev 24, 992-1009.
Chung, W.J., Agius, P., Westholm, J.O., Chen, M., Okamura, K., Robine, N., Leslie, C.S., and
Lai, E.C. (2011). Computational and experimental identification of mirtrons in Drosophila
melanogaster and Caenorhabditis elegans. Genome Res 21, 286-300.
Cifuentes, D., Xue, H., Taylor, D.W., Patnode, H., Mishima, Y., Cheloufi, S., Ma, E., Mane, S.,
Hannon, G.J., Lawson, N.D., et al. (2010). A novel miRNA processing pathway independent
of Dicer requires Argonaute2 catalytic activity. Science 328, 1694-1698.
93
Denli, A.M., Tops, B.B., Plasterk, R.H., Ketting, R.F., and Hannon, G.J. (2004). Processing of
primary microRNAs by the Microprocessor complex. Nature 432, 231-235.
Duan, R., Pak, C., and Jin, P. (2007). Single nucleotide polymorphism associated with mature
miR-125a alters the processing of pri-miRNA. Hum Mol Genet 16, 1124-1131.
Ellington, A.D., and Szostak, J.W. (1990). In vitro selection of RNA molecules that bind specific
ligands. Nature 346, 818-822.
Fujita, P.A., Rhead, B., Zweig, A.S., Hinrichs, A.S., Karolchik, D., Cline, M.S., Goldman, M.,
Barber, G.P., Clawson, H., Coelho, A., et al. (2011). The UCSC Genome Browser database:
update 2011. Nucleic Acids Res 39, D876-882.
Gottwein, E., Cai, X., and Cullen, B.R. (2006). A novel assay for viral microRNA function
identifies a single nucleotide polymorphism that affects Drosha processing. J Virol 80, 53215326.
Gregory, R.I., Yan, K.P., Amuthan, G., Chendrimada, T., Doratotaj, B., Cooch, N., and
Shiekhattar, R. (2004). The Microprocessor complex mediates the genesis of microRNAs.
Nature 432, 235-240.
Grimson, A., Srivastava, M., Fahey, B., Woodcroft, B.J., Chiang, H.R., King, N., Degnan, B.M.,
Rokhsar, D.S., and Bartel, D.P. (2008). Early origins and evolution of microRNAs and Piwiinteracting RNAs in animals. Nature 455, 1193-1197.
Grishok, A., Pasquinelli, A.E., Conte, D., Li, N., Parrish, S., Ha, I., Baillie, D.L., Fire, A.,
Ruvkun, G., and Mello, C.C. (2001). Genes and mechanisms related to RNA interference
regulate expression of the small temporal RNAs that control C. elegans developmental
timing. Cell 106, 23-34.
Guil, S., and Caceres, J.F. (2007). The multifunctional RNA-binding protein hnRNP A1 is
required for processing of miR-18a. Nat Struct Mol Biol 14, 591-596.
Guo, H., Ingolia, N.T., Weissman, J.S., and Bartel, D.P. (2010). Mammalian microRNAs
predominantly act to decrease target mRNA levels. Nature 466, 835-840.
Han, B.W., Hung, J.H., Weng, Z., Zamore, P.D., and Ameres, S.L. (2011). The 3′-to-5′
Exoribonuclease Nibbler Shapes the 3′ Ends of MicroRNAs Bound to Drosophila
Argonaute1. Curr Biol.
Han, J., Lee, Y., Yeom, K.H., Kim, Y.K., Jin, H., and Kim, V.N. (2004). The Drosha-DGCR8
complex in primary microRNA processing. Genes Dev 18, 3016-3027.
Han, J., Lee, Y., Yeom, K.H., Nam, J.W., Heo, I., Rhee, J.K., Sohn, S.Y., Cho, Y., Zhang, B.T.,
and Kim, V.N. (2006). Molecular basis for the recognition of primary microRNAs by the
Drosha-DGCR8 complex. Cell 125, 887-901.
Han, J., Pedersen, J.S., Kwon, S.C., Belair, C.D., Kim, Y.K., Yeom, K.H., Yang, W.Y.,
Haussler, D., Blelloch, R., and Kim, V.N. (2009). Posttranscriptional crossregulation
between Drosha and DGCR8. Cell 136, 75-84.
Hofacker, I.L., and Stadler, P.F. (2006). Memory efficient folding algorithms for circular RNA
secondary structures. Bioinformatics 22, 1172-1176.
Hutvagner, G., McLachlan, J., Pasquinelli, A.E., Balint, E., Tuschl, T., and Zamore, P.D. (2001).
A cellular function for the RNA-interference enzyme Dicer in the maturation of the let-7
small temporal RNA. Science 293, 834-838.
Jazdzewski, K., Murray, E.L., Franssila, K., Jarzab, B., Schoenberg, D.R., and de la Chapelle, A.
(2008). Common SNP in pre-miR-146a decreases mature miR expression and predisposes to
papillary thyroid carcinoma. Proc Natl Acad Sci U S A 105, 7269-7274.
94
Jolma, A., Kivioja, T., Toivonen, J., Cheng, L., Wei, G., Enge, M., Taipale, M., Vaquerizas,
J.M., Yan, J., Sillanpaa, M.J., et al. (2010). Multiplexed massively parallel SELEX for
characterization of human transcription factor binding specificities. Genome Res 20, 861873.
Ketting, R.F., Fischer, S.E., Bernstein, E., Sijen, T., Hannon, G.J., and Plasterk, R.H. (2001).
Dicer functions in RNA interference and in synthesis of small RNA involved in
developmental timing in C. elegans. Genes Dev 15, 2654-2659.
Khvorova, A., Reynolds, A., and Jayasena, S.D. (2003). Functional siRNAs and miRNAs exhibit
strand bias. Cell 115, 209-216.
Kozomara, A., and Griffiths-Jones, S. (2011). miRBase: integrating microRNA annotation and
deep-sequencing data. Nucleic Acids Res 39, D152-157.
Lamontagne, B., and Elela, S.A. (2004). Evaluation of the RNA determinants for bacterial and
yeast RNase III binding and cleavage. J Biol Chem 279, 2231-2241.
Landgraf, P., Rusu, M., Sheridan, R., Sewer, A., Iovino, N., Aravin, A., Pfeffer, S., Rice, A.,
Kamphorst, A.O., Landthaler, M., et al. (2007). A mammalian microRNA expression atlas
based on small RNA library sequencing. Cell 129, 1401-1414.
Landthaler, M., Yalcin, A., and Tuschl, T. (2004). The human DiGeorge syndrome critical
region gene 8 and Its D. melanogaster homolog are required for miRNA biogenesis. Curr
Biol 14, 2162-2167.
Lee, Y., Ahn, C., Han, J., Choi, H., Kim, J., Yim, J., Lee, J., Provost, P., Radmark, O., Kim, S.,
et al. (2003). The nuclear RNase III Drosha initiates microRNA processing. Nature 425, 415419.
Lee, Y., Jeon, K., Lee, J.T., Kim, S., and Kim, V.N. (2002). MicroRNA maturation: stepwise
processing and subcellular localization. EMBO J 21, 4663-4670.
Lee, Y., Kim, M., Han, J., Yeom, K.H., Lee, S., Baek, S.H., and Kim, V.N. (2004). MicroRNA
genes are transcribed by RNA polymerase II. EMBO J 23, 4051-4060.
Lee, Y., and Kim, V.N. (2007). In vitro and in vivo assays for the activity of Drosha complex.
Methods Enzymol 427, 89-106.
Lim, L.P., Glasner, M.E., Yekta, S., Burge, C.B., and Bartel, D.P. (2003a). Vertebrate
microRNA genes. Science 299, 1540.
Lim, L.P., Lau, N.C., Weinstein, E.G., Abdelhakim, A., Yekta, S., Rhoades, M.W., Burge, C.B.,
and Bartel, D.P. (2003b). The microRNAs of Caenorhabditis elegans. Genes Dev 17, 9911008.
Liu, J., Carmell, M.A., Rivas, F.V., Marsden, C.G., Thomson, J.M., Song, J.J., Hammond, S.M.,
Joshua-Tor, L., and Hannon, G.J. (2004). Argonaute2 is the catalytic engine of mammalian
RNAi. Science 305, 1437-1441.
Lund, E., Guttinger, S., Calado, A., Dahlberg, J.E., and Kutay, U. (2004). Nuclear export of
microRNA precursors. Science 303, 95-98.
Macrae, I.J., Zhou, K., Li, F., Repic, A., Brooks, A.N., Cande, W.Z., Adams, P.D., and Doudna,
J.A. (2006). Structural basis for double-stranded RNA processing by Dicer. Science 311,
195-198.
Meister, G., Landthaler, M., Patkaniowska, A., Dorsett, Y., Teng, G., and Tuschl, T. (2004).
Human Argonaute2 mediates RNA cleavage targeted by miRNAs and siRNAs. Mol Cell 15,
185-197.
Michlewski, G., Guil, S., Semple, C.A., and Caceres, J.F. (2008). Posttranscriptional regulation
of miRNAs harboring conserved terminal loops. Mol Cell 32, 383-393.
95
Moore, M.J. (1999). Joining RNA molecules with T4 DNA ligase. Methods Mol Biol 118, 1119.
Okamura, K., Hagen, J.W., Duan, H., Tyler, D.M., and Lai, E.C. (2007). The mirtron pathway
generates microRNA-class regulatory RNAs in Drosophila. Cell 130, 89-100.
Paddison, P.J., Caudy, A.A., Bernstein, E., Hannon, G.J., and Conklin, D.S. (2002). Short hairpin
RNAs (shRNAs) induce sequence-specific silencing in mammalian cells. Genes Dev 16,
948-958.
Pan, T., and Uhlenbeck, O.C. (1992). In vitro selection of RNAs that undergo autolytic cleavage
with Pb2+. Biochemistry 31, 3887-3895.
Park, J.E., Heo, I., Tian, Y., Simanshu, D.K., Chang, H., Jee, D., Patel, D.J., and Kim, V.N.
(2011). Dicer recognizes the 5′ end of RNA for efficient and accurate processing. Nature
475, 201-205.
Pitman, J. (1993). Probability (New York, Springer-Verlag).
Pitt, J.N., and Ferre-D'Amare, A.R. (2010). Rapid construction of empirical RNA fitness
landscapes. Science 330, 376-379.
Ruby, J.G., Jan, C.H., and Bartel, D.P. (2007). Intronic microRNA precursors that bypass Drosha
processing. Nature 448, 83-86.
Schurer, H., Lang, K., Schuster, J., and Morl, M. (2002). A universal method to produce in vitro
transcripts with homogeneous 3′ ends. Nucleic Acids Res 30, e56.
Schwarz, D.S., Hutvagner, G., Du, T., Xu, Z., Aronin, N., and Zamore, P.D. (2003). Asymmetry
in the assembly of the RNAi enzyme complex. Cell 115, 199-208.
Sempere, L.F., Cole, C.N., McPeek, M.A., and Peterson, K.J. (2006). The phylogenetic
distribution of metazoan microRNAs: insights into evolutionary complexity and constraint. J
Exp Zool B Mol Dev Evol 306, 575-588.
Slattery, M., Riley, T., Liu, P., Abe, N., Gomez-Alcala, P., Dror, I., Zhou, T., Rohs, R., Honig,
B., Bussemaker, H.J., et al. (2011). Cofactor binding evokes latent differences in DNA
binding specificity between Hox proteins. Cell 147, 1270-1282.
Sun, G., Yan, J., Noltner, K., Feng, J., Li, H., Sarkis, D.A., Sommer, S.S., and Rossi, J.J. (2009).
SNPs in human miRNA genes affect biogenesis and function. RNA 15, 1640-1651.
Trabucchi, M., Briata, P., Garcia-Mayoral, M., Haase, A.D., Filipowicz, W., Ramos, A., Gherzi,
R., and Rosenfeld, M.G. (2009). The RNA-binding protein KSRP promotes the biogenesis of
a subset of microRNAs. Nature 459, 1010-1014.
Viswanathan, S.R., Daley, G.Q., and Gregory, R.I. (2008). Selective blockade of microRNA
processing by Lin28. Science 320, 97-100.
Warf, M.B., Johnson, W.E., and Bass, B.L. (2011). Improved annotation of C. elegans
microRNAs by deep sequencing reveals structures associated with processing by Drosha and
Dicer. RNA 17, 563-577.
Witten, D., Tibshirani, R., Gu, S.G., Fire, A., and Lui, W.O. (2010). Ultra-high throughput
sequencing-based small RNA discovery and discrete statistical biomarker analysis in a
collection of cervical tumours and matched controls. BMC Biol 8, 58.
Yeom, K.H., Lee, Y., Han, J., Suh, M.R., and Kim, V.N. (2006). Characterization of
DGCR8/Pasha, the essential cofactor for Drosha in primary miRNA processing. Nucleic
Acids Res 34, 4622-4629.
Yi, R., Qin, Y., Macara, I.G., and Cullen, B.R. (2003). Exportin-5 mediates the nuclear export of
pre-microRNAs and short hairpin RNAs. Genes Dev 17, 3011-3016.
96
Zeng, Y., and Cullen, B.R. (2005). Efficient processing of primary microRNA hairpins by
Drosha requires flanking nonstructured RNA sequences. J Biol Chem 280, 27595-27603.
Zeng, Y., Yi, R., and Cullen, B.R. (2005). Recognition and cleavage of primary microRNA
precursors by the nuclear processing enzyme Drosha. EMBO J 24, 138-148.
Zhang, H., Kolb, F.A., Jaskiewicz, L., Westhof, E., and Filipowicz, W. (2004). Single processing
center models for human Dicer and bacterial RNase III. Cell 118, 57-68.
Zhang, K., and Nicholson, A.W. (1997). Regulation of ribonuclease III processing by doublehelical sequence antideterminants. Proc Natl Acad Sci U S A 94, 13437-13441.
Zhang, X., and Zeng, Y. (2010). The terminal loop region controls microRNA processing by
Drosha and Dicer. Nucleic Acids Res 38, 7689-7697.
Zykovich, A., Korf, I., and Segal, D.J. (2009). Bind-n-Seq: high-throughput analysis of in vitro
protein-DNA interactions using massively parallel sequencing. Nucleic Acids Res 37, e151.
97
0.0
cel-lsy-6
X
cel-mir-50
Query pri-miRNA
cel-mir-40
B
cel-lin-4
Unprocessed
pri-miRNA X
Gppp
hsa-mir-1-1
11
oc (o
n
m k1 ly)
oc
lin k2
-4
ls
y6
m
ir4
m 0
ir5
m 0
ir2
le 40
t-7
m
ir1
m
ir2
m
ir3
m 4
ir4
m 3/4
ir- 4
4
m 6
ir5
m 9
ir6
m 0
ir1
m 24
ir23
5
ir-
m
m
CMV Promoter
Relative pri-miRNA
ir1
ir2a
m -1
ir2
le 86
t-7 /4
/5
m
ir3
m 4
ir9
m 2
ir12
m 5
ir2
m 79
ir2
le 81
t-7
m a
ir9
m 2a
ir2
m 05
ir1
m 7~
ir- 20
1
m 25
ir- a
1
m 28
ir- -1
1
m 38
ir- -2
1
m 22
ir1
m 33
ir- a14 1
2
m
m
A
Control
pri-miRNA
hsa-mir-1-1
AAAAA
dme miRNAs
miR-1 mature
miR-1-1 pre
U6 snRNA
Glu tRNA
2.5
2.0
1.5
1.0
0.5
TK pA
cel miRNAs
miR-1 mature
miR-1-1 pre
U6 snRNA
Glu tRNA
hsa miRNAs
Supplemental Materials
Supplemental Figures
Figure S1. Human, fly, and worm pri-miRNA transcription in human cells, related to Figure 1.
(A) Schematic of HEK293 overexpression system and detection of miR-1-1 processing products.
HEK293 cells were individually transfected with plasmids bearing a human, D. melanogaster, or C.
elegans pri-miRNAs transcriptionally fused to human pri-mir-1-1. Mature miR-1 and pre-miR-1-1
derived from the transcriptional fusion were detected by RNA blot. Results from vectors in which let-7
and mir-1 were the query pri-miRNAs are shown. However, results from these let-7 and mir-1 vectors
are not shown in Figure 1A because the corresponding mature miRNAs were indistinguishable from
those of other transfected vectors after total RNA was pooled for small-RNA sequencing.
(B) Direct detection of C. elegans pri-miRNA transcription. Selected human and C. elegans pri-miRNAs
were detected by a ribonuclease protection assay and signals were normalized to that of neomycin
phosphotransferase mRNA expressed from the same plasmid.
99
A
Basal Stem
mir125.wt
(Wildtype)
mir125.1
mir125.2
UCU
GCCAG
CUAGG
CGGUC
GGU
CCAACC
UGC
pre-miRNA
CAUGAA
UCU
GCCAG
CUAGG
CGGUC
GGU
CCAACC
UGC
pre-miRNA
CAUGCC
UCU
GCCAG
CUAGG
CGGUC
GGU
CCAACC
UGC
pre-miRNA
UCU
GGGCCAG
CUAGG
CCCGGUC
GGU
CCAA
UGC
pre-miRNA
mir125.4
CAUGUUA
UCU
CCAG
CUAGG
GGUC
GGU
CCAACCC
UGC
pre-miRNA
mir125.5
CAUGUUAA
UCU
CAG
CUAGG
GUC
GGU
CCAACCCG
UGC
pre-miRNA
mir125.6
CAUGUUAAG UCU
AG
CUAGG
UC
GGU
CCAACCCGG UGC
pre-miRNA
mir125.3
B
CAUGUU
CAUG
Control miRNA
(WT mir-125a)
C
Query
miRNA
*P
?
*P
?
1
CMV Promoter
3
4
5
AAAAA
mir-125a mutant
wt
2
Control
miRNA
?
Gppp
mir-125a
mutants
1
2
3
hsa-mir-1-1
4
5
TK pA
6
605040-
Competitive Cleavage
Prediction from mir-125a selection (1 mM Mg)
wt
?
6
30-
1
hsa-miR-125a
20-
hsa-miR-1
0.25
0.125
0.0625
0.03125
1
Relative expression
Relative Cleavage
0.5
0.5
0.25
wt
1
2
3
4
5
6
Figure S2. Confirmation of hsa-mir-125a selection results in vitro and in HEK293T cells, related to Figure
2.
(A) Predicted basal stem secondary structure of mir-125a variants tested in the experiment.
(B) Competitive cleavage of individual mir-125a variants, relative to wildtype mir-125a. Variants were
mixed with a reference wildtype mir-125a, which was longer at its 5′ end, and incubated in whole-cell
lysate from HEK293T cells overexpressing Drosha and DGCR8. Cleavage products were separated
on denaturing gels, and the ratio of wildtype and variant products quantified (blue, geometric mean ±
standard error, n = 3), together with the relative cleavage inferred from the selection experiment
(gray).
(C) Evaluation of mir-125a variants in HEK293T cells. Variants were transcriptionally fused to pri-mir-1-1
as in Figure S1A and expressed in HEK293T cells. Accumulation of mature miR-125a was quantified
by RNA blot and normalized to the level of mature miR-1 (geometric mean ± standard error, n = 3).
101
A
Position 12
C
G
A
0.29
C
–1.14 –1.57 0.99
Pair 12
A
0.1
–0.2
1.65
U
–0.32 –0.06
–1.85
–0.16 –0.02
–0.08 –0.64 –0.97 –0.78
A
1.48
Position 13
C
G
0.06
0.54
U
–0.67
C
1.03
–0.34 –1.01 –1.11
G
0.99
–0.94 –0.04 –1.31
U
1.79
1.61
1.48
Position –12
A
–1.82 –1.77 –0.91 0.24
C
–1.99 –2.18 0.69 –1.51
G
–0.94 1.00 –1.41 0.23
U
0.47 –0.21 0.12 –0.13
hsa-mir-223
Wildtype basal stem
–1 C
U
C-G +1
G-U
C-G
A
C
C-G
C-G
G-C
–10 U-A +8
–11 G-C +9
–12 N-N +10
–13 N-N +11
–14 N-N +12
N +13
–15 N
N
N
N
N
N
N
N
N
B
hsa-mir-125a
1024
256
256
64
64
16
1
0
0.5
1
0
Threshold score
hsa-mir-30a
1024
256
256
64
64
16
16
4
4
1
0
0.5
Threshold score
1
1
0
C
–1.25 –0.83
1.62
–0.89
G
–0.51
1.74
–0.16
0.78
U
0.86
–0.27
0.35
–0.37
A
Position 11
C
G
U
A
–1.36 –1.00 –1.21 –0.05
C
–1.58
G
–1.44 0.75
U
–0.48 –1.51 –1.25 –0.81
A
–1.10 –0.14 0.68
–1.63
Position 12
C
G
A
–0.59 –0.53 0.47
0.2
U
–1.08
C
–0.71
G
–1.03 –1.11 –0.34 –1.59
U
0.64
–0.70 –0.11 –1.25
0.76
1.82
0.13
Timepoint 2
2
1
9
10
11
12
Basal stem pairs
0.5
1
Threshold score
hsa-mir-223
1024
0.8
Timepoint 1
4
1
U
–0.15
4
16
Basal stem
4
0.55
8
hsa-mir-16-1
1024
Position 10
C
G
hsa-mir-223
Alternative pairs
16
0.5
C
–0.48
Pair 12
0.73
A
A
Pair 11
hsa-mir-223
Alternative basal stem
Relative cleavage
G
U
Position –14
A
U
C-G +1
G-U
C-G
A
C
C-G
C-G
G-C
–10 U-A +8
–11 G-C +9
| A +10
–12 A-U +11
–13 C-G +12
–14 G-C +13
U
U
C
U
C
A
U
C
Candidate pairs
Position –13
Pair 11
–1 C
0.45
U
Position –13
1.93
A
Pair 10
Position –14
0.23
Position –11
1.26
Candidate pairs
Position –12
U
pre-miRNA
Pair 9
1
–0.21 –0.21
Odds ratio (log2)
–2.23 0.49
Position 9
C
G
0.8
1.64
–1.69 –1.19 –0.98
0.6
1.73
–0.9
1.67
0.4
–1.22 –1.08
U
0.2
–1.24
C
G
–1.32
0
0.69
0.00
−0.2
–2.26 –1.79 –1.39
–1.22 –1.78 0.22
–1.47
pre-miRNA
U
A
C
G
−0.4
Position 11
C
G
U
–2.44 –0.59 –2.89 0.33
−0.6
A
Position 8
C
G
−0.8
Pair 10
A
A
−1
Position –10
Pair 8
0.5
Threshold score
1
13
Figure S3. Analysis of mir-223 basal stem structure, related to Figure 3.
(A) Wild-type (left) and alternative (right) basal stem structures for hsa-mir-223. In the wild-type basal
stem, the A at +10 is predicted to be bulged, whereas some variants are predicted to have an
alternative structure in which the nucleotide at +10 is Watson–Crick paired within a contiguous helix.
Covariation matrices for both conformations were calculated as in Figure 3A.
(B) Relative cleavage of variants with different lengths of the alternative basal stem. Cleavage values
were calculated as in Figure 3B and normalized to the 9 bp stem.
(C) Screen for Watson–Crick pairs involving any two varied positions. A Watson–Crick-pairing score was
calculated for each of the >3000 possible pairs of varied positions in each of the four pri-miRNAs.
The number of Watson–Crick candidates is plotted as a function of threshold score, in which a pair is
considered a Watson–Crick candidate if its score exceeds the threshold. The number of pairs
corresponding to the basal stem is shown (dashed line).
103
A
Linear pri-miRNA substrate
(pool of variants)
ppp
DroshaTN
DGCR8
ppp
ppp
Nonfunctional variants
Functional variants
Nitrocellulose
filtration
X
Library preparation for
paired-end sequencing
B
1
A
0.8
G
hsa-mir-125a
C
U
–1
0.4
0
+1
–14 +12
Invariant
residues
0.2
43
45
45
41
39
37
Stem-loop
–1
Cleavage selection
0.8
35
33
31
29
27
25
23
21
19
17
15
13
–15
–17
–19
–21
–23
–25
–27
–29
–31
hsa-mir-125a
1
43
1.2
–33
–35
–37
–39
–41
–43
–45
–47
–49
–0.2
–0.4
–9
0.6
+1
+9
–14 +12
0.4
Invariant
residues
0.2
0
41
39
37
35
33
31
29
27
25
23
21
19
17
15
13
9
11
–9
–11
–13
–15
–17
–19
–21
–23
–25
–27
–29
–31
–33
–35
–37
–39
–41
–43
–45
–47
–0.4
–49
–0.2
–51
Information content (bits)
Stem-loop
TN binding selection
Round 1
0.6
–51
Information content (bits)
1.2
Figure S4. Selection for Microprocessor-binding variants of hsa-mir-125a, related to Figure 4.
(A) Schematic of the in vitro selection. Linear, partially-randomized variants of mir-125a were incubated
with immunopurified DGCR8 and catalytically-inactive Drosha (DroshaTN). Bound variants were
recovered by nitrocellulose filtration, reverse-transcribed, and amplified for high-throughput
sequencing.
(B) Information content after one round of selection for Microprocessor binding. Information content was
calculated as in Figure 2D. Information content after one round of cleavage selection (Figure 2D) is
reproduced here for comparison. The nucleotides varied in the initial pools are shown for each
selection (insets, red inner lines).
105
A
Timepoint 1
Odds ratio of CNNC
Timepoint 2
hsa-mir-16-1
16
16
8
8
4
4
2
2
1
1
0.5
0.5
Odds ratio of CNNC
0.25
hsa-mir-30a
Odds ratio of CNNC
14 15 16 17 18 19 20 21
0.25
8
8
4
4
2
2
1
1
0.5
0.5
0.25
hsa-mir-223
Sequences without
wildtype CNNC
All Sequences
14 15 16 17 18 19 20 21
0.25
8
8
4
4
2
2
1
1
0.5
0.5
0.25
14 15 16 17 18 19 20 21
0.25
NA
NA
14 15 16 17 18 19 20 21
NA
NA
NA
NA
?
?
Position
Control
pri-miRNA
Gppp
mir-30a mutant
CMV Promoter
Construct
Basal Stem
mir30.wt
mir30.1
mir30.2
mir30.3
mir30.4
mir30.5
mir30.6
mir30.7
WT
–12 Paired
WT
WT
WT
WT
WT
7-pair stem
mir30.wt
AAAAA
hsa-mir-1-1
TK pA
UG(–14) CNNC(+17)
UG
UG
AG
UG
UG
UG
AG
UG
CNNC
CNNC
CNNC
CNNU
UNNC
UNNU
UNNU
CNNC
hsa-miR-30a probe binding site
UGUU A
A
A
UC
GUGAAG
G CAGUG GCG CUGUAAACAUCC GACUGGAAGCU
C
C GUCAU CGU GACGUUUGUAGG CUGACUUUCGG
C
GGCU C
C
C
-GUAGACA
mir30.1
UGUU
A
A
UC
UGUGAAG
GCCAGUG GCG CUGUAAACAUCC GACUGGAAGC
C
CGGUCAU CGU GACGUUUGUAGG CUGACUUUCG
C
C
C
-GGUAGACA
GGCU
mir30.7
UGUUCCGC
NA
14 15 16 17 18 19 20 21
A
A
UC
UGUGAAG
GUG GCG CUGUAAACAUCC GACUGGAAGC
C
CAU CGU GACGUUUGUAGG CUGACUUUCG
C
GGCUCCGU
C
C
-GGUAGACA
wt 1
70605040-
2
3
4
5
6
7
30hsa-miR-30a
20hsa-miR-1-1
hsa-miR-30a expression
Query
pri-miRNA
NA
14 15 16 17 18 19 20 21
Position
B
NA
2
1
0.5
0.25
0.125
0.063
0.031
wt 1
2
3
4
5
6
7
Figure S5. Contribution of the CNNC motif in vitro and in HEK293T cells, related to Figure 5.
(A) CNNC odds ratios at alternative positions. Odds ratios were calculated for CNNC dinucleotides
starting at the indicated of positions downstream of the Drosha cleavage site. Plotted are odds ratios
for all sequences (left) and for sequences that lack both wild-type C residues (right).
(B) Contributions of basal features, including the CNNC motif, to the accumulation of hsa-miR-30a in
HEK293T cells. The listed variants of hsa-mir-30a were transcriptionally fused to hsa-mir-1-1 (left).
Predicted secondary structures for variants with non-wild-type structure are shown (center), with the
annotated Drosha cleavage sites (purple arrowheads). The accumulation of miR-30a was quantified
by RNA blot, normalized to miR-1 (right, geometric mean ± standard error, n = 3).
107
B
Pair 18
P1
0.8
CCA
0.6
G
0.4
A
0.2
A
G
0
U
G
P24 U-G P40
P23 C-G P41
P22 G-C P42
P21 A-U P43
P20 A-U P44
17th Pair
G-U
G-C
U-A
C-G
A-U
G-C
C |
U |
hsa-mir-30a
C-G
pre-miRNA
C-G
U-A
A-U
C-G
A-U
A-U
A-U
U-G
G-C
P1 U-A
G
C P63
0.4
0.2
0
P44
P42
P40
P38
P36
P34
P32
P30
P28
P26
P24
P22
-0.2
0.8
0.6
0.4
0.2
0
P44
P42
P40
P38
P36
P34
P32
P30
P28
P26
-0.4
P24
-0.2
P22
hsa-mir-30a
1
P20
Information content (bits)
1.2
0.11
C
–0.11
–0.1
–0.54
–0.15
G
0.11
–0.05
–0.08
–0.24
U
0
0.27
–0.05
–0.21
A
Position P43
C
G
U
Position P21
0.05
–0.31
0.03
C
0.06
–0.29
–0.17
G
0.3
0.23
0.14
0.24
U
0.39
0.5
–0.02
–0.06
A
Candidate pairs
Position P24
0.2
0
0.5
Threshold score
A
–0.12
0.22
0.4
C
–0.32
–0.54
–0.01
–0.41
0.2
G
–0.28
0.15
0.01
0.42
0
U
0.18
–0.14
0.08
–0.41
−0.2
A
Position P41
C
G
U
A
–0.48
–0.41
–0.7
0.54
C
–0.49
–0.66
0.47
–0.61
G
–0.04
0.37
-0.06
0.44
U
0.44
–0.29
0.07
–0.13
A
A
–0.61
−0.6
−0.8
−1
Position P40
C
G
U
–0.5
–0.66
0.04
C
–0.42 –0.41 –0.07 –0.66
G
–0.17 –0.06 –0.49 –0.11
U
0.03
–0.21
0.32
–0.09
P46
P44
P42
P40
P38
P36
P34
P32
P28
P26
P24
P30
hsa-mir-16-1
Apical Stem
0
0.6
–0.25
-0.2
hsa-mir-125a
256
128
64
32
16
8
4
2
1
0.8
Position P42
C
G
U
–0.41
Pair 21
0.4
-0.4
C
1
−0.4
0.6
P22
Information content (bits)
hsa-mir-223
0.16
–0.13
Pair 22
0.8
–0.13
–0.15
1.2
1
–0.04
A
Pair 20
Position P22
0.6
Position P44
C
G
U
A
Pair 19
Position P23
P42
P40
P38
P36
P34
P32
P30
P28
P26
P24
P22
P20
0.8
P20
Information content (bits)
1
-0.4
A
G
1.2
hsa-mir-16-1
C
G
A
U
-0.2
-0.4
Position P20
1
P18
hsa-mir-125a
Information content (bits)
1.2
A
Odds ratio (log2)
A
1
128
64
32
16
8
4
2
1
0
0.5
Threshold score
hsa-mir-30a
1
128
64
32
16
8
4
2
1
0
0.5
Threshold score
hsa-mir-223
1
128
64
32
16
8
4
2
1
0
0.5
Threshold score
1
Figure S6. Primary-sequence and structural elements in the apical stem and terminal loop, related to
Figure 6.
(A) Enrichment and depletion at variable residues in pri-miRNA variants selected from a pool with varied
nucleotide identities in the apical stem and loop. At each varied position (inset, red inner line),
information content was calculated for each residue (green, cyan, black, and red for A, C, G, and U,
respectively), as in Figure 2D.
(B) Apical stem secondary structure of mir-30a. Predicted secondary structure and covariation matrices
for were calculated as in Figure 3A.
(C) Screen for Watson–Crick pairs involving any to varied positions of the apical stem and loop. A
Watson–Crick pairing score was calculated as in Figure S3C for each of the >275 possible pairs of
varied positions in each of the four pri-miRNAs. The number of pairs corresponding to the apical
stem is shown (dashed line).
109
A
Query
pri-miRNA ?
Control
pri-miRNA
?
Gppp
CMV Promoter
mir-50 mutant
AAAAA
hsa-mir-1-1
TK pA
Construct
Basal Stem
UG(–14)
CNNC
mir50.wt
WT
UG
None
mir50.1
11-pair stem
AC
None
mir50.2
11-pair stem
UC
None
40-
mir50.3
11-pair stem
UC
CNNC(+19)
30-
mir50.4
11-pair stem
UC
CNNC(+18)
mir50.5
11-pair stem
UC
CNNC(+17)
mir50.6
11-pair stem
UC
CNNC(+16)
mir50.7
11-pair stem
UC
CNNC(+15)
mir50.8
11-pair stem
UG
None
mir50.9
11-pair stem
UG
CNNC(+18)
wt 1
3
4
5
6
7
8
9
cel-miR-50
20-
CUG
C
UCU
G
UUC
UAUU CCUG CCCGCCGGCCG UGAUAUGUCUGGUAU
UGGGUUU AAC
\
GUAA GGAC GGGCGGCCGGC GCUAUGCAGAUUAUA
GCCCAAG UUG
C
UCG
U
G
A
C-CGA
cel-miR-50 relative expression
hsa-miR-1
cel-miR-50 Probe Binding Site
mir50.wt
2
706050-
64
32
16
8
4
2
0
0.5
0.25
wt 1
2
3
4
5
6
7
8
9
mir50.2
CUGUAUUCCUC
UCU
G
UUC
CCCCGCCGGCCG UGAUAUGUCUGGUAU
UGGGUUU AAC
\
GGGGCGGCCGGC GCUAUGCAGAUUAUA
GCCCAAG UUG
C
UCGGUAAU--AC
A
C-CGA
Query
pri-miRNA ?
Control
pri-miRNA
?
Gppp
CMV Promoter
mir-40 mutant
AAAAA
hsa-mir-1-1
TK pA
Construct
Basal Stem
UG(–14)
CNNC
mir40.wt
WT
CC
None
None
mir40.1
(mutant)
UG
mir40.2
WT reverted
UG
None
mir40.3
WT reverted
UG
CNNC(+16)
mir40.4
WT reverted
UG
CNNC(+17)
mir40.5
WT reverted
UG
CNNC(+18)
mir40.6
WT reverted
UG
CNNC(+19)
mir40.7
WT reverted
UG
CNNC(+20)
mir40.wt
GUCUC
CCU-
-C
G
A A
A
AUC
CCUGU CCGCACCU AGU GAUGUAUGCC UG UGAUA GAU
\
GGACA GGCGUGGA UCG CUACAUGUGG GC ACUAU CUA
A
ACGU
UG
A
A
- C
C
AAG
CGAA
GUUU
A----
cel-miR-40 Probe Binding Site
mir40.1
GUCUC
CCU- G
-C
G
A A
A
AUC
U UGU CCGCACCU AGU GAUGUAUGCC UG UGAUA GAU
\
G ACA GGCGUGGA UCG CUACAUGUGG GC ACUAU CUA
A
ACGU G
UG
A
A
- C
C
AAG
CGAA
GUUU
A---mir40.2
GUCUC
CCU-
-C
G
A A
A
AUC
UGUGU CCGCACCU AGU GAUGUAUGCC UG UGAUA GAU
\
ACACA GGCGUGGA UCG CUACAUGUGG GC ACUAU CUA
A
ACGU
UG
A
A
- C
C
AAG
A----
CGAA
GUUU
706050-
wt 1
2
3
4
5
6
7
4030cel-miR-40
20hsa-miR-1
cel-miR-40 relative expression
B
8
4
2
1
0.5
wt 1
2
3
4
5
6
7
Figure S7. Rescue of C. elegans pri-miRNA processing in human cells, related to Figure 7.
(A) Effects of adding human pri-miRNA motifs to C. elegans mir-50. Changes that introduced the listed
features were incorporated into mir-50 within the bicistronic expression vector (left). Secondary
structures are shown for changes that were predicted to affect the wild-type basal stem (middle), with
the annotated Drosha cleavage sites (purple arrowheads). After transfection into HEK293T cells,
accumulation of miR-50 was assessed on an RNA blot, normalizing to the accumulation of the miR-1
control, and increased miR-50 expression is plotted (geometric mean ± standard error, n = 3, right).
(B) Effects of adding human pri-miRNA motifs to C. elegans mir-40, otherwise as in (A).
111
Supplemental Table S1: Oligonucleotides used in the in vitro
selections
Related to the Experimental Procedures. Degenerate and conventional oligonucleotides
were commercially synthesized (IDT). Oligonucleotides are DNA unless otherwise noted.
HDV 5′ Polishing Primer
hsa-mir-125a circular selection
CTTCTCCCTTAGCCTACCGAAGTAGCCCAGG
Note: Underlined bases were degenerate.
S125circ.004 T7 Adaptor
CAGAGATGCATAATACGACTCACTATAGGGTCACAG
S125circ.001a Left arm
S125circ.003 HDV adaptor
GACTCACTATAGGGTCACAGGTGAGGTTCTTGGGAGCCTGGCGTCTGGCCCA
ACCACACACCTGGGGAATTGCTGGCCTGACTTCTGACCCCTGACTCCT
TCCTCACAGGTTAAAGGGTCTCAGGGACCTAGAGACTGGCAACATGGTGTGC
GGTGGCCCGGTAGACCCTGGGGTGGGGGTATGAGGAGTCAGGGGTCAG
CTTCTCCCTTAGCCTACCGAAGTAGCCCAGGTCGGACCGCGAGGAGGTGGAG
ATGCCATGCCGACCCTGGATGTCCTCACAGGTTAAAGG
S125circ.006 3p Arm Splint
AGACGCCAAGATCGGA
S125circ.007 5p Arm Splint
ACGTGTACCCTAGAGA
S125circ.009 Ref RT
TGGATGTCCTCACAGGTTAAAGGGTCTCAGGGACCTAG
S125circ.014 Ref-II Fwd Primer
S125circ.015 Ref-II Rev Primer
CTTTCCCTACACGACGCTCTTCCGATCTCAGGTGAGGTTCTTGGGAGCCTGG
C
GCATTCCTGCTGAACCGCTCTTCCGATCTTTAAAGGGTCTCAGGGACCTAGA
G
hsa-mir-16-1 circular selection
Note: Underlined bases were degenerate.
S16-1circ.004 T7 Adaptor
S16-1circ.003 HDV Adaptor
CAGAGATGCATAATACGACTCACTATACTAAAATTATCTCCAGTATTAACTG
TGC
ATACTAAAATTATCTCCAGTATTAACTGTGCTGCTGAAGTAAGGTTGACCAT
ACTCTACAGTTGTGTTTTAATGTATATTAATGTTGCTTAATTAAGGAC
ACCCAATCTTAACGCCAATATTTACGTGCTGCTAAGGCACTGCTGACATTGC
TATCATAAGAGCTATGAATAAAAAGAAATATGTCCTTAATTAAGCAAC
GCCTACCGAAGTAGCCCAGGTCGGACCGCGAGGAGGTGGAGATGCCATGCCG
ACCCAATCTTAACGCCAATATTTAC
S16-1circ.006 3p Arm Splint
AACCTTACAGATCGGA
S16-1circ.006b 3p Arm Splint Variant
AACCTTACTAGATCGGA
S16-1circ.007 5p Arm Splint
ACGTGTACAGGCACTG
S16-1circ.009 Ref RT Primer
AATCTTAACGCCAATATTTACGTGCTGCTAAGGC
S16-1circ.010 Ref Fwd Primer
CTTTCCCTACACGACGCTCTTCCGATCTTCCAGTATTAACTGTGCTGCTGA
S16-1circ.011 Ref Rev Primer
GCATTCCTGCTGAACCGCTCTTCCGATCTCCAATATTTACGTGCTGCTA
hsa-mir-30a circular selection
Note: Underlined bases were degenerate.
S30acirc.004 T7 Adaptor
CAGAGATGCATAATACGACTCACTATAGCCACAGATGGGCTTTCAGTCGG
S30acirc.001a Left Arm
S30a.003 HDV Adaptor
CTATAGCCACAGATGGGCTTTCAGTCGGATGTTTGCAGCTGCCTACTGCCTC
GGACTTCAAGGGGCTACTTTAGGAGCAATTATCTTGTTAATTAAGGTT
CGACCCTTCACAGCTTCCAGTCGAGGATGTTTACAGTCGCTCACTGTCAACA
GCAATATACCTTCTTTAGCCTTCTGTTGGGTTAACCTTAATTAACAAG
GCCTACCGAAGTAGCCCAGGTCGGACCGCGAGGAGGTGGAGATGCCATGCCG
ACCCTTCACAGCTTCCAGTCGAGG
S30acirc.006 3p Arm Splint
AGTAGGCAAGATCGGA
S30acirc.007 5p Arm Splint
ACGTGTACGTCGCTCA
S125circ.002a Right arm
S16-1circ.001a Left Arm
S16-1circ.002a Right Arm
S30acirc.002a Right Arm
112
S30acirc.009 Ref RT Primer
CCGATCTTTCCAGTCGAGGATGTTTACAGTCGC
S30acirc.010 Ref Fwd Primer
S30acirc.011 Ref Rev Primer
CTTTCCCTACACGACGCTCTTCCGATCTGGGCTTTCAGTCGGATGTTTGCAG
CTGC
GCATTCCTGCTGAACCGCTCTTCCGATCTTTCCAGTCGAGGATGTTTACAGT
CGC
hsa-mir-223 circular selection
Note: Underlined bases were degenerate.
S223circ.004 T7 Adaptor
CAGAGATGCATAATACGACTCACTATAGGTAGAGTGTCAGTTTGTC
S223circ.001a Left Arm
S223.003 HDV Adaptor
ACTATAGGTAGAGTGTCAGTTTGTCAAATACCCCAAGTGCGGCACATGCTTA
CCAGCTCTAGGCCAGGGCAGATGGGATATGACGAATTTAATTAAGATC
ACATGGAGTGTCCAACTCAGCTTGTCAAATACACGGAGCGTGGCACTGCAGG
AGGCCAGGCCAAGAGCTTCTGTGGGGAAGTGAGATCTTAATTAAATTC
CTTCTCCCTTAGCCTACCGAAGTAGCCCAGGTCGGACCGCGAGGAGGTGGAG
ATGCCATGCCGACCCACATGGAGTGTCCAACTCAGC
S223circ.006 3p Arm Splint
GCCGCACTAGATCGGA
S223circ.007 5p Arm Splint
ACGTGTACGAGCGTGG
S223circ.009 Ref RT Primer
ACATGGAGTGTCCAACTCAGCTTGTCAAATACAC
S223circ.010 Ref Fwd Primer
S223circ.011 Ref Rev Primer
CTTTCCCTACACGACGCTCTTCCGATCTGTCAGTTTGTCAAATACCCCAAGT
G
GCATTCCTGCTGAACCGCTCTTCCGATCTACTCAGCTTGTCAAATACACGGA
GC
Common oligonucleotides for circular
cleavage selection
Note: “p” indicates a phosphate and lowercase
letters denote RNA
S0circ.005 3p Arm Adaptor
S0circ.007B.007.A 5p Arm Adaptor, CAT
barcode
S0circ.007B.007.B 5p Arm Adaptor, ATG
barcode
S0circ.007B.007.C 5p Arm Adaptor, TGA
barcode
S0circ.007B.007.D 5p Arm Adaptor, TAG
barcode
S0circ.008 RT Primer (-1)
GAGATCTACACTCTTTCCCTACACGACGCTCTuccgaucu
S0.001 Solexa Fwd Seq
S0.002 Solexa Rev Seq, -1 short
AATGATACGGCGACCACCGAGATCTACACTCTTTCCCTACACGACGCTCTTC
CGATCT
CAAGCAGAAGACGGCATACGAGATCGGTCTCGGCATTCCTGCTGAACCGCTC
TTCCGATC
hsa-mir-125a binding selection
Note: Underlined bases were degenerate.
S0.003 T7-fwdSeq
CAGAGATGCATAATACGACTCACTATAGGACACGACGCTCTTCCGATCT
S125.001 Left arm
S125.002 Right arm
ACGCTCTTCCGATCTCCCCAGGGTCTACCGGGCCACCGCACACCATGTTGCC
AGTCTCTAGGTCCCTGAGACCCTTTAACCTGTGAGGACATCCAGGGTC
AGGAGTCAGGGGTCAGAAGTCAGGCCAGCAATTCCCCAGGTGTGTGGTTGGG
CCAGACGCCAGGCTCCCAAGAACCTCACCTGTGACCCTGGATGTCCTC
S125.003 RT primer
TATGAGGAGTCAGGGGTCAG
S125.005.A Solexa Reverse, barcoded
CAAGCAGAAGACGGCATACGAGATCGGTCTCGGCATTCCTGCTGAACCGCTC
TTCCGATCTCATTATGAGGAGTCAGGGGTCAG
CAAGCAGAAGACGGCATACGAGATCGGTCTCGGCATTCCTGCTGAACCGCTC
TTCCGATCTATGTATGAGGAGTCAGGGGTCAG
CAAGCAGAAGACGGCATACGAGATCGGTCTCGGCATTCCTGCTGAACCGCTC
TTCCGATCTTGATATGAGGAGTCAGGGGTCAG
CAAGCAGAAGACGGCATACGAGATCGGTCTCGGCATTCCTGCTGAACCGCTC
TTCCGATCTTAGTATGAGGAGTCAGGGGTCAG
S223circ.002a Right Arm
S125.005.B Solexa Reverse, barcoded
S125.005.C Solexa Reverse, barcoded
S125.005.D Solexa Reverse, barcoded
p-guacacguaugaGATCGGAAGAGCGGTTCAGCAGGAATGC
p-guacacgucauaGATCGGAAGAGCGGTTCAGCAGGAATGC
p-guacacguucaaGATCGGAAGAGCGGTTCAGCAGGAATGC
p-guacacgucuaaGATCGGAAGAGCGGTTCAGCAGGAATGC
GCATTCCTGCTGAACCGCTCTTCCGATC
113
hsa-mir-125a loop and apical selection
Note: Underlined bases were degenerate.
S125loop.001a Central
S125loop.002 Left Arm
ACCATGTTGCCAGTCTCTAGGTCCCTGAGACCCTTTAACCTGTGAGGACATC
CAGGGTCACAGGTGAGGTTCTTGGGAGCCTGGCGTCTGGCCCAACCAC
CAGAGATGCATAATACGACTCACTATAgCCCCCACCCCAGGGTCTACCGGGC
CACCGCACACCATGTTGCCAGTCTCTAGG
S125loop.003 Right Arm
GGTCAGAAGTCAGGCCAGCAATTCCCCAGGTGTGTGGTTGGGCCAGACGCCA
S125loop.004 RT Primer
GGCATAGGCTCCCAAGAACCTC
S125loop.005 Init. PCR Fwd
GACGATCTCCCTGAGACCCTTTAA
S125loop.005a Init. PCR Fwd, barcoded
GACGATCgaTCCCTGAGACCCTTTAA
S125loop.005b Init. PCR Fwd, barcoded
GACGATCctTCCCTGAGACCCTTTAA
S125loop.006 Solexa-R Adaptor
CAAGCAGAAGACGGCATACGAGGCTCCCAAGAACCTC
S125loop.007 Solexa-Seq Adaptor
AATGATACGGCGACCACCGACAGGTTCAGAGTTCTACAGTCCGACGATCTCC
CTGAGACCCTTTAA
AATGATACGGCGACCACCGACAGGTTCAGAGTTCTACAGTCCGACGATCgaT
CCCTGAGACCCTTTAA
S125loop.007a Solexa-Seq Adaptor,
barcoded
S125loop.007b Solexa-Seq Adaptor,
barcoded
AATGATACGGCGACCACCGACAGGTTCAGAGTTCTACAGTCCGACGATCctT
CCCTGAGACCCTTTAA
hsa-mir-16-1 loop and apical selection
Note: Underlined bases were degenerate.
S16-1loop.001a Central
S16-1loop.003 Right Arm
GCAATGTCAGCAGTGCCTTAGCAGCACGTAAATATTGGCGTTAAGATTCTAA
AATTATCTCCAGTATTAACTGTGCTGCTGAAGTAAGGTTGACCATAC
CAGAGATGCATAATACGACTCACTATAgATATTTCTTTTTATTCATAGCTCT
TATGATAGCAATGTCAGCAGTGCCTTAG
AACATTAATATACATTAAAACACAACTGTAGAGTATGGTCAACCTTACTTCA
G
S16-1loop.004 RT Primer
GACGGCATATTCAGCAGCACAGTTAATAC
S16-1loop.005 Init. PCR Fwd
CCGACGATCTAGCAGCACGTAAATATT
S16-1loop.005a Init. PCR Fwd, Barcoded
CCGACGATCgaTAGCAGCACGTAAATATT
S16-1loop.005b Init. PCR Fwd, Barcoded
CCGACGATCctTAGCAGCACGTAAATATT
S16-1loop.006 Solexa-R Adaptor
CAAGCAGAAGACGGCATACGATTCAGCAGCACAGTTAATAC
S16-1loop.007 Solexa-Seq Adaptor
AATGATACGGCGACCACCGACAGGTTCAGAGTTCTACAGTCCGACGATCTAG
CAGCACGTAAATATT
AATGATACGGCGACCACCGACAGGTTCAGAGTTCTACAGTCCGACGATCgaT
AGCAGCACGTAAATATT
S16-1loop.002 Left Arm
S16-1loop.007a Solexa-Seq Adaptor,
barcoded
S16-1loop.007b Solexa-Seq Adaptor,
barcoded
AATGATACGGCGACCACCGACAGGTTCAGAGTTCTACAGTCCGACGATCctT
AGCAGCACGTAAATATT
hsa-mir-30a loop and apical selection
Note: Underlined bases were degenerate.
S30aloop.001a Central
S30aloop.003 Right Arm
GCTGTTGACAGTGAGCGACTGTAAACATCCTCGACTGGAAGCTGTGAAGCCA
CAGATGGGCTTTCAGTCGGATGTTTGCAGCTGCCTACTGCCTCGGAC
CAGAGATGCATAATACGACTCACTATAGGTTAACCCAACAGAAGGCTAAAGA
AGGTATATTGCTGTTGACAGTGAGCGACTGTAAACATC
GTAAACAAGATAATTGCTCCTAAAGTAGCCCCTTGAAGTCCGAGGCAGTAGG
CAGCTGCAAACATC
S30aloop.004 RT Primer
GACGGCATAGCTGCAAACATCCGACTGA
S30aloop.005 Init. PCR Fwd
CCGACGATCTGTAAACATCCTCGACTG
S30aloop.005a Init. PCR Fwd, Barcoded
CCGACGATCgaTGTAAACATCCTCGACTG
S30aloop.005b Init. PCR Fwd, Barcoded
CCGACGATCctTGTAAACATCCTCGACTG
S30aloop.006 Solexa-R Adaptor
CAAGCAGAAGACGGCATACGAGCTGCAAACATCCGACTGA
S30aloop.007 Solexa-Seq Adaptor
AATGATACGGCGACCACCGACAGGTTCAGAGTTCTACAGTCCGACGATCTGT
AAACATCCTCGACTG
AATGATACGGCGACCACCGACAGGTTCAGAGTTCTACAGTCCGACGATCgaT
GTAAACATCCTCGACTG
S30aloop.002 Left Arm
S30aloop.007a Solexa-Seq Adaptor,
barcoded
114
S30aloop.007b Solexa-Seq Adaptor,
barcoded
AATGATACGGCGACCACCGACAGGTTCAGAGTTCTACAGTCCGACGATCctT
GTAAACATCCTCGACTG
hsa-mir-223 loop and apical selection
Note: Underlined bases were degenerate.
S223loop.001a Central
S223loop.003 Right Arm
TCCTGCAGTGCCACGCTCCGTGTATTTGACAAGCTGAGTTGGACACTCCATG
TGGTAGAGTGTCAGTTTGTCAAATACCCCAAGTGCGGCACATGCTTAC
CAGAGATGCATAATACGACTCACTATAGATCTCACTTCCCCACAGAAGCTCT
TGGCCTGGCCTCCTGCAGTGCCACGCTCCGTGTATTTGACAAGCTGAG
ATTCGTCATATCCCATCTGCCCTGGCCTAGAGCTGGTAAGCATGTGCCGCAC
TTGGGGTATTTGACAAAC
S223loop.004 RT Primer
GACGGCATATGGGGTATTTGACAAAC
S223loop.005 Init. PCR Fwd
CCGACGATCCGTGTATTTGACAAGCTGA
S223loop.005a Init. PCR Fwd, Barcoded
CCGACGATCgaCGTGTATTTGACAAGCTGA
S223loop.005b Init. PCR Fwd, Barcoded
CCGACGATCctCGTGTATTTGACAAGCTGA
S223loop.006 Solexa-R Adaptor
CAAGCAGAAGACGGCATACGATGGGGTATTTGACAAAC
S223loop.007 Solexa-Seq Adaptor
AATGATACGGCGACCACCGACAGGTTCAGAGTTCTACAGTCCGACGATCCGT
GTATTTGACAAGCTGA
AATGATACGGCGACCACCGACAGGTTCAGAGTTCTACAGTCCGACGATCgaC
GTGTATTTGACAAGCTGA
S223loop.002 Left Arm
S223loop.007a Solexa-Seq Adaptor,
barcoded
S223loop.007b Solexa-Seq Adaptor,
barcoded
AATGATACGGCGACCACCGACAGGTTCAGAGTTCTACAGTCCGACGATCctC
GTGTATTTGACAAGCTGA
115
Supplemental Table S2. Pri-miRNA collections
Related to the Experimental Procedures. Pri-miRNAs used in phylogenetic analyses were
chosen as described in the Experimental Procedures.
H. sapiens
D. rerio
hsa-mir-455, hsa-mir-451, hsa-mir-452, hsa-mir-181b-1, hsa-mir-568, hsa-mir-302a,
hsa-mir-7-1, hsa-mir-19a, hsa-mir-708, hsa-mir-219-2, hsa-mir-28, hsa-mir-431,
hsa-mir-301a, hsa-mir-432, hsa-mir-21, hsa-mir-433, hsa-mir-22, hsa-mir-574, hsamir-598, hsa-mir-599, hsa-mir-101-1, hsa-mir-592, hsa-mir-18a, hsa-mir-590, hsamir-3960, hsa-mir-449a, hsa-mir-320a, hsa-mir-1306, hsa-mir-125b-1, hsa-mir-31,
hsa-mir-32, hsa-mir-216a, hsa-mir-1193, hsa-mir-582, hsa-mir-212, hsa-mir-376a-1,
hsa-mir-214, hsa-mir-210, hsa-mir-215, hsa-mir-217, hsa-mir-129-2, hsa-mir-124-2,
hsa-mir-412, hsa-mir-411, hsa-mir-224, hsa-mir-328, hsa-mir-222, hsa-mir-325,
hsa-mir-223, hsa-mir-326, hsa-mir-718, hsa-mir-324, hsa-mir-1912, hsa-mir-711,
hsa-mir-425, hsa-mir-424, hsa-mir-423, hsa-mir-1277, hsa-mir-200a, hsa-mir-615,
hsa-mir-96, hsa-mir-105-1, hsa-mir-802, hsa-mir-337, hsa-mir-24-1, hsa-mir-338,
hsa-mir-339, hsa-mir-335, hsa-mir-330, hsa-mir-331, hsa-mir-744, hsa-mir-218-1,
hsa-mir-135a-2, hsa-mir-202, hsa-mir-203, hsa-mir-551b, hsa-mir-199a-2, hsa-mir345, hsa-mir-346, hsa-mir-204, hsa-mir-205, hsa-mir-342, hsa-let-7a-1, hsa-mir340, hsa-mir-544, hsa-mir-542, hsa-mir-541, hsa-mir-138-1, hsa-mir-33a, hsa-mir1298, hsa-mir-760, hsa-mir-668, hsa-mir-764, hsa-mir-150, hsa-mir-761, hsa-mir23b, hsa-mir-762, hsa-mir-664, hsa-mir-767, hsa-mir-665, hsa-mir-9-2, hsa-mir-3652, hsa-mir-149, hsa-mir-107, hsa-mir-142, hsa-mir-144, hsa-mir-1249, hsa-mir-143,
hsa-mir-1247, hsa-mir-145, hsa-mir-140, hsa-mir-193b, hsa-mir-652, hsa-mir-653,
hsa-mir-654, hsa-mir-759, hsa-mir-139, hsa-mir-153-2, hsa-mir-137, hsa-mir-136,
hsa-mir-134, hsa-mir-196a-2, hsa-mir-1264, hsa-mir-194-1, hsa-mir-127, hsa-mir126, hsa-mir-122, hsa-mir-505, hsa-mir-506, hsa-mir-1251, hsa-mir-770, hsa-mir503, hsa-mir-504, hsa-mir-10b, hsa-mir-1-2, hsa-mir-133a-1, hsa-mir-128-2, hsamir-99a, hsa-mir-486, hsa-mir-147b, hsa-mir-489, hsa-mir-488, hsa-mir-208a, hsamir-483, hsa-mir-485, hsa-mir-484, hsa-mir-384, hsa-mir-383, hsa-mir-191, hsa-mir190, hsa-mir-185, hsa-mir-186, hsa-mir-187, hsa-mir-188, hsa-mir-27b, hsa-mir892b, hsa-mir-490, hsa-mir-374c, hsa-mir-491, hsa-mir-3074, hsa-mir-146a, hsamir-375, hsa-mir-184, hsa-mir-500b, hsa-mir-378, hsa-mir-183, hsa-mir-182, hsamir-30c-1, hsa-mir-15b, hsa-mir-367, hsa-mir-362, hsa-mir-361, hsa-mir-363, hsamir-370, hsa-mir-1224, hsa-mir-371, hsa-mir-29a, hsa-mir-92a-1, hsa-mir-2861,
hsa-mir-496, hsa-mir-670, hsa-mir-26a-1, hsa-mir-495, hsa-mir-148a, hsa-mir-493,
hsa-mir-671, hsa-mir-499, hsa-mir-676, hsa-mir-675, hsa-mir-497, hsa-mir-875,
hsa-mir-876, hsa-mir-877, hsa-mir-155, hsa-mir-450a-2, hsa-mir-297, hsa-mir-3064,
hsa-mir-298, hsa-mir-296, hsa-mir-3065, hsa-mir-34a, hsa-mir-873, hsa-mir-299,
hsa-mir-874
dre-let-7f, dre-mir-100-2, dre-mir-101b, dre-mir-107, dre-mir-10d, dre-mir-122, dremir-124-2, dre-mir-125b-1, dre-mir-126b, dre-mir-128-1, dre-mir-129-2, dre-mir-1322, dre-mir-133a-2, dre-mir-135b, dre-mir-137-2, dre-mir-1388, dre-mir-139, dre-mir140, dre-mir-142b, dre-mir-145, dre-mir-146b-1, dre-mir-148, dre-mir-153a, dre-mir155, dre-mir-15c, dre-mir-1788, dre-mir-181b-1, dre-mir-182, dre-mir-183, dre-mir184, dre-mir-187, dre-mir-190, dre-mir-192, dre-mir-193a-1, dre-mir-194a, dre-mir196a-2, dre-mir-199-3, dre-mir-19b, dre-mir-202, dre-mir-203a, dre-mir-204-2, dremir-206-2, dre-mir-20a, dre-mir-210, dre-mir-21-1, dre-mir-214, dre-mir-216a-1, dremir-2188, dre-mir-218a-1, dre-mir-219-3, dre-mir-222, dre-mir-223, dre-mir-22a,
dre-mir-23a-3, dre-mir-24-1, dre-mir-26a-3, dre-mir-27a, dre-mir-29b-3, dre-mir301c, dre-mir-30a, dre-mir-31, dre-mir-338-1, dre-mir-34, dre-mir-363, dre-mir-3652, dre-mir-375-2, dre-mir-429, dre-mir-430c-3, dre-mir-451, dre-mir-454b, dre-mir455, dre-mir-456, dre-mir-458, dre-mir-460, dre-mir-489, dre-mir-499, dre-mir-736,
dre-mir-7b, dre-mir-92a-1, dre-mir-9-3, dre-mir-96
116
C. intestinalis
A. gambiae
D. melanogaster
cin-mir-4003b, cin-mir-4001a-2, cin-mir-4092, cin-mir-4001a-1, cin-mir-4091, cinmir-4094, cin-mir-4093, cin-mir-4098, cin-mir-196, cin-mir-4097, cin-mir-4099, cinmir-4219, cin-mir-4003d, cin-mir-4003c, cin-mir-4217, cin-mir-4090, cin-mir-4207,
cin-mir-4209, cin-mir-4208, cin-mir-4213, cin-mir-4214, cin-mir-4215, cin-mir-4216,
cin-mir-4006a-1, cin-mir-4212, cin-mir-4014-1, cin-mir-4006a-3, cin-mir-4006a-2,
cin-mir-4010-1, cin-mir-4001d, cin-mir-4001c, cin-mir-4001h, cin-mir-4001g, cin-mir4001f, cin-mir-4001e, cin-mir-4001i, cin-mir-1502c, cin-mir-1502b, cin-mir-1502d,
cin-mir-4013a, cin-mir-4013b, cin-mir-4120, cin-mir-4123, cin-mir-4124, cin-mir4121, cin-mir-4122, cin-mir-4127, cin-mir-4128, cin-mir-4118, cin-let-7f, cin-mir4000d, cin-mir-4000e, cin-mir-4000f, cin-mir-4000g, cin-mir-15, cin-mir-375, cin-mir4000c, cin-mir-4110, cin-mir-4112, cin-mir-219, cin-mir-4113, cin-mir-4114, cin-mir4115, cin-mir-4116, cin-mir-4011a, cin-mir-4011b, cin-mir-4000i, cin-mir-4000h, cinmir-4002, cin-mir-29, cin-mir-200, cin-mir-4007, cin-mir-4004, cin-mir-4106, cin-mir4103, cin-mir-4101, cin-mir-367, cin-mir-4100, cin-mir-4201, cin-mir-31, cin-mir4077d, cin-mir-4077c, cin-mir-4205, cin-mir-34, cin-mir-4203, cin-mir-124-1, cin-mir3598, cin-mir-3599, cin-mir-92c, cin-mir-92a, cin-mir-4159, cin-mir-4158, cin-mir4157, cin-mir-4156, cin-mir-4155, cin-mir-92e, cin-mir-4154, cin-mir-4153, cin-mir4019, cin-mir-4169, cin-mir-4166, cin-mir-4165, cin-mir-4021, cin-mir-4022, cin-mir4025, cin-mir-4178a, cin-mir-4163, cin-mir-4024, cin-mir-126, cin-mir-4012-2, cinmir-125, cin-mir-4026, cin-mir-4029, cin-mir-4028, cin-mir-3575, cin-mir-4020b, cinmir-4129, cin-mir-4134, cin-mir-4031, cin-mir-4133, cin-mir-4030, cin-mir-4132, cinmir-4139, cin-mir-135, cin-mir-133, cin-mir-4137, cin-mir-4136, cin-mir-4039, cinmir-4038, cin-mir-4037, cin-mir-4131, cin-mir-4036, cin-mir-4130, cin-mir-4035, cinmir-4034, cin-mir-4033, cin-mir-4041, cin-mir-4144, cin-mir-4143, cin-mir-4146, cinmir-4043, cin-mir-141, cin-mir-4145, cin-mir-4009b, cin-mir-4147, cin-mir-4009a,
cin-mir-4009c, cin-mir-4049, cin-mir-4048, cin-mir-4140, cin-mir-4045, cin-mir-4044,
cin-mir-4142, cin-mir-4047, cin-mir-4046, cin-mir-4141, cin-mir-4057, cin-mir-4196,
cin-mir-7, cin-mir-4197, cin-mir-4055, cin-mir-4194, cin-mir-4195, cin-mir-4056, cinmir-4192, cin-mir-4001b-2, cin-mir-4193, cin-mir-4190, cin-mir-4059, cin-mir-9, cinmir-4001b-1, cin-mir-4008b, cin-mir-4008c, cin-mir-4050, cin-mir-4008a, cin-mir155, cin-mir-4053, cin-mir-153, cin-mir-4054, cin-mir-4198, cin-mir-4051, cin-mir-1,
cin-mir-4052, cin-mir-1473, cin-mir-96, cin-mir-4066, cin-mir-4067, cin-mir-4068,
cin-mir-4069, cin-let-7a-2, cin-mir-4060, cin-mir-4061, cin-mir-4062, cin-mir-4063,
cin-mir-4064, cin-mir-4065, cin-mir-132-2, cin-mir-4171, cin-mir-4014-2, cin-mir4174, cin-mir-4079, cin-mir-4006f, cin-mir-4172, cin-mir-4078, cin-mir-4006g, cinmir-4173, cin-mir-4006d, cin-mir-4076, cin-mir-4006e, cin-mir-4179, cin-mir-4073,
cin-mir-4006b, cin-mir-4176, cin-mir-4074, cin-mir-4006c, cin-mir-4177, cin-mir4071, cin-mir-4072, cin-mir-4070, cin-mir-182, cin-mir-1497, cin-mir-4017-2, cin-mir4018b, cin-mir-281, cin-mir-4180, cin-mir-4181, cin-mir-4015-1, cin-mir-4088, cinmir-4183, cin-mir-4089, cin-mir-4185, cin-mir-4186, cin-mir-4084, cin-mir-4005a,
cin-mir-183, cin-mir-4085, cin-mir-4086, cin-mir-4005c, cin-mir-4087, cin-mir-40162, cin-mir-4081, cin-mir-4083
aga-mir-9a, aga-mir-1000, aga-mir-957, aga-mir-100, aga-mir-iab-4, aga-mir-275,
aga-mir-965-1, aga-mir-278, aga-mir-279, aga-mir-276, aga-mir-277, aga-mir-993,
aga-mir-1, aga-mir-8, aga-mir-305, aga-mir-996, aga-mir-137, aga-mir-10, aga-mir11, aga-mir-929, aga-mir-12, aga-mir-927, aga-mir-14, aga-mir-283, aga-mir-286,
aga-mir-375-2, aga-mir-281, aga-mir-282, aga-mir-309, aga-mir-308, aga-mir-307,
aga-bantam, aga-mir-1890, aga-mir-315, aga-mir-981, aga-let-7, aga-mir-87, agamir-184, aga-mir-219, aga-mir-2-1, aga-mir-1891, aga-mir-124, aga-mir-125, agamir-190, aga-mir-210, aga-mir-988, aga-mir-989, aga-mir-317, aga-mir-970, agamir-34, aga-mir-92b, aga-mir-263, aga-mir-1175
dme-mir-31a, dme-mir-932, dme-mir-8, dme-mir-5, dme-mir-4, dme-mir-7, dme-mir125, dme-mir-1, dme-mir-124, dme-mir-3, dme-mir-318, dme-mir-219, dme-mir-316,
dme-mir-317, dme-mir-34, dme-mir-33, dme-mir-193, dme-mir-190, dme-mir-281-2,
dme-mir-92a, dme-mir-210, dme-mir-315, dme-mir-314, dme-mir-313, dme-mir-312,
dme-mir-305, dme-mir-306, dme-mir-308, dme-mir-375, dme-mir-959, dme-mir-958,
117
C. elegans
C. briggsae
P. pacificus
dme-mir-184, dme-mir-957, dme-mir-955, dme-mir-100, dme-mir-2494, dme-mir304, dme-mir-303, dme-mir-969, dme-mir-965, dme-mir-9a, dme-mir-968, dme-mir967, dme-mir-961, dme-mir-962, dme-mir-276a, dme-mir-963, dme-mir-964, dmemir-960, dme-mir-2489, dme-mir-307b, dme-mir-263a, dme-mir-971, dme-mir-970,
dme-mir-975, dme-mir-976, dme-mir-977, dme-mir-978, dme-bantam, dme-mir307a, dme-mir-1001, dme-mir-1002, dme-mir-1003, dme-mir-1005, dme-mir-1006,
dme-mir-6-3, dme-mir-982, dme-mir-981, dme-mir-980, dme-mir-986, dme-mir1000, dme-mir-989, dme-mir-987, dme-mir-988, dme-mir-iab-4, dme-mir-iab-8,
dme-mir-252, dme-mir-2a-2, dme-mir-1012, dme-mir-1013, dme-mir-1010, dme-mir995, dme-mir-1011, dme-mir-994, dme-mir-87, dme-mir-996, dme-mir-991, dmemir-993, dme-mir-992, dme-let-7, dme-mir-288, dme-mir-289, dme-mir-999, dmemir-286, dme-mir-284, dme-mir-285, dme-mir-282, dme-mir-283, dme-mir-280,
dme-mir-927, dme-mir-929, dme-mir-279, dme-mir-11, dme-mir-275, dme-mir-12,
dme-mir-277, dme-mir-14, dme-mir-278, dme-mir-133, dme-mir-10, dme-mir-274,
dme-mir-137, dme-mir-983-2
cel-mir-2208b, cel-mir-237, cel-mir-90, cel-mir-238, cel-mir-55, cel-mir-233, cel-mir58, cel-mir-234, cel-mir-57, cel-mir-239a, cel-mir-235, cel-mir-236, cel-mir-59, celmir-2209b, cel-mir-230, cel-mir-231, cel-mir-232, cel-mir-50, cel-mir-52, cel-mir-51,
cel-mir-4922-1, cel-lin-4, cel-mir-2214, cel-mir-67, cel-mir-65, cel-mir-248, cel-mir249, cel-mir-246, cel-mir-247, cel-mir-244, cel-mir-245, cel-mir-240, cel-mir-241,
cel-mir-63, cel-mir-62, cel-mir-790, cel-mir-61, cel-mir-791, cel-mir-60, cel-mir-7892, cel-mir-2, cel-mir-4812, cel-mir-1, cel-mir-788, cel-mir-787, cel-mir-786, cel-mir785, cel-mir-71, cel-mir-255, cel-mir-72, cel-mir-35, cel-mir-73, cel-mir-38, cel-mir353, cel-mir-74, cel-mir-259, cel-mir-392, cel-mir-34, cel-mir-70, cel-mir-79, cel-mir358, cel-mir-359, cel-mir-250, cel-mir-354, cel-mir-75, cel-mir-251, cel-mir-355, celmir-76, cel-mir-252, cel-mir-356, cel-mir-77, cel-mir-253, cel-mir-357, cel-mir-254,
cel-let-7, cel-mir-1822, cel-lsy-6, cel-mir-39, cel-mir-124, cel-mir-84, cel-mir-268,
cel-mir-49, cel-mir-85, cel-mir-269, cel-mir-48, cel-mir-82, cel-mir-83, cel-mir-46,
cel-mir-80, cel-mir-360, cel-mir-44, cel-mir-43, cel-mir-42, cel-mir-228, cel-mir-86,
cel-mir-87
cbr-let-7, cbr-mir-238, cbr-mir-236, cbr-mir-237, cbr-mir-234, cbr-mir-87, cbr-mir235, cbr-mir-233, cbr-mir-47, cbr-mir-48, cbr-mir-49, cbr-mir-43, cbr-mir-124, cbrmir-42, cbr-mir-83, cbr-mir-242, cbr-mir-239b, cbr-mir-84, cbr-mir-241, cbr-mir-85,
cbr-mir-240, cbr-mir-86, cbr-mir-80, cbr-mir-81, cbr-mir-248, cbr-mir-249, cbr-mir232-1, cbr-mir-35b-6, cbr-mir-244, cbr-mir-245, cbr-mir-76, cbr-mir-246, cbr-mir791, cbr-mir-39, cbr-mir-73b, cbr-mir-34, cbr-mir-789, cbr-mir-786, cbr-mir-787, cbrmir-784, cbr-mir-785, cbr-mir-357-1, cbr-mir-74, cbr-mir-251, cbr-mir-250, cbr-mir72, cbr-mir-253, cbr-mir-392, cbr-mir-252, cbr-mir-70, cbr-mir-71, cbr-mir-90b, cbrmir-67, cbr-mir-359, cbr-mir-255, cbr-mir-358, cbr-mir-254, cbr-mir-355, cbr-mir-353,
cbr-mir-259, cbr-mir-354, cbr-mir-360, cbr-mir-45-2, cbr-mir-790-1, cbr-mir-2222-2,
cbr-mir-60, cbr-mir-61, cbr-mir-1822, cbr-mir-57, cbr-mir-55, cbr-mir-228, cbr-mir-58,
cbr-mir-1, cbr-mir-268, cbr-mir-54a, cbr-mir-77-2, cbr-lsy-6, cbr-mir-52, cbr-mir-35d,
cbr-mir-231, cbr-mir-50, cbr-mir-230
ppc-mir-2265, ppc-mir-2266, ppc-mir-2235-1, ppc-mir-2267, ppc-mir-2268, ppc-mir2269, ppc-mir-279, ppc-mir-72, ppc-mir-71, ppc-mir-63b, ppc-mir-67, ppc-mir-65,
ppc-mir-2239-1, ppc-mir-993, ppc-mir-2271, ppc-mir-236, ppc-mir-2270, ppc-mir2233, ppc-mir-2273, ppc-mir-234, ppc-mir-2272, ppc-mir-2234a, ppc-mir-232, ppcmir-2274, ppc-lin-4, ppc-mir-2237b, ppc-mir-1, ppc-mir-2, ppc-mir-86, ppc-mir-55a,
ppc-mir-55b, ppc-mir-81, ppc-mir-124, ppc-mir-79, ppc-mir-2253b, ppc-mir-2247,
ppc-mir-2248, ppc-mir-2249, ppc-let-7, ppc-mir-2243, ppc-mir-252, ppc-mir-2244,
ppc-mir-38, ppc-mir-2245, ppc-mir-37, ppc-mir-2246, ppc-mir-312, ppc-mir-2241a-2,
ppc-mir-45, ppc-mir-46, ppc-mir-2240c, ppc-mir-2250, ppc-mir-2236a, ppc-mir2236b, ppc-mir-87, ppc-mir-2258, ppc-mir-2259, ppc-mir-2256, ppc-mir-240, ppcmir-84b, ppc-mir-2255, ppc-mir-2234b, ppc-mir-242, ppc-mir-2242-2, ppc-mir-22321, ppc-mir-2238a-2, ppc-mir-56, ppc-mir-42b, ppc-mir-42a, ppc-mir-2264, ppc-mir2263, ppc-mir-2262, ppc-mir-2261, ppc-mir-239-2
118
C. teleta
L. gigantea
S. mediterranea
N. vectensis
cte-mir-1999, cte-mir-87a, cte-mir-1997, cte-mir-1998, cte-mir-1991, cte-mir-124,
cte-mir-1992, cte-mir-1993, cte-mir-1994, cte-mir-193, cte-mir-277b, cte-mir-216b,
cte-mir-12, cte-mir-750, cte-mir-1989, cte-mir-216c, cte-mir-2694, cte-mir-137, ctemir-1987, cte-mir-2693, cte-mir-2692, cte-mir-2691, cte-mir-133, cte-mir-2690, ctemir-2686b, cte-mir-2699, cte-mir-2686c, cte-mir-2695, cte-mir-2696, cte-mir-29b,
cte-mir-92c, cte-mir-92b, cte-mir-153, cte-mir-219, cte-mir-277a, cte-mir-210, ctemir-210b, cte-mir-71, cte-mir-182, cte-mir-1175, cte-mir-1996b, cte-mir-1996a, ctemir-281, cte-mir-981, cte-mir-278, cte-mir-279, cte-mir-67, cte-mir-2001, cte-mir375, cte-mir-2000, cte-mir-7, cte-mir-1, cte-mir-9, cte-mir-8, cte-mir-2703, cte-mir2706, cte-mir-993, cte-mir-2705, cte-mir-2708, cte-mir-2707, cte-mir-996, cte-mir2709, cte-mir-365, cte-mir-2700, cte-mir-2702, cte-mir-317, cte-mir-2687, cte-mir1990c, cte-mir-2g, cte-mir-2685, cte-mir-2f, cte-mir-2e, cte-mir-315, cte-mir-1990b,
cte-mir-2d, cte-mir-2c, cte-mir-2689, cte-mir-2688, cte-mir-2719, cte-mir-2718, ctemir-2717, cte-mir-252b, cte-mir-2716, cte-mir-2714, cte-mir-2713, cte-mir-2712, ctemir-2711, cte-mir-2710, cte-mir-184b, cte-mir-31, cte-let-7, cte-bantam, cte-mir-10b,
cte-mir-10a, cte-mir-10d, cte-mir-10c, cte-mir-745a, cte-mir-745b, cte-mir-242, ctemir-2721, cte-mir-33, cte-mir-34, cte-mir-2720, cte-mir-36
lgi-mir-31, lgi-mir-34, lgi-mir-33, lgi-mir-2001, lgi-mir-745b, lgi-mir-252, lgi-let-7, lgimir-87, lgi-mir-981, lgi-mir-193, lgi-mir-216a, lgi-mir-2c, lgi-mir-317, lgi-mir-1, lgi-mir315, lgi-mir-124, lgi-mir-745a, lgi-mir-96a, lgi-mir-96b, lgi-mir-71, lgi-mir-2d, lgi-mir182, lgi-mir-184, lgi-mir-183, lgi-mir-375, lgi-mir-8, lgi-mir-10, lgi-mir-9, lgi-mir-7, lgimir-12, lgi-mir-1993, lgi-mir-137, lgi-mir-133, lgi-mir-281, lgi-mir-1990, lgi-mir-1992,
lgi-mir-1991, lgi-mir-92, lgi-mir-67, lgi-mir-29, lgi-mir-242, lgi-mir-1985, lgi-mir-279,
lgi-mir-1984, lgi-mir-278, lgi-mir-1989, lgi-mir-1994b, lgi-mir-1988, lgi-mir-750, lgimir-1986, lgi-mir-252b, lgi-mir-100, lgi-mir-1175
sme-mir-1175, sme-mir-1a, sme-mir-1c, sme-mir-1b, sme-mir-71c, sme-mir-125a,
sme-mir-125b, sme-mir-61b, sme-mir-61a, sme-mir-2156b, sme-mir-2156a, smemir-92, sme-lin-4, sme-mir-7d, sme-mir-7c, sme-mir-7b, sme-mir-7a, sme-mir-745,
sme-mir-31a, sme-mir-10b, sme-mir-10a, sme-mir-190b, sme-mir-190a, sme-mir12, sme-mir-2150, sme-mir-8b, sme-mir-13, sme-mir-8a, sme-mir-2152, sme-mir2151, sme-mir-2153, sme-mir-2155, sme-mir-1993, sme-mir-2158, sme-mir-1992,
sme-mir-2157, sme-mir-278, sme-mir-2159, sme-mir-96b, sme-mir-315, sme-mir9b, sme-mir-281, sme-mir-2154-1, sme-mir-2149, sme-mir-2148, sme-mir-2200,
sme-mir-2201, sme-mir-2202, sme-let-7d, sme-mir-124e, sme-mir-277d, sme-mir2203, sme-mir-124d, sme-mir-277c, sme-let-7c, sme-mir-2204, sme-mir-31b-2,
sme-mir-277b, sme-mir-2205, sme-mir-277a, sme-mir-2206, sme-mir-2147b, smemir-2e, sme-mir-2f, sme-mir-2177, sme-mir-2b, sme-mir-2178, sme-mir-2179, smemir-36c, sme-mir-2173, sme-mir-36b, sme-mir-36a, sme-mir-2175, sme-mir-2176,
sme-mir-2170, sme-mir-2171, sme-mir-2172, sme-bantam-a, sme-mir-133a, smemir-133b, sme-mir-2168, sme-mir-2169, sme-mir-2166, sme-mir-2167, sme-mir2164, sme-mir-2165, sme-mir-2162, sme-mir-2163, sme-mir-2161, sme-mir-67,
sme-mir-748, sme-mir-749, sme-mir-746, sme-mir-87d, sme-mir-747, sme-mir-87b,
sme-mir-87c, sme-mir-219, sme-mir-87a, sme-mir-752, sme-mir-751, sme-mir-750,
sme-mir-756, sme-mir-755, sme-mir-79, sme-mir-76, sme-mir-753b, sme-mir-184,
sme-mir-2160-1, sme-mir-216, sme-mir-754c-1, sme-mir-2181, sme-let-7b,
nve-mir-2049, nve-mir-2023, nve-mir-2022, nve-mir-2047, nve-mir-2048, nve-mir100, nve-mir-2041, nve-mir-2042, nve-mir-2029, nve-mir-2028, nve-mir-2046, nvemir-2027, nve-mir-2026, nve-mir-2044, nve-mir-2025, nve-mir-2040a-1, nve-mir2024a-3, nve-mir-2033, nve-mir-2034, nve-mir-2031, nve-mir-2030, nve-mir-2050,
nve-mir-2035-2, nve-mir-2037, nve-mir-2036, nve-mir-2039, nve-mir-2043b, nvemir-2032b-2
119
120
Chapter 3.
Future directions
Contents
Overview ..................................................................................................................................... 122
Expanding the understanding of human Microprocessor substrates........................................... 122
Recognition of human pri-miRNAs without known motifs ............................................... 122
Dynamics of pri-miRNA recognition in different cellular contexts ................................... 123
Recognition of non-miRNA substrates of Drosha and DGCR8/Pasha............................... 127
Defining nematode pri-miRNAs ................................................................................................. 128
Gaining insight into mechanisms of pri-miRNA recognition ..................................................... 132
Applying selection approaches to stubborn questions in RNA biology ..................................... 136
121
Overview
In Chapter 2, in vitro selection of partially-randomized pri-miRNAs and high-throughput
sequencing of the functional variants confirmed the importance of secondary structure and
revealed three novel and conserved primary sequence determinants. These experiments provided
a detailed perspective on the substrate specificity of the human Microprocessor, and,
correspondingly, the features that define most conserved human pri-miRNAs. The results in
Chapter 2 also inspire several additional questions related to the Microprocessor and the
biological definition of a pri-miRNA.
I will discuss some attractive avenues of further
investigation, ranging from a wider search for determinants in human Microprocessor substrates
to the application of in vitro and in vivo selection strategies to find functional elements in other
RNAs that have resisted detailed characterization.
Expanding the understanding of human Microprocessor
substrates
Recognition of human pri-miRNAs without known motifs
The three recognition motifs identified in Chapter 2 are generally utilized by conserved
human pri-miRNAs, with 79% having one or more motifs. Despite the success of the study, the
motifs that define the remaining 21% are still unknown.
Many of these pri-miRNAs are
conserved across vertebrates, and three are conserved across bilaterian animals, suggesting that
they have distinct, evolutionarily-conserved processing motifs that were simply not present in the
four pri-miRNAs studied (Figure 1A). Partial-randomization and selection of these pri-miRNAs
will therefore expand the catalog of pri-miRNA recognition motifs, and perhaps uncover
additional determinants that are deeply conserved in bilaterian animals.
Some of these pri-miRNAs are located in clusters of miRNAs, including hsa-mir-15b,
hsa-mir-24-1, and hsa-mir-200a, and other members of these clusters have the downstream
CNNC motif. These pri-miRNAs may simply use separate processing motifs that have not been
discovered yet (Figure 1B), but an intriguing possibility is that the processing of clustered primiRNAs is somehow coordinated, and that a single positioned CNNC instance in the cluster is
sufficient to recruit the Microprocessor to process the cluster hairpins (Figure 1C). Consistent
with this idea, the ~600 kDa Microprocessor complex has been suggested to consist of multiple
122
functional Drosha and DGCR8 units (Han et al., 2004; Sohn et al., 2007), and a single complex
could be capable of simultaneously cleaving multiple pri-miRNAs in a cluster. At the same
time, clustered miRNAs could interact at a tertiary level to promote or inhibit joint processing by
the Microprocessor (Chakraborty et al., 2012).
This question could be addressed by
computationally analyzing clustered pri-miRNAs and determining whether the probability that
any individual member has a processing motif is dependent on the presence or absence of
processing motifs in other cluster members. Suggestive results could be followed by in vitro and
in vivo studies of the processing of clustered miRNAs.
Dynamics of pri-miRNA recognition in different cellular contexts
Processing motifs may exist that are dependent on cell identity or cell state, and may not
have been identified in Chapter 2 because the lysate used for pri-miRNA cleavage was derived
from a single cell line (HEK293T) under standard culture conditions. Such conditional motifs,
and the proteins that recognize them, are likely to help regulate levels of pri-miRNA processing
and may play important, dynamic roles during physiological changes such as during
development or in response to extracellular signals; or during pathological changes, such as
oncogenic transformation.
For example, the regulation and dysregulation of miRNAs has a crucial role in
oncogenesis, primary tumor growth, and distant metastasis (Garzon et al., 2009). On a cellular
and molecular level, studies have demonstrated the importance of miRNAs as both tumor
suppressors and oncogenes (Zhang et al., 2007).
Indeed, many highly-expressed miRNAs
overlap with regions that are genomically unstable in cancer, including deletions, duplications,
translocations, or losses of heterozygozity (Calin et al., 2005).
At an epidemiological level,
miRNA expression patterns are highly correlated with both cancer type (Lu et al., 2005) and
clinical course (Calin et al., 2004a; Yanaihara et al., 2006), and thus are of both diagnostic and
prognostic value.
In particular, pri-miRNA processing is dysregulated in cancer. For example, a SNP in
the hsa-mir-16-1 locus severely impairs the processing of mir-16-1 by the Microprocessor. Mir16-1 has been shown to be frequently deleted in chronic lymphoid leukemia (CLL) (Calin et al.,
2002), leading to dysregulation of the key apoptosis factor BCL2 (Cimmino et al., 2005).
Correspondingly, the SNP identified in mir-16-1 that impairs processing was found in two CLL
123
patients, one meeting the criteria for familial CLL, and was associated with loss of
heterozygozity in both (Calin et al., 2005). More broadly, as discussed in Chapter 1, several
oncogenic and tumor-suppressor proteins have been shown to interact with pri-miRNAs and
regulate their cleavage by the Microprocessor complex, including p53 (Suzuki et al., 2009),
Smad family members (Davis et al., 2008; Davis et al., 2010; Hata and Davis, 2011), and Lin-28
[reviewed in (Viswanathan and Daley, 2010)]. In most cases, the RNA motifs that these proteins
recognize and bind are poorly understood.
Identification of conditional motifs will require two expansions to the in vitro selection
approach described in Chapter 2 (Figure 1D). First, a more extensive panel of pri-miRNAs must
be examined, preferably ones for which there is evidence of dynamic processing and important
function in physiology or pathology. For example, four pri-miRNAs might be fertile ground for
additional study: hsa-mir-206, hsa-mir-199a-2, hsa-mir-155, and hsa-let-7e. SNPs have been
identified in cancer samples for all four of these miRNAs, and these polymorphisms may be
responsible for inhibited or enhanced miRNA processing in specific cells or tissues (Calin et al.,
2005; Wu et al., 2008).
mir-206 is a tumor suppressor miRNA that is downregulated in
metastatic subsets of the human breast cancer cell line MDA-MB-231 and the human lung cancer
line LM2; restoration of miR-206 expression decreases the colonization capacity of these cells
(Tavazoie et al., 2008).
mir-199a has been implicated in the regulation of the SWI/SNF
complex, and hence may be both a tumor suppressor and an oncogenic miRNA, depending on
the tumor (Sakurai et al., 2011). Likewise, miR-155, the major product of the BIC locus, is an
oncogenic miRNA activated in human and bird B-lymphomas (Tam et al., 1997; Eis et al.,
2005), and its overexpression is sufficient to cause lymphoproliferation (Costinean et al., 2006)
and myeloproliferation (O'Connell et al., 2008) disorders in mice. Finally, the let-7 family of
miRNAs is a well-studied class of tumor-suppressor miRNAs that have been implicated in a
variety of oncologic settings in both molecular and epidemiological studies [reviewed in
(Viswanathan and Daley, 2010)].
The second expansion is the preparation and use of lysates from different cell lines or
tissues, in order to maximize opportunity to detect cancer or cell-type specific motifs. For
example, one might use four readily-available cell lines: HeLa, derived from a cervical
carcinoma; Huh7, derived from a hepatocellular carcinoma; MCF7, derived from a breast
carcinoma; and K562, derived from a chronic myelogenous leukemia. These lines divide rapidly
124
and are efficiently transfected using cationic lipids, making them suitable for medium-scale
preparation of lysates containing the Microprocessor complex and any auxiliary proteins
expressed in those cells. These lysates would be used to cleave the expanded panel of partiallyrandomized miRNA substrates. As in Chapter 2, any novel motifs should be confirmed by in
vitro and in vivo analysis.
High-throughput sequencing of the selected pri-miRNA variants is likely to reveal a
variety of sequence and structural features important to processing. Some of these elements will
recur in selections from all four cell lines and 293T cells, suggesting that these elements are
important for general recognition of pri-miRNAs for cleavage. In contrast, other features could
be present in some selections but not others; such motifs are of special interest. For example,
some motifs may only be observable in the four human cancer lines, and not in the 293T
transformed embryonic line; or in some cancer lines, but not others. This heterogeneity suggests
that the utilization of such features is dependent on the cellular milieu, including the expression
and/or activity of unknown proteins that may dynamically regulate miRNA processing. Variable
motifs may therefore be binding sites for such proteins.
In addition to novel motif discovery, the in vitro selection will also provide data on the
effect of cancer SNPs previously identified in the four example miRNAs described above. These
polymorphisms will be represented millions of times in the proposed selection pool, in addition
to the billions of variants mutated at other positions. Analysis of the selected variants will
provide quantitative data on the cleavage efficiency of pri-miRNAs with mutations that mimic
cancer SNPs.
For let-7e, where the specific polymorphism is already known to impair
processing of the pri-miRNA (Wu et al., 2008), further study would help define the complete
sequence or structural motif that is mutated in prostate cancer. For mir-206, mir-199a-2, and
mir-155, detailed studies of processing motifs would help establish whether the cancer
polymorphisms affect pri-miRNA processing, in addition to helping define the affected motif.
To the extent that novel elements enhance miRNA biogenesis, this study will also aid
efforts towards miRNA-based cancer gene therapy. Several approaches are based on stably
reconstituting expression of tumor suppressor miRNAs to inhibit tumor growth or metastasis
(Esquela-Kerscher et al., 2008; Kumar et al., 2008; Kota et al., 2009; Aigner, 2011), while others
are attempting to stably express miRNA-derived hairpins engineered to downregulate specific
oncogenes (Wang et al., 2009). In both cases, highly efficient miRNA processing will be critical
125
A
Mammalian
>1 known motif
79%
No known motifs
21%
hsa-mir-302a
hsa-mir-320a
hsa-mir-325
hsa-mir-326
hsa-mir-328
hsa-mir-335
hsa-mir-337
hsa-mir-383
hsa-mir-384
hsa-mir-423
hsa-mir-484
hsa-mir-486
hsa-mir-505
hsa-mir-568
hsa-mir-582
Human pri-miRNAs
B
hsa-mir-653
hsa-mir-654
hsa-mir-664
hsa-mir-665
hsa-mir-668
hsa-mir-676
hsa-mir-711
hsa-mir-718
hsa-mir-759
hsa-mir-764
hsa-mir-767
hsa-mir-877
hsa-mir-1193
hsa-mir-1251
hsa-mir-1912
hsa-mir-2861
Vertebrate
Bilaterian
hsa-mir-15b
hsa-mir-24-1
hsa-mir-143
hsa-mir-155
hsa-mir-181b-1
hsa-mir-214
hsa-mir-223
hsa-mir-301a
hsa-mir-153-2
hsa-mir-200a
hsa-mir-216a
C
Microprocessor
CNNC
D
???
Partially randomized pri-miRNA sequences
mir-206
let-7 family
mir-199a-2
mir-155
In vitro selection
HeLa
Huh7
MCF7
K562
Lysate overexpressing Drosha and DGCR8
Unselected reference
pools
High-throughput sequencing
Novel motif discovery
to the efficacy of the therapy, and these efforts will be able to use the miRNA processing motifs
identified in this study. In addition, if cancer-specific or tissue-specific motifs are identified,
gene therapy platforms could use this information to maximize therapeutic miRNA expression
specifically in cancer cells.
Recognition of non-miRNA substrates of Drosha and DGCR8/Pasha
Several groups have recently described substrates of Drosha and DGCR8/Pasha that are
not processed into mature miRNAs, at least not efficiently. Instead, these substrates are mRNAs,
and cleavage destabilizes the target messages. The best-described example of this is DGCR8 in
humans; its mRNA contains two hairpins, one in the 5′ untranslated region and another in the
beginning of the coding region, which are cleaved by Drosha (Han et al., 2009). Since the
DGCR8 protein helps stabilize Drosha, the cleavage of DGCR8 would in turn lead to
destabilization of Drosha. Thus, the activity of the Drosha-DGCR8 complex is self-limiting
(Han et al., 2009). Two mature miRNAs, hsa-mir-3618 and hsa-mir-1306 have been annotated
that are derived from the DGCR8 hairpins, but the importance of these miRNAs is unclear; the
fact that mature miRNAs can be produced might be considered good illustrations of how the
specificity of the processing machinery downstream of the Microprocessor is largely dependent
on the biochemistry of the previous steps (Chapter 1).
DGCR8 is not the only mRNA target of Drosha cleavage. A microarray comparison of
gene expression after Drosha knockdown to expression after Dicer knockdown revealed
Figure 1. Open questions in the recognition of human pri-miRNAs
(A) How does the Microprocessor recognize pri-miRNAs without known primary sequence determinants?
Among conserved human pri-miRNA families, 21% do not have any of the three primary sequence
determinants elucidated in Chapter 2. The representatives of these human pri-miRNA families are
listed, classified by their conservation.
(B) To what degree is the recognition of individual pri-miRNAs biochemically independent from the
recognition of other pri-miRNAs in a cluster? One model is that recognition of clustered pri-miRNAs
is distinct, implying that pri-miRNAs in the cluster that do not have apparent primary sequence
determinants utilize other motifs that have yet to be discovered.
(C) To what degree is the recognition of clustered pri-miRNAs coordinated? In contrast to (B), sequence
determinants in one pri-miRNA could be sufficient to recruit the Microprocessor to the cluster, so that
multiple pri-miRNAs are dependent on a single set of determinants for processing.
(D) To what extent is the use of individual recognition determinants conditional on cell state or identity?
Performing in vitro selection on additional pri-miRNAs in cell lysates derived from a panel of cancer
cell lines could identify motifs that are differentially utilized in different cellular environments.
127
hundreds of mRNAs upregulated after Drosha knockdown but not after Dicer knockdown,
suggesting that Drosha could be directly regulating these mRNAs (Han et al., 2009). Similar
results were obtained with Dicer knockout and Drosha knockout cell lines in mice (Chong et al.,
2010), and an analogous experiment in Drosophila S2 cells similar found hundreds of genes
upregulated by Drosha knockdown but not Dicer knockdown (Kadener et al., 2009). Thus
Drosha could cleave many mRNAs as a mechanism for directly regulating of gene expression.
There are several open questions in the direct cleavage of mRNAs. First, are hairpins in
mRNA targets recognized in the same manner as pri-miRNA hairpins? A glance at the DGCR8
hairpins indicates that the upstream hairpin does not have any of the three motifs identified in
Chapter 2, while the downstream hairpin has both the loop GUG and the downstream CNNC. A
more systematic analysis is needed to determine whether mRNA cleavage substrates have the
same propensity as pri-miRNAs to have one or more processing motifs. Second, is the DroshaDGCR8 complex that cleaves mRNA hairpins the same as the Microprocessor complex that
cleaves pri-miRNAs? Cleavage of mRNA hairpins has the same core protein requirements
(Drosha and Pasha/DGCR8), but it would be interesting to know whether the ~600 kDa
Microprocessor complex isolated by gel filtration (Gregory et al., 2004) is fully competent to
cleave these hairpins, or if a distinct complex of Drosha, Pasha/DGCR8, and other unidentified
proteins is utilized instead. If these hairpins use processing motifs distinct from those of primiRNAs, it would be reasonable to hypothesize that the proteins that recognize those motifs are
members of a distinct Drosha-DGCR8 complex.
Finally, is Drosha-DGCR8 cleavage of
individual mRNAs regulated in a dynamic manner? From a teleological standpoint, constitutive
cleavage of the DGCR8 mRNA is comprehensible, since this negative feedback makes the
activity of Drosha and DGCR8 self-limiting. However, it is not obvious why Drosha and
DGCR8 would constitutively cleave hundreds of other putative mRNA targets. As with primiRNA processing, perhaps Drosha and DGCR8 cleavage of individual mRNAs is dynamically
regulated.
Defining nematode pri-miRNAs
As described in Chapter 2, a species barrier exists between C. elegans pri-miRNAs and
human pri-miRNAs, leading to poor processing of C. elegans pri-miRNAs in human cells. In
Chapter 2, I described the discovery of additional motifs that promote efficient pri-miRNAs
128
processing in humans, which are absent in C. elegans and other nematode pri-miRNAs. The lack
of human processing motifs in nematode pri-miRNAs explains why the nematode pri-miRNAs
are poorly processed in human cells, and the species barrier can be bridged by adding the human
processing motifs to nematode pri-miRNAs.
Although this work has improved the understanding of human pri-miRNA processing,
nematode processing continues to be poorly understood. In fact, pri-miRNA processing in
nematodes may be unique among bilaterian animals: the downstream CNNC motif described in
Chapter 2 is conserved in pri-miRNAs throughout bilaterian animals, but is strikingly absent in
nematodes. One interpretation is that mechanisms of pri-miRNA recognition have generally
diverged in nematodes from recognition in other bilaterian animals.
In fact, pri-miRNA
processing would not be alone in this divergence. Nematodes have 2-3 fold higher substitution
rates in their 18S ribosomal sequences than other metazoan animals (Aguinaldo et al., 1997),
and, among nematodes, the evolution of the C. elegans 18S ribosomal sequence is particularly
fast (Holterman et al., 2006). Rapid evolution is reflected in protein coding genes as well; the
nematodes have lost several homeobox (Hox) genes compared to the Hox clusters in other
animals, and there are high substitution rates in the retained Hox genes (Aboobaker and Blaxter,
2003). Again, even among nematodes, C. elegans and its close relatives have evolved more
rapidly, with more Hox gene losses and even a unique expansion of the posterior Hox genes
(Aboobaker and Blaxter, 2003).
The approach used in Chapter 2 to detail the processing motifs of human pri-miRNAs
used a cell lysate to cleave pri-miRNA variants in vitro. This approach can be applied to
nematode pri-miRNAs, with some considerations for biological and practical differences
between C. elegans and other model animals. One example of a biological difference that could
impinge on pri-miRNA processing studies is trans-splicing. C. elegans is unique among the
model animals in that some 70% of pre-messenger RNA transcripts are spliced in trans to 22 nt
leader exons called SL1 and SL2 (Blumenthal, 2005). Since the splice leaders RNAs form the 5′
end of the mature mRNA, splice leaders provide the methylguanosine cap, and trans-splicing
serves to physically excise individual mRNAs from polycistronic transcripts.
In many
transcripts, SL1 and SL2 form almost the entirety of the 5′ untranslated region in the mature
mRNA, and may play a role in efficient translational initiation (Blumenthal, 2005).
129
In a similar way, trans-splicing may play a critical role in pri-miRNA processing. The
processing of the cel-let-7 pri-miRNA is thought to be dependent on trans-splicing to SL1,
which occurs 41 nt upstream of the 5p cleavage site (Bracht et al., 2004). The degree to which
trans-splicing occurs and is important in other pri-miRNAs is not clear, although analysis of
whole-transcriptome sequencing in C. elegans suggests that trans splicing to pri-miRNAs may
be relatively common (J.-W. Nam, personal communication). This poses an inconvenience for
the synthesis of pri-miRNA substrates for in vitro studies. For Drosophila and humans, primiRNA substrates have been inferred by simply examining the sequence of the corresponding
genome. By contrast, it will be necessary to determine the 5′ ends of the C. elegans pri-miRNAs
by examining transcriptome sequencing data when possible, or by direct experimental
approaches like amplification and sequencing of cDNA ends (e.g., 5′-RACE).
When trans-splicing to cel-let-7 was characterized, it was suggested that the SL1 exon
changed the optimal overall secondary structure of the pri-miRNA, presumably converting a
nonfunctional substrate into a recognizable pri-miRNA substrate (Bracht et al., 2004). However,
computationally-predicted folding (and changes in folding) is notoriously difficult to interpret
when RNAs are hundreds of nucleotides long. To the extent that trans-splicing is generally
important in pri-miRNA processing, it seems unlikely that the SL1 sequence optimizes the
folding of all these divergent RNAs. An alternative possibility is that the SL1 exon contains a
processing motif that marks nearby hairpins as bona-fide pri-miRNAs. For example, the SL1
exon contains an AGUU tetraloop (Greenbaum et al., 1996), and the structure of this tetraloop is
thought to be very similar to the AGNN tetraloops recognized by the S. cerevisae RNase III
protein Rnt1p (Wu et al., 2001). Perhaps one of the dsRBDs in the C. elegans Microprocessor
complex has convergently evolved to recognize the AGNN tetraloop in SL1, and this recognition
serves to recruit the nematode Microprocessor to pri-miRNAs. If the AGUU tetraloop in SL1 is
important, it should be easy to see enrichment of the SL1 stem and tetraloop after in vitro
selection.
On a practical level, C. elegans has been a difficult platform for biochemistry. Since no
C. elegans cell line is currently available, whole animals would have to be raised and used to
make cell-free lysates, including their protease- and nuclease-rich gastrointestinal tracts.
Nevertheless, lysates derived from C. elegans larvae have been used to study miRNA turnover,
indicating that lysates with cytosolic activities and probably nuclear activities can be successfully
130
prepared (Chatterjee and Grosshans, 2009). In addition, a cell-free system from the parasitic
nematode Ascaris lumbricoides was developed to study transcription and co-transcriptional
events, including capping (Maroney et al., 1990) and trans-splicing (Hannon et al., 1990).
Recently, this technique was adapted to generate splicing-competent extracts in C. elegans
(Lasda et al., 2010). It is reasonable to believe that either of these extracts from C. elegans could
also be competent for pri-miRNA processing.
Regardless of the mechanism of pri-miRNA recognition in nematodes, the divergence
from other animals is an opportunity to explore the co-evolution of an enzyme complex and its
substrates.
At first glance, an enzyme with multiple substrates seems likely to evolve its
substrate preferences very slowly, since any significant change would cause the enzyme to lose
the capacity to catalyze the reaction on many (perhaps most) of its substrates. To the extent that
the enzyme activity on these substrates is important, changes to enzyme specificity would surely
be detrimental, and thus the enzyme would be subjected to strong purifying selection. For
example, the 3′ splice site consensus sequence YAG recognized by the U2AF complex is
conserved in canonical U2-dependent splice sites in animals, plants, and fungi (Sharp and Burge,
1997; Spingola et al., 1999), perhaps because any alteration to the specificity of U2AF would
negatively affect thousands of 3′ splice sites. By contrast, proteins that interact with just one
binding partner may be capable of much more rapid evolution in their binding specificities. One
of the best-studied examples of this is the abalone sperm enzyme lysin, which specifically binds
the egg vitelline protein VERL and induces a conformational change that opens a pore for sperm
heads to access the egg membrane. VERL and lysin are among the most rapidly evolving
proteins in the abalone genome, perhaps because there is evolutionary pressure to continuously
change sperm-egg specificity, particularly in organisms with external fertilization (Swanson and
Vacquier, 2002). It is reasonable to believe that this rapid evolution would be severely impeded
if VERL had to simultaneously maintain critical interactions with dozens or hundreds of ligands.
Nevertheless, Microprocessor specificity does appear to have evolved in nematodes.
One potential path to altered enzyme specificity is duplication and sub- or neofunctionalization;
the redundancy allows one duplicate to evolve, while the other maintains the biologically
important substrate interactions. Over time, substrates could then co-evolve to be recognized by
one or both duplicates. For example, a functionally distinct spliceosome with different snRNPs,
including U12 in place of U2, tends to utilize 3′ splice sites that have an AC rather than AG
131
(Tarn and Steitz, 1996).
In worms and other animals, there could be variant Microprocessor
complexes that would have allowed the substrate specificity of the complex to change over
evolutionary time.
Another potential path is modularity in the substrate, where multiple
recognition motifs collaborate to distinguish true substrates from other substrates, even though
each individual motif contributes relatively little to recognition. In this scenario, if the enzyme
or enzyme complex alters its specificity for any one of the motifs, overall recognition is not
significantly impacted; over time, the specificity for all the motifs could change quite
considerably. For example, the rapid evolution of VERL is likely facilitated by the presence of
22 tandem repeats of the VERL binding domain, since variations in one repeat will not be too
detrimental to lysin binding; in time, the overall VERL-lysin interaction can change
dramatically, resulting in speciation (Swanson and Vacquier, 2002). Likewise, recognition of
human pri-miRNAs appears to be modular, and it would be reasonable to suppose that
recognition of nematode pri-miRNAs is also modular, although the modules themselves have
diverged between the two lineages.
Once the recognition motifs have been determined in C. elegans, phylogenetic analysis of
pri-miRNA sequences across the nematode clade could permit a reconstruction of the
evolutionary history of pri-miRNA recognition. For example, it is possible that the less rapidlyevolving members of the nematode clade might continue to use the downstream CNNC motif,
while gaining recognition elements that are used throughout the nematode clade. C. elegans
might use the common nematode elements, and may have gained additional elements specific to
the rhabditid subclade, in lieu of the CNNC motif. If the members of the Microprocessor
complex are elucidated that recognize these putative motifs, it may even be possible to correlate
the evolutionary history of pri-miRNA motifs with the history of individual Microprocessor
components. The next section will describe possible strategies for finding protein components of
the Microprocessor that recognize pri-miRNA motifs.
Gaining insight into mechanisms of pri-miRNA recognition
Each sequence or structural preference found in the selection corresponds to a protein
component of the Microprocessor that recognizes and binds to the motif, or to an RNA structure
whose formation is dependent upon the motif. What proteins bind the motifs identified in
Chapter 2, and any additional motifs that might be identified in pri-miRNAs? Given a functional
132
RNA motif, it is possible to specifically incorporate a photoreactive nucleoside analog in the
motif, crosslink the nucleoside to a bound protein, and identify the proteins by mass
spectrometry. This technique of incorporating a photoreactive nucleoside at a specific site in a
long RNA substrate was initially used to identify spliceosome proteins that recognize the 5′
splice site motif (Wyatt et al., 1992). For snoRNAs and tRNAs, structured RNAs with multiple
important motifs and protein interactions, this form of site-specific crosslinking has provided
physical maps of the functional elements of the ribonucleoprotein complex relative to the
substrate RNA (Mishima and Steitz, 1995; Cahill et al., 2002).
The classic site-specific crosslinking protocol (Sontheimer, 1994) can be adapted to
purify protein/pri-miRNA complexes and identify the peptide components by mass spectrometry.
For example, to identify proteins bound to the CNNC motif, a pri-miRNA substrate could be
assembled by splint ligation of three RNA fragments: a synthetic RNA oligonucleotide with a
4-thiouridine moiety in the CNNC motif, a synthetic RNA oligonucleotide with a 3′ biotin
modification, and an RNA that includes the 5p flank and the hairpin (Figure 2A). This substrate
would be crosslinked to candidate proteins in the Microprocessor-containing lysate described in
Chapter 2, and covalently-linked protein-RNA complexes purified by binding to streptavidincoated beads (Figure 2B). As in the classic protocol, RNase T1 or another suitable RNase would
be used to digest away the RNA substrate; in the proposed protocol, RNase digestion would
additionally serve to elute the protein-RNA complexes from the streptavidin-coated beads. The
peptides in the eluted protein-RNA complex would then be identified by mass spectrometry.
Nearly all pri-miRNA motifs should be amenable to this crosslinking approach.
However, one potential pitfall is that, for certain motifs, bound proteins may not be efficiently
crosslinked; the 4-thiouridine moiety can only be crosslinked to peptide moieties within a few
angstroms (Sontheimer, 1994). This spatial specificity is advantageous for identifying only those
proteins which are directly bound to the motif. However, for some direct binding interactions, it
is possible that the thio group may be too distant from a reactive amino acid to crosslink
efficiently. An alternative strategy could use 5′-[(4-azidophenacyl)thio]-uridine, a photoreactive
uridine analog whose reactive moiety is about 10 angstroms distant from the base, and is thus
more likely to react with nearby amino acids (Hanna, 1989). Additionally, one could attempt to
co-precipitate the bound protein with biotinylated pri-miRNA without crosslinking. Binding of
biotinylated, wildtype pri-miRNA molecules would be carried out in the presence of competing
133
A
Splint ligation
s
s
p CUUC
Bio
GApCUUC
Bio
p
B
Crosslinking
?
s
s
Bio
GApCUUC
Bio
Binding to
solid support
?
s
GApCUUC
Bio
Streptavidin-coated bead
GApCUUC
Aggressive
washing
Elution by
RNAse T1
Elution by
RNAse T1
?
s
GApCUUC
Peptide identification by mass spectrometry
Bio
pri-miRNA molecules in which the motif has been mutated. This approach would selectively coprecipitate proteins whose binding depends on the motif; these proteins can then be identified by
either mass spectrometry or Western blot. In any of these approaches, the contributions of the
candidate proteins should be delineated by in vitro and in vivo processing assays after loss-offunction or gain-of-function of the candidate proteins.
The identification of proteins that recognize pri-miRNA processing motifs has
implications for the regulation and dysregulation of miRNA processing in physiology and
pathology. As discussed above, processing motifs may be mutated in cancer, as the CNNC motif
is mutated in human chronic lymphoid leukemia. The proteins that recognize these motifs may
themselves be dysregulated in cancer, and may be possible to infer the nature of this
dysregulation based on the regulated miRNA (tumor-suppressor or oncogene) and the type of
motif (promoting or inhibiting Microprocessor cleavage). For example, proteins that enhance
processing of a tumor-suppressor miRNA may themselves be tumor suppressors. One instance
of this is found in the biogenesis of let-7, a family of tumor-suppressor miRNAs that target key
oncogenes such as c-MYC, and HMGA2 (Mayr et al., 2007; Chang et al., 2008). Lin-28A protein
inhibits Microprocessor cleavage of let-7 family pri-miRNAs and thus reduces mature levels of
this key tumor-suppresor miRNA (Piskounova et al., 2011). Correspondingly, both isoforms of
Lin-28 are upregulated in cancer samples, particularly in samples from advanced malignancy
(Viswanathan et al., 2009).
It could be possible to characterize the functional roles of identified proteins by analyzing
analyze publicly-available databases on cancer gene expression, mutations, clinical course, and
drug sensitivity (Cowin et al., 2010). For example, candidate proteins could be cross-referenced
against protein-coding genes known to be mutated in cancer (Futreal et al., 2004). Data on the
Figure 2.
Proposed strategy for identifying proteins that bind identified pri-miRNA sequence
determinants. Identification of CNNC-binding proteins is shown as an example.
(A) Assembly of a biotinylated pri-miRNA substrate containing a 4-thiouridine moiety specifically
incorporated in the CNNC motif. Three synthetic RNAs would be assembled into a pri-miRNA
32
substrate by splint ligation. The red “p” denotes a P phosphate.
(B) Purification of crosslinked RNA-protein complexes. The assembled pri-miRNA would be incubated in
whole-cell lysate. Proteins that bind in close proximity to the 4-thiouridine moiety would be
crosslinked to the biotinylated pri-miRNA by 365 nm ultraviolet light. Crosslinked complexes would
be purified by streptavidin bead binding with aggressive washing, and released by RNase T1
digestion of the pri-miRNA. Protein components of the complex would be identified by massspectrometry.
135
NCI-60 cancer cell lines would be particularly appealing for analysis, since many types of data
are available for all 60 lines, including protein-coding gene expression data (Ross et al., 2000;
Liu et al., 2010) and miRNA expression data (Blower et al., 2007; Liu et al., 2010; Sokilde et al.,
2011). For proteins that have not been characterized in the literature, gene expression clustering
analysis may help reveal the molecular pathways that impinge on the candidate proteins and thus
on miRNA biogenesis.
Identifying the proteins that bind processing motifs and characterizing their effects on
miRNA biogenesis would open new avenues of investigation and may lead to opportunities for
the development of diagnostic tools or novel therapeutics. At a basic level, identifying proteins
that bind pri-miRNAs and regulate their biogenesis would expand the catalog of proteins
comprise the Microprocessor complex, and shed light on the mechanisms by which the activity
of the complex can be regulated. In addition, since little is known of the structure of the
Microprocessor, determination of which proteins bind to various parts of the pri-miRNA would
yield a schematic of the three-dimensional structure of the Microprocessor, albeit at low spatial
resolution. From a regulatory pathway standpoint, classification of previously uncharacterized
proteins as miRNA regulators may give insight into novel oncogenic or tumor suppressing
regulatory pathways. On an epidemiological level, the expression levels of identified proteins
may be useful as biomarkers for tumor progression and prognosis, as is the case for Drosha,
Dicer, and Lin-28 (Merritt et al., 2008; Viswanathan et al., 2009). To the extent that identified
proteins are mutated in cancer, the proposed work may also aid the interpretation of loci
identified in cancer genome efforts. On a therapeutic level, these proteins may also be novel
drug targets for regulating miRNA levels.
Applying selection approaches to stubborn questions in RNA
biology
The availability of hundreds of organismal genomes has made it possible to identify and
characterize functional elements, including those in functional RNAs, simply by computationally
analyzing sequence data. The power of comparative sequence analysis is illustrated by some of
the original work to identify Dicer: knowing that the precursors of small interfering RNAs were
double-stranded and that RNase III enzymes in other organisms cleaved RNA duplexes,
investigators simply searched the D. melanogaster and C. elegans genomes for sequences that
136
would code for proteins resembling RNase III (Bernstein et al., 2001). Computational analysis
can even provide powerful evidence for biological importance when actual function is unknown,
since conservation of nucleic acid or protein sequence over a long evolutionary timespan is ipso
facto evidence for purifying selection.
However, computational analysis of genomic sequence data is limited when functional
sequence elements are degenerate, as motifs for RNA–protein binding tend to be. The study of
pri-miRNA determinants described in Chapter 2 was situated just beyond the limits of
computational analysis, given the relatively unsuccessful efforts to find determinants by
comparing individual pri-miRNAs to their orthologs, or by comparing pri-miRNAs to other primiRNAs in the same species. In fact, the determinants identified by in vitro selection are
composed of just two or three informative nucleotides, which are generally too short to be found
by multiple-sequence alignment or de novo motif discovery algorithms. In particular, discovery
of the CNNC motif by pure computational analysis was probably complicated because it is a
“spaced” motif, where two informative nucleotides are separated by less informative nucleotides,
and the CNNC position can vary across a narrow range of positions relative to the Drosha
cleavage site. Even when subtle enrichment signals were detected in pure sequence analysis, it
was often difficult to interpret their significance, since the sequence data was not intrinsically
rooted in pri-miRNA processing. These difficulties were overcome by generating hundreds of
billions of sequence variants, selecting in vitro for those that were functional in a specific
biochemical context, and coupling the in vitro selection to high-throughput sequencing.
This approach is not limited to the identification of motifs in pri-miRNAs; it can be
applied to the study of any functional nucleic acid, and is uniquely suited to study RNA species
which have not been amenable to computational analysis either because of extensive divergence
that impairs evolutionary conservation analysis, or because the elements to be found are too
subtle to be detected above noise by computational algorithms.
application is the study of long noncoding RNAs (lncRNAs).
One potentially fertile
These RNAs are important
regulators of gene expression, but studies of lncRNAs have been hampered by diversity in both
sequence and regulatory mechanism (Wang and Chang, 2011).
When blocks of sequence
conservation can be found, the sequences have been shown to be important for the function of
the lncRNA and for normal vertebrate development (Ulitsky et al., 2011).
137
A
Incubation with
immunopurified PRC2
Nitrocellulose
filtration
Pool of partially-randomized
lncRNA variants
High-throughput
sequencing
??
Pool of selected
lncRNA variants
B
Novel motifs that
define PRC2 binding
High-throughput
plasmid sequencing
Poly-A
signals
Barcode
Reporter
ORF
Database mapping barcodes
to lncRNA variant sequences
Enhancer-like
lncRNA
A
AAAA
In vivo expression
of reporter mRNA
AAA
A
AAA
AAAA
A
AAA
High-throughput
barcode sequencing
AAAA
Plasmid pool of partially-randomized
lncRNA variants
A
AAA
AAAA
??
Novel motifs that
promote lncRNA function
C
Model
ORF
Putative zipcode
region
Soma
compartment
In vivo expression
of model mRNA
Poly-A
signal
Plasmid pool of partially-randomized
3’ UTR variants
High-throughput
sequencing
NGF-containing
axon compartment
A
AAAA
AAA
A
??
AAAA
AAA
AAAA
Novel motifs that
regulate mRNA localization
However, the vast majority of lncRNAs are poorly conserved at the sequence level, even
when the presence of a noncoding transcript is conserved in syntenic regions of the genome in
other organisms (Wang and Chang, 2011). In vitro and in vivo selection techniques, coupled
with high throughput sequencing, could rapidly identify and characterize the functional elements
in such lncRNAs. For example, many lncRNAs are probably molecular scaffolds that assist in
the assembly of chromatin-regulation complexes in specific genomic regions. The best-studied
example of this is the recruitment of polycomb repressive complex 2 (PRC2) to the HOXD locus
by the lncRNA HOTAIR (Rinn et al., 2007). In fact, up to 20% of the annotated lncRNAs in
mammalian genomes may bind PRC2, although some of these may be interacting nonspecifically
(Khalil et al., 2009; Guttman et al., 2011). The region of HOTAIR that binds PRC2 components
has been isolated to a 300 nt region that is thought to have some secondary structure, but the
importance of this structure or any potential primary sequence elements is unclear (Tsai et al.,
2010). A relatively straightforward approach to studying this region would be to partially
randomize the 300 nt region and select for variants that retain binding to immunopurified PRC2
complexes, followed by sequencing of those variants (Figure 3A). This approach resembles that
first used to define the HIV Rev-binding site (Bartel et al., 1991). Any motifs that are found in
HOTAIR might directly translate to some of the hundreds of other lncRNAs that are thought to
bind PRC2, thus defining a functional domain in many lncRNAs with a single experiment.
Another set of lncRNAs that could be studied by in vivo selection and sequencing are the
enhancer-like lncRNAs that promote the expression of nearby protein-coding genes, including a
Figure 3. Possible selection strategies to study three functional RNAs.
(A) Scheme for identifying motifs that specify binding substrates of PRC2. Variants of lncRNAs that bind
the immunopurified PRC2 complex would be isolated by nitrocellulose filtration and sequenced.
(B) Scheme for mapping functional elements in enhancer-like lncRNAs. The enhancer lncRNA
sequences would be partially randomized and cloned into an expression plasmid with a nearby
reporter mRNA containing a random barcode. An initial sequencing run would associate the
sequence of the enhancer-lncRNA variants with the reporter barcode. Successful variants would
promote the transcription of the reporter mRNA, including the barcode, and would be identified by
sequencing the barcode.
(C) Scheme for identifying and characterizing zipcode motifs that promote the localization of mRNAs to
axon terminals. Variant 3′ UTR sequences corresponding to a putative zipcode region would be
cloned into a plasmid downstream of a model localized protein-coding sequence. Neurons
expressing these variants would be grown in a compartmentalized chamber that permits easy
separation of axons from the neuronal soma. Successful variants would be isolated from the axon
compartment and sequenced.
139
set of transcription factors that are master regulators of hematopoiesis (Orom et al., 2010). The
enhancement function of these lncRNAs is dependent on the lncRNA molecule, as opposed to
the act of transcription, but the important elements in the lncRNA are unclear. Since the
enhancing effects of the lncRNAs can be recapitulated in a plasmid-based reporter gene system,
the lncRNAs could be partially randomized and cloned into a reporter plasmid, and functional
variants could be selected on the basis of reporter expression. The randomization, selection, and
sequencing would be carried out following a strategy developed to study DNA enhancer
elements (Patwardhan et al., 2012). This strategy uses a two-step approach to selection and
sequencing. In the first step, the plasmid pool is sequenced to determine the sequence of the
partially randomized region, which is correlated to a barcode sequence located in the reporter
transcript.
In the second step, plasmids are transfected into cells, where functional DNA
enhancer elements drive the transcription of the reporter mRNA, including the barcode. These
barcoded mRNAs are reverse transcribed and sequenced to determine which variants
successfully promoted transcription.
For enhancer-like lncRNAs, the strategy would be
identical, except the partially-randomized DNA enhancer sequence would be replaced by the
partially-randomized lncRNA sequence (Figure 3B).
Another potentially fertile application of selection and high-throughput sequencing is the
study of “zipcode” elements that specify the localization of mRNAs. Dynamic mRNA transport
is an important component of many biological processes, such as the formation of spatial
gradients in developmental syncytia and polarized cells, local regulation at the leading edge of
migrating cells, and the specification of cell identity in daughter cells after mitosis (Meignin and
Davis, 2010). The regulated localization of mRNAs is particularly important in neurons, where
transported mRNAs play important roles in the growth and migration of new dendrites and
axons, and in the maintenance of functional synaptic terminals (Jung et al., 2012). The bestcharacterized zipcode is a bipartite element in the 3′ UTR of the β-actin mRNA, consisting of
two 6 nt motifs separated by a 15 nt spacer; this spacer is thought to allow the zipcode region to
loop around the zipcode-binding protein (ZBP1) so that the two motifs can be simultaneously
recognized by a pair of KH domains (Patel et al., 2012). However, this zipcode is present in just
a fraction of synapse-localized mRNAs; for the rest, neither the zipcodes nor the proteins that
recognize them are known. For example, the mRNA of the chaperone protein calreticulin is
localized to axon terminals, and two 100 nt conserved regions in the 3′ UTR are individually
140
sufficient to drive mRNA transport, but the actual motifs and the proteins that bind them are
unknown (Vuppalanchi et al., 2010). Since the 3′ UTR is sufficient to drive mRNA transport, a
relatively simple selection approach could be undertaken where a pool of variant 3′ UTRs would
be cloned into an expression plasmid downstream of a reporter mRNA; the plasmid pool would
be transfected into neurons grown in a compartmentalized culture system that permits easy
separation of axon terminals from the neuron soma (Campenot et al., 2009). Successful 3′ UTR
variants would be transported into the axon terminals, collected by the compartmentalized
culture system, amplified and sequenced (Figure 3C). Once the specific zipcode motifs have
been delineated, the proteins that bind the zipcode and recruit the mRNA to transport granules
would be identified by candidate testing or by site-specific crosslinking, as described above.
These three applications illustrate the diversity of questions that could be addressed by
the careful design of a selection experiment, coupled to a high-throughput sequencing of the
selected variants.
This type of strategy is becoming increasingly attractive as the cost of
sequencing falls, and the capacity increases to generate long, high-quality sequences. More
broadly, prospects are excellent for approaches that combine experimental savvy with the power
of computational analysis to crack stubborn but critical problems in biology.
141
Bibliography and References Cited
Aboobaker, A., and Blaxter, M. (2003). Hox gene evolution in nematodes: novelty conserved.
Curr Opin Genet Dev 13, 593-598.
Aguinaldo, A.M., Turbeville, J.M., Linford, L.S., Rivera, M.C., Garey, J.R., Raff, R.A., and
Lake, J.A. (1997). Evidence for a clade of nematodes, arthropods and other moulting
animals. Nature 387, 489-493.
Aigner, A. (2011). MicroRNAs (miRNAs) in cancer invasion and metastasis: therapeutic
approaches based on metastasis-related miRNAs. J Mol Med (Berl) 89, 445-457.
Akey, D.L., and Berger, J.M. (2005). Structure of the nuclease domain of ribonuclease III from
M. tuberculosis at 2.1 A. Protein Sci 14, 2744-2750.
Altuvia, S., Locker-Giladi, H., Koby, S., Ben-Nun, O., and Oppenheim, A.B. (1987). RNase III
stimulates the translation of the cIII gene of bacteriophage lambda. Proc Natl Acad Sci U S A
84, 6511-6515.
Ambros, V., Bartel, B., Bartel, D.P., Burge, C.B., Carrington, J.C., Chen, X., Dreyfuss, G., Eddy,
S.R., Griffiths-Jones, S., Marshall, M., et al. (2003). A uniform system for microRNA
annotation. RNA 9, 277-279.
Arvey, A., Larsson, E., Sander, C., Leslie, C.S., and Marks, D.S. (2010). Target mRNA
abundance dilutes microRNA and siRNA activity. Mol Syst Biol 6, 363.
Ashburner, M., Ball, C.A., Blake, J.A., Botstein, D., Butler, H., Cherry, J.M., Davis, A.P.,
Dolinski, K., Dwight, S.S., Eppig, J.T., et al. (2000). Gene ontology: tool for the unification
of biology. The Gene Ontology Consortium. Nat Genet 25, 25-29.
Babiarz, J.E., Ruby, J.G., Wang, Y., Bartel, D.P., and Blelloch, R. (2008). Mouse ES cells
express endogenous shRNAs, siRNAs, and other Microprocessor-independent, Dicerdependent small RNAs. Genes Dev 22, 2773-2785.
Baek, D., Villen, J., Shin, C., Camargo, F.D., Gygi, S.P., and Bartel, D.P. (2008). The impact of
microRNAs on protein output. Nature 455, 64-71.
Bar, M., Wyman, S.K., Fritz, B.R., Qi, J., Garg, K.S., Parkin, R.K., Kroh, E.M., Bendoraite, A.,
Mitchell, P.S., Nelson, A.M., et al. (2008). MicroRNA discovery and profiling in human
embryonic stem cells by deep sequencing of small RNA libraries. Stem Cells 26, 2496-2505.
Bardwell, J.C., Regnier, P., Chen, S.M., Nakamura, Y., Grunberg-Manago, M., and Court, D.L.
(1989). Autoregulation of RNase III operon by mRNA processing. EMBO J 8, 3401-3407.
Barry, G., Squires, C., and Squires, C.L. (1980). Attenuation and processing of RNA from the
rplJL--rpoBC transcription unit of Escherichia coli. Proc Natl Acad Sci U S A 77, 33313335.
Bartel, D.P. (2004). MicroRNAs: genomics, biogenesis, mechanism, and function. Cell 116, 281297.
Bartel, D.P. (2009). MicroRNAs: target recognition and regulatory functions. Cell 136, 215-233.
Bartel, D.P., Zapp, M.L., Green, M.R., and Szostak, J.W. (1991). HIV-1 Rev regulation involves
recognition of non-Watson-Crick base pairs in viral RNA. Cell 67, 529-536.
Bass, B.L. (2000). Double-stranded RNA as a template for gene silencing. Cell 101, 235-238.
Bazzini, A.A., Lee, M.T., and Giraldez, A.J. (2012). Ribosome profiling shows that miR-430
reduces translation before causing mRNA decay in zebrafish. Science 336, 233-237.
Beltrame, M., and Tollervey, D. (1995). Base pairing between U3 and the pre-ribosomal RNA is
required for 18S rRNA synthesis. EMBO J 14, 4350-4356.
142
Bentwich, I., Avniel, A., Karov, Y., Aharonov, R., Gilad, S., Barad, O., Barzilai, A., Einat, P.,
Einav, U., Meiri, E., et al. (2005). Identification of hundreds of conserved and nonconserved
human microRNAs. Nat Genet 37, 766-770.
Berezikov, E., van Tetering, G., Verheul, M., van de Belt, J., van Laake, L., Vos, J., Verloop, R.,
van de Wetering, M., Guryev, V., Takada, S., et al. (2006). Many novel mammalian
microRNA candidates identified by extensive cloning and RAKE analysis. Genome Res 16,
1289-1298.
Bernstein, E., Caudy, A.A., Hammond, S.M., and Hannon, G.J. (2001). Role for a bidentate
ribonuclease in the initiation step of RNA interference. Nature 409, 363-366.
Bernstein, E., Kim, S.Y., Carmell, M.A., Murchison, E.P., Alcorn, H., Li, M.Z., Mills, A.A.,
Elledge, S.J., Anderson, K.V., and Hannon, G.J. (2003). Dicer is essential for mouse
development. Nat Genet 35, 215-217.
Blow, M.J., Grocock, R.J., van Dongen, S., Enright, A.J., Dicks, E., Futreal, P.A., Wooster, R.,
and Stratton, M.R. (2006). RNA editing of human microRNAs. Genome Biol 7, R27.
Blower, P.E., Verducci, J.S., Lin, S., Zhou, J., Chung, J.H., Dai, Z., Liu, C.G., Reinhold, W.,
Lorenzi, P.L., Kaldjian, E.P., et al. (2007). MicroRNA expression profiles for the NCI-60
cancer cell panel. Mol Cancer Ther 6, 1483-1491.
Blumenthal, T. (2005). Trans-splicing and operons. WormBook, 1-9.
Bohmert, K., Camus, I., Bellini, C., Bouchez, D., Caboche, M., and Benning, C. (1998). AGO1
defines a novel locus of Arabidopsis controlling leaf development. EMBO J 17, 170-180.
Bohnsack, M.T., Czaplinski, K., and Gorlich, D. (2004). Exportin 5 is a RanGTP-dependent
dsRNA-binding protein that mediates nuclear export of pre-miRNAs. RNA 10, 185-191.
Bohnsack, M.T., Regener, K., Schwappach, B., Saffrich, R., Paraskeva, E., Hartmann, E., and
Gorlich, D. (2002). Exp5 exports eEF1A via tRNA from nuclei and synergizes with other
transport pathways to confine translation to the cytoplasm. EMBO J 21, 6205-6215.
Bracht, J., Hunter, S., Eachus, R., Weeks, P., and Pasquinelli, A.E. (2004). Trans-splicing and
polyadenylation of let-7 microRNA primary transcripts. RNA 10, 1586-1594.
Bram, R.J., Young, R.A., and Steitz, J.A. (1980). The ribonuclease III site flanking 23S
sequences in the 30S ribosomal precursor RNA of E. coli. Cell 19, 393-401.
Breaker, R.R., Banerji, A., and Joyce, G.F. (1994). Continuous in vitro evolution of
bacteriophage RNA polymerase promoters. Biochemistry 33, 11980-11986.
Brownawell, A.M., and Macara, I.G. (2002). Exportin-5, a novel karyopherin, mediates nuclear
export of double-stranded RNA binding proteins. J Cell Biol 156, 53-64.
Brummelkamp, T.R., Bernards, R., and Agami, R. (2002). A system for stable expression of
short interfering RNAs in mammalian cells. Science 296, 550-553.
Butcher, S.E., Dieckmann, T., and Feigon, J. (1997). Solution structure of the conserved 16 Slike ribosomal RNA UGAA tetraloop. J Mol Biol 268, 348-358.
Cahill, N.M., Friend, K., Speckmann, W., Li, Z.H., Terns, R.M., Terns, M.P., and Steitz, J.A.
(2002). Site-specific cross-linking analyses reveal an asymmetric protein distribution for a
box C/D snoRNP. EMBO J 21, 3816-3828.
Calabrese, J.M., Seila, A.C., Yeo, G.W., and Sharp, P.A. (2007). RNA sequence analysis defines
Dicer's role in mouse embryonic stem cells. Proc Natl Acad Sci U S A 104, 18097-18102.
Calado, A., Treichel, N., Muller, E.C., Otto, A., and Kutay, U. (2002). Exportin-5-mediated
nuclear export of eukaryotic elongation factor 1A and tRNA. EMBO J 21, 6216-6224.
Calin, G.A., Dumitru, C.D., Shimizu, M., Bichi, R., Zupo, S., Noch, E., Aldler, H., Rattan, S.,
Keating, M., Rai, K., et al. (2002). Frequent deletions and down-regulation of micro- RNA
143
genes miR15 and miR16 at 13q14 in chronic lymphocytic leukemia. Proc Natl Acad Sci U S
A 99, 15524-15529.
Calin, G.A., Ferracin, M., Cimmino, A., Di Leva, G., Shimizu, M., Wojcik, S.E., Iorio, M.V.,
Visone, R., Sever, N.I., Fabbri, M., et al. (2005). A MicroRNA signature associated with
prognosis and progression in chronic lymphocytic leukemia. N Engl J Med 353, 1793-1801.
Calin, G.A., Liu, C.G., Sevignani, C., Ferracin, M., Felli, N., Dumitru, C.D., Shimizu, M.,
Cimmino, A., Zupo, S., Dono, M., et al. (2004a). MicroRNA profiling reveals distinct
signatures in B cell chronic lymphocytic leukemias. Proc Natl Acad Sci U S A 101, 1175511760.
Calin, G.A., Sevignani, C., Dumitru, C.D., Hyslop, T., Noch, E., Yendamuri, S., Shimizu, M.,
Rattan, S., Bullrich, F., Negrini, M., et al. (2004b). Human microRNA genes are frequently
located at fragile sites and genomic regions involved in cancers. Proc Natl Acad Sci U S A
101, 2999-3004.
Campenot, R.B., Lund, K., and Mok, S.A. (2009). Production of compartmented cultures of rat
sympathetic neurons. Nat Protoc 4, 1869-1887.
Caudy, A.A., Myers, M., Hannon, G.J., and Hammond, S.M. (2002). Fragile X-related protein
and VIG associate with the RNA interference machinery. Genes Dev 16, 2491-2496.
Cenik, E.S., Fukunaga, R., Lu, G., Dutcher, R., Wang, Y., Tanaka Hall, T.M., and Zamore, P.D.
(2011). Phosphate and R2D2 restrict the substrate specificity of Dicer-2, an ATP-driven
ribonuclease. Mol Cell 42, 172-184.
Chakraborty, S., Mehtab, S., Patwardhan, A., and Krishnan, Y. (2012). Pri-miR-17-92a transcript
folds into a tertiary structure and autoregulates its processing. RNA 18, 1014-1028.
Chakravarthy, S., Sternberg, S.H., Kellenberger, C.A., and Doudna, J.A. (2010). Substratespecific kinetics of Dicer-catalyzed RNA processing. J Mol Biol 404, 392-402.
Chanfreau, G., Buckle, M., and Jacquier, A. (2000). Recognition of a conserved class of RNA
tetraloops by Saccharomyces cerevisiae RNase III. Proc Natl Acad Sci U S A 97, 3142-3147.
Chanfreau, G., Legrain, P., and Jacquier, A. (1998). Yeast RNase III as a key processing enzyme
in small nucleolar RNAs metabolism. J Mol Biol 284, 975-988.
Chang, T.C., Yu, D., Lee, Y.S., Wentzel, E.A., Arking, D.E., West, K.M., Dang, C.V., ThomasTikhonenko, A., and Mendell, J.T. (2008). Widespread microRNA repression by Myc
contributes to tumorigenesis. Nat Genet 40, 43-50.
Chatterjee, S., and Grosshans, H. (2009). Active turnover modulates mature microRNA activity
in Caenorhabditis elegans. Nature 461, 546-549.
Chelladurai, B.S., Li, H., and Nicholson, A.W. (1991). A conserved sequence element in
ribonuclease III processing signals is not required for accurate in vitro enzymatic cleavage.
Nucleic Acids Res 19, 1759-1766.
Cheloufi, S., Dos Santos, C.O., Chong, M.M., and Hannon, G.J. (2010). A dicer-independent
miRNA biogenesis pathway that requires Ago catalysis. Nature 465, 584-589.
Chen, C.Z., Li, L., Lodish, H.F., and Bartel, D.P. (2004). MicroRNAs modulate hematopoietic
lineage differentiation. Science 303, 83-86.
Chendrimada, T.P., Gregory, R.I., Kumaraswamy, E., Norman, J., Cooch, N., Nishikura, K., and
Shiekhattar, R. (2005). TRBP recruits the Dicer complex to Ago2 for microRNA processing
and gene silencing. Nature 436, 740-744.
Chiang, H.R., Schoenfeld, L.W., Ruby, J.G., Auyeung, V.C., Spies, N., Baek, D., Johnston,
W.K., Russ, C., Luo, S., Babiarz, J.E., et al. (2010). Mammalian microRNAs: experimental
evaluation of novel and previously annotated genes. Genes Dev 24, 992-1009.
144
Chong, M.M., Zhang, G., Cheloufi, S., Neubert, T.A., Hannon, G.J., and Littman, D.R. (2010).
Canonical and alternate functions of the microRNA biogenesis machinery. Genes Dev 24,
1951-1960.
Chung, W.J., Agius, P., Westholm, J.O., Chen, M., Okamura, K., Robine, N., Leslie, C.S., and
Lai, E.C. (2011). Computational and experimental identification of mirtrons in Drosophila
melanogaster and Caenorhabditis elegans. Genome Res 21, 286-300.
Cifuentes, D., Xue, H., Taylor, D.W., Patnode, H., Mishima, Y., Cheloufi, S., Ma, E., Mane, S.,
Hannon, G.J., Lawson, N.D., et al. (2010). A novel miRNA processing pathway independent
of Dicer requires Argonaute2 catalytic activity. Science 328, 1694-1698.
Cimmino, A., Calin, G.A., Fabbri, M., Iorio, M.V., Ferracin, M., Shimizu, M., Wojcik, S.E.,
Aqeilan, R.I., Zupo, S., Dono, M., et al. (2005). miR-15 and miR-16 induce apoptosis by
targeting BCL2. Proc Natl Acad Sci U S A 102, 13944-13949.
Costinean, S., Zanesi, N., Pekarsky, Y., Tili, E., Volinia, S., Heerema, N., and Croce, C.M.
(2006). Pre-B cell proliferation and lymphoblastic leukemia/high-grade lymphoma in E(mu)miR155 transgenic mice. Proc Natl Acad Sci U S A 103, 7024-7029.
Cowin, P.A., Anglesio, M., Etemadmoghadam, D., and Bowtell, D.D. (2010). Profiling the
cancer genome. Annu Rev Genomics Hum Genet 11, 133-159.
Czech, B., Malone, C.D., Zhou, R., Stark, A., Schlingeheyde, C., Dus, M., Perrimon, N., Kellis,
M., Wohlschlegel, J.A., Sachidanandam, R., et al. (2008). An endogenous small interfering
RNA pathway in Drosophila. Nature 453, 798-802.
Czech, B., Zhou, R., Erlich, Y., Brennecke, J., Binari, R., Villalta, C., Gordon, A., Perrimon, N.,
and Hannon, G.J. (2009). Hierarchical rules for Argonaute loading in Drosophila. Mol Cell
36, 445-456.
Daniels, D.L., Subbarao, M.N., Blattner, F.R., and Lozeron, H.A. (1988). Q-mediated late gene
transcription of bacteriophage lambda: RNA start point and RNase III processing sites in
vivo. Virology 167, 568-577.
Davis, B.N., Hilyard, A.C., Lagna, G., and Hata, A. (2008). SMAD proteins control DROSHAmediated microRNA maturation. Nature 454, 56-61.
Davis, B.N., Hilyard, A.C., Nguyen, P.H., Lagna, G., and Hata, A. (2010). Smad proteins bind a
conserved RNA sequence to promote microRNA maturation by Drosha. Mol Cell 39, 373384.
Denli, A.M., Tops, B.B., Plasterk, R.H., Ketting, R.F., and Hannon, G.J. (2004). Processing of
primary microRNAs by the Microprocessor complex. Nature 432, 231-235.
Doench, J.G., and Sharp, P.A. (2004). Specificity of microRNA target selection in translational
repression. Genes Dev 18, 504-511.
Dong, Z., Han, M.H., and Fedoroff, N. (2008). The RNA-binding proteins HYL1 and SE
promote accurate in vitro processing of pri-miRNA by DCL1. Proc Natl Acad Sci U S A 105,
9970-9975.
Dreyfuss, G., Matunis, M.J., Pinol-Roma, S., and Burd, C.G. (1993). hnRNP proteins and the
biogenesis of mRNA. Annu Rev Biochem 62, 289-321.
Duan, R., Pak, C., and Jin, P. (2007). Single nucleotide polymorphism associated with mature
miR-125a alters the processing of pri-miRNA. Hum Mol Genet 16, 1124-1131.
Dunn, J.J., and Studier, F.W. (1973a). T7 early RNAs and Escherichia coli ribosomal RNAs are
cut from large precursor RNAs in vivo by ribonuclease 3. Proc Natl Acad Sci U S A 70,
3296-3300.
145
Dunn, J.J., and Studier, F.W. (1973b). T7 early RNAs are generated by site-specific cleavages.
Proc Natl Acad Sci U S A 70, 1559-1563.
Eis, P.S., Tam, W., Sun, L., Chadburn, A., Li, Z., Gomez, M.F., Lund, E., and Dahlberg, J.E.
(2005). Accumulation of miR-155 and BIC RNA in human B cell lymphomas. Proc Natl
Acad Sci U S A 102, 3627-3632.
Elbashir, S.M., Harborth, J., Lendeckel, W., Yalcin, A., Weber, K., and Tuschl, T. (2001a).
Duplexes of 21-nucleotide RNAs mediate RNA interference in cultured mammalian cells.
Nature 411, 494-498.
Elbashir, S.M., Lendeckel, W., and Tuschl, T. (2001b). RNA interference is mediated by 21- and
22-nucleotide RNAs. Genes Dev 15, 188-200.
Elela, S.A., Igel, H., and Ares, M., Jr. (1996). RNase III cleaves eukaryotic preribosomal RNA at
a U3 snoRNP-dependent site. Cell 85, 115-124.
Ellington, A.D., and Szostak, J.W. (1990). In vitro selection of RNA molecules that bind specific
ligands. Nature 346, 818-822.
Esquela-Kerscher, A., Trang, P., Wiggins, J.F., Patrawala, L., Cheng, A., Ford, L., Weidhaas,
J.B., Brown, D., Bader, A.G., and Slack, F.J. (2008). The let-7 microRNA reduces tumor
growth in mouse models of lung cancer. Cell Cycle 7, 759-764.
Fagard, M., Boutet, S., Morel, J.B., Bellini, C., and Vaucheret, H. (2000). AGO1, QDE-2, and
RDE-1 are related proteins required for post-transcriptional gene silencing in plants, quelling
in fungi, and RNA interference in animals. Proc Natl Acad Sci U S A 97, 11650-11654.
Faller, M., Toso, D., Matsunaga, M., Atanasov, I., Senturia, R., Chen, Y., Zhou, Z.H., and Guo,
F. (2010). DGCR8 recognizes primary transcripts of microRNAs through highly cooperative
binding and formation of higher-order structures. RNA 16, 1570-1583.
Farh, K.K., Grimson, A., Jan, C., Lewis, B.P., Johnston, W.K., Lim, L.P., Burge, C.B., and
Bartel, D.P. (2005). The widespread impact of mammalian MicroRNAs on mRNA repression
and evolution. Science 310, 1817-1821.
Fellmann, C., Zuber, J., McJunkin, K., Chang, K., Malone, C.D., Dickins, R.A., Xu, Q.,
Hengartner, M.O., Elledge, S.J., Hannon, G.J., et al. (2011). Functional identification of
optimized RNAi triggers using a massively parallel sensor assay. Mol Cell 41, 733-746.
Feng, Y., Zhang, X., Song, Q., Li, T., and Zeng, Y. (2011). Drosha processing controls the
specificity and efficiency of global microRNA expression. Biochim Biophys Acta 1809, 700707.
Filippov, V., Solovyev, V., Filippova, M., and Gill, S.S. (2000). A novel type of RNase III
family proteins in eukaryotes. Gene 245, 213-221.
Fire, A., Albertson, D., Harrison, S.W., and Moerman, D.G. (1991). Production of antisense
RNA leads to effective and specific inhibition of gene expression in C. elegans muscle.
Development 113, 503-514.
Fire, A., Xu, S., Montgomery, M.K., Kostas, S.A., Driver, S.E., and Mello, C.C. (1998). Potent
and specific genetic interference by double-stranded RNA in Caenorhabditis elegans. Nature
391, 806-811.
Flynt, A.S., Greimann, J.C., Chung, W.J., Lima, C.D., and Lai, E.C. (2010). MicroRNA
biogenesis via splicing and exosome-mediated trimming in Drosophila. Mol Cell 38, 900907.
Forstemann, K., Horwich, M.D., Wee, L., Tomari, Y., and Zamore, P.D. (2007). Drosophila
microRNAs are sorted into functionally distinct argonaute complexes after production by
dicer-1. Cell 130, 287-297.
146
Forstemann, K., Tomari, Y., Du, T., Vagin, V.V., Denli, A.M., Bratu, D.P., Klattenhoff, C.,
Theurkauf, W.E., and Zamore, P.D. (2005). Normal microRNA maturation and germ-line
stem cell maintenance requires Loquacious, a double-stranded RNA-binding domain protein.
PLoS Biol 3, e236.
Frank, F., Sonenberg, N., and Nagar, B. (2010). Structural basis for 5'-nucleotide base-specific
recognition of guide RNA by human AGO2. Nature 465, 818-822.
Friedman, R.C., Farh, K.K., Burge, C.B., and Bartel, D.P. (2009). Most mammalian mRNAs are
conserved targets of microRNAs. Genome Res 19, 92-105.
Fujita, P.A., Rhead, B., Zweig, A.S., Hinrichs, A.S., Karolchik, D., Cline, M.S., Goldman, M.,
Barber, G.P., Clawson, H., Coelho, A., et al. (2011). The UCSC Genome Browser database:
update 2011. Nucleic Acids Res 39, D876-882.
Fukuda, T., Yamagata, K., Fujiyama, S., Matsumoto, T., Koshida, I., Yoshimura, K., Mihara, M.,
Naitou, M., Endoh, H., Nakamura, T., et al. (2007). DEAD-box RNA helicase subunits of the
Drosha complex are required for processing of rRNA and a subset of microRNAs. Nat Cell
Biol 9, 604-611.
Futreal, P.A., Coin, L., Marshall, M., Down, T., Hubbard, T., Wooster, R., Rahman, N., and
Stratton, M.R. (2004). A census of human cancer genes. Nat Rev Cancer 4, 177-183.
Gan, J., Shaw, G., Tropea, J.E., Waugh, D.S., Court, D.L., and Ji, X. (2008). A stepwise model
for double-stranded RNA processing by ribonuclease III. Mol Microbiol 67, 143-154.
Gan, J., Tropea, J.E., Austin, B.P., Court, D.L., Waugh, D.S., and Ji, X. (2006). Structural insight
into the mechanism of double-stranded RNA processing by ribonuclease III. Cell 124, 355366.
Garcia, D.M., Baek, D., Shin, C., Bell, G.W., Grimson, A., and Bartel, D.P. (2011). Weak seedpairing stability and high target-site abundance decrease the proficiency of lsy-6 and other
microRNAs. Nat Struct Mol Biol 18, 1139-1146.
Garzon, R., Calin, G.A., and Croce, C.M. (2009). MicroRNAs in Cancer. Annu Rev Med 60,
167-179.
Ghildiyal, M., Xu, J., Seitz, H., Weng, Z., and Zamore, P.D. (2010). Sorting of Drosophila small
silencing RNAs partitions microRNA* strands into the RNA interference pathway. RNA 16,
43-56.
Gottwein, E., Cai, X., and Cullen, B.R. (2006). A novel assay for viral microRNA function
identifies a single nucleotide polymorphism that affects Drosha processing. J Virol 80, 53215326.
Grad, Y., Aach, J., Hayes, G.D., Reinhart, B.J., Church, G.M., Ruvkun, G., and Kim, J. (2003).
Computational and experimental identification of C. elegans microRNAs. Mol Cell 11, 12531263.
Greenbaum, N.L., Radhakrishnan, I., Patel, D.J., and Hirsh, D. (1996). Solution structure of the
donor site of a trans-splicing RNA. Structure 4, 725-733.
Gregory, R.I., Yan, K.P., Amuthan, G., Chendrimada, T., Doratotaj, B., Cooch, N., and
Shiekhattar, R. (2004). The Microprocessor complex mediates the genesis of microRNAs.
Nature 432, 235-240.
Griffiths-Jones, S., Grocock, R.J., van Dongen, S., Bateman, A., and Enright, A.J. (2006).
miRBase: microRNA sequences, targets and gene nomenclature. Nucleic Acids Res 34,
D140-144.
147
Grimson, A., Farh, K.K., Johnston, W.K., Garrett-Engele, P., Lim, L.P., and Bartel, D.P. (2007).
MicroRNA targeting specificity in mammals: determinants beyond seed pairing. Mol Cell
27, 91-105.
Grimson, A., Srivastava, M., Fahey, B., Woodcroft, B.J., Chiang, H.R., King, N., Degnan, B.M.,
Rokhsar, D.S., and Bartel, D.P. (2008). Early origins and evolution of microRNAs and Piwiinteracting RNAs in animals. Nature 455, 1193-1197.
Grishok, A., Pasquinelli, A.E., Conte, D., Li, N., Parrish, S., Ha, I., Baillie, D.L., Fire, A.,
Ruvkun, G., and Mello, C.C. (2001). Genes and mechanisms related to RNA interference
regulate expression of the small temporal RNAs that control C. elegans developmental
timing. Cell 106, 23-34.
Guil, S., and Caceres, J.F. (2007). The multifunctional RNA-binding protein hnRNP A1 is
required for processing of miR-18a. Nat Struct Mol Biol 14, 591-596.
Guo, H., Ingolia, N.T., Weissman, J.S., and Bartel, D.P. (2010). Mammalian microRNAs
predominantly act to decrease target mRNA levels. Nature 466, 835-840.
Guo, S., and Kemphues, K.J. (1995). par-1, a gene required for establishing polarity in C.
elegans embryos, encodes a putative Ser/Thr kinase that is asymmetrically distributed. Cell
81, 611-620.
Guttman, M., Donaghey, J., Carey, B.W., Garber, M., Grenier, J.K., Munson, G., Young, G.,
Lucas, A.B., Ach, R., Bruhn, L., et al. (2011). lincRNAs act in the circuitry controlling
pluripotency and differentiation. Nature 477, 295-300.
Gwizdek, C., Ossareh-Nazari, B., Brownawell, A.M., Doglio, A., Bertrand, E., Macara, I.G., and
Dargemont, C. (2003). Exportin-5 mediates nuclear export of minihelix-containing RNAs. J
Biol Chem 278, 5505-5508.
Haase, A.D., Jaskiewicz, L., Zhang, H., Laine, S., Sack, R., Gatignol, A., and Filipowicz, W.
(2005). TRBP, a regulator of cellular PKR and HIV-1 virus expression, interacts with Dicer
and functions in RNA silencing. EMBO Rep 6, 961-967.
Hagan, J.P., Piskounova, E., and Gregory, R.I. (2009). Lin28 recruits the TUTase Zcchc11 to
inhibit let-7 maturation in mouse embryonic stem cells. Nat Struct Mol Biol 16, 1021-1025.
Hamilton, A.J., and Baulcombe, D.C. (1999). A species of small antisense RNA in
posttranscriptional gene silencing in plants. Science 286, 950-952.
Hammond, S.M., Bernstein, E., Beach, D., and Hannon, G.J. (2000). An RNA-directed nuclease
mediates post-transcriptional gene silencing in Drosophila cells. Nature 404, 293-296.
Hammond, S.M., Boettcher, S., Caudy, A.A., Kobayashi, R., and Hannon, G.J. (2001).
Argonaute2, a link between genetic and biochemical analyses of RNAi. Science 293, 11461150.
Han, B.W., Hung, J.H., Weng, Z., Zamore, P.D., and Ameres, S.L. (2011). The 3'-to-5'
Exoribonuclease Nibbler Shapes the 3' Ends of MicroRNAs Bound to Drosophila
Argonaute1. Curr Biol.
Han, J., Lee, Y., Yeom, K.H., Kim, Y.K., Jin, H., and Kim, V.N. (2004). The Drosha-DGCR8
complex in primary microRNA processing. Genes Dev 18, 3016-3027.
Han, J., Lee, Y., Yeom, K.H., Nam, J.W., Heo, I., Rhee, J.K., Sohn, S.Y., Cho, Y., Zhang, B.T.,
and Kim, V.N. (2006). Molecular basis for the recognition of primary microRNAs by the
Drosha-DGCR8 complex. Cell 125, 887-901.
Han, J., Pedersen, J.S., Kwon, S.C., Belair, C.D., Kim, Y.K., Yeom, K.H., Yang, W.Y.,
Haussler, D., Blelloch, R., and Kim, V.N. (2009). Posttranscriptional crossregulation
between Drosha and DGCR8. Cell 136, 75-84.
148
Hanna, M.M. (1989). Photoaffinity cross-linking methods for studying RNA-protein interactions.
Methods Enzymol 180, 383-409.
Hannon, G.J., Maroney, P.A., Denker, J.A., and Nilsen, T.W. (1990). Trans splicing of nematode
pre-messenger RNA in vitro. Cell 61, 1247-1255.
Hartig, J.V., Esslinger, S., Bottcher, R., Saito, K., and Forstemann, K. (2009). Endo-siRNAs
depend on a new isoform of loquacious and target artificially introduced, high-copy
sequences. EMBO J 28, 2932-2944.
Hata, A., and Davis, B.N. (2011). Regulation of pri-miRNA Processing Through Smads. Adv
Exp Med Biol 700, 15-27.
Hendrickson, D.G., Hogan, D.J., McCullough, H.L., Myers, J.W., Herschlag, D., Ferrell, J.E.,
and Brown, P.O. (2009). Concordant regulation of translation and mRNA abundance for
hundreds of targets of a human microRNA. PLoS Biol 7, e1000238.
Heo, I., Joo, C., Cho, J., Ha, M., Han, J., and Kim, V.N. (2008). Lin28 mediates the terminal
uridylation of let-7 precursor MicroRNA. Mol Cell 32, 276-284.
Heo, I., Joo, C., Kim, Y.K., Ha, M., Yoon, M.J., Cho, J., Yeom, K.H., Han, J., and Kim, V.N.
(2009). TUT4 in concert with Lin28 suppresses microRNA biogenesis through premicroRNA uridylation. Cell 138, 696-708.
Hofacker, I.L., and Stadler, P.F. (2006). Memory efficient folding algorithms for circular RNA
secondary structures. Bioinformatics 22, 1172-1176.
Holterman, M., van der Wurff, A., van den Elsen, S., van Megen, H., Bongers, T., Holovachov,
O., Bakker, J., and Helder, J. (2006). Phylum-wide analysis of SSU rDNA reveals deep
phylogenetic relationships among nematodes and accelerated evolution toward crown Clades.
Mol Biol Evol 23, 1792-1800.
Hughes, J.A., Brown, L.R., and Ferro, A.J. (1987). Nucleotide sequence and analysis of the
coliphage T3 S-adenosylmethionine hydrolase gene and its surrounding ribonuclease III
processing sites. Nucleic Acids Res 15, 717-729.
Hutvagner, G., McLachlan, J., Pasquinelli, A.E., Balint, E., Tuschl, T., and Zamore, P.D. (2001).
A cellular function for the RNA-interference enzyme Dicer in the maturation of the let-7
small temporal RNA. Science 293, 834-838.
Izant, J.G., and Weintraub, H. (1984). Inhibition of thymidine kinase gene expression by antisense RNA: a molecular approach to genetic analysis. Cell 36, 1007-1015.
Jacobsen, S.E., Running, M.P., and Meyerowitz, E.M. (1999). Disruption of an RNA
helicase/RNAse III gene in Arabidopsis causes unregulated cell division in floral meristems.
Development 126, 5231-5243.
Jazdzewski, K., Murray, E.L., Franssila, K., Jarzab, B., Schoenberg, D.R., and de la Chapelle, A.
(2008). Common SNP in pre-miR-146a decreases mature miR expression and predisposes to
papillary thyroid carcinoma. Proc Natl Acad Sci U S A 105, 7269-7274.
Jiang, F., Ye, X., Liu, X., Fincher, L., McKearin, D., and Liu, Q. (2005). Dicer-1 and R3D1-L
catalyze microRNA maturation in Drosophila. Genes Dev 19, 1674-1679.
Jolma, A., Kivioja, T., Toivonen, J., Cheng, L., Wei, G., Enge, M., Taipale, M., Vaquerizas,
J.M., Yan, J., Sillanpaa, M.J., et al. (2010). Multiplexed massively parallel SELEX for
characterization of human transcription factor binding specificities. Genome Res 20, 861873.
Jones-Rhoades, M.W., Bartel, D.P., and Bartel, B. (2006). MicroRNAS and their regulatory roles
in plants. Annu Rev Plant Biol 57, 19-53.
149
Jung, H., Yoon, B.C., and Holt, C.E. (2012). Axonal mRNA localization and local protein
synthesis in nervous system assembly, maintenance and repair. Nat Rev Neurosci 13, 308324.
Kadener, S., Rodriguez, J., Abruzzi, K.C., Khodor, Y.L., Sugino, K., Marr, M.T., 2nd, Nelson,
S., and Rosbash, M. (2009). Genome-wide identification of targets of the droshapasha/DGCR8 complex. RNA 15, 537-545.
Kawahara, Y., Megraw, M., Kreider, E., Iizasa, H., Valente, L., Hatzigeorgiou, A.G., and
Nishikura, K. (2008). Frequency and fate of microRNA editing in human brain. Nucleic
Acids Res 36, 5270-5280.
Kawahara, Y., Zinshteyn, B., Chendrimada, T.P., Shiekhattar, R., and Nishikura, K. (2007).
RNA editing of the microRNA-151 precursor blocks cleavage by the Dicer-TRBP complex.
EMBO Rep 8, 763-769.
Kawamata, T., Seitz, H., and Tomari, Y. (2009). Structural determinants of miRNAs for RISC
loading and slicer-independent unwinding. Nat Struct Mol Biol 16, 953-960.
Kawamata, T., Yoda, M., and Tomari, Y. (2011). Multilayer checkpoints for microRNA
authenticity during RISC assembly. EMBO Rep 12, 944-949.
Ketting, R.F. (2011). The many faces of RNAi. Dev Cell 20, 148-161.
Ketting, R.F., Fischer, S.E., Bernstein, E., Sijen, T., Hannon, G.J., and Plasterk, R.H. (2001).
Dicer functions in RNA interference and in synthesis of small RNA involved in
developmental timing in C. elegans. Genes Dev 15, 2654-2659.
Khalil, A.M., Guttman, M., Huarte, M., Garber, M., Raj, A., Rivea Morales, D., Thomas, K.,
Presser, A., Bernstein, B.E., van Oudenaarden, A., et al. (2009). Many human large
intergenic noncoding RNAs associate with chromatin-modifying complexes and affect gene
expression. Proc Natl Acad Sci U S A 106, 11667-11672.
Khvorova, A., Reynolds, A., and Jayasena, S.D. (2003). Functional siRNAs and miRNAs exhibit
strand bias. Cell 115, 209-216.
Knight, S.W., and Bass, B.L. (2001). A role for the RNase III enzyme DCR-1 in RNA
interference and germ line development in Caenorhabditis elegans. Science 293, 2269-2271.
Kota, J., Chivukula, R.R., O'Donnell, K.A., Wentzel, E.A., Montgomery, C.L., Hwang, H.W.,
Chang, T.C., Vivekanandan, P., Torbenson, M., Clark, K.R., et al. (2009). Therapeutic
microRNA delivery suppresses tumorigenesis in a murine liver cancer model. Cell 137,
1005-1017.
Kozomara, A., and Griffiths-Jones, S. (2011). miRBase: integrating microRNA annotation and
deep-sequencing data. Nucleic Acids Res 39, D152-157.
Krinke, L., and Wulff, D.L. (1990). The cleavage specificity of RNase III. Nucleic Acids Res 18,
4809-4815.
Kumar, M.S., Erkeland, S.J., Pester, R.E., Chen, C.Y., Ebert, M.S., Sharp, P.A., and Jacks, T.
(2008). Suppression of non-small cell lung tumor development by the let-7 microRNA
family. Proc Natl Acad Sci U S A 105, 3903-3908.
Kumar, M.S., Lu, J., Mercer, K.L., Golub, T.R., and Jacks, T. (2007). Impaired microRNA
processing enhances cellular transformation and tumorigenesis. Nat Genet 39, 673-677.
Kurihara, Y., and Watanabe, Y. (2004). Arabidopsis micro-RNA biogenesis through Dicer-like 1
protein functions. Proc Natl Acad Sci U S A 101, 12753-12758.
Lagos-Quintana, M., Rauhut, R., Lendeckel, W., and Tuschl, T. (2001). Identification of novel
genes coding for small expressed RNAs. Science 294, 853-858.
150
Lamontagne, B., and Elela, S.A. (2004). Evaluation of the RNA determinants for bacterial and
yeast RNase III binding and cleavage. J Biol Chem 279, 2231-2241.
Lamontagne, B., Ghazal, G., Lebars, I., Yoshizawa, S., Fourmy, D., and Elela, S.A. (2003).
Sequence dependence of substrate recognition and cleavage by yeast RNase III. J Mol Biol
327, 985-1000.
Landgraf, P., Rusu, M., Sheridan, R., Sewer, A., Iovino, N., Aravin, A., Pfeffer, S., Rice, A.,
Kamphorst, A.O., Landthaler, M., et al. (2007). A mammalian microRNA expression atlas
based on small RNA library sequencing. Cell 129, 1401-1414.
Landthaler, M., Yalcin, A., and Tuschl, T. (2004). The human DiGeorge syndrome critical
region gene 8 and Its D. melanogaster homolog are required for miRNA biogenesis. Curr
Biol 14, 2162-2167.
Lasda, E.L., Allen, M.A., and Blumenthal, T. (2010). Polycistronic pre-mRNA processing in
vitro: snRNP and pre-mRNA role reversal in trans-splicing. Genes Dev 24, 1645-1658.
Lau, N.C., Lim, L.P., Weinstein, E.G., and Bartel, D.P. (2001). An abundant class of tiny RNAs
with probable regulatory roles in Caenorhabditis elegans. Science 294, 858-862.
Lau, P.W., Guiley, K.Z., De, N., Potter, C.S., Carragher, B., and MacRae, I.J. (2012). The
molecular architecture of human Dicer. Nat Struct Mol Biol 19, 436-440.
Lee, R.C., and Ambros, V. (2001). An extensive class of small RNAs in Caenorhabditis elegans.
Science 294, 862-864.
Lee, R.C., Feinbaum, R.L., and Ambros, V. (1993). The C. elegans heterochronic gene lin-4
encodes small RNAs with antisense complementarity to lin-14. Cell 75, 843-854.
Lee, Y., Ahn, C., Han, J., Choi, H., Kim, J., Yim, J., Lee, J., Provost, P., Radmark, O., Kim, S.,
et al. (2003). The nuclear RNase III Drosha initiates microRNA processing. Nature 425, 415419.
Lee, Y., Hur, I., Park, S.Y., Kim, Y.K., Suh, M.R., and Kim, V.N. (2006). The role of PACT in
the RNA silencing pathway. EMBO J 25, 522-532.
Lee, Y., Jeon, K., Lee, J.T., Kim, S., and Kim, V.N. (2002). MicroRNA maturation: stepwise
processing and subcellular localization. EMBO J 21, 4663-4670.
Lee, Y., Kim, M., Han, J., Yeom, K.H., Lee, S., Baek, S.H., and Kim, V.N. (2004a). MicroRNA
genes are transcribed by RNA polymerase II. EMBO J 23, 4051-4060.
Lee, Y., and Kim, V.N. (2007). In vitro and in vivo assays for the activity of Drosha complex.
Methods Enzymol 427, 89-106.
Lee, Y.S., Nakahara, K., Pham, J.W., Kim, K., He, Z., Sontheimer, E.J., and Carthew, R.W.
(2004b). Distinct roles for Drosophila Dicer-1 and Dicer-2 in the siRNA/miRNA silencing
pathways. Cell 117, 69-81.
Lehrbach, N.J., Armisen, J., Lightfoot, H.L., Murfitt, K.J., Bugaut, A., Balasubramanian, S., and
Miska, E.A. (2009). LIN-28 and the poly(U) polymerase PUP-2 regulate let-7 microRNA
processing in Caenorhabditis elegans. Nat Struct Mol Biol 16, 1016-1020.
Lim, L.P., Glasner, M.E., Yekta, S., Burge, C.B., and Bartel, D.P. (2003a). Vertebrate
microRNA genes. Science 299, 1540.
Lim, L.P., Lau, N.C., Weinstein, E.G., Abdelhakim, A., Yekta, S., Rhoades, M.W., Burge, C.B.,
and Bartel, D.P. (2003b). The microRNAs of Caenorhabditis elegans. Genes Dev 17, 9911008.
Liu, H., D'Andrade, P., Fulmer-Smentek, S., Lorenzi, P., Kohn, K.W., Weinstein, J.N., Pommier,
Y., and Reinhold, W.C. (2010). mRNA and microRNA expression profiles of the NCI-60
integrated with drug activities. Mol Cancer Ther 9, 1080-1091.
151
Liu, J., Carmell, M.A., Rivas, F.V., Marsden, C.G., Thomson, J.M., Song, J.J., Hammond, S.M.,
Joshua-Tor, L., and Hannon, G.J. (2004). Argonaute2 is the catalytic engine of mammalian
RNAi. Science 305, 1437-1441.
Liu, Q., Rand, T.A., Kalidas, S., Du, F., Kim, H.E., Smith, D.P., and Wang, X. (2003). R2D2, a
bridge between the initiation and effector steps of the Drosophila RNAi pathway. Science
301, 1921-1925.
Liu, X., Park, J.K., Jiang, F., Liu, Y., McKearin, D., and Liu, Q. (2007). Dicer-1, but not
Loquacious, is critical for assembly of miRNA-induced silencing complexes. RNA 13, 23242329.
Llave, C., Xie, Z., Kasschau, K.D., and Carrington, J.C. (2002). Cleavage of Scarecrow-like
mRNA targets directed by a class of Arabidopsis miRNA. Science 297, 2053-2056.
Lu, J., Getz, G., Miska, E.A., Alvarez-Saavedra, E., Lamb, J., Peck, D., Sweet-Cordero, A.,
Ebert, B.L., Mak, R.H., Ferrando, A.A., et al. (2005). MicroRNA expression profiles classify
human cancers. Nature 435, 834-838.
Lund, E., Guttinger, S., Calado, A., Dahlberg, J.E., and Kutay, U. (2004). Nuclear export of
microRNA precursors. Science 303, 95-98.
Ma, J.B., Ye, K., and Patel, D.J. (2004). Structural basis for overhang-specific small interfering
RNA recognition by the PAZ domain. Nature 429, 318-322.
MacRae, I.J., Zhou, K., and Doudna, J.A. (2007). Structural determinants of RNA recognition
and cleavage by Dicer. Nat Struct Mol Biol 14, 934-940.
Macrae, I.J., Zhou, K., Li, F., Repic, A., Brooks, A.N., Cande, W.Z., Adams, P.D., and Doudna,
J.A. (2006). Structural basis for double-stranded RNA processing by Dicer. Science 311,
195-198.
Maroney, P.A., Hannon, G.J., and Nilsen, T.W. (1990). Transcription and cap trimethylation of a
nematode spliced leader RNA in a cell-free system. Proc Natl Acad Sci U S A 87, 709-713.
Mateos, J.L., Bologna, N.G., Chorostecki, U., and Palatnik, J.F. (2010). Identification of
microRNA processing determinants by random mutagenesis of Arabidopsis MIR172a
precursor. Curr Biol 20, 49-54.
Mayr, C., Hemann, M.T., and Bartel, D.P. (2007). Disrupting the pairing between let-7 and
Hmga2 enhances oncogenic transformation. Science 315, 1576-1579.
Meignin, C., and Davis, I. (2010). Transmitting the message: intracellular mRNA localization.
Curr Opin Cell Biol 22, 112-119.
Meister, G., Landthaler, M., Patkaniowska, A., Dorsett, Y., Teng, G., and Tuschl, T. (2004).
Human Argonaute2 mediates RNA cleavage targeted by miRNAs and siRNAs. Mol Cell 15,
185-197.
Merritt, W.M., Lin, Y.G., Han, L.Y., Kamat, A.A., Spannuth, W.A., Schmandt, R., Urbauer, D.,
Pennacchio, L.A., Cheng, J.F., Nick, A.M., et al. (2008). Dicer, Drosha, and outcomes in
patients with ovarian cancer. N Engl J Med 359, 2641-2650.
Mian, I.S. (1997). Comparative sequence analysis of ribonucleases HII, III, II PH and D. Nucleic
Acids Res 25, 3187-3195.
Michlewski, G., and Caceres, J.F. (2010). Antagonistic role of hnRNP A1 and KSRP in the
regulation of let-7a biogenesis. Nat Struct Mol Biol 17, 1011-1018.
Michlewski, G., Guil, S., Semple, C.A., and Caceres, J.F. (2008). Posttranscriptional regulation
of miRNAs harboring conserved terminal loops. Mol Cell 32, 383-393.
152
Mishima, Y., and Steitz, J.A. (1995). Site-specific crosslinking of 4-thiouridine-modified human
tRNA(3Lys) to reverse transcriptase from human immunodeficiency virus type I. EMBO J
14, 2679-2687.
Miyoshi, K., Miyoshi, T., and Siomi, H. (2010). Many ways to generate microRNA-like small
RNAs: non-canonical pathways for microRNA production. Mol Genet Genomics 284, 95103.
Miyoshi, K., Okada, T.N., Siomi, H., and Siomi, M.C. (2009). Characterization of the miRNARISC loading complex and miRNA-RISC formed in the Drosophila miRNA pathway. RNA
15, 1282-1291.
Moffat, J., Grueneberg, D.A., Yang, X., Kim, S.Y., Kloepfer, A.M., Hinkle, G., Piqani, B.,
Eisenhaure, T.M., Luo, B., Grenier, J.K., et al. (2006). A lentiviral RNAi library for human
and mouse genes applied to an arrayed viral high-content screen. Cell 124, 1283-1298.
Montgomery, M.K., and Fire, A. (1998). Double-stranded RNA as a mediator in sequencespecific genetic silencing and co-suppression. Trends Genet 14, 255-258.
Moore, M.J. (1999). Joining RNA molecules with T4 DNA ligase. Methods Mol Biol 118, 1119.
Murphy, D., Dancis, B., and Brown, J.R. (2008). The evolution of core proteins involved in
microRNA biogenesis. BMC Evol Biol 8, 92.
Nakamura, T., Canaani, E., and Croce, C.M. (2007). Oncogenic All1 fusion proteins target
Drosha-mediated microRNA processing. Proc Natl Acad Sci U S A 104, 10980-10985.
Nam, J.W., Shin, K.R., Han, J., Lee, Y., Kim, V.N., and Zhang, B.T. (2005). Human microRNA
prediction through a probabilistic co-learning model of sequence and structure. Nucleic
Acids Res 33, 3570-3581.
Nam, Y., Chen, C., Gregory, R.I., Chou, J.J., and Sliz, P. (2011). Molecular Basis for Interaction
of let-7 MicroRNAs with Lin28. Cell.
Napoli, C., Lemieux, C., and Jorgensen, R. (1990). Introduction of a Chimeric Chalcone
Synthase Gene into Petunia Results in Reversible Co-Suppression of Homologous Genes in
trans. Plant Cell 2, 279-289.
Newman, M.A., Thomson, J.M., and Hammond, S.M. (2008). Lin-28 interaction with the Let-7
precursor loop mediates regulated microRNA processing. RNA 14, 1539-1549.
Nishikura, K. (2010). Functions and regulation of RNA editing by ADAR deaminases. Annu
Rev Biochem 79, 321-349.
Nykanen, A., Haley, B., and Zamore, P.D. (2001). ATP requirements and small interfering RNA
structure in the RNA interference pathway. Cell 107, 309-321.
O'Connell, R.M., Rao, D.S., Chaudhuri, A.A., Boldin, M.P., Taganov, K.D., Nicoll, J., Paquette,
R.L., and Baltimore, D. (2008). Sustained expression of microRNA-155 in hematopoietic
stem cells causes a myeloproliferative disorder. J Exp Med 205, 585-594.
Okada, C., Yamashita, E., Lee, S.J., Shibata, S., Katahira, J., Nakagawa, A., Yoneda, Y., and
Tsukihara, T. (2009). A high-resolution structure of the pre-microRNA nuclear export
machinery. Science 326, 1275-1279.
Okamura, K., Chung, W.J., Ruby, J.G., Guo, H., Bartel, D.P., and Lai, E.C. (2008). The
Drosophila hairpin RNA pathway generates endogenous short interfering RNAs. Nature 453,
803-806.
Okamura, K., Hagen, J.W., Duan, H., Tyler, D.M., and Lai, E.C. (2007). The mirtron pathway
generates microRNA-class regulatory RNAs in Drosophila. Cell 130, 89-100.
153
Okamura, K., Liu, N., and Lai, E.C. (2009). Distinct mechanisms for microRNA strand selection
by Drosophila Argonautes. Mol Cell 36, 431-444.
Orom, U.A., Derrien, T., Beringer, M., Gumireddy, K., Gardini, A., Bussotti, G., Lai, F.,
Zytnicki, M., Notredame, C., Huang, Q., et al. (2010). Long noncoding RNAs with enhancerlike function in human cells. Cell 143, 46-58.
Paddison, P.J., Caudy, A.A., Bernstein, E., Hannon, G.J., and Conklin, D.S. (2002). Short hairpin
RNAs (shRNAs) induce sequence-specific silencing in mammalian cells. Genes Dev 16,
948-958.
Pan, T., and Uhlenbeck, O.C. (1992). In vitro selection of RNAs that undergo autolytic cleavage
with Pb2+. Biochemistry 31, 3887-3895.
Park, J.E., Heo, I., Tian, Y., Simanshu, D.K., Chang, H., Jee, D., Patel, D.J., and Kim, V.N.
(2011). Dicer recognizes the 5' end of RNA for efficient and accurate processing. Nature 475,
201-205.
Park, M.Y., Wu, G., Gonzalez-Sulser, A., Vaucheret, H., and Poethig, R.S. (2005). Nuclear
processing and export of microRNAs in Arabidopsis. Proc Natl Acad Sci U S A 102, 36913696.
Park, W., Li, J., Song, R., Messing, J., and Chen, X. (2002). CARPEL FACTORY, a Dicer
homolog, and HEN1, a novel protein, act in microRNA metabolism in Arabidopsis thaliana.
Curr Biol 12, 1484-1495.
Pasquinelli, A.E., Reinhart, B.J., Slack, F., Martindale, M.Q., Kuroda, M.I., Maller, B.,
Hayward, D.C., Ball, E.E., Degnan, B., Muller, P., et al. (2000). Conservation of the
sequence and temporal expression of let-7 heterochronic regulatory RNA. Nature 408, 86-89.
Patel, V.L., Mitra, S., Harris, R., Buxbaum, A.R., Lionnet, T., Brenowitz, M., Girvin, M., Levy,
M., Almo, S.C., Singer, R.H., et al. (2012). Spatial arrangement of an RNA zipcode
identifies mRNAs under post-transcriptional control. Genes Dev 26, 43-53.
Patwardhan, R.P., Hiatt, J.B., Witten, D.M., Kim, M.J., Smith, R.P., May, D., Lee, C., Andrie,
J.M., Lee, S.I., Cooper, G.M., et al. (2012). Massively parallel functional dissection of
mammalian enhancers in vivo. Nat Biotechnol 30, 265-270.
Pertzev, A.V., and Nicholson, A.W. (2006). Characterization of RNA sequence determinants and
antideterminants of processing reactivity for a minimal substrate of Escherichia coli
ribonuclease III. Nucleic Acids Res 34, 3708-3721.
Piskounova, E., Polytarchou, C., Thornton, J.E., LaPierre, R.J., Pothoulakis, C., Hagan, J.P.,
Iliopoulos, D., and Gregory, R.I. (2011). Lin28A and Lin28B inhibit let-7 microRNA
biogenesis by distinct mechanisms. Cell 147, 1066-1079.
Piskounova, E., Viswanathan, S.R., Janas, M., LaPierre, R.J., Daley, G.Q., Sliz, P., and Gregory,
R.I. (2008). Determinants of microRNA processing inhibition by the developmentally
regulated RNA-binding protein Lin28. J Biol Chem 283, 21310-21314.
Pitman, J. (1993). Probability (New York, Springer-Verlag).
Pitt, J.N., and Ferre-D'Amare, A.R. (2010). Rapid construction of empirical RNA fitness
landscapes. Science 330, 376-379.
Portier, C., Dondon, L., Grunberg-Manago, M., and Regnier, P. (1987). The first step in the
functional inactivation of the Escherichia coli polynucleotide phosphorylase messenger is a
ribonuclease III processing at the 5' end. EMBO J 6, 2165-2170.
Regnier, P., and Grunberg-Manago, M. (1989). Cleavage by RNase III in the transcripts of the
met Y-nus-A-infB operon of Escherichia coli releases the tRNA and initiates the decay of the
downstream mRNA. J Mol Biol 210, 293-302.
154
Regnier, P., and Portier, C. (1986). Initiation, attenuation and RNase III processing of transcripts
from the Escherichia coli operon encoding ribosomal protein S15 and polynucleotide
phosphorylase. J Mol Biol 187, 23-32.
Reinhart, B.J., Slack, F.J., Basson, M., Pasquinelli, A.E., Bettinger, J.C., Rougvie, A.E., Horvitz,
H.R., and Ruvkun, G. (2000). The 21-nucleotide let-7 RNA regulates developmental timing
in Caenorhabditis elegans. Nature 403, 901-906.
Reinhart, B.J., Weinstein, E.G., Rhoades, M.W., Bartel, B., and Bartel, D.P. (2002). MicroRNAs
in plants. Genes Dev 16, 1616-1626.
Rinn, J.L., Kertesz, M., Wang, J.K., Squazzo, S.L., Xu, X., Brugmann, S.A., Goodnough, L.H.,
Helms, J.A., Farnham, P.J., Segal, E., et al. (2007). Functional demarcation of active and
silent chromatin domains in human HOX loci by noncoding RNAs. Cell 129, 1311-1323.
Rivas, F.V., Tolia, N.H., Song, J.J., Aragon, J.P., Liu, J., Hannon, G.J., and Joshua-Tor, L.
(2005). Purified Argonaute2 and an siRNA form recombinant human RISC. Nat Struct Mol
Biol 12, 340-349.
Robertson, H.D. (1982). Escherichia coli ribonuclease III cleavage sites. Cell 30, 669-672.
Robertson, H.D., and Dunn, J.J. (1975). Ribonucleic acid processing activity of Escherichia coli
ribonuclease III. J Biol Chem 250, 3050-3056.
Robertson, H.D., Webster, R.E., and Zinder, N.D. (1967). A nuclease specific for doublestranded RNA. Virology 32, 718-719.
Robertson, H.D., Webster, R.E., and Zinder, N.D. (1968). Purification and properties of
ribonuclease III from Escherichia coli. J Biol Chem 243, 82-91.
Romano, N., and Macino, G. (1992). Quelling: transient inactivation of gene expression in
Neurospora crassa by transformation with homologous sequences. Mol Microbiol 6, 33433353.
Ross, D.T., Scherf, U., Eisen, M.B., Perou, C.M., Rees, C., Spellman, P., Iyer, V., Jeffrey, S.S.,
Van de Rijn, M., Waltham, M., et al. (2000). Systematic variation in gene expression patterns
in human cancer cell lines. Nat Genet 24, 227-235.
Rotondo, G., and Frendewey, D. (1996). Purification and characterization of the Pac1
ribonuclease of Schizosaccharomyces pombe. Nucleic Acids Res 24, 2377-2386.
Ruby, J.G., Jan, C., Player, C., Axtell, M.J., Lee, W., Nusbaum, C., Ge, H., and Bartel, D.P.
(2006). Large-scale sequencing reveals 21U-RNAs and additional microRNAs and
endogenous siRNAs in C. elegans. Cell 127, 1193-1207.
Ruby, J.G., Jan, C.H., and Bartel, D.P. (2007a). Intronic microRNA precursors that bypass
Drosha processing. Nature 448, 83-86.
Ruby, J.G., Stark, A., Johnston, W.K., Kellis, M., Bartel, D.P., and Lai, E.C. (2007b). Evolution,
biogenesis, expression, and target predictions of a substantially expanded set of Drosophila
microRNAs. Genome Res 17, 1850-1864.
Ruvkun, G., Wightman, B., and Ha, I. (2004). The 20 years it took to recognize the importance
of tiny RNAs. Cell 116, S93-96, 92 p following S96.
Saetrom, P., Heale, B.S., Snove, O., Jr., Aagaard, L., Alluin, J., and Rossi, J.J. (2007). Distance
constraints between microRNA target sites dictate efficacy and cooperativity. Nucleic Acids
Res 35, 2333-2342.
Saito, K., Ishizuka, A., Siomi, H., and Siomi, M.C. (2005). Processing of pre-microRNAs by the
Dicer-1-Loquacious complex in Drosophila cells. PLoS Biol 3, e235.
Sakurai, K., Furukawa, C., Haraguchi, T., Inada, K., Shiogama, K., Tagawa, T., Fujita, S., Ueno,
Y., Ogata, A., Ito, M., et al. (2011). MicroRNAs miR-199a-5p and -3p target the Brm
155
subunit of SWI/SNF to generate a double-negative feedback loop in a variety of human
cancers. Cancer Res 71, 1680-1689.
Schauer, S.E., Jacobsen, S.E., Meinke, D.W., and Ray, A. (2002). DICER-LIKE1: blind men and
elephants in Arabidopsis development. Trends Plant Sci 7, 487-491.
Schurer, H., Lang, K., Schuster, J., and Morl, M. (2002). A universal method to produce in vitro
transcripts with homogeneous 3' ends. Nucleic Acids Res 30, e56.
Schwarz, D.S., Hutvagner, G., Du, T., Xu, Z., Aronin, N., and Zamore, P.D. (2003). Asymmetry
in the assembly of the RNAi enzyme complex. Cell 115, 199-208.
Seitz, H., Tushir, J.S., and Zamore, P.D. (2011). A 5'-uridine amplifies miRNA/miRNA*
asymmetry in Drosophila by promoting RNA-induced silencing complex formation. Silence
2, 4.
Sempere, L.F., Cole, C.N., McPeek, M.A., and Peterson, K.J. (2006). The phylogenetic
distribution of metazoan microRNAs: insights into evolutionary complexity and constraint. J
Exp Zool B Mol Dev Evol 306, 575-588.
Sharp, P.A., and Burge, C.B. (1997). Classification of introns: U2-type or U12-type. Cell 91,
875-879.
Shi, Y., Wang, Y.F., Jayaraman, L., Yang, H., Massague, J., and Pavletich, N.P. (1998). Crystal
structure of a Smad MH1 domain bound to DNA: insights on DNA binding in TGF-beta
signaling. Cell 94, 585-594.
Shin, C., Nam, J.W., Farh, K.K., Chiang, H.R., Shkumatava, A., and Bartel, D.P. (2010).
Expanding the microRNA targeting code: functional sites with centered pairing. Mol Cell 38,
789-802.
Slattery, M., Riley, T., Liu, P., Abe, N., Gomez-Alcala, P., Dror, I., Zhou, T., Rohs, R., Honig,
B., Bussemaker, H.J., et al. (2011). Cofactor binding evokes latent differences in DNA
binding specificity between Hox proteins. Cell 147, 1270-1282.
Sohn, S.Y., Bae, W.J., Kim, J.J., Yeom, K.H., Kim, V.N., and Cho, Y. (2007). Crystal structure
of human DGCR8 core. Nat Struct Mol Biol 14, 847-853.
Sokilde, R., Kaczkowski, B., Podolska, A., Cirera, S., Gorodkin, J., Moller, S., and Litman, T.
(2011). Global microRNA analysis of the NCI-60 cancer cell panel. Mol Cancer Ther 10,
375-384.
Song, L., Axtell, M.J., and Fedoroff, N.V. (2010). RNA secondary structural determinants of
miRNA precursor processing in Arabidopsis. Curr Biol 20, 37-41.
Sontheimer, E.J. (1994). Site-specific RNA crosslinking with 4-thiouridine. Mol Biol Rep 20,
35-44.
Spingola, M., Grate, L., Haussler, D., and Ares, M., Jr. (1999). Genome-wide bioinformatic and
molecular analysis of introns in Saccharomyces cerevisiae. RNA 5, 221-234.
Stark, A., Brennecke, J., Bushati, N., Russell, R.B., and Cohen, S.M. (2005). Animal
MicroRNAs confer robustness to gene expression and have a significant impact on 3'UTR
evolution. Cell 123, 1133-1146.
Stark, A., Kheradpour, P., Parts, L., Brennecke, J., Hodges, E., Hannon, G.J., and Kellis, M.
(2007). Systematic discovery and characterization of fly microRNAs using 12 Drosophila
genomes. Genome Res 17, 1865-1879.
Steiner, F.A., Hoogstrate, S.W., Okihara, K.L., Thijssen, K.L., Ketting, R.F., Plasterk, R.H., and
Sijen, T. (2007). Structural features of small RNA precursors determine Argonaute loading in
Caenorhabditis elegans. Nat Struct Mol Biol 14, 927-933.
156
Sun, G., Yan, J., Noltner, K., Feng, J., Li, H., Sarkis, D.A., Sommer, S.S., and Rossi, J.J. (2009).
SNPs in human miRNA genes affect biogenesis and function. RNA 15, 1640-1651.
Suzuki, H.I., Yamagata, K., Sugimoto, K., Iwamoto, T., Kato, S., and Miyazono, K. (2009).
Modulation of microRNA processing by p53. Nature 460, 529-533.
Swanson, W.J., and Vacquier, V.D. (2002). The rapid evolution of reproductive proteins. Nat
Rev Genet 3, 137-144.
Tam, W., Ben-Yehuda, D., and Hayward, W.S. (1997). bic, a novel gene activated by proviral
insertions in avian leukosis virus-induced lymphomas, is likely to function through its
noncoding RNA. Mol Cell Biol 17, 1490-1502.
Tang, G., Reinhart, B.J., Bartel, D.P., and Zamore, P.D. (2003). A biochemical framework for
RNA silencing in plants. Genes Dev 17, 49-63.
Tarn, W.Y., and Steitz, J.A. (1996). A novel spliceosome containing U11, U12, and U5 snRNPs
excises a minor class (AT-AC) intron in vitro. Cell 84, 801-811.
Tavazoie, S.F., Alarcon, C., Oskarsson, T., Padua, D., Wang, Q., Bos, P.D., Gerald, W.L., and
Massague, J. (2008). Endogenous human microRNAs that suppress breast cancer metastasis.
Nature 451, 147-152.
Thai, T.H., Calado, D.P., Casola, S., Ansel, K.M., Xiao, C., Xue, Y., Murphy, A., Frendewey,
D., Valenzuela, D., Kutok, J.L., et al. (2007). Regulation of the germinal center response by
microRNA-155. Science 316, 604-608.
Tomari, Y., Du, T., and Zamore, P.D. (2007). Sorting of Drosophila small silencing RNAs. Cell
130, 299-308.
Trabucchi, M., Briata, P., Filipowicz, W., Ramos, A., Gherzi, R., and Rosenfeld, M.G. (2010).
KSRP promotes the maturation of a group of miRNA precursors. Adv Exp Med Biol 700,
36-42.
Trabucchi, M., Briata, P., Garcia-Mayoral, M., Haase, A.D., Filipowicz, W., Ramos, A., Gherzi,
R., and Rosenfeld, M.G. (2009). The RNA-binding protein KSRP promotes the biogenesis of
a subset of microRNAs. Nature 459, 1010-1014.
Tsai, M.C., Manor, O., Wan, Y., Mosammaparast, N., Wang, J.K., Lan, F., Shi, Y., Segal, E.,
and Chang, H.Y. (2010). Long noncoding RNA as modular scaffold of histone modification
complexes. Science 329, 689-693.
Tsutsumi, A., Kawamata, T., Izumi, N., Seitz, H., and Tomari, Y. (2011). Recognition of the premiRNA structure by Drosophila Dicer-1. Nat Struct Mol Biol 18, 1153-1158.
Tuschl, T., Zamore, P.D., Lehmann, R., Bartel, D.P., and Sharp, P.A. (1999). Targeted mRNA
degradation by double-stranded RNA in vitro. Genes Dev 13, 3191-3197.
Ui-Tei, K., Naito, Y., Nishi, K., Juni, A., and Saigo, K. (2008). Thermodynamic stability and
Watson-Crick base pairing in the seed duplex are major determinants of the efficiency of the
siRNA-based off-target effect. Nucleic Acids Res 36, 7100-7109.
Ulitsky, I., Shkumatava, A., Jan, C.H., Sive, H., and Bartel, D.P. (2011). Conserved function of
lincRNAs in vertebrate embryonic development despite rapid sequence evolution. Cell 147,
1537-1550.
van der Krol, A.R., Mur, L.A., Beld, M., Mol, J.N., and Stuitje, A.R. (1990). Flavonoid genes in
petunia: addition of a limited number of gene copies may lead to a suppression of gene
expression. Plant Cell 2, 291-299.
Vermeulen, A., Behlen, L., Reynolds, A., Wolfson, A., Marshall, W.S., Karpilow, J., and
Khvorova, A. (2005). The contributions of dsRNA structure to Dicer specificity and
efficiency. RNA 11, 674-682.
157
Viswanathan, S.R., and Daley, G.Q. (2010). Lin28: A microRNA regulator with a macro role.
Cell 140, 445-449.
Viswanathan, S.R., Daley, G.Q., and Gregory, R.I. (2008). Selective blockade of microRNA
processing by Lin28. Science 320, 97-100.
Viswanathan, S.R., Powers, J.T., Einhorn, W., Hoshida, Y., Ng, T.L., Toffanin, S., O'Sullivan,
M., Lu, J., Phillips, L.A., Lockhart, V.L., et al. (2009). Lin28 promotes transformation and is
associated with advanced human malignancies. Nat Genet 41, 843-848.
Vuppalanchi, D., Coleman, J., Yoo, S., Merianda, T.T., Yadhati, A.G., Hossain, J., Blesch, A.,
Willis, D.E., and Twiss, J.L. (2010). Conserved 3'-untranslated region sequences direct
subcellular localization of chaperone protein mRNAs in neurons. J Biol Chem 285, 1802518038.
Wang, D., Zhang, Z., O'Loughlin, E., Lee, T., Houel, S., O'Carroll, D., Tarakhovsky, A., Ahn,
N.G., and Yi, R. (2012). Quantitative functions of Argonaute proteins in mammalian
development. Genes Dev 26, 693-704.
Wang, K.C., and Chang, H.Y. (2011). Molecular mechanisms of long noncoding RNAs. Mol
Cell 43, 904-914.
Wang, S.L., Yao, H.H., and Qin, Z.H. (2009). Strategies for short hairpin RNA delivery in
cancer gene therapy. Expert Opin Biol Ther 9, 1357-1368.
Wang, Y., Medvid, R., Melton, C., Jaenisch, R., and Blelloch, R. (2007). DGCR8 is essential for
microRNA biogenesis and silencing of embryonic stem cell self-renewal. Nat Genet 39, 380385.
Warf, M.B., Johnson, W.E., and Bass, B.L. (2011). Improved annotation of C. elegans
microRNAs by deep sequencing reveals structures associated with processing by Drosha and
Dicer. RNA 17, 563-577.
Weinberg, D.E., Nakanishi, K., Patel, D.J., and Bartel, D.P. (2011). The inside-out mechanism of
Dicers from budding yeasts. Cell 146, 262-276.
Wightman, B., Ha, I., and Ruvkun, G. (1993). Posttranscriptional regulation of the heterochronic
gene lin-14 by lin-4 mediates temporal pattern formation in C. elegans. Cell 75, 855-862.
Witten, D., Tibshirani, R., Gu, S.G., Fire, A., and Lui, W.O. (2010). Ultra-high throughput
sequencing-based small RNA discovery and discrete statistical biomarker analysis in a
collection of cervical tumours and matched controls. BMC Biol 8, 58.
Wu, H., Henras, A., Chanfreau, G., and Feigon, J. (2004). Structural basis for recognition of the
AGNN tetraloop RNA fold by the double-stranded RNA-binding domain of Rnt1p RNase
III. Proc Natl Acad Sci U S A 101, 8307-8312.
Wu, H., Xu, H., Miraglia, L.J., and Crooke, S.T. (2000). Human RNase III is a 160-kDa protein
involved in preribosomal RNA processing. J Biol Chem 275, 36957-36965.
Wu, H., Yang, P.K., Butcher, S.E., Kang, S., Chanfreau, G., and Feigon, J. (2001). A novel
family of RNA tetraloop structure forms the recognition site for Saccharomyces cerevisiae
RNase III. EMBO J 20, 7240-7249.
Wu, M., Jolicoeur, N., Li, Z., Zhang, L., Fortin, Y., L'Abbe, D., Yu, Z., and Shen, S.H. (2008).
Genetic variations of microRNAs in human cancer and their effects on the expression of
miRNAs. Carcinogenesis 29, 1710-1716.
Wyatt, J.R., Sontheimer, E.J., and Steitz, J.A. (1992). Site-specific cross-linking of mammalian
U5 snRNP to the 5' splice site before the first step of pre-mRNA splicing. Genes Dev 6,
2542-2553.
158
Yanaihara, N., Caplen, N., Bowman, E., Seike, M., Kumamoto, K., Yi, M., Stephens, R.M.,
Okamoto, A., Yokota, J., Tanaka, T., et al. (2006). Unique microRNA molecular profiles in
lung cancer diagnosis and prognosis. Cancer Cell 9, 189-198.
Yang, W., Chendrimada, T.P., Wang, Q., Higuchi, M., Seeburg, P.H., Shiekhattar, R., and
Nishikura, K. (2006). Modulation of microRNA processing and expression through RNA
editing by ADAR deaminases. Nat Struct Mol Biol 13, 13-21.
Yekta, S., Shih, I.H., and Bartel, D.P. (2004). MicroRNA-directed cleavage of HOXB8 mRNA.
Science 304, 594-596.
Yeom, K.H., Lee, Y., Han, J., Suh, M.R., and Kim, V.N. (2006). Characterization of
DGCR8/Pasha, the essential cofactor for Drosha in primary miRNA processing. Nucleic
Acids Res 34, 4622-4629.
Yi, R., Qin, Y., Macara, I.G., and Cullen, B.R. (2003). Exportin-5 mediates the nuclear export of
pre-microRNAs and short hairpin RNAs. Genes Dev 17, 3011-3016.
Yoda, M., Kawamata, T., Paroo, Z., Ye, X., Iwasaki, S., Liu, Q., and Tomari, Y. (2010). ATPdependent human RISC assembly pathways. Nat Struct Mol Biol 17, 17-23.
Young, R.A., and Steitz, J.A. (1978). Complementary sequences 1700 nucleotides apart form a
ribonuclease III cleavage site in Escherichia coli ribosomal precursor RNA. Proc Natl Acad
Sci U S A 75, 3593-3597.
Zamore, P.D., Tuschl, T., Sharp, P.A., and Bartel, D.P. (2000). RNAi: double-stranded RNA
directs the ATP-dependent cleavage of mRNA at 21 to 23 nucleotide intervals. Cell 101, 2533.
Zeng, Y., and Cullen, B.R. (2005). Efficient processing of primary microRNA hairpins by
Drosha requires flanking nonstructured RNA sequences. J Biol Chem 280, 27595-27603.
Zeng, Y., Yi, R., and Cullen, B.R. (2005). Recognition and cleavage of primary microRNA
precursors by the nuclear processing enzyme Drosha. EMBO J 24, 138-148.
Zhang, B., Pan, X., Cobb, G.P., and Anderson, T.A. (2007). microRNAs as oncogenes and tumor
suppressors. Dev Biol 302, 1-12.
Zhang, H., Kolb, F.A., Brondani, V., Billy, E., and Filipowicz, W. (2002). Human Dicer
preferentially cleaves dsRNAs at their termini without a requirement for ATP. EMBO J 21,
5875-5885.
Zhang, H., Kolb, F.A., Jaskiewicz, L., Westhof, E., and Filipowicz, W. (2004). Single processing
center models for human Dicer and bacterial RNase III. Cell 118, 57-68.
Zhang, K., and Nicholson, A.W. (1997). Regulation of ribonuclease III processing by doublehelical sequence antideterminants. Proc Natl Acad Sci U S A 94, 13437-13441.
Zhang, X., and Zeng, Y. (2010). The terminal loop region controls microRNA processing by
Drosha and Dicer. Nucleic Acids Res 38, 7689-7697.
Zhou, R., Czech, B., Brennecke, J., Sachidanandam, R., Wohlschlegel, J.A., Perrimon, N., and
Hannon, G.J. (2009). Processing of Drosophila endo-siRNAs depends on a specific
Loquacious isoform. RNA 15, 1886-1895.
Zykovich, A., Korf, I., and Segal, D.J. (2009). Bind-n-Seq: high-throughput analysis of in vitro
protein-DNA interactions using massively parallel sequencing. Nucleic Acids Res 37, e151.
159
160
Appendix 1.
Experimental protocols
Contents
In Vitro Cleavage Selection for Functional pri-miRNAs ............................................................164
Preparation of Drosha/DGCR8 Overexpression Lysate ..................................................... 164
Overview ................................................................................................................... 164
Cell culture and transfection ..................................................................................... 164
Lysate preparation ..................................................................................................... 165
Assay for microprocessor activity ............................................................................ 165
In vitro cleavage selection of circular pri-miRNA substrates ............................................. 168
Overview ................................................................................................................... 168
Assembly of partially-randomized, circular pri-miRNA substrate ........................... 169
Selection of functional variants from circular pri-miRNA pool ............................... 174
Library preparation for Illumina high-throughput sequencing ................................. 175
In vitro cleavage selection of linear pri-miRNA substrates ................................................ 184
Overview ................................................................................................................... 184
Assembly of linear, partially-randomized, pri-miRNA substrate ............................. 185
Selection of functional variants from linear pri-miRNA pool .................................. 187
Library preparation for Illumina high-throughput sequencing ................................. 188
In Vitro Binding Selection for Functional pri-miRNAs ..............................................................191
Preparation of immunopurified Drosha-TN/DGCR8 ......................................................... 191
Overview ................................................................................................................... 191
Cell culture and transfection ..................................................................................... 191
161
Lysate preparation ..................................................................................................... 192
Immunoprecipitation with anti-FLAG beads ............................................................ 192
In vitro binding selection of linear pri-miRNA substrates.................................................. 194
Overview ................................................................................................................... 194
Assembly of linear, partially-randomized, pri-miRNA substrate ............................. 195
Selection of functional variants from linear pri-miRNA pool .................................. 197
PCR amplification for downstream applications ...................................................... 200
Identification of Motif-Binding Proteins by Site-Specific Crosslinking .....................................203
Overview ............................................................................................................................. 203
Assembly of 4-thiouridine containing RNA substrate ........................................................ 204
RNA-protein crosslinking and complex analysis ............................................................... 205
Optimization of 365 nm UV dose ............................................................................. 205
Large-scale purification of crosslinked complexes for analysis ............................... 207
Candidate protein testing by immunoprecipitation of crosslinked complexes ......... 209
Supplement 1: Standard Protocols ...............................................................................................211
Crush-and-soak technique for acrylamide gel extraction ................................................... 211
T7 transcription (“Midi” scale) ........................................................................................... 212
Phosphorylation of nucleic acid 5′ Ends ............................................................................. 213
Cold ........................................................................................................................... 213
Hot............................................................................................................................. 213
Dephosphorylation of nucleic acid 5′ ends ......................................................................... 214
Dephosphorylation of 2′-3′ cyclic phosphates and 3′ phosphates ....................................... 214
Splint ligation of RNA ........................................................................................................ 215
Using T4 RNA Ligase 2 (RNL2) .............................................................................. 215
Partial hydrolysis of RNA................................................................................................... 216
162
Supplement 2: Standard Reagents ...............................................................................................217
Buffers................................................................................................................................. 217
Commercial products .......................................................................................................... 219
163
In Vitro Cleavage Selection for Functional pri-miRNAs
Preparation of Drosha/DGCR8 Overexpression Lysate
Overview
293T cells are transiently transfected with plasmids encoding C-terminal FLAG tagged human
Drosha and N-terminal FLAG-HA tagged human DGCR8. Cells are harvested and sonicated.
After centrifuge clearing of precipitates and insoluble material, the whole-cell lysate is stored in
single-use aliquots. This protocol is modified from Lee and Kim, Meth. Enz. 427 (2007), with
additional input from Jinju Han and V. Narry Kim.
Plasmids:
•
•
•
pMAX-GFP (transfection marker)
pFLAG-HA-DGCR8 (courtesy of T. Tuschl)
pCK-Drosha-FLAG (courtesy of V. Narry Kim)
Cell culture and transfection
Maintain 293T cells under standard conditions. We use Dulbecco’s Modified Eagle Medium
(DMEM) supplemented to 10% Fetal Bovine Serum.
To transfect cells, start with three 15-cm dishes at 90-100% confluence. Split all of these cells
into eight 15-cm dishes the day before transfection. Use 15 ml media per plate.
The day after splitting (18-24 hours later), begin transfection.
Assemble Lipofectamine dilution in one 50 ml Falcon tube:
720 µl Lipofectamine 2000
24 ml Opti-mem I Serum-Free Media
Incubate at room temperature for 5 min.
Meanwhile, assemble plasmid mixture in another 50 ml Falcon tube:
24 µg pMAX-GFP
60 µg pCK-Drosha-FLAG
60 µg pFLAG-HA-DGCR8
24 ml Opti-mem I Serum-Free Media
Add plasmid mixture to Lipofectamine dilution, and gently invert twice to mix.
complexes to form by incubating at room temperature for 20-30 min.
Allow
After complexes have formed, add 6 ml of mixture dropwise to each 15-cm dish. Rock plate
gently 1-2 times to ensure even dispersal of transfection complexes.
164
12-24 hours after transfection, check for expression of GFP. At this relatively early point, 50%75% of cells will already be expressing GFP.
Split all the transfected cells into twenty 15-cm dishes.
Allow cells to grow for 48 hr. After this period, cells should be at or just past confluence.
Frequently, with a successful transfection containing pMAX-GFP, the monolayer is faintly green
in room lighting. Cells are now ready for harvesting.
Lysate preparation
Harvest overexpressing cells by removing media and pipetting PBS onto the monolayer. Due to
the poor adherence of 293T cells, this is typically enough to dislodge the monolayer. Collect
PBS suspension of cells, and keep on ice.
Pellet PBS suspension by centrifugation. This pellet should be green from GFP in the cells;
there should be little or no visible fluorescence in the supernatant.
Prepare 1X Sonic Buffer + protease inhibitors by dissolving a Mini EDTA-Free Protease
Inhibitor tablet in 10 ml of 1X Sonic Buffer. Keep on ice.
Resuspend cell pellet in 10 ml of 1X Sonic Buffer + protease inhibitors.
Prepare sonicator for use. We use a Branson Sonifier 250 at 50% duty cycle and output level 4.
Clean the probe by sonicating RNase-ZAP solution for 10 pulses (approximately 20 seconds at
50% duty cycle), then sonicating deionized water for 10 pulses.
Lyse 293T cells by sonicating for 10 pulses.
Clear the lysate once by centrifuging lysate at 3500 x g for 15 min. If lysis is successful, the
pellet should be yellow and perhaps light-green, while the supernatant is green. Collect
supernatant.
Clear supernatant once more by centrifuging at 3500 x g for 15 min. Collect supernatant. This is
the cleared whole-cell extract.
Aliquot whole-cell extract into single-use aliquots and store aliquots in liquid nitrogen vapor.
Assay for microprocessor activity
We assess Microprocessor activity by cleavage of a model miRNA substrate, pri-mir-125a.
Cleavage of this substrate appears to be essentially complete at 10 minutes, and shows features
consistent with a single-turnover reaction. Accordingly, we use this substrate to estimate a
functional concentration of Microprocessor complex.
165
Synthesis of pri-mir-125a reference substrate
Amplify a runoff template for T7 transcription by PCR using standard protocols. Human
genomic DNA or a mir-125a expression plasmid can be used as a template for this reaction.
T7::mir-125a Forward Primer:
TAATACGACTCACTATAggTCTCTGACCCCCACCCCAGGG
mir-125a Reverse Primer
ATGAGGAGTCAGGGGTCAGAAGTCAGGCCAGC
See Appendix I for a T7 transcription protocol using the PCR product as a template. The
expected product is below (underlined region is the pre-miRNA product).
Pri-mir-125a reference (187 nt)
ggUCUCUGACCCCCACCCCAGGGUCUACCGGGCCACCGCACACCAUGUUGCCAGUCUCUAGGUCC
CUGAGACCCUUUAACCUGUGAGGACAUCCAGGGUCACAGGUGAGGUUCUUGGGAGCCUGGCGUCU
GGCCCAACCACACACCUGGGGAAUUGCUGGCCUGACUUCUGACCCCUGACUCCUCAU
After purification on urea-polyacrylamide, dephosphorylate the 5′-triphosphorylated substrate
RNA using CIP, according to the standard protocol in Appendix I. After the reaction is
complete, phenol extract the reaction and ethanol precipitate the product; resuspend in deionized
water.
5′-label the substrate using γ-[32P]-ATP and T4 Polynucleotide Kinase, as described in Appendix
I. After the reaction is complete, run over a size-exclusion column such as G-25 or P-30 to
remove unincorporated ATP. Purify again on urea-polyacrylamide. After resuspension, measure
concentration of hot substrate or dilute into a known concentration of cold substrate. We
typically maintain substrates at 1 µM or greater to minimize concentration changes due to tube
wall adsorption.
Titration-timecourse assay for microprocessor activity
Set up a titration of the whole-cell lysate for estimating the effective concentration of
Microprocessor complex:
10X Cleavage Buffer (5 mM Mg)
Microprocessor Lysate
1X Sonic Buffer
1 µM pri-mir-125a Ref. Substrate
1% Rxn
10 µl
1 µl
80 µl
10 µl
100 µl
5% Rxn
10 µl
5 µl
75 µl
10 µl
100 µl
10% Rxn
10 µl
10 µl
70 µl
10 µl
100 µl
20% Rxn
10 µl
20 µl
50 µl
10 µl
100 µl
50% Rxn
10 µl
50 µl
30 µl
10 µl
100 µl
Before adding substrate, prewarm the reaction mixture to 37º. Be sure to add the substrate last,
as addition of the substrate marks timepoint “0.” After addition of substrate, withdraw 20 µl of
166
the reaction at 0 min, 1 min, 3 min, 9 min, and 27 min. Each 20 µl timepoint should be
immediately mixed with at least 100 µl ice-cold Tri-Reagent to stop the reaction.
After extraction from Tri-Reagent and precipitation, load and run a 10% Urea-polyacrylamide
gel. Labeled Decade markers can be used as a size ladder. The expected product size is 61 nt.
Measure the fraction of substrate cleaved at each time point to estimate the concentration of
active enzyme. Results from a representative experiment is shown below. Since the initial
substrate concentration was 100 nM, the Microprocessor appears to cleave hsa-mir-125a with
burst kinetics. We have used this behavior to estimate the concentration of Microprocessor at
about 75 nM.
50
45
40
nM Product
35
30
1% Lysate
25
5% Lysate
20
10% Lysate
15
20% Lysate
10
50% Lysate
5
0
0
5
10
15
20
Time (min)
167
25
30
In vitro cleavage selection of circular pri-miRNA substrates
Overview
The goal of this selection is to explore Microprocessor cleavage determinants in sequences
flanking the pre-miRNA hairpin. The overall strategy is to generate a large pool of variant
molecules and subject this pool to Microprocessor cleavage. Products of cleavage are recovered
and sequenced by high-throughput sequencing to detect motifs that are enriched in the product
(Microprocessor-cleaved) population, relative to the original pool.
A major hindrance to this strategy is that after cleavage of linear pri-miRNA substrates, there are
two flanking RNA products, a 5p (upstream) product, and a 3p (downstream) product.
We would like to clone and sequence both products, maintaining the relationship between
upstream fragment and downstream fragment for any given partially-randomized RNA molecule.
For example, the basal stem of a pri-miRNA hairpin involves pairing between bases in the 5p
product and complementary bases in the 3p product. To detect evidence of covariation, we must
recover and sequence both bases from the 5p and 3p product of the same substrate molecule. We
have accomplished this by circularizing pri-miRNA substrates, and isolating linearized products
for high-throughput sequencing.
168
Assembly of partially-randomized, circular pri-miRNA substrate
The overall strategy is summarized in the following diagram. A T7 template corresponding to
the desired RNA is assembled by PCR, then transcribed. An HDV ribozyme is included in the
sequence to homogenize the 3′ ends; self-cleavage occurs in the transcription reaction. After gel
purification, the 5′ and 3′ ends are treated to produce ligatable ends. The circularization is
achieved using T4 RNA Ligase 1.
Template assembly by PCR
Templates for T7 transcription are assembled from synthetic oligonucleotides. IDT will
synthesize long oligos with both constant sequence and partially randomized positions (for a
price). In addition to the pri-miRNA sequence (both constant and partially-randomized),
sequence corresponding to the HDV ribozyme is also added to the template amplicon.
By way of example, oligonucleotide sequences are provided for the constant (unrandomized)
C125circ template. However, the clonal pool (composed many copies of the wildtype miRNA
sequence) should be prepared in parallel with the partially randomized pool (composed of a few
copies of many variant sequences).
C125circ.001a Left arm
GACTCACTATAGGGTCACAGGTGAGGTTCTTGGGAGCCTGGCGTCTGGCCCAACCACACACCTGGGGAATT
GCTGGCCTGACTTCTGACCCCTGACTCCT
169
C125circ.002a Right arm (ordering orientation)
TCCTCACAGGTTAAAGGGTCTCAGGGACCTAGAGACTGGCAACATGGTGTGCGGTGGCCCGGTAGACCCTG
GGGTGGGGGTATGAGGAGTCAGGGGTCAG
C125circ.003 HDV adaptor
CTTCTCCCTTAGCCTACCGAAGTAGCCCAGGTCGGACCGCGAGGAGGTGGAGATGCCATGCCGACCCTGGA
TGTCCTCACAGGTTAAAGG
C125circ.004 T7 Adaptor
CAGAGATGCATAATACGACTCACTATAGGGTCACAG
HDV 5′ Polishing Primer
CTTCTCCCTTAGCCTACCGAAGTAGCCCAGG
First, PAGE-purify long primers. In general, primers longer than 60 nt are difficult even for
commercial operations to synthesize, and should be purified before use. Resuspend purified
oligos to 100 µM where possible.
An initial primer extension between the left and right primers is used to generate a small amount
of template for PCR. Assemble the following reaction:
15 µl
15 µl
270 µl
40 µl
40 µl
20 µl
400 µl
Left Arm oligo (100 µM)
Right Arm oligo (100 µM)
H2O
10X “House” PCR Buffer
10X “House” dNTP Mix
“House” Taq
Perform a primer extension using the program “L/R Ext.”
30 sec
(ramp)
1 min
1 min
(hold)
95°C
Ramp to 37°C at 0.1°C/sec
37°C
72°C
Repeat for 5 total cycles
10°C
Without purifying or concentrating the primer extension, assemble the following reaction:
30 µl
355 µl
5 µl
5 µl
50 µl
50 µl
5 µl
500 µl
Primer extension reaction
H2O
C125circ.003 HDV Adaptor (100 µM)
C125circ.004 T7 Adaptor (100 µM)
10X “House” PCR Buffer
10X “House” dNTP Mix
“House” Taq
170
Perform an initial PCR amplification using the program “L/R PCR.” Note that the annealing
temperature is based on the amount of homology between the T7 adaptor and the left arm oligo,
and the amount of homology between the HDV adaptor and the right arm oligo.
30 sec
30 sec
30 sec
30 sec
(hold)
95°C
95°C
45°C or appropriate annealing temp.
72°C
Repeat for 4 total cycles
10°C
Perform a second PCR amplification to “polish” the template for use. Frequently, the long HDV
oligo has internal lesions that prevent T7 RNA polymerase from proceeding through the
template. Incomplete transcripts will not have intact HDV ribozyme sequences, and will not
self-cleave to the correct size. A few additional rounds of PCR will selectively amplify
molecules that have fewer lesions.
50 µl
335 µl
5 µl
5 µl
50 µl
50 µl
5 µl
500 µl
Initial PCR reaction
H2O
HDV Polish Primer (100 µM)
C125circ.004 T7 Adaptor (100 µM)
10X “House” PCR Buffer
10X “House” dNTP Mix
“House” Taq
Repeat L/R PCR program. The annealing temperature may be raised based on the melting
temperatures of the HDV Polish Primer and the T7 Adaptor.
Ethanol precipitate the reaction and resuspend in 50 µl H2O.
Transcription of linear pri-miRNA pool
Use T7 RNA Polymerase to transcribe the template prepared in the previous step. A 200 µl
reaction scale should be sufficient for most applications, but more is always possible.
25 µl
85 µl
20 µl
40 µl
10 µl
10 µl
10 µl
200 µl
Concentrated template pool
H2O
10X “House” T7 Buffer
5X “House” NTP Mix
DTT (100 mM)
Fresh α-[32P]-UTP
“House” T7 RNA Polymerase
Incubate 37º x 2-3 hr
171
Add 5 µl Turbo DNAse (Ambion)
Incubate 37º x 30 min
Add 1/20 volume 500 mM EDTA
Add 1/10 volume 3M NaCl
Add 1 volume 100% ethanol
Final:
T7 reaction mixture with whatever NTP/Salt/Mg/PPi is present
~25 mM EDTA
~75 mM NaCl (supplemental)
50% ethanol
Incubate at least 15 min in -20
Spin 15 min at 4º
After precipitation, the reaction should be purified on a 5% Urea-polyacrylamide gel with an
appropriate size marker. The expected products are (1) a full-length transcript that has not been
self-cleaved; (2) the mature, self-cleaved transcript, and (3) the HDV ribozyme RNA after
liberation from the rest of the transcript. Cut out and elute the mature, self-cleaved transcript.
This is the linear pri-miRNA pool.
For the C125circ primers given above, the expected product sequence is given below. The
S125circ partially randomized transcript was twin to this sequence, except that positions colored
red were partially randomized at 79% wildtype (indicated base) and 7% each of the other three
bases. These positions comprise the basal miRNA stem and 35-45 nt of flanking sequence.
ppp-5′GGGTCACAGGTGAGGTTCTTGGGAGCCTGGCGTCTGGCCCAACCACACACCTGGGGAATTGCTGG
CCTGACTTCTGACCCCTGACTCCTCATACCCCCACCCCAGGGTCTACCGGGCCACCGCACACCAT
GTTGCCAGTCTCTAGGTCCCTGAGACCCTTTAACCTGTGAGGACATCCA-[2′-3′
cyclic
phosphate]
Treatment of transcript ends for ligation
After transcription and self-cleavage, the pool molecules have a 5′ triphosphate and a 2′-3′ cyclic
phosphate. The goal of this phase is to convert these ends to a 5′ monophosphate and a 3′-OH,
which are the substrates of T4 RNA Ligase 1.
First, the triphosphate is removed using Calf Intestinal Phosphatase or Alkaline Phosphatase
(CIP). Assemble the following reaction. It is often convenient to reconstitute the pellet from gel
extraction directly in the appropriate amount of water.
172
(pellet)
50 µl
6 µl
4 µl
60 µl
Linear pri-miRNA pool from previous
H2O
NEBuffer 3
CIP
Incubate at 37°C x 1 hr
Phenol-extract the reaction and precipitate. It is often convenient to directly resuspend the pellet
in the next reaction’s buffer.
Next, the 2′-3′ cyclic phosphate is removed using T4 Polynucleotide Kinase, which has this
activity in addition to its kinase activity, in keeping with its role in repairing cleaved tRNA
anticodon loops. Assemble the following:
(pellet)
50 µl
120 µl
10 µl
180 µl
First dephosphorylation from above
H2O
1.5X MES Dephosphorylation Buffer
T4 PNK
Incubate at 37°C x 6 hr
Since PNK is also used to 5′ phosphorylate the RNA, it is not necessary to phenol extract.
Instead, ethanol precipitate the reaction and resuspend it in the phosphorylation reaction mix.
Note: T4 DNA Ligase buffer from NEB is equivalent to T4 PNK Buffer supplemented with 1
mM ATP (final). If using PNK buffer, supplement the reaction to 1 mM ATP (final).
(pellet)
170 µl
20 µl
10 µl
200 µl
Dephosphorylated pri-miRNA pool
H2O
T4 DNA Ligase Buffer
T4 PNK
Incubate at 37°C x 1 hr
Phenol-extract the reaction and precipitate. Resuspend in 340 µl H2O. Half will be used
directly for circularization.
Circularization of pri-miRNA pool using T4 RNA Ligase 1
Consistent with its role in repairing cleaved tRNA anticodon loops, T4 RNA Ligase 1 is
reasonably good at ligating single stranded RNA ends that are held close to each other by
double-stranded RNA. This is the design of the circularization strategy. Assemble the following
reaction:
170 µl
Phosphorylated pri-miRNA pool
Heat to 85°C x 1 min and cool to room temperature
20 µl
T4 RNA Ligase 1 Buffer
10 µl
T4 RNA Ligase 1
200 µl
Incubate at 37°C x 2 hr
173
Purify the ligation product on a 5% urea-polyacrylamide gel along with an appropriate size
ladder. It is frequently convenient to include a small amount of the phosphorylated pri-miRNA
pool as a reference. The circularized RNA runs slightly higher than its linearized form. The
apparent linear RNA from the ligation lane is a mixture of unligated RNA and nicked circular
RNA. Based on experiments using hot 5′ phosphorylated RNA, it appears that much of the
apparently linear RNA is actually nicked circularized RNA, suggesting that the ligation reaction
is relatively efficient, but that hydrolysis occurs during the reaction for reasons not clear. The
nicked RNA does not appear to accumulate with time, indicating that its occurrence is not due to
contaminating activity.
If desired, the circular nature of the substrate can be verified by two methods. First, cleave the
substrate with the Microprocessor to verify that two products form, corresponding to the premiRNA and a single molecule with the flanking sequence. Second, perform a partial hydrolysis
time course from 30 sec to 10 min, and compare the products with the uncut circular substrate
and the phosphorylated, unligated, linear substrate. At early time points, hydrolysis of the
circular substrate should produce a product at the size of the unligated linear substrate. Further
hydrolysis will produce a smear of degradation products.
Selection of functional variants from circular pri-miRNA pool
Cleavage timecourse of clonal wildtype pool and partially-randomized pool
This experiment serves three purposes: (1) verification that the circularized pri-miRNA
substrates are cleavage competent circular molecules, (2) estimation of the overall contribution
of flanking RNA sequence by comparing the cleavage of the partially-randomized pool to the
cleavage of wildtype pri-miRNA, and (3) optimization of cleavage timing for the actual
selection.
Assemble the following reaction, assuming 1 µM stock solutions of RNA and 50 nM
Microprocessor complex in whole-cell lysate. As concentrations from different substrate and
protein preps vary, volumes should be adjusted to match these final concentrations.
39 µl
50 µl
10 µl
1 µl
100 µl
1X Sonic Buffer
Microprocessor Lysate
10X Cleavage/Binding Buffer (1 mM Mg)
Circularized clonal or randomized pool (1 µM)
Prior to addition of the RNA substrate, the reaction mixture should be prewarmed at 37º. After
addition of substrate, withdraw 10 µl at the following times: 0 min, 20 sec, 40 sec, 1 min, 2 min,
4 min, 6 min, 8 min, 10 min, and 12 min. Each timepoint should be mixed immediately with 100
µl of ice-cold Tri-Reagent to stop the reaction.
After extraction from Tri-Reagent, run the reactions on an 8% urea-polyacrylamide gel and
estimate the amount of cleavage at each timepoint for both the clonal (wildtype) substrate and
174
the partially-randomized pool. For miRNAs where the flanking sequence contributes to
recognition and cleavage, the clonal (wildtype) substrate is cleaved 4-5 fold more rapidly than
the partially randomized pool. For miRNAs studied so far, this ratio appears to be constant for
most early time points (between 20 sec and 6 min). When this is the case, we have selected a
timepoint where ~1% of the partially randomized pool and used this timepoint for selection. In
some situations, to achieve 1% cleavage it was necessary to adjust the ratio between
Microprocessor complex and partially-randomized pool substrate.
Selection of functional variants from partially-randomized circular pri-miRNA pool
Using conditions optimized in the previous experiment, perform a selection for functional,
Microprocessor-cleaved variants in the partially-randomized circular pri-miRNA pool. The
reaction should be scaled such that ~100 fmol of cleaved product can be obtained. This
corresponds to a selection for 6x1010 product molecules. After accounting for yield losses in
library preparation, ~108 to 109 molecules will be available for sequencing; as of early 2012, a
single lane of Illumina HiSeq is capable of producing upwards of 108 sequences. Thus, this scale
is the minimum scale needed to fully leverage the current sequencing capacity.
A target of 100 fmol product requires a moderately large cleavage reaction. For example, for
cleavage conditions using a (typical) final reaction concentration of 10 nM pri-miRNA pool, a
1000 µl cleavage reaction will use a total of 10,000 fmol substrate. Cleavage of 1% of this
amount yields 100 fmol of product for library prepation for high-throughput sequencing.
After performing the selection reaction under optimized conditions, phenol-extract the reaction
and precipitate it. The precipitate should be resuspended in the minimum amount of water
needed to fully dissolve the pellet, and the resulting nucleic acid should separated on a 5% ureapolyacrylamide gel. Cut out the band corresponding to the flanking, partially-randomized RNA.
For the S125circ example given above, the product is expected to be 119 nt. Since yield is
valuable, it is advisable to use the crush-and-soak technique to purify this band (see Standard
Protocols). Resuspend in 20 µl water.
Library preparation for Illumina high-throughput sequencing
Library preparation from the selected pool
The goal of this phase is to ligate adaptors to the ends of the product sequence; these adaptors
will permit amplification of the variant molecules, and contain some necessary sequence for the
current (Feb. 2012) Illumina paired-end sequencing technology. We use splint ligation to assist
in ligation of adaptors to the correct arm. Because both T4 RNA Ligase 2 and T4 DNA Ligase
are highly sensitive to gaps or overhangs in the substrate, the splint ligation is highly selective
for defined ends. In general, we have designed splints corresponding to the MirBase annotated
cleavage site, which in turn is usually based on the observed ends by deep-sequencing mature
miRNAs and their * species. However, recent work by the Zamore lab has suggested that the 3p
arm cleavage site annotations may be incorrect, and this view is supported by our experience
with splint ligation. Instead, we recommend that the 5p splint be designed according to the
175
annotated 5p arm cleavage, and the 3p arm splint be designed against an inferred cleavage site
that leaves a 2-nt overhang.
176
3p Arm
5p Arm
p-uggcgucu---------------------ucucuagg-OH
S125circ.006 3p Arm Splint (ordering orientation)
AGACGCCAAGATCGGA
S125circ.007 5p Arm Splint
ACGTGTACCCTAGAGA
S125circ.006 3p Arm Splint (ordering orientation)
AGACGCCAAGATCGGA
S125circ.005 3p Arm Adaptor
S125circ.007.A 5p Arm Adaptor, CAT barcode
GAGATCTACACTCTTTCCCTACACGACGCTCTuccgaucuuggcgucu---------------------ucucuagg guacacguaugaGATCGGAAGAGCGGTTCAGCAGGAATGC
S125circ.007 5p Arm Splint
ACGTGTACCCTAGAGA
(Splints do not cross-hybridize)
S0.002 Solexa Rev Seq, -1 short
CAAGCAGAAGACGGCATACGAGATCGGTCTCGGCATTCCTGCTGAACCGCTCTTCCGATC
S125circ.008 RT Primer (-1)
GCATTCCTGCTGAACCGCTCTTCCGATC
S0.001 Solexa Fwd Seq
AATGATACGGCGACCACCGAGATCTACACTCTTTCCCTACACGACGCTCTTCCGATCT
GAGATCTACACTCTTTCCCTACACGACGCTCTuccgaucuuggcgucu---------------------ucucuaggguacacguaugaGATCGGAAGAGCGGTTCAGCAGGAATGC
On a sequence level, this is the strategy for
ligating and amplifying the S125circ product.
The same adaptors can be used for other miRNA
products, provided that a new splint is designed.
Due to the circular permutation of the substrate, the 5p arm of the miRNA is at the 3′ end, and the
3p arm of the miRNA is at the 5′ end. To avoid confusion, this protocol will continue to use the
5p and 3p arm convention relative to the miRNA sequence. By way of example, after cleavage
and gel purification, the desired product from S125circ selection is as follows. “NNNN” denotes
partially randomized sequence in the S125circ.
p-TGGCGTCTNNNNNNNNNNNNN-------------------------------NNNNNNNNNNNNNTCTCTAGG-OH
There are several features of note. First, as noted above, due to circular permutation, the 5p arm
basal stem sequence (TCTCTAGG) is at the 3′ end of the product molecule, while the 3p arm
basal stem sequence (TGGCGTCT) is at the 5p end of the product. In the intact circular
substrate, the pre-miRNA sequence joined these two ends to form a closed circle. Second, the
partially randomized pool was designed such that eight unrandomized nucleotides cap the
product. This will serve as the splint binding site during ligation. Third, since this product was
generated by Drosha (RNase III) cleavage, it has a 5′ phosphate and 3′-OH and is therefore ready
for ligation. Molecules in the product pool that were generated by other endonuclease activities,
such as metal cation or base catalyzed hydrolysis and cleavage by RNase A or RNase T1, would
have a 5-hydroxyls and 3′ phosphates (or cyclic 2′-3′ phosphates) and are therefore unligatable.
Thus the splint ligation approach also selects for true products of Microprocessor activity, rather
than unintended endonucleolytic degradation products.
First, ligate adaptors to the 5p side. The following sequences were used to ligate the S125circ
product pool. Note lowercase letters are RNA sequence. Also, the adaptors are synthetically 5′phosphorylated at the time of synthesis; this service can be requested from most oligo synthesis
facilities. Finally, four separate adaptors were synthesized, each with a slight sequence variation.
This provides the opportunity to do up to four selections using the same parent miRNA pool, and
sequence the selections in the same Illumina lane (provided that each selection library uses a
different adaptor barcode). Three nucleotides were used for the barcode, so in principle one
could design a total of 43 or 64 barcodes.
S125circ.007 5p Arm Splint
ACGTGTACCCTAGAGA
S125circ.007.A 5p Arm Adaptor, CAT barcode
p-guacacguaugaGATCGGAAGAGCGGTTCAGCAGGAATGC
S125circ.007.B 5p Arm Adaptor, ATG barcode
p-guacacgucauaGATCGGAAGAGCGGTTCAGCAGGAATGC
S125circ.007.C 5p Arm Adaptor, TGA barcode
p-guacacguucaaGATCGGAAGAGCGGTTCAGCAGGAATGC
S125circ.007.D 5p Arm Adaptor, TAG barcode
p-guacacgucuaaGATCGGAAGAGCGGTTCAGCAGGAATGC
178
Assemble the ligation reaction.
1.2 µl
5 µl
1.0 µl
0.8 µl
8 µl
5p Arm Adaptor (100 µM)
Gel-purified selection product
5p Arm Splint (100 µM)
H2O
Heat to 85º for 5 min, then air cool to room temperature
+1 µl
+1 µl
10 µl
10X DNA Ligase Buffer
T4 DNA Ligase
Incubate at 25º x overnight. Due to the short splint, do not incubate at 37º.
Add 0.5 µl Turbo DNAse and incubate at 37º x 10 min to degrade the splint.
After the reaction is complete, gel purify the product on a 5% urea-polyacrylamide gel using
appropriate size markers. Yield may range from 10-50%, depending on the amount of product
available, and the fraction of product with the desired ends. Cut out the band, elute using the
crush-and-soak method, precipitate, and resuspend in 10 µl H2O.
Next, ligate adaptors to the 3p arm side. The following sequences were used to ligate the
S125circ product pool. Note lowercase letters are RNA sequence. Do not phosphorylate
adaptors, (a) because it is not necessary, and (b) phosphorylation may make certain side products
possible.
S125circ.005 3p Arm Adaptor
Do not phosphorylate.
GAGATCTACACTCTTTCCCTACACGACGCTCTuccgaucu
S125circ.006 3p Arm Splint
AGACGCCAAGATCGGA
Assemble the 3p arm ligation reaction.
1.2 µl
5 µl
1.0 µl
0.8 µl
8 µl
3p Arm Adaptor (100 µM)
Gel-purified 5p ligation product
3p Arm Splint (100 µM)
H2O
Heat to 85º for 5 min, then air cool to room temperature
+1 µl
+1 µl
10 µl
10X DNA Ligase Buffer
T4 DNA Ligase
Incubate at 25º x overnight. Due to the short splint, do not incubate at 37º.
179
Add 0.5 µl Turbo DNAse and incubate at 37º x 10 min to degrade the splint.
After the reaction is complete, gel purify the product on a 5% urea-polyacrylamide gel using
appropriate size markers. Yield may range from 10-50%, depending on the amount of product
available, and the fraction of product with the desired ends. Cut out the band, elute using the
crush-and-soak method, precipitate, and resuspend in 10 µl H2O.
Reverse transcribe the ligated product. Since the product is partially-randomized, the reverse
transcription will be primed from the 5p arm ligation adaptor. Note that reverse transcriptase
appears to be error-prone in the first few nucleotides after the primer; ideally, one or more
nucleotides of constant sequence should be included before a critical sequence (e.g. the barcode)
is encountered. For our selections, we used this primer.
S125circ.008 RT Primer (-1)
GCATTCCTGCTGAACCGCTCTTCCGATC
Assemble the RT reaction:
5 µl
3p Ligation Product
5 µl
H2O
6.25 µl
10X “House” dNTP Mix
0.3 µl
RT primer
16.55 µl
Heat to 85º for 5 min, then air cool to room temperature
+5 µl
+1.25 µl
+2 µl
~25 µl
5X First Strand Buffer
100 mM DTT
Superscript III
Incubate at 55º x 2hr
Base hydrolyze the RT reaction by adding 5 µl 1N NaOH and heating to 85º x 10
min. Add 25 µl of 1M HEPES-NaOH pH 7.0 and desalt over a P30 column.
PCR amplify the cDNA of the selected pool. The primers hybridize to the cDNA and add the
remaining sequences necessary for Illumina paired-end sequencing. It is crucial to use PAGE or
some other technique to purify the primers. The sequences are:
S0.001 Solexa Fwd Seq
GAGATCTACACTCTTTCCCTACACGACGCTCTTCCGATCT
S0.002 Solexa Rev Seq, -1 short
CAAGCAGAAGACGGCATACGAGATCGGTCTCGGCATTCCTGCTGAACCGCTCTTCCGATC
180
Assemble the following reaction:
38 µl
0.5 µl
0.5 µl
5 µl
5 µl
1 µl
50 µl
Desalted RT reaction
S0.001 Solexa Fwd Primer (100 µM)
S0.002 Solexa Rev Primer -1 (100 µM)
10X “House” dNTP
10X “House” Taq Buffer
“House” Taq
30 sec
30 sec
30 sec
30 sec
95°C
95°C
62°C
72°C
Repeat for 10 total cycles
10°C
Use the following program.
(hold)
After 10 cycles are complete, withdraw 5 µl of the reaction and run on a 1.5% agarose gel. If
product is visible, stop and purify the product. If no product is visible, amplify another 3 cycles
and withdraw a second aliquot. If still no product is visible, amplify another 3 cycles and
withdraw a third aliquot. If no product is visible after this stage (16 total cycles), it is likely that
one must repeat the selection at a larger scale, taking greater care to preserve yield.
If product is visible at any point, stop and purify. Use a 10% formamide-acrylamide gel and
stain with Sybr Gold. After elution, precipitation, and resuspension, the purified DNA is ready
for Illumina sequencing.
Library preparation from the reference (uncleaved) pool
To discover contributing sequence elements, we need to determine whether the sequences in the
selected pool deviate in abundance from expectation. To minimize bias, we directly sample the
unselected pool by amplifying the unselected pool RNA and preparing high-throughput
sequencing
libraries.
Since no cleavage has occurred, the ligation-RT-amplification approach taken for the selected is
neither necessary nor possible. Instead, the uncircularized, phosphorylated pool (before
circularization ligation) is directly reverse transcribed, amplified, and sequenced.
For the unselected S125circ pool, the reverse transcription primer was:
S125circ.009 Ref RT
TGGATGTCCTCACAGGTTAAAGGGTCTCAGGGACCTAG
181
Assemble the following reaction:
0.5 µl
Phosphorylated, uncircularized pool RNA
9.5 µl
H2O
6.25 µl
10X “House” dNTP Mix
0.3 µl
RT primer
16.55 µl
Heat to 85º for 5 min, then air cool to room temperature
+5 µl
+1.25 µl
+2 µl
~25 µl
5X First Strand Buffer
100 mM DTT
Superscript III
Incubate at 55º x 2hr
Base hydrolyze the RT reaction by adding 5 µl 1N NaOH and heating to 85º x 10
min. Add 25 µl of 1M HEPES-NaOH pH 7.0 and desalt over a P30 column.
We will add the Illumina sequences needed for sequencing by PCR, in two stages. The primers
for the first stage amplification of S125circ are as follows.
S125circ.014 Ref-II Fwd Primer
CTTTCCCTACACGACGCTCTTCCGATCTCAGGTGAGGTTCTTGGGAGCCTGGC
S125circ.015 Ref-II Rev Primer
GCATTCCTGCTGAACCGCTCTTCCGATCTTTAAAGGGTCTCAGGGACCTAGAG
Assemble the following reaction:
38 µl
0.5 µl
0.5 µl
5 µl
5 µl
1 µl
50 µl
Desalted RT reaction
S125circ.014 Fwd (100 µM)
S125circ.015 Rev (100 µM)
10X “House” dNTP
10X “House” Taq Buffer
“House” Taq
30 sec
30 sec
30 sec
30 sec
95°C
95°C
55°C
72°C
Repeat for 3 total cycles
10°C
Use the following program.
(hold)
Polish the pool for Illumina sequencing using the following primers (second stage):
182
S0.001 Solexa Fwd Seq
GAGATCTACACTCTTTCCCTACACGACGCTCTTCCGATCT
S0.002 Solexa Rev Seq, -1 short
CAAGCAGAAGACGGCATACGAGATCGGTCTCGGCATTCCTGCTGAACCGCTCTTCCGATC
Assemble the following reaction:
5 µl
0.5 µl
0.5 µl
5 µl
5 µl
33 µl
1 µl
50 µl
Stage 1 PCR Reaction
S0.001 Solexa Fwd (100 µM)
S0.002 Solexa Rev (100 µM)
10X “House” dNTP
10X “House” Taq Buffer
H2O
“House” Taq
30 sec
30 sec
30 sec
30 sec
95°C
95°C
55°C
72°C
Repeat for 8 total cycles
10°C
Use the following program.
(hold)
Run a 5 µl aliquot on a 1.5% agarose gel. If no product is visible, amplify an additional 3 cycles.
When product is faintly visible on agarose, run the remaining reaction on a 10% formamideacrylamide gel and stain with Sybr Gold. After elution, precipitation, and resuspension, the
purified DNA is ready for Illumina sequencing.
183
In vitro cleavage selection of linear pri-miRNA substrates
Overview
The goal of this selection is to explore Microprocessor cleavage determinants the stem-loop
region (what will µltimately be the pre-miRNA). The overall strategy is to generate a large pool
of variant molecules and subject this pool to Microprocessor cleavage. Products of cleavage are
recovered and sequenced by high-throughput sequencing to detect motifs that are enriched in the
product (Microprocessor-cleaved) population, relative to the original pool.
For RNA motifs flanking the pre-miRNA, a major hindrance to this strategy was that after
cleavage of linear pri-miRNA substrates, there will be three products: (1) the pre-miRNA, (2) a
5p (upstream) product, and (3) a 3p (downstream) product.
By contrast, motifs in the loop and apical stem are relatively easy to explore, since the entire
partially randomized region is contained within a single product molecule, which can be
recovered and sequenced.
We would like to generate a large number of variants, recover the variants that are cleaved by the
Microprocessor, and sequence the successful variants.
184
Assembly of linear, partially-randomized, pri-miRNA substrate
Template assembly by PCR
Templates for T7 transcription are assembled from synthetic oligonucleotides. IDT will
synthesize long oligos with both constant sequence and partially randomized positions (for a
price).
By way of example, oligonucleotide sequences are provided for the constant (unrandomized)
C125loop template. However, the clonal pool (composed many copies of the wildtype miRNA
sequence) should be prepared in parallel with the partially randomized pool (composed of a few
copies of many variant sequences). Oligos for the clonal pool are the same as oligos for the
randomized pool, except that nucleotides in red were partially randomized to 79% wildtype base
and 7% chance of each nonwildtype base.
S125loop.001a Central
ACCATGTTGCCAGTCTCTAGGTCCCTGAGACCCTTTAACCTGTGAGGACATCCAGGGTCACAGGT
GAGGTTCTTGGGAGCCTGGCGTCTGGCCCAACCAC
S125loop.002 Left Arm
CAGAGATGCATAATACGACTCACTATAgCCCCCACCCCAGGGTCTACCGGGCCACCGCACACCAT
GTTGCCAGTCTCTAGG
S125loop.003 Right Arm
GGTCAGAAGTCAGGCCAGCAATTCCCCAGGTGTGTGGTTGGGCCAGACGCCA
First, PAGE-purify long primers. In general, primers longer than 60 nt are difficult even for
commercial operations to synthesize, and should be purified before use. Resuspend purified
oligos to 100 µM where possible.
The T7 template can be generated by a single PCR reaction. In the first cycle, the “Central”
oligo acts as a template for an initial primer extension by the “Right Arm” oligo. In the second
cycle, the extended “Right Arm” DNA acts a template for a second primer extension by the “Left
Arm.” After two cycles, the reaction should proceed as a normal PCR. Assemble the following:
370 µl
0.3 µl
10 µl
10 µl
50 µl
50 µl
10 µl
500 µl
H2O
S125loop.001a (100 µM)
S125loop.002 (100 µM)
S125loop.003 (100 µM)
10X “House” PCR Buffer
10X “House” dNTP Mix
“House” Taq
185
Use this program:
30 sec
30 sec
30 sec
30 sec
(hold)
95°C
95°C
52°C or appropriate annealing temp.
72°C
Repeat for 4 total cycles
10°C
Ethanol precipitate the reaction and resuspend in 50 µl H2O.
Transcription of linear pri-miRNA pool
Use T7 RNA Polymerase to transcribe the template prepared in the previous step. A 200 µl
reaction scale should be sufficient for most applications, but more is always possible.
25 µl
195 µl
40 µl
80 µl
20 µl
20 µl
20 µl
400 µl
Concentrated template pool
H2O
10X “House” T7 Buffer
5X “House” NTP Mix
DTT (100 mM)
Fresh α-[32P]-UTP
“House” T7 RNA Polymerase
Incubate 37º x 2-3 hr
Add 5 µl Turbo DNAse (Ambion)
Incubate 37º x 30 min
Add 1/20 volume 500 mM EDTA
Add 1/10 volume 3M NaCl
Add 1 volume 100% ethanol
Final:
T7 reaction mixture with whatever NTP/Salt/Mg/PPi is present
~25 mM EDTA
~75 mM NaCl (supplemental)
50% ethanol
Incubate at least 15 min in -20
Spin 15 min at 4º
After precipitation, the reaction should be purified on a 5% Urea-polyacrylamide gel with an
appropriate size marker. Cut out and elute the mature, self-cleaved transcript. This is the linear
pri-miRNA pool.
186
For the C125loop primers given above, the expected product sequence is given below. The
S125loop partially randomized transcript was twin to this sequence, except that positions colored
red were partially randomized at 79% wildtype (indicated base) and 7% each of the other three
bases. These positions comprise the apical miRNA stem and the loop.
ppp-5′gCCCCCACCCCAGGGTCTACCGGGCCACCGCACACCATGTTGCCAGTCTCTAGGTCCCTGAGACCCTTTA
ACCTGTGAGGACATCCAGGGTCACAGGTGAGGTTCTTGGGAGCCTGGCGTCTGGCCCAACCACACACCTG
GGGAATTGCTGGCCTGACTTCTGACC-3′-OH
Selection of functional variants from linear pri-miRNA pool
Cleavage timecourse of clonal wildtype pool and partially-randomized pool
This experiment serves three purposes: (1) verification that the linear pri-miRNA substrates are
cleavage competent molecules, (2) estimation of the overall contribution of apical stem and/or
loop RNA sequence by comparing the cleavage of the partially-randomized pool to the cleavage
of wildtype pri-miRNA, and (3) optimization of cleavage timing for the actual selection.
Assemble the following reaction, assuming 1 µM stock solutions of RNA and 50 nM
Microprocessor complex in whole-cell lysate. As concentrations from different substrate and
protein preps vary, volumes should be adjusted to match these final concentrations.
39 µl
50 µl
10 µl
1 µl
100 µl
1X Sonic Buffer
Microprocessor Lysate
10X Cleavage/Binding Buffer (1 mM Mg)
Linear clonal or randomized pool (1 µM)
Prior to addition of the RNA substrate, the reaction mixture should be prewarmed at 37º. After
addition of substrate, withdraw 10 µl at the following times: 0 min, 20 sec, 40 sec, 1 min, 2 min,
4 min, 6 min, 8 min, 10 min, and 12 min. Each timepoint should be mixed immediately with 100
µl of ice-cold Tri-Reagent to stop the reaction.
After extraction from Tri-Reagent, run the reactions on an 10% urea-polyacrylamide gel and
estimate the amount of cleavage at each timepoint for both the clonal (wildtype) substrate and
the partially-randomized pool. For selection (next section), we have previously selected
timepoints where ~1% of the partially randomized pool and used this timepoint for selection. In
some situations, to achieve 1% cleavage it was necessary to adjust the ratio between
Microprocessor complex and partially-randomized pool substrate.
Selection of functional variants from partially-randomized linear pri-miRNA pool
Using conditions optimized in the previous experiment, perform a selection for functional,
Microprocessor-cleaved variants in the partially-randomized linear pri-miRNA pool. The
reaction should be scaled such that ~10 fmol of cleaved product can be obtained. This
187
corresponds to a selection for 6x109 product molecules. There is very little loss in the library
preparation, so essentially all of these molecules will be available for sequencing.
A target of 10 fmol product requires just a small cleavage reaction. For example, for cleavage
conditions using a (typical) final reaction concentration of 10 nM pri-miRNA pool, a 100 µl
cleavage reaction will use a total of 1,000 fmol substrate. Cleavage of 1% of this amount yields
10 fmol of product for library prepation for high-throughput sequencing.
After performing the selection reaction under optimized conditions, phenol-extract the reaction
and precipitate it. The precipitate should be resuspended in the minimum amount of water
needed to fully dissolve the pellet, and the resulting nucleic acid should separated on a 10% ureapolyacrylamide gel. Cut out the band corresponding to the pre-miRNA. For the S125loop
example given above, the product is expected to be 60 nt. Since yield is valuable, it is advisable
to use the crush-and-soak technique to purify this band (see Standard Protocols). Resuspend in
20 µl water.
Library preparation for Illumina high-throughput sequencing
Library preparation from the selected pool and reference pool
The goal of this phase is to add adaptors to the ends of the product sequence; these adaptors will
permit amplification of the variant molecules, and contain some necessary sequence for the
current (Feb. 2012) Illumina paired-end sequencing technology. We use PCR with long primers
to achieve this.
Reverse transcribe the pre-miRNA product, or the reference pool RNA. Note that reverse
transcriptase appears to be error-prone in the first few nucleotides after the primer; ideally, one
or more nucleotides of constant sequence should be included before a critical sequence (e.g.
randomized sequence) is encountered. For our selections, we used this primer.
S125loop.004 RT Primer
GGCATAGGCTCCCAAGAACCTC
Assemble the RT reaction:
10 µl
Pre-miRNA product or reference pool
6.25 µl
10X “House” dNTP Mix
0.3 µl
RT primer
16.55 µl
Heat to 85º for 5 min, then air cool to room temperature
+5 µl
+1.25 µl
+2 µl
~25 µl
5X First Strand Buffer
100 mM DTT
Superscript III
Incubate at 55º x 2hr
188
Base hydrolyze the RT reaction by adding 5 µl 1N NaOH and heating to 85º x 10 min. Add 25
µl of 1M HEPES-NaOH pH 7.0 and desalt over a P30 column.
PCR amplify the cDNA of the selected pool. The primers hybridize to the cDNA and will add
(1) a short barcode, and (2) part of the sequence needed for Illumina single-end sequencing. It is
crucial to use PAGE or some other technique to purify the primers. The reverse primer will be
the RT primer used above. The forward primers vary (each contains a different barcode, or no
barcode at all). We used:
S125loop.005 Init. PCR Fwd
GACGATCTCCCTGAGACCCTTTAA
S125loop.005a Init. PCR Fwd, barcoded
GACGATCgaTCCCTGAGACCCTTTAA
S125loop.005b Init. PCR Fwd, barcoded
GACGATCctTCCCTGAGACCCTTTAA
Assemble the following reaction:
20 µl
1 µl
1 µl
10 µl
10 µl
57 µl
1 µl
100 µl
Desalted RT reaction
S125loop.005 Series Primer (100 µM)
S125loop.004 RT Primer (100 µM)
10X “House” dNTP
10X “House” Taq Buffer
H2O
“House” Taq
30 sec
30 sec
30 sec
30 sec
95°C
95°C
50°C
72°C
Repeat for 5 total cycles
10°C
Use the following program.
(hold)
Without purifying the PCR reaction, assemble the second-stage PCR, which adds the remaining
sequences necessary for Illumina single-end sequencing. The primers used are specific to the
barcode used in the initial reaction, i.e. if 005a was used for the first reaction, 007a should be
used for the second reaction. It is essential to purify these primers to get optimal sequencing
results.
S125loop.006corr Solexa-R Adaptor
CAAGCAGAAGACGGCATACGAGGCTCCCAAGAACCTC
189
S125loop.007 Solexa-Seq Adaptor
AATGATACGGCGACCACCGACAGGTTCAGAGTTCTACAGTCCGACGATCTCCCTGAGACCCTTTAA
S125loop.007a Solexa-Seq Adaptor, barcoded
AATGATACGGCGACCACCGACAGGTTCAGAGTTCTACAGTCCGACGATCgaTCCCTGAGACCCTTTAA
S125loop.007b Solexa-Seq Adaptor, barcoded
AATGATACGGCGACCACCGACAGGTTCAGAGTTCTACAGTCCGACGATCctTCCCTGAGACCCTTTAA
Assemble the following reaction:
20 µl
2 µl
2 µl
20 µl
20 µl
2 µl
100 µl
Initial PCR Reaction
S125loop.007 Series Primer (100 µM)
S125loop.006corr R. Primer (100 µM)
10X “House” dNTP
10X “House” Taq Buffer
“House” Taq
30 sec
30 sec
30 sec
30 sec
95°C
95°C
55°C
72°C
Repeat for 5 total cycles
10°C
Use the following program.
(hold)
Run a 5 µl aliquot on a 1.5% agarose gel. If no product is visible, amplify an additional 3 cycles.
When product is faintly visible on agarose, run the remaining reaction on a 10% formamideacrylamide gel and stain with Sybr Gold.
Due to the nature of the PCR-addition approach, there may be four dominant products visible on
the gel: (1) Product from the initial PCR reaction, which may be further amplified by the dilute
oligos that are still around, at 73-75 bp; (2) products of reactions where only the reverse
extended arm has been added, at 85-87 bp; (3) products of reactions where only the forward
extended arm has been added, at 115-117 bp; and (4) full-length product for Illumina
sequencing, at 127-129 bp. This is the band that should be excised.
After elution, precipitation, and resuspension, the purified DNA is ready for Illumina
sequencing.
190
In Vitro Binding Selection for Functional pri-miRNAs
Preparation of immunopurified Drosha-TN/DGCR8
Overview
293T cells are transiently transfected with plasmids encoding C-terminal FLAG tagged human
DroshaTN (a dominant-negative form of Drosha in which the two RNase III domains are
mutated), and N-terminal FLAG-HA tagged human DGCR8. Cells are harvested and sonicated.
After centrifuge clearing of precipitates and insoluble material, the whole-cell lysate is stored in
single-use aliquots. This protocol is modified from Lee and Kim, Meth. Enz. 427 (2007), with
additional input from Jinju Han and V. Narry Kim.
Plasmids:
•
•
•
pMAX-GFP (transfection marker)
pFLAG-HA-DGCR8 (courtesy of T. Tuschl)
pCK-DroshaTN-FLAG (courtesy of V. Narry Kim)
Cell culture and transfection
Maintain 293T cells under standard conditions. We use Dulbecco’s Modified Eagle Medium
(DMEM) supplemented to 10% Fetal Bovine Serum.
To transfect cells, start with three 15-cm dishes at 90-100% confluence. Split all of these cells
into eight 15-cm dishes the day before transfection. Use 15 ml media per plate.
The day after splitting (18-24 hours later), begin transfection.
Assemble Lipofectamine dilution in one 50 ml Falcon tube:
720 µl Lipofectamine 2000
24 ml Opti-mem I Serum-Free Media
Incubate at room temperature for 5 min.
Meanwhile, assemble plasmid mixture in another 50 ml Falcon tube:
24 µg pMAX-GFP
60 µg pCK-Drosha-FLAG
60 µg pFLAG-HA-DGCR8
24 ml Opti-mem I Serum-Free Media
Add plasmid mixture to Lipofectamine dilution, and gently invert twice to mix.
complexes to form by incubating at room temperature for 20-30 min.
Allow
After complexes have formed, add 6 ml of mixture dropwise to each 15-cm dish. Rock plate
gently 1-2 times to ensure even dispersal of transfection complexes.
191
12-24 hours after transfection, check for expression of GFP. At this relatively early point, 50%75% of cells will already be expressing GFP.
Split all the transfected cells into twenty 15-cm dishes.
Allow cells to grow for 48 hr. After this period, cells should be at or just past confluence.
Frequently, with a successful transfection containing pMAX-GFP, the monolayer is faintly green
in room lighting. Cells are now ready for harvesting.
Lysate preparation
Harvest overexpressing cells by removing media and pipetting PBS onto the monolayer. Due to
the poor adherence of 293T cells, this is typically enough to dislodge the monolayer. Collect
PBS suspension of cells, and keep on ice.
Pellet PBS suspension by centrifugation. This pellet should be green from GFP in the cells;
there should be little or no visible fluorescence in the supernatant.
Prepare 1X Sonic Buffer without any reducing agents. Add protease inhibitors by dissolving a
Mini EDTA-Free Protease Inhibitor tablet in 10 ml of 1X Sonic Buffer (no reducing agents).
Keep on ice.
Resuspend cell pellet in 10 ml of 1X Sonic Buffer + protease inhibitors.
Prepare sonicator for use. We use a Branson Sonifier 250 at 50% duty cycle and output level 4.
Clean the probe by sonicating RNase-ZAP solution for 10 pulses (approximately 20 seconds at
50% duty cycle), then sonicating deionized water for 10 pulses.
Lyse 293T cells by sonicating for 10 pulses.
Clear the lysate once by centrifuging lysate at 3500 x g for 15 min. If lysis is successful, the
pellet should be yellow and perhaps light-green, while the supernatant is green. Collect
supernatant.
Clear supernatant once more by centrifuging at 3500 x g for 15 min. Collect supernatant. This is
the cleared whole-cell extract.
Immunoprecipitation with anti-FLAG beads
Prepare 100 µl Anti-FLAG agarose or magnetic beads according to manufacturer’s directions.
Wash twice more with 1X Sonic Buffer, without reducing agents.
Add beads to cleared whole-cell extract. Rotate or agitate at 4º x 4-18hr.
192
Pellet the beads by centrifugation or magnet. Pull off the supernatant and wash 3 times in 1X
Sonic Buffer + protease inhibitors, without reducing agents. Each wash should be at least 10fold larger than the volume of the packed beads, and each wash should incubate with the beads at
4º for at least 10 minutes.
After the final wash, elute with a combination of FLAG peptide and reducing agent. Prepare the
following elution buffer:
1X Sonic Buffer with Protease Inhibitors
0.7 µl/ml 2-mercaptoethanol
50% glycerol
100 ng/ul RNase-free BSA
100 ng/ul Yeast total RNA
Use at least 3-fold more than the packed bead volume. Aliquot eluted complex into single-use
aliquots and store in the vapor phase of liquid nitrogen.
193
In vitro binding selection of linear pri-miRNA substrates
Overview
The goal of this selection is to explore Microprocessor cleavage determinants the flanking
regions upstream and downstream of the pre-miRNA. The overall strategy is to generate a large
pool of variant molecules and subject this pool to a binding reaction with immunopurified
Drosha-TN and DGCR8. We would like to generate a large number of variants, recover the
variants that are bound by the Microprocessor, and sequence the successful variants. We seek
motifs that are enriched in the product (Microprocessor-bound) population, relative to the
original pool.
Since dominant-negative Drosha with mutations in the RNase III domains is used in this
experiment, the pri-miRNA substrate should remain intact after binding. Thus it is possible to
simply reverse-transcribe and amplify the bound RNA for Illumina high-throughput sequencing.
194
Assembly of linear, partially-randomized, pri-miRNA substrate
Template assembly by PCR
Templates for T7 transcription are assembled from synthetic oligonucleotides. IDT will
synthesize long oligos with both constant sequence and partially randomized positions (for a
price).
By way of example, oligonucleotide sequences are provided for the constant (unrandomized)
C125 template. However, the clonal pool (composed many copies of the wildtype miRNA
sequence) should be prepared in parallel with the partially randomized pool (composed of a few
copies of many variant sequences). Oligos for the clonal pool are the same as oligos for the
randomized pool, except that nucleotides in red were partially randomized to 79% wildtype base
and 7% chance of each nonwildtype base. Note that, on the left arm, 15 nucleotides of
nongenomic sequence are included. These are part of the sequences needed for Illumina pairedend sequencing.
C125.001 Left Arm
acgctcttccgatctCCCCAGGGTCTACCGGGCCACCGCACACCATGTTGCCAGTCTCTAGGTCC
CTGAGACCCTTTAACCTGTGAGGACATCCAGGGTC
C125.002 Right Arm
AGGAGTCAGGGGTCAGAAGTCAGGCCAGCAATTCCCCAGGTGTGTGGTTGGGCCAGACGCCAGGC
TCCCAAGAACCTCACCTGTGACCCTGGATGTCCTC
First, PAGE-purify long primers. In general, primers longer than 60 nt are difficult even for
commercial operations to synthesize, and should be purified before use. Resuspend purified
oligos to 100 µM where possible.
An initial primer extension between the left and right primers is used to generate a small amount
of template for PCR. Assemble the following reaction:
15 µl
15 µl
270 µl
40 µl
40 µl
20 µl
400 µl
Left Arm oligo (100 µM)
Right Arm oligo (100 µM)
H2O
10X “House” PCR Buffer
10X “House” dNTP Mix
“House” Taq
195
Perform a primer extension using the program “L/R Ext.”
30 sec
(ramp)
1 min
1 min
(hold)
95°C
Ramp to 37°C at 0.1°C/sec
37°C
72°C
Repeat for 5 total cycles
10°C
The T7 template can be generated a PCR reaction, using primers that add the T7 promoter, along
with two template C residues (so that T7 can initiate with GTP):
S0.002 T7-fwdSeq
CAGAGATGCATAATACGACTCACTATAggacacgacgctcttccgatct
S125.003 RT primer
TATGAGGAGTCAGGGGTCAG
Without purifying or concentrating the primer extension, assemble the following reaction:
30 µl
355 µl
5 µl
5 µl
50 µl
50 µl
5 µl
500 µl
Primer extension reaction
H2O
S0.002 T7-fwdSeq (100 µM)
S125.003 RT Primer (100 µM)
10X “House” PCR Buffer
10X “House” dNTP Mix
“House” Taq
Perform an initial PCR amplification using the program “L/R PCR.” Note that the annealing
temperature is based on the amount of homology between the T7 adaptor and the left arm oligo,
and the amount of homology between the RT primer and the right arm oligo.
30 sec
30 sec
30 sec
30 sec
(hold)
95°C
95°C
55°C or appropriate annealing temp.
72°C
Repeat for 4 total cycles
10°C
Ethanol precipitate reaction and resuspend in 50 µl H2O.
Transcription of linear pri-miRNA pool
Use T7 RNA Polymerase to transcribe the template prepared in the previous step. A 400 µl
reaction scale should be sufficient for most applications, but more is always possible.
196
25 µl
195 µl
40 µl
80 µl
20 µl
20 µl
20 µl
400 µl
Concentrated template pool
H2O
10X “House” T7 Buffer
5X “House” NTP Mix
DTT (100 mM)
Fresh α-[32P]-UTP
“House” T7 RNA Polymerase
Incubate 37º x 2-3 hr
Add 10 µl Turbo DNAse (Ambion)
Incubate 37º x 30 min
Add 1/20 volume 500 mM EDTA
Add 1/10 volume 3M NaCl
Add 1 volume 100% ethanol
Final:
T7 reaction mixture with whatever NTP/Salt/Mg/PPi is present
~25 mM EDTA
~75 mM NaCl (supplemental)
50% ethanol
Incubate at least 15 min in -20
Spin 15 min at 4º
After precipitation, the reaction should be purified on a 5% Urea-polyacrylamide gel with an
appropriate size marker. Cut out and elute the full-length transcript. This is the linear primiRNA pool.
For the C125 primers given above, the expected product sequence is given below. The S125
partially randomized transcript was twin to this sequence, except that positions colored red were
partially randomized at 82% wildtype (indicated base) and 6% each of the other three bases.
These positions comprise the flanking RNA sequence, up to (but not including) the basal stem.
ppp-5′ggacacgacgctcttccgatctCCCCCACCCCAGGGTCTACCGGGCCACCGCACACCATGTTGCCAGTCT
CTAGGTCCCTGAGACCCTTTAACCTGTGAGGACATCCAGGGTCACAGGTGAGGTTCTTGGGAGCCTGGCG
TCTGGCCCAACCACACACCTGGGGAATTGCTGGCCTGACTTCTGACCCCTGACTCCTCATA-3′-OH
Selection of functional variants from linear pri-miRNA pool
Trial competitive binding between clonal wildtype RNA and partially-randomized pool
This experiment serves to estimate the overall contribution of flanking RNA sequence by
comparing the binding of the partially-randomized pool to the binding of wildtype pri-miRNA.
197
First, transcribe a reference pri-miRNA sequence that is 15-20 nt shorter than the pool RNAs.
For mir-125a, the following sequence can be ordered from IDT and transcribed using T7
polymerase (see Standard Protocols). The RNA can be body labeled, or 5′ dephosphorylated and
5′ labeled with ATP. See Standard Protocols.
TATGAGGAGTCAGGGGTCAGAAGTCAGGCCAGCAATTCCCCAGGTGTGTGGTTGGGCCA
GACGCCAGGCTCCCAAGAACCTCACCTGTGACCCTGGATGTCCTCACAGGTTAAAGGGT
CTCAGGGACCTAGAGACTGGCAACATGGTGTGCGGTGGCCCGGTAGACCCTGGGGTGGG
GGcTATAGTGAGTCGTATTATGCATCTCTG
To perform the competitive binding experiment, assemble the following reaction, assuming 5
µM stock solutions of RNA. Although we do not have a measurement of the functional
Microprocessor concentration, it is likely to be much less than the concentration of RNA used in
the experiment.
20 µl
1.25 µl
1.25 µl
2.5 µl
25 µl
DroshaTN/DGCR8 eluate
Short Reference pri-miRNA (5 µM)
C125 wildtype or S125 pool RNA (5 µM)
10X Cleavage/Binding Buffer (1 mM Mg)
Allow binding reaction to proceed at room temperature for 15-30 min.
“unbound” sample and store in 1X urea loading dye.
Withdraw 0.5 µl
Prepare nitrocellulose filters for binding. Place pedestals on vacuum system (we use an older
vacuum setup that is totally unlabeled and inherited from David P. Bartel’s younger days; in
principle, a new Qiagen vacuum manifold could be used). Wet filters by adding 500 µl 1X Sonic
Buffer.
Apply binding reaction to the center of a nitrocellulose filter, under vacuum.
Wash the filter three times. For each wash, add at least 10-fold more than the reaction volume of
1X Sonic Buffer.
Elute protein-RNA complexes from the filter by carefully lifting the filter and placing in a
microcentrifuge tube. Add 500 µl 1X VCA Elution Buffer, and heat to 85º x 10 min. Vortex.
Withdraw elution buffer and ethanol precipitate by adding 1000 µl 100% ethanol.
After precipitation, estimate the number of filter-retained counts in the pellet. Resuspend the
pellet in an appropriate amount of 1X urea loading dye. Run an equal counts between the bound
and unbound sample (retained above) on a 4% urea-polyacrylamide gel. The gel should be run
such that the xylene dye is almost at the bottom of the gel in order to maximize the resolution at
~180 nt.
198
Quantify the ratio of long (pool) RNA to short (reference) RNA, and normalize by the ratio in
the input lane. The normalized ratio is an estimate of the relative binding of the long RNA vs.
the shorter RNA. Ideally, the relative binding between the reference RNA and C125 (the clonal,
wildtype, long RNA) is ~1, while the relative binding between the reference RNA and S125 (the
partially-randomized, long RNA) is >1, i.e. binding favors the reference RNA.
Selection of functional variants from partially-randomized linear pri-miRNA pool
Perform a selection for functional, Microprocessor-binding variants in the partially-randomized
linear pri-miRNA pool. Assemble the following reaction:
40 µl
5 µl
5 µl
50 µl
DroshaTN/DGCR8 eluate
C125 wildtype or S125 pool RNA (5 µM)
10X Cleavage/Binding Buffer (1 mM Mg)
Apply binding reaction to the center of a nitrocellulose filter, under vacuum. Wash the filter
three times. For each wash, add at least 10-fold more than the reaction volume of 1X Sonic
Buffer.
Elute protein-RNA complexes from the filter by carefully lifting the filter and placing in a
microcentrifuge tube. Add 500 µl 1X VCA Elution Buffer, and add 1 µl Yeast Total RNA to act
as a tube-blocking and co-precipitating agent. Heat to 85º x 10 min. Vortex. Withdraw elution
buffer and ethanol precipitate by adding 1000 µl 100% ethanol. Resuspend the pellet directly in
the following reaction.
Reverse transcribe the RNA. For the pri-mir-125 binding selection, the following primer was
used:
S125.003 RT primer
TATGAGGAGTCAGGGGTCAG
Assemble the RT reaction.
(pellet)
Eluted bound RNA
10 µl
H2O
6.25 µl
10X “House” dNTP Mix
0.3 µl
RT primer
16.55 µl
Heat to 85º for 5 min, then air cool to room temperature
+5 µl
+1.25 µl
+2 µl
~25 µl
5X First Strand Buffer
100 mM DTT
Superscript III
Incubate at 55º x 2hr
Base hydrolyze the RT reaction by adding 5 µl 1N NaOH and heating to 85º x 10
min. Add 25 µl of 1M HEPES-NaOH pH 7.0 and desalt over a P30 column.
199
PCR amplification for downstream applications
Template preparation for the next round of selection
The goal of this phase is to add adaptors to the ends of the product sequence; these adaptors will
permit amplification and then transcription of the variant molecules. We use PCR with long
primers to achieve this.
PCR amplify the cDNA of the selected pool using the same primers used to generate the original
T7 template.
S0.002 T7-fwdSeq
CAGAGATGCATAATACGACTCACTATAggacacgacgctcttccgatct
S125.003 RT primer
TATGAGGAGTCAGGGGTCAG
Assemble the following reaction:
55 µl
4 µl
4 µl
40 µl
40 µl
253 µl
4 µl
400 µl
Desalted RT reaction
S125.003 RT Primer (100 µM)
S0.002 T7-fwdSeq (100 µM)
10X “House” dNTP
10X “House” Taq Buffer
H2O
“House” Taq
30 sec
30 sec
30 sec
30 sec
95°C
95°C
55°C
72°C
Repeat for 6 total cycles
10°C
Use the following program.
(hold)
Run a 10 µl aliquot on a 1.5% agarose gel. If no product is visible, amplify an additional 3
cycles, and check again; we often get visible product around 10 cycles. When product is faintly
visible on agarose. Ethanol precipitate the PCR reaction and resuspend in 50 µl H2O. This
reaction can be used for T7 transcription for the next round of selection.
Library preparation for Illumina high-throughput sequencing
Any selected T7 template pool can be “polished” for Illumina high-throughput sequencing. To
sequence the initial pool (Pool 0), the pool RNA should be reverse-transcribed and amplified as
200
above, in order to fully capture any biases that were introduced by reverse transcription and
PCR.
The “polishing” primers add (1) a short barcode, and (2) part of the sequence needed for Illumina
single-end sequencing. It is crucial to use PAGE or some other technique to purify the primers.
The reverse primers vary (each contains a different barcode, in blue). We used:
S0.001 Solexa Fwd Seq
AATGATACGGCGACCACCGAGATCTACACTCTTTCCCTACACGACGCTCTTCCGATCT
S125.005.A Solexa Reverse Paired-End Sequencing, barcoded
CAAGCAGAAGACGGCATACGAGATCGGTCTCGGCATTCCTGCTGAACCGCTCTTCCGATCTCATTATGAGG
AGTCAGGGGTCAG
S125.005.B Solexa Reverse Paired-End Sequencing, barcoded
CAAGCAGAAGACGGCATACGAGATCGGTCTCGGCATTCCTGCTGAACCGCTCTTCCGATCTATGTATGAGG
AGTCAGGGGTCAG
S125.005.C Solexa Reverse Paired-End Sequencing, barcoded
CAAGCAGAAGACGGCATACGAGATCGGTCTCGGCATTCCTGCTGAACCGCTCTTCCGATCTTGATATGAGG
AGTCAGGGGTCAG
S125.005.D Solexa Reverse Paired-End Sequencing, barcoded
CAAGCAGAAGACGGCATACGAGATCGGTCTCGGCATTCCTGCTGAACCGCTCTTCCGATCTTAGTATGAGG
AGTCAGGGGTCAG
Assemble the following reaction:
0.3 µl
59 µl
1 µl
1 µl
10 µl
10 µl
1 µl
100 µl
Pool DNA
H2O
S125.005 Series Primer (100 µM)
S0.001 Solexa Fwd Seq (100 µM)
10X “House” dNTP
10X “House” Taq Buffer
“House” Taq
30 sec
30 sec
30 sec
30 sec
95°C
95°C
55°C
72°C
Repeat for 4 total cycles
10°C
Use the following program.
(hold)
201
Ethanol precipitate the PCR reaction and run on a 10% formamide-acrylamide gel. Stain with
Sybr Gold and cut out the product. After elution, precipitation, and resuspension, the purified
DNA is ready for Illumina sequencing.
202
Identification of Motif-Binding Proteins by Site-Specific
Crosslinking
Overview
This protocol outlines a method to identify proteins bound to a particular RNA motif. Typically,
a specific region of RNA is thought to be functionally important, and the underlying assumption
of this protocol is that this region provides a sequence-specific binding site for important
proteins. Given the motif, a 4-thiouridine (or, in principle, another reactive nucleoside analog) is
incorporated in specific positions, preferably at positions known to be important for function.
After 365 nm UV-mediated crosslinking of the 4-thiouridine to protein candidates, the
covalently-linked protein-RNA complex is subjected to further analysis to characterize the
unknown protein. A schematic of this process, in the context of pri-miRNAs, is provided below.
Protocol development was heavily influenced by Sontheimer EJ. Site-Specific Crosslinking with
4-thiouridine. Mol. Biol. Rep. (1994).
203
Assembly of 4-thiouridine containing RNA substrate
A 4-thiouridine containing substrate must be assembled or synthesized such that the following
conditions are met:
1. The 4-thiouridine is incorporated at a position thought to be important for binding of the
protein-of-interest.
2. A radioactive phosphate is incorporated at or near the 4-thiouridine site, such that after
complete digestion of the RNA by RNase (preferably RNase T1), the fragment of RNA
that contains the 4-thiouridine also contains the radioactive phosphate.
3. A biotin moiety is incorporated such that after binding to the protein complex of interest,
the piece of RNA that contains the 4-thiouridine also contains the biotin moiety.
However, the biotin should be incorporated such that cleavage by RNase (preferably
RNase T1) separates the biotin from the 4-thiouridine.
We have assembled substrates with these properties by splint ligating together a mix of RNAs
containing a particular 4-thiouridine, RNAs containing biotin, and RNAs containing a 5′-32P.
RNAs with 4-thiouridine were synthesized commercially by Dharmacon (we have not been able
to find a supplier for the (4-S-U)pG suggested by Sontheimer); likewise, RNAs containing 3′ or
5′ biotin were synthesized commercially by Dharmacon or IDT. Other RNAs were synthesized
by T7 transcription, where possible. See Supplement 1 for protocols related to splint ligation, 5′
phosphorylation, and T7 transcription. It is essential to perform all operations, to the greatest
extent possible, under light-protected conditions.
As an example, we have constructed a pri-miRNA substrate that contains a 4-thiouridine in the
CNNC motif, located in the 3p flanking region of the pri-miRNA.
The first step is to use T7 transcription to synthesize the “left” arm of the substrate, which
includes the 5p flanking region of the miRNA, the 5p arm of the hairpin, the loop, the 3p arm,
and part of the 3p flanking region. This is the T7 template used for this purpose:
CX30.017 (E) T7-L.Arm Template
TCCGAGGCAGTAGGCAGCTGCAAACATCCGACTGAAAGCCCATCTGTGGCTTCACAGCTTCCAGT
CGAGGATGTTTACAGTCGCTCACTGTCAACAGCAATATACCTTCTTTAGCCTTCTGTTGGGTTAA
CCTATAGTGAGTCGTATTATGCATCTCTG
The transcription product was gel purified. Next, a 4-thiouridine containing RNA was purchased
from Dharmacon:
CX30.020RNA
CU(4-S-U)CAAGGG
This RNA was 5′ phosphorylated with γ-32P-ATP and polynucletide kinase, then ligated to the
CX30.017 transcript by splint ligation with T4 RNA Ligase 2 and the following DNA splint from
IDT.
204
CX30.029 3p Splint
GCTCCTAAAGTAGCCCCTTGAAGTCCGAGGCAGTA
Finally, the following RNA was purchased from Dharmacon.
CX30.028 3p Biotin Fragment
GCTACTTTAGGAGCAATTATC-3′-Biotin
This RNA was 5′ phosphorylated with cold ATP and polynucleotide kinase, then ligated to the
017+020 ligation product using T4 RNA Ligase 2 and the same DNA splint (CX30.029).
The resulting ligation product looks like this:
5′pppGGTTAACCCAACAGAAGGCTAAAGAAGGTATATTGCTGTTGACAGTGAGCGACTGTAAA
CATCCTCGACTGGAAGCTGTGAAGCCACAGATGGGCTTTCAGTCGGATGTTTGCAGCTGCCT
ACTGCCTCGGACT(4-S-U)CAAGGGGCTACTTTAGGAGCAATTATC-3′-biotin
The product was validated by RNaseH cleavage at various places in the 5p sequence, the hairpin
loop, and the 3p sequence. In addition, the ligation product is cleavage-competent in
Drosha/DGCR8 overexpressing lysate.
RNA-protein crosslinking and complex analysis
Optimization of 365 nm UV dose
The goal of UV dose optimization is to maximize the amount of crosslinking to proteins of
interest, while minimizing other, possibly nonspecific crosslinks. In practice it seems to be
difficult to distinguish between the two, so we generally choose a UV dose to achieve
crosslinking just shy of maximum.
First, assemble a binding reaction under conditions where the putative protein of interest is
known to bind the RNA. Presumably, conditions in which the RNA is functional should be
appropriate conditions to perform the binding reaction.
For the Microprocessor complex and binding to the CNNC motif, we have used the following
reaction binding and/or cleavage reaction:
39 µl
50 µl
10 µl
1 µl
100 µl
1X Sonic Buffer
Microprocessor Lysate from 293T
10X Cleavage/Binding Buffer (1 mM Mg)
4-S-U containing RNA substrate (1 µM)
205
Prepare a platform for crosslinking. We wrap an aluminum heat block or aluminum tube rack in
parafilm, and inverted it so that the binding reaction can be spotted on to a flat, parafilm-covered
surface, according to this cross-sectional diagram. The advantage of using aluminum as a solid
support is that the block can be preheated or prechilled to a desired temperature, and will
reasonably hold this temperature during the crosslinking process.
Binding Reaction (20100 ul droplet)
Parafilm layer
Aluminum block
Spot aliquots of the reaction onto the parafilm surface in 20-100 µl volumes, taking care not to
puncture the parafilm layer with the pipette tip (if the RNA is radioactive, the aluminum block
will adsorb the radioactive RNA and is quite difficult to clean).
Place the block with the binding reaction droplets into the crosslinker. We use a Stratagene UV
2400 Stratalinker with 365 nm bulbs.
Crosslink the RNA. We use a constant-energy setting rather than a timer for greater consistency,
particularly as individual bulbs wear down and become dimmer or fail. For the UV dose
titration, a good starting point is to place 5 spots of 20 µl, and expose them to UV in multiple
iterations, removing one spot at a time, as described in the following table:
NB1: 500 mJ = “5000 x 100 micro Joules”
NB2: The maximum setting is about 1000 mJ = “9999 x 100 micro Joules”
Cycle
1
2
3
4
5
Setting
500 mJ
500 mJ
1000 mJ
1000 mJ
1000 mJ
Total Dose:
Spot #1
Spot #2
Spot #3
Spot #4
Spot #5
X
X
X
X
X
0 mJ
500 mJ
X
X
X
X
500 mJ
500 mJ
500 mJ
X
X
X
1000 mJ
500 mJ
500 mJ
1000 mJ
X
X
2000 mJ
500 mJ
500 mJ
1000 mJ
1000 mJ
1000 mJ
4000 mJ
After each dose, remove the appropriate spot and place in a light-protected (e.g. amber)
Eppendorf tube.
For each dose, aliquot half the recovered reaction (usually about 80% of the original volume, less
for higher doses) into separate tubes. Keep one set of crosslinked reactions at room temperature,
which will be a set of undigested controls.
206
For the other set of crosslinked reactions, digest the RNA in a manner such that the RNA is
largely degraded, but the hot phosphate label is still associated with the crosslinked RNA-protein
complex. For the CNNC crosslinking experiment described as an example, we used RNase T1.
We added 0.5 µl high-concentration RNase T1 to each 20 µl reaction and incubated at room
temperature for 15 minutes.
After RNase digestion, add SDS-PAGE buffer, e.g. Laemmli Buffer, to every sample and boil
for 5 minutes. Run the reactions on an appropriate SDS-PAGE gel. For most crosslinking
reactions, we have used Invitrogen Novex Nu-Page 4-12% Bis-Tris gels and MOPS buffer.
After separation on SDS-PAGE, expose the gel to film or phosphorimager plate. Proteins of
interest should be visible as discrete, UV-dependent, RNase-resistant bands.
Large-scale purification of crosslinked complexes for analysis
The goal of this section is to purify substantial amounts of crosslinked RNA-protein complex
using streptavidin-biotin affinity pulldown. The first step is to simply increase the scale of the
binding reaction, and crosslink using optimized UV dosing. If complex is being prepared for
mass-spectrometry analysis, it is worth determining whether the mass-spectrometry operator is
willing to accept radioactive samples. If not, it will be necessary to prepare a non-radioactive or
“cold” 4-thiouridine substrate, and do experiments in parallel with both hot and cold substrates.
For the CNNC binding experiment, a 10-fold scale up was sufficient for mass spectrometry,
although just barely so.
390 µl
500 µl
100 µl
10 µl
1000 µl
1X Sonic Buffer
Microprocessor Lysate from 293T
10X Cleavage/Binding Buffer (1 mM Mg)
4-S-U containing RNA substrate (1 µM)
As before, half the binding reaction should be kept at room temperature without exposure to UV
light. The other half should be crosslinked using previously optimized conditions.
The goal of the purification is to only retain proteins that are covalently linked to the RNA in a
UV-dependent manner. Consequently, we purify both crosslinked and uncrosslinked complexes;
the uncrosslinked preparation is used as a negative control in downstream analysis experiments.
Wash streptavidin beads according to manufacturer’s instructions. Wash twice more with 1X
Sonic buffer to equilibrate the beads. Use enough beads so that the binding capacity is in
reasonable excess over the amount of RNA used in the binding reaction. For example, in the
10X scale CNNC binding experiment, 10 pmol biotinylated RNA was used. The binding
capacity of the beads is 500 pmol/mg, or 0.5 pmol RNA per µl of beads. A reasonable amount to
use would therefore be 40 µl of beads (20 pmol capacity).
Add beads to binding reactions (uncrosslinked and crosslinked). Incubate at 4º, rotating or
agitating, for 15 min. Magnetically pellet the beads and withdraw the unbound supernatant. We
207
routinely keep all purification intermediates, including the unbound supernatant. For radioactive
substrates, compare counts on the beads to counts in the supernatant to ensure that most or all
counts have been captured on the beads. Additional incubation time is occasionally needed to
fully capture the biotinylated RNA.
Wash the beads thoroughly. Wash buffers should be at least 10-fold excess over packed bead
volume, and beads should be incubated in each wash cycle for at least 15 min at room
temperature, agitated or rotated.
We have used the following washing pattern:
Wash 1: 1X Laemmli buffer
Wash 2: 1X VCA-EB (see Standard Reagents)
Move beads to fresh light-protected tubes
Wash 3: 1X Laemmli buffer
Wash 4: 1X VCA-EB
After the final wash has been removed, elute the RNA-protein complex from the beads by
cleaving the RNA with RNase. We use RNase T1 for our experiments; other RNases can be
used subject to the conditions outlined at the beginning of this section.
For the 10X scale CNNC complex pulldown, we diluted high-concentration RNase T1 1:25 in
1X Sonic Buffer and added 20 µl of diluted RNase T1 to the beads. Note that small volumes
were used in order to preserve our ability to load the sample on to 1.5 mm thick SDS-PAGE gels.
Incubate at room temperature, rotating or agitating, for at least 15 min. Magnetically pellet the
beads and withdraw the supernatant. If using radioactive substrate, verify that all or nearly all
counts are released into the supernatant. This is the eluate containing the RNA-protein complex
(and a lot of RNase).
We have primarily used the eluate for mass-spectrometry. Check with the mass-spectrometry
facility or operator before purifying individual bands for this experiment. We have previously
run the sample on 1.5 mm 4-12% Bis-Tris gels using MOPS buffer. Hot and cold complex
purifications were carried out in parallel, and run on adjacent lanes in the gel. The radioactive
lane was used to mark the gel cutting location for the cold lane, and the cold gel slices submitted
for mass spectrometry. We have tried using Invitrogen SilverQuest silver staining to visualize
individual protein-RNA complex bands, but this appears to severely impede mass-spectrometric
workup.
Another approach is to run the eluate on a 1 mm SDS-PAGE gel, transfer to nitrocellulose or
PVDF membrane, and Western blot the eluate for candidate proteins. Due to the low amount of
protein purified, we have not had much success with this approach.
208
If candidate proteins are available by “educated guess” or from mass-spectrometry results, a
more robust approach immunoprecipitate the protein-RNA complex, as suggested by Sontheimer
and as described below.
Candidate protein testing by immunoprecipitation of crosslinked
complexes
The goal of this section is to determine whether a protein of interest binds the target RNA site (4thiouridine location). If an antibody is available for immunoprecipitation, a reasonably robust
approach is to crosslink the RNA-protein complex, immunoprecipitate the protein, and
demonstrate the association of radioactive RNA with the immunoprecipitated protein.
Assemble a sufficiently large binding reaction such that immunoprecipitated protein-RNA
complex can be easily visualized by phosphorimager or film exposure. You will need enough
binding reaction to do immunoprecipitation with the candidate protein’s antibody, as well as an
isotype control antibody. For the CNNC motif binding experiment, a 100 µl binding reaction
was used (for four total immunoprecipitations).
39 µl
50 µl
10 µl
1 µl
100 µl
1X Sonic Buffer
Microprocessor Lysate from 293T
10X Cleavage/Binding Buffer (1 mM Mg)
4-S-U containing RNA substrate (1 µM)
Crosslink using optimized crosslinking conditions and recover all droplets. Divide into aliquots.
For the CNNC binding experiment, four immunoprecipitations were carried out: (1) Mouse IgG,
(2) Anti-FLAG M2 antibody, (3) Anti-Candidate1 antibody, and (4) Anti-Candidate2 antibody.
For each immunoprecipitation, add 1 µl high-concentration RNase T1 per 25 µl binding reaction
(for other RNases, use an appropriate concentration). The RNase T1 will digest the substrate
while the immunoprecipitation is ongoing.
For each immunoprecipitation, add an appropriate dilution of antibody for immunoprecipitation.
For example, 1:500 (10 µg/ml) anti-FLAG is a typical dilution for immunoprecipitation.
Incubate at 4º 1 hour to overnight.
Wash Protein A or Protein G agarose beads or magnetic beads according to manufacturer’s
instructions. Wash twice more with 1X Sonic Buffer to equilibrate the beads. Add sufficient
beads to bind the amount of antibody used in the immunoprecipitation. For example, Sigma
Protein G agarose has a binding capacity of 8 mg IgG per ml of beads. Thus 10 µl of beads is
usually more than enough to bind the added IgG.
Add beads to the immunoprecipitation mixes. Incubate one hour at 4º, agitated or rotating.
209
Pellet beads and pull off supernatant. Wash 2-3 times in 1X Sonic Buffer. If a more aggressive
wash is desired, we have previously supplemented Sonic Buffer to 500 mM KCl (final) and 0.1%
Tween-20 (final).
Elute the RNA-protein complex from the beads by adding 20 µl 1X Laemmli Buffer and boiling
beads for 10 minutes. Pellet the beads and carefully pull off the eluate. Beads that carry over
may clog pipette tips and make gel loading difficult.
Run the eluate on an appropriate SDS-PAGE gel, e.g. 4-12% Bis-Tris with MOPS buffer.
Visualize radioactive bands by phosphorimager or film. Candidate proteins that bind the target
RNA site will show a radioactive band at or near the expected size of the protein, since the
immunoprecipitated protein is covalently linked to the radioactive phosphate in the RNA. If a
radioactive band is seen but is not at the target protein size, consider the possibility that the
candidate protein is in a complex with another protein and the target RNA, but the candidate
protein does not directly bind the RNA (i.e. the candidate protein coprecipitates with the actual
binding protein, but is not directly crosslinked to the RNA).
210
Supplement 1: Standard Protocols
Crush-and-soak technique for acrylamide gel extraction
This technique uses a centrifuge to force a gel slice through a 22-gauge needle hole. This
produces an acrylamide gel slurry, and decreases the diffusion distance between RNA molecules
and the elution fluid.
This technique requires that the gel slice be small enough to fit in a 0.5 ml Sarstedt Eppendorf
tube.
First, the 0.5 ml tube must be prepared for use. Close the tube cap and stand the tube on its cap.
Unpackage a new, unused and sterile 22-gauge needle. Attach this needle to a syringe for
convenient handling. Heat the tip of the needle using a Bunsen burner to red hot. Obviously, it
is important to handle the plastic portion, and not to overheat (i.e. to avoid melting the plastic
and the human parts).
Carefully stab the hot needle through the very bottom of the conical portion of the tube (sitting
inverted on the bench). For obvious safety reasons, it is critical to avoid holding the tube with
one’s hand, as this places the offending hand at needlestick risk. Although this needle is sterile,
it is likely that a needlestick with a red-hot needle will be quite painful.
After the hole has been created, carefully place the 0.5 ml tube inside a standard 1.5 ml
eppendorf tube. Minimize the amount of glove contact with the exterior of the 0.5 ml tube, since
this will be in contact with the interior of the 1.5 ml tube.
Place the gel slice in 0.5 ml eppendorf tube. Place the entire assembly (gel slice in the 0.5 ml
tube; 0.5 ml tube inside the 1.5 ml Eppendorf tube) into a centrifuge. Centrifuge at full speed
(~13,000 x g) for 1 min or until the gel slice is spun through the hole. Discard 0.5 ml tube.
Suspend the gel particles in 500 µl of 300 mM NaCl solution, and rotate overnight for elution.
After elution is complete, filter the slurry through a Costar Spin-X filter.
Precipitate the filtrate by adding 1000 µl ethanol, incubating in cold, and centrifuging as usual.
211
T7 transcription (“Midi” scale)
Note: 1 µl of 1 µM solution = 1 pmol solute
NB1. T7 RNA polymerase requires a double-stranded promoter region. The enzyme will
extend on a single-stranded antisense template. If a PCR product is used for runoff
transcription, the T7 promoter sense strand oligo should be excluded.
NB2. It is crucial that the first incorporated nucleotide be G, i.e. all T7 transcription products
will begin with pppG…
Minimal T7 promoter sequence and template strand
T7 promoter
5′ TAATTACGACTCACTATA 3′
Template strand
3′ ATTAATGCTGAGTGATATCNNNNNNNNNNNNN…NNNNNN 5′
5 µl
40 µl
80 µl
20 µl
10 µl
500 pmol oligo template
500 pmol T7 promoter sense strand (100 µM)
10 X House T7 Buffer
5X House NTP Mix
100 mM DTT
House T7 Enzyme
H2O
400 µl
Incubate 37º x 2-3 hr
Add 10 µl Turbo DNAse (Ambion)
Incubate 37º x 30 min
Add 1/20 volume 500 mM EDTA
Add 1/10 volume 3M NaCl
Add 1 volume 100% ethanol
Final:
T7 reaction mixture with whatever NTP/Salt/Mg/PPi is present
~25 mM EDTA
~75 mM NaCl (supplemental)
50% ethanol
Incubate at least 15 min in -20
Spin 15 min at 4º
212
Phosphorylation of nucleic acid 5′ Ends
Note: 1 µl of 1 µM solution = 1 pmol solute
Cold
NB1. For cold phosphorylation, a 1 mM final ATP concentration is used. This is
sufficiently higher than Km such that PNK quantitatively phosphorylates the substrate in 15
min, as measured by PAGE shift.
NB2. For the 10X Buffer, either supplement NEB 10X PNK buffer to 1 mM ATP (final) or
use 10X T4 DNA Ligase Buffer, which has the same buffer composition with 1 mM ATP
(final).
1 µl
1 µl
Up to 100 pmol 5′ ends
10X NEB T4 DNA Ligase Buffer (Yes, Ligase.)
T4 PNK
H2O
10 µl
Incubate 37º x 30 min
Hot
NB3. For hot phosphorylation, the hot ATP concentration is limiting and is sub-Km.
NB4. NEN/PE ATP is 6000 Ci/mmol, 150 mCi/ml = 2.5e-4 mmol/ml = 25 µM. For
quantitative phosphorylation, use less than 25 pmol 5′ ends and be prepared to wait for >1 hr.
1 µl
1 µl
1 µl
Up to 25 pmol 5′ ends
6000 Ci/mmol y-ATP
10X NEB T4 PNK Buffer
T4 PNK
H2O
10 µl
Incubate 37º x 60 min
213
Dephosphorylation of nucleic acid 5′ ends
Note: 1 µl of 1 µM solution = 1 pmol solute
NB1. Unit conditions in typical CIP reactions is sufficient to dephosphorylate recessed 5′ ends.
Nevertheless, if significant secondary structure shields the 5′ end, reaction can be carried out at
50º; CIP is quite heat stable.
Up to 100 pmol 5′ ends
10X NEBuffer 3
T4 CIP
H2O
1 µl
1 µl
10 µl
Incubate 37-50º x 60 min
Dephosphorylation of 2′-3′ cyclic phosphates and 3′ phosphates
NB1. Adapted from Huili Guo’s RNaseq protocol
NB2. I have not tested this protocol to strictly define the parameters for RNA concentration.
20 µl
1 µl
RNA, preferably in water
1.5X MES dephosphorylation buffer
T4 PNK
H2O
30 µl
Incubate 37º x 6 hrs
214
Splint ligation of RNA
Note: 1 µl of 1 µM solution = 1 pmol solute
Using T4 RNA Ligase 2 (RNL2)
NB1. T4 RNL2 is exceptionally efficient when the concentration of the nicked helical substrate
is around 10 µM. Below 10 µM, yield is either low or dominated by side products.
NB2. Splint should have 40-50º Tm so that the nicked substrate is stable at 37º.
100 pmol RNA with 5′-P (3p substrate)
100 pmol RNA with 3′-OH (5p substrate)
100 pmol Splint Oligo
H2O
8 µl
Heat to 85º for 5 min, then air cool to room temperature
+1 µl
+1 µl
10 µl
10X RNL2 Buffer
T4 RNL2
Incubate 37º x 4 hr or at 25º x overnight
Optional: add 0.5 µl Turbo DNAse and incubate at 37º x 10 min to degrade the splint.
Using T4 DNA Ligase
NB3. T4 DNA Ligase was used for the Moore and Sharp ligation protocol. The Km of DNA
Ligase is in the low nM range. Although it is less efficient than RNL2 with high concentration
of substrates, it is less sensitive to substrate concentration.
NB4. Typically a ligation target is at low concentration and the other components are in excess.
If this is the case, the splint oligo should be at least higher in concentration than the lower of the
two substrates.
1-100 pmol RNA with 5′-P (3p substrate)
1-100 pmol RNA with 3′-OH (5p substrate)
1-100 pmol Splint Oligo.
H2O
8 µl
Heat to 85º for 5 min, then air cool to room temperature
+1 µl
+1 µl
10 µl
10X DNA Ligase Buffer
T4 DNA Ligase
Incubate 37º x 4 hr or at 25º x overnight
Optional: add 0.5 µl Turbo DNAse and incubate at 37º x 10 min to degrade the splint.
215
Partial hydrolysis of RNA
NB1: Adapted from Huili Guo’s RNaseq protocol
10 µl
RNA, preferably in water
2X Fragmentation Buffer
H2O
20 µl
Incubate 85º x variable time, ranging from 30 sec to 20 min. Individual substrates should
be optimized.
Chill immediately on ice and neutralize.
Neutralization option 1 (Huili Guo): Add 380 µl of 300 mM NaOAc pH 5.2.
Neutralization option 2: Add 20 µl of 1M Tris-Cl pH 7.6. Add 360 µl of 300 mM NaCl.
Precipitate RNA by adding 1000 µl 100% ethanol.
216
Supplement 2: Standard Reagents
Buffers
1X Sonic Buffer
Modified from Lee and Kim, Meth. Enz. 427 (2007), with additional input from
Jinju Han and V. Narry Kim.
Concentrations:
20 mM Tris-Cl pH 8.0
100 mM KCl
0.2 mM EDTA
5 mM DTT (equivalent to 0.7 µl/ml 2-mercaptoethanol); add fresh
10X Drosha Cleavage/Binding Buffer
Designed to be added to reactions assembled in 1X Sonic Buffer. Due to large
volume of Yeast Total RNA, this buffer requires assembly of 10X Sonic Buffer
concentrate.
5 mM Mg version – concentrations at 10X:
1X Sonic Buffer
3 µg/ul Yeast Total RNA (Ambion)
50 mM MgCl
1 mM Mg version – concentrations at 10X:
1X Sonic Buffer
3 µg/ul Yeast Total RNA (Ambion)
10 mM MgCl
1.5X MES dephosphorylation buffer
Concentrations at 1.5X:
150 mM MES-NaOH, pH 5.5
15 mM MgCl2
15 mM β-mercaptoethanol (add fresh)
450 mM NaCl
10X “House” T7 Transcription Buffer
Concentrations at 10X:
400mM Tris-Cl, pH 7.9 at 20C
25mM Spermidine
260mM MgCl2
217
0.1% Triton X-100
Urea Loading Buffer
Concentrations at 2X:
8
M
25 mM EDTA
Add bromophenol blue and xylene cyanol FF as needed.
Urea
5X “House” NTP Mix
Note: From powder, resuspend 1 gram in 2 mls of H2O. Empirically pH with 1M
NaOH to pH 7. Can estimate equivalents of NaOH needed based on salt form of
the NTP and charge of 3.5 at pH 7.
Concentrations at 5X:
40 mM GTP
25 mM CTP
25 mM ATP
10 mM UTP
10X “House” dNTP Mix
Concentrations at 10X:
2 mM dATP
2 mM dCTP
2 mM dGTP
2 mM dTTP
10X “House” PCR Buffer
Concentrations at 10X:
100 mM TRIS pH 8.3 @20°
500 mM KCl
15 mM MgCl2
0.1% Gelatin
1X VCA Elution Buffer (VCA-EB)
Modified from DPB Elution buffer (DPBEB)
Concentrations at 1X:
8M Urea
25 mM EDTA
300 mM NaCl
218
2X Fragmentation Buffer
Modified from Huili Guo.
Concentrations at 2X:
2 mM EDTA
10 mM Na2CO3
90 mM NaHCO3
(Final carbonate buffer pH 9.3)
Commercial products
γ-[32P]-ATP (NEN/PE)
Catalog number: NEG035C001MC
Concentration: 6000ci/mmol, 150 mCi/ml
α-[32P]-UTP (NEN/PE)
Catalog number: BLU007X500UC
Concentration: 800Ci/mmol, 10mCi/ml
T4 DNA Ligase (NEB)
Catalog number M0202
Concentration: 400,000 cohesive end units/ml
Storage Temperature: -20°C
10X T4 DNA Ligase Buffer (NEB)
Catalog number B0202
Concentrations at 1X:
50 mM Tris-HCl
10 mM MgCl2
10 mM Dithiothreitol
1 mM ATP
pH 7.5 @ 25°C
Storage Temperature: -20°C
T4 RNA Ligase 2 (NEB)
Catalog number M0239
Concentration: 10,000 units/ml
Storage Temperature: -20°C
219
10X T4 RNA Ligase 2 Buffer (NEB)
Concentrations at 1X:
50 mM Tris-HCl
2 mM MgCl2
1 mM DTT
400 µM ATP
pH 7.5 @ 25°C
T4 Polynucleotide Kinase (NEB)
Catalog number M0201
Concentration: 10,000 units/ml
Storage Temperature: -20°C
10X T4 Polynucleotide Kinase Buffer (NEB)
Concentrations at 1X:
70 mM Tris-HCl
10 mM MgCl2
5 mM Dithiothreitol
pH 7.6 @ 25°C
Yeast Total RNA (Ambion)
Catalog number: AM7118
Concentration: 10 mg/ml
Storage Temperature: -20°C
Lipofectamine 2000 (Invitrogen)
Catalog number: 11668-027
Storage Temperature: 4°C
Opti-MEM I Reduced Serum Medium (Invitrogen)
Catalog number: 31985-062
Storage Temperature: 4°C
Complete Mini EDTA-Free Protease Inhibitor Tablets (Roche)
Catalog Number: 1836170001
Storage Temperature: 4°C
220
MicroSpin G-25 Columns (GE)
Catalog Number: 28917922
Storage Temperature: Room Temperature
Note: G-25 columns have higher overall RNA retention than P-30 columns, and
are somewhat less consistent. However, these columns reasonably pass 10-mer
RNAs and are therefore useful for preparation of Decade markers
Micro Bio-Spin 30 (aka P30) Columns (Bio-Rad)
Catalog Number: 732-6250
Storage Temperature: 4°C
Note: P-30 columns are highly consistent with low overall RNA retention. The
cutoff for this column is somewhere between 10 and 20 nt.
Decade Markers (Ambion)
Catalog Number: AM7778
Storage Temperature: -20°C
Tri-Reagent (Ambion)
Catalog Number: AM9738
Storage Temperature: 4°C
Turbo DNAse (Ambion)
Catalog Number: AM2238
Concentration: 2 units/ul
Storage Temperature: -20°C
0.5 ml Microcentrifuge Tube (Sarstedt)
Catalog Number: 72.699
Spin-X centrifuge tube filters (Costar / Corning)
Catalog Number: 8161
Pore size: 0.22 um
Superscript III Reverse Transcriptase (Invitrogen)
Catalog Number: 18080044
Concentration: 200 units/ul
Storage Temperature: -20°C
SYBR Gold nucleic acid stain (Invitrogen)
Catalog Number: S11494
221
Concentration: 10,000X
Storage Temperature: -20°C
Stratalinker 2400 bulbs, 365 nm (Thermo Fisher)
Catalog Number: 50125580
RNase T1 (Cloned), a.k.a High-Concentration (Ambion)
Catalog Number: AM2280
Concentration: 1000 units/ul
Storage Temperature: -20°C
Dynabeads MyOne Streptavidin C1 Magnetic Beads (Invitrogen)
Catalog Number: 650-02
Concentration: 10 mg/ml
Binding Capacity: 500 pmol/mg ssDNA oligo
EZview Red Protein G Affinity Gel (Sigma)
Catalog Number: E3403
Binding Capacity: 8 mg/ml rabbit IgG
EZview Red Anti-FLAG M2 (Sigma)
Catalog Number: F2426
Binding Capacity: 0.6 mg tagged bacterial alkaline phosphatase / ml packed slurry
Storage Temperature: -20°C
Anti-FLAG M2 Magnetic Beads (Sigma)
Catalog Number: M8823
Binding Capacity: 0.6 mg tagged bacterial alkaline phosphatase / ml packed slurry
Storage Temperature: -20°C
Ultrapure BSA (Applied Biosystems)
Catalog Number: AM2616
Concentration: 50 mg/ml
Storage Temperature: -20°C
3X FLAG Peptide (Sigma)
Catalog Number: F4799
Storage Temperature: 4°C (lyophilized)
222
Immobilon-NC Membrane Filters (Millipore)
Catalog Number: HATF01300
Pore size: 0.45 um
Diameter: 13 mm
“Surfactant-Free”
223
224
Appendix 2.
Statistical methods
Contents
Overview ......................................................................................................................................226
Calculation of relative cleavage ...................................................................................................226
Calculation of the odds ratio score...............................................................................................229
Calculation of the Watson–Crick base pairing score ...................................................................230
Calculation of Information Content .............................................................................................231
225
Overview
After selection and sequencing, analysis of the data is rooted in measuring frequencies of
motifs in the reference pool and the selected pool. Intuitively, these frequencies reflect the
preference of the Microprocessor for a given motif: if the Microprocessor prefers the motif, it
would be enriched by the selection, and would be present more frequently in the selected pool
than in the reference pool. By contrast, a disfavored motif would be depleted by the selection, so
that the motif would be present less frequently in the selected pool than in the reference pool. The
following discussion formalizes this intuition in order to quantify the relative contribution of
different motifs to recognition and cleavage by the Microprocessor.
Calculation of relative cleavage
For each base at any given position, we know two values based on sequencing of selected
and reference pools. We want to estimate the selectivity of the complex for a particular base, or a
particular motif. Bayes’ Theorem, written in odds form, can be used to compare one motif to one
other motif at that position:
𝑃(𝑚𝑜𝑡𝑖𝑓𝑖 ) 𝑃(𝑐𝑙𝑒𝑎𝑣𝑎𝑔𝑒|𝑚𝑜𝑡𝑖𝑓𝑖 ) 𝑃(𝑚𝑜𝑡𝑖𝑓𝑖 |𝑐𝑙𝑒𝑎𝑣𝑎𝑔𝑒)
×
=
𝑃(𝑚𝑜𝑡𝑖𝑓𝑗 ) 𝑃(𝑐𝑙𝑒𝑎𝑣𝑎𝑔𝑒|𝑚𝑜𝑡𝑖𝑓𝑗 ) 𝑃(𝑚𝑜𝑡𝑖𝑓𝑗 |𝑐𝑙𝑒𝑎𝑣𝑎𝑔𝑒)
(Eq. 1A: Bayesian statement for motifi vs. motifj)
The first term is the proportion of motifi to motifj in the initial, unselected pool of
sequences, estimated by sequencing the reference pool. The third term is the proportion of motifi
to motifj in the selected pool, estimated by sequencing the cleavage/selection product.
The middle term describes the relative cleavage of motifi vs. motifj, and is the observed
relative cleavage of motifi over motifj, at a particular position. This term is ideally a property of
the enzyme cleavage conditions, independent of the distribution of motifs in the initial pool. For
every motifi of length n, there are 4n terms that describe its relative cleavage with respect to each
of every other possible motifj of length n at that position:
𝑃(𝑚𝑜𝑡𝑖𝑓𝑖 ) 𝑃(𝑐𝑙𝑒𝑎𝑣𝑎𝑔𝑒|𝑚𝑜𝑡𝑖𝑓𝑖 ) 𝑃(𝑚𝑜𝑡𝑖𝑓𝑖 |𝑐𝑙𝑒𝑎𝑣𝑎𝑔𝑒)
×
=
𝑃(𝑚𝑜𝑡𝑖𝑓1 ) 𝑃(𝑐𝑙𝑒𝑎𝑣𝑎𝑔𝑒|𝑚𝑜𝑡𝑖𝑓1 ) 𝑃(𝑚𝑜𝑡𝑖𝑓1 |𝑐𝑙𝑒𝑎𝑣𝑎𝑔𝑒)
226
𝑃(𝑚𝑜𝑡𝑖𝑓𝑖 ) 𝑃(𝑐𝑙𝑒𝑎𝑣𝑎𝑔𝑒|𝑚𝑜𝑡𝑖𝑓𝑖 ) 𝑃(𝑚𝑜𝑡𝑖𝑓𝑖 |𝑐𝑙𝑒𝑎𝑣𝑎𝑔𝑒)
×
=
𝑃(𝑚𝑜𝑡𝑖𝑓2 ) 𝑃(𝑐𝑙𝑒𝑎𝑣𝑎𝑔𝑒|𝑚𝑜𝑡𝑖𝑓2 ) 𝑃�𝑚𝑜𝑡𝑖𝑓𝑗 �𝑐𝑙𝑒𝑎𝑣𝑎𝑔𝑒�
…
𝑃(𝑚𝑜𝑡𝑖𝑓𝑖 ) 𝑃(𝑐𝑙𝑒𝑎𝑣𝑎𝑔𝑒|𝑚𝑜𝑡𝑖𝑓𝑖 ) 𝑃(𝑚𝑜𝑡𝑖𝑓𝑖 |𝑐𝑙𝑒𝑎𝑣𝑎𝑔𝑒)
×
=
𝑃(𝑚𝑜𝑡𝑖𝑓𝑖 ) 𝑃(𝑐𝑙𝑒𝑎𝑣𝑎𝑔𝑒|𝑚𝑜𝑡𝑖𝑓𝑖 ) 𝑃(𝑚𝑜𝑡𝑖𝑓𝑖 |𝑐𝑙𝑒𝑎𝑣𝑎𝑔𝑒)
…
𝑃(𝑚𝑜𝑡𝑖𝑓𝑖 )
𝑃(𝑐𝑙𝑒𝑎𝑣𝑎𝑔𝑒|𝑚𝑜𝑡𝑖𝑓𝑖 )
𝑃(𝑚𝑜𝑡𝑖𝑓𝑖 |𝑐𝑙𝑒𝑎𝑣𝑎𝑔𝑒)
×
=
𝑃(𝑚𝑜𝑡𝑖𝑓4𝑛 ) 𝑃(𝑐𝑙𝑒𝑎𝑣𝑎𝑔𝑒|𝑚𝑜𝑡𝑖𝑓4𝑛 ) 𝑃(𝑚𝑜𝑡𝑖𝑓4𝑛 |𝑐𝑙𝑒𝑎𝑣𝑎𝑔𝑒)
In practice, however, we rarely consider a single, isolated motif relative to another isolated
motif, due to the design of the selection. In addition, the analysis occasionally necessitates
comparison of degenerate motifs, e.g. Watson-Crick pairs vs. nonpairs, or wildtype vs.
nonwildtype motifs. In these cases, each probability in Eq. 1A must be partitioned and expressed
as the sum of those sub-motifs. Let I be a partition of motifi into individual submotifs, and let J
be a partition of motifj into submotifs, i.e.
𝑃(𝑚𝑜𝑡𝑖𝑓𝑖 ) = ∑𝐼 𝑃(𝑠𝑢𝑏𝑚𝑜𝑡𝑖𝑓 ∈ 𝐼)
and
𝑃�𝑚𝑜𝑡𝑖𝑓𝑗 � = ∑𝐽 𝑃(𝑠𝑢𝑏𝑚𝑜𝑡𝑖𝑓 ∈ 𝑗)
∑𝐼 𝑃(𝑠𝑢𝑏𝑚𝑜𝑡𝑖𝑓 ∈ 𝐼) × 𝑃(𝑐𝑙𝑒𝑎𝑣𝑎𝑔𝑒|𝑠𝑢𝑏𝑚𝑜𝑡𝑖𝑓 ∈ 𝐼) ∑𝐼 𝑃(𝑠𝑢𝑏𝑚𝑜𝑡𝑖𝑓 ∈ 𝐼|𝑐𝑙𝑒𝑎𝑣𝑎𝑔𝑒)
=
∑𝐽 𝑃(𝑠𝑢𝑏𝑚𝑜𝑡𝑖𝑓 ∈ 𝐽) × 𝑃(𝑐𝑙𝑒𝑎𝑣𝑎𝑔𝑒|𝑠𝑢𝑏𝑚𝑜𝑡𝑖𝑓 ∈ 𝐽) ∑𝐽 𝑃(𝑠𝑢𝑏𝑚𝑜𝑡𝑖𝑓 ∈ 𝐽|𝑐𝑙𝑒𝑎𝑣𝑎𝑔𝑒)
Factoring,
∑𝐼 𝑃(𝑠𝑢𝑏𝑚𝑜𝑡𝑖𝑓∈𝐼)×𝑃(𝑐𝑙𝑒𝑎𝑣𝑎𝑔𝑒|𝑠𝑢𝑏𝑚𝑜𝑡𝑖𝑓∈𝐼)
� 𝑃(𝑚𝑜𝑡𝑖𝑓𝑖 |𝑐𝑙𝑒𝑎𝑣𝑎𝑔𝑒)
𝑃(𝑚𝑜𝑡𝑖𝑓𝑖 ) �
∑𝐼 𝑃(𝑠𝑢𝑏𝑚𝑜𝑡𝑖𝑓∈𝐼)
× ∑ 𝑃(𝑠𝑢𝑏𝑚𝑜𝑡𝑖𝑓∈𝐽)×𝑃(𝑐𝑙𝑒𝑎𝑣𝑎𝑔𝑒|𝑠𝑢𝑏𝑚𝑜𝑡𝑖𝑓∈𝐽)
=
𝑃(𝑚𝑜𝑡𝑖𝑓𝑗 ) � 𝐽
� 𝑃(𝑚𝑜𝑡𝑖𝑓𝑗 |𝑐𝑙𝑒𝑎𝑣𝑎𝑔𝑒)
∑𝐽 𝑃(𝑠𝑢𝑏𝑚𝑜𝑡𝑖𝑓∈𝐽)
(Eq. 1B: Bayesian statement after partition of motifi and motifj)
Thus, the relative cleavage (middle) term is the ratio of the average cleavage of submotifs
in I to the average cleavage of submotifs in J, weighted by the frequencies of the individual
submotifs in the initial pool. The equation is even more straightforward in three special cases:
•
Probabilities of cleavage are roughly equal between submotifs. This case is useful in this
particular analysis, because each sequence is unique by stipulation. Thus, in order to
227
consider any motif at all, we may reasonably assume that sequences outside the motif
contribute equally to cleavage.
Similarly, when considering the relative cleavage of
Watson-Crick pairs vs. nonpairs, we assume for the sake of argument that individual pairs
are equivalent, and that individual nonpairs are equivalent.
•
Submotifs are approximately equiprobable in the initial pool. This case primarily occurs
when comparing motifs composed exclusively of wildtype bases to motifs composed of
nonwildtype bases, e.g. CNNC vs. (not-C)NN(not-C). The initial pool was designed such
that nonwildtype bases are equiprobable, and sequencing of the initial pool shows
nonwildtype bases are indeed nearly equiprobable.
Probabilities of cleavage and submotif frequencies are independent. We cannot rely on
this case in much of the relevant analysis, since motifs composed of wildtype nucleotides
are more frequent in the pool, and are more likely to have higher probabilities of cleavage
than nonwildtype motifs, assuming that the wildtype sequence has evolved in nature to
optimize cleavage. Nevertheless, in this condition the weighted average converges to the
arithmetic average as the number of submotifs increases.
combinations
of
independent
cleavage
probabilities
demonstrates this:
8
4
2
1
0.5
Number of submotifs
228
1048576
65536
4096
256
16
0.25
1
(Weighted average)
(arithmetic average)
•
and
A simulation of 2000
submotif
frequencies
In any of these possibly overlapping situations, the equation collapses to:
∑ 𝑃(𝑐𝑙𝑒𝑎𝑣𝑎𝑔𝑒|𝑠𝑢𝑏𝑚𝑜𝑡𝑖𝑓∈𝐼)
𝐼
𝑃(𝑚𝑜𝑡𝑖𝑓𝑖 |𝑐𝑙𝑒𝑎𝑣𝑎𝑔𝑒)
𝑃(𝑚𝑜𝑡𝑖𝑓𝑖 )
|𝐼|
× ∑ 𝑃(𝑐𝑙𝑒𝑎𝑣𝑎𝑔𝑒|𝑠𝑢𝑏𝑚𝑜𝑡𝑖𝑓∈𝐽) =
𝐽
𝑃(𝑚𝑜𝑡𝑖𝑓𝑗 )
𝑃(𝑚𝑜𝑡𝑖𝑓𝑗 |𝑐𝑙𝑒𝑎𝑣𝑎𝑔𝑒)
|𝐽|
(Eq. 1C: Bayesian statement after equipartition of motifi and motifj)
and the relative cleavage (middle) term is simply the ratio of (unweighted) average cleavages of
the submotifs.
Calculation of the odds ratio score
The calculation of relative cleavage requires us to nominate a specific reference motif
(motifj). In situations where there is no obvious reference motif, and when rapidly screening for
enriched motifs, it is convenient to consider a single motif at a time, and simply compare it to the
aggregation of all other motifs. We use a score called the “odds ratio.” In principle, our
formulation is:
𝑃(𝑚𝑜𝑡𝑖𝑓𝑖 )
𝑃(𝑚𝑜𝑡𝑖𝑓𝑖 |𝑐𝑙𝑒𝑎𝑣𝑎𝑔𝑒)
× 𝑂𝑑𝑑𝑠 𝑅𝑎𝑡𝑖𝑜 =
1 − 𝑃(𝑚𝑜𝑡𝑖𝑓𝑖 )
1 − 𝑃(𝑚𝑜𝑡𝑖𝑓𝑖 |𝑐𝑙𝑒𝑎𝑣𝑎𝑔𝑒)
(Eq. 2A: “Odds Ratio” measure of relative cleavage for motifi over all other motifs)
This is a specific case of Eq. 1B, rewritten with the set I as before, and Ic, the complement
of I, composed of all submotifs of length n that are not members of set I:
∑𝐼 𝑃(𝑠𝑢𝑏𝑚𝑜𝑡𝑖𝑓∈𝐼)×𝑃(𝑐𝑙𝑒𝑎𝑣𝑎𝑔𝑒|𝑠𝑢𝑏𝑚𝑜𝑡𝑖𝑓∈𝐼)
�
�
𝑃(𝑚𝑜𝑡𝑖𝑓𝑖 )
𝑃(𝑚𝑜𝑡𝑖𝑓𝑖 |𝑐𝑙𝑒𝑎𝑣𝑎𝑔𝑒)
∑𝐼 𝑃(𝑠𝑢𝑏𝑚𝑜𝑡𝑖𝑓∈𝐼)
× ∑ 𝑃(𝑠𝑢𝑏𝑚𝑜𝑡𝑖𝑓∈𝐼
=
𝑐
𝑐
)×𝑃(𝑐𝑙𝑒𝑎𝑣𝑎𝑔𝑒|𝑠𝑢𝑏𝑚𝑜𝑡𝑖𝑓∈𝐼 )
1 − 𝑃(𝑚𝑜𝑡𝑖𝑓𝑖 ) � 𝐼𝑐
� 1 − 𝑃(𝑚𝑜𝑡𝑖𝑓𝑖 |𝑐𝑙𝑒𝑎𝑣𝑎𝑔𝑒)
𝑐
∑𝐼𝑐 𝑃(𝑠𝑢𝑏𝑚𝑜𝑡𝑖𝑓∈𝐼 )
(Eq. 2A: Reformulation of Eq. 1B for the Odds Ratio)
Thus, the odds ratio is the ratio of the average cleavage of submotifs in I to the average
cleavage of submotifs in Ic, weighted by the frequencies of the individual submotifs in the initial
pool. The odds ratio can be further simplified under the same special conditions as with the
relative cleavage measure.
One important caveat is that odds ratios cannot strictly be used to compare two motifs. To
see why this is so, consider two motifs of length n at the same position, motifi and motifj.
229
𝑂𝑑𝑑𝑠 𝑅𝑎𝑡𝑖𝑜𝑖 =
𝑂𝑑𝑑𝑠 𝑅𝑎𝑡𝑖𝑜𝑗 =
�∑𝐼 𝑃(𝑠𝑢𝑏𝑚𝑜𝑡𝑖𝑓∈𝐼)×𝑃(𝑐𝑙𝑒𝑎𝑣𝑎𝑔𝑒|𝑠𝑢𝑏𝑚𝑜𝑡𝑖𝑓∈𝐼)
�
∑ 𝑃(𝑠𝑢𝑏𝑚𝑜𝑡𝑖𝑓∈𝐼)
𝐼
𝑐 )×𝑃(𝑐𝑙𝑒𝑎𝑣𝑎𝑔𝑒|𝑠𝑢𝑏𝑚𝑜𝑡𝑖𝑓∈𝐼 𝑐 )
∑
� 𝐼𝑐 𝑃(𝑠𝑢𝑏𝑚𝑜𝑡𝑖𝑓∈𝐼
�
∑𝐼𝑐 𝑃(𝑠𝑢𝑏𝑚𝑜𝑡𝑖𝑓∈𝐼 𝑐 )
∑𝐽 𝑃(𝑠𝑢𝑏𝑚𝑜𝑡𝑖𝑓∈𝐽)×𝑃(𝑐𝑙𝑒𝑎𝑣𝑎𝑔𝑒|𝑠𝑢𝑏𝑚𝑜𝑡𝑖𝑓∈𝐽)
�
∑𝐽 𝑃(𝑠𝑢𝑏𝑚𝑜𝑡𝑖𝑓∈𝐽)
�
∑𝐽𝑐 𝑃(𝑠𝑢𝑏𝑚𝑜𝑡𝑖𝑓∈𝐽𝑐 )×𝑃(𝑐𝑙𝑒𝑎𝑣𝑎𝑔𝑒|𝑠𝑢𝑏𝑚𝑜𝑡𝑖𝑓∈𝐽𝑐 )
�
∑𝐽𝑐 𝑃(𝑠𝑢𝑏𝑚𝑜𝑡𝑖𝑓∈𝐽𝑐 )
�
Because the sets I and J are mutually exclusive, it follows that Ic≠Jc and the denominators
are not equal. Therefore, when directly comparing two motifs, the relative cleavage score (Eq.
1B) is superior. However, the Odds Ratio is still useful for considering many motifs at the same
time. Since
𝐼 𝑐 = 𝐽𝑐 + 𝐽 − 𝐼
it is intuitive that, as the number of motifs being considered increases,
𝑃(𝑚𝑜𝑡𝑖𝑓𝑖 ) → 0
and
𝑃�𝑚𝑜𝑡𝑖𝑓𝑗 � → 0
∑𝐽𝑐 𝑃(𝑠𝑢𝑏𝑚𝑜𝑡𝑖𝑓∈𝐽𝑐 )×𝑃(𝑐𝑙𝑒𝑎𝑣𝑎𝑔𝑒|𝑠𝑢𝑏𝑚𝑜𝑡𝑖𝑓∈𝐽𝑐 )
∑𝐽𝑐 𝑃(𝑠𝑢𝑏𝑚𝑜𝑡𝑖𝑓∈𝐽𝑐 )
∑𝐼𝑐 𝑃(𝑠𝑢𝑏𝑚𝑜𝑡𝑖𝑓∈𝐼 𝑐 )×𝑃(𝑐𝑙𝑒𝑎𝑣𝑎𝑔𝑒|𝑠𝑢𝑏𝑚𝑜𝑡𝑖𝑓∈𝐼 𝑐 )
∑𝐼𝑐 𝑃(𝑠𝑢𝑏𝑚𝑜𝑡𝑖𝑓∈𝐼 𝑐 )
→
and the ability to compare two odds ratio scores improves accordingly.
Calculation of the Watson–Crick base pairing score
To screen for specifically for Watson–Crick pairing between all possible combinations of
randomized positions, we used a scoring metric to compare the geometric average of odds ratios
for Watson–Crick pairing to that of odds ratios for non-Watson–Crick pairs. The score has no
fundamental meaning, and simply serves to identify position pairs where Watson–Crick
nucleotide identities are, on average, more preferred than non-Watson–Crick identities. It is also a
useful metric for prioritizing position pairs for followup analysis.
Pairing score = �
�
Watson–Crick
1/4
Odds ratio�
230
−�
�
non−Watson–Crick
1/12
Odds ratio�
Calculation of Information Content
To calculate the information content at each position, we use Bayes’ Theorem to infer the
distribution of bases after selection from a completely random pool, then calculate an information
content score based on the post-selection distribution.
We can use Bayes’ Theorem to infer how a distribution of the four bases changes after
selection, for any initial pool. Because we are considering all four bases at once, we must
consider the relative cleavage of any given base vs. the other three bases.
For clarity, the formula is shown for one base, A. According to Bayes’ Theorem for Total
Probability:
=
𝑃(𝑏𝑎𝑠𝑒 → 𝐴 𝑎𝑓𝑡𝑒𝑟 𝑠𝑒𝑙𝑒𝑐𝑡𝑖𝑜𝑛) = 𝑃(𝑏𝑎𝑠𝑒 → 𝐴|𝑐𝑙𝑒𝑎𝑣𝑎𝑔𝑒)
𝑃(𝑐𝑙𝑒𝑎𝑣𝑎𝑔𝑒|𝑏𝑎𝑠𝑒 → 𝐴) × 𝑃(𝑏𝑎𝑠𝑒 → 𝐴)
𝑃(𝑐𝑙𝑒𝑎𝑣𝑎𝑔𝑒|𝑏𝑎𝑠𝑒 → 𝐴) × 𝑃(𝑏𝑎𝑠𝑒 → 𝐴)
�+𝑃(𝑐𝑙𝑒𝑎𝑣𝑎𝑔𝑒|𝑏𝑎𝑠𝑒 → 𝐶) × 𝑃(𝑏𝑎𝑠𝑒 → 𝐶)
+𝑃(𝑐𝑙𝑒𝑎𝑣𝑎𝑔𝑒|𝑏𝑎𝑠𝑒 → 𝐺) × 𝑃(𝑏𝑎𝑠𝑒 → 𝐺)
+𝑃(𝑐𝑙𝑒𝑎𝑣𝑎𝑔𝑒|𝑏𝑎𝑠𝑒 → 𝑈) × 𝑃(𝑏𝑎𝑠𝑒 → 𝑈)
Reorganization shows:
𝑃(𝑏𝑎𝑠𝑒 → 𝐴|𝑐𝑙𝑒𝑎𝑣𝑎𝑔𝑒)
𝑃(𝑐𝑙𝑒𝑎𝑣𝑎𝑔𝑒|𝑏𝑎𝑠𝑒→𝐶) 𝑃(𝑏𝑎𝑠𝑒→𝐶)
𝑃(𝑏𝑎𝑠𝑒→𝐺)
×
+ 𝑃(𝑐𝑙𝑒𝑎𝑣𝑎𝑔𝑒|𝑏𝑎𝑠𝑒→𝐺)
×
= �1 + 𝑃(𝑐𝑙𝑒𝑎𝑣𝑎𝑔𝑒|𝑏𝑎𝑠𝑒→𝐴)
𝑃(𝑏𝑎𝑠𝑒→𝐴)
𝑃(𝑐𝑙𝑒𝑎𝑣𝑎𝑔𝑒|𝑏𝑎𝑠𝑒→𝐴) 𝑃(𝑏𝑎𝑠𝑒→𝐴)
𝑃(𝑏𝑎𝑠𝑒→𝑈)
+ 𝑃(𝑐𝑙𝑒𝑎𝑣𝑎𝑔𝑒|𝑏𝑎𝑠𝑒→𝑈)
×
�
𝑃(𝑐𝑙𝑒𝑎𝑣𝑎𝑔𝑒|𝑏𝑎𝑠𝑒→𝐴) 𝑃(𝑏𝑎𝑠𝑒→𝐴)
−1
In other words, the new probability depends on the relative cleavage of A vs. the other
nucleotides, and the relative abundance of A vs. the other nucleotides in the initial pool.
We can calculate the relative cleavage of one base vs. another, e.g. A vs. C, using the
selection data and Eq. 1A.
𝑅𝑎𝑡𝑖𝑜 𝑜𝑓 𝐶 𝑡𝑜 𝐴
𝑅𝑒𝑙𝑎𝑡𝑖𝑣𝑒 𝐶𝑙𝑒𝑎𝑣𝑎𝑔𝑒
𝑅𝑎𝑡𝑖𝑜 𝑜𝑓 𝐶 𝑡𝑜 𝐴
×
=
(𝑝𝑟𝑒 − 𝑠𝑒𝑙𝑒𝑐𝑡𝑖𝑜𝑛)
𝐶 𝑣𝑠. 𝐴
(𝑝𝑜𝑠𝑡 − 𝑠𝑒𝑙𝑒𝑐𝑡𝑖𝑜𝑛)
𝑃(𝑏𝑎𝑠𝑒 → 𝐶) 𝑃(𝑐𝑙𝑒𝑎𝑣𝑎𝑔𝑒|𝑏𝑎𝑠𝑒 → 𝐶) 𝑃(𝑏𝑎𝑠𝑒 → 𝐶|𝑐𝑙𝑒𝑎𝑣𝑎𝑔𝑒)
×
=
𝑃(𝑏𝑎𝑠𝑒 → 𝐴) 𝑃(𝑐𝑙𝑒𝑎𝑣𝑎𝑔𝑒|𝑏𝑎𝑠𝑒 → 𝐴) 𝑃(𝑏𝑎𝑠𝑒 → 𝐴|𝑐𝑙𝑒𝑎𝑣𝑎𝑔𝑒)
231
𝑃(𝑏𝑎𝑠𝑒 → 𝐶|𝑐𝑙𝑒𝑎𝑣𝑎𝑔𝑒)
𝑃(𝑐𝑙𝑒𝑎𝑣𝑎𝑔𝑒|𝑏𝑎𝑠𝑒 → 𝐶) 𝑃(𝑏𝑎𝑠𝑒 → 𝐴|𝑐𝑙𝑒𝑎𝑣𝑎𝑔𝑒)
�
=
𝑃(𝑏𝑎𝑠𝑒 → 𝐶)
𝑃(𝑐𝑙𝑒𝑎𝑣𝑎𝑔𝑒|𝑏𝑎𝑠𝑒 → 𝐴)
𝑃(𝑏𝑎𝑠𝑒 → 𝐴)
For a totally random initial pool, the bases are equiprobable, i.e.
Hence:
𝑃(𝑏𝑎𝑠𝑒 → 𝐴) = 𝑃(𝑏𝑎𝑠𝑒 → 𝐶) = 𝑃(𝑏𝑎𝑠𝑒 → 𝐺) = 𝑃(𝑏𝑎𝑠𝑒 → 𝑈)
𝑃𝑖𝑛𝑓𝑒𝑟𝑟𝑒𝑑 𝐴 = 𝑃(𝑏𝑎𝑠𝑒 → 𝐴|𝑐𝑙𝑒𝑎𝑣𝑎𝑔𝑒)
𝑃(𝑐𝑙𝑒𝑎𝑣𝑎𝑔𝑒|𝑏𝑎𝑠𝑒→𝐶)
= �1 + 𝑃(𝑐𝑙𝑒𝑎𝑣𝑎𝑔𝑒|𝑏𝑎𝑠𝑒→𝐴)
+ 𝑃(𝑐𝑙𝑒𝑎𝑣𝑎𝑔𝑒|𝑏𝑎𝑠𝑒→𝐺)
+ 𝑃(𝑐𝑙𝑒𝑎𝑣𝑎𝑔𝑒|𝑏𝑎𝑠𝑒→𝑈)
�
𝑃(𝑐𝑙𝑒𝑎𝑣𝑎𝑔𝑒|𝑏𝑎𝑠𝑒→𝐴)
𝑃(𝑐𝑙𝑒𝑎𝑣𝑎𝑔𝑒|𝑏𝑎𝑠𝑒→𝐴)
−1
This is the inferred fraction of A bases after selection from a random pool. Knowing this
fraction, we can calculate the information “contribution” for that base:
𝐼𝐴 = 𝑃𝑖𝑛𝑓𝑒𝑟𝑟𝑒𝑑 𝐴 × �𝑙𝑜𝑔2 �𝑃𝑖𝑛𝑓𝑒𝑟𝑟𝑒𝑑 𝐴 � + 2�
And the total information at that position is:
𝐼𝑡𝑜𝑡𝑎𝑙 = 2 + ∑𝑎𝑙𝑙 𝑏𝑎𝑠𝑒𝑠 𝑃𝑖𝑛𝑓𝑒𝑟𝑟𝑒𝑑 × 𝑙𝑜𝑔2 �𝑃𝑖𝑛𝑓𝑒𝑟𝑟𝑒𝑑 �
232
233
234
Appendix 3.
Mammalian microRNAs: experimental evaluation of novel and
previously annotated genes
H. Rosaria Chiang1,2, Lori W. Schoenfeld1,2, J. Graham Ruby3, Vincent C. Auyeung1,2,4, Noah
Spies1,2, Daehyun Baek1,2, Wendy K. Johnston1,2, Carsten Russ5, Shujun Luo6, Joshua E.
Babiarz7, Robert Blelloch7, Gary P. Schroth6, Chad Nusbaum5, David P. Bartel1,2
1
Whitehead Institute for Biomedical Research, Cambridge, MA 02142, USA
2
Howard Hughes Medical Institute and Department of Biology, Massachusetts Institute of
Technology, Cambridge, MA 02139, USA
3
Department of Biochemistry and Biophysics, University of California San Francisco, San
Francisco, CA 94158, USA
4
Harvard-MIT Division of Health Sciences and Technology, Cambridge, MA 02139, USA
5
Broad Institute of MIT and Harvard, Cambridge, MA 02141, USA
6
Illumina, Inc., Hayward, CA 94545, USA
7
Institute for Regeneration Medicine, Center for Reproductive Sciences, and Department of
Urology, University of California San Francisco, San Francisco, CA 94143, USA
Published as:
Chiang, et al. Mammalian microRNAs: experimental evaluation of novel and previously
annotated genes. Genes and Development. 2010 May 15;24(10):992-1009.
Contribution: V.C.A. performed the analysis of RNA editing.
235
Mammalian microRNAs: experimental
evaluation of novel and previously
annotated genes
H. Rosaria Chiang,1,2 Lori W. Schoenfeld,1,2 J. Graham Ruby,1,2,7 Vincent C. Auyeung,1,2,3 Noah Spies,1,2
Daehyun Baek,1,2 Wendy K. Johnston,1,2 Carsten Russ,4 Shujun Luo,5 Joshua E. Babiarz,6
Robert Blelloch,6 Gary P. Schroth,5 Chad Nusbaum,4 and David P. Bartel1,2,8
1
Whitehead Institute for Biomedical Research, Cambridge, Massachusetts 02142, USA; 2Howard Hughes Medical Institute and
Department of Biology, Massachusetts Institute of Technology, Cambridge, Massachusetts 02139, USA; 3Harvard-Massachusetts
Institute of Technology Division of Health Sciences and Technology, Cambridge, Massachustts 02139, USA; 4Broad Institute of
Massachusetts Institute of Technology and Harvard, Cambridge, Massachusetts 02141, USA; 5Illumina, Inc., Hayward,
California 94545, USA; 6Institute for Regeneration Medicine, Center for Reproductive Sciences, and Department of Urology,
University of California at San Francisco, San Francisco, California 94143, USA
MicroRNAs (miRNAs) are small regulatory RNAs that derive from distinctive hairpin transcripts. To learn
more about the miRNAs of mammals, we sequenced 60 million small RNAs from mouse brain, ovary, testes,
embryonic stem cells, three embryonic stages, and whole newborns. Analysis of these sequences confirmed 398
annotated miRNA genes and identified 108 novel miRNA genes. More than 150 previously annotated miRNAs
and hundreds of candidates failed to yield sequenced RNAs with miRNA-like features. Ectopically expressing
these previously proposed miRNA hairpins also did not yield small RNAs, whereas ectopically expressing the
confirmed and newly identified hairpins usually did yield small RNAs with the classical miRNA features,
including dependence on the Drosha endonuclease for processing. These experiments, which suggest that previous
estimates of conserved mammalian miRNAs were inflated, provide a substantially revised list of confidently
identified murine miRNAs from which to infer the general features of mammalian miRNAs. Our analyses also
revealed new aspects of miRNA biogenesis and modification, including tissue-specific strand preferences,
sequential Dicer cleavage of a metazoan precursor miRNA (pre-miRNA), consequential 59 heterogeneity, newly
identified instances of miRNA editing, and evidence for widespread pre-miRNA uridylation reminiscent of
miRNA regulation by Lin28.
[Keywords: MicroRNA; miRNA biogenesis; noncoding RNA genes; high-throughput sequencing]
Supplemental material is available at http://www.genesdev.org.
Received November 11, 2009; revised version accepted March 19, 2010.
MicroRNAs (miRNAs) are endogenous ;22-nucleotide
(nt) RNAs that post-transcriptionally regulate gene expression (Bartel 2004). miRNAs mature through three
intermediates: a primary miRNA transcript (pri-miRNA),
a precursor miRNA (pre-miRNA), and a miRNA:miRNA*
duplex. RNA Polymerase II transcribes the pri-miRNA,
which contains one or more segments that each fold into
an imperfect hairpin. For canonical metazoan miRNAs,
the RNase III enzyme Drosha together with its partner,
the RNA-binding protein DGCR8, recognize the hairpin,
and Drosha cleaves both strands ;11 base pairs (bp) from
the base of the stem (Han et al. 2006). The cut leaves a
7
Present address: Department of Biochemistry and Biophysics, University of California San Francisco, San Francisco, CA 94158, USA.
8
Corresponding author.
E-MAIL dbartel@wi.mit.edu; FAX (617) 258-6768.
Article published online ahead of print. Article and publication date are
online at http://www.genesdev.org/cgi/doi/10.1101/gad.1884710.
992
59 phosphate and 2-nt 39 overhang (Lee et al. 2003). The
liberated pre-miRNA hairpin is then exported to the
cytoplasm by Exportin-5 (Yi et al. 2003; Lund et al.
2004). There, the RNase III enzyme Dicer cleaves off
the loop of the pre-miRNA, ;22 nt from the Drosha cut
(Lee et al. 2003), again leaving a 59 monophosphate and
2-nt 39 overhang. The resulting miRNA:miRNA* duplex,
comprised of ;22-nt strands from each arm of the original
hairpin, then associates with an Argonaute protein such
that the miRNA strand is usually the one that becomes
stably incorporated, while the miRNA* strand dissociates and is degraded.
In addition to canonical miRNAs, some miRNAs mature through pathways that bypass Drosha/DGCR8 recognition and cleavage. Members of the mirtron subclass of
pre-miRNAs are excised as intron lariats from the primiRNA by the spliceosome and, following debranching,
fold into Dicer substrates (Okamura et al. 2007; Ruby et al.
GENES & DEVELOPMENT 24:992–1009 Ó 2010 by Cold Spring Harbor Laboratory Press ISSN 0890-9369/10; www.genesdev.org
236
Mammalian microRNAs
2007a). For some mirtrons, known as tailed mirtrons,
a longer intron is excised such that only one end of the
pre-miRNA is generated by the spliceosome, whereas the
other end of the pre-miRNA matures through the Droshaindependent trimming of a 59 or 39 tail (Ruby et al. 2007a;
Babiarz et al. 2008). Members of another subclass of premiRNAs, called endogenous shRNAs, are suitable Dicer
substrates without preprocessing by either Drosha or the
spliceosome (Babiarz et al. 2008). Other small silencing
RNAs are generated from the sequential processing of long
hairpins or long bimolecular duplexes. These small RNAs
are classified as endogenous siRNAs rather than miRNAs
because they derive from extended duplexes that produce
many different small RNA species, whereas miRNAs
derive from distinctive hairpins that produce one or two
dominant species (Bartel 2004).
The first indication of the abundance of miRNA genes
came from sequencing small RNAs from mammals, flies,
and worms (Lagos-Quintana et al. 2001; Lau et al. 2001;
Lee and Ambros 2001). Hundreds of mammalian miRNAs
have been identified by Sanger sequencing of cloned
small RNA-derived cDNAs (Lagos-Quintana et al. 2001,
2002, 2003; Houbaviy et al. 2003; Berezikov et al. 2006b;
Landgraf et al. 2007). Some miRNAs, however, are expressed only in a limited number of cells or through a limited portion of development, and their rarity makes them
difficult to detect. Computational methods have been used
to identify mammalian miRNAs initially missed by sequencing, and some of these predicted miRNAs have been
evaluated experimentally—e.g., by rapid amplification of
cDNA ends (RACE) (Lim et al. 2003; Xie et al. 2005),
hybridization to RNA blots (Berezikov et al. 2005), microarrays (Bentwich et al. 2005), and RNA-primed array-based
Klenow extension (RAKE) (Berezikov et al. 2006b). Each of
these experimental methods, however, can yield false
positives. Indeed, recent work in invertebrates and plants
(Rajagopalan et al. 2006; Ruby et al. 2006, 2007b) has
shown that the fraction of erroneously annotated miRNAs
can be quite high, depending on the quality of the initial
computational predictions. Even when miRNA genes
are predicted correctly, the resolution of the prediction is
often insufficient to confidently determine the precise
59 end of the mature miRNA. Because miRNAs repress target mRNAs by pairing to the seed sequence, which is defined relative to the position of the miRNA 59 end, singlenucleotide resolution of 59-end annotations is required for
useful downstream analysis of their physiological consequences (Bartel 2009).
Another approach for finding miRNAs and other small
RNAs missed in the early discovery efforts is highthroughput sequencing (Lu et al. 2005). In mammals,
high-throughput sequencing methods that have contributed to miRNA discovery efforts have included massively
parallel signature sequencing (MPSS) (Mineno et al. 2006),
miRNA serial analysis of gene expression (miRAGE)
(Cummins et al. 2006), 454 pyrosequencing (Berezikov
et al. 2006a, 2007; Calabrese et al. 2007), and Illumina
sequencing (Babiarz et al. 2008; Kuchenbauer et al. 2008).
Here we use the Illumina sequencing-by-synthesis platform (Seo et al. 2004) for miRNA discovery in mice.
Analyses of these reads, combined with experimental
evaluation of newly identified miRNAs as well as previous annotations, led us to substantially revise the set
of confidently identified murine miRNAs, thereby providing a more accurate picture of the general features of
mammalian miRNAs and their abundance in the genome. In addition, our results revealed new aspects of
miRNA biogenesis and modification, including tissuespecific strand preferences, sequential Dicer cleavage of a
metazoan pre-miRNA, cases of consequential 59 heterogeneity, newly identified instances of miRNA editing,
and widespread pre-miRNA uridylation reminiscent of
Lin28-like miRNA regulation.
Results
We sequenced small-RNA libraries from three mouse
tissues—brain, ovary, and testes—as well as embryonic
day 7.5 (E7.5), E9.5, E12.5, and newborn. Combining these
data with data collected similarly from mouse embryonic
stem (ES) cells (Babiarz et al. 2008) yielded 28.7 million
reads between 16 nt and 27 nt in length that perfectly
matched the mouse genome assembly (Supplemental
Table 1). Of these reads, 79.3% mapped to miRNA
hairpins, and 7.1% mapped to other annotated noncoding
RNA genes (Supplemental Table 2). Because the sequencing protocol was selective for RNAs with 59 monophosphate and 39 hydroxyl groups, this dominance of miRNA
species was expected (Lau et al. 2001).
miRNA gene discovery
As when analyzing high-throughput data from invertebrates (Ruby et al. 2006, 2007b; Grimson et al. 2008), we
identified miRNA genes in mice by applying the following criteria: (1) expression of the candidate miRNA, with
a relatively uniform 59 terminus; (2) pairing characteristics of the predicted hairpin; (3) absence of annotation
suggesting non-miRNA biogenesis; (4) absence of proximal reads suggesting that the candidate is a degradation
intermediate; and (5) presence of reads corresponding to
a miRNA* species with potential to pair to the miRNA
candidate with ;2-nt 39 overhangs. Using a low-stringency genomic search strategy that considered the first
four criteria, 736 miRNA candidates were identified from
the total data set of mouse reads. Manual inspection of
these candidates, focusing on all five criteria, narrowed
the list to 465 canonical miRNA genes, 377 of which were
already annotated in miRBase version 14.0 (GriffithsJones 2004) and 88 of which were novel (Fig. 1A; Supplemental Fig. S1; Supplemental Table 3). We also found 14
mirtrons (including 10 tailed mirtrons), four of which
were already annotated, and 16 endogenous shRNAs, six
of which were annotated previously (Fig. 1B). When added
to the 88 novel canonical miRNA genes, the newly
identified mirtons and shRNAs raised the total number
of novel genes to 108.
Of these 108 genes, 36 appeared to be close paralogs of
previously annotated miRNA genes (most of which were
paralogs of mir-466, mir-467, or mir-669), producing
GENES & DEVELOPMENT
237
993
Chiang et al.
Figure 1. Mouse miRNAs and candidates initially identified by high-throughput sequencing. (A) Overlap between previously annotated miRNA hairpins (miRBase
version 14.0; green), miRNA candidates identified in the
current study, and the subset of these candidates that met
our criteria for classification as confidently identified
canonical miRNAs (red). Additional considerations increased the number of confidently identified canonical
miRNAs to 475. (B) Overlap between previously annotated mirtrons and shRNAs and the mirtrons and
shRNAs supported by our study, colored as in A.
miRNA reads that were identical to the previously
annotated miRNAs, creating ambiguity as to which loci
contributed to the sequenced reads. Most of these close
paralogs (35 of 36), as well as 14 other novel loci, were
clustered with annotated miRNAs. The 72 novel genes
with reads distinguishable from those of previously identified genes were expressed at a lower level than the previously annotated genes (median read counts 27 and 8206,
respectively), and, compared with previously annotated
miRNAs, a higher fraction of these novel miRNAs were
located within introns of annotated RefSeq (Pruitt et al.
2005) mRNAs (47% and 26%, respectively).
Experimental evaluation of unconfirmed miRNAs
Of 564 miRBase-annotated miRNA genes (including four
confirmed mirtons and six confirmed shRNAs) that map
to mm8 genome assembly, 157 annotated miRNAs did
not pass the filters for miRNA candidates (Fig. 1A,B;
Supplemental Fig. S1; Supplemental Table 4). Of these
157, 26 mapped to annotated rRNA and tRNA loci, 52
had no reads mapping to them, and another 72 had some
reads but in numbers deemed insufficient for confident
annotation. The remaining seven either had reads with
very heterogeneous 59 ends, which suggested nonspecific
degradation of a non-pri-miRNA transcript (mir-464, mir1937a, and mir-1937b); had many reads that mapped well
into the loop of the putative hairpin, which were inconsistent with Dicer processing (mir-451, mir-469, and mir805); or did not give a predicted fold with the requisite
pairing involving the candidate and predicted miRNA*
(mir-484) (Supplemental Fig. S2). For five of these seven,
we have no reason to suspect that they might be authentic miRNA genes. Among the remaining two, mir-484
might be regarded as a miRNA candidate because manual
refolding was able to generate a hairpin with the requisite
pairing, but, even so, this candidate lacked reads for the
predicted miRNA*. miR-451 is a noncanonical miRNA
generated from an unusual hairpin without production
of a miRNA:miRNA* duplex (S Cheloufi and G Hannon,
pers comm.). We do not suspect that any other annotated
miRNA genes failed to pass our filters for the same reason
as mir-451.
An additional 20 annotated miRNA hairpins were in
our set of candidates but failed the manual inspection
because they lacked predicted miRNA* reads even after
allowing for alternate hairpin structures. Hundreds of
994
candidates from other miRNA discovery efforts (Xie et al.
2005; Berezikov et al. 2006b) also failed to pass the filters,
usually because no reads mapped to them.
One of the annotated miRNA genes missing from our
data sets was mir-220, which had been predicted computationally using MiRscan as a miRNA gene candidate
conserved in humans, mice, and fish, and was supported
experimentally using RACE analysis of zebrafish small
RNAs (Lim et al. 2003). In contrast, the other 37 miRNAs
newly annotated by Lim et al. (2003) were among our
confirmed miRNAs. The absence of mir-220 in our data
sets might have reflected either very low expression in the
sequenced samples or inaccuracy of its annotation. Similarly, mir-207, annotated in a contemporaneous study that
cloned novel miRNAs from mouse tissues, was missing
from our data set, but another 27 miRNAs annotated from
that study were confirmed (Lagos-Quintana et al. 2003).
To evaluate whether the missing annotated miRNAs and
candidates represented authentic miRNAs, we developed
a moderate-throughput assay to examine if their respective
hairpins could be processed as miRNAs in cultured cells
(Fig. 2A). If these putative miRNAs were missing from our
data sets because they were not expressed in the sequenced
tissues or stages, we reasoned that they would probably be
detected in cells ectopically expressing their respective
hairpins, because most authentic miRNAs are processed
correctly from heterologous transcripts that include the
full hairpin flanked by ;100 nt of genomic sequence on
each side of the hairpin (Chen et al. 2004; Voorhoeve et al.
2006). Alternatively, if these putative miRNAs were missing because they were not authentic miRNAs and therefore
lacked the features needed for Drosha and Dicer processing,
they would not be sequenced from cells ectopically expressing their hairpins. To evaluate many hairpins simultaneously, we transfected pools of hairpin-expressing constructs into HEK293T cells and isolated small RNAs for
high-throughput sequencing.
The performance of 26 positive controls, chosen from
canonical human/mouse miRNAs confirmed by our sequencing from mice, illustrated the value of the assay. For
all but one of these controls, miRNA and miRNA* reads
were more abundant in the cells ectopically expressing the
hairpin than in the cells without the hairpin constructs
(Fig. 2B–D; Supplemental Figs. S3, S4). For example, both
hsa-miR-193b and mmu-miR-137 (from humans and mice,
respectively) were >10 fold overexpressed (Fig. 2B). The
positive controls included genes of tissue-specific miRNAs,
GENES & DEVELOPMENT
238
Mammalian microRNAs
Figure 2. Experimental evaluation of annotated miRNAs and previously proposed candidates. (A) Schematic of the expression vector
transfected into HEK293T cells. (B) Examples of the standard ectopic expression assay, transfecting plasmids indicated in the key. Reads
from the control transfection (no hairpin plasmid) were from endogenous expression in HEK293T cells. (C) Assay results for annotated
human miRNAs and published candidates. Bars are colored as in B; asterisks indicate detectable overexpression ($1 read from both the
anticipated miRNA and miRNA*, with miRNA and miRNA* combined expressed more than threefold over endogenous levels). (D)
Assay results for unconfirmed annotated mouse miRNAs and published candidates. Mouse controls were selected from miRNAs that
were sequenced from our mouse samples. Bars are colored as in B; detectable overexpression is indicated (asterisks). Shown are the
results compiled from two experiments (Supplemental Figs. S3, S4).
including mir-122 (liver), mir-133 (muscle), mir-223 (neutrophil), and several neuron-specific miRNAs, with the idea
that hairpins of tissue-specific miRNAs might require
tissue-specific factors for their processing, and therefore
might be sensitive to the potential absence of such factors
in HEK293T cells. Differences were observed, ranging from
;100 to 10,000 reads above the control transfection (Fig.
2C, hsa-mir-214 and hsa-mir-9-1, respectively), consistent
with the idea that factors absent in HEK293T cells might
play a role in processing of some miRNAs. Alternatively,
some miRNA hairpins might be processed less efficiently in
all cell types, perhaps because our vectors might not present
the hairpins in an optimal context for processing. Perhaps
hsa-mir-192, the control gene that did not overexpress in
our assay, lacked crucial processing determinants needed in
all cells. In either scenario, the very high sensitivity of highthroughput sequencing enabled miRNAs to be observed
from most of the less efficiently processed hairpins.
GENES & DEVELOPMENT
239
995
Chiang et al.
From the 52 annotated mouse miRNAs that our study
did not sequence, 17 miRNAs, including mir-220 and mir207, were tested in the ectopic expression assay. One, mir698, generated a single read corresponding to the annotated miRNA, and the rest failed to generate any reads
representing the annotated miRNA (Fig. 2D). From the 72
annotated miRNAs that we could not identify due to insufficient number of reads, 28 were tested, and only four
of these were found to be overexpressed (Fig. 2D). The
difficulty in overexpressing a canonical control miRNA
(hsa-miR-192) illustrates that our ectopic expression assay
cannot be used to prove conclusively that a particular
hairpin does not represent an authentic miRNA gene.
However, the inability to overexpress each of the 17
unsequenced miRNAs, as well as most of the 28 insufficiently sequenced miRNAs, strongly indicated that, overall, these annotations have been faulty, and that our failure
to detect previously annotated miRNAs in mouse samples
was not merely due to inadequate sequencing coverage.
We also tested 10 of the 20 annotated miRNA genes
that we identified as candidates but did not confidently
classify as miRNA genes because the predicted miRNA*
species was not sequenced. Four of seven genes without
a miRNA* read and one of three genes with substantially
offset miRNA* reads produced the predicted miRNA*
species in our ectopic expression assay (Fig. 2D). mir-184
and mir-489, both of which tested positive in this assay,
are conserved. mir-184 is conserved throughout mammals, and mir-489 is conserved to chicken, although the
miRNA seed, which is highly conserved in mammals and
chickens, differs in mice and rats. Thus, these two genes,
as well as mir-875, which is a broadly conserved gene without a miRNA* read, were added to our set of confidently
identified miRNA genes. Also added were mir-290, mir291a, mir-291b, mir-292, mir-293, mir-294, and mir-295,
which were missing in the genome assembly (mm8) used
in our analysis because they fall in the region of the genome
that is difficult to assemble. Including these 10 genes, plus
mir-451, brings the total number of confidently identified
miRNA genes to 506, which includes 475 canonical genes.
Our sets of confirmed and novel murine miRNAs also
provided the opportunity to evaluate results of more
recent computational efforts to find miRNAs conserved
among mammals. One set of studies predicted miRNAs
based on phylogenetic conservation, and then tested these
and additional murine-specific hairpins using RAKE and
cloning (Berezikov et al. 2005, 2006b). Among the 322
candidates supported by these experiments, 11 were in
our sets of miRNAs (two in our confirmed set, and nine in
our novel set), and another nine did not satisfy our annotation criteria but had at least one read consistent with
the predictions. Another study started with MiRscan predictions conserved in four mammals, and filtered these
predictions for potential seed pairing to conserved motifs
in 39 untranslated regions (UTRs) (Xie et al. 2005). Of
their 144 final candidates, 45 were paralogs of miRNAs
already published at the time of prediction. Of the
remaining 99 candidates, 27 were in our sets of miRNAs
(26 in our confirmed set and one in our novel set), and one
did not satisfy our annotation criteria but had three reads
996
consistent with the miRNA* of the predicted miRNA.
However, only four of the 27 confirmed miRNA genes
(4% of the 99 novel predictions) gave rise to the mature
miRNA with the predicted seed, suggesting that filtering
MiRscan predictions for potential seed pairing provided
little, if any, added benefit. This conclusion concurs
with a recent analysis of miRNA targeting: miRNAs that
are not conserved beyond mammals do not have enough
preferentially conserved sites to place these sites as
among the most conserved UTR motifs (Friedman et al.
2009). Therefore, it stands to reason that preferentially
conserved UTR motifs would provide little value for
predicting such miRNAs.
To investigate whether the computational candidates
might have been missed because of low expression in
tissues and stages from which we sequenced, we included
representatives from each study in our ectopic expression
assay. We randomly selected 12 Xie et al. (2005) candidates and eight Berezikov et al. (2006b) candidates that our
study did not sequence, as well as four human candidates
from the Berezikov et al. (2005) set whose mouse orthologs were not sequenced. None generated reads representing the candidate miRNAs (Fig. 2C,D). Taken together,
our results raise new questions regarding the authenticity
of these candidates, and suggest that previous extrapolation from these candidates, which had suggested that
mammals have a surprisingly high number of conserved
miRNA genes (as many as 1000) (Berezikov et al. 2005),
should be revised accordingly.
Experimental evaluation of novel miRNAs
and new candidates
We also used the ectopic expression assay to evaluate
novel miRNAs identified from our sequencing. Of the 25
evaluated hairpins, 18 (72%) generated a significant number of miRNA-like reads in HEK293T cells, indicating
that most, although perhaps not all, of our 108 novel
annotations represented authentic miRNAs (Fig. 3; Supplemental Figs. S5, S6). These 25 hairpins were selected
arbitrarily for evaluation, except for a preference for rare
miRNAs; i.e., those that had <10 mature miRNA reads.
The rare miRNAs and the higher-abundance miRNAs
performed similarly (five of seven and 11 of 14 positives,
respectively).
To evaluate Drosha and Dicer dependence of the overexpressed hairpins, the experiment was repeated with and
without a plasmid encoding a dominant-negative allele of
either Drosha or Dicer (Fig. 3A; Han et al. 2009). All but two
canonical miRNA controls and most of the novel canonical
miRNAs (16 of 17) responded to TNdrosha coexpression
(Fig. 3B; Supplemental Fig. S7). Fewer responded to
TNdicer, suggesting that this construct was less disruptive
of normal miRNA processing (Supplemental Fig. S7).
The tested hairpins included several noncanonical
miRNA precursors. The level of mmu-miR-1224, an annotated mirtronic miRNA (Berezikov et al. 2007), increased in the presence of TNdrosha, as expected if this
pre-miRNA had more access to Exportin-5 and Dicer
when the canonical pre-miRNAs were reduced (Grimm
GENES & DEVELOPMENT
240
Mammalian microRNAs
et al. 2006). Although mmu-miR-1839, an annotated
shRNA (Babiarz et al. 2008), did not overexpress, mmumiR-344e and mmu-miR-344f, novel shRNAs, did over-
express from our vector, and, as expected for shRNAs,
their biogenesis was Drosha-independent (Fig. 3B; Supplemental Figs. S5–S7). Repeating the ectopic expression
assay in Dicer knockout and control cells confirmed that
mmu-miR-344e biogenesis was Dicer-dependent (data
not shown).
We also evaluated our candidates that had not satisfied
our criteria for confident annotation as miRNAs, usually
because they lacked reads representing the predicted
miRNA*. We tested three sets of these candidates. One
set represented our candidates that lacked predicted
miRNA* reads, yet, based on small RNA sequencing results from wild-type and mutant ES cells (Babiarz et al.
2008), appeared DGCR8- and Dicer-dependent. Another
set represented candidates that appeared conserved in
syntenic regions of other mammalian genomes, and the
third set was selected at random from among the remaining candidates. All but one of the 28 tested candidates
failed to generate miRNA-like reads, and the processing
of the candidate that did generate miRNA-like reads in
HEK293T cells was not dependent on Dicer, based on its
presence in Dicer knockout ES cells (Babiarz et al. 2008).
The results evaluating the novel miRNAs and candidates illustrated the importance of requiring a convincing
miRNA* read as a criterion for confident miRNA annotation. Five previously annotated miRNAs that were
initially rejected due to lack of a convincing miRNA*
read had tested positive in our overexpression assay (Fig.
2D), which indicated that this criterion was too stringent
for some of the previously annotated genes. However, the
results for the newly identified miRNAs and candidates
showed that the presence of a convincing miRNA* read
was the primary criterion that distinguished the novel
canonical miRNAs (most of which tested positive) from
the remaining candidates (nearly all of which tested
negative). By requiring a convincing miRNA* read in addition to the other four annotation criteria, our approach
accurately distinguished miRNA reads from the millions
of other small RNA reads generated by high-throughput
sequencing, with relatively few false positives among the
novel annotations and few false negatives among the
rejected candidates.
miRNA expression profiles
To compare expression levels of each miRNA in different
sequenced samples, we constructed relative miRNA
expression profiles (Fig. 4; Supplemental Table 5), and to
compare the relative expression of various miRNAs with
Figure 3. Experimental evaluation of novel miRNAs and
candidates. (A) Examples of assays evaluating Drosha dependence, transfecting plasmids indicated in the key. (B) Assay
results for control miRNAs, novel miRNAs, and miRNA
candidates. Bars are colored as in A; detectable overexpression
(black asterisks), overexpression attempted but not detected
(black minus sign), detectable Drosha dependence (orange asterisks), and Drosha dependence assayed but not detected
(orange minus sign) are all indicated. Shown are the results
compiled from three experiments (Supplemental Figs. S5–S7).
GENES & DEVELOPMENT
241
997
Chiang et al.
Figure 4. miRNA relative expression profiles. Profiles of mature miRNAs were constructed as described (Ruby et al. 2007b). The
relative contribution of each miRNA from each sample and the sum of the normalized reads of all samples are provided (Supplemental
Table 5).
each other, we generated a table of overall miRNA
abundance (Supplemental Table 5). Most miRNAs had
substantially stronger expression in some tissues or
stages than in others, in agreement with previous observations (Wienholds et al. 2005). We expect that strong
tissue- or stage-specific expression preferences inferred
from our limited sample set will be revised as more
tissues and stages are surveyed.
General features of mammalian miRNAs
Our analyses of high-throughput sequencing data and
subsequent experimental evaluation reshaped the set of
known murine miRNAs, setting aside 173 questionable
998
annotations and adding 108 novel miRNA genes to bring
the total number of confidently identified murine genes
to 506. A majority (60%) of the 506 genes appeared conserved in other mammals (Supplemental Fig. S1; Supplemental Table 6). However, only 15 of the 108 novel
miRNA genes were conserved in other mammals, suggesting that the number of nonconserved miRNA genes
will soon surpass that of conserved ones as high-throughput sequencing is applied more deeply and more broadly.
Five novel miRNAs (mir-3065, mir-3071, mir-3074-1, mir3074-2, and mir-3111) mapped to the antisense strand of
previously annotated miRNAs (mir-338, mir-136, mir-24-1,
mir-24-2, and mir-374, respectively), which, when added to
the previously identified mir-1-2/mir-1-2-as pair, brings
GENES & DEVELOPMENT
242
Mammalian microRNAs
the total number of sense/antisense miRNA pairs to six. In
addition, the mir-486 hairpin has a palindromic sequence,
which resulted in the same reads mapping to both the
sense (mir-486) and antisense (mir-3107) hairpins. Analysis
of the antisense loci of all 498 miRNA genes identified six
additional loci that gave rise to some antisense reads
resembling miRNAs (antisense loci of mir-21, mir-126,
mir-150, mir-337, mir-434, and mir-3073). As more highthroughput data is acquired, these as well as other antisense loci are likely to be annotated as miRNA genes.
However, <0.00002 of our miRNA reads corresponded to
miRNAs from antisense loci (excluding the reads mapping
ambiguously to mir-486/mir-3107), raising the possibility
that none of the murine antisense miRNAs have a function
comparable with that of miR-iab-as in flies (Bender 2008;
Stark et al. 2008; Tyler et al. 2008).
Our substantially revised set of miRNA genes provided
the opportunity to speak to the general features of 475
canonical miRNAs in mice, with the properties of the 295
conserved genes applying also to the conserved genes of
humans and other mammals (Table 1). Most canonical
miRNA genes (61%) were clustered in the genome, falling
within 50 kb of another miRNA gene, on the same
genomic strand. Even when excluding the four known
megaclusters (Calabrese et al. 2007), which are on chromosomes 2, 12 (two clusters), and X (with 69, 35, 16, and
18 genes, respectively), a sizable fraction of the remaining
genes (153 of 337) were in clusters of two to seven genes.
As observed in humans (Baskerville and Bartel 2005),
miRNAs from these loci within 50 kb of each other
tended to have correlated expression, consistent with
their processing from polycistronic pri-miRNA transcripts (Supplemental Fig. S8). In a scenario of one
transcript per cluster, the 475 canonical miRNA genes
would derive from 245 transcription units. In addition,
many miRNA hairpins mapped to introns. Just over a
third (38%) of the hairpins fell within introns of annotated mRNAs. Several lines of evidence—including coexpression correlations, chromatin marks, and directed
experiments—indicate that miRNAs can be processed
from introns (Baskerville and Bartel 2005; Kim and Kim
2007; Marson et al. 2008). In this scenario, as many as 107
Table 1.
Properties of canonical miRNAs
Total Conserved Nonconserved
Hairpins
Cluster analysis
In clusters
In small clusters
In large clusters
Not in clusters
Intron overlap
In introns (same strand)
Opposite introns
Not in introns
Arm preferences
With miRNA from 59 arm
With miRNA from 39 arm
With miRNAs from
both arms
475
295
180
291
153
138
184
163
129
34
132
128
24
104
52
180
22
273
77
18
200
103
4
73
202
141
137
102
65
39
132
56
76
(44%) of the 245 transcription units could double as premRNAs. Other hairpins were found within transcripts
that lacked other annotated functions, falling either
within introns or exons, or in transcripts without evidence of splicing.
miRNA hairpins are generally thought to each give
rise to a single dominant mature guide RNA. This was
usually the case for the murine miRNAs, although, as in
other species, this result relied on grouping together as
a single functional species all the isoforms that share the
same 59 terminus. This grouping is justified based on the
current understanding of miRNA target recognition, which
stipulates that heterogeneity often observed at miRNA 39
termini should have no effect on miRNA target recognition (Bartel 2009). Most mature miRNA reads (97%) were
20–24 nt in length, with 20mer, 21mer, 22mer, 23mer, and
24mer comprising 5%, 19%, 47%, 21%, and 4% of the
reads, respectively (Supplemental Fig. S9). Although a
single dominant mature species appears to be the most
frequent outcome of miRNA biogenesis, some miRNA
hairpins give rise to two or more species that each could
function to target different sets of mRNAs. This expanded
targeting potential arises from multiple mechanisms, including utilization of both strands of the miRNA:miRNA*
duplex with similar frequency, 59 heterogeneity, sequential
Dicer cleavage, and RNA editing. Addition of untemplated
nucleotides to the 39 termini of the miRNAs can also
occur, and although not thought to change targeting
specificity, these changes could indicate post-transcriptional regulation of miRNA stability. Occurrence of each
of these phenomena is described below.
miRNAs from both arms, with occasional
tissue-specific differences in the preferred arm
Most canonical miRNA genes produced one dominant
mature miRNA species, from either the 59 or 39 arm of the
pre-miRNA hairpin, with an overall tendency to derive
from the 59 arm (Table 1), as reported for previously annotated human miRNAs (Hu et al. 2009). Some, however,
yielded a similar number of reads from both arms, suggesting that the two species enter the silencing complex
with similar frequencies. For these genes, mature species
from the 59 and 39 arms were annotated using the -5p and
-3p suffixes, as is conventional in such cases (GriffithsJones 2004). Discrimination favoring one arm over the
other was less pronounced for both the nonconserved
miRNAs and the less highly expressed miRNAs (Fig. 5A),
although for the miRNAs with very few reads this trend
was likely enhanced by our requirement for a miRNA*
read. Overall, the discrimination was high, with the
species from the less dominant arm comprising 4.1% of
the reads that map to a miRNA or miRNA*. For the 10
most abundant miRNAs (sampling just the most abundant
member in cases of repetitive miRNAs), discrimination
was even higher, with the less dominant arm comprising
only 1.3% of the reads. Nevertheless, the miRNA* species
of these more highly expressed miRNAs were sequenced
at a median frequency 13-fold greater than that of the
median nonconserved miRNA, suggesting that a search for
GENES & DEVELOPMENT
243
999
Chiang et al.
Figure 5. Reads from both arms of a hairpin, and
sequential reads from the same arm. (A) Fraction and
abundance of miRNA reads from each miRNA hairpin. To calculate the fraction, the miRNA reads were
divided by the total number of miRNA and miRNA*
reads, considering on each arm only the major 59 terminus. The dashed lines indicate the median fraction of
miRNA reads and the median number of miRNA reads
for conserved (red) and nonconserved (blue) miRNAs.
(B) Switching of the dominant arm in different samples.
For each sample, the fold enrichment of miRNA reads
produced from the 59 arm over those produced from the
39 arm and vice versa was calculated. Shown are results
for nonrepetitive miRNAs that switch dominant arms,
with at least a fivefold differential between two samples. The samples are color-coded (key), and an asterisk
indicates samples with statistically significant enrichment of miRNAs produced from one arm over the
other (P < 0.05, x2 test). (C) Sequential Dicer cleavage.
Predicted secondary structure of mmu-mir-3102 premiRNA (Hofacker et al. 1994).
biological function for these miRNA* species might be at
least as fruitful as that for the poorly expressed nonconserved miRNAs.
If the mature miRNA accumulated preferentially from
one arm of the pre-miRNA hairpin, the preferred arm
generally remained consistent across the various libraries.
For a few miRNAs, however, the preferred arms switched
between samples (Fig. 5B), as reported previously using
PCR-based miRNA quantification (Ro et al. 2007). For
example, miR-142-5p was sequenced more frequently in
ovary, testes, and brain, and miR-142-3p was sequenced
more frequently in embryonic and newborn samples.
These results imply a developmental switch in targeting
preferences. A similar arm-switching phenomena has been
reported for a sponge miRNA (Grimson et al. 2008), and
was observed for 20 other nonrepetitive mouse miRNA
genes (Fig. 5B).
1000
Sequential Dicer cleavage of a mirtron hairpin
In plants, a few pri-miRNA hairpins with long, continuous RNA duplexes are cleaved sequentially by Dicer
to generate two adjacent miRNA:miRNA* duplexes
(Kurihara and Watanabe 2004; Rajagopalan et al. 2006).
Those precursors bear little resemblance to the shorter,
imperfectly base-paired hairpins of metazoan miRNA
genes. In mice, similar precursors are found in the form
of hairpin siRNA (hp-siRNA) precursors, but their expression appears to be limited to germline tissues and
totipotent ES cells, which lack a robust interferon response to intracellular dsRNA (Babiarz et al. 2008; Tam
et al. 2008; Watanabe et al. 2008). However, we detected
two miRNA:miRNA* duplexes deriving from the mmumir-3102 pre-miRNA hairpin, an apparent mirtron as
evidenced by reads mapping to both boundaries of an
GENES & DEVELOPMENT
244
Mammalian microRNAs
intron (Fig. 5C; Supplemental Table 3). After splicing and
debranching, the excised intron was predicted to fold into
a 104-nt pre-miRNA hairpin—substantially longer than
the average pre-miRNA length of 61 nt (calculated from
the set of confirmed miRNAs). Reads from this locus
suggested that Dicer cleaved this pre-miRNA twice, with
the first cut generating the outer miRNA:miRNA* duplex and the second cut generating the inner miRNA:
miRNA* duplex (Fig. 5C). The inner miRNA (miR3102.2-3p) was among a set of proposed miRNA candidates (Berezikov et al. 2006b), but the most frequently
sequenced species from this hairpin was the outer
miRNA (miR-3102.1) (Fig. 5C). Of the 16 genomes examined, the extended mir-3102 hairpin with both the inner
and outer miRNAs appeared conserved only in rats,
although the orthologous loci in cows, dogs, and humans
also could fold into shorter hairpins, with miR-3102.1
potentially conserved in cows.
We suspect that it is more than a coincidence that the
single metazoan example of a sequentially diced miRNA
is initially processed by the spliceosome rather than by
Drosha. One way to explain this observation is that
DGCR8/Drosha interacts directly with the loop of primiRNA stem–loops when recognizing its substrates
(Zeng et al. 2005), and that the lack of sequentially diced
Drosha-dependent miRNA hairpins in animals reflects
the limited reach of this complex.
59 Heterogeneity
Most conserved miRNAs had very precise 59 processing,
with alternative 59 isoforms comprising only 8% of all
miRNA reads (Fig. 6A,B). These results, analogous to
those observed in worms and flies (Ruby et al. 2006,
2007b), are consistent with the idea that selective pressure to avoid off-targeting acts to optimize precision of
the cleavage event that produces the 59 terminus of the
dominant species so as to prevent a consequential number of molecules with seed sequences in the wrong register. Moreover, 59 termini of conserved miRNAs were
more precise than those of miRNA* reads (4% and 12%
offset reads, respectively, excluding those that produce
comparable numbers of small RNAs from each arm). For
cases in which Dicer produced the 59 terminus of the
miRNA, the Dicer cut appeared somewhat more precise
than the Drosha cut (5% offset reads for miRNAs on the
39 arm, compared with 7% offset reads for miRNA* on
the 59 arm), hinting that features of the pre-miRNA structure may supplement the distance from the Drosha cut as
determinants of Dicer cleavage specificity (Ruby et al.
2006, 2007b).
A few miRNAs had less uniform 59 termini (Fig. 6A,B).
For some miRNAs, 59 heterogeneity has been documented
previously (Ruby et al. 2007b; Stark et al. 2007; AzumaMukai et al. 2008; Wu et al. 2009), the most prominent
example being hsa-miR-124, a conserved neuronal miRNA
for which the 59-shifted isoform was initially annotated as
the miRNA and eventually replaced by the more prominent isoform following more extensive sequencing (LagosQuintana et al. 2002; Landgraf et al. 2007). Another pro-
minent miRNA with unusually diverse 59 termini was
miR-133a. This conserved miRNA, which is highly
expressed in heart and muscle, had a second dominant
isoform (miR-133a.2) that was shifted 1 nt downstream
from the annotated miRNA (miR-133a.1) (Fig. 6C; Supplemental Table 3). To test whether this heterogeneity might
be explained by differential processing of the two mir-133a
paralogous hairpins, as observed for the two Drosophila
mir-2 hairpins (Ruby et al. 2007b), we tested the two mir133a hairpins in our ectopic expression assay. Although
mir-133a-1 was somewhat more prone to produce the miR133a.2 isoform, both hairpins produced a substantial
amount of both isoforms (Fig. 6C).
To investigate the functional consequences of miRNA
59 heterogeneity, we examined published array data
showing the responses of mRNAs after deleting either
mir-223, a miRNA with substantial heterogeneity, or mir155, a miRNA with little heterogeneity. miR-223 is
highly expressed in neutrophils, and analysis of small
RNA sequences from isolated neutrophils (Baek et al.
2008) was consistent with our sequencing results (Supplemental Table 3) in showing 59 heterogeneity, with 81%
of the reads mapping to the 59 end of the major isoform
miRNA and 12% mapping to the 59 end of a second
isoform that was shifted by 1 nt in the 39 direction (Fig.
6D). As expected, mRNAs with canonical 7–8mer sites
(Bartel 2009) matching the seed of the major isoform were
significantly derepressed in the mir-223 deletion mutant
(P < 10 12, Kolmogorov–Smirnov [K–S] test, compared
with no site distribution). mRNAs with canonical sites
matching the minor isoform also showed a significant
tendency to be derepressed, albeit to a lesser degree (P =
0.0022 3 10 7, 0.013 3 10 7, and 1.7 3 10 7, for 8mer,
7mer-m8, and 7–8mers combined, respectively) (Fig. 6D).
This result could not be attributed to the overlap between
sites matching the major and minor isoforms because all
mRNAs with a 6mer seed match to the major isoform
(ACUGAC) were excluded, and additional analyses ruled
out participation of the ‘‘shifted 6mer’’ match (Friedman
et al. 2009) to the major isoform (AACUGA) (Supplemental
Fig. S10A). Analogous analysis of miR-155 yielded strong
evidence for function of the major isoform (Rodriguez et al.
2007) but no sign of function for the minor isoform, which
comprised very few (1%) of our miR-155 reads (Fig. 6E;
Supplemental Table 3).
Taken together, our results show that some miRNAs
have alternative 59 miRNA isoforms that are expressed at
levels sufficient to direct the repression of a distinct set of
endogenous targets and thereby broaden the regulatory
impact of the miRNA genes. Therefore, we suggest that,
rather than choosing one isoform over the other for
annotation as the authentic miRNA, more of these alternative isoforms should be annotated, with the expectation that, for some highly expressed miRNAs, more
than one 59 isoform contributes to miRNA function.
RNA editing
RNA editing in which adenosine is deaminated and
thereby converted to inosine (I) has been reported for
GENES & DEVELOPMENT
245
1001
Chiang et al.
Figure 6. miRNAs with 59 heterogeneity. (A) The distribution of conserved (red) and nonconserved (blue) miRNAs with reads #5 nt
offset at their 59 terminus. (B) The fraction of offset reads and abundance of reads for each miRNA hairpin, colored as in A. The dashed
lines indicate the median level of reads for conserved (red) and nonconserved (blue) miRNAs. (C) 59 Heterogeneity of miR-133a. Data
from mouse heart (Rao et al. 2009) and newborn are mapped to the mir-133a-1 hairpin (top), and data from the ectopic expression assay
are mapped to the indicated transfected hairpin (bottom). The lines indicate miR-133a.1 (dark blue) and miR-133a.2 (light blue), and red
nucleotides indicate those that differ between mir-133a-1 and mir-133a-2. (D) Effect of losing miR-223 on messages with 39 UTR sites
for miR-223 major and minor isoforms. (Top) Small RNA sequencing data from mouse neutrophils (Baek et al. 2008) were mapped to the
mir-223 hairpin as in C. For each set of messages with the indicated 39 UTR site for miR-233 (major isoform sites, bottom left; minor
isoform sites, bottom right), the fraction that changed at least to the degree indicated following loss of miR-223 is plotted, using data
published for neutrophils differentiated in vivo (Baek et al. 2008). (E) Effect of losing miR-155 on messages with 39 UTR sites for miR155 major and minor isoforms, plotted as in D using published data from T cells (Rodriguez et al. 2007). (Top) Sequencing data from our
study are mapped to the mir-155 hairpin as in C. The mRNAs with 8mer and 7mer-A1 sites for the minor isoform were excluded from
the analysis because these sites overlapped with 7mer-m8 sites for the major isoform.
1002
GENES & DEVELOPMENT
246
Mammalian microRNAs
some miRNA precursors (Blow et al. 2006; Landgraf et al.
2007; Kawahara et al. 2008). Because I pairs with C, such
edits could change miRNA target recognition. Reasoning
that the mammalian adenosine deaminases (ADARs) responsible for A-to-I editing are expressed primarily in the
brain, we searched for sequencing reads from the brain that
did not match the genome and had as their closest match
a mature miRNA or miRNA*. After filtering for mismatches occurring >2 nt from the 39 end, a step taken to
avoid considering instances of untemplated 39-terminal
addition, only 4% of the reads had single mismatches to
the genome (Supplemental Fig. S11A). Moreover, the
fraction of sequences with A-to-G changes (indicative of
A-to-I editing) was only 0.61%, a fraction resembling that
of other mismatches (Supplemental Fig. S11A). This
fraction was also similar to that of the A-to-G changes in
our synthetic internal standards used for preparing the
sequencing libraries. These results indicate that mature
edited miRNAs are very rare and difficult to distinguish
above the background level of sequencing errors. The low
frequency of editing in mature miRNAs was consistent
with the findings that edited processed miRNAs are more
than fourfold less common in mice relative to humans
(Landgraf et al. 2007), and are less common than edited
miRNA precursors (Kawahara et al. 2008). The latter observation might be due to rapid degradation or impaired
processing, which has been shown for miR-142 (Yang et al.
2006) and miR-151 (Kawahara et al. 2007a).
Although editing did not appear to be a widespread
phenomenon among all mature miRNAs, editing at
specific sites might still be important for a few individual
miRNAs. To investigate this possibility, mismatch fractions were calculated as the fraction of reads bearing
a particular mismatch over all reads covering that genomic
position. For each library, a change was considered significant if the fraction exceeded 5% and at least 10 reads contained the mismatch. Additional filters designed to remove sequencing errors, alignment artifacts, and instances
of untemplated nucleotide addition preferentially retained
A-to-G changes while removing nearly all other events
(Supplemental Fig. S11B). Sixteen A-to-G events passed the
filters and subsequent manual examination, all of which
occurred only in the brain library (Table 2). Five of these
inferred editing sites were also observed in a low-throughput sequencing effort in human brain samples (Kawahara
et al. 2008), indicating that editing of some miRNAs is
conserved between mammals. Consistent with that study,
eight of 16 editing sites occurred in a UAG motif. A separate examination of read alignments with up to three
mismatches showed that the vast majority of edited reads
were edited at one position, suggesting that either editing
of multiple sites in the same RNA molecule is rare, or
multiply edited RNAs are degraded more rapidly.
A-to-I editing of a seed nucleotide would dramatically
affect targeting. In addition to editing in the miR-376 cluster described previously (Kawahara et al. 2007b, 2008), we
found another eight miRNAs that are edited within the
seed of either the miRNA or the miRNA*. A-to-I editing
could also affect miRNA loading, and thereby indirectly
affect targeting. Indeed, the editing of miR-540 might
Table 2.
Inferred A-to-I editing sites in miRNAs
miRNA
miR-219-2-3p
miR-337-3p
miR-376a*
miR-376b-3p
miR-376c
miR-378
miR-379*
miR-381
miR-411-5p
miR-421
miR-467d
miR-497
miR-497*
miR-540*
miR-1251
miR-3099
Position
Fraction edited
15
10
4
6
6
16
5
4
5
14
3
2
20
3
6
7
0.064
0.062
0.297
0.501
0.311
0.087
0.095
0.125
0.239
0.054
0.094
0.104
0.699
0.080
0.431
0.209
help explain why the 59 arm is more abundant in the brain
than in other tissues, although editing is too infrequent to
fully explain the switch in strand bias. Altering Drosha
and Dicer processing could also indirectly affect targeting. Analysis of 59 ends showed that seven of 16 instances
of editing were associated with a statistically significant
(P < 0.05) shift in the 59 nucleotide, presumably due to
changes in the Drosha and Dicer cleavage site (Supplemental Fig. S11D).
Untemplated nucleotide addition
Much more prevalent than editing of internal nucleotides
was addition of untemplated nucleotides to miRNA 39
termini. As reported previously for miRNAs in mammals
(Landgraf et al. 2007), and also observed for those of worms
and flies (Ruby et al. 2006, 2007b), nucleotides most
frequently added to murine miRNAs were U and A (Fig.
7A). Addition of C or G was no higher than background, as
estimated by monitoring apparent addition to tRNA
fragments (Fig. 7A). Possible sources of the background
rate could be sequencing error, transcription error, or a low
level of biological nucleotide addition. Some miRNAs
were much more frequently extended than others (Supplemental Table 7). One very frequently extended miRNA
was miR-143, for which the extended reads outnumbered
the nonextended ones (196,565 compared with 114,980
reads, respectively).
For extension by U, RNAs from the pre-miRNA 39 arm
were three times more frequently extended than were
those from the 59 arm (Fig. 7A,B, P = 2.3 3 10 4, K–S test).
This preference, not observed for the A extension (Fig.
7A,C), suggests that much of the U extension occurs to the
pre-miRNA, prior to Dicer cleavage—a state in which the
39 arm but not the 59 arm would be available for extension
(Fig. 7D). TUT4-catalyzed poly(U) addition to the let-7 premiRNA, which is specified by Lin28, plays an important
role in post-transcriptional repression of let-7 expression
(Heo et al. 2008, 2009; Hagan et al. 2009). Our analyses
indicating untemplated U extension to many other premiRNAs hint that this type of regulation may not be
GENES & DEVELOPMENT
247
1003
Chiang et al.
Figure 7. Untemplated nucleotide addition. (A)
Untemplated nucleotide addition rate for miRNA
and miRNA* reads from the indicated arm. Rates for
each miRNA are provided (Supplemental Table 6).
As a control, tRNA degradation fragments were
analyzed similarly. Numbers of genes analyzed are
indicated in parentheses. (B) Distribution of rates for
untemplated U addition to RNAs from the 59 arm
(blue) and from the 39 arm (red). (C) Distribution of
rates for untemplated A addition to RNAs from the
59 arm (blue) and from the 39 arm (red). (D) Schematic of the biogenesis stage in which U could be
added to the RNA of only one arm (pre-miRNA,
left), and the stage in which U could be added to the
RNA of either arm (mature miRNA and miRNA*,
right).
limited to let-7, but that analogous pathways, presumably
using mediators other than Lin28, act to regulate the
expression of other murine miRNAs.
Discussion
The status of miRNA gene discovery in mammals
Our current study sets aside nearly a third (173 of 564) of
the miRBase version 14.0 gene annotations for lack of convincing evidence that these produce authentic miRNAs.
It also adds another 108 novel miRNA loci, raising the
question of how many more authentic loci remain undiscovered. This question is difficult to answer. Ever
since the recognition that the poorly conserved miRNAs
are also the ones expressed at lower levels in mammals,
and thus are the most difficult to detect by both computational and experimental methods, we have known that
it is impossible to provide a meaningful estimate of the
number of mammalian miRNA genes remaining to be
discovered (Bartel 2004). The broadly conserved miRNAs
are another matter. Only three of the 88 novel canonical
miRNAs had recognizable orthologs sequenced in chickens,
lizards, frogs, or fish, and these three were antisense to previously annotated broadly conserved miRNA genes. Therefore, apart from miRNAs expressed at very low levels from
the antisense strand of known genes, we suspect that the
list of broadly conserved miRNA gene families is nearing
completion. The current set of murine miRNA genes
includes 192 genes that fall into 89 broadly conserved
miRNA gene families (Supplemental Table 6).
Another 107 miRNA gene families appeared conserved
in other mammals (Supplemental Table 6). These were
represented by 120 murine genes, including 14 novel
1004
genes. Of these novel genes, 11 were founding members
of novel conserved gene families. Some of these were
identified with only 11 reads, indicating that additional
pan-mammalian gene families remain to be found, although we have no evidence supporting the idea that the
number of conserved gene families will rise to the very
high levels suggested by some earlier computational
studies (Berezikov et al. 2005, 2006b; Xie et al. 2005). For
now, we can say that mammals have at least 196 conserved miRNA gene families represented in mice by at
least 312 pre-miRNA hairpins (303 canonical and nine
noncanonical hairpins) produced from at least 194 unique
transcription units.
Because a single miRNA hairpin can produce multiple
functional isoforms, generated by either 59 processing
heterogeneity or utilization of both arms of the miRNA
duplex, a single conserved hairpin can produce more than
one conserved miRNA isoform. Because the different
isoforms have different seed sequences, they fall into
different families of mature miRNAs. Thus, the number
of conserved families of miRNAs (i.e., mature guide
RNAs) will exceed the number of conserved families of
genes (i.e., hairpins). Perhaps the best known example of
a hairpin with two broadly conserved isoforms is mir-9,
for which conserved miRNAs from both arms of the
hairpin are readily detected by using in situ hybridization
in both zebrafish and marine annelids (Wienholds et al.
2005; Christodoulou et al. 2010). Numerous conserved
genes produce more than one miRNA isoform (Figs. 5A,
6A), but for most of these we do not yet know whether
production of the alternative isoform is conserved in
other species. High-throughput sequencing from other
species will help identify many additional conserved
GENES & DEVELOPMENT
248
Mammalian microRNAs
isoforms. We anticipate that the discovery of multiple
conserved isoforms will contribute much more to the
future growth in the list of broadly conserved miRNA
families than will the discovery of new conserved genes.
As expected, the conserved miRNAs tended to be expressed at much higher levels than were the nonconserved ones, with the median read frequency of conserved
miRNAs 44-fold greater than that of the nonconserved
miRNAs (Figs. 5A, 6B). Therefore, even if many nonconserved miRNA genes remained to be found, these
would add little to the number of annotated miRNA
molecules in a given cell or tissue, and presumably even
less to the impact of miRNAs on gene expression (Bartel
2009). Indeed, even more pressing than the question of
how many poorly conserved miRNAs remain undetected
is the question of whether any of the known poorly
conserved miRNAs have any consequential function in
the animal.
Most of these poorly conserved miRNAs could have
derived from transcripts that fortuitously acquired hairpin regions with features needed for some Drosha/Dicer
processing. In this scenario, most of these newly emergent miRNAs will be lost during the course of evolution
before ever acquiring the expression levels needed to have
a targeting function sufficient for their selective retention
in the genome. Consistent with the hypothesis that most
of these miRNAs play inconsequential regulatory roles,
these miRNAs generally accumulated to much lower
levels in our ectopic expression assay, (Fig. 3B, median
read frequencies of 58 and 844 for nonconserved and conserved miRNAs, respectively), and they displayed weaker
specificity for one arm of the hairpin (Fig. 5A), as would be
expected if there was no advantage for the cell to efficiently use their respective hairpins. Nonetheless, some
were processed efficiently, and at least a few poorly conserved miRNAs probably have acquired consequential
species-specific functions. Although none have known
functions, such hairpins are worthy of annotation as
miRNA loci (just as protein-coding genes can be annotated before the protein is known to be functional), and as
a class these newly emergent miRNAs could provide an
important evolutionary substrate for the emergence of
new regulatory activities.
The major challenge for miRNA gene discovery stems
from the difficulty in proving that a nonconserved, poorly
expressed candidate is an authentic miRNA, combined
with the even greater difficulty in proving that a questionable candidate is not an authentic miRNA. This challenge has become all the more acute as miRNA discovery
has reached the point to which nearly all of the novel
candidates are both nonconserved and poorly expressed.
Our approach of testing pools of candidates in an ectopic
expression assay provides useful data for evaluating
miRNA authenticity. However, our approach cannot
provide conclusive proof for or against the authenticity
of a proposed candidate, leaving open the possibility that
some of the nonconserved, poorly expressed candidates
that we classify as ‘‘confidently identified miRNAs’’ are
false positives. When considering the limitations of the
current tools for miRNA gene identification, this possi-
bility cannot be avoided. Therefore, if any nonconserved,
poorly expressed miRNAs are annotated as miRNAs, the
resulting list of miRNAs will have to be somewhat fuzzy,
with an expectation that some of the annotated genes will
not be authentic miRNAs. This expectation should not
be viewed as advocating the indiscriminant annotation of
all candidates as miRNAs. Our proposal is that miRNA
gene discovery efforts should annotate as miRNAs only
those novel candidates that both are found in highthoughput sequencing libraries and pass a set of criteria
that is sufficiently stringent such that a majority of
the novel canonical miRNAs are cleanly processed in a
Drosha-dependent manner when using the ectopic expression assay. Although implementing this proposal would
not prevent all false positives from entering the databases,
it would preserve a higher quality set of miRNAs while
eliminating few authentic annotations. Those wanting to
take additional measures to avoid false positives could
focus on only the subset of miRNAs that both meet these
criteria and are conserved in other species.
Unknown features required for Drosha/Dicer
processing
Before learning the results of our experiments, we wondered whether any ectopically overexpressed hairpin of
suitable length would be processed as if it were a miRNA,
a result that would have rendered our assay too permissive to be of value. In this scenario, most of the specificity
that distinguished authentic miRNA genes from other
regions of the genome with the potential to produce
transcripts that fold into seemingly miRNA-like hairpins
would have been a function of whether or not the regions
were transcribed. This scenario was not realized, however, and our assay turned out to be informative, which
illustrates how much of Drosha/Dicer substrate recognition still remains unknown. Many of the previously
proposed miRNA hairpins that had no reads in our mouse
samples were indistinguishable from authentic miRNA
hairpins with regard to the known determinants for
Drosha/Dicer recognition, yet none of these unconfirmed
hairpins produced miRNA and miRNA* molecules in our
very sensitive assay (Fig. 2C,D). These results showing
that major processing specificity determinants still remain undiscovered point to the importance of finding
these determinants—efforts that, if successful, will mark
the next substantive advance in accurately predicting and
annotating metazoan miRNAs.
Materials and methods
Library preparation
Total RNA samples from mouse ovary, testes, and brain were
purchased from Ambion, and total RNA from mouse E7.5, E9.5,
E12.5, and newborn were obtained from the Chess laboratory. The
small RNA cDNA libraries were made as described (Grimson
et al. 2008), except for the 39 adaptor ligation, which was 59
adenylated pTCGTATGCCGTCTTCTGCTTGidT. For a detailed
protocol, see http://web.wi.mit.edu/bartel/pub/protocols.html.
GENES & DEVELOPMENT
249
1005
Chiang et al.
miRNA discovery
The reads with inserts of 16–27 nt were processed as described
(Babiarz et al. 2008). The miRNA candidates were identified
using reads matching genomic regions that were not very highly
repetitive (reads with <500 genomic matches). Reads from all
data sets were combined and grouped by their 59-terminal loci,
requiring that each candidate 59 locus pass five criteria listed in
the text. (1) To pass the expression criterion, a candidate required
$10 normalized reads. (2) To address the hairpin requirement,
the secondary structure of the candidate was evaluated by
selecting for each 59-terminal locus the most abundant sequence
and extending its 59 end by 2 nt to define the range of the
potential miRNA/miRNA* duplex. Three genomic windows
were extracted with the 59 end extended an additional 10 nt
and the 39 end extended either 50 nt, 100 nt, or 150 nt. Three
more windows were extracted extending the 39 end by 10 nt and
the 59 end another 50 nt, 100 nt, or 150 nt. The secondary structure of each of the six windows was predicted using RNAfold
(Hofacker et al. 1994), and the number of hairpin base pairs (denoted using bracket notation) involving the 59-extended miRNA
candidate was calculated as the absolute value of ([number of
59-facing brackets] [number of 39-facing brackets]). A candidate
with a minimum of 16 bp using at least one of the six genomic
windows satisfied the hairpin criteria. (3) The candidates with
non-miRNA biogenesis were found by mapping to annotated
noncoding RNA loci (rRNA, tRNA, snRNA, and srpRNA). (4)
The candidates likely produced by degradation were defined
as those failing the 59 homogeneity requirement. A candidate
satisfied the 59 homogeneity requirement if at least half of the
reads within 30 nt of the candidate 59 end were present within
2 nt of the candidate 59 end and if the candidate 59 end comprised
at least half of the reads within 2 nt of the candidate 59 end, or
if there was only one other 59 end within 30 nt of the candidate 59 end that had more than half of the reads mapping to the
candidate 59 end. (5) Manual inspection of reads mapped to
predicted secondary structures identified candidates accompanied by potential miRNA* reads. For 10 previously annotated
miRNAs and seven novel miRNAs, a suitable miRNA* read
was found only after considering alternative hairpin folds predicted to be suboptimal using mfold (Mathews et al. 1999; Zuker
2003).
For the analysis of mir-290, mir-291a, mir-291b, mir-292,
mir293, mir-294, and mir-295, which are not present in mm8
genome assembly, we mapped all reads to mm9 genome assembly corresponding to the region [chr7(+): 3,218,627–3,220,842].
For conservation analysis, a candidate was considered broadly
conserved if the hairpin structure and the seed sequence were
conserved to chickens, fish, frogs, or lizards (galGal3, danRer5,
xenTro2, and anoCar1, respectively) in the University of California at Santa Cruz whole-genome alignments (Kuhn et al.
2009). To identify a candidate conserved in mammals, we looked
at 12 additional genomes (bosTau3, canFam2, cavPor2, equCab1,
hg18, loxAfr1, monDom4, ornAna1, panTro2, ponAbe2, rheMac2, and rn4) and calculated the branch length score from
a phylogenetic tree trained on mouse 39 UTR data (Friedman
et al. 2009), using the cutoff score of 0.7. A gene was considered
to be in a conserved miRNA gene family if the hairpin produced
a miRNA with a seed matching that of a conserved miRNA
(Supplemental Table 6).
Ectopic overexpression assays
To generate expression constructs, pre-miRNA hairpins and the
surrounding regions were amplified from human genomic DNA
(NCI-BL2126) or from mouse BL6 genomic DNA using Pfu Ultra II
1006
polymerase (Stratagene) and primers with Gateway (Invitrogen)compatible ends designed to anneal ;100 nt upstream of and
downstream from the miRNA hairpins. PCR products were inserted into Gateway vector pDONR221 and subsequently into
pcDNA3.2/V5-DEST, and the resulting plasmids were transformed
into DH5-a cells. Positive clones were selected by colony PCR and
were sequenced. Clones that did not have a mutation within premiRNA hairpins were selected. Plasmid DNA from the confirmed
expression clones was purified for transfection using the Plasmid
Mini Kit (Qiagen). For each standard assay, plasmids for up to 10
hairpin expression constructs were mixed in equal amounts to
create seven or eight pools of ;1.4 mg of DNA each, with each pool
including one to three positive control hairpins.
HEK293T cells were cultured in DMEM supplemented with
10% FBS, and were plated in 12-well plates ;24 h prior to
transfection to reach ;80%–90% confluency. Each well of cells
was transfected with one pool of DNA using Lipofectamine 2000
(Invitrogen). For the standard assays, 145–200 ng of pMaxGFP
(Amaxa) was cotransfected with each pool to enable transfection
efficiency to be confirmed by GFP expression. Control wells (no
hairpin plasmid) were transfected only with 145 ng of pMaxGFP.
For the Drosha/Dicer dependency assays, seven to eight hairpin
constructs were combined to create six pools of ;400 ng each.
Each pool was mixed with 1.2 mg of the pCK-Drosha-Flag(TN)
(TNdrosha), pCK-Flag-Dicer(TN) (TNdicer), or pCK-dsRed.T4
(control vector, constructed by replacing the Drosha-coding
sequence of TNdrosha with dsRed-coding sequence) and used to
transfect one well of HEK293T cells as above. Control wells were
transfected with 1.2 mg of either TNdrosha, TNdicer, or control
vector. For the dependency assays, each transfection was performed in duplicate wells. Cells from all assays were harvested
39–48 h after transfection. Cells from each treatment were
combined, total RNA was extracted using TriReagent (Ambion),
and small RNA libraries were prepared for Illumina sequencing.
The reads were processed as above, and RNA species were
matched to the transfected hairpins. In the standard assay, reads
were normalized by the median of the 30 most frequently
sequenced endogenous miRNAs. For assays testing Drosha/Dicer
dependency, reads were normalized based on the number of
reads corresponding to an 18-nt internal standard that had been
spiked into equivalent amounts of total RNA prior to beginning
library preparation. Reads matching the transfected hairpins were
grouped by their 59 termini (59-terminal locus). The locus with the
largest number of reads was considered the 59-terminal locus of
the mature miRNA produced by the hairpin, and similarly, the
most dominant 59 locus on the opposite arm was considered the
miRNA*. The normalized miRNA and miRNA* read numbers
were summed to calculate the expression level.
If an overexpressed hairpin generated mature miRNA with the
dominant 59-terminal locus corresponding to the expected locus
and at least one read corresponding to the miRNA* with an ;2-nt
39 overhang, it was considered expressed. A hairpin was classified
as overexpressed if there were at least threefold more reads in the
hairpin transfection than in the control transfection, after adding
psuedocounts of five to both. A hairpin was classified as Droshaor Dicer-dependent if the knockdown was at least threefold.
Identification of arm-switching miRNAs
To determine the read numbers from the 59 and 39 arms, reads
from each sample were grouped based on their 59 termini, and
the read numbers were tallied for those corresponding to the
miRNA or miRNA* 59 terminus. Only samples with five or
more reads on either arm were considered. The fold enrichment
was calculated as the ratio of 59 and 39 arm reads after adding
pseudocounts of one.
GENES & DEVELOPMENT
250
Mammalian microRNAs
RNA editing analysis
Sequencing libraries from individual tissues were combined and
mapped to the genome using the Bowtie alignment tool (Langmead
et al. 2009). The alignments were filtered for sequences that
uniquely aligned to the genome, contained at most one mismatch
to the genome, and had 59 ends that mapped to within 1 nt of an
annotated miRNA or miRNA* 59 end. The 12 possible mismatch
types were then quantified at each position covered by the filtered
reads. For example, to screen for A-to-G mismatches indicative of
A-to-I editing sites, the editing fraction was calculated as the
number of reads containing an A-to-G mismatch at a particular
position, divided by the number of filtered reads covering that
position. Sites were considered editing candidates if the editing
fraction was >5%, had at least 10 A-to-G mismatch reads, and did
not occur in the last 2 nt of the corresponding miRNA or miRNA*.
Candidate editing sites were then manually examined and discarded if an alternative explanation was more parsimonious. For
example, the only nonbrain editing candidate mapped to let-7c-1,
but was most likely due to a handful of let-7b reads containing
untemplated nucleotide additions that fortuitously matched the
let-7c-1 locus. Consistent with this explanation, the putatively
edited reads were unusually long and at unusually low abundance.
Candidate editing sites were also checked in the Perlegen SNP
database (Frazer et al. 2007) and dbSNP; no editing candidates
corresponded to known SNPs.
Untemplated nucleotide analysis
To examine untemplated nucleotide addition, non-genome-mapping reads were filtered for those that matched miRNA or miRNA*
sequences but also included a nongenomic poly(N) at the 39 end.
The untemplated nucleotide addition rate was calculated as the
ratio of reads with the untemplated nucleotide to the sum of the
reads with and without the untemplated nucleotide. After excluding miRNAs that map to multiple loci, and any miRNAs or
miRNA*s with a genomic T at the position immediately 39 of the
annotated sequence, there were 343 miRNA/miRNA* species
with untemplated U on the 59 arm and 318 on the 39 arm. Similarly, there were 287 59 arm species with untemplated A on the
59 arm and 324 on the 39 arm. The background tRNA untemplated
U addition rate was calculated similarly. A two-sided K–S test was
used to assess significant differences in distributions.
Accession numbers
All small RNA reads are available at the GEO database with
accession number GSE20384.
Acknowledgments
We thank N. Lau and A. Chess for embryonic and newborn total
RNA, R. Friedman for calculating branch length scores for the
analysis of conservation, A. Marson and N. Hannet for technical
advice, and V.N. Kim for TNdrosha and TNdicer plasmids. This
work was supported by a grant from the NIH (GM067031) to D.B.
References
Azuma-Mukai A, Oguri H, Mituyama T, Qian ZR, Asai K, Siomi
H, Siomi MC. 2008. Characterization of endogenous human
Argonautes and their miRNA partners in RNA silencing.
Proc Natl Acad Sci 105: 7964–7969.
Babiarz JE, Ruby JG, Wang YM, Bartel DP, Blelloch R. 2008.
Mouse ES cells express endogenous shRNAs, siRNAs, and
other Microprocessor-independent, Dicer-dependent small
RNAs. Genes & Dev 22: 2773–2785.
Baek D, Villén J, Shin C, Camargo FD, Gygi SP, Bartel DP. 2008.
The impact of microRNAs on protein output. Nature 455:
64–71.
Bartel DP. 2004. MicroRNAs: Genomics, biogenesis, mechanism, and function. Cell 116: 281–297.
Bartel DP. 2009. MicroRNAs: Target recognition and regulatory
functions. Cell 136: 215–233.
Baskerville S, Bartel DP. 2005. Microarray profiling of microRNAs reveals frequent coexpression with neighboring
miRNAs and host genes. RNA 11: 241–247.
Bender W. 2008. MicroRNAs in the Drosophila bithorax complex. Genes & Dev 22: 14–19.
Bentwich I, Avniel A, Karov Y, Aharonov R, Gilad S, Barad O,
Barzilai A, Einat P, Einav U, Meiri E, et al. 2005. Identification of hundreds of conserved and nonconserved human
microRNAs. Nat Genet 37: 766–770.
Berezikov E, Guryev V, van de Belt J, Wienholds E, Plasterk
RHA, Cuppen E. 2005. Phylogenetic shadowing and computational identification of human microRNA genes. Cell 120:
21–24.
Berezikov E, Thuemmler F, van Laake LW, Kondova I, Bontrop
R, Cuppen E, Plasterk RHA. 2006a. Diversity of microRNAs
in human and chimpanzee brain. Nat Genet 38: 1375–1377.
Berezikov E, van Tetering G, Verheul M, van de Belt J, van Laake
L, Vos J, Verloop R, van de Wetering M, Guryev V, Takada S,
et al. 2006b. Many novel mammalian microRNA candidates
identified by extensive cloning and RAKE analysis. Genome
Res 16: 1289–1298.
Berezikov E, Chung WJ, Willis J, Cuppen E, Lai EC. 2007.
Mammalian mirtron genes. Mol Cell 28: 328–336.
Blow MJ, Grocock RJ, van Dongen S, Enright AJ, Dicks E, Futreal
PA, Wooster R, Stratton MR. 2006. RNA editing of human
microRNAs. Genome Biol 7: R27. doi: 10.1186/gb-2006-7-4-r27.
Calabrese JM, Seila AC, Yeo GW, Sharp PA. 2007. RNA
sequence analysis defines Dicer’s role in mouse embryonic
stem cells. Proc Natl Acad Sci 104: 18097–18102.
Chen C-Z, Li L, Lodish HF, Bartel DP. 2004. MicroRNAs modulate hematopoietic lineage differentiation. Science 303: 83–86.
Christodoulou F, Raible F, Tomer R, Simakov O, Trachana K,
Klaus S, Snyman H, Hannon GJ, Bork P, Arendt D. 2010.
Ancient animal microRNAs and the evolution of tissue
identity. Nature 463: 1084–1088.
Cummins JM, He YP, Leary RJ, Pagliarini R, Diaz LA, Sjoblom
T, Barad O, Bentwich Z, Szafranska AE, Labourier E, et al.
2006. The colorectal microRNAome. Proc Natl Acad Sci
103: 3687–3692.
Frazer KA, Eskin E, Kang HM, Bogue MA, Hinds DA, Beilharz
EJ, Gupta RV, Montgomery J, Morenzoni MM, Nilsen GB,
et al. 2007. A sequence-based variation map of 8.27 million
SNPs in inbred mouse strains. Nature 448: 1050–1053.
Friedman RC, Farh KKH, Burge CB, Bartel DP. 2009. Most
mammalian mRNAs are conserved targets of microRNAs.
Genome Res 19: 92–105.
Griffiths-Jones S. 2004. The microRNA registry. Nucleic Acids
Res 32: D109–D111. doi: 10.1093/nar/gkh023.
Grimm D, Streetz KL, Jopling CL, Storm TA, Pandey K, Davis
CR, Marion P, Salazar F, Kay MA. 2006. Fatality in mice due
to oversaturation of cellular microRNA/short hairpin RNA
pathways. Nature 441: 537–541.
Grimson A, Srivastava M, Fahey B, Woodcroft BJ, Chiang HR,
King N, Degnan BM, Rokhsar DS, Bartel DP. 2008. Early
origins and evolution of microRNAs and Piwi-interacting
RNAs in animals. Nature 455: 1193–1197.
Hagan JP, Piskounova E, Gregory RI. 2009. Lin28 recruits the
TUTase Zcchc11 to inhibit let-7 maturation in mouse
embryonic stem cells. Nat Struct Mol Biol 16: 1021–1025.
GENES & DEVELOPMENT
251
1007
Chiang et al.
Han JJ, Lee Y, Yeom KH, Nam JW, Heo I, Rhee JK, Sohn SY, Cho
YJ, Zhang BT, Kim VN. 2006. Molecular basis for the
recognition of primary microRNAs by the Drosha–DGCR8
complex. Cell 125: 887–901.
Han J, Pedersen JS, Kwon SC, Belair CD, Kim Y-K, Yeom K-H,
Yang W-Y, Haussler D, Blelloch R, Kim VN. 2009. Posttranscriptional crossregulation between Drosha and DGCR8.
Cell 136: 75–84.
Heo I, Joo C, Cho J, Ha M, Han JJ, Kim VN. 2008. Lin28
mediates the terminal uridylation of let-7 precursor microRNA. Mol Cell 32: 276–284.
Heo I, Joo C, Kim Y-K, Ha M, Yoon M-J, Cho J, Yeom K-H, Han J,
Kim VN. 2009. TUT4 in concert with Lin28 suppresses
microRNA biogenesis through pre-microRNA uridylation.
Cell 138: 696–708.
Hofacker IL, Fontana W, Stadler PF, Bonhoeffer LS, Tacker M,
Schuster P. 1994. Fast folding and comparison of rna secondary structures. Monatsh Chem 125: 167–188.
Houbaviy HB, Murray MF, Sharp PA. 2003. Embryonic stem
cell-specific microRNAs. Dev Cell 5: 351–358.
Hu H, Yan Z, Xu Y, Hu H, Menzel C, Zhou Y, Chen W, Khaitovich
P. 2009. Sequence features associated with microRNA strand
selection in humans and flies. BMC Genomics 10: 413.
Kawahara Y, Zinshteyn B, Chendrimada TP, Shiekhattar R,
Nishikura K. 2007a. RNA editing of the microRNA-151
precursor blocks cleavage by the Dicer–TRBP complex.
EMBO Rep 8: 763–769.
Kawahara Y, Zinshteyn B, Sethupathy P, Iizasa H, Hatzigeorgiou
AG, Nishikura K. 2007b. Redirection of silencing targets by
adenosine-to-inosine editing of miRNAs. Science 315: 1137–
1140.
Kawahara Y, Megraw M, Kreider E, Iizasa H, Valente L,
Hatzigeorgiou AG, Nishikura K. 2008. Frequency and fate
of microRNA editing in human brain. Nucleic Acids Res 36:
5270–5280.
Kim Y-K, Kim VN. 2007. Processing of intronic microRNAs.
EMBO J 26: 775–783.
Kuchenbauer F, Morin RD, Argiropoulos B, Petriv OI, Griffith
M, Heuser M, Yung E, Piper J, Delaney A, Prabhu AL, et al.
2008. In-depth characterization of the microRNA transcriptome in a leukemia progression model. Genome Res 18:
1787–1797.
Kuhn RM, Karolchik D, Zweig AS, Wang T, Smith KE, Rosenbloom KR, Rhead B, Raney BJ, Pohl A, Pheasant M, et al.
2009. The UCSC Genome Browser Database: Update 2009.
Nucleic Acids Res 37: D755–D761. doi: 10.1093/nar/gkn875.
Kurihara Y, Watanabe Y. 2004. Arabidopsis micro-RNA biogenesis through Dicer-like 1 protein functions. Proc Natl
Acad Sci 101: 12753–12758.
Lagos-Quintana M, Rauhut R, Lendeckel W, Tuschl T. 2001.
Identification of novel genes coding for small expressed
RNAs. Science 294: 853–858.
Lagos-Quintana M, Rauhut R, Yalcin A, Meyer J, Lendeckel W,
Tuschl T. 2002. Identification of tissue-specific microRNAs
from mouse. Curr Biol 12: 735–739.
Lagos-Quintana M, Rauhut R, Meyer J, Borkhardt A, Tuschl T.
2003. New microRNAs from mouse and human. Rna 9: 175–
179.
Landgraf P, Rusu M, Sheridan R, Sewer A, Iovino N, Aravin A,
Pfeffer S, Rice A, Kamphorst AO, Landthaler M, et al. 2007.
A mammalian microRNA expression atlas based on small
RNA library sequencing. Cell 129: 1401–1414.
Langmead B, Trapnell C, Pop M, Salzberg S. 2009. Ultrafast and
memory-efficient alignment of short DNA sequences to the
human genome. Genome Biol 10: R25. doi: 10.1186/gb-200910-3-r25.
1008
Lau NC, Lim LP, Weinstein EG, Bartel DP. 2001. An abundant
class of tiny RNAs with probable regulatory roles in Caenorhabditis elegans. Science 294: 858–862.
Lee RC, Ambros V. 2001. An extensive class of small RNAs in
Caenorhabditis elegans. Science 294: 862–864.
Lee Y, Ahn C, Han JJ, Choi H, Kim J, Yim J, Lee J, Provost P,
Radmark O, Kim S, et al. 2003. The nuclear RNase III Drosha
initiates microRNA processing. Nature 425: 415–419.
Lim LP, Glasner ME, Yekta S, Burge CB, Bartel DP. 2003.
Vertebrate microRNA genes. Science 299: 1540.
Lu C, Tej SS, Luo S, Haudenschild CD, Meyers BC, Green PJ.
2005. Elucidation of the small RNA component of the
transcriptome. Science 309: 1567–1569.
Lund E, Guttinger S, Calado A, Dahlberg JE, Kutay U. 2004.
Nuclear export of microRNA precursors. Science 303: 95–98.
Marson A, Levine SS, Cole MF, Frampton GM, Brambrink T,
Johnstone S, Guenther MG, Johnston WK, Wernig M,
Newman J, et al. 2008. Connecting microRNA genes to
the core transcriptional regulatory circuitry of embryonic
stem cells. Cell 134: 521–533.
Mathews DH, Sabina J, Zuker M, Turner DH. 1999. Expanded
sequence dependence of thermodynamic parameters improves prediction of RNA secondary structure. J Mol Biol
288: 911–940.
Mineno J, Okamoto S, Ando T, Sato M, Chono H, Izu H,
Takayama M, Asada K, Mirochnitchenko O, Inouye M,
et al. 2006. The expression profile of microRNAs in mouse
embryos. Nucleic Acids Res 34: 1765–1771.
Okamura K, Hagen JW, Duan H, Tyler DM, Lai EC. 2007. The
mirtron pathway generates microRNA-class regulatory
RNAs in Drosophila. Cell 130: 89–100.
Pruitt KD, Tatusova T, Maglott DR. 2005. NCBI Reference
Sequence (RefSeq): A curated non-redundant sequence database of genomes, transcripts and proteins. Nucleic Acids Res
33: D501–D504. doi: 10.1093/nar/gki025.
Rajagopalan R, Vaucheret H, Trejo J, Bartel DP. 2006. A diverse
and evolutionarily fluid set of microRNAs in Arabidopsis
thaliana. Genes & Dev 20: 3407–3425.
Rao PK, Toyama Y, Chiang HR, Gupta S, Bauer M, Medvid R,
Reinhardt F, Liao R, Krieger M, Jaenisch R, et al. 2009.
Loss of cardiac microRNA-mediated regulation leads to
dilated xardiomyopathy and heart failure. Circ Res 105:
585–594.
Ro S, Park C, Young D, Sanders KM, Yan W. 2007. Tissuedependent paired expression of miRNAs. Nucleic Acids Res
35: 5944–5953.
Rodriguez A, Vigorito E, Clare S, Warren MV, Couttet P, Soond
DR, van Dongen S, Grocock RJ, Das PP, Miska EA, et al.
2007. Requirement of bic/microRNA-155 for normal immune function. Science 316: 608–611.
Ruby JG, Jan C, Player C, Axtell MJ, Lee W, Nusbaum C, Ge H,
Bartel DP. 2006. Large-scale sequencing reveals 21U-RNAs
and additional microRNAs and endogenous siRNAs in
C. elegans. Cell 127: 1193–1207.
Ruby JG, Jan CH, Bartel DP. 2007a. Intronic microRNA precursors that bypass Drosha processing. Nature 448: 83–86.
Ruby JG, Stark A, Johnston WK, Kellis M, Bartel DP, Lai EC. 2007b.
Evolution, biogenesis, expression, and target predictions of
a substantially expanded set of Drosophila microRNAs. Genome Res 17: 1850–1864.
Seo TS, Bai XP, Ruparel H, Li ZM, Turro NJ, Ju JY. 2004.
Photocleavable fluorescent nucleotides for DNA sequencing
on a chip constructed by site-specific coupling chemistry.
Proc Natl Acad Sci 101: 5488–5493.
Stark A, Kheradpour P, Parts L, Brennecke J, Hodges E, Hannon
GJ, Kellis M. 2007. Systematic discovery and characterization
GENES & DEVELOPMENT
252
Mammalian microRNAs
of fly microRNAs using 12 Drosophila genomes. Genome Res
17: 1865–1879.
Stark A, Bushati N, Jan CH, Kheradpour P, Hodges E, Brennecke
J, Bartel DP, Cohen SM, Kellis M. 2008. A single Hox locus in
Drosophila produces functional microRNAs from opposite
DNA strands. Genes & Dev 22: 8–13.
Tam OH, Aravin AA, Stein P, Girard A, Murchison EP, Cheloufi
S, Hodges E, Anger M, Sachidanandam R, Schultz RM, et al.
2008. Pseudogene-derived small interfering RNAs regulate
gene expression in mouse oocytes. Nature 453: 534–538.
Tyler DM, Okamura K, Chung W-J, Hagen JW, Berezikov E,
Hannon GJ, Lai EC. 2008. Functionally distinct regulatory
RNAs generated by bidirectional transcription and processing of microRNA loci. Genes & Dev 22: 26–36.
Voorhoeve PM, le Sage C, Schrier M, Gillis AJM, Stoop H, Nagel
R, Liu Y-P, van Duijse J, Drost J, Griekspoor A, et al. 2006. A
genetic screen implicates miRNA-372 and miRNA-373 as
oncogenes in testicular germ cell tumors. Cell 124: 1169–
1181.
Watanabe T, Totoki Y, Toyoda A, Kaneda M, KuramochiMiyagawa S, Obata Y, Chiba H, Kohara Y, Kono T, Nakano
T, et al. 2008. Endogenous siRNAs from naturally formed
dsRNAs regulate transcripts in mouse oocytes. Nature 453:
539–543.
Wienholds E, Kloosterman WP, Miska E, Alvarez-Saavedra E,
Berezikov E, de Bruijn E, Horvitz HR, Kauppinen S, Plasterk
RHA. 2005. MicroRNA expression in zebrafish embryonic
development. Science 309: 310–311.
Wu H, Ye C, Ramirez D, Manjunath N. 2009. Alternative
processing of primary microRNA transcripts by Drosha
generates 59 end variation of mature microRNA. PLoS One
4: e7566. doi: 10.1371/journal.pone.0007566.
Xie XH, Lu J, Kulbokas EJ, Golub TR, Mootha V, Lindblad-Toh
K, Lander ES, Kellis M. 2005. Systematic discovery of
regulatory motifs in human promoters and 39 UTRs by
comparison of several mammals. Nature 434: 338–345.
Yang W, Chendrimada TP, Wang Q, Higuchi M, Seeburg PH,
Shiekhattar R, Nishikura K. 2006. Modulation of microRNA
processing and expression through RNA editing by ADAR
deaminases. Nat Struct Mol Biol 13: 13–21.
Yi R, Qin Y, Macara IG, Cullen BR. 2003. Exportin-5 mediates
the nuclear export of pre-microRNAs and short hairpin
RNAs. Genes & Dev 17: 3011–3016.
Zeng Y, Yi R, Cullen BR. 2005. Recognition and cleavage of
primary microRNA precursors by the nuclear processing
enzyme Drosha. EMBO J 24: 138–148.
Zuker M. 2003. Mfold web server for nucleic acid folding and
hybridization prediction. Nucleic Acids Res 31: 3406–3415.
GENES & DEVELOPMENT
253
1009
Chiang135681_SuppFig1
Undetected
annotated miRNAs
(157)
Not sequenced (52)
Not enough reads (72)
Failed other filters (33)
DGCR8 & Dicer-dependent (290, 226)
DGCR8-dependent (2, 2)
Confirmed miRNAs (387)
Annotated miRNAs (407)
Dicer-dependent (7, 3)
Not strongly dependent (3, 3)
miRNA candidates not
confidently confirmed (20)
Cannot determine (85, 49)
DGCR8 & Dicer-dependent (37, 0)
Total candidates
(736)
Novel miRNAs (108)
Dicer-dependent (1, 0)
Not strongly dependent (1, 0)
Cannot determine (69, 15)
New candidates (329)
DGCR8 & Dicer-dependent (45, 0)
miRNA candidates (221)
Dicer-dependent (5, 0)
Not strongly dependent (42, 8)
Cannot determine (129, 9)
Supplementary Figure 1. Mouse miRNA candidates identified by Illumina sequencing. MicroRNAs that are
annotated in miRBase v.14.0 are boxed in green. The miRNA hairpin loci were further categorized by DGCR8- and
Dicer-dependency using sequencing data from wild-type an mutant ES cells (Babiarz et al. 2008). The number in
parenthesis is the total number of loci in the category. If followed by another number, the second number is the number of conserved loci. A candidate was considered DGCR8- and Dicer-dependent using criteria of a previous study
(Babiarz et al. 2008), except that predicted hairpin loci replaced the 100-nt windows, with the read cutoffs scaled to
the hairpin lengths.
254
Chiang135681_FigureS3
hsa-mir-124-1
hsa-mir-125a
hsa-mir-128-1
hsa-mir-142
hsa-mir-150
hsa-mir-192
hsa-mir-193b
hsa-mir-205
hsa-mir-214
hsa-mir-455
hsa-mir-483
hsa-mir-499
hsa-mir-888
hsa-mir-9-1
hsa-mir-220a
cand141
cand142
cand181
cand316
mmu-mir-122
mmu-mir-133a-1
mmu-mir-137
mmu-mir-138-1
mmu-mir-139
mmu-mir-153
mmu-mir-208a
mmu-mir-216a
mmu-mir-217
mmu-mir-223
mmu-mir-224
mmu-mir-375
mmu-mir-105
mmu-mir-207
mmu-mir-220
mmu-mir-327
mmu-mir-343
mmu-mir-453
mmu-mir-568
mmu-mir-654
mmu-mir-678
mmu-mir-680-3
mmu-mir-687
mmu-mir-697
mmu-mir-698
mmu-mir-717
mmu-mir-719
mmu-mir-761
mmu-mir-882
mmu-mir-682
mmu-mir-690
mmu-mir-707
mmu-mir-763
mmu-mirc-niob-MM_28
mmu-mirc-niob-MM_57
mmu-mirc-niob-MM_76
mmu-mirc-niob-MM_155
mmu-mirc-niob-MM_185
mmu-mirc-niob-MM_227
mmu-mirc-niob-MM_290
mmu-mirc-niob-MM_298
MIR90
MIR103
MIR146
MIR165
MIR170
MIR174
MIR181
MIR192
MIR213
MIR223
MIR237
MIR252
≤1
*
*
*
*
*
Human
miRNA
controls
*
*
*
*
*
*
*
*
Lim 2003
Berezikov
2005
*
*
*
*
*
*
*
*
*
Mouse
miRNA
controls
*
*
Not
sequenced
Not enough
reads
Berezikov
2006b
Xie 2005
10
100
1000
10000 100000
Reads
No hairpin plasmid
Hairpin plasmid
Supplemental Figure S3. Ectopic-expression assay evaluating unconfirmed annotated miRNAs and predicted
miRNAs. Either GFP (red) or miRNA hairpins (blue) were expressed in HEK293T. Asterisk indicates positive results.
Results of this experiment were compiled with those of Supplemental Figure S4 to produce Figure 2C and D.
255
Chiang135681_FigureS4
*
hsa-mir-193b
*
mmu-mir-122
Human miRNA
control
*
mmu-mir-133a-1
*
mmu-mir-137
*
mmu-mir-138-1
*
*
mmu-mir-139
mmu-mir-153
Mouse
miRNA
controls
*
mmu-mir-208a
*
mmu-mir-216a
*
mmu-mir-217
*
mmu-mir-223
*
mmu-mir-224
*
mmu-mir-375
mmu-mir-599
mmu-mir-669i
*
mmu-mir-684-1
mmu-mir-684-2
mmu-mir-685
mmu-mir-688
mmu-mir-690
*
mmu-mir-693
mmu-mir-704
mmu-mir-705
mmu-mir-707
mmu-mir-763
mmu-mir-1187
Not enough
reads
mmu-mir-1192
mmu-mir-1894
mmu-mir-1903
mmu-mir-1904
mmu-mir-1907
mmu-mir-1927
*
*
mmu-mir-1929
mmu-mir-1936
mmu-mir-1937c
mmu-mir-1940
mmu-mir-1959
mmu-mir-1960
mmu-mir-1966
mmu-mir-1970
*
mmu-mir-184
mmu-mir-297a-6
mmu-mir-466f-4
*
mmu-mir-489
mmu-mir-1191
mmu-mir-1953
No miRNA*
*
*
mmu-mir-1969
mmu-mir-449c
*
mmu-mir-677
mmu-mir-1944
≤1
Incorrect
miRNA*
10
100
1000
10000 100000
Reads
No hairpin plasmid
Hairpin plasmid
Supplemental Figure S4. Ectopic-expression assay evaluating unconfirmed annotated miRNAs. Either GFP (red) or
miRNA hairpins (blue) were expressed in HEK293T. Asterisk indicates positive results. Results of this experiment were
compiled with those of Supplemental Figure S3 to produce Figure 2C and D.
256
Chiang135681_FigureS5
*
hsa-mir-124-1
hsa-mir-125a
hsa-mir-128-1
hsa-mir-142
hsa-mir-150
hsa-mir-192
hsa-mir-193b
hsa-mir-205
hsa-mir-214
hsa-mir-455
hsa-mir-483
hsa-mir-499
hsa-mir-888
hsa-mir-9-1
hsa-mir-220a
cand141
cand142
cand181
cand316
mmu-mir-122
mmu-mir-133a-1
mmu-mir-137
mmu-mir-138-1
mmu-mir-139
mmu-mir-153
mmu-mir-208a
mmu-mir-216a
mmu-mir-217
mmu-mir-223
mmu-mir-224
mmu-mir-375
mmu-mir-1941
mmu-mir-1964
mmu-mir-1968
mmu-mir-1912
mmu-mir-3061
mmu-mir-3072
mmu-mir-3073
mmu-mir-3075
mmu-mir-3081
mmu-mir-3089
mmu-mir-3090
mmu-mir-3093
mmu-mir-3095
mmu-mir-3108
mmu-mir-3109
mmu-mir-3110
mmu-mir-344f
mmu-mir-3104
noStar-014
noStar-033
noStar-043
noStar-073
noStar-080
noStar-087
noStar-117
noStar-135
noStar-150
noStar-154
noStar-166
wrongStar-016
noStar-149
≤1
*
*
*
Human
miRNA
controls
*
*
*
*
*
*
*
*
Lim 2003
Berezikov
2005
*
*
*
*
*
*
*
*
*
*
*
Mouse
miRNA
controls
*
*
*
*
*
*
*
*
*
Novel
miRNAs
*
*
*
*
*
Novel
shRNAs
*
DGCR8- &
DCR-dependent
candidates
Other candidate
10
100
1000
10000 100000
Reads
No hairpin plasmid
Hairpin plasmid
Supplemental Figure S5. Ectopic-expression assay evaluating predicted miRNAs, novel miRNAs, and miRNA candidates. Either GFP (red) or miRNA hairpins (blue) were expressed in HEK293T. Asterisk indicates positive results. Results of
this experiment were compiled with those of Supplemental Figures S6 and S7 to produce Figure 3B.
257
Chiang135681_FigureS6
*
hsa-mir-124-1
hsa-mir-125a
hsa-mir-128-1
hsa-mir-142
hsa-mir-150
hsa-mir-192
hsa-mir-193b
hsa-mir-205
hsa-mir-214
hsa-mir-455
hsa-mir-483
hsa-mir-499
hsa-mir-888
hsa-mir-9-1
hsa-mir-220a
cand141
cand142
cand181
cand316
mmu-mir-122
mmu-mir-133a-1
mmu-mir-137
mmu-mir-138-1
mmu-mir-139
mmu-mir-153
mmu-mir-208a
mmu-mir-216a
mmu-mir-217
mmu-mir-223
mmu-mir-224
mmu-mir-375
mmu-mir-1188
mmu-mir-1197
mmu-mir-1933
mmu-mir-1947
mmu-mir-1224
mmu-mir-1839
mmu-mir-509
mmu-mir-3059
mmu-mir-3063
mmu-mir-3065
mmu-mir-3067
mmu-mir-3079
mmu-mir-3086
mmu-mir-3091
mmu-mir-3100
mmu-mir-3112
mmu-mir-344e
mmu-mir-3111
noStar-046
noStar-148
wrongStar-017
noStar-020
noStar-034
noStar-054
noStar-056
noStar-068
noStar-093
noStar-122
noStar-126
noStar-160
wrongStar-002
wrongStar-007
wrongStar-009
≤1
*
*
*
*
Human
miRNA
controls
*
*
*
*
*
*
*
*
Lim 2003
Berezikov
2005
*
*
*
*
*
*
*
*
*
*
*
Noncanonical
controls
Early miRBase
Novel miRNA
*
*
Mouse
miRNA
controls
*
*
*
*
*
*
*
Rare novel
miRNAs
*
*
Novel
shRNAs
Conserved
candidates
Other
candidates
*
10
100
1000
10000 100000
Reads
No hairpin plasmid
Hairpin plasmid
Supplemental Figure S6. Ectopic-expression assay evaluating novel miRNAs, miRNA candidates, predicted miRNAs,
and an unconfirmed annotated miRNA (mmu-mir-509). Either GFP (red) or miRNA hairpins (blue) were expressed in
HEK293T. Asterisk indicates positive results. Results of this experiment were compiled with those of Supplemental Figures
S5 and S7 to produce Figure 3B.
258
Chiang135681_FigureS7
Human miRNA
control
Lim 2003
Mouse
miRNA
controls
Noncanonical
controls
Early miRBase
Novel
miRNAs
Novel rare
miRNAs
Novel
shRNAs
Candidates
hsa-mir-193b
hsa-mir-220a
mmu-mir-122
mmu-mir-133a-2
mmu-mir-137
mmu-mir-138-1
mmu-mir-139
mmu-mir-153
mmu-mir-208a
mmu-mir-216a
mmu-mir-217
mmu-mir-223
mmu-mir-224
mmu-mir-375
mmu-mir-1933
mmu-mir-1941
mmu-mir-1947
mmu-mir-1964
mmu-mir-1968
mmu-mir-1224
mmu-mir-1839
mmu-mir-509
mmu-mir-1912
mmu-mir-3059
mmu-mir-3061
mmu-mir-3072
mmu-mir-3073
mmu-mir-3075
mmu-mir-3081
mmu-mir-3090
mmu-mir-3095
mmu-mir-3108
mmu-mir-3109
mmu-mir-3110
mmu-mir-3063
mmu-mir-3065
mmu-mir-3079
mmu-mir-3086
mmu-mir-3091
mmu-mir-344e
mmu-mir-344f
noStar-020
noStar-056
noStar-122
noStar-148
wrongStar-002
wrongStar-009
<=1
10
No hairpin plasmid + no TNdrosha/TNdicer plasmid
Hairpin plasmid + no TNdrosha/TNdicer plasmid
100
Reads
1000
No hairpin plasmid + TNdrosha plasmid
Hairpin plasmid + TNdrosha plasmid
10000
100000
No hairpin plasmid + TNdicer plasmid
Hairpin plasmid + TNdicer plasmid
Supplemental Figure S7. Drosha/Dicer-dependent biogenesis of novel miRNAs. The selected hairpins were transfected
into HEK293T with a control vector (blue), TNdrosha (red), or TNdicer (green). Similar transfections using the control vector
instead of the hairpins are shown in light blue, orange, and light green, respectively. Results of this experiment were compiled
with those of Supplemental Figures S5 and S6 to produce Figure 3B.
259
Chiang135681_FigureS8
1
0.8
0.6
Correlation coefficient
0.4
0.2
0
-0.2
-0.4
-0.6
-0.8
-1
10
102
103
104
105
106
107
108
109
Genomic distance
Not clustered
Clustered
Supplemental Figure S8. Correlation of expression and genomic distance. The correlation of expression with clustering
was calculated as previously (Baskerville and Bartel 2005), except that miRNAs that mapped to the same pre-mRNA
transcript were considered clustered regardless of genomic distance. The clustered miRNAs (red) were more correlated than
non-clustered miRNAs (blue). Some miRNA pairs more than 50,000 nt apart were categorized as clustered with each other
due to joint proximity to intervening miRNAs, and their correlated expressions supported this clustering method. Other
miRNAs that are within 50,000 nt of each other were not considered clustered because one mapped within a pre-mRNA,
whereas the other one did not; each of these three pairs of miRNAs were not correlated in expression. Correlated expression
observed for many miRNAs located ~130,000 nt apart was due to likely co-expression of two megaclusters on chr12.
260
Chiang135681_FigureS9
A
B
12,000,000
350
300
10,000,000
250
miRNAs
Reads
8,000,000
6,000,000
200
150
4,000,000
100
2,000,000
50
0
16
17
18
19
20
21
22
23
24
25
26
0
27
miRNA length
Conserved
18
19
20
21
22
23
24
25
26
miRNA length
Nonconserved
Conserved
Nonconserved
Supplemental Figure S9. The distribution of lengths of conserved (red) and nonconserved (blue) mature miRNAs. (A)
Size distribution plotted in terms of number of normalized reads. (B) Size distribution plotted in terms of the dominant read
length for each miRNA.
261
Chiang135681_FigureS10
A
UCAGUUG
UCAGUUC
0.50
8mer
7mer-m8
7mer-A1
6mer
No Site
0.25
-0.5
-0.25 0.0
0.25
0.5
Fold Change (log2)
0.75
0.50
8mer
7mer-m8
7mer-A1
6mer
No Site
0.25
0.00
-0.75
0.75
Cummulative Fraction
0.75
0.00
-0.75
UCAGUUA
1.00
1.00
Cummulative Fraction
Cummulative Fraction
1.00
-0.5
0.5
-0.25 0.0
0.25
Fold Change (log2)
0.75
0.50
0.25
0.00
-0.75
0.75
8mer
7mer-m8
7mer-A1
6mer
No Site
-0.5
-0.25 0.0
0.25
0.5
Fold Change (log2)
0.75
B
AAUGCUU
AAUGCUG
0.50
8mer
7mer-m8
7mer-A1
6mer
No Site
0.25
-0.5
-0.25 0.0
0.25
0.5
Fold Change (log2)
0.75
1.00
Cummulative Fraction
Cummulative Fraction
Cummulative Fraction
0.75
0.00
-0.75
AAUGCUC
1.00
1.00
0.75
0.50
8mer
7mer-m8
7mer-A1
6mer
No Site
0.25
0.00
-0.75
-0.5
0.5
-0.25 0.0
0.25
Fold Change (log2)
0.75
0.75
0.50
8mer
7mer-m8
7mer-A1
6mer
No Site
0.25
0.00
-0.75
-0.5
0.5
-0.25 0.0
0.25
Fold Change (log2)
0.75
Supplemental Figure S10. Controls to ensure that observed mRNA derepression attributed to the minor isoform was not
due to overlap of its sites with offset 6mer sites of the major isoform. (A) Lack of statistically significant derepression by the
three control motifs that differed from the miR-223 minor site by a single nt at position 8. (B) Same as in A except for the
miR-155 minor site. The mRNAs with 8mer and 7mer-A1 sites for the minor isoform were excluded from the analysis because
these sites overlapped with 7mer-m8 sites for the major isoform.
262
Chiang135681_FigureS11
Distance from
3’ end of read
Mismatch
type
T>C
C>T (1.2%)
C>G
C>A
One mismatch
(12%)
Distance from
3’ end of read
T>C
T>G
T>A
1-2 nt
(2.0%)
Perfect match
(86%)
Brain miRNA-matching sequences
Mismatch
type
1-2 nt
(13%)
Perfect match
(80%)
G>T
One mismatch
(16%)
G>T
>2 nt
(9.6%)
G>C
G>A
A>T
G>C
G>A
>2 nt
(4.0%)
A>T
A>C
A>G (0.32%)
Three mismatches
(0.92%)
Two mismatches
(2.0%)
C
Significant mismatch events
Thresholds
Fraction edited: >5%
Edited reads: >10
250
A>G
C>T
All others
100
40
50
30
20
0
10
0
Single
mismatches
Read filters
miRNA
mature or *
Brain – mir-381
chr12(+): 110965025 - 110965112
GTTTGGTACTTAAAGCGAGGTTGCCCTTTGTATATTCGGTTTATTGACATGGAATATACAAGGGCAAGCTCTCTGTGAGTATCAAACC
((((((((((((.((.(((.((((((((.((((((((.(((....)))...)))))))))))))))).))).)).)))))))))))).
.............AGCGAGGTTGCCCTTTGTAA.......................................................
.............GGCGAGGTTGCCCTTTGTATATT....................................................
.....................................................CTATACAAGGGCAAGCTCTCTGT............
.....................................................ATATACAAGGGCAAGCTCTCTGA............
......................................................TATACAAGGGCAAGCTCTCTGC............
.......................................................ATGCAAGGGCAAGCTCTCTGT............
......................................................TATACAAGGGCACGCTCTCTGT............
......................................................TATACAAGGGCAAGCTCTCTGTT...........
......................................................TATACAAGGGCAAGCTCTCTGA............
......................................................TATACAGGGGCAAGCTCTCTGT............
......................................................TATACAAGGGCAAGCTCTCTGTA...........
......................................................TATGCAAGGGCAAGCTCTCTGT............
.............AGCGAGGTTGCCCTTTGTA........................................................
.............AGCGAGGTTGCCCTTTGTAT.......................................................
.............AGCGAGGTTGCCCTTTGTATA......................................................
.............AGCGAGGTTGCCCTTTGTATAT.....................................................
.............AGCGAGGTTGCCCTTTGTATATT....................................................
.............AGCGAGGTTGCCCTTTGTATATTC...................................................
.....................................................ATATACAAGGGCAAGCTCTC...............
.....................................................ATATACAAGGGCAAGCTCTCT..............
.....................................................ATATACAAGGGCAAGCTCTCTG.............
.....................................................ATATACAAGGGCAAGCTCTCTGT............
......................................................TATACAAGGGCAAGCTCTC...............
......................................................TATACAAGGGCAAGCTCTCT..............
......................................................TATACAAGGGCAAGCTCTCTG.............
......................................................TATACAAGGGCAAGCTCTCTGT............
.......................................................ATACAAGGGCAAGCTCTCTG.............
.......................................................ATACAAGGGCAAGCTCTCTGT............
150
Unique
alignments
A>G (0.61%)
217 Sequences mapped (sequences with at least 5 reads shown)
200
Genome
matching
A>C
Three mismatches
(0.65%)
Two mismatches
(2.9%)
B
300
T>G
T>A
C>T (0.26%)
C>G
C>A
>2 nt from
3’ end
Event filter
1mm
1mm
1mm
1mm
1mm
1mm
1mm
1mm
1mm
1mm
1mm
1mm
Perf.
Perf.
Perf.
Perf.
Perf.
Perf.
Perf.
Perf.
Perf.
Perf.
Perf.
Perf.
Perf.
Perf.
Perf.
Perf.
Reads
Spiked-in sequence controls
Mismatches
A
5
7
31
36
37
39
41
100
200
236
260
2054
9
43
19
57
96
32
5
90
57
212
33
407
547
11639
15
239
150
120
90
60
30
miR-337
3p arm (star strand)
24
32
40
p = 3.27e-13
Edited reads
Edited reads
16
3000
2500
2000
1500
1000
500
miR-411
5p arm (star strand)
GAGATAGTAGACCGTATAGCGTACG
0
CCATTCAGCTCCTATATGATGCCTTT
8
3500
Perfect match reads
180
p < 2.2e-16
300
600
900
1200
900
750
600
450
300
miR-376a
150
5p arm (star strand)
AAAAGGTAGATTCTCCTTCTATGAGT
Edited reads
210
Perfect match reads
D
Perfect match reads
A>G Rate:
0.125
70
p < 2.2e-16
140
210
280
350
1500
Supplemental Figure S11. RNA editing. (A) An overview of mismatches from the sequences indicated. In the two spiked-in
synthetic RNAs of known sequence, mismatches were distributed throughout the length of the sequence, with no preference for
A-to-G mismatches. In miRNA-mapping small RNA sequences from brain, mismatches were concentrated in the last 2 nt of the
read, probably due to cellular terminal-transferase activity. (B) Loss of most mismatch events after applying filters expected to
distinguish editing events from background. Mismatch events were considered significant if a position had at least 10 reads
corresponding to a particular mismatch, and these reads accounted for at least 5% of reads covering that position. As successive
filters were applied to the genome-mapping reads, the number of significant A-to-G mismatch events remained relatively
unaffected, whereas nearly all other mismatch events were eliminated. In particular, C-to-T mismatches were mostly eliminated,
indicating that C-to-U RNA editing does not occur to any significant degree in miRNAs. A-to-G mismatch events that passed all
filters were considered editing candidates and manually examined to see if other plausible models could explain the mismatches.
(C) A display of most abundant perfectly-matching and single-mismatch reads from the mmu-mir-381 locus illustrates that
inferred A-to-I editing accounts for essentially all mismatches at the edited position, and the great majority of all mismatched
reads mapping to the miRNA or miRNA*. An analogous pattern was found for all 16 miRNAs that passed filters and manual
validation. (D) Editing of a miRNA or miRNA* was associated with significantly altered 5' end specificity. In the cases of
mmu-mir-337 and mmu-mir-411, edited reads had more homogeneous 5' ends than unedited reads.
263