BIOLOGY 475: MOLECULAR BIOLOGY
SPRING 2007
RNA Interference
(from Section 10.4.2)
Eukaryotes also possess other RNA degradation mechanisms that have evolved largely to protect the cell from
attack by foreign RNAs such as the genomes of viruses. An example is the pathway called RNA interference, a
name that will be familiar because RNA interference has been adopted by genome researchers as a means of
inactivating selected genes in order to study their function (Section 7.2.2). The target DNA for RNA interference
must be double stranded, which excludes cellular mRNAs but encompasses many viral genomes. The doublestranded RNA is cleaved by a ribonuclease called Dicer into short interfering RNAs (siRNAs) of 21 25 nucleotides
in length ( Ambros, 2001). This inactivates the virus genome, but what if the virus genes have already been
transcribed? If this has occurred then the harmful effects of the virus will already have been initiated and RNA
interference would appear to have failed in its attempt to protect the cell from damage. One of the more
remarkable discoveries of recent years has revealed a second stage of the interference process that is directed
specifically at the viral mRNAs. The siRNAs produced by cleavage of the viral genome are separated into individual
strands, one strand of each siRNA subsequently base-pairing to any viral mRNAs that are present in the cell. The
double-stranded regions that are formed are target sites for the RDE-1 nuclease, which destroys the mRNAs (see
Figure 7.16).
Figure 7.16. RNA interference. The double-stranded
RNA molecule is broken down by the Dicer
ribonuclease into 'short interfering RNAs' (siRNAs)
of 21 25 bp in length. One strand of each siRNA base
pairs to the target mRNA, which is then degraded by
the RDE-1 nuclease. For more details on RNA
interference, see Section 10.4.2. (above paragraph)
Simplified animation from Promega:
http://www.promega.com/paguide/animation/selector.htm?coreName=rnai01
complex appealing Animation – view online – from Nature Reviews
http://www.nature.com/focus/rnai/animations/index.html
Slide show from Howard Hughes Medical Institute:
http://www.hhmi.org/biointeractive/rna/rnai/index.html
Complex diagrammatic animation from Imgenex (beware it froze IE when I tried to save the animation
http://www.imgenex.com/rnai_anim.php
116104104
1
2/12/2016
BIOLOGY 475: MOLECULAR BIOLOGY
SPRING 2007
Interference in the Secondary
Guy Riddihough Sci. STKE, 16 January 2007 Vol. 2007, Issue 369, p.
tw25 [DOI: 10.1126/stke.3692007tw25] Science, AAAS, Washington, DC 20005, USA
The effector molecules in RNA interference (RNAi) are small interfering RNAs (siRNAs). The initial population of
"primary" siRNAs, ~22 nucleotides in length with 5'-monophosphates groups, is generated by the Dicer nuclease.
Amplification and "spreading" of the initial trigger population are thought to contribute to strength of the RNAi
response in a number of systems and involve an RNA-dependent RNA polymerase (RDRP) (see the Perspective by
Baulcombe). To investigate the nature of this secondary response, Pak and Fire and Sijen et al. analyzed the course
of an experimentally induced RNAi reaction in the nematode worm Caenorhabditis elegans and also examined
endogenous small RNAs. They found distinct populations of "secondary" siRNAs that are antisense to the
messenger RNA target, that have a di- or triphosphate moiety at their 5' ends, and that may map both upstream
and downstream of the original dsRNA trigger. Primary siRNAs do not appear to act as primers for RdRP but
rather guide RdRP to targeted messages for the de novo synthesis of secondary siRNAs that further boost the
RNAi response.
J. Pak, A. Fire, Distinct populations of primary and secondary effectors during RNAi in C. elegans. Science 315,
241-244 (2007). [Abstract] [Full Text]
T. Sijen, F. A. Steiner, K. L. Thijssen, R. H. A. Plasterk, Secondary siRNAs result from unprimed RNA synthesis and
form a distinct class. Science 315, 244-247 (2007). [Abstract] [Full Text]
D. C. Baulcombe, Amplified silencing. Science 315, 199-200 (2007). [Summary] [Full Text]
Citation: G. Riddihough, Interference in the Secondary. Sci. STKE 2007, tw25 (2007).
Amplified Silencing,
David C. Baulcombe,* Science 12 January 2007: Vol. 315. no. 5809, pp. 199 - 200
DOI: 10.1126/science.1138030
Ten years ago, we knew nothing about how double-stranded RNA blocks gene expression through the silencing of
targeted RNA. We now have a good understanding of this process, and current interest is turning to variations on
the basic mechanism. Recent studies involving plants and the nematode Caenorhabditis elegans continue this trend,
including those reported in this issue by Pak and Fire on page 241 (1) and Sijen et al. on page 244 (2). Two other
papers by Axtell et al. (3) and Ruby et al. (4) are also relevant. These studies deal with the amplification of
silencing-related RNA and explain how strong, persistent silencing can be initiated with small amounts of "initiator"
double-stranded RNA. The amplification process has implications for application of RNA interference to control
gene expression in biotechnology and for understanding the effects of silencing RNAs on cell function and organism
development.
Specifically, these new studies investigate how the target of silencing can spread (or transit) within a single strand
of RNA. The initiator of transitivity is a double-stranded RNA that is first processed by Dicer, a ribonuclease IIIlike enzyme, into short interfering RNA (siRNA) or a related type of RNA referred to as microRNA (miRNA).
These 21-to 25-nucleotide single-stranded RNAs are the primary silencing RNAs in the transitive process. A
primary silencing RNA binds to a ribonuclease H-like protein of the Argonaute class. The resulting Argonaute
ribonucleoprotein can target long RNA molecules by Watson-Crick base pairing. The targeted RNA then becomes a
source of secondary siRNAs. Transitivity occurs when the secondary siRNAs correspond to regions adjacent to the
target sites of the primary silencing RNA.
RNA-directed RNA polymerases (RdRPs) produce secondary siRNA, and the new results indicate that they catalyze
two different mechanisms of silencing amplification. One mechanism is characterized by Axtell et al. (3), who
investigated endogenous secondary siRNAs in plants. They show that efficient secondary siRNA production occurs
if a single-stranded RNA has two target sites for the Argonaute ribonucleoprotein. Optimal secondary siRNA
116104104
2
2/12/2016
BIOLOGY 475: MOLECULAR BIOLOGY
SPRING 2007
production occurs when the targeted RNA is cleaved by Argonaute. Cleaved RNA then recruits RdRP, which
generates double-stranded RNA. Dicer then produces transitive secondary siRNAs (see the figure).
Another biogenesis mechanism of secondary siRNAs has, so far, only been described in C. elegans. The discovery of
this distinct mechanism by Sijen et al., Pak and Fire, and Ruby et al. follows from the observation that a type of
siRNA is underrepresented in sequence databases. This scarceness is because these siRNAs have a 5′-triphosphate
and are thus excluded by the standard methods for cloning and sequence analysis. These methods are normally
specific for RNA with a 5′-monophosphate, the hallmark of Dicer cleavage.
Secondary siRNA production in plants and animals. Secondary siRNAs are produced by RdRP-mediated
transcription of RNA that has been targeted by a primary siRNA or miRNA. In C. elegans (left), an Argonaute
protein associated with a primary siRNA targets a long single-stranded RNA and recruits an RdRP that synthesizes
22-23 nucleotide secondary siRNAs directly. In plants (right), the recruitment of RdRP is optimal when the long
single-stranded RNA has two targets for primary siRNA or miRNA (only one is shown). The targeted RNA is then
converted to long double-stranded RNA by the RdRP and secondary siRNAs are generated after cleavage by Dicer.
In addition to their 5′-triphosphorylation, these siRNAs are distinct from the primary silencing RNAs in that they
have a strand bias. They predominantly are antisense to the target of the primary silencing RNA. The secondary
siRNAs also have the surprising characteristic that they are phased relative to each other (2, 4): The first siRNA
covers 22 nucleotides starting close to the target site of the primary siRNA, the second siRNA is then the
adjacent 22 nucleotides, and so on (see the figure). One explaination for these features might be that the 5′triphosphorylated secondary siRNAs are generated when RdRPs are recruited to a target of the primary silencing
RNA. Short antisense RNAs are then synthesized de novo, and the presence of the 5′-triphosphate in the first
incorporated nucleotide is diagnostic of secondary siRNAs made by this mechanism. Sijen et al. rule out primary
116104104
3
2/12/2016
BIOLOGY 475: MOLECULAR BIOLOGY
SPRING 2007
siRNA as a primer in this mechanism because mismatches in its sequence relative to that of a target RNA are
absent in the secondary siRNAs. To explain the rather precise size (22 or 23 nucleotides) of the secondary sRNAs,
this model requires that the RdRP automatically terminates RNA synthesis at a defined site or that the
transcription products be cleaved at their 3′?end by an unidentified endonuclease.
What is the natural role of these transitive secondary siRNAs? In plants, they target messenger RNAs (mRNAs)
(3), and it is likely that they do the same in C. elegans because endogenous siRNAs with 5′-triphosphate correspond
to the antisense of mRNA coding sequences (1). Moreover, Yigit et al. (5) describe how secondary siRNAs are bound
to a specific class of Argonaute proteins and that they direct RNA cleavage. It is likely, therefore, that secondary
siRNAs regulate gene expression in situations where amplification of silencing is important.
A clue to the type of situation in which secondary siRNA might be important comes from experimental RNA
interference in C. elegans and transitive transgene silencing in plants. In both systems, transitivity and secondary
siRNA production amplify silencing-related RNAs so that silencing persists in the absence of the initiator doublestranded RNA. In some instances associated with this persistence, there are epigenetic effects at the DNA or
chromatin level (6, 7). On the basis of these observations, and reasoning that experimental systems may illustrate
elements of the natural mechanisms, it seems likely that the endogenous secondary siRNAs could mediate effects
of silencing that persist in the absence of the initiator double-stranded RNA. Perhaps the amplified secondary
siRNAs influence processes such as developmental timing in which the effects of a silencing trigger might persist
after their initial induction. Consistent with this idea, secondary siRNAs in the plant Arabidopsis thaliana affect
the timing of the developmental transition between adult and juvenile growth phases (8).
In addition to the biological implications of the amplification mechanisms, there are two technical issues. First,
from a biotechnological perspective, it would be advantageous if the amplification mechanisms could be harnessed
to enhance silencing in therapeutic or genomic applications. The absence of RdRP genes in the fly Drosophila
melanogaster and in mammalian genomes indicates that this effect might not be possible in all organisms. However,
there are recently described siRNA-like species in Drosophila (9) with the phased and strand-bias characteristics
of secondary siRNAs in C. elegans. Perhaps there are other enzymes in mammals that can substitute for the RdRP
proteins in an amplification process. The second technical point is a cautionary message about methods for highthroughput sequencing of siRNA populations. Secondary siRNAs with 5′-triphosphates are excluded from many of
the methods associated with this technology, and amplified siRNAs would be missed. Fortunately, two of the C.
elegans papers (1, 2) describe methods for cloning and sequencing siRNA with 5′-triphosphate. We will now see to
what extent the existing sequence databases will need to be revised to account for 5′-triphosphorylated siRNAs.
References
1.
J. Pak, A. Fire, Science 315, 241 (2007); published online 23 November 2006 (10.1126/science.1132839).
2.
T. Sijen, F. A. Steiner, K. L. Thijssen, R. H. A. Plasterk, Science 315, 244 (2007); published online 7
December 2006 (10.1126/science.1136699).
3.
M. J. Axtell, C. Jan, R. Rajagopalan, D. P. Bartel, Cell 127, 565 (2006).
4.
J. G. Ruby et al., Cell 127, 1193 (2006).
5.
E. Yigit et al., Cell 127, 747 (2006).
6.
N. L. Vastenhouw et al., Nature 442, 882 (2006).
7.
O. Voinnet, P. Vain, S. Angell, D. C. Baulcombe, Cell 95, 177 (1998).
8.
N. Fahlgren et al., Curr. Biol. 16, 939 (2006).
9.
V. V. Vagin et al., Science 313, 320 (2006).
10.1126/science.1138030
The author is in the Sainsbury Laboratory, John Innes Centre, Norwich NR4 7UH, UK. E-mail:
david.baulcombe@sainsbury-laboratory.ac.uk
A Third Way to Silence RNA
116104104
4
2/12/2016
BIOLOGY 475: MOLECULAR BIOLOGY
SPRING 2007
Two well-characterized RNA silencing pathways use small RNAs. Small interfering RNAs (siRNAs) act as targeting
molecules in RNA interference (RNAi), and microRNAs (miRNAs) are encoded in the genome as tiny noncoding RNA
genes. Although distinct, these pathways share a number of components, such as the endonuclease Dicer, which
produces RNAs with a characteristic length of ~22 nucleotides (nt). Vagin et al. and Lau et al. report the initial
characterization of a third putative RNA silencing pathway in animals, characterized by ~30-nt small RNAs in the
germline--so-called repeat associated (ra) siRNAs in Drosophila and Piwi-interacting RNAs (piRNAs) in mammals
(see the Perspective by Carthew). In both cases, these RNAs map specifically either to the sense or antisense
strand, but rarely to both, which suggests that, in contrast to siRNAs and miRNAs, they do not arise from doublestranded precursors. The rasi- and piRNAs purify with Piwi proteins, homologs of the Ago proteins found in RNAi
and miRNA pathways. Dicer enzymes do not appear to be involved in the generation of the rasiRNAs and,
intriguingly, a weak slicing activity is associated with the piRNA complex.
V. V. Vagin, A. Sigova, C. Li, H. Seitz, V. Gvozdev, P. D. Zamore, A distinct small RNA pathway silences selfish
genetic elements in the germline. Science 313, 320-324 (2006). [Abstract] [Full Text]
N. C. Lau, A. G. Seto, J. Kim, S. Kuramochi-Miyagawa, T. Nakano, D. P. Bartel, R. E. Kingston, Characterization of the
piRNA complex from rat testes. Science 313, 363-367 (2006). [Abstract] [Full Text]
R. W. Carthew, A new RNA dimension to genome control. Science 313, 305-306 (2006). [Summary] [Full Text]
Citation: A Third Way to Silence RNA. Sci. STKE 2006, tw245 (2006).
A New RNA Dimension to Genome Control, Richard W. Carthew* Science 21 July 2006:
Vol. 313. no. 5785, pp. 305 – 306 DOI: 10.1126/science.1131186
MicroRNAs (miRNAs) and small interfering RNAs (siRNAs) are 21- to 25-nucleotide RNA molecules that influence
their much bigger relatives, the messenger RNAs (mRNAs). Over the past few years, these small RNA species have
captivated the study of gene regulation and modified our notions about how gene expression is controlled. A recent
clutch of papers describe for the first time a class of small RNA cousins that are distinct from miRNAs and
siRNAs (1-6). They promise to yield fascinating new insights into genome control.
The genesis of the discovery of a third type of small RNAs is linked to the Argonaute family of proteins. Certain
Argonaute proteins such as Ago1 and Ago2 associate with miRNAs and siRNAs to form ribonucleoprotein complexes
that associate and repress the expression of target mRNAs. Sometimes, mRNA targets are cleaved by a
mechanism that is catalyzed by the particular associated Ago protein. However, a subclade of Argonaute proteins
are phylogenetically distinct from the Ago1/Ago2 subclade and do not appear to associate with siRNAs or miRNAs
(7). Led by its founding member, the Piwi protein of Drosophila melanogaster, this subfamily appears to play an
important role in germline development. For example, genetic analysis of Piwi and its mouse orthologs (Miwi, Mili,
Miwi2) indicates that they are essential for spermatogenesis (7-9). How the Piwi subfamily regulates the germline
has remained for the most part elusive.
A recent breakthrough regarding this question has emerged from the identification of the small RNA partners
that associate with Piwi proteins. In research reported on page 363 of this issue, Lau et al. (1) partially purified a
ribonucleoprotein complex from extracts of rat testis and found testis-specific RNAs of 25 to 31 nucleotides, with
a dominant subpopulation of 29- to 30-nucleotide RNAs. These RNAs are distinct in size from miRNAs and are
associated with distinct protein complexes. Lau et al. purified the RNA-protein complex by conventional
chromatography and identified the rat homologs to Piwi (Riwi) and the human RecQ1 protein as subunits of the
complex. On this basis, the RNAs have been named Piwi-interacting RNAs (piRNAs) and the complex is called the
Piwi-interacting RNA complex (piRC).
piRC exhibits adenosine triphosphate-dependent DNA helicase activity, which is likely attributable to RecQ1.
Interestingly, the RecQ1 homolog in Neurospora crassa has been implicated in gene silencing (10). Lau et al. (1) also
116104104
5
2/12/2016
BIOLOGY 475: MOLECULAR BIOLOGY
SPRING 2007
found that piRC will cleave RNA targets in a manner dependent on piRNA complementarity, much like Ago2 cleavage
of siRNA targets. This might imply that piRC has some posttranscriptional role in gene silencing. Indeed, piRNAs
are associated with polysomes fractionated from mouse testis extract (4). However, genetic studies have
implicated Piwi proteins in transcriptional gene silencing by altering chromatin conformation (7). Consistent with
the idea that piRC plays a role in transcriptional silencing, RNAs that are longer than 24 nucleotides (such as
piRNAs) have been associated with this mode of gene silencing in a wide variety of species (11).
One of the real surprises has been the nature of piRNAs themselves. Deep sequencing of complementary DNAs
derived from piRNAs revealed that they correspond to regions of the genome previously thought not to be
transcribed (1-3). These regions are limited in number to about 100 clusters ranging in size from 1 to 100 kilobases
and are distributed across the genome. Very few clusters contain repetitive DNA; in fact, repetitive DNA is
underrepresented. A greater surprise is that piRNAs in a typical cluster exclusively map to either one or the other
strand of genomic DNA (1-5). A minority of clusters generates piRNAs from both strands, but plus-strand piRNAs
are segregated from minus-strand piRNAs into distinct regions that are separated by a gap of a few hundred base
pairs (see the figure). The paucity of evidence for overlapping complementary RNAs or potential foldback RNA
precursors suggests that piRNAs are not derived from double-stranded RNA precursors. This would suggest a
biogenesis mechanism distinct from that of siRNA and miRNA, both of which are derived from dsRNA through
enzymatic cleavage by Dicer.
The known and unknown features of piRNAs. Shown is a genomic region that generates a cluster of piRNAs. The
left side of the region generates antisense RNA transcripts (blue) and the right side generates sense transcripts
(green); a short gap in between likely acts as the promoter for transcription of both sides (divergent red arrows).
An RNA polymerase of unknown identity is shown in active transcription. These transcripts are processed into 25to 31-nucleotide piRNAs by an unknown mechanism. piRNAs then associate with Piwi and RecQ1 homologs to form
piRC complexes. These complexes might regulate the genome at the level of DNA or histones, or at a
posttranscriptional level. The events that are under direct control of the piRC mechanism within developing sperm
cells are currently unknown.
piRNAs and piRC complexes are not restricted to rats; they have been detected in testes of other mammals,
including mouse and human (1-5). The organization of piRNA genomic clusters is conserved in other mammalian
species as well. Most clusters in the rat, mouse, and human are homologous or syntenic, even extending to strand
116104104
6
2/12/2016
BIOLOGY 475: MOLECULAR BIOLOGY
SPRING 2007
specificity. Nonetheless, piRNA sequences are not conserved between species. Sequence heterogeneity is
consistent with the idea that the genomic clusters are subject to neutral selection pressure.
What testis cells express piRNAs? In the mouse, Mili is expressed in male germ cells from primordial to pachytene
stages, whereas Miwi is expressed from pachytene to spermatid stages (8, 9). Both Mili and Miwi associate with
piRNAs, which suggests that piRNAs are produced within the developing male germ cells. Consistent with a
germline-specific expression pattern, piRNAs are not detectable in WV mutant mice, which are missing
differentiating germ cells (2, 5), and piRNAs are reduced in Miwi mutants (4). piRNAs are detected throughout
sperm development but appear to peak in abundance at the round-spermatid stage (1-5). The abundance of piRNAs
in spermatids is staggering; about 1 million molecules are estimated per round-spermatid cell (3).
Although mammalian piRNAs are not associated with repetitive DNA, the situation might be different in
Drosophila. On page 320 of this issue, Vagin et al. (6) describe repeat-associated siRNAs (rasiRNAs) in the fly
germline as 24- to 29-nucleotide species that arise primarily from the antisense strand of repetitive sequences
such as retrotransposons. These RNAs are associated with Piwi and another member of the Piwi subclade, and
mutations in the Piwi class of genes cause derepressed retrotransposon silencing coupled with altered levels of
rasiRNA abundance. Interestingly, these effects are not restricted to the male germline but also apply to the
female germline. Perhaps rasiRNAs in Drosophila use a molecular mechanism similar to that of mammalian piRNAs to
silence portions of the genome.
Many questions ensue from these studies. Are testis-specific piRNAs found in species other than mammals? Does
piRC regulate male meiosis by regulating genome organization, or is it a surveillance mechanism to ensure genome
integrity during germ cell maturation, including suppression of selfish elements? Is piRC male-specific, or are other
classes of RNAs associated with Piwi in the female germline? How are piRNAs produced? Their structures might
suggest a ribonuclease III-independent origin. Indeed, neither of the two Dicers from Drosophila is essential for
rasiRNA biogenesis and repeat DNA silencing (6), although it is possible that each is redundant for the other or
that a third enzyme, Drosha, carries out processing. Further investigation should reveal how piRCs regulate the
genome.
References
1.
N. C. Lau et al., Science 313, 363 (2006); published online 15 June 2006 (10.1126/science.1130164).
2.
A. Girard, R. Sachidanandam, G. J. Hannon, M. A. Carmell, Nature 10.1038/nature04917 (4 June 2006)
[CrossRef].
3.
A. Aravin et al., Nature 10.1038/nature 04916 (4 June 2006) [CrossRef].
4.
S. T. Grivna, E. Beyret, Z. Wang, H. Lin, Genes Dev. 10.1101/gad.1434406 (9 June 2006) [CrossRef].
5.
T. Watanabe et al., Genes Dev. 10.1101/gad.1425706 (9 June 2006) [CrossRef].
6.
V. V. Vagin et al., Science 313, 320 (2006); published online 29 June 2006 (10.1126/science.1129333).
7.
M. A. Carmell, Z. Xuan, M. Q. Zhang, G. J. Hannon, Genes Dev. 16, 2733 (2002) [CrossRef].
8.
W. Deng, H. Lin, Dev. Cell 2, 819 (2002) [CrossRef].
9.
S. Kuramochi-Miyagawa et al., Development 131, 839 (2004) [CrossRef].
10.
C. Cogoni, G. Macino, Science 286, [2342] (1999).
11.
M. A. Matzke, J. A. Birchler, Nat. Rev. Genet. 6, 24 (2005) [CrossRef].
116104104
7
2/12/2016