Strategies for Combinatorial Development of
siRNA Conjugate Delivery Systems
0z
C\J
Q LL
=
-U
By
Rosemary L. Kanasty
Submitted to the Department of Chemical Engineering in partial fulfillment of
the requirements for the degree of Doctor of Philosophy in Chemical Engineering at
the Massachusetts Institute of Technology
June 2015
2015 Massachusetts Institute of Technology
All rights reserved.
Signature of Au thor......
Sicinature redacted ..Rosemary
L. Kanasty
Department of Chemical Engineering
&Certified by......
Signature red acted...Rbr
.Lne
.......................... Robert S. Langer
David H4 Koch (1962?) Instituten Professor
S.
SSignature redacted
Certified by.........
............
v........
.........
....
Thesis Supervisor
T
..................... Daniel G. Anderson
Samuel A. Goldblith Professor of Applied Biology,
Chemical Engineering and Health Sciences & Technology
Thesis Supervisor
Signature redacted .....................
Accepted by ....... ,V......... . .,
...............................
Richard D. Braatz
<ofessor of Chemical Engineering
Chairman, Committee for Graduate Students
02
z
Strategies for Combinatorial Development of siRNA
Conjugate Delivery Systems
By Rosemary Lynn Kanasty
Abstract
RNA interference (RNAi), which can reversibly silence the expression of any gene,
has vast potential as a therapeutic to treat many diseases. A wide variety of small
interfering RNA (siRNA) delivery systems has been studied, including lipid nanoparticles
(LNPs) and small molecule conjugates. LNPs are among the most potent and diverse nonviral delivery systems, and they have enjoyed continued improvement in efficacy as new
generations of lipids have been explored. This steady progress in lipid performance has
relied heavily on studies using combinatorial synthesis and high-throughput screening of
large libraries to discover novel delivery materials. Conjugate delivery systems are
attractive for their purity, well-defined structures, low ratios of delivery material to siRNA,
and potentially broad therapeutic windows, yet relatively few efficacious conjugates are
reported in the literature. As synthetic challenges prevent the study of large conjugate
libraries, the development of conjugate systems has relied on individual synthesis of new
materials. There is currently a need for new directions in the development of conjugate
delivery systems, which demands methods to enable high-throughput screening of
conjugate libraries.
Here we present a viable synthetic strategy for creating combinatorial libraries of
hundreds of siRNA conjugate delivery materials. We first demonstrate the synthesis of
novel sequence-defined polymers that can incorporate a broad diversity of chemical
properties relevant to delivery into oligomers with high purity. We develop methods of
synthesis, purification, and siRNA conjugation that are translatable to high-throughput
synthesis in multiwell plates. We then demonstrate the high-throughput capability of this
method by synthesizing a library of over 500 novel siRNA conjugate materials. We explore
applications of these novel conjugates as potential delivery materials in cell-based screens.
We also develop strategies for conjugating materials to siRNA multivalently, which may
improve delivery. This work enables the synthesis of hundreds of novel conjugate delivery
materials in parallel. The study of these combinatorial libraries has the potential to open
vast new avenues in the development of therapeutic siRNA conjugates.
2
Table of Contents
A B ST R A CT ......................................................................................................................................... 2
A CK N O W LED G EM EN T S ................................................................................................................. 6
CH A PT ER 1:BA CK G R O U N D ......................................................................................................... 7
1.1 M OTIVATION FOR siRN A THERAPEUTICS ............................................................................................. 7
1.2 CHALLENGES OF IN VIVO siRN A DELIVERY ........................................................................................... 7
1.2.1 Pharmacodynamic considerations............................................................................................. 8
1.2.1.1 RNA structure and Gene Silencing ...................................................................................................................... 8
1.2.1.2 Im m unostim ulation ............................................................................................................................................... 10
1.2.1.3 Off-target gene silencing ...................................................................................................................................... 14
1.2.1.4 Saturation of the RNAi m achinery ................................................................................................................... 16
1.2.1.5 Delivery m aterial toxicity .................................................................................................................................... 17
1.2.2 Pharmacokinetic considerations.............................................................................................. 19
1.2.2.1 siRNA in circulation ................................................................................................................................................20
1.2.2.2 Extravasation ............................................................................................................................................................21
1.2.2.3 Cell uptake ..................................................................................................................................................................23
1.2.2.4 Elim ination .................................................................................................................................................................26
1.2.3 Conclusions.........................................................................................................................................27
1.3 STRATEGIES FOR siRN A DELIVERY ...................................................................................................... 28
1.3.1 Cy clodextrin polym er nanoparticles....................................................................................... 30
1.3.2 Lipid nanoparticles.........................................................................................................................31
1.3.2.1 Cationic and ionizable lipids .............................................................................................................................. 32
1.3.2.2 Shielding lipid ........................................................................................................................................................... 34
1.3.2.3 Cholesterol ................................................................................................................................................................. 35
1.3.2.4 Targeting Ligands ................................................................................................................................................... 35
1.3.3 Conjugate delivery system s ......................................................................................................... 36
1.3.3.1 Dynam ic PolyConjugates ..................................................................................................................................... 36
3
1.3.3.2
T rianten nary G al N A c-siR N A ...............................................................................................................................
39
1.3.3.3 O ligon ucleotid e N an op articles ..........................................................................................................................
40
1 .3.4 Co n clusio ns .........................................................................................................................................
1 .4 T H E SIS G OA LS............................................................................................................................................
41
43
CHAPTER 2: SEQUENCE-DEFINED OLIGOMERIC MATERIALS FOR SYNTHESIS OF
NUCLEIC ACID CONJUGATE LIBRARIES ...................................................................................
44
2 .1 IN TRO D U C TIO N ........................................................................................................................................
44
2.2: POTENTIAL CHEMICAL STRATEGIES FOR SYNTHESIZING CONJUGATE LIBRARIES ..................
45
2 .2.1 In trod u ction .......................................................................................................................................
45
2.2.2 DegradableEster-linked Oligomers.................................................................................
46
2.2.3 Oligocarbamatesbased on 2-amino 4-hydroxybutyric acid....................................
48
2.2.4 Oligocarbamatesbased on 4-amino 3-hydroxybutyric acid.....................................
49
2.2.5 Oligocarbamatesbased on 4-amino 2-hydroxybutyric acid.....................................
51
2.2.6 Oligocarbamatesbased on hydroxyproline...................................................................
54
2.3 HYDROXYPROLINE-BASED OLIGOCARBAMATES PROVIDE A ROBUST PLATFORM FOR THE
SYNTHESIS OF SIRNA CONJUGATE LIBRARIES ...............................................................................................
57
CHAPTER 3. HIGH-THROUGHPUT SYNTHESIS OF A HYDROXYPROLINE OLIGOMERSIRNA CONJUGATE LIBRARY............................................................................................................65
3 .1 IN T R O D U CTIO N ........................................................................................................................................
65
3.2 HIGIH-THROIJGHPUT SYNTIESIS OF IYDROXYPROLINE-BASED OLIGOCARBAMATES............ 66
3.3 CONJUGATION OF OLIGOCARBAMATES TO SIRNA AND CONJUGATE PURIFICATION................ 70
3.4 CONCLUSIONS AND RECOMMENDATIONS.......................................................................................
71
CHAPTER 4: POTENTIAL APPLICATIONS OF HYDROXYPROLINE-BASED
OLIGOCARBAMATES
............................................................................................................................73
4 .1 IN T R O D U C T IO N ........................................................................................................................................
73
4
4.2 LIBRARY SCREENING IN DUAL HELA CELLS ....................................................................................
74
4.2 LIBRARY SCREENING IN K B CELLS ....................................................................................................
76
4.3
81
HYDROXYPROLINE OLIGOMERS AS LIPID-LIKE NANOPARTICLE DELIVERY SYSTEMS.............
4.4. CONJUGATE MODIFICATIONS AS ENHANCERS IN OTHER DELIVERY SYSTEMS.........................85
4.5. RECOMMENDATIONS FOR FUTURE IN VITRO SCREENS ...............................................................
88
CHAPTER 5. STRATEGIES FOR MULTIVALENT CONJUGATION TO SIRNA.............90
CHAPTER 6. CONCLUSIONS....................................................................................................94
APPENDIX 1. ABBREVIATIONS AND SYMBOLS ...............................................................
96
APPENDIX 2. MATERIALS AND METHODS.......................................................................97
A PPEN DIX 2.1 LIST OF M ATERIALS .............................................................................................................
97
A PPEN DIX 2 .2 M ETH OD S ..............................................................................................................................
97
APPENDIX 3. SUPPLEMENTARY DATA ...............................................................................
104
APPENDIX 3.1 'H NM R AND 13C N M R SHIFTS ......................................................................................
104
APPENDIX 3.2 LCMS ANALYSIS OF TRIMER AND HEXAMER SYNTHESIS ............................................
110
APPENDIX 3.3 HIGH-THROUGHPUT PURIFICATION OF SIRNA CONJUGATES .....................................
122
APPENDIX 3.4 HPLC ANALYSIS OF SIRNA CONJUGATES ......................................................................
123
APPENDIX 3.5 SYNTHESIS OF MULTIVALENT DBCO AND BCN LINKERS ..........................................
125
APPENDIX 4. REFERENCES .....................................................................................................
128
5
Acknowledgements
I would like to thank my thesis advisors, Prof. Bob Langer and Prof. Dan Anderson, for
their unwavering support, positivity, and their fostering of a laboratory culture that inspires
openness and imagination. Bob is both a considerate mentor and an inspiring role model
who is very deserving of the widespread respect and admiration he has garnered. Dan's
guidance, along with his creativity, ambition, and penchant for big ideas has motivated me
and made me a better scientist.
I am grateful to have enjoyed the support of a thesis committee that offers unparalleled
scientific expertise, consistent encouragement, and exemplary personal mentorship. For
this I would like to thank Prof. Sangeeta Bhatia and Prof. Dane Wittrup.
I am extremely fortunate to have had the daily guidance of Dr. Arturo Vegas, who has
generously contributed his time to training me throughout my graduate career. I have him
to thank not only for my laboratory skill set, but for years of scientific guidance,
experimental troubleshooting, professional mentorship, and emotional support through all
of challenges of graduate school.
Among the many brilliant scientists I've met in the Langer and Anderson labs, I would
like to especially thank Luke Ceo, Owen Fenton, Omar Khan, Ben Tang, Karsten Olejnik,
Omid Veiseh, Kevin Kauffman, Hok Hei Tam, Abel Cortinas, Robert Dorkin, Abigail LyttonJean, Katie Whitehead, Kevin Love, Ana Jaklenec, Josh Doloff, Chris Alabi, Leon Bellan,
H6loise Ragelle, Eunha Kim, Minglin Ma, Chris Levins, Eddie Eltoukhy, David Nguyen, Pam
Basto, Eric Pridgen, Aleks Radovic-Moreno. Yizhou Dong, Weiheng Wang, Guarav Sahay, Hao
Yin, Patrick Fenton, Beata Chertok, and Avi Schroeder.
I would also like to thank our collaborators at Alnylam: Klaus Charisse, Martin Maier,
Don Foster, and Stuart Milstein. For funding I owe my gratitude to the National Science
Foundation, Alnylam Pharmaceuticals, and the MIT Skoltech Initiative.
I must also thank my parents for enthusiasm for science and engineering and for
encouraging me as I pursued my goals. Finally, I would like to thank my friends and family
for their support, confidence, patience, and comfort over the last several years.
6
Chapter 1: Background
Reproduced with permission from: Kanasty et al., Molecular Therapy, 20121; Kanasty et al.,
Nature Materials, 20132; and Yin et al., Nature Reviews Genetics, 20143.
1.1 Motivation for siRNA therapeutics
Since the discovery of RNA interference (RNAi) in mammalian cells, there has been
great interest in harnessing this pathway for the treatment of disease. RNAi is an
endogenous pathway for post-transcriptional silencing of gene expression that is triggered
by double-stranded RNA, including endogenous microRNA (miRNA) and synthetic short
interfering RNA (siRNA). By activating this pathway, siRNAs can silence the expression of
virtually any gene with high efficiency and specificity, including targets traditionally
considered to be "undruggable." The therapeutic potential of this method is far-reaching,
and siRNA-based therapeutics are under development for the treatment of diseases ranging
from viral infections 4s to hereditary disorders 6 to cancers 7,8. Large amounts of effort and
capital have been invested in bringing siRNA therapeutics to the market. At least 22 RNAibased drugs have entered clinical trials (Table 1), and many more are in the developmental
pipeline.
1.2 Challenges of in vivo siRNA delivery
siRNA molecules must be delivered intracellularly in order to trigger RNAi, but their
large, anionic, and hydrophilic structure prevents them from diffusing across cell
membranes to reach their site of action. A broad array of delivery materials has been
developed to mediate delivery and to improve the overall pharmacokinetics of siRNA when
administered systemically (reviewed elsewhere, see reference 9). While the design of siRNA
delivery systems is usually focused on the immediate goal of delivering enough of the
injected dose into enough target cells to generate a particular therapeutic effect, a broader
awareness of the fate of the entire injected dose and its many other interactions with the
body is essential in the development of safe and effective systems. siRNAs and their
delivery vehicles can induce immunogenicity, toxicity, and off-target effects in addition to
target gene silencing. Various factors can mediate the degradation, clearance, and cellular
uptake of siRNA delivery systems, affecting their potency. This chapter presents a brief
summary of important pharmacodynamic and pharmacokinetic considerations in siRNA
delivery and intends to introduce the reader to the numerous interactions between the
body and siRNA, but is not a comprehensive review of PD and PK data for various delivery
systems. For more detailed information on particular topics, the reader is referred to
reviews of more limited scope and of greater depth.
7
1.2.1 Pharmacodynamic considerations
siRNA delivery systems can elicit both intended effects (e.g. target gene silencing) as
well as unintended consequences, including immune stimulation, toxicity, and off-target
silencing. The pharmacodynamics of siRNA delivery systems are dependent not only upon
the siRNA therapeutic, but also upon the biomaterials included in the delivery vehicle.
.- rxjfriva.j
Pd III
-
sR A
/0nmotm
*\
av
.e
-BE=~A
*-
'if andii
/
ci
XeA
KS oalang
K
jrd
Perfect hybricabt*on
Partia i,? rizANA
to 3-1t-Rg
I
Vaarni
.ciivmin
fr ot1ple hyt1
up% rreRNA
I tutin
T,ge rtiOiNA
siRNA pathway
m~RNA pathway
Figure 1-1. sIRNA and miRNA pathways. Exogenous double-stranded RNA dsRNA) introduced into the cytoplasm is
cleaved by Dicer into 21-nt fragments with 2-nt overhangs on the 3' ends. The antisense strand, or guide strand, is loaded
into the RNA-tnduced silencing complex and the sense or passenger strand is cleaved by Argonaute-2 Ago-2). The
activated RISC-siRNA complex seeks out messenger RNA (mRNA) that hybridizes with the guide strand and cleaves this
mRNA. The same silenang machinery particpates in gene silencing by endogenous miRNAs MiRNAs usually hybridize
imperfectly with target mRNA. which leads to translational repression rather than cleavage of target mRNA. Off-target
gene silencing by synthetic siRNA is thought to occur when the siRNA seed region is partialy complementary to 3'UTR
sequences, causing siRNAs to enter the miRNA pathway
1.2.1.1 RNA structure and Gene Silencing
8
The central component of any siRNA formulation is the siRNA itself. The silencing of
specific genes through RNA interference machinery (Figure 1.1) is the intended
pharmacodynamic effect of siRNA. Over the past few years, a number of design criteria
have been developed to improve the specificity and potency of RNA, some of which has
been driven by an increased understanding of the biochemistry driving RNA-interference
10,11
When an siRNA molecule is delivered inside of a cell, it can be loaded into the RNAinduced silencing complex (RISC) (Figure 1.1). Argonaute 2 (Ago 2, Figure 1.1), a protein
within RISC, unwinds the siRNA. The two strands of siRNA are referred to the passenger
strand and the guide strand. The passenger strand is cleaved while the activated RISC-guide
strand complex seeks out and cleaves mRNA that is complementary to the guide strand.
The nucleotide sequence of the siRNA affects the efficiency of each of these steps and
therefore affects the potency of gene silencing. A number of guidelines have emerged for
selection of the most active siRNA sequences, and these principles have been incorporated
into various siRNA design algorithms 12. The enhancement of siRNA asymmetry and
minimization of secondary structure in mRNA target sites are both important
considerations in sequence selection.
When siRNA enters the RNAi pathway (Figure 1.1), one strand is loaded into the RISC
complex to become the guide strand and its complementary strand is cleaved. If the
incorrect strand is selected for RISC loading, the strand intended to be the guide strand is
cleaved, leading to reduced potency of target gene silencing. Incorrect strand selection can
also trigger the silencing of off-target genes complementary to the intended passenger
strand. siRNA asymmetry refers to the preference for one strand to be loaded into RISC
over the other. Increasing the asymmetry of the duplex helps to ensure that the desired
strand is loaded into RISC to direct gene silencing. Strand selection by RISC is determined
by the thermodynamic stability of the ends of the duplex: the 5' end with the lowest
hybridization stability is loaded into RISC to become the guide strand 13-15. Thermodynamic
asymmetry is increased by enriching the A-U base pair content in the 5' end of the antisense
strand. Unlocked nucleic acids (UNA, Figure 1.4) destabilize duplexes and can enhance
asymmetry when incorporated into the 5' end of the guide strand 16-18.
Asymmetry can also be enhanced by modification of passenger strand ends. For siRNAs
to be functional, the 5' end of the antisense strand must have a terminal phosphate 15. 5'-OH
-terminated siRNAs are rapidly phosphorylated in the cytoplasm 19, but modification of the
terminal hydroxyl can prevent phosphorylation and prevent RISC loading 20. Methylation of
9
the 5' end of the passenger strand is a common strategy to ensure proper strand selection
by RISC 20.
The presence of delivery materials can affect RISC loading. Delivery materials that
complex with siRNA to form nanoparticles must dissociate from siRNA once inside the cell
to allow RISC loading. For example, the cationic lipids of many lipoplex delivery systems
undergo a phase transition in acidic environments such as endosomes. The conversion
from a lamellar phase to hexagonal packing promotes both endosomal release and release
of siRNA from the delivery material 21. In other systems, delivery materials are conjugated
directly to the ends of siRNA strands. As the 5' end of the guide strand is essential for RISC
loading, conjugation of delivery materials is usually avoided in this position. Delivery
materials are usually conjugated to the 3' or 5' ends of the passenger strand, though
conjugation to the 3' end of the guide strand can be effective as well 13,22,23.
The level of accessibility of the mRNA target site also affects the potency of siRNA 14,24,25.
Folding of mRNA can block access to the target site, as activated RISC cannot unfold
secondary structures in mRNA 24. Secondary structure prediction algorithms can help to
choose mRNA target sites with high accessibility 12,14. Self-folding of the siRNA guide strand
can also impede target site recognition, and inverted repeats are often avoided in siRNA
sequences 26,27. An example of a siRNA design algorithm that considers RNA secondary
structures is the OligoWalk web server, which selects active siRNA sequences based on their
free energy changes upon hybridization with target mRNA 12.
The internal stability of the length of the duplex can also affect activity. A systematic
analysis of a siRNA library determined that the most active sequences had a GC content of
36% - 52% 11. It has been proposed that higher GC content may inhibit RISC loading and
release of the cleaved sense strand, while lower GC content may inhibit activity by
weakening hybridization between the guide strand and mRNA 10.
Several studies of siRNA sequences have confirmed a preference for certain nucleotides
in specific positions along siRNA strands 11,28. Several of these positional preferences are
consistent with the idea of thermodynamic asymmetry of siRNA ends: A or U is preferred at
position 1 of the antisense strand, G or C is preferred at position 19, and AU-richness in
positions 1 - 7 is favored. In addition, A or U is favored at position 10, the site of substrate
cleavage by Ago2.
1.2.1.2 Immunostimulation
10
TLR-Indspondent Pathways
TLR-Oependmnt Pathways
WVAYO%
SARNA
'I
V
cytos
A
A
ME
h
I
Figure 1-2 sMA invokes Ihmunmostmulation via pathogen recognitIn. receptor. TLR3 (red' exists on
both the cell surface and in subcellular compartments of select cells. TLR7 (orange) and TLR8 (yellow) exist
solely in the endosomes and lysosomes of specialized immune cells and recogruze siRNA in a sequencedependent manner. PKR, and RIG-1 are present in the cytoplasm of certain cell types and can detect and react
to siRNA in a sequence-independent fashion. When activated, receptors trigger an immune signaling cascade
that causes an increased transcription of mRNA that encodes Type I interferons and inflammatory cytokines.
The cascade and resulting inflammatory response are unique for each receptor.
siRNA can be recognizedby the innate immune system
In addition to mediating RNA interference, siRNA molecules have the potential to induce
the innate immune system 29. The interactions between siRNA and the immune system are
complex and only an overview is provided here. For a detailed review, see reference 30.
The process of immunological recognition and response to non-self nucleic acids such as
siRNA is governed by the innate immune
system. Innate
immune responses are
11
characterized by an induction of small signaling molecules called cytokines, including
interleukins 31, Type I interferons 32 and tumor necrosis factor alpha (TNF-a) 33.
The stimulation of pattern recognition receptors (PRRs) triggers the innate immune
response. PRRs recognize distinct pathogenic patterns that are not present on self-cells 34,35
Importantly for siRNA delivery, the innate immune system has evolved to include multiple
ZPRRs that recognize different aspects of RNA structure, providing a redundancy that
makes immunostimulation difficult to escape (Figure 1.2). Two general classes of siRNArecognizing PRRs include Toll-like receptors and cytoplasmic receptors 36.
Toll-like receptors (TLRs) are a class of PRRs that recognize structurally conserved
regions of foreign pathogens, and each member of the TLR family is responsible for the
detection of varying pathogen-associated molecules. Of the 10 human TLRs and 12 murine
TLRs that have been discovered to date 37, at least three have relevance to siRNA delivery.
These include TLR3, which recognizes dsRNA, and TLR7 and TLR8, which recognize singlestranded RNA (ssRNA). TLR3 responds to dsRNA, which is typically a characteristic of viral
replication found in lysed or apoptotic virally-infected cells 38. In humans, TLR3 is expressed
in endosomes and on the cell surface of select cell populations. TLRs 7 and 8 are important
PRRs that respond to ssRNA in a sequence-specific manner. They are located exclusively in
the intracellular vesicles, including endosomes, lysosomes, and the endoplasmic reticulum
of plasmacytoid dendritic cells, B cells, and myeloid cells 35. More specifically, TLR 7 is
located primarily in the endosomes of plasmacytoid dendritic cells and TLR 8 in endosomes
of myeloid cells
39.
In addition to endosomal TLRs, several cytoplasmic receptors mediate immune
responses to siRNA (Figure 1.2). PKR, a protein kinase that responds to dsRNA 40 and
siRNA 35, can cause the inhibition of protein translation and an interferon response when
activated 4 0. RIG-I, another cytoplasmic PRR expressed in fibroblasts and dendritic cells 41,
can provoke a strong interferon response in the presence of various forms of siRNA 35.
Several features of siRNA delivery systems influence the nature and degree of
immunostimulation. These include the siRNA sequence, the siRNA structure and chemistry,
and the materials used to construct the delivery vehicle 29. The sequence of siRNA
influences immune stimulation through TLRs 7 and 8, which recognize ssRNA in a
sequence-specific manner 42,43. Generally, siRNAs rich in GU-rich motifs tended to provoke
more immunostimulatory activity while a decrease in the presence of uridine residues had
the opposite effect. The substitution of certain residues within an RNA molecule has been
12
demonstrated to diminish pro-inflammatory response. In particular, substituting guanosine
or uridine residues with adenosine resulted in decreased cytokine and interferon activity 43.
The chemical structure of the siRNA duplex can also influence the degree of innate
immune response. RIG-1, for example, has been shown to bind to single-stranded or doublestranded RNA containing uncapped 5'-triphosphate groups resulting in an interferonmediated immune response 44-46. Uncapped RNA is a sign of viral infection, and
subsequently induces inflammatory action. Additionally, it has been reported that bluntended dsRNA (no overhangs) is also capable of provoking immunostimulatory activity
through recognition by RIG-I 35.
Nucleotide structure also has a profound effect on innate immune system activation. For
example, it has been established that modifying the 2' group on the ribose ring of the RNA
backbone can reduce or even eliminate the innate immune response. Locked nucleic acids
47, unlocked nucleic acids 48, as well as 2'-F, 2'-H (i.e. 2'-deoxy) and 2'-O-methyl
modifications (2'-O-Me) (Figure 1.4) 4,22 have all been shown inhibit immune activity to
some extent. Relative to other types of nucleotide modification, 2'-O-Me modifications
successfully inhibit TLR7/8-mediated recognition of siRNA without diminishing RNAi
.
potency 4 2
siRNA Delivery Vehicles Influence Immunostimulation
Although single doses of naked siRNA do not necessarily cause innate immune
activation 49, siRNA delivery materials can have a substantial effect on siRNA-mediated
immune activity. Delivery vehicles can account for immune responses varying up to two
orders of magnitude 39,50,51 and can act through a variety of mechanisms. Different delivery
materials escort siRNA molecules across cell membranes in unique ways and into differing
subcellular compartments, exposing the siRNA to differing numbers and types of PRRs.
For example, cationic materials are frequently employed for siRNA delivery
applications, as they readily condense RNA due to electrostatic interactions. In general,
cationic and lipid materials are believed to deliver siRNA through the mediation of
endosomal uptake into the cell and subsequent endosomal escape into the cytoplasm of the
cell 52. When formulated with a cationic delivery material, siRNA is trafficked through
several subcellular locations, allowing it to interact with Toll-like receptors in the
13
endosomal compartments of immune cells as well as with RIG-I and PKR (Figure 1.2), which
are present in the cytoplasm of most cell types. Therefore, this type of delivery material
would expose its cargo to many pattern recognition receptors, rendering the siRNA more
prone to innate immune recognition than a delivery material that avoided endosomal and
lysosomal delivery routes.
The size, charge, and biodistribution of a delivery vehicle may also alter the intensity
and nature of a siRNA-induced immune response. Judge and colleagues found that SNALPencapsulated and polyethylenimine-complexed siRNA caused a different immune response
than siRNA incorporated with polylysine, which forms a larger delivery particle. It has also
been shown that the immunostimulatory properties of the lipid-like delivery vehicle ND985 (Figure 1.3) depend significantly on formulation approach 53.
1.2.1.3 Off-target gene silencing
Early reports described siRNA-induced silencing as extremely specific: a single
mismatch in the target site completely abolished silencing 54. However, it was soon
demonstrated that siRNAs with partial sequence homology to off-target mRNAs could affect
the expression of many genes 55,56. Off-target silencing both complicates experimental
results and can lead to toxicity 57. One cause of off-target silencing is improper strand
selection by RISC. Passenger-strand silencing can be avoided by selecting siRNA sequences
with high thermodynamic asymmetry or by chemically modifying the sense strand, as
described above.
Guide strand off-targeting commonly occurs through complementarity of the siRNA 5'
end to the 3' untranslated regions (UTRs) of mRNA 29. Positions 1-8 of the 5' end of the
guide strand are referred to as the "seed region," a sequence that plays a key role in target
recognition by endogenous miRNAs 58. miRNAs regulate gene expression through
incomplete hybridization of the seed region with 3'UTRs, which results in gene suppression
by translational inhibition and deadenylation (Figure 1.1) 59,60. As 3' UTR sequences are
shared by many mRNAs, a single miRNA can regulate hundreds of genes 61-65. siRNAs and
miRNAs share the same silencing machinery, and siRNA off-targeting is thought to result
from siRNAs entering the miRNA pathway (Figure 1.1) 29. These miRNA-like effects can
result in downregulation of scores of off-target genes 18,66,67.
14
Though miRNA-like effects are responsible for most off-target gene regulation 10,68,
partial complementarity of siRNAs to coding regions of mRNA can also cause unintended
silencing. Significant gene silencing has been observed with as little as 11 nt of continuous
sequence homology in coding regions 56.
To avoid miRNA-like off-targeting, siRNA design algorithms attempt to select sequences
with minimal seed-region complementarity to 3' UTRs 69. However, partial homology with
some UTRs is unavoidable: an analysis of the human 3'UTR database revealed that even the
most infrequent 7-nt seed sequence still had complementarity with 17 different 3' UTR
sequences 70. Alternatively, chemical modification can be used to minimize miRNA-like
effects. Efforts have focused on modifications that destabilize hybridization between the
seed sequence and the 3'UTR.
The most common modification for minimizing off-target effects is 2'-OMe in the guide
strand seed region, specifically at position 2. In a microarray study, this modification
mitigated effects on 80% of off-target transcripts 67. It has been suggested that the
bulkiness of the 2'-OMe affects the conformation of the complex formed from Ago, the guide
strand, and the target mRNA in a way that inhibits binding to partially-complementary
targets more than to fully-complementary targets. Incorporation of UNA in the seed region,
particularly in position 7, also destabilizes siRNA-3'UTR hybridization 48, and was able to
reduce the number of off-target transcripts by >90% 16,17. Both methylation and UNA were
able to reduce off-targeting without reducing the potency of on-target silencing 16,17,67.
While destabilization of the seed region duplex has been shown to mitigate offtargeting, stabilization of the length of the duplex may improve specificity by other
mechanisms.
Of the four human Argonaute proteins, Ago-2 is the only one with
In
target cleavage
24.
for siRNA-mediated
activity required
endonuclease
coimmunoprecipitation studies, siRNA duplexes interacted with all four Ago proteins,
however, strands of duplexes with increased hybridization stability were only efficiently
separated by Ago2. In the same study, duplexes stabilized by LNA demonstrated both
reduced seed-region off-targeting and improved on-target potency 71.
Modifications such as methylation, UNA, and LNA can provide the additional benefit of
reducing the immunostimulatory effects of siRNA without reducing potency 4,22,42,48.
Optimization of siRNA sequence along with combinations of chemical modifications enables
15
the design of siRNAs that potently silence target genes with minimal off-target silencing or
immunogenicity.
1.2.1.4 Saturation of the RNAi machinery
Exogenous siRNAs share components of the RNAi pathway with endogenous miRNAs,
and competition for RNAi machinery may inhibit normal gene regulation by miRNAs.
Several studies have demonstrated the saturability of the RNAi pathway and the toxic
effects that can result from saturation 72-76. RNAi pathway saturation was first observed in
experiments using high expression of short hairpin RNAs (shRNAs).
ShRNAs are
transcribed from virus or plasmid vectors and have a stem-loop structure similar to that of
pre-miRNAs that enters the RNAi pathway. In cultured cells, overexpression of shRNAs
from plasmid vectors delivered by transfection reagents inhibited miRNA activity by
saturating Exportin-5 77, a protein required for transport of shRNAs and miRNA from the
nucleus to the cytoplasm. In adult mouse livers, sustained high expression of shRNAs,
achieved using viral delivery vectors, resulted in dose-dependent liver injury and fatality
that was associated with decreased expression of liver miRNAs 72.
Subsequent studies demonstrated that competition for RNAi machinery could also occur
downstream from Exportin-5, in the pathway relevant to silencing by synthetic siRNAs.
When multiple siRNAs were co-transfected into cells, efficacies of individual sequences
were reduced, suggesting competition for downstream silencing machinery 73,78. siRNAs
were also observed to affect the activity of endogenous miRNAs and of co-transfected
shRNAs 73. Overexpression of Expo-5 only partially relieved this competition, suggesting
saturation of downstream components of RISC 73. Competition between transfected siRNAs
and endogenous miRNAs was also suggested by an examination of genome-wide transcript
levels from a set of 150 published experiments. The analysis found that transfection of
siRNA, while silencing genes complementary to siRNA sequences, also led to increased
expression of common miRNA targets 76.
While Xpo-5 was previously identified as one saturable component of the RNAi
pathway, recent overexpression studies with shRNA have implicated the Ago proteins as
key factors in the saturation of RNAi. In adult mice, overexpression of Xpo-5 enhanced
shRNA efficiency but increased hepatotoxicity. Ago-2/Xpo-5 coexpression enhanced shRNA
silencing, reduced hepatotoxicity, and increased RNAi stability 79. Of the four human
Argonautes, Ago-2 was identified as the primary determinant of RNAi efficacy, toxicity, and
persistence both in cell culture and in mouse liver 79.
16
The harmful effects of RNAi saturation may be avoidable by limiting doses of exogenous
RNA 79. Hepatotoxic effects in mice were abrogated when shRNAs were expressed under
milder promoters 80. Therapeutic levels of gene silencing in vivo have been observed
without perturbing miRNA function. In mice and hamsters, LNP-delivered siRNAs achieved
80% silencing of clinically-relevant gene targets without affecting endogenous miRNA
activity 80.
1.2.1.5 Delivery material toxicity
Systemic or cellular toxicity can also be caused by delivery reagents themselves. In
many of the most efficient in vivo delivery systems, the therapeutic dose of siRNA is
accompanied by more than 7 times its weight in delivery materials 81-83, making vehiclerelated toxicity an important consideration. Delivery material toxicity can occur through
various mechanisms including immunostimulation (see above), membrane destabilization,
and alterations in cellular signaling or gene expression.
One element of the toxicity of lipoplexes and polyplexes is often attributed to their
positive charge, which is exacerbated in low pH environment of the endosomes and
lysosomes 84. Mice treated with certain cationic lipid nanoparticles showed signs of
hepatotoxicity and a systemic interferon type I response, attributed in part to activation of
TLR 4 85. This effect was not observed in mice treated with neutral or negatively-charged
particles 85. Cellular toxicity of cationic lipids has been linked to increased production of
reactive oxygen species 86,87 and subsequent rise in cellular calcium levels 86, which were not
Cholesterol-derived cationic
observed with neutral or negatively-charged lipids.
amphiphiles are known to inhibit protein kinase c (PKC) 84,88, which can cause cytotoxicity.
Cationic lipid toxicity varies greatly with structure and formulation. Cholesterolderived lipids with quaternary ammonium head groups can be potent PKC inhibitors and
are more toxic than those with tertiary amines 84,88. Introduction of heterocyclic rings into
head groups can delocalize charge and in some cases reduce this toxicity 84,89,90. Lipids that
contain biodegradable linkages between their polar head groups and hydrophobic tails are
also associated with lower toxicity 84,91. Use of ionizable cationic lipids, which are neutral in
circulation but become positively charged in the acidic endosomal environment, may
reduce nonspecific disruption of cell membranes 92. Formulation of cationic lipids with
17
nucleic acids partially neutralizes their positive charge, and the toxicity profiles of lipids
alone can be distinct from those of lipids in formulation with nucleic acid 84,93.
Unmodified PEI has been shown to induce systemic and/or cellular toxicity depending
on structure and formulation. Both linear and branched PEI can cause cytotoxicity by
damage to cell membranes and by depolarization of mitochondrial membranes, leading to
apoptosis 94. PEI formulations with high ratios of delivery material to siRNA can also cause
systemic toxicity due to their high positive charges. Particles with high surface charges
interact with the negatively charged membranes of blood cells, which can lead to
erythrocyte aggregation. Large aggregates can have difficulty passing through capillaries,
leading to serious systemic effects 95. Grafting of PEG to PEI can mitigate the problem of
erythrocyte aggregation 9s and reduce overall cytotoxicity 95,96. PEG-PEI also is not inert and
can affect cellular processes differently than PEI alone. PEG-PE has been shown to activate
apoptotic pathways in a cell-line and concentration-dependent manner 97.
c. Cholesterol-siRNA conjugate
a. Stable nucleic acid lipid particle (SNALP)
0
H
d. Lipidoid 98N12-5(1)
H
N
H
N
N
H
H
0
H
Cationic lipid Artf\j Poly(ethylene glycol)
Neutral lipid
Cholesterol
e. Lipidoid C12-200
HO
b. Cyclodextrin
OH
OH
0
HO
0
\fOH
0
HO
HO
H
H
-H9H
O
N
f. Poly(ethylene imine)
H
OH
NH,
H
OH
00
NH
NH-,
Figure 1-3. Examples of siRNA delivery systems. (a) Stable nucleic acid lipid particles (SNALPs) encapsulate siRNA as
cargo. They are formulated with cationic and neutral lipids, cholesterol, and poly(ethylene glycol) (PEG). (b) Cyclodextrin polymers form nanoparticles when complexed with siRNA. (c) Delivery materials such as cholesterol can be directly
conjugated to siRNA. (d,e) Lipid-like delivery agents are often formulated with cholesterol and PEG when administered
in vivo. (f) Poly(ethylene imine) can be linear or branched and is often grafted with PEG.
18
Microarray studies have begun to reveal that delivery materials can induce gene
expression changes, which can potentially cause toxicity and other off-target effects.
Commonly-used transfection reagents have been shown to induce their own gene
expression signatures even in the absence of siRNA 96-98. The cationic lipids Oligofectamine
and Lipofectin altered the expression of dozens of genes involved in various cellular
processes including proliferation, differentiation, and apoptosis pathways 98. PEI (25kDa)
and PEI-PEG altered the expression of several genes involved in apoptosis and
inflammatory signaling 96,97. In some stably-transfected cell lines, signaling triggered by
PEG-PEI increased CMV promoter activity, causing an increase in reporter gene expression
rather than siRNA-mediated knockdown 97. Gene signatures of these materials can change
when the material is formulated with nucleic acid 99 and can vary greatly for different
Structure of the delivery material
delivery materials, cell types, and treatment times.
affects gene expression signatures as well. When injected intratumorally, branched PEI
induced more gene expression changes that linear PEI 99, consistent with the higher
reported toxicity of branched PEI 100. An understanding of the alterations in gene
expression due to delivery materials is necessary to properly interpret experimental results
and to develop delivery formulations with minimal off-target effects.
While delivery materials alone can induce gene expression changes and cause toxicity,
delivery materials in formulation with siRNA may cause toxicity and off-target effects
Delivering siRNAs into
simply introducing siRNA into different cellular environments.
endosomes increases their potential to stimulate endosomal TLRs to elicit an immune
response. Delivering siRNAs into the cytoplasm increases their potential to saturate the
RNAi machinery or to enter the miRNA pathway, causing off-target silencing. As such,
delivery materials alone and in formulation with siRNA (even siRNA sequences that lack
targets) can have distinct gene signatures 99.
1.2.2 Pharmacokinetic considerations
Pharmacokinetic considerations determine the ability of delivery systems to reach the
cytoplasm of target cells and therefore to induce a therapeutic effect.
The fate of
intravenously administered siRNA depends heavily upon its structure, chemical
modification, and delivery formulation. These factors affect a delivery system's stability in
circulation, its tendency to extravasate, or exit blood vessels, its ability to enter cells, and its
likelihood of being cleared from circulation without being delivered.
19
Pharmacokinetic parameters such as circulation half-lives, area under the
concentration-time curve, elimination rates, and bioavailability are functions of these
various physiological processes.
For example, a short circulation half-life could be
explained by rapid excretion of the formulation in urine, or it may result from the rapid
uptake of the formulation by tissues. This section aims to describe the mechanisms and
processes that shape pharmacokinetic behavior of siRNA delivery systems.
Detailed
pharmacologic data for various delivery systems are beyond the scope of this review.
1.2.2.1 siRNA in circulation
Intravenously-administered siRNA is exposed to serum proteins, nucleases, and cells of
the innate immune system.
The
base
base
base
resulting
processing
and
/5 0
biodistribution of injected
siRNA
0
0
varies with both the structure of the
H
CH 3
siRNA and the delivery system in
which it is carried. Successful siRNA
RNA
2'-F-RNA
2'-OMe-RNA
delivery has often required the use of a
base
base
delivery
vehicle
because
naked,
unmodified siRNA is degraded by
0
serum RNAses and cleared from the
5/
bloodstream within minutes
0
LNA
base
0
0
nanoparticles
H
04__0
In
contrast, many delivery formulations,
such as lipid and polymer systems,
protect siRNA from nucleases by
encapsulating it as cargo inside
UNA
base
4101102.
103,104.
H
base
0
o4s
_
base
0
H
Phosphodiester
H
Phosphorothioate
Figure 1-4. Common chemical modifications to the
siRNA backbone. The 2'-OH of the ribose ring can be
replaced with methoxy (2'-OMe) or fluorine (2'-F)
moieties. Locked nucleic acids (LNA) contain a methylene bridge connecting the 2' oxygen to the 4' carbon.
Unlocked nucleic acids (UNA) lack a bond between the 2'
and 3'carbons.
The
nuclease
stability
of
unformulated siRNA can also be
dramatically improved by making
chemical modifications to the RNA
backbone
105.
A common strategy
involves modification of the ribose 2'OH group, as this functional group is
critical to the mechanism of many
serum RNAses
106,107.
Among the most
effective backbone modifications for
serum stability improvement are the
20
substitution of the ribose 2' hydroxyl with 2'-Fluorine or 2'-Methoxy groups (Figure 1.4). If
placed judiciously within the duplex, such modifications can add stability without affecting
RNAi activity 22,102. A similar effect is achieved by incorporation of locked nucleic acid (LNA)
nucleotides, which contain a methylene bridge connecting the 2' oxygen to the 4' carbon
47,108 (Figure 1.4).
Additionally, modifications to the linkages between sugars can be
incorporated into the 3' overhangs of siRNA to confer resistance to exonucleases. The most
common of these is the replacement of certain phosphodiester linkages with
phosphorothioate linkages (Figure 4) 109. Combinations of chemical modifications have
been shown to improve siRNA stability in plasma 22,110 as well as increase its circulation
time 4,22,102. The mechanism by which chemical modification improves circulation time in
vivo is unknown. As renal filtration is expected to be slower for large molecules (such as
intact siRNA) than for small molecules (such as fragments of degraded siRNA), an increase
in serum stability may partly explain the increase in circulation time for chemically
stabilized siRNA.
In addition to enzymes, circulating siRNAs also interact with or bind to various
components of the blood, such as red blood cells and serum proteins, which can influence
their pharmacokinetics. In some cases, interaction between serum proteins and siRNA can
improve delivery outcomes by increasing circulation time and enhancing uptake into target
tissues 111,112. For example, the incorporation of cholesterol-conjugated siRNA into serum
lipoproteins has been shown to improve its circulation time, reduce renal clearance, and
facilitate uptake into hepatocytes via lipoprotein receptors 111. It has also been found that
certain ionizable lipid nanoparticles are subject to apolipoprotein E adsorption, which can
mediate their uptake by hepatocytes via the low-density lipoprotein receptor 112. In other
cases, interaction with serum proteins or red blood cells can be detrimental, leading to
formation of aggregates that are readily opsonized or worse. Large aggregates can have
difficulty passing through capillaries, which can lead to accumulation in lung capillary beds
and even lung hemorrhage 9s. More commonly, adsorption of serum proteins, especially
components of the complement system, leads to rapid clearance of particles by the cells of
the mononuclear phagocyte system (MPS) 113,114. Phagocytosis of particles is stimulated by
the interaction of adsorbed proteins with receptors on phagocytes 113. Adsorption of these
proteins can be minimized by functionalizing the surface of particles with hydrophilic
polymers such as poly(ethylene glycol) (PEG) 113,11s,116. PEGylation of siRNA delivery
systems is a widely-used strategy for minimizing protein adsorption and reducing
nonspecific uptake to create long-circulating delivery systems 113,114.
1.2.2.2 Extravasation
21
Circulating siRNA must leave the bloodstream by crossing the vascular endothelium in
order to reach many target tissues. The structure and permeability of the endothelium vary
across different tissue types, making some tissues more accessible to macromolecular
therapeutics than others 117. In healthy tissues, endothelia are described as continuous or
discontinuous and fenestrated or non-fenestrated (Figure 1.5). In continuous endothelium,
cells are connected by
tight
junctions
and
adherens junctions that
*
allow the passage of
water and small solutes
but prevent the passage
of molecules larger than
3
Endothel
Let
nm
Macromolecules such as
addr-fer1 iuntiuit'
!Urdeut
albumin
Continuous, nonfenestrated
capillary
hetirt. lunys .Mi %Lk'trit
I
spae uf D'e -f
117
Continuous, fenestrated
capillary
iterti
1'.1d
Iy
I
~and
rtneKtae
traverse
endothelia
continuous
by transcytosis 118. In
fenestrated
endothelia,
contain
cells
channels,
transcellular
or fenestrae, that allow
direct passage of water
small solutes across
endothelial
cells.
Fenestrae in continuous
5irus .dstEC
-
uendothelia
Liver sinusoidal capillary
$N-isitated endothel idi' ..'t ,
are typically
40 - 60 nm in diameter
119 and are spanned by a
glycoprotein
thin
diaphragm that prevents
of
passage
the
macromolecules
Figure 1-5. Endoth"elia heterogeneity. Capillary endothelta can be
continuous or discontinuous. fenestrated or nonfenestrated. The structure of the endothelium affects the permeability of the vasculature to the
passage of siRNA therapeutic formulations.
117,120.
Discontinuous
found
endothelium,
liver
in
primarily
is
sinusoids,
The
diaphragm.
lack
a
that
nm
117,121)
200
(100
fenestrae
by
large
characterized
sinusoidal endothelium also lacks an organized basement membrane 117 and therefore
provides minimal resistance to the passive transport of macromolecules and nanoparticles
from the bloodstream to the liver interstitium. This, along with the well-perfused nature of
the liver, may explain why the liver is one of the primary sites of accumulation of many
systemically-administered siRNA formulations 4,81,92,122-124. Though the structure of the liver
endothelium allows access to hepatocytes, therapeutics can also be absorbed by resident
22
macrophages in the liver, called Kupffer cells 125.
In some cases, accumulation of
therapeutic particles the liver is due mainly to phagocytosis by Kupffer cells rather than
uptake by hepatocytes 125.
Another site that can have increased vascular permeability is malignant tissue. In
tumors, blood vessels can be heterogeneous in size, spacing, tortuosity, and permeability
119,126-128. Tumor endothelia can have large fenestrae 129, unusually thick or thin basement
membranes 119, and intercellular gaps that can be as wide as a few microns 130. As a result,
tumor endothelia are sometimes more permeable than those of healthy tissues 131,132,
allowing delivery systems to extravasate relatively easily. Certain tumors also have poorlyfunctioning lymphatic vessels 133,134, interstitial fluid is not efficiently drained from tumor
tissue and nanoparticles tend to accumulate in the interstitium. This accumulation, due
mostly to high vascular permeability in addition to poor lymphatic drainage, is known as
the enhanced permeation and retention (EPR) effect, which is exploited by many
nanoparticle systems for passive tumor targeting 119,135,136. The EPR effect aids larger
particles (-100 nm) to accumulate in regions of leaky vasculature in tumors; however,
smaller particles (< 60 nm) diffuse more readily through the interstitium and more deeply
penetrate tissues 119,137.
Phagocytosis can also play a major role in the clearance of siRNA delivery systems from
the blood. As a result, accumulation of siRNA in immune cells is also observed 138. In one
study, fluorescence molecular tomography imaging showed that siRNA delivered by lipidoid
nanoparticles exited the blood pool primarily into the spleen, bone marrow, and liver, with
a blood half-life of 8.1 minutes. siRNA taken up by the liver was excreted by the
hepatobiliary tract, while siRNA in the spleen and bone marrow localized to phagocytic cells
including neutrophils, macrophages, dendritic cells, and monocytes. This biodistribution
pattern was exploited to use siRNA to reduce the accumulation of monocytes in sites of
inflammation by silencing of the chemokine receptor CCR2. Knockdown of this target in
inflammatory monocytes attenuated the damaging effects caused by inflammation in
multiple disease models, including atherosclerotic plaques, myocardial infarction, diabetes,
and cancer
138.
1.2.2.3 Cell uptake
Delivery systems that extravasate and diffuse into tissues must cross cell membranes
into the cytoplasm to trigger RNAi. Most siRNA delivery formulations traverse cell
membranes through membrane-derived transport vesicles in the tightly regulated process
23
of endocytosis.
Endocytosis encompasses two broad categories: phagocytosis and
pinocytosis. While phagocytosis is only associated with specific cell subtypes, pinocytosis is
common to all cells.
Phagocytosis is used by specialized immune cells including macrophages, neutrophils,
and monocytes to clear the body of large particles such as pathogens or dead cells 139-141.
Prior to involution by phagocytosis, foreign materials are opsonized, which involves their
binding to opsonin proteins such as immunoglobulins, complement proteins, and blood
serum proteins. Complexation with these proteins marks foreign materials for phagocytosis
140-143. Rho-family GTPases together with specific cell-surface receptors and signaling
cascades determine the mode of phagocytosis 139,144. Macrophages, for example, utilize Fc
receptors to recognize antibody-bound targets. Resulting activation of Cdc42 and Rac
triggers actin assembly and creates finger-like protrusions that engulf antibody-bound
targets 140,141,145. The formation of the resulting vesicle, called a phagosome, stimulates the
cell's inflammatory responses and the entrapped target is bombarded with hydrolases,
oxygen radicals, and a low pH environment in an attempt to destroy the foreign material 146.
Macropinocytosis is a non-specific endocytic process similar to phagocytosis that occurs
in most cells. Rho-GTPase mediates actin-driven membrane protrusions that extend out and
collapse over large volumes of extracellular milieu, entrapping particles in large vesicles
called macropinosomes 147-149. These large vesicles can be up to 1-5 tm in diameter, and
their intracellular fate is uncertain, although there is evidence that they shrink and fuse
with lysosomal compartments for degradation of entrapped materials.
Clathrin-mediated endocytosis (CME), the most prolific internalization mechanism,
occurs constitutively in all tissue types and participates in the uptake of vital nutrients and
in intercellular signaling through receptor-dependent and receptor-independent pathways
148,150-153. Iron-bound transferrin protein, for example, is internalized after association with
transferrin receptors initiates invagination of the plasma membrane. Pits that form in the
membrane around protein-bound receptors are coated by clathrin, a triskelion-structured
protein that assembles into polygonal cages on the cytoplasmic face of the membrane
150,151,153. Receptor-independent CME is able to induce these internalization events simply
by non-specific ionic or hydrophobic interactions with the plasma membrane. Fission of the
coated pit from the membrane is mediated by the GTPase dynamin, forming a clathrincoated vesicle roughly 100-120 nm in size. Internalization occurs on the order of minutes,
and resulting vesicles undergo endosomal maturation. Early endosomes have a pH around
6 and later fuse with acid hydrolase-containing prelysosomal vesicles to form late
24
endosomes (pH -5). The acidic and enzymatic environment of the lysosome is aimed at
degradation of the vesicular cargo 154.
Caveolae-mediated endocytosis (CvME) is used by cells to internalize serum proteins
such as albumin and essential metabolites such as cholesterol and folic acid. CvME involves
flask-shaped invaginations, on the order of 50-100 nm, of the plasma membrane called
The dimeric protein caveolin is responsible for the shape and
caveolae 150,1S2,155,156.
structure of caveolae, whose membranes are enriched with cholesterol and sphingolipids.
CvME is a tightly regulated process that is believed to be driven by receptor-ligand
interactions, with vesicle formation being mediated by the GTPase dynamin. Caveolinassociated vesicles do not contain any enzymatic mixtures and do not enter a degradative
pathway by maturing into lysosomes like CME. The internalization kinetics of CvME are
much slower than those of CME, with an uptake half-time on the order of 20 minutes.
Recently, endocytic pathways have been uncovered that do not fall into any of the
previously-described categories 1s2,155-157. These processes are often referred to as clathrinand caveolae-independent pathways, and occur at cholesterol-rich microdomains of the
plasma membrane that are similar to caveolae. These lipid "rafts" are small structures
approximately 40-50 nm in size, and have unique membrane protein compositions that may
play a role in vesicle formation. The mechanisms that stimulate this endocytic pathway are
still poorly understood, and currently the only way to determine the contribution of this
pathway to a cargo's internalization is by elimination of the other pathways.
Some systems
In general, siRNA delivery particles enter cells via endocytosis.
incorporate targeting ligands designed to stimulate receptor-mediated endocytosis 158.
Other systems, such as lipophilic conjugates and some ionizable lipid nanoparticles, exploit
endogenous targeting ligands by associating with serum lipoproteins, which are taken up by
hepatocytes by receptor-mediated endocytosis 111. Untargeted siRNA lipoplexes are also
taken up mostly by endocytosis, but a small number of complexes enter cells through
another pathway, possibly by direct fusion of the lipoplex with the cell membrane.
Interestingly, this minor pathway was shown to be responsible for functional siRNA
delivery in certain systems, as blocking endocytosis did not abolish gene silencing 159.
Uptake of siRNA delivery systems by endocytosis leads to sequestration of the delivery
system inside intracellular vesicles. Escape from these vesicles may represent a major
barrier to triggering RNAi and thereby affect the efficacy of delivery systems. While the
mechanisms of vesicular escape of synthetic materials are still poorly-understood, polymers
25
with many protonatable groups are thought to induce lysis of endosomes by pH buffering
52,160.
Protons are transported into endosomes accompanied by counter ions, and
absorption of protons by polymers leads to increased concentrations of counter ions,
osmotic swelling, and endosomolysis. Certain ionizable lipid systems undergo a phase
transition at low pH, resulting in membrane destabilization that promotes release of
material from endosomes 92,161. By converting from a lamellar to an inverted hexagonal
phase, these lipids are thought to increase their interactions with the endosomal
membrane, leading membrane destabilization and inducing endosomal release. Still other
systems employ acid-labile materials 162 or modified cell-penetrating peptides 163.
Dominska and Dykxhoorn provide a more detailed review of how various delivery systems
overcome the problem of endosomal release 164. A recent report showed that for a leading
lipid delivery system, approximately 70% of the siRNA taken up by cells underwent
00
.
endocytic recycling and exocytosis
1.2.2.4 Elimination
siRNA that is not degraded or endocytosed is removed from circulation by excretion in
urine or bile. Renal filtration occurs in the
specialized
capillaries
of the
kidney
glomerulus (Figure 1.6). Water and small
molecules are able to pass through capillary
walls
into the urinary space, while
Gamerulmr
macromolecules remain in circulation. The
E6 %e:,ent
glomerular filtration barrier is composed of
Mem~rbrane
the
the endothelium,
three
layers:
glomerular basement membrane (GBM), and
a layer of epithelial cells called podocytes 165.
Fe-e-,t ated endotheiuf
Fgure 1-6. The glomrular filtration barrier. The
capillaries of the kidney glomerutus participate in
the filtration of blood to form urine. The glomerular
filtration barrier consists of three layers endothelial
cells, the glomnerular basement membrane, and
podocytes. Molecules smaller than -8 nm in diameter are able to pass through this barer into the
urinary space.
The glomerular endothelium is continuous
and populated with abundant 60-100 nm
166.
Endothelial cells are
fenestrations
supported by the glomerular basement
membrane, which provides an additional
Podocytes play an
diffusion barrier 167.
important role in the size selectivity of the
Slit
168.
barrier
filtration
glomerular
diaphragms, composed of proteins that
bridge the space between podocytes, create
-8 nm pores 169-171 that filter waste prior to
excretion in the urine 172,173.
26
The liver, which is also part of the body's clearance system, mediates the secretion of
compounds into bile ducts, and are eliminated from the body via the intestine 174-176.
Hepatocytes secrete bile into hepatic ducts, where it can be emptied directly into the small
intestine or stored in the gallbladder. Various studies have shown that naked, unstabilized
siRNA is rapidly eliminated from the bloodstream via renal filtration, with serum half-lives
of less than 5 min 4,101,110. Chemically-stabilized siRNAs have been reported to circulate for
30 - 50 min 4,110, and also to undergo renal clearance 4,110. While kidney filtration has often
been assumed to be the major route of siRNA elimination, a recent study demonstrated the
importance of the hepatobiliary pathway in the clearance of chemically-stabilized siRNAs,
both naked and in formulation 177.
1.2.3 Conclusions
The many interactions between siRNA and the body continue to be elucidated as more
delivery systems are studied. In the design of delivery vehicles, it is natural to focus on the
aspects of pharmacokinetics and pharmacodynamics that contribute directly to the desired
therapeutic effect, yet many other interactions exist that affect the overall potency and
tolerability of a formulation. The information contained in this review provides a brief
overview of these interactions, which can serve as a guide in design of new systems and in
the optimization of existing systems.
27
M
Table 1. Selected RNAi drugs in clinical trials (continued on next page)
Delivery
System
Drug
ALN-RSV01
TD101
Sponsor
Alnylam
Pharmaceuticals
Pachyonychia
Congenita
Project
Status
ClinicalTrials.gov
Identifier
Disease
Phase
RSV
nucleocapsid
Respiratory Syncytial
Virus Infections
2
Completed
NCT00658086
K6a (N171K
mutation)
Pachyonychia
congenita
1
Completed
NCT00716014
2
Terminated
NCT00363714
1
Active
NCT01064505
Acute Primary AngleClosure Glaucoma
2
Recruiting
NCT01965106
Kidney injury, acute
renal failure
Delayed Graft
1
Completed
NCT00554359
1,2
Active
NCT00802347
2
Active
NCT01445899
2
Completed
NCT00713518
2
Completed
NCT00306904
2
Completed
NCT00259753
Ocular Pain, Dry eye
1, 2
Recruiting
NCT01776658
ADR32
Ocular Hypertension,
Open Angle
Glaucoma
2
Completed
NCT01739244
CTGF
Cicatrix, Scar
prevention
1
Active
NCT01780077
Target gene
Age-Related Macular
AGN211745
Allergan
VEGFR1
Degeneration,
Choroidal
Neovascularization
Optic Atrophy, Nonarteritic Anterior
Quark
QPI-1007
Naked
siRNA
PF-655 (PF-
04523655)
bevasiranib
Pharmaceuticals
CASP2
Quark
PharmaceuticQuNP
als
p53
Quark
Pharmaceutica
Is
RTP8O1 (Propri
etary target)
Opko
Inc.
Health'
VEGF
Neuropathyic
Function,
Complications of
kidney transplant
Choroidal
Neovascularization,
Diabetic Retinopathy,
Diabetic Macular
Edema
Age Related Macular
Degeneration
Diabetic Macular
Edema
Macular
Degeneration
SYL1001
SYL040012
RXi-109
I
Sylentis, S.A.
___________
Sylentis, S.A.
RMi
PharmaceuticI als
TRPV1
_____________syndrome__________
1.3 Strategies for siRNA delivery
A key challenge to realizing the broad potential of siRNA-based therapeutics is the need
for safe and effective delivery methods. Unmodified siRNA is unstable in the bloodstream,
can be immunogenic, and does not readily cross membranes to enter cells 9. Therefore,
chemical modifications and/or delivery materials are required to bring siRNA to its site of
28
action without adverse effects. A broad diversity of materials is under exploration to
address the challenges of in vivo delivery. These include polymers 178, lipids 81,92, peptides
179, antibodies 180, aptamers 181,182, or small molecules 101,183 as delivery materials. Successful
systems have been developed by rational design or discovered utilizing high-throughput
screens
81,92.
Table 1 (continued). Selected RNAi drugs in
Delivery
System
Drug
Sponsor
clinical trials
Target
gene
Disease
Phase
ClinicalTrials.gov
Identifier
Status
ALN-VSPO2
Alnylam
Pharmaceuticals
KSP and
VEGF
Solid tumors
1
Completed
NCT01158079
siRNA-
MD Anderson
EphA2
Advanced
1
Cancer Center
Active
NCT01591356
EphA2-DOPC
1, 2
Recruiting
NCT01808638
1, 2
Recruiting
NCT01262235
1
Recruiting
NCT01518881
Atu027
ence
cancers
PKN3
Therapeutics AG
TKM-080301
Tekmira
Pharmaceutical
Company
Pancreatic
Ductal
Carcinoma
PLK1
Cancer
VP24,
Lipid
nanoparticle
s (LNPs)
TKM-100201
s (LNs)
Tekmira
Pharmaceutical
CopanyEbola
Company
VP35,
Zaire
Lpolymera
se
Tekmira
PRO-040201
Pharmaceutical
Ebola Virus
Infection
Hypercholestero
ApoB
lemia
1
Terminated
NCT00927459
Hypercholestero
lemia
Transthyretin
mediated
amyloidosis
1
Completed
NCT01437059
3
Active
NCT01960348
Fibrosis
1
Recruiting
Transthyretin
mediated
amyloidosis
1
Recruiting
NCT01814839
Hepatitis B
1
Recruiting
NCT01872065
Company
siRNAGaINAc
conjugate
Dynamic
Poly-
ALN-PCSO2
Alnylam
Pharmaceuticals
PCSK9
ALN-TTR02
Alnylam
Pharmaceuticals
TTR
ND-L02s0201
Nitto Denko
Corporation
ALN-TTRsc
Alnylam
Pharmaceuticals
TTR
Arrowhead
Research
conserved
regions of
Corporation
HBV
Calando
Pharmaceuticals
RRM2
Solid tumors
1
Active
Silenseed Ltd
KRAS
Pancreatic
Cancer
2
Active
ARC-520
Conjugate
Cyclodextrin
NP
LODER
polymer
CALAA-01
siG12D
LODER
HSP47
NCT01858935
I
NCT00689065
NCT01676259
Here we review a selection of promising systems for systemic siRNA delivery, with a
focus on the design of materials and the approach to the delivery problem. These systems,
all with reported in vivo efficacy, are diverse in size, shape, structure, chemistry, and
mechanism of action. This diversity reflects our still-developing understanding of the
mechanisms that underlie much of the delivery process and highlights the still vast space
for innovation and creativity in the field of siRNA delivery.
29
1.3.1 Cyclodextrinpolymer nanoparticles
.
Cyclodextrin polymer (CDP)-based nanoparticles (Figure 1.7) entered clinical trials for
siRNA delivery less than a decade after their introduction, 184. Davis and coworkers first
introduced the cyclodextrin delivery system for plasmid DNA in 1999, with re-optimization
of the system for siRNA delivery years later 185-192. This was the first targeted nanoparticle
siRNA delivery system to enter clinical trials for cancer 184
PrEGMW=5SO
AD-PEG
AD-PEG-Tf
siRNA
nn
Fgm 1-7. Cyludwdln podmmv .mnapwllms. Composition ofthe cyclodextrin delivery system developed by
the Davis group.
CDPs are polycationic oligomers (n-5) synthesized by a step-growth polymerization
between diamine-bearing cyclodextrin monomers and dimethyl suberimidate, yielding
oligomers with amidine functional groups 185. The strong basicity of these amidine groups
mediates efficient condensation of nucleic acids with the CDPs at N/P ratios as low as 3.
End-capping of the polymer termini with imidazole functional groups was shown to
facilitate endosomal escape 193, resulting in improved delivery efficacy of both plasmid DNA
and siRNA
192,194.
30
Though nanocomplexes composed of only CDP and siRNA were able to mediate efficient
delivery in vitro, these complexes required additional formulation components for
Both adamantane-PEG (AD-PEG) and
stabilization and efficacy in vivo 188,194-196.
(AD-PEG-TO
were incorporated to improve particle
adamantane-PEG-transferrin
properties in vivo 192,196,197. Adamantane (AD) is a hydrophobic molecule that forms a stable
inclusion complex with the cyclic core of the cyclodextrin structure. This noncovalent
interaction allowed for surface modification of CDP-siRNA nanoparticles with AD-modified
excipients by a chemical interaction that is orthogonal to the ionic forces that facilitate
siRNA binding. PEG shielding was necessary to prevent protein-induced aggregation in
serum, but PEGylation also reduced cellular uptake and silencing efficacy. To recover
efficacy, the protein transferrin was conjugated to the free end of AD-PEG as a targeting
agent. Inclusion of this AD-PEG-Tf conjugate enabled multivalent binding to the CD71
transferrin receptor, improving efficacy 198.
The CDP-siRNA delivery system has been evaluated in several therapeutically relevant
animal models. In a xenograft model for Ewing's Sarcoma 192, CDP nanoparticles formulated
with siRNA targeting the oncogenic EWS-FLl1 fusion gene induced gene knockdown and
when
had antiproliferative effects with no measured innate immune responses
administered intravenously 192. In a syngeneic subcutaneous mouse tumor model, the
targeted CDP delivery system showed potent silencing against the validated cancer target
ribonucleotide reductase subunit 2 (RRM2) 199. The clinical translatability of the delivery
system was evaluated in cynomolgus monkeys, which indicated that the nanoparticles can
be tolerated up to 27 mg siRNA/kg with translated efficacy in the range of 0.6-1.2 mg
siRNA/kg 200. Finally, clinical potential was established when Davis and co-workers showed
RNAi-specific gene inhibition in human melanoma patients (phase I clinical trial) by
monitoring siRNA-mediated cleavage of RRM2 mRNA 8.
Several factors contributed to the translation of this delivery system: 1) the low toxicity
of the cationic polymer, 2) its facile condensation with nucleic acids, 3) the steric
stabilization by PEGylation of the CDP-siRNA nanoparticles in a stable and non-covalent
fashion, and 4) the inclusion of a ligand to enhance in vivo uptake and efficacy. These
lessons provide useful guidelines for designing future polycationic delivery vehicles.
1.3.2 Lipid nanoparticles
The activity of liposomal siRNA formulations was first reported in non-human
primates in 200682. Since then a number of lipid nanoparticle (LNP) RNAi drugs have
entered clinical trials (Table 1), including treatments targeting hypercholesterolemia,
transthyretin mediated amyloidosis, and cancer 2 01 20 2 . Prior to their use with siRNA,
liposomes were studied for decades as delivery vectors for DNA-based drugs, due to their
31
.
ability to protect entrapped oligonucleotides from nuclease degradation and renal
clearance, and for their ability to promote cellular uptake and endosomal escape 203. To
date, a number of different lipid and lipid-like structures and formulation methods have
been developed, generating a wide variety of LNPs2O4,20s. In the study of these diverse
systems, several features have emerged as particularly effective for siRNA delivery. They
include the use of cationic or ionizable lipids, shielding lipids, cholesterol, and targeting
moieties, 206
1.3.2.1 Cationic and ionizable lipids
Many liposomes used for siRNA delivery include a cationic or ionizable lipid.
Positively charged lipids serve several functions:
improving the entrapment of the
negatively charged siRNA, increasing cellular uptake, and facilitating endosomal escape.
Several studies have determined that cationic lipids, which have a constitutive net positive
charge, have been shown to be less efficacious 20 7 and more toxic 88 than ionizable lipids,
whose charge is dependent upon the pH of the surrounding environment. As a result,
recent research has focused on the development of novel ionizable lipids. The composition
of these lipids is generally divided into three parts: an amine head group, a linker group,
and hydrophobic tails (Figure 1.8a). Many of these lipids have been synthesized in a
combinatorial manner 83 208 by altering these three sections in a systematic fashion in
attempt to develop structure-function correlations.
To minimize toxicity without sacrificing efficacy, the pKa of an ionizable lipid should
be low enough to remain unprotonated during circulation, but high enough to become
protonated in either the early or late endosome. Protonation is necessary to promote
membrane fusion and lipid mixing with the anionic lipids in the endosomal
membrane61 2 09
z ,2 1 0 . In a study of the effect of lipid pKa on in vivo gene silencing using 53
ionizable lipids, an efficacious range of pKa was found to occur between 5.4 and 7.6. Within
this range, efficacy increased as pKa approached an optimum value of 6.44211. Overall
liposomal pKa appears to be more important for efficacy than the pKa of individual lipid
components. When two lipids of pKas 5.64 and 6.93 were co-formulated, the mixed
liposomes had an overall pKa 6.93 and exhibited four-fold greater in vivo silencing than
liposomes formulated with either lipid alone 211. It should be noted that though there is
strong correlation between pKa and efficacy, particles with comparable pKas can have
disparate levels of efficacy in ViVo92,209.
Lipid transition temperature refers to the temperature at which lipid membranes
shift from the more stable lamellar phase to the less-stable hexagonal phase configuration
(Figure 1.8d) 2 12. This transition promotes destabilization of the endosomal membrane and
release of siRNA from both nanoparticles and endosomes 213. Lipids with lower transition
temperatures, which more readily shift from lamellar to hexagonal phase 214 to promote
32
endosomal release, have a small polar head groups and large, unsaturated hydrophobic tails
(Figure 1.8b). Lipids with large polar head groups and fully saturated hydrophobic tails are
more likely to adopt the stable lamellar phase configuration. Successful lipid formulations
are engineered to remain in the lamellar phase during circulation and to transition to the
hexagonal phase within endosomal compartments.
Amine Head Group
a
0
0
Linker Group
b
a
0
c
Conical
Hydrophobic Tails
I
I
I
I
II
I
Cylindrical
DLinDMA
Liposomal
d
membrane
DLin-KC2-DMA
En dosomal
m embrane
Lamellar Phase
Hexagonal Phase
Figure 1-8. Lipid structures and shapes. a) Ionizable lipids are composed of three sections, the amine head
group, the linker group, and the hydrophobic tails. b) Lipids with a large head groups and tails composed of
saturated hydrocarbons tend to adopt a cylindrical structure, while lipids with small head groups and unsaturated
tails tend to adopt a conical structure. c) The structures of siRNA delivery lipids DLinDMA and DLin-KC2-DMA. d)
The mixing of cationic (orange) and anionic (blue) lipids promotes the transition from the more stable lamellar
phase to the less stable hexagonal phase, facilitating fusion of the liposomal and endosomal membranes.
The structure of a lipid's hydrophobic tail affects its pKa, transition temperature,
and potency.
In a study of lipids with identical head groups, linkers, and tail lengths,
33
.
increased unsaturation of hydrophobic tails resulted in decreased pKa, decreased transition
temperature, and increased in in vivo efficacy. The most efficacious of the lipids in this
study, DLinDMA (Figure 1.8c) was subsequently tested in cynomolgus monkeys where it
was shown to be capable of silencing 90% of mRNA in hepatocytes 8 2
Variations on the amine head group and linker portions of lipids also affect in vivo
efficacy. For the lipid DLinDMA (Figure 1.8c), replacing the ether linker with an ester group
reduced efficacy, while a ketal linker improved efficacy 9 2. The amine head group was
modified by altering the length of the carbon chain connecting the amine to the ketal group.
As the chain length increased, the pKa of the formulation increased and the transition
temperature remained constant. Maximal efficacy was obtained using DLin-KC2-DMA
(Figure 1.8c), which had a pKa of 6.792.
.
While many of the ionizable structures examined are two-tailed species with a
single amine, it has also been demonstrated that multi-tailed species with numerous amines
can be highly efficacious. Several combinatorial libraries have been generated using multi83 2 08
amine head groups bound to numerous hydrophobic tails of various lengths 8 1.
1.3.2.2 Shielding lipid
Lipid-anchored PEG is a common component in liposomes. PEG groups serve many
purposes: reducing particle size 21 s.2 1 6, preventing aggregation during storage, increasing
circulation time, and reducing uptake by unintended targets such as red blood cells and
macrophages 2 16,217. Shielding lipids can also reduce cellular uptake by target cells, and have
been shown to reduce efficacy in vitro and in vivo 2 1 6. After endocytosis, PEG can sterically
and electrostatically block the interaction between the liposome and endosomal membrane,
hindering membrane fusion and preventing endosomal release.
One strategy for improving the efficacy of PEGylated nanoparticles involves
incorporation of acid-sensitive bonds connecting PEG to the liposome. In a comparison of
stable carbamate linkers and oxime linkers designed to degrade at pH 5.5, liposomes with
oxime-linked PEG were stable at pH 7.4 but showed improved release of siRNA at pH 5.5 in
vitro. Oxime-linked PEGylated liposomes also demonstrated improved gene silencing in
2 17
.
ViV0
Another method to reduce the negative effects of shielding components involves the
use of a pH-sensitive modified PEG that binds to liposomes through ionic interactions. The
liposomal core consists of a HEMA-lysine-modified cholesterol as well as DOPE, and the PEG
is covalently modified with HEMA-histidine-methacrylic acid. At neutral pH the PEG
copolymer has a net negative charge while the liposomal core has a net positive charge. In
the endosome, imidazole and methacrylic acid residues become protonated, and the net
charge of the PEG becomes positive. This results in the release of PEG from the lipid core,
34
.
exposing the positively charged liposomal membrane and allowing it to fuse with the
endosome 21 8
1.3.2.3 Cholesterol
Many liposomal formulations include cholesterol, which can associate with lipid
bilayers 219 . Up to 25% cholesterol, increasing cholesterol content lowers the transition
temperature of liposomal membranes containing conical-shaped lipids, facilitating their
conversion from lamellar to hexagonal phase 220. Cationic liposomes with less than 10%
cholesterol released less than 5% of entrapped drug within two hours when mixed with
anionic liposomes, but liposomes with 17% cholesterol released more than 90% of
2 13
.
entrapped drug within 5 minutes
1.3.2.4 Targeting Ligands
To improve the biodistribution of liposomes, many formulations employ
endogenous or exogenous targeting ligands. An endogenous targeting ligand is a molecule,
often a serum protein, which binds to the liposome during circulation and directs the
particle to the ligand's natural cellular target. Exogenous targeting ligands are added to
liposomal formulations before injection to bind desired surface proteins on target cells.
Advancements have been made in developing and understanding both forms of delivery,
thereby improving control of the biodistribution of liposomes in vivo.
.
The lipoprotein ApoE has been utilized as an endogenous targeting ligand by DLinKC2-DMA based ionizable liposomes. In wild type mice, the formulation silenced greater
than 90% of factor VII serum protein with a dose of 0.2 mg siRNA/kg. However, in ApoE
knockout mice the same formulation achieved less than 20% silencing at the same dose.
Silencing efficacy was restored by incubation of liposomes with recombinant ApoE prior to
injection into ApoE knockout mice' 1 2
Retinol binding protein (RBP) is also used as an endogenous targeting ligand. This
serum protein binds vitamin A and transports it to cells expressing the RBP receptor,
including hepatic stellate and pancreatic stellate cells as well as stellate cells that have
become proliferative, fibrogenic, and contractile myofibroblasts 22 ,222. Delivery of RBPmodified nanoparticles to stellate cells has been reported to have efficacy that is five-fold
higher in cirrhotic rats as compared to normal rats. These liposomes have been reported to
reduce collagen production, thereby reducing liver and pancreatic fibrosis and improving
222
.
survival rates in rats 2 2'
The use of exogenous ligands has also been examined as a means of targeting
distribution and improving efficacy in vivo. The small molecule N-acetylgalactosamine
(GalNAc) has been studied as a ligand to target liposomes to the liver, due to its ability to
35
.
bind to the asialoglycoprotein receptor on the surface of hepatocytes11 2, while folate has
been used to target delivery to rapidly dividing cancer cells 2 2 3 - 2 2 5 . Exogenous ligands are
generally attached to the distal end of the lipid anchored PEG groups 12 ,2 2 3 - 2 2 5
.
-
The LNP drug ALN-VSP, under development at Alnylam Pharmaceuticals, is an
example of a lipid delivery system that achieves success by careful incorporation of each of
these components. The particles contain the ionizable lipids DLin-DMA and DPPC, the PEGlipid MPEG200-C-DMA, cholesterol, and two siRNAs targeting two genes - vascular
endothelial growth factor and kinesin spindle protein 226,227. The resulting particles are 80
100 nm in diameter, have essentially no surface charge at physiological pH, and accumulate
in the liver and spleen in preclinical models 226. ALN-VSP was recently evaluated in Phase I
clinical trials for the treatment of advanced solid tumors with liver involvement. In this
dose-escalation study the drug was generally well tolerated up to 1.0 mg/kg and
demonstrated antitumor activity, including complete response in one patient with
endometrial cancer. RNAi mechanism of action was confirmed by the detection of mRNA
cleavage products in tissue biopsies 226
1.3.3 Conjugate delivery systems
A number of promising systems have been developed by directly conjugating
delivery material to the siRNA cargo.
This approach leads to well-defined, singlecomponent systems that employ only equimolar amounts of delivery material and siRNA.
The first conjugate delivery systems to show efficacy in vivo consisted of siRNA conjugated
to cholesterol 101 and other lipophilic molecules 111. Other conjugate delivery systems have
been developed by attaching siRNA to polymers, peptides, antibodies, aptamers, and small
molecules
228.
This review will discuss in detail two of the most clinically advanced conjugate
platforms, Dynamic Polyconjugates and N-acetylgalactosamine (GalNAc) conjugates, as well
as one newly developed system based on oligonucleotide nanoparticles. The first two
systems are among the most developed conjugate delivery systems and are both drug
candidates. However this work is not well represented in the scientific literature because
they are under development by private companies as proprietary technologies. Though not
all widely discussed, these systems provide useful lessons on the design of conjugate
materials that may benefit the design of future systems.
1.3.3.1 Dynamic PolyConjugates
In 2007, Rozema and coworkers reported the development of siRNA-polymer
conjugate delivery systems designed to respond to intracellular environments, which they
termed Dynamic PolyConjugates 229. These conjugates incorporate several components,
each intended to play a particular role in the delivery process (Figure 1.9). The siRNA cargo
36
is attached to a membrane-disrupting polymer via a hydrolysable disulfide linker. The
activity of the polymer is masked by PEG side chains while the system is in circulation.
Ligands are incorporated to induce uptake by target cells via receptor-mediated
endocytosis. The PEG is designed to be shed in the acidic environment inside the
endosome, exposing the membrane-active polymer and triggering endosomal release. The
disulfide linkage is cleaved in the reducing environment of the cytosol, releasing the siRNA
from the delivery polymer. The siRNA itself is chemically modified to improve nuclease
stability and to reduce off-target effects.
?M_
BF
h
rrrl
0
011N H'NH
H3
H
1
01
**'
'NH3
+H3
Nll 'NHI
+
0
o
IQ
+ 3
_0
0
Is
VW4
0
SH
wlAV
AQM0
Cytoplasm
1X
II1
52 1
If
Endosome
HLHNINH
I
+
0
0
0
+
+
00""
'S
)OWMllr'
Figure 1-9. Dynamic PolyConjugates. Dynamic PolyConjugate materials are designed to respond to the acidic environment of
the endosome and to the reducing environment of the cytoplasm. In circulation, the membrane-disrupting PBAVE polymer
(black) is shielded by polyethylene glycol (PEG). After cell uptake, the PEG chains are shed as the pH of the endosome lowers,
exposing the polymer and causing endosomal release. In the cytoplasm, the disulfide bond linking the siRNA to the polymer is
reduced, freeing siRNA to trigger RNAi. N-Acetylgalactosamine (GaINAc) is included in the formulation as a targeting ligand to
facilitate uptake by hepatocytes.
The membrane-active polymer, poly(butyl amino vinyl ether) or PBAVE (Figure 19), has amphipathic side chains that include alkyl groups interspersed with amines that are
reversibly linked to the PEG shielding agent and N-acetylgalactosamine (GalNAc) targeting
ligands. The alkyl chains are important for membrane activity, as longer alkyl chains
(propyl or butyl) were shown to improve the polymer's ability to lyse liposomes in solution
230.
PEG and targeting ligands are reversibly linked to the polymer backbone using
carboxylated dimethyl maleic acid (CDM) chemistry (Figure 1-9), which allows for the
release of the PEG shielding agent in the acidic environment of the endosome.
37
The Dynamic PolyConjugate system was effective at silencing two different genes in
the liver when administered intravenously, apolipoprotein B (Apo-B) and peroxisome
proliferator-activated receptor alpha (ppara). PolyConjugates were designed to target the
liver by incorporation of GalNAc ligands, which bind to the asialoglycoprotein receptor
(ASGPR) on hepatocytes.
Silencing of Apo-B was dose-dependent and produced the
expected phenotypic effects, including reduction in serum cholesterol 229.
The importance of individual components of the PolyConjugate was explored,
revealing some structure-function insight. The attached GaINAc ligand was essential for
both uptake by hepatocytes and in vivo silencing activity. Replacement of GalNAc with
glucose greatly reduced hepatocellular uptake, and replacement with mannose directed
uptake to nonparenchymal liver cells that express mannose receptors.
In primary
hepatocytes, it was reported that the PEG shielding moiety must be linked to the polymer
via a reversible linkage, as attachment with a nonhydrolyzable linkage abolished silencing
activity 229.
Newer generations of Dynamic PolyConjugates are under development at
Arrowhead Research Corporation 231. The original PBAVE polymers were synthesized by
uncontrolled polymerization, resulting in heterogeneity in size and composition. Newer
generation DPC polymers are synthesized using controlled radical polymerizations,
including atom transfer radical polymerization (ATRP) and reversible additionfragmentation chain transfer (RAFT) to produce homogenous polymers that are more
amenable to optimization 231. Hydrolyzable bonds are incorporated into different positions,
including the polymer backbone and side chains, a strategy aimed at reducing toxicity. The
company also reports the development of longer-circulating DPCs by employing more
stable bonds between the membrane-active polymer and the PEG shielding agent 231. The
longer circulation time is intended to facilitate the targeting of tissues other than the liver.
To this end, the use of other classes of targeting ligands have been explored, including
peptides, antibodies, small molecules, glycans, lectins, and nucleic acids. Latest generation
DPCs have been reported to induce 99% knockdown of liver genes after a single 0.2 mg/kg
dose in nonhuman primates, with the effect lasting nearly 7 weeks 231.
A 2012 report on DPC showed that the masked PBAVE polymer could be co-injected
with cholesterol-siRNA to induce gene silencing in the liver 232. Though the polymer was
not covalently attached to the siRNA in this case, both were targeted to the hepatocytes and
colocalized in endosomes, though they did not interact with each other in solution. It was
demonstrated that the GalNAc ligand, ASGPR, and the cholesterol moiety were all essential
for gene silencing. Interestingly, LDL and LDLR were not required for silencing, though they
are required for silencing by cholesterol-siRNA alone 111.
This co-injection strategy is reported to be used by the Arrowhead clinical candidate
for the treatment of Hepatitis B (HBV), ARC-520 233. This drug contains two cholesterolsiRNAs targeting conserved regions of HBV transcripts. The original PBAVE polymer has
38
been replaced with a melittin-like peptide with similar reversibly-masked endosomolytic
properties 234,235. The drug is currently in Phase I clinical trials (Table 1).
1.3.3.2 Triantennary GaINAc-siRNA
A liver-targeted siRNA conjugate composed simply of chemically stabilized siRNA
with a trivalent targeting ligand has shown promise in the treatment of several diseases.
Under development at Alnylam Pharmaceuticals, the conjugates ALN-TTRsc, ALN-PCS, and
ALN-AT3 are being studied for the treatment of transthyretin amyloidosis,
hypercholesterolemia, and hemophilia, respectively. In this system the 3' terminus of the
siRNA sense strand is attached to three GalNAc molecules via a triantennary spacer (Figure
1-10).
HO
H
H
OH
0
X FF fNOH
OH
HAc
00
0'N
N
0
*-
NA_
0
OH
0
NHAC
Figure 1-10. N-Acetylgalactosamine (GaINAc)-siRNA conjugates. Structure of the triantennary GaINAcsiRNA conjugate used in several drug candidates from Alnylam Pharmaceuticals.
The structure of this delivery material is designed for high-affinity binding to its
target, ASGPR on hepatocytes. Variations of this triantennary ligand have previously been
studied for the purposes of targeting drugs or liposomes to the liver 236,237, and certain
trends relating geometry to binding affinity and cell uptake are documented. Multivalency
of the sugar ligand greatly improves cell uptake 238,239 and spacing of the sugar moieties also
plays a role. In a study of triantennary galactose ligands, binding affinity increased with
spacer length over a range of 4 - 20 angstroms 236.
Alnylam's GalNAc conjugate similarly employs a triantennary GalNAc ligand with
20-Angstrom spacing that binds to the ASGPR with high affinity (Kd = 2.48 nM 240). A
comparison of subcutaneous and intravenous administration of this conjugate revealed
both greater accumulation of siRNA in the liver and improved knockdown of the target gene
by subcutaneous administration 240. The three drug candidates based on this conjugate are
therefore administered subcutaneously.
ALN-TTRsc, designed to silence transthyretin (TTR) for the treatment of TTRmediated amyloidosis, is the most clinically advanced of Alnylam's GalNAc conjugates. In
nonhuman primates, ALN-TTRsc administered subcutaneously reduced circulating TTR
protein by 70% after one week of daily dosing at 2.5 mg/kg 6. This level of TTR expression
39
was maintained by weekly administration of the same dose, and serum TTR gradually
returned to pre-dosing levels upon cessation of treatment. Circulating TTR mRNA levels
decreased concomitantly. No evidence of cytokine induction, complement activation, or
other adverse effects was observed at a dose of 300 mg/kg, indicating a wide therapeutic
window. The expected therapeutic phenotype, the reduction of TTR deposits in peripheral
tissues, was confirmed in mice by histology 6. A Phase IlIl clinical trial of ALN-TTRsc
(Revusiran) was initiated in late 2014 (Tablel).
Two other drugs are under investigation using the same GaINAc targeting ligand to
deliver siRNA to hepatocytes. By changing the siRNA sequence, this conjugate has been
used to silence the expression of two circulating proteins: PCSK9, which plays a role in
hepatocellular uptake of low-density lipoprotein, and antithrombin, which regulates
thrombin and plays a role in blood clotting. ALN-PCSsc, a conjugate targeting PCSK9 for the
treatment of hypercholesterolemia, showed dose-dependent silencing of the target gene in
humanized mice, with an EC5O of 0.3 mg/kg 6. ALN-AT3 targets antithrombin (AT) for the
treatment of hemophilia and rare bleeding disorders. A single dose of 1.0 mg/kg ALN-AT3
reduced serum AT protein levels in nonhuman primates by 50% and weekly doses as low as
0.5 mg/kg stably reduced serum AT levels by 75-80%. The expected phenotype, increased
in serum thrombin levels, was observed to occur in a dose-dependent manner 6.
1.3.3.3 OligonucleotideNanoparticles
The construction of 3D nanoparticles of defined composition from nucleic acids has
generated significant interests, due to its unique ability to produce a population of
molecularly identical nanoparticles with strictly-defined characteristics 241. This approach
has been adapted to deliver siRNA molecules 2 4 2 ,2 4 3 . Oligonucleotide nanoparticles (ONPs)
were composed of complementary DNA fragments designed to hybridize into predefined
three-dimensional structures.
Lee and coworkers 242 modified a previously-described
method 244 of constructing DNA tetrahedra by incorporating single-stranded overhangs on
each edge. SiRNAs were modified by extension of the 3' sense strands with DNA overhangs
that enabled hybridization to the edges of the tetrahedra (Figure 1.11). By using unique
overhang sequences, six siRNA strands could be attached to each particle, each in a specified
position. The resulting particles had a hydrodynamic diameter of about 29 nanometers.
ONPs modified with folate ligands were used to study the minimum number of
targeting ligands required for delivery and to probe the optimal arrangement of these
ligands, questions that are difficult or impossible to address using many other nanoparticle
systems. As the position of each siRNA on the tetrahedron could be controlled, siRNAs with
or without targeting ligand were assembled into the particles to achieve the desired number
and position of folate ligands. A minimum of three folate ligands was required to achieve
significant gene silencing. Incorporation of more than three ligands did not drastically
improve silencing efficiency. In addition, the positioning of the three ligands was critical.
ONPs with three ligands arranged to maximize local density (all three ligands arranged
40
.
around one side or one vertex) showed efficient silencing. Arranging ligands to be distant
from one another abolished silencing activity. Interestingly, ligand positioning did not
affect the rate of cellular uptake - it was hypothesized that the arrangement of these ligands
may affect intracellular trafficking of particles 242
Six single-stranded oligonucleotides
One-step
self-assembly
N/v\a
AW _i
'
tP
l
Six folate-modified siRNAs with 3'overhangs
Figure 1-11. Self-assembly of oligonucleotide nanopartictes. DNA tetrahedra carrying six siRNAs were synthesized in a
single step through hybridization of complementary strands. Positioning of each siRNA on the surface of the tetrahedron
could be controlled by using unique sequences in each of the siRNA 3'overhangs. This precise synthesis method enabled
the study of structure-function relationships such as number and spatial orientation of folate targeting ligands.
Folate-ONPs were evaluated for both biodistribution and gene silencing ability in
After intravenous injection,
mice with folate receptor-expressing tumor xenografts.
particles accumulated in both the tumor and kidneys, a pattern consistent with the
expression profile of the high affinity folate receptor 245. At a dose of 2.5 mg/kg, folate-ONPs
without significant
silenced
luciferase expression in the tumor by -60%
immunostimulation. The DNA particle was required for this efficacy, as free folate-siRNA
demonstrated no significant silencing in this model.
ONPs are set apart from other nanoparticle systems by the exceptional control over
particle structure that is afforded by DNA self-assembly. Particle size, shape, and surface
chemistry are known to influence many aspects of performance 246, but the heterogeneity of
ONPs
many systems precludes the study of specific structure-function relationships.
provide a platform to gain structure-function information that can benefit many delivery
technologies.
1.3.4 Conclusions
41
The delivery systems that have shown efficacy in vivo exhibit great diversity in
structure, size, chemistry, and overall approach to the delivery problem. The delivery
systems discussed here are all highly efficacious, promising drug candidates, yet they are
each unique in many aspects of their designs. Some have precisely-defined structures while
others are heterogeneous. Their sizes range from hundreds of nanometers to about the size
of a single siRNA. Some are held together firmly by covalent bonds or precise hydrogen
bonding; others are associated by hydrophobic or ionic interactions.
Nanoparticles
composed of synthetic lipids are among the most potent formulations in the developmental
pipeline. Conjugate systems enjoy precisely defined molecular structures with minimal
amounts of delivery material, apparently broad therapeutic windows, and their efficacy
continues to improve.
Though a variety of delivery systems has been developed in the laboratory,
challenges remain in bringing the full potential of RNAi to the clinic. The most advanced
systems are nanoparticles formed by mixing and self-assembly of various components. This
formulation method presents additional challenges in the scale-up of the manufacturing
process. Tightly-controlled mixing processes are required to achieve consistent quality of
the drug product 2 47 . Microfluidic methods are already in use to improve the quality and
reproducibility of LNP formulations 21s, 248. Currently, the most clinically advanced systems
target siRNA to well-perfused tissues such as the liver where fenestrated or discontinuous
endothelium allows the passage of macromolecules to target tissues. Delivery to less
accessible tissues remains a major challenge. Safe and effective delivery of siRNA to the
many tissues where it has therapeutic potential will likely require development of a variety
of delivery systems, each optimized to address the specific challenges of delivery to a
particular tissue.
As our mechanistic understanding of the delivery process remains incomplete, we
have only some guidelines about optimal delivery material characteristics. Nanoparticulate
delivery systems aim for particle sizes of 20 - 200 nm, large enough to avoid renal filtration
but small enough to evade phagocytic clearance. Shielding agents such as PEG have proven
valuable in preventing nonspecific interactions and avoiding immune recognition in
circulation. Chemical modification of siRNA is usually employed to improve nuclease
stability and reduce immunostimulation.
Exploitation of endogenous or exogenous
targeting ligands is often beneficial to improve uptake by target cells. Membrane-disrupting
materials that are shielded or neutralized in circulation but become active in the endosome
have proven efficacious. Yet even these established guidelines are not set in stone. For
example, it is assumed that siRNA must escape from endosomes in order to trigger RNAi,
yet we observe efficient silencing from delivery vehicles that incorporate no moiety aimed
at achieving this, such as GalNAc conjugates. Such delivery systems reach the RNAi
machinery via unknown avenues - they may follow undiscovered intracellular trafficking
pathways, they may escape from endosomes by an unidentified mechanism, or they may
enter cells in an unknown non-endocytic manner. Continued research into diverse delivery
platforms will help to elucidate the biological phenomena that are currently unclear and
more guidelines will emerge to aid in the design of delivery materials. Today, important
42
pieces of the delivery process remain unexplained, and there is much room for creativity
and innovation in the design of siRNA delivery materials.
1.4 Thesis goals
The goal of this thesis is to build the capacity to study siRNA conjugate delivery systems
using combinatorial methods. With the recent advancement of certain conjugate platforms
to clinical trials, there is great interest in developing novel conjugate delivery systems to
However, compared to an
target various tissues and treat various disease states.
materials,
relatively few new
abundance of lipid and polymer nanoparticle delivery
conjugate delivery materials are reported in the literature. The vast diversity of effective
lipid and polymer nanoparticle systems has been developed by the synergy of two
approaches: rational design and combinatorial screening. By contrast, the development
conjugate delivery systems has relied on rational design alone.
Combinatorial library screens have the potential to open new avenues in siRNA
conjugate delivery, but conjugate libraries have not been studied due to the many technical
hurdles to their synthesis. This thesis aims to overcome these technical hurdles to develop
a comprehensive strategy for creating siRNA conjugate libraries that can incorporate vast
diversity in chemical structures.
In Chapter 2 of this thesis, chemical strategies are developed for library synthesis and
siRNA conjugation that are shown to be robust and amenable to high-throughput synthesis.
In Chapter 3 this chemical strategy is used to synthesize a library of over 500 siRNA
conjugates using a semi-automated process. In Chapter 4 various applications of these
materials are explored. Chapter 5 presents strategies for conjugating multiple copies of
these materials to each siRNA duplex.
43
Chapter 2: Sequence-defined oligomeric materials for synthesis
of nucleic acid conjugate libraries
2.1 Introduction
RNA interference-based drugs represent a promising new class of therapeutics with
the potential to treat a wide range of human diseases. The major barrier to clinical
translation of these drugs is the safe and efficient delivery of nucleic acid to target cells. A
wide variety of delivery materials has been studied for more than a decade. Some of these
materials have advanced to clinical trials and begun to fulfill the promise of RNAi
therapeutics. Yet there is still a need for the discovery of novel materials to target different
tissues, increase efficacy, and provide new mechanistic insight into the delivery process.
Some of the most potent RNA delivery systems have been developed using a
combinatorial chemistry approach, in which large libraries of novel materials are
synthesized rapidly and then are screened for their ability to complex with and functionally
deliver RNA. This high-throughput approach has led to the discovery of a number of
efficacious lipid-like materials that would not likely have been developed by relying on
rational design alone.
Furthermore, the structure-function information gained from
studying libraries can help to guide the rational design of new generations of materials.
Conjugate delivery systems, in which delivery material is covalently attached to RNA
strands, have been of interest recently. These single-component systems require only an
equimolar ratio of delivery material to RNA, and the most advanced clinical candidates are
reported to have very broad therapeutic windows 240 . A major limitation to the discovery of
new conjugate delivery materials is that they are less amenable to the combinatorial
chemistry and high-throughput screening approach that has led to the discovery of many
potent nanoparticle formulations. Conjugate systems are usually designed and synthesized
individually, making the exploration novel materials laborious.
Here we report a synthetic strategy for the high-throughput synthesis of siRNA
conjugate libraries.
We develop a general method of synthesizing sequence-defined
oligomeric materials using a hydroxyproline backbone and a fluorous purification tag. By
modifying the carboxylic acid of hydroxyproline, we create a diverse set of monomer units
that incorporate various functionalities of interest for siRNA delivery. The hydroxyprolinebased monomers are linked together sequentially to form oligocarbamates. The same
chemistry is used to incorporate functionalities for siRNA attachment. Our method enables
44
the exploration of vast chemical space in well-defined materials with high conversion and
high purity. Mild reaction conditions and simple purification methods make this strategy
readily scalable to multiwell plate format for high-throughput synthesis.
In this chapter, we first describe several alternative chemical strategies for the
synthesis of synthetic oligomers that were explored and found to be insufficiently robust.
The insight gained from studying these potential strategies led us to develop the very
This method is then thoroughly
robust hydroxyproline-based synthetic method.
characterized to demonstrate its applicability to high-throughput synthesis of siRNA
conjugates.
2.2: Potential chemical strategies for synthesizingconjugate libraries.
2.2.1 Introduction
The aim of this project is to develop a systematic method of making libraries of siRNA
conjugate delivery materials. These materials should be able to incorporate a wide variety
chemical structures and properties, including particular functionalities that have been
observed to promote delivery in other systems. For example, in lipid delivery systems
protonatable nitrogen groups, such as dimethylamino groups, are associated with several
effective delivery systems 92,249. Alkyl groups are also a major component of these systems.
PEG chains are regularly incorporated to improve in vivo performance11 3 114 . A useful
synthetic strategy would enable the researcher to incorporate such structures at will and
combine these structures to yield predictable products with high purity.
Our general strategy for creating such libraries is modeled on peptide chains.
Peptides, formed from combinations of amino acid building blocks, can have enormous
structural diversity created from a limited number of starting monomers. Each monomer
contains the same functionalities necessary for linking monomers together into peptide
chains, along with a variable side chain that introduces structural diversity. We envisioned
a library of synthetic peptide-like molecules, where monomer side chains are chosen by the
researcher.
45
In creating a set of peptide-inspired synthetic oligomers, there are many potential
avenues ranging from naturally occurring peptides to completely synthetic materials.
Naturally occurring cell penetrating peptides have been studied in various delivery
applications 250 . These have the advantages of being non-immunogenic, endogenous
sequences, but they are limited in number and allow only the incorporation of natural
amino acids. Attachment of natural peptides to siRNA can be challenging. An alternative
strategy could pursue the combination of synthetic amino acids, forming a peptide
backbone with synthetic side chains. This allows more room for the incorporation of
structures of choice, including functionalities for siRNA attachment. Still, peptide synthesis
requires multiple protection and deprotection steps, which adds complexity to highthroughput synthesis. We hypothesized that we could create a set of entirely synthetic
monomer units that would allow the incorporation of non-natural side chains and that
could be linked together without the need for protection and deprotection steps.
A synthetic strategy that is scalable and amenable to high-throughput synthesis
must meet several requirements. First, the chemical reactions for linking monomers
together in to dimers, trimers, and longer chains must proceed smoothly, with high
conversion and minimal side reactions, regardless of the properties of the individual
monomers being incorporated. Though monomer side chains may be very hydrophobic,
very hydrophilic, charged, neutral, large, or small, coupling chemistry must be robust for
any combination. Second, a general purification strategy is needed to separate oligomer
chains from excess reagents. Purification must also be effective regardless of the properties
of each oligomer. Third, functionalities for siRNA attachment must be incorporated, such as
azide or cyclooctyne groups. Fourth, a general strategy for conjugating the materials to
siRNA and purifying the conjugates is needed. Finally, all of theses steps must be amenable
to 96-well plate format, and should be mild and robust enough to be largely carried out by a
liquid-handling robot.
In this section, we discuss various strategies that were explored for achieving this
aim. We discuss the limitations of various chemistries and the observations that led us to
develop a strategy that is robust and reliable.
2.2.2 Degradable Ester-linked Oligomers
46
a.
0
3
OH
HO
HN
NH 2
0ZNH
0
R1
b.
0
R ,0OH
HN, Rl
O
Cl H
TMS
0
0
Cl
HO
0
DMF
HN, R
R2
2
NH
0
0
R
0
O
OH
R
HN, Rl
O
Cl
0
0
Cl
HO
0
DMF
HN, R
0
HN'
R..
TMS
0
O
3
0
0
HN R
OH
0
HN
RR 3
Figure 2.1 Degradable ester-linked oligomers. a.) Monomers were
synthesized from amine-substituted butyrolactone.
b.) Monomers were coupled together using oxalyl chloride. Recyclization of the lactone monomers reduced the efficiency of monomer
coupling.
Biodegradability
of delivery materials is
often desirable as it is
observed to reduce the
toxicity
of
some
251
One
.
systems
for
strategy
synthesizing degradable
materials is by linking a
hydroxyl group to a
carboxylic acid to form
ester linkages.
Esters
are hydrolyzed over
in
aqueous
time
environments.
To
create
degradable
oligomers,
we
synthesized
model
monomers from amino
butyrolactone
(Figure
2.1). The amine could
be readily modified with
the desired side chain
by amidation. Opening
the lactone ring made
acid
and
hydroxyl
groups
available for
monomers
linking
together into chains.
These
lactonebased monomers were
linked together using
oxalyl chloride (Figure
2.1).
In
pilot
experiments, formation
from
dimer
of
monomers
was
detectable by LC-MS
analysis,
but
was
inefficient. Reclosure of
the lactone ring was highly favored over formation of dimer, even when protecting groups
47
were added to the free hydroxyl (Figure 2.1). The stability of the five-membered lactone
ring relative to the linear ester may make these oligomers too unstable to be usable. For
this reason, we moved to a synthetic strategy that prevented ring closure, as described in
the next section.
2.2.3 Oligocarbamates based on 2-amino 4-hydroxybutyricacid
To improve the stability of oligomer chains, we explored replacing the ester linkages
with more stable carbamates. Using the same monomer backbone structure, we modified
the acid moiety with our side chain of choice, leaving the amine free for linking monomers
together (Figure 2.2). Using a starting monomer with a protected amine, the free hydroxyl
was activated with excess 1,1-carbonyldiimidazole (CDI). After quenching of excess reagent
and thorough drying of the intermediate, the second monomer was added in excess,
yielding the carbamate dimer.
In pilot experiments, the formation of dimer from monomer proceeded with about
40% conversion when left overnight. Adjusting reaction conditions including temperature,
solvent, and amount of excess amine, and addition of 4-(dimethylamino)pyridine (DMAP)
catalyst did not reliably improve the yield of product. The reaction of the starting monomer
with CDI to form the imidazole-N-carboxylic ester intermediate proceeded smoothly,
suggesting that the amine of the second monomer was limiting the progress of the reaction.
We hypothesized that a less hindered amine might react more efficiently with the
imidazolide intermediate. To test this, we coupled a starting monomer to a model amine
compound (amine-PEG, Figure 2.2c). This reaction proceeded smoothly for a variety of
reaction conditions and consistently gave complete conversion of starting monomer to
carbamate product, according to LC-MS analysis. Based on these results, we moved to a
new monomer backbone structure with a less hindered amine.
48
a.
PG
NH2
PGNH
NH
PG NH
OH
NH
O
0
CO
0=
NH 2
ROH
0
ROH
0
b.
c.
R1
NH
0
Ri
PG OH
H
PG
OH
H
R2
0
0
N#N AN'
OH
H2NX
R
NH
0
N'N KN I
H 2N
in
R2
NH
0
0
PG
OH
ON
N
PGH
H
H
flO
n
N,
40%
100%
0
N'N
C)
NH
OH
H-jN I.
RI
PG
N
H
N
from amine-substituted butyrolactone. b.) Monomers
FR?2
Ri
NH
0
Figure 2.2 Oligocarbamates based on 2-amino
4-hydroxybutyric acid. a.) Monomers were synthesized
N
0
N
H
O
N
H
oH
were coupled together using carbonyldiimidazole (CDI)
with limited efficiency. c.) The same reaction proceeded
efficiently when a model amine was used, suggesting
that changing the structure of the monomer backbone
may improve coupling efficiency.
2.2.4 Oligocarbamates based on 4-amino 3-hydroxybutyricacid
We next explored 4-amino 3-hydroxybutyric acid as a monomer backbone structure
(Figure 2.3) because the amine in this molecule is less hindered than in the previous
backbone structure and our previous results suggested that this change could improve the
reactivity of the amine.
As before, we modified the acid moiety of the backbone with the desired side chain
by amidation. In the first monomer of the chain, the amine was protected leaving the
49
alcohol to react with CDI. In pilot studies, formation of the imidazole-N-carboxylic ester
intermediate proceeded smoothly, and carbamate was formed efficiently upon addition of
the second monomer. Conversion of monomer to dimer was consistently over 90% based
on LC-MS analysis. These pilot studies were done using monomers whose side chains were
uncharged and hydrophobic, such as benzyl groups.
OH
'NH
0
H 2N
lip
H
O
Figure 2.3
Oligocarbamates based on
4-amino 3-hydroxybutyric acid. Monomer
coupling proceeded smoothly with some
monomer side chains, but with others
coupling efficiency was reduced by elimination.
0
OH
PG'N
0
N
N'
HN'Ri
0
H
0
0
PGIN
R
N
N
IN
R = dimethylarnino,
morpholino
R
2
R*NH
1
*NH
benzyl,
0
H 2N
dodecyl
0
H
OH
PG'N
HNRi
0
H
G
0
OH
itN
H
0'N
HNR1
0
NN
0
H
N
\zJ
PG
,N
0
O
N
H
R2N
I~1
HN
NH
0
PG'
0
O
N
H
0H'
Y
0
HN, R 2
50
To study the versatility and robustness of this method, we synthesized a set of
monomers with varying side chains, including aliphatic (dodecyl), aromatic (benzyl),
hydrophilic (dioxolane), and positively charged (dimethylamino, morpholino). We then
linked these monomers into dimers and trimers in various combinations and observed the
efficiency of each reaction by LC-MS.
Monomer coupling proceeded smoothly for
hydrophobic and uncharged side chains, but was less efficient when side chains were
positively charged. With positively charged side chains we observed elimination of the
hydroxyl group upon activation with CDI (Figure 2.3). The elimination byproduct was
observed in significant amounts (up to 50%), preventing elongation of the chain. These
results indicated that in order to incorporate monomer side chains that are positively
charged, we would need to change the monomer backbone to a structure that was less likely
to undergo elimination.
2.2.5 Oligocarbamates based on 4-amino 2-hydroxybutyricacid
Based on prior results, we next explored 4-amino 2-hydroxybutyric acid as a
monomer backbone structure. As before, the carboxylic acid group was modified to
incorporate variable side chains, and the alcohol of the first monomer was linked to the
amine of the second monomer using CDI to form oligocarbamates (Figure 2.4).
In pilot experiments, monomer coupling proceeded smoothly and with high yield
based on LC-MS analysis.
Pilot reactions were carried out using monomers with
hydrophobic side chains, usually benzyl. As before, we tested the versatility of this method
by synthesizing a set of monomers with varying side chains, including aliphatic (dodecyl),
aromatic (benzyl), hydrophilic (dioxolane), and positively charged (dimethylamino,
morpholino).
We then linked these monomers into dimers and trimers in various
combinations and observed the efficiency of each reaction by LC-MS. As in our previous
strategy, monomer coupling proceeded smoothly for hydrophobic and uncharged side
chains, but was less efficient when side chains were positively charged. In this case, the
major side reaction was the formation of an oxazolidinedione ring (Figure 2.4) upon
addition of CDI. This byproduct was present in significant amounts (up to 40%) for certain
monomer side chains.
51
0
OH
J
H 2N
Figure 2.4 Oligocarbamates
based on 4-amino
2-hydroxybutyric acid.
Monomer coupling proceeded
smoothly with some monomer
side chains, but with others
coupling efficiency was reduced
by cyclization of the imidazoleN-carboxylic ester intermediate.
OH
0
NH
PG% N
OH
H
0
IN
IIN
N
R1
O
NH
PG N
H
JO IN *N
R2
O
OH
H 2N
H 2N
1
NH
O
PG 1N Q
H
O
O
N
H
N
\:zj
0OH
)
N
GN
PGN
H
NH
0
NH
HI N-i
R2
Ri
0
0
1
NH
0
N
R1
1
N
0
N JO
HH
NH
NH
H
52
H2
OH
Figure 2.5 Oligocarbamates based
on 4-amino 2-hydroxybutyric acid.
Monomer side chains were modified
to afford a fully substituted amide
nitrogen. This change was designed
to prevent the cyclization reaction
shown in Figure 2.4.
OH
H 2N
1
R
O
N
PG
OH
H
o
N'N
I
N-'
a WN
R1
N
PGO
H
R2
O
N.
OH
H 2N
1
R2
O
O NPGN
O
H
&
0
HN
NOH
N
H
OH
R
0
NN
N
N'
(OH
HN
O
0
'
H
O
PGN
N
O-4
R2
IN1
PG
N
N
H
N
N
H
H
53
53
To avoid the inadvertent formation of oxazolidinediones, we synthesized a new set
of monomers in which all of the side chains were formed by amidation using secondary
amines, yielding tertiary amides unlikely to cyclize (Figure 2.5). We observed a similar
trend in monomer coupling - reactions proceeded smoothly with hydrophobic and
uncharged side chains, but yield of desired product was lower with positively charged side
chains.
Two side reactions contributed to this decreased efficiency. First, we observed that
at least one monomer (dimethylamino) was unstable over time and tended to undergo
cyclization to form a 5-membered amide ring (Figure 2.5). Though this problem could be
mitigated by using freshly prepared monomer in excess, it would be preferable to be able to
Second, formation of a different
store monomers for some time before use.
oxazolidinedione ring structure was observed after monomers had been linked together
into dimers (Figure 2.5). This rearrangement resulted in loss of a monomer side chain.
These results demonstrated that the formation of stable 5-membered ring
structures is favored when certain side chains are present, and would likely limit the utility
of this synthetic strategy for the synthesis of libraries. For this reason, we moved to a new
monomer backbone to avoid this ring formation.
2.2.6 Oligocarbamates based on hydroxyproline
In our previous synthetic strategies for making oligocarbamates, the efficiency of
coupling monomers together into chains was limited by several side reactions that involved
the cyclization of monomers or monomer chains to form five-membered rings. As five- and
six-membered rings are generally quite stable, it may be difficult to avoid the
intramolecular cyclization of a linear oligomer backbone over time.
Based on our
observations, we hypothesized that a rigid monomer backbone that intentionally includes
five-membered rings may be more stable. For this reason we chose hydroxyproline as a
monomer backbone structure.
54
HN
Figure 2.6 Oligocarbamates based on
hydroxyproline. This strategy affords
highly efficient monomer coupling as long
as the imidazole-N-carboxylic ester intermediate is kept dry.
-H
OH
N
PG
oOH
0)2
NRT
0
N
N
H 20
0
PG, N
N'\N
NZ/
0
N
R1
00H
Q
HN
0
NR2
0
PG, N
0
N ON
00
N
0
N
N"'
0
N-Rj
oOH
R2
Ri N
0FI
N^N
)-N
55
NR2
As before, monomer side chains are added to the backbone by amidation of the
carboxylic acid. As in our previous strategy, we used tertiary amides to avoid unwanted
reaction of the amide nitrogen during oligomer synthesis. The hydroxyl and amine on
hydroxyproline were then coupled together using CDI to form carbamates. In this case,
monomer coupling proceeded smoothly for all side chains tested, including aliphatic
(methyldodecyl), aromatic (methylbenzyl), hydrophilic uncharged (dioxolane), and
Complete
positively charged (dimethylamino, morpholino, imidazole) (Figure 2.6).
conversion of monomer to dimer was consistently observed when a 2- to 6-fold molar
excess of amine (second monomer) was used.
The only side reaction observed using this synthetic strategy was the gradual
hydrolysis of the imidazole-N-carboxylic ester intermediate. This side reaction is easily
avoided by limiting its exposure to moisture. Simple procedural adjustments were
sufficient to avoid this degradation. First, reactions were carried out in a dry nitrogen
environment using dry solvents and thoroughly dried monomers. Second, quenching of
excess CDI reagent was done with an equimolar amount of water to CDI, rather than excess
water. Finally, an extra drying step was added to the workup of the imidazole carbonyl
ester intermediate - benzene was added to the intermediate, which was dried under
vacuum to remove residual water in the form of the benzene-water azeotrope.
Using this method, we were able to consistently achieve high conversion of
monomer to dimer, trimer, and longer chains. These chains were highly stable over time,
even when exposed to heat and acidic environments. When coupled with a fluorous
purification strategy (discussed in the next section), we were able to achieve good recovery
of product with excellent purity.
This method is a robust candidate for scale-up to a high-throughput strategy.
Reactions proceed reliably for a variety of different side chains, and the only observed side
reaction (hydrolysis of the intermediate), results in the loss of a monomer from a chain.
While this side reaction is undesirable, it is both avoidable and predictable. In highthroughput synthesis, some impurities will invariably be present in some wells.
Preliminary screening of materials is usually carried out on crude mixtures, the composition
of which is deconvoluted after hits are identified. As this process can be difficult and timeconsuming, it is extremely helpful to know what impurities to expect. In the synthetic
strategy developed here, it is advantageous that the side reactions that do occasionally
occur are very predictable, which will facilitate the detailed analysis of results in the future.
56
2.3 Hydroxyproline-basedoligocarbamates provide a robust platform for
the synthesis of siRNA conjugate libraries
Pilot studies of various chemical strategies indicated that hydroxyproline-based
oligocarbamates could provide an effective method for synthesizing siRNA conjugate
libraries. To test the robustness and versatility of this method we first synthesized a set of 6
diverse monomers and then combined these into 14 different trimers. Monomers were
synthesized by amidation of the carboxylic acid on CbZ-protected hydroxyproline (Sigma),
followed by deprotection of the amine (Figure 2.7a). Fluorous-CbZ-protected starting
monomers were synthesized by coupling of fluorous-CbZ-NHS ester (Fluorous
Technologies) to hydroxyproline, followed by amidation of the carboxylic acid. Coupling of
monomers was carried out by adding excess carbonyldiimidazole (CDI, Sigma) to starting
monomer in dimethylformamide (DMF) (Figure 2.7b). After 1 h, excess CDI was quenched
with water and the activated imidazole-N-carboxylic ester was dried thoroughly. The
second monomer was then added in excess and reacted overnight to yield the carbamate
product. After the addition of each monomer, solid phase extraction (SPE) on fluorous silica
(Fluorous Technologies) provided a simple and general method for purifying fluoroustagged oligomer chains.
For a general and scalable purification strategy, we take advantage of a fluorous
Molecules tagged with this
protecting group that serves as a purification handle.
perfluorocarbon chain can be rapidly recovered from a reaction mixture by solid phase
extraction through fluorous silica (FSPE, Figure 2.7c). The reaction mixture in DMF is
loaded onto the silica matrix and washed with an aqueous solvent, removing excess
monomer and any impurities that lack a fluorous tag. The fluorous product is then eluted
with acetone. This method afforded efficient separation of oligomers from excess reagents
for a wide range of monomer and oligomer properties. Fluorous tags provide the benefit of
facile purification afforded by solid-phase synthesis, but with the advantage of solutionphase reaction kinetics and direct analysis of materials without the need for cleavage of
materials from the solid support. This method can be readily scaled for high-throughput
purification, as fluorous columns are available in 96-well plate format.
57
a.
0
O
OH
N
0
0
-
HN
OH
O
RH0
R NHRNH
(:
OH
b.
-OH
0
0
C.
OH
N
O
C8 F 1 7
Reaction
mixture
10
Ri
N.) N
N''N
2.)
HN
80%
MeOH/H 20
Acetone
-OH
NAN
Fluorous
R2
silica
Silica
0
0
0
C8 F 17
N
0
oR
-
-OH
o
C8 F17
RiR2
1.)
N
2
0
o
0
0_
S0
coupling using CDI c.) Purification by fluorous solid phase
extraction (FSPE)
3
0
O
o
O
N
*~ON
0O N
o-
R1
R2
R3
C0F17
product
oligocarbamates. a.) Monomer synthesis b.) Monomer
OH
N
R
Fluorous
Figure 2.7 Synthesis and purification of hydroxyproline
.) HN
K
1
Impurities
OH
H
Reactions were monitored by LCMS (Figures 2.8, 2.10, 2.11, Appendix 2.2).
Monomer coupling consistently proceeded smoothly and with high conversion, based on the
shift in elution time and the detected product mass. Minimal impurities were observed,
with the most common impurity being unreacted starting material. This was most likely
due to hydrolysis of the imidazole-N-carboxylic ester intermediate, as formation of the
intermediate proceeded reliably with complete conversion. Hydrolysis was easily avoided
by thorough drying of reagents and by minimizing exposure of the intermediate to
moisture. Purity of products was calculated by integration of the LC trace at 214 nm.
Product structure was confirmed by 1H NMR and
13
C
NMR (Appendix 2.1).
58
811.2047 [M+H]+
AJL
C.F,
00
C0,
O
O
'-0H
1120,2130
[M+HI+
A
I
CF,.
..
'''.'j..i'....''i'',..'..'.
[M+HI+
02
0.50
1.00
1.60
2.06 ... 2.50
3.00
min
500
100
1380.5671
1
m/Z
'00
Figure 2.8 LCMS analysis of trimer synthesis. LC traces of monomer, dimer, and trimer show UV absorbance at
214 nm. Mass spectra show ions contributing to the corresponding peak in the MS trace.
For the 14 trimers synthesized, product purity was consistently high (averaging
94% for dimers and 92% for timers), and recovered yield was variable, averaging 72% per
step. We observed reduced yield for the most hydrophilic oligomers, which was due to
partial elution of product in the aqueous wash. Reducing the amount of washing improved
recovery in subsequent steps. This may also be avoided by using longer perfluorocarbon
tags.
59
1
a.
2
3
4
6
N,
I
N
5
N
N
R= N
N
N
0
Rf'~O N
b.
N
R0,'
0
OH
R4
0
R1
R2
Purity (%) / Yield (%)
R1\R2
1
1
2
90/55
3
4
2
100/48
3
6
96/80
90/94
100/90
100/48
1_
5
97/76
5
6
4
95/54
80/68
87/86
97/85
1_1100/68
C.
0
Rf^'O fl NH
0O
K
Rl.0
RI
Ri
R2
R3
R2
-OH
Purity (%) / Yield (%)
R1R2\R3
1
2
3
4
6
71/23
12
14
100/48
21
73/83
89/14
78/34
1100/67
25
34
5
100/64
90/54
36
100/82
42
100/87
44
91/58
46
100/97
53
54
99/86
63
95/60
Figure 2.9 Purity and Yield of various monomer combinations. A set of
14 unique trimers was synthesized from six starting monomers. a.)
Monomer side chains for monomers 1- 6. b.) Purity and yield are
reported for each dimerization reaction. c.) Purity and yield are
reported for each trimerization reaction.
60
To evaluate the utility of this method for synthesizing longer chains, we selected a
subset of the trimers, which exhibited varying hydrophilicity and charge, and extended
these chains to hexamers (Figure 2.10, Appendix 2.2). As before, monomer coupling
proceeded smoothly, with essentially complete conversion. Recovered yield averaged 80%
per step, an improvement over the average yield observed during trimer synthesis. This
improvement is most likely due to adjustments made to the purification process - the
amount of aqueous washing was reduced by half, which improved yields without sacrificing
purity.
ZNOH
Z
Z
[M+H]+
N-
N-
N-
N-
1728.9980
+
[M+2H] 2
864.967 6
1020.0685
[M+2H] 2
.
43,
yO -OH
0
N-
- QO
~N
OZ1 OZ DOH
O
C8F17
0
0
783.0773
0
[M+3H]
3
.
O
N
1174.6709
[M+2H]24
0.5
1.0
1.5
2.0
2.5
3.0
mn
500
1000
1500
m/z
Figure 2.10 LCMS analysis of hexamer synthesis. LC traces of tetramer, pentamer, and hexamer show UV absorbance at
214 nm. Mass spectra show ions contributing to the corresponding peak in the MS trace.
We next explored whether hydroxyproline-based oligomers could be conjugated to
biomacromolecules. Using the same CDI chemistry, we attached a short polyethylene glycol
with an azide functionality to a subset of five selected trimers (Figure 2.11). After
purification by FSPE, we removed the fluorous protecting group by heating in
trifluoroacetic acid (Figure 2.11).
61
815.2253
cg7^ o
C.F
N
.<)r
N
[M+H]+
[M+H]+
~
1153.3219
N
N- NH
[M+H]+
1394.6890
N'
N
N
N
~5N
3
Ok<
...
0
[M+H]+
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
[M+2H]
2
+
.
1638.8097
N
.
.
.
N
N
0
.
O
0 <
.
0
CsFI?--
529.8145
[M+H]+
1058.7411
0.50
0.75
1.00
1.25
1.50
1.75
2.00
2.25
'2.50
2.75
3.00
min
500
1000
1500
m/z
Figure 2.11 LCMS analysis of trimer synthesis, coupling to azido PEG, and removal of purification tag. LC
traces show UV absorbance at 214 nm. Mass spectra show ions contributing to the product peak in the MS trace.
The final LC trace shows peaks representing deprotected product (1.2 min), fluorous-tagged starting material (2.35
min), and free fluorous tag (2.8 min).
The azide-modified trimers were conjugated to dibenzocyclooctyne (DBCO)modified siRNA using a 10-fold molar excess of the azide to DBCO. Both fluorous-protected
trimers and deprotected trimers conjugated readily to the siRNA using a solvent system of
30% acetonitrile in phosphate-buffered saline (PBS). Deprotected trimers were conjugated
without purification, and the presence of the free fluorous protecting group did not appear
to affect conjugation efficiency. Excess azide and free fluorous tag were removed from the
mixture by ethanol precipitation. Conjugation efficiency was analyzed by HPLC (Figure
2.12, Appendix 2.3). Conjugation consistently proceeded essentially to completion, based
on the shift in elution time by H PLC.
62
siRNA Conjugates by HPLC
14 0
12:0
0
C
80
.0
0
-
Unmodified siRNA
--
341-siRNA
-F341-siRNA
6
0
0
2
0
5
10
15
20
25
30
35
40
45
50
55
60
Time (min)
Conjugation of fluorous-protected trimers to siRNA
-Unmodified
E60
o
120
siRNA
s13-F214
ci10
N120
100
s13-F341
80
M
60
0
40
--
s13-F365
40
$ 0
..---
__
5
10
15
20
25
30
35
40
45
50
55
60
Time (min)
Conjugation of deprotected trimers to siRNA
140
-341
E 120
8
-443
100
-215
@ 80
-Unmodified
C60
0
siRNA
40
20
0'
5
10
15
20
25
30
35
40
45
50
55
60
Time (min)
Figure 2.12 HPLC analysis of oligocarbamate conjugation to siRNA. Top trace: Azide-modified trimer 341
(structure shown in Figure 2.11) was conjugated to DBCO-modified siRNA. Unmodified siRNA = siRNA-DBCO alone,
F341 -siRNA = fluorous protected trimer 341 conjugated to siRNA-DBCO, 341 -siRNA = deprotected trimer 341 conjugated to siRNA-DBCO. Middle trace: Samples of other fluorous-tagged trimers conjugated to siRNA. Side chains
corresponding to each number shown in Figure 2.9a. Bottom trace: Samples of other deprotected trimers conjugated
to siRNA.
Our results indicate that hydroxyproline-based oligocarbamates offer a viable
strategy for the high-throughput synthesis of siRNA conjugate libraries. Conversion of
monomer to dimer, trimer, and longer chains proceeded smoothly with this chemistry
regardless of the properties of the individual monomers. Oligocarbamate chains were all
purified by one general method, solid phase extraction through fluorous silica, and were
isolated with good yields and outstanding purity. We were able to easily incorporate a
functional group for conjugation to biomolecules using the same CDI chemistry and the
Deprotection was achieved by simple heating in
same FSPE purification method.
63
trifluoroacetic acid, and deprotected oligomers could be purified by FSPE or conjugated
directly to siRNA with out purification. Conjugation of various trimers to siRNA was
demonstrated using a general method.
Finally, a simple and scalable strategy for
purification of conjugates, ethanol precipitation, was also demonstrated. Importantly, all of
the steps of the synthesis of these carbamate materials, their purification, and their
conjugation to biomacromolecules can be carried out in 96-well plate format.
64
Chapter 3.
High-Throughput synthesis of a hydroxyproline
oligomer-siRNA conjugate library
3.1 Introduction
In the previous chapter, we showed that hydroxyproline-based oligomers are wellsuited to high-throughput library synthesis. The chemistry of linking monomers together
proceeds efficiently and the FSPE purification method provides excellent purity, regardless
of monomer properties. Functional groups for siRNA attachment are easily incorporated
using the same chemistry and purification method. Purification tags are easily removed,
and deprotected products can be conjugated to biomolecules with or without purification.
In this chapter, we apply this strategy to the synthesis of hundreds of diverse
materials in parallel. The process is semi-automated, making use of liquid-handling robots
to mix reactants in various combinations. Purification is carried out in 96-well plate format
using Fluorous SPE plates. We use LCMS analysis to characterize a subset of the materials.
We also develop methods of high-throughput conjugation to siRNA as well as highthroughput purification of siRNA conjugates.
-N
N/
OH
OH
N0
HN
HNO
OH
HN D-OH
No
ow
/N
HN
NH
"*0H
H 04
wHN D
4H
Figure 3.1 Monomers for library synthesis. A set of seven unique starting mono-
mers was synthesized from hydroxyproline.
65
Using this method we are able to synthesize 504 unique siRNA conjugates. We are
able to incorporate a variety of material properties that are relevant to siRNA delivery in a
manner that allows for the study of structure-function relationships.
The strategy
developed here is a robust method for synthesizing libraries of diverse materials conjugated
to biomolecules, which can enable the discovery new therapeutically relevant materials,
including siRNA conjugate delivery materials.
3.2 High-throughputsynthesis of hydroxyproline-basedoligocarbamates
Beginning with seven unique monomers (Figure 3.1), we synthesized a set of 252
unique trimers in multiwell plates. Individual monomers were synthesized conventionally
and loaded onto a liquid-handling robot for library synthesis (Figure 3.2). To reduce the
exposure of the reactions to moisture, all reaction vessels were dried overnight in a 150'C
oven and the space inside the robot was purged with dry nitrogen.
First, solutions of starting, fluorous-protected monomers were distributed into
wells of a 96-well plate, followed by a solution of CDI. After one hour, plates were cooled to
00 C excess CDI was quenched with 3:1 acetonitrile: water. After quenching, plates were
removed from the robot and transferred to a Genevac EZ-2 Evaporator to remove solvent.
Once concentrated, the imidazole-N-carboxylic ester intermediates were further dried by
redissolving them in benzene and again removing solvents under vacuum in the Genevac.
Once thoroughly dried, the plates were transferred back to the liquid-handling robot, which
distributed solutions of (deprotected) monomers into each of the wells. The reactions were
then stirred overnight at room temperature.
66
Conventional synthesis of 7 starting monomers
Purification was achieved by
FSPE in 96-well plates. Silica was
preconditioned with 15% water in
DMF, the reaction mixture was
loaded, and 15% water in DMF was
used to wash away excess monomer
and other impurities.
Fluorous
product was eluted with THF, and
the silica was cleaned with a 0.5%
solution
of
TFA
in
1:1
methanol:THF.
0
o
HN
H
N
1N
0N)
0-P
HNOH
HN
o 0_
OH
HN
'OH
HN
OH
Automated trimer synthesis --> 252 compounds
LCMS analysis showed that
synthesis proceeded largely as
expected
(Figures
3.3,
3.5).
Monomer
coupling
proceeded
smoothly for various combinations
of monomers. The most common
impurity observed was unconverted
starting material, most likely due to
the hydrolysis of the imidazole-N-
High-throughput purification in 96-we 11
Fluorous SPE plates
(Figure
3.4).
carboxylic
ester
Hydrolysis was observed more
frequently than in pilot studies.
This is most likely due to the longer
time required for the workup after
CDI quenching. In individual benchtop reactions, solvents could be
evaporated quickly due to their
LCMS Characterization
Upon
relatively small volumes.
scale-up to hundreds of reactions,
the larger solvent volume required
greater
time
for evaporation,
allowing more time for exposure to
moisture and hydrolysis of the
intermediate. This problem could
be mitigated by using a more
Figure 3.2 High-throughput synthesis workflow
volatile solvent for this step, such as
Overall,
tetrahydrofuran (THF).
purity of trimers averaged 68% based on integration of LC traces at 214 nm.
67
0
893.3083
0K
[M+H]+
OH
0
CNF!
,N
1785.6145
1149.5431
[M+H]+
1409.7048
0
[M+H]+
2.0
1.5
min
3.0
2.5
1000
500
1500 m/z
Figure 3.3 LCMS analysis of high-throughput trimer synthesis. LC traces of monomer,
dimer, and trimer show UV absorbance at 214 nm. Mass spectra show ions contributing to
the corresponding peak in the MS trace.
0o
~
OO
0
3 K
No..O 0
COF1D
1.0
1.5
Figure 3.4 Impurities. The most
common impurity detected in the
library was unreacted starting
material. In this case, LCMS
analysis shows the presence of
unreacted dimer.
N
_-OH
_
H
2.5
2.0
0
O
0
Cs
CgF17
3.0
1395.6764
[M+H]+
-).oj
-OH
O
min
1153.5270
[M+H]+
t
500
1000
1
500
1000
1500 m/z
An azide functionality was then added to the trimer library to enable siRNA
attachment. Additional washing steps were added to the purification protocol to remove
any traces of excess azide-PEG amine, which, if not completely removed, could reduce the
68
efficiency siRNA conjugation. This extra purification may have reduced recovered yield, but
this was acceptable because very small amounts of material are required for siRNA
conjugation. A selection of the trimer azides was weighed, and the average recovery was
9.6 mg, roughly a 51% theoretical yield.
811.2047
[M+H]+
H
COF17
OLO
"I''"I''"..''"." ".
[M+H]+
-
N
1120.3473
....
.............
"I"M'"
I"'
" I" !"""'l "if
" "-
O
C&IF
OO
O
SN\
N
4
N
1380.5056
OH
[M+H]+
[M+2H] +
2
N
690.7457
-
N
_-0
S 7
C F1
O
O
N
1.0
1.5
2.0
N
N
O
)
,
0
1624.7058
'N0
[M+H]+
N
500
1000
1500 M/z
Figure 3.5 LCMS analysis of high-throughput trimer synthesis and attachement of
azido-PEG. LC traces show UV absorbance at 214 nm. Mass spectra show ions contributing to the corresponding peak in the MS trace.
Each of the 252 trimer-azides was split into two batches, one of which underwent
deprotection to remove the fluorous purification tag. This doubled the total number of
compounds in the library to 504 (252 trimer-azides with fluorous tags and 252 without
Deprotection was carried out using iodotrimethylsilane (TMSI) in
fluorous tags).
dichloromethane at 0*C. After quenching with methanol, the reaction mixture was
incubated with sodium thiosulfate beads to remove the iodine byproduct before
purification by FSPE. Pilot studies of this reaction showed complete deprotection of the CbZ
group without detectable degradation of the trimer backbone (Appendix 3.2).
69
3.3 Conjugation of oligocarbamates to siRNA and conjugate purification
We next developed methods of conjugating the carbamate-azide library to siRNA
and purifying the siRNA conjugates. While conjugation could easily be carried out in 96well plates, the purification strategy required adjustments to accommodate this format.
The standard method of purifying nucleic acid by ethanol precipitation involves briefly
freezing samples followed by centrifugation at very high forces (10,000 - 20,000 x g).
Adapting this procedure to 96-well plates required adjustments because standard
centrifuges designed to run plates reach only forces of 3,000 x g. We compensated for this
difference by freezing the samples longer (overnight) and centrifuging longer (90 min).
We tested whether purification by ethanol precipitation in 96-well plates afforded
consistent recovery of siRNA conjugates. As conjugation of a hydrophobic material to siRNA
may alter the solubility of the nucleic acid in ethanol, we were concerned that we may
experience a loss of conjugated siRNA when purifying by this method. Also, the slower
centrifugation speeds may be insufficient to pellet out precipitated nucleic acid.
To test this, we conjugated a hydrophobic trimer (fluorous-protected benzyl-benzylbenzyl, F333) to two different siRNA sequences (siLuc and GaINAc-modified siTTR) across
the diagonal of a 96-well PCR plate. The samples were purified by ethanol precipitation as
described above, dried, and redissolved in PBS. The concentration of siRNA was measured
using a NanoDrop spectrophotometer. Based on the absorbance of the samples at 260 and
280 nm, the average recovery of siRNA was 96%, with all eight samples over 89%. To test
whether the added carbamate material contributed significantly to absorbance at 260 (and
therefore could skew the measurement of siRNA concentration) we also measured the
absorbance spectrum of a concentrated sample of the carbamate alone, which showed peak
absorbance at 230 nm and negligible absorbance at 260 nm (Appendix 3.3). We concluded
that the standard spectrophotometry method for measuring siRNA concentration (using the
absorbance of the sample at 260 and 280 nm) is a reliable method for measuring the
concentration of siRNA conjugated to the carbamate materials, and that ethanol
precipitation in 96-well plate format affords consistently high recovery of siRNA.
We then used this method to conjugate the library of 504 carbamate azide materials
HPLC analysis of selected wells showed consistently efficient
to siRNA sequences.
conjugation and good recovery (Figure 3.6). We observed the same for several siRNA
sequences (siLuc, Figure 3.6; siGFP, siAHA1) as well as for siRNA sequences that were also
conjugated to a targeting ligand (Folate-siGFP, Figure 2.12, GalNAc-siTTR, Figure 3.6).
70
siLuc conjugates by HPLC
140
E 120
100
80
U
C
siRNA
-Unmodified
siLuc-F126
60
-siLuc-111
o 40
20
0
3
7
5
9
11
13
15
17
19
23
21
25
27
29
31
35
33
37
39
Time (min)
E
GaINAc-siTTR conjugates
80
unmodified
0-
-
GaINAc-siTTR-72
o-
-
GaNAc-siTTR-F12 6
0
20
7
9
11
13
15
17
19
21
23
25
27
29
31
33
35
37
39
41
43
45
47
49
51
53
Time (min)
Figure 3.6 HPLC analysis of oligocarbamate library conjugation to siRNA. Top trace: Azide-modified trimers
were conjugated to DBCO-modified luciferase siRNA (top trace) and to an siRNA targeting transthyretin that was
modified with both the DBCO moiety and a trivalent N-acetylgalactosamine ligand (GaINAc-siTTR, bottom trace).
3.4 Conclusionsand recommendations
Our results show that this method of making hydroxyproline-based oligocarbamates
is a viable strategy for the synthesis of siRNA conjugate libraries. In synthesizing over 500
compounds in parallel, we generally observed the results expected based on pilot studies.
Monomer coupling and oligomer purification proceeded smoothly, azide functionalities
were easily added to chains, and conjugation to siRNA was efficient.
71
We did observe some reduction in efficiency and purity compared to pilot studies,
and we believe these issues are easily addressable. For example, reduced efficiency of the
CDI coupling reaction was likely due to the increased handling time of the imidazole-Ncarboxylic ester intermediate. This problem can be avoided by using a more volatile solvent
for the addition of CDI (THF rather than DMF).
Removal of fluorous purification tags can also be made simpler. While pilot studies
of this reaction using TMSI showed efficient deprotection without degradation of the
trimers, later data suggested that in some cases a small amount of degradation could occur.
If a small amount of unwanted cleavage occurs, free azide-PEG-amine may be present in the
deprotected product. Even a small amount of this free azide could greatly reduce the
efficiency of siRNA conjugation and therefore delivery efficiency.
We found that
deprotection by heating in trifluoroacetic acid avoided this degradation. In addition, this
obviates the need for incubation with thiosulfate beads, leaving a cleaner final product free
of thiosulfate salts. We also found that TFA-deprotected trimers could be conjugated to
siRNA without FSPE purification, as free fluorous tag does not affect the conjugation
reaction. The free fluorous protecting group is then removed during ethanol precipitation
of siRNA conjugates.
The semi-automated nature of this method places some limitations on the possible
throughput. FSPE purifications were not automated and were carried out using a
multichannel pipette. Automation of this process is possible but may require some effort.
When plate-based FSPE was done by hand, the FSPE plates were fitted to a vacuum
manifold to draw solvents through the silica quickly. This plate-compatible vacuum system
is not available on most liquid handling robots. FSPE purification can also be done by
gravity alone; the plates are loaded with coarse-grained silica for this reason. Using our
semi-automated method, the largest number of compounds one could practically synthesize
in parallel would be 384, equivalent to four 96-well plates. Splitting each into two batches
before deprotection, as we have done here, can double this number to 768.
Our method of hydroxyproline oligomer synthesis enabled us to synthesize 504
unique siRNA conjugates in parallel. While purity and yield of materials were acceptable in
this pilot study, we also found some simple ways to improve the efficiency and simplicity of
the process. Our data demonstrate the robustness of this method and its utility for the
synthesis and study of libraries of siRNA conjugate delivery materials.
72
Chapter 4:
Potential applications of hydroxyproline-based
oligocarbamates
4.1 Introduction
In our efforts to develop the capacity to synthesize combinatorial libraries of siRNA
conjugates, we have also developed a novel strategy for the synthesis of sequence-defined
polymers. These polymers, which can incorporate a wide variety of side chains, could find
applications in many fields, including molecular recognition, molecular self-assembly,
peptidomimetics, novel enzyme inhibitors, and various other applications that can benefit
from library screening. As our original purpose for developing this methodology was for
the study of siRNA delivery materials, we chose to focus first on this application, though we
hope that other researchers may find these materials useful in other fields.
There are a number of strategies available to assess the ability of these materials to
deliver RNA to cells. For in vitro assays, cell types, gene targets, siRNA sequences, and
Delivery materials can be conjugated to siRNA
siRNA modifications can all vary.
monovalently or multivalently. They can be the only material conjugated to the siRNA, or
they can be conjugated to an siRNA strand that is also modified with a targeting ligand.
They can be administered as purified conjugates or with an added excess of delivery
material, as a colloidal nanoparticle system. Conjugates can be combined with other
delivery materials in an effort to improve efficacy of other systems.
Our initial screening efforts focused on assessing the ability of the materials to
mediate delivery 1) on their own as monovalent conjugates, 2) as monovalent conjugates
alongside a targeting ligand, 3) as conjugates in synergy with delivery lipids, and 4) as
delivery lipids themselves, forming nanoparticles. We chose cell types and gene targets for
ease of analysis high-throughput, primarily using reporter genes in common cell lines.
For the 504 materials screened, we did not observe functional delivery that proved
to be reproducible in the assays chosen. Though no outstanding hits were discovered in
this first attempt at a library screen, our data provide a number of lessons to guide future
screens. We discuss possible explanations for our data as well as our recommendations for
future researchers who may be interested in screening this library or other conjugate
libraries for functional delivery. Our methodology and recommendations provide a viable
73
platform for researchers to study conjugate libraries with a higher probability of
discovering efficacious delivery materials.
4.2 Library screening in dual HeLa cells
The first in vitro assay that we chose to screen our conjugate library was in dual
HeLa cells, using an siRNA sequence targeted to the reporter gene firefly luciferase. The
dual HeLa cell line expresses two luciferase reporter genes (Firefly luciferase and Renilla
luciferase) whose expression levels can be measured by a simple luminescence reading.
Renilla luciferase serves as an internal control for cell number and cytotoxicity, while firefly
luciferase is the reporter gene targeted for knockdown by siRNA. We chose this assay
because it is a well-established method for studying functional siRNA delivery in vitro and
because it is well suited to high-throughput screening.
Reagents are commercially
available (Dual-Glo Assay, Promega), the assay protocol is easily carried out in multiwell
plates, and luminescence measurements can be taken quickly using a plate reader.
We screened the library in these cells at several doses and with varying cell culture
conditions. Our initial screens included doses of 500 nM and 200 nM with and without
serum for all 504 compounds. We noticed significant well-to-well variation, even among
untreated wells, resulting in large error bars. We nonetheless identified several lead
compounds that appeared to show knockdown of firefly luciferase without knockdown of
Renilla.
-
The selected compounds were rescreened at multiple doses ranging from 500 nM
5 nM (Figure 4.1). We did not observe the dose-dependent silencing that would be expected
for true siRNA delivery. Furthermore, performance at the 500 nM dose did not replicate the
findings from the original screen at this dose for any of the 20 compounds. Most showed no
silencing at all at the highest dose. For further verification, we conducted an additional
repeat screen at 500 nM on 40 of the top-performing compounds from our original screen.
None of these compounds showed significant silencing.
Overall, results of our initial library screens showed a lack of reproducibility, a lack
of dose-responsiveness, and consistently large error bars (Figures 4.1). Silencing by
positive control formulations was evident, however, suggesting that compounds with strong
74
enough silencing ability should be detectable despite the noisy readout. Our data thus far
suggested that none of the compounds screened was potent enough to be detected by this
assay at the current dosing levels. In attempt to overcome this issue, we screened the
compounds again at higher doses (2 ptM, repeated twice, and 5p.M). Results were similarly
noisy, with a coefficient of variation averaging 0.3 for each treatment, even for non-treated
wells.
- - - -- -
--
- - ---
-
Rescreen of top-performing compounds: 500 nM
- --
2 .5-
E Firefly 0 Renilla
2- -
-
-
- -
- - - -
-
-T-
1 .5
1
0.5
0
Z
~00
I- 0
<
ri(4
m
n
<<
LA
<c
-1 C- n
ao
n
oU
i
4
M
-T
UUU
n
1r
M
-I
n
r
JL_-
i
-i
n V
LL 00U i 00(f 0 XT X
,
m
q
,
m
-t
M
Random variation?
12
Firefly
---- - Re
fl-a
___1.01
10
--
-
.T -T
___
M
Increased gene expression: Cell stress?
-;o 1.2
-
088
-
Firefly
0.44
. Renilla
0.2
2
0.0
0
.
06
,
V1N
Dose response: Conjugate 331
--
1.4
0.8
Z
00000WU.WLULUL
Figure 4.1 Conjugate screens in dual HeLa cells. This assay produced very noisy data, obscuring results. Compounds that appeared to be
demonstrating silencing of firefly luciferase did not show the same effect upon repeat screening and did not demonstrate dose-responsive
silencing. In some cases increased gene expression was observed, which could be due to cell stress or random noise.
We explored whether various aspects of our methods could be causing the high
well-to-well variability in this assay. Using 96-well plates of non-treated cells, we carried
out the assay according to the manufacturer's protocol and examined well-to-well variation.
We adjusted various aspects of our methods of cell culture, plating, assay timing, and
robotics procedures. More specifically we explored plating cells at different densities
(20,000 - 100,000k cells/well), adjusting robotics methods and carrying out the assay by
hand, changing time points, plating cells at different levels of confluency, thawing new
batches of cells, plating cells at various passage numbers, using fresh reagents and reagents
that had been previously prepared and frozen, adjusting the ratio of media to reagent, and
having different researchers plate cells. None of these changes reliably reduced the
The coefficient of variation in luciferase readings
observed well-to-well variability.
averaged 0.2 or higher (Figure 4.2). This level of error would make it difficult to identify
75
any compounds that silence relatively weakly; only highly potent delivery systems could be
identified over the background noise. As our data indicated that none of our conjugates
silenced strongly enough to be detected by this assay, we decided to move to a different cell
line that exhibited less background noise in reporter gene readout.
Variation in non-treated cells
Renilla
Firefly
300000
MH
30000
H
250000
MG
25000
SG
200000
SF
20000
MF
150000
mE
15000
EE
100000
MD
10000
MD
50000
C
5000
SC
0
EB
0
KB
MA
8
9
10
11
12
Firefly/Renilla
'H
20
SG
15
SF
10
KE
MD
5B
0'
1
2
3
4
5
6
MB
B
6 78
9
10
11
12
MA
Figure 4.2 Random variation in non-treated dual HeLa cells. 96-well plates of non-treated dual HeLa cells showed significant levels of
background variation in luciferase expression.
4.2 Library screening in KB cells
We next screened the library in a KB cell line, which expresses the reporter gene
green fluorescent protein (GFP). Expression level of GFP was measured by quantifying total
fluorescence of cell lysates using a multiwell plate reader 48 h after transfection. Control
experiments showed relatively low well-to-well variation in non-treated cells (CV=0.07),
making this cell line an attractive alternative to the dual HeLas discussed in the previous
section.
76
We conjugated the 504 library materials to siGFP and screened them for activity at a
dose of 1 piM in KB cells (Figure 4.3). Background noise levels in this assay were much
improved over the previous assay. However, none of the materials showed significant
improvement in silencing ability over siRNA alone.
KB cell screen: 1 uM, no targeting ligand
-
1.6
-------
-.--..
1.4
-
1.8
- ~--~
0.
6
-
-
1.2
0-
Figure 4.3 siGFP conjugates in KB cells. Selected results of high-throughput screen of 504 siRNA conjugates targeting GFP in KB cells.
We also conjugated the library materials to an siGFP sequence bearing a folate
targeting ligand. KB cells are known to express folate receptors and are used to study
folate-targeted delivery 223,24 2. Folate targeting of drug delivery systems is thought to aid in
245
, but folate-targeted siRNA alone does
cellular uptake via receptor-mediated endocytosis
not induce RNAi (Figure 4.4). Presumably folate-targeting is sufficient to aid cellular uptake
but not endosomal escape' 83. By combining our library materials with folate targeting, we
hoped to identify compounds that may aid in endosomal release or trafficking to the RNAi
machinery. Such compounds could potentially benefit a number of ligand-targeted delivery
systems.
In our first screen using the folate-targeted sequence, we identified five compounds
that showed modest silencing at a 500 nM dose (Figure 4.4a). We repeated the transfection
with these five lead compounds at doses of 500 nM and 1 lM (Figure 4.4b). Of the five, two
showed silencing that replicated the data from the initial screen at 500 nM, and both
showed dose-dependent silencing. A parallel cytotoxicity assay (Cell Titer-Glo, Promega)
showed no detectable toxicity at either dose (Figure 4.4c). The two lead compounds were
and FluorousF214 and F215 (Fluorous-protected-dioxolane-dimethylamino-C12
carbamates
two
these
Delivery by
protected-dioxolane-dimethylamino-morpholino).
appeared to be folate-dependent, as they did not induce silencing in the original screen
without folate targeting (Figure 4.4d).
77
1.4
-t------
-
1.6
w 0.8
0-2
I-<
0:
0
r-
r- r
rN r- r- rN
N-
N-
oo o
0
opo oo oo
co
oo o
o
o-a)q)a o-) 0, a)~ q0) o'l
a) o- c
U
Repeat screen:
Folate-siGFP-carbamates
b.
1.4 T_
1.2
C
0
la
#A
LU
Cell Viability
ATP concentration by Cell Titer-Glo
MluM
1000000
M 500 nM
1
800000
0.8
600000
0.6
ti
1 uM
400000
0.4
500 nM
200000
0.2
0
'-
0
.,
q
iN
4
Control siRNA IF D4 1F D5
14
KB Cell screen
Folate-siGFP-carbamates
d.
of performance of compounds F214 and F215
when conjugated to siRNA sequences with and
without folate targeting ligand seems to
indicate folate-dependent silencing.
a
X
W
a
1.2
-
0.8
0.6
0 .4
0.2
* 500 nM
--
-
1
-
- - --
-
-
-
C
0
*;
-
compounds that demonstrated reproducible,
dose-dependent silencing. d.) The comparison
1.4
-
Figure 4.4
siGFP conjugates with folate
targeting ligand in KB cells, a.) Selected results
of high-throughput screen of 504 folate-siRNA
conjugates targeting GFP in KB cells. Arrows
indicate compounds that were selected for
rescreening. b.) Repeated screen of selected
compounds. c.) Cell viability assay results for
.
------
KB cell screen: 500 nM, folate-targeted siRNA
a.
0
Z
-
<<
<
Encouraged by these results, we synthesized larger batches of the two lead
compounds by conjugating more of the trimers to the folate-modified siRNA. The same
batch of carbamate material was used as had been used previously, it was conjugated to the
same siRNA sequence, and conjugation was confirmed by HPLC. Upon transfection of KB
cells with the new batch of conjugate, we did not observe the same gene silencing effect. No
significant reduction in GFP expression was observed (Figure 4.5).
78
We took several approaches to attempt to replicate the original silencing data with
the two lead compounds. The original batch of conjugate, which had repeatably showed
silencing, had been used completely in previous experiments, leaving none of the original
efficacious batch available for comparison to new batches. As the only difference between
the newest batch and the previous batch was scale at conjugation, we first repeated the
siRNA conjugation step at the original scale, following exactly the same protocol used
originally. We also made additional batches with adjusted conjugation conditions, varying
concentration, time, and ratios of reactants. Upon transfection, these new batches also did
not show significant silencing, even at micromolar doses (Figure 4.5).
F214 and F215 conjugates with folate, different batches
1.2
__
T i.T
T
1
T
I I
0
I I- 1
4--
III
0.8
CL
W 0.6 t
~ 250 nM
0.
500nM
0.4
0.2
0
I
1 uM
I_____
kF
o
(~V
~~
\0
((f
t>1- <V
N
Different batches of lead conjugates
Figure 4.5 Loss of activity in new batches of lead compounds.
We also explored the possibility that the differences in results may have resulted
from variation in cell culture or plating conditions. We repeated the transfection
experiments at different cell plating densities, with freshly thawed cells, and with cells at
various passage numbers. These adjustments did not recover silencing activity. We also
explored delivery with these lead compounds in another GFP-expressing cell line, 4T1 cells.
We did not observe significant silencing by either F214 or F215 at several plating densities.
79
We next explored whether degradation of the carbamate material or the siRNA
sequence may have resulted in the loss of activity. The carbamates were purified by HPLC
and fractions of the purified, intact trimers were isolated and confirmed by LCMS. We also
isolated fractions of some impurities from these materials (which appeared from LCMS
analysis to be dimer-azides, F21-azide). These materials were conjugated to the freshest
available batches of the siRNA sequence. At doses of 500 nM and 1 pM, these conjugates did
not induce silencing (Figure 4.6). These experiments were repeated again later, with newly
synthesized carbamate materials, and the results were analyzed by flow cytometry. Again,
significant silencing again was not observed, indicating that these particular materials as
monovalent purified conjugates most likely do not effectively deliver siRNA in this system,
and that the effects observed in the original data sets must have been caused by something
unpredicted.
HPLC-purified carbamate conjugates (folate-targeted)
1.2
0.8
0.6
0.4
0 nM
EM50
El uM
0.2
0
Figure 4.6 HPLC-purified carbamate materials. HPLC purification of carbamate materials did not recover silencing activity in
KB cells.
We also explored whether excess carbamate material present in the conjugate
solution could have contributed to the original silencing effect. Here, we did see some
evidence of silencing with a large excess of carbamate (roughly a 2:1 weight ratio of
carbamate: siRNA, Figure 4.7). Though it is unclear whether this could explain the silencing
observed in the first set of experiments, it does suggest that increasing the amount of
delivery material relative to siRNA may improve delivery. Adding excess delivery material
also improved the efficacy of some of another material from the library (Figure 4.7). This
led us to explore methods of multivalent conjugation and of formulation of these materials
as nanoparticles, discussed in other sections.
80
Conjugates with excess carbamate
1.4
0
0.
-;
1.2
M
1
Conjugate
W +2 uL carbamate
0.8
0.6
+10 uL
- X-
C:L 0.4
k +30 uL
0.2
0
~
.O
P~~ D <N
Figure 4.7 Conjugates formulated with excess carbamate material. Addition of excess carbamate material to purified
conjugtes showed some degree of recovered silencing, suggesting that the additional material may improve delivery.
4.3 Hydroxyprolineoligomers as lipid-likenanoparticledelivery systems
In the process of characterizing conjugate materials, we observed evidence that
some of the carbamate materials may form complexes with siRNA. For example, excess
carbamate material was regularly added to siRNA in conjugation reactions, and this excess
material had to be removed by ethanol precipitation before HPLC analysis. If the excess
carbamate material was not removed, the siRNA product did not elute from an HPLC
column (Figure 4.8).
HPLC analysis of purified conjugates vs. siRNA with excess carbamate
50:1 mol carbamate F126: mol Luc siRNA
100
E
C
0
M
0i
U
unmodified siRNA
80
siRNA + excess
carbamate
purified siRNA
conjugate
60
40
C
20
0
5
7
9
11
13
15
17
19
21
23
25
27
29
31
33
35
Time (min)
Figure 4.8 HPLC studies suggest the formation of complexes when excess carbamate is added to siRNA. Conjugates were usually
prepared by the addition of a large excess of carbamate-azide material to DBcO-modified siRNA, followed by removal of excess
carbamate by ethanol precipitation and HPLC analysis. If excess conjugate was not removed by ethanol precipitation prior to HPLC
analysis, no siRNA was observed to elute from the HPLC column.
To test whether the carbamate materials could form particles when mixed with
siRNA, a subset of the library materials was formulated with siRNA and particle size was
measured by dynamic light scattering (DLS, Malavern ZetaSizer). Formulation was done by
81
ethanol injection, a method that is typically used for making nanoparticles from nucleic acid
and cationic lipids. The materials selected were the ones most similar to delivery lipids:
those with at least two C12 chains and some cationic component (dimethylamino,
morpholino, or imidazole). Of eighteen library materials tested, five formed nanoparticles
after simple formulation, with average diameters ranging from 260 - 540 nm. The same
materials were also formulated with 1% PEG lipid, C14-PEG2000, a lipid commonly used in
lipid nanoparticle formulations. With 1% PEG lipid add, all 18 materials tested formed
nanoparticles (Figure 4.9), with average diameters ranging from 100 - 500 nm. Some of
these particles showed stability over time, with particle size remaining constant over a 24-h
period (Figure 4.9).
Carbamate Nanoparticle Size
25
-
Particle stability over time
Formulated with 1% PEG lipid
F414 + 1% PEG lipid
I
-20
-
-i30
-F414
-F441
15
-
10
-1 5
446
0X
464
IA
C
'544
"-
min
24 h
0
M
5
1
10
100
Diameter (nm)
1000
10000
1
10
100
1000
Diameter (nm)
10000
Figure 4.9 Carbamate-siRNA particle sizing by dynamic light scattering. Lipid-like carbamates formed nanoparticles when
formulated with siRNA and PEG-lipid. Some formulations showed long-term stability.
We next explored whether the particles could deliver siRNA to KB cells. Particles
were prepared by the same ethanol injection method but with varying amounts of PEG lipid
(none, 0.25%, and 1.25% PEG lipid) and at a higher concentration (10X). The particles were
The purpose of formulating at a higher
then diluted in PBS and added to cells.
concentration was to reduce the final concentration of ethanol added to cells, as toxicity was
observed when particles were not diluted. At a dose of 65 nM in KB cells, some
formulations stood out as showing silencing of GFP (Figure 4.11). The transfection was
repeated with these formulations at two doses (65 nM and 130 nM). One particular
carbamate (F144), which showed the strongest silencing of any of the materials, also
showed silencing in the repeated experiment. Silencing activity for this material was dosedependent and relatively independent of the amount of PEG lipid present (Figure 4.12). We
also explored different ratios of carbamate material to siRNA and found increasing silencing
with increasing ratio of carbamate to siRNA (Figure 4.12). A cytotoxicity assay indicated
82
good cell viability for all formulations (Figure 4.13). However, one repeated transfection
showed weaker silencing for the same formulations (those with F144) that had showed
silencing in previous experiments (not shown).
Cationic Lipid complexes
0
1.5
0.
5:1 wt:wt siRNA
------
1
T-
No PEG
L
aJ
0.5
Lo PEG
Hi PEG
0
F_
0
4~
0
-4
1-T
-J
I~T rq :
t r- ZT R LI, -* - LI, -t -1 .014 TT k.0
1:
.- 1
Iz
L
U
I,
'-4 .-4 r-
j
'Z3
LnLICIC o'0 Ir '0 -T -1 ui W.
-4 .-4 LI
mNY1
U
Figure 4.11
Carbamate-siRNA complexes in KB cells.
Lipid-like carbamates chosen from the library were formulated as
nanoparticle complexes with varying amounts of PEG lipid for delivery to KB cells. Hi PEG = 1.25 mol% PEG-lipid, Lo PEG = 0.25
mol% PEG-lipid.
Nanoparticle complexes: repeated
F144 at different wt:wt siRNA
-- - -
-- - -
No PEG
-
1 .2
1 _TTI
C
.0
0.8
1.2
T---
1
Wt:wt
1A
0. 6
-~
0.4
0.2
U 0.8
a No PEG
*2:1
e0.60.6
* lo PEG
0.4
hi PEG
0.2
10:1
0
0
--
<<<<N
<)t
Figure 4.12 Carbamate-siRNA complexes in KB cells, varying in dose and formulation composition. Only one compound
showed repeated, dose-dependent silencing when formulated as particles. Increasing the ratio of carbamate to siRNA appeared
to improve silencing. Hi PEG = 1.25 mol% PEG-lipid, Lo PEG = 0.25 mol% PEG-lipid.
83
Cell viability
ATP levels measured by Cell Titer-Glo
2
1.8
1.6
1.4
1.2-
- ------* no PEG
M low PEG
0.4
high PEG
0.2
0
Figure 4.13 Carbamate-siRNA complexes in KB cells, cell viability.
To address the discrepancy in results from experiment to experiment, we repeated
the transfection and analyzed the results by flow cytometry. Previous experiments relied
on average measurements of fluorescence in each well, and we hoped that data points on
individual cells could provide clearer information as to whether the observed changes in
GFP levels were truly due to efficacious siRNA delivery. Several of the carbamates were
formulated with siRNA and different ratios of PEG lipid (no PEG, 0.25%, and 1% PEG, 8:1 wt
A propidium iodide stain was included in the flow cytometry
carbamate: wt siRNA).
analysis to gate out dead cells. The data revealed very few GFP-negative cells, even for
formulations that had showed silencing of GFP in previous experiments. Two of the
formulations showed a very modest population of GFP-negative cells (Figure 4.14),
amounting to of cells 4-5% transfected.
Carbamate NP transfection in KB cells, FACS
analysis
~No
PEG
m
12
10
"0
10
8
10
Lo PEG
_H
40F
C
M.
100-
1
1
Z
4
2
0
102
F
CA
0
PBS
F144
F146
F341
F414
F141
Positive controls: Lipo (95% transfection), C12-200
(98% transfection)
Figure 4.14 Carbamate-siRNA complexes in KB cells, flow cytometry analysis. Formulations that previously appeared to show silencing did
not show activity when GFP expression was analyzed by flow cytometry. a.) Percent transfected cells by formulation b.) Histograms of GFP
expression for non treated cells (red) and positive controls (blue = C12-200, green = Lipofectamine RNAiMax) c.) Histograms of GFP
expression for non-treated cells (red) and F144 Lo-PEG formulation (blue). Hi PEG = 1.25 mol% PEG-lipid, Lo PEG = 0.25 mol% PEG-lipid.
84
There are several possible reasons for the variation in performance of these
formulations across experiments. Formulation procedures, concentration, and ratios of
components were adjusted for different experiments and particle size and stability were not
characterized for all formulations. These changes in formulation may affect particle
formation and stability, and this should be explored more thoroughly if such nanoparticle
formulations are to be pursued as delivery agents. The data do suggest that higher ratios of
delivery material to siRNA may improve delivery efficiency, suggesting that multivalent
conjugation may be helpful to the performance of conjugates.
4.4. Conjugate modificationsas enhancers in other delivery systems
Another application of the carbamate library could be to enhance the efficacy of
other delivery systems. One efficacious delivery system, a trivalent N-acetylgalactosamine
(GalNAc) conjugate targeting hepatocytes, is currently in clinical trials for the treatment of
transthyretin amyloidosis. The drug consists of a chemically stabilized siRNA sequence
targeting transthyretin (TTR) conjugated to a trivalent linker and three GaINAc ligands. Our
collaborators at Alnylam Pharmaceuticals generously provided a version of this molecule
that also contained a DBCO group for conjugation to our library of azides. We conjugated
the 504 materials to this sequence and screened the conjugates for delivery to primary
hepatocytes. TTR mRNA levels were measured by a branched DNA (bDNA) assay.
The library of conjugates was added to plated primary hepatocytes at doses ranging
from 500 nM - 0.1 nM. Most of the materials showed no improvement over the parent
sequence, which achieved 50% silencing at the lowest dose. A small number of these
materials (four trimers) showed some modest improvement at the lowest dose (Figure
4.15), but the improvement was not large enough to merit extensive follow up.
85
GaINAc-siTTR conjugates in primary hepatocytes
70.0
50.0
-
-I
M 500nM
-
60.0
40.0
0 1OnM
30.0
M 1nM
20.0
* 0.lnM
10.0
0.0
Parent
766
611
351
F766
726
646
331
666
334
335
411
F333
Figure 4.15. Carbamates conjugated to GaINAc-modified siTTR in primary hepatocytes. Topperforming compounds.
As a control experiment to verify that the siRNA sequences could still induce RNAi
when modified with the carbamates, the library of conjugates was delivered to primary
hepatocytes using Lipofectamine. This confirmed that the sequence was still effective when
conjugated to the carbamate trimers. Interestingly, when combined with this transfection
reagent a number of the materials showed improved silencing over the parent (Figure
4.16). Analysis of the top 25 most effective materials showed that this group of materials
was enriched for monomer 1 (dimethylamino) and monomer 3 (benzyl). In particular,
these groups appeared twice in the same monomer about four times as frequently as they
did on average in the library.
Primary hepatocyte transfection: Synergy with Lipofectamine
70.0
60.0
50.0
40.0
30.0
TT
T
20.0
J.U
C.0a
T
i iI
I
I II
L
u
ii1ti
0 1OnM
4i-i
-
M0.1nM
--- -- ~~-i
U11iEEi~ ii I ii ILdI iUiUJJIIhI
-
10.0
I I
0~
Figure 4.16 Carbamate-siTTR-GaINAc conjugtes transfected with Lipofectamine. Top -performing
compounds.
86
We further explored the possibility that our carbamate materials might synergize
with delivery lipids in other systems. We first explored transfection with Lipofectamine in
dual HeLa cells using the top-performing compounds from the previous screen (Figure
4.17). Again the modification did not hinder silencing activity, but we did not observe the
same synergistic improvement in silencing either. We next explored whether we could
replicate the original results in primary hepatocytes. Using the top four carbamates from
the original screen combined with Lipofectamine, we did not observe any improvement in
silencing over the parent sequence (Figure 4.18). We also explored possible synergy with
the lipid C12-200, but did not observe significant improvement in efficacy with the addition
of the carbamate materials (Figure 4.18).
Transfection with Lipofectamine in Dual HeLas
1.6
0 1.4
1.2
T
T
* 60 nM
::I
0.1
>. 0.8
JI E
I
E 30nM
S0.6
U.
a
ElO nM
0.4
M1
0.2
nM
0
NT
Parent F111
F251
345
336
Figure 4.17 Carbamate-siLuc conjugates transfected with
Lipofectamine in Dual HeLa cells
Transfection with C12-200 in primary
hepatocytes
Transfection with Lipofectamine in
primary hepatocytes
C 1.6
.0 1.4
0. 1.2
11
400.8
0.26
.0
C
40InM
2
2
1.5
T
C1
X
1 40 nM
0.5
& 20 nM
0
Figure 4.18 Carbamate-siTTR-GaINAc conjugates transfected with Lipofectamine and C12-200 in primary hepatocytes. The synergy
between the carbamates and Lipofectamine observed in Figure 4.16 was not observed upon repetition of the experiment.
87
It is unclear why synergistic silencing observed in the library screen was not
replicated in subsequent experiments, even in the same cell type with the same siRNA
sequence. The original observation could be an artifact of noise in the assay readout: in a
screen of hundreds of wells there is inevitably a distribution in readout due experimental
error, and the lower tail of the distribution may appear to be silencing. As silencing was not
observed upon replication of the experiment, we were not convinced of an actual effect of
enhanced delivery in this system.
4.5. Recommendationsfor future in vitro screens
As a first attempt to screen our conjugate library for functional delivery, we
explored various cell-based assays that had previously been useful for screening libraries of
lipid nanoparticles. These assays are useful for high-throughput screening because they are
simple to execute and they allow for quick measurement of gene expression levels using a
microplate reader. However, our results indicate plate reader measurements, which report
a total level of gene expression for a well of cells, may not be sensitive enough for conjugate
library applications. Interpretation of screening data was complicated by signal-to-noise
issues, lack of reproducibility, and even inconsistent controls. Though these cell-based
assays were sufficiently sensitive for LNP screens in the past, our data suggest that
improvements are needed in order to study libraries of conjugates. None of the conjugate
materials in our library appear to silence strongly enough to be consistently detected over
the background variation that is unavoidably present in the assays used.
There are several reasons why our conjugate library cannot be expected to induce in
vitro silencing with the same potency as previous lipid libraries. First, past lipid libraries
have been based around a particular delivery lipid that is already known to be efficacious.
As we lack a comparable starting point for our conjugate library, we have developed a
random library whose range of potency is unknown. Second, LNPs are able to deliver
hundreds of siRNA molecules per particle, so a single internalization event can have a
significant impact on gene expression, leading to strong signals in cellular assays.
Conjugates must achieve hundreds of internalization events to achieve the same level of
silencing. Third, the dose of delivery material in conjugate assays is significantly lower than
that used in LNP screens. A typical LNP screen (100 ng siRNA in 100 uL, 10:1 wt lipid: wt
siRNA) uses only -70 nM siRNA, but this is delivered with roughly 1 ptg delivery lipid per
well. A comparable conjugate screen at 1 ptM siRNA includes only -150 ng delivery
material per well. Overall, it is not surprising that the high-throughput assays studied may
not be sensitive enough to detect the modest levels of silencing we might reasonably expect
out of our first attempt at a conjugate library.
88
There are two ways to address this issue: one is to improve assay sensitivity and the
other is to improve the potency of the delivery conjugates. To improve assay sensitivity,
our results suggest that measuring gene expression levels by flow cytometry is superior to
microplate reader methods. By flow cytometry, we observe consistent performance of
positive controls and we are able to reproducibly detect very modest levels of silencing
(Figure 4.14). Another strategy may be to fluorescently label the siRNA conjugates and
observe their internalization or association with cells by flow cytometry, rather than
measuring functional delivery. To improve the potency of delivery materials, one method
may be to increase the representation of delivery material per siRNA strand via multivalent
conjugation (Chapter 5). We observe some evidence that addition of excess delivery
carbamate may improve performance (Figure 4.7). Also, we have performed our assays
using sequences developed several years ago, and since then significant progress has been
made in sequence design for improved potency. Use of improved stabilization chemistry on
the siRNA backbone may also improve results.
Our work on cell-based screening methods has provided important information to
guide future assay development for in vitro screening of conjugate libraries. We have
demonstrated the need for more sensitive measurements of gene expression levels and
have provided several recommendations for improving upon current assays. In the next
chapter, we provide methods for multivalent conjugation of delivery materials, which may
improve the potency of conjugates in future screens.
89
Chapter 5. Strategies for multivalentconjugationto siRNA
The ability to conjugate multiple copies of a material to siRNA would be beneficial for
several applications. In the case of the delivery materials developed here, increasing the
representation of the carbamate library materials on each siRNA strand may increase the
chances of observing delivery in cell-based screens. Multivalent conjugation would also be
useful in the study of novel targeting ligands, which often benefit from multiple ligandreceptor interactions 23 ,240. We developed methods for making siRNA with trivalent
clickable groups (azides and cyclooctynes) in order to allow facile attachment of multiple
groups to siRNA. This simple method employs commercially available multiarm PEGs
(Creative PEGWorks) and can be used to convert any DBCO-modified sequence into
trivalent azide, trivalent DBCO, or trivalent bicyclononyne (BCN).
a.
+
/
N3-\-(H2H2,nO 3 oCHCH
Oln,-N3
2H2)~
\
CH CH2O
NaN
n
N3
9'C
(CH 2CH 2 O)n-N3
b.
N3
N3
excess
70
Trivalent azide siRNA
-
E 60
50
-siRNA-(N3)3
40
--- siRNA
C30
(0 201
10
0
0
10
20
30
40
50
60
Time (min)
Figure 5.1 Synthesis of trivalent azide-siRNA. a.) Reaction scheme b.) HPLC analysis of DBCO-modified siRNA alone and the same
siRNA after conjugation to 4-arm PEG azide
We first conjugated DBCO-modified siRNA to 4-arm PEG azide (Figure 5.1). Under
dilute conditions, the siRNA was added slowly to a large excess (50-100 eq) of the multiarm
PEG to afford one-to-one coupling of the two components. HPLC analysis indicated efficient
coupling of the two components in a single product peak (Figure 5.1).
90
We also synthesized 4-arm PEG DBCO and 4-arm PEG BCN (Figure 5.2) from purchased
4-arm PEG- NHS ester (Creative PEGWorks) coupled to excess DBCO amine and BCN amine
(Sigma).
Completely substituted products were isolated by reversed phase flash
chromatography. Azide-modified siRNA was made by adding a large excess of diazido PEG
(Sigma) to the same DBCO-modified siRNA (Figures 5.3, 5.4). The 4-arm PEG-cyclooctynes
were then conjugated to the azide- siRNA (Figures 5.3, 5.4). After each conjugation step,
siRNA was thoroughly purified by ethanol precipitation to remove excess reagents.
0
H2N--
N
(CH 2CH 2O)n,
0
\\
O
(CHH2O)n
2 CH2
0
'kCCH
,(CH 2 CH 2
H)fCH
(CH 2 CH2O)n
9
(CH2 CH 20)n
(CH 2CH 2O)
I
-
O)n
YO
0
(CH2CH20)n
0
H
H
H2N 'O' 0
O
O-,
0
H
H
(CH 2CH 2O)n,0
H
0
II
'(CH 2CH 20)n
OxH
:
(CH 2CH 20)n
9
H
H
Figure 5.2 Synthesis of 4-arm PEG-DBCO and 4-arm PEG BCN.
91
\-
(CH2CHz0)n,
.(CHYc-l
CH
(-JCHCH2(CH
0)
excess
excess
N
3
b.
Trivalent DBCO siRNA
E 80
C 70
60
j 50
a 40
m 30
.0
_-S13
_S13 azide
f1~S13
azide
113-3DBCO
o 20
.010
0
15
10
20
30
25
35
40
45
50
6
55
0
Time (min)
Figure 5.3 Synthesis of trivalent DBCO-siRNA. a.) Reaction scheme b.) HPLC analysis of DBCO-modified siRNA alone ("S13"), the
same siRNA after conjugation to diazido PEG ("S13 azide"), and after conjugation to 4-arm PEG DBCO ("S13-3DBCO")
dCH20J..
-:D
\_CH2C2~
a.
n
\CHCHOJ)
N3
eoes
excess
JC CH
H1
HCpH~
_N3
excess
HH
HH
H
b.
Trivalent BCN siRNA
513-3BCN
E 70
60
-
50
_
40
c
513 azide
S13
30
-
0 20
10
10
15
20
25
30
35
Time (min)
40
45
50
55
60
Figure 5.4 Synthesis of trivalent BCN-siRNA, a.) Reaction scheme b.) HPLC analysis of DBCO-modified siRNA alone ("S13"), the same
siRNA after conjugation to diazido PEG ("S13 azide"), and after conjugation to 4-arm PEG BCN ("S13-3BCN")
We next explored whether the library materials developed in Chapter 2 could be
conjugated to the trivalent DBCO-siRNA and trivalent BCN-siRNA. Roughly 3 eq of a sample
oligocarbamate azide (trimer 341, Figure 2.11) were added to the trivalent DBCO and BCN
sequences. By HPLC we observed three shifted peaks in each case (Figures 5.5 and 5.6). We
expect that these peaks correspond to mono-, di-, and tri-substituted products, but
additional analysis is needed.
92
N
-N
-N
-N
'
a.
H
>N
AA
p
-,p
C.
b.
N
Trivalent DBCO conjugates
E50
J 30
S13-3DBCO
-
40
S13-3DBCO-341
-
o
20
.0
IO 10
.0
35
30
40
45
50
55
60
Time (min)
Figure 5.5 Conjugation of carbamate trimers to trivalent DBCO-siRNA. a.) Reaction scheme b.) HPLC analysis of
trivalent DBCO-siRNA ("S13-3DBCO") and the same conjugated to trimer 341 ("S13-3DBCO-341")
N,-'O-O--oO-
NY 0O-ko
N
IH
-N
0,;
N
a.
]
H
H
H
JOOOQ
HH
H
b.
Trivalent BCN conjugates
45
E 40
S13-3BCN
U 335
-S13-3BCN-341
0
o 10
5
0
30
35
40
45
50
55
60
Time (min)
Figure 5.6 Conjugation of carbamate trimers to trivalent BCN-siRNA. a.) Reaction scheme b.) HPLC analysis of
trivalent BCN-siRNA ("S13-3BCN") and the same conjugated to trimer 341 ("S13-3BCN-341")
93
Chapter 6. Conclusions
The field of RNAi therapeutics is coming of age as dozens of RNAi drugs are now in
clinical trials (Table 1). The clinical success of certain conjugate platforms has engendered
renewed interest in the development of conjugate delivery systems. However, when
compared to the vast diversity LNP delivery materials, there are relatively few efficacious
conjugate systems reported in the literature.
The diversity of successful LNP systems derives from the cooperation of two
development approaches: rational design and combinatorial screening. In a rational design
approach, individual materials are synthesized with the aim of testing a particular
hypothesis.
In a combinatorial approach, large libraries of diverse molecules are
synthesized and screened to identify hits whose structures may not have been predictable.
Hypothesis-driven, rational design methods allow us to systematically test our
understanding of the delivery mechanism. However, there are important gaps in our
understanding of the delivery process. In particular the intracellular mechanism of
transport from the endosome to the RNAi machinery in the cytosol is poorly understood.
For many systems, including successful conjugate systems, we have little to no information
about how this process happens or how to make it more efficient. When our understanding
of mechanism is lacking to this degree, it can be difficult to develop hypotheses to test.
Combinatorial screens play the essential role of providing new leads, informing us of what
hypotheses to make. Data gathered from screening results can provide structure-activity
information to guide the rational design of new materials, and mechanistic understanding
gained from rational design experiments can guide the choice of chemical space to explore
in new combinatorial screens. The synergy of these two methods has propelled the field of
nanoparticle-based delivery systems forward at an impressive rate.
Conjugate delivery materials have thus far been developed exclusively by rational
design, as the molecularly defined nature of conjugate systems makes a combinatorial
chemistry approach challenging. The chemical methods used to make lipid libraries are
poorly suited to making conjugate libraries as they do not allow for facile incorporation of
functionalities for siRNA attachment and they rarely provide the high purity that is required
when materials are attached to siRNA in just a 1:1 molar ratio. Without combinatorial
methods for developing new conjugate delivery materials, the discovery of novel conjugate
systems has been much slower than that of nanoparticle systems.
94
In this thesis, we have developed a reliable synthetic strategy for creating libraries
of conjugate delivery materials. In the process we have demonstrated a novel method for
synthesizing sequence-defined polymers that may find applications in a number of fields.
We showed that our methodology is scalable to high-throughput automated synthesis and
we were able to synthesize a library of over 500 novel siRNA conjugates in parallel. We
have done significant work on the development of cell-based assays for the screening of
conjugate libraries. We have also developed platforms for multivalent conjugation of
library materials to siRNA.
Recent clinical success has reinvigorated interest in conjugates, and there is a need
for the discovery of novel conjugate delivery materials. Combinatorial library screening is
essential to providing new leads in the development of these systems, but synthetic
challenges have thus far precluded the use of this method. The synthetic approach
developed in this thesis enables the study of conjugate libraries of vast structural diversity,
including hundreds of conjugate materials in parallel. The study of conjugate libraries,
enabled by this work, has the potential to open vast new directions in the field of siRNA
delivery.
95
Appendix 1. Abbreviationsand Symbols
Oligomer naming system: "F253" refers to fluorous-tagged ("F") trimer with side chains 2,
5, and 3, with 2 being nearest the fluorous tag. The structure of F253 is shown in Figure 2.8.
Structures of monomer side chains corresponding to each number are shown in Figure 2.9a.
"253" refers to the same trimer after removal of the fluorous tag. Addition of an azide
(Figure 2.11) is indicated by "F341-azide" or "341-azide," or is implied if, for example the
trimer is attached to DBCO-modified siRNA.
ASGPR - Asialoglycoprotein receptor
BCN - Bicyclononyne, (1R,8S,9s)-Bicyclo[6.1.0]non-4-yn
C12-200 - Lipid formulation from Anderson et al. (Figure 1.5e) used as positive control for
siRNA delivery
DBCO - Dibenzocyclooctyne
DCM - Dichloromethane
DMF - Dimethylformamide
FSPE - Fluorous solid phase extraction
GalNAc - N-acetylgalactosamine
HPLC- High performance liquid chromatography
LCMS - Liquid chromatography with tandem Mass Spectrometry
Lipo - Lipofectamine 2000 or Lipofectamine RNAiMax
MeCN - Acetonitrile
MeOH - Methanol
NMR - Nuclear magnetic resonance spectroscopy
PBS - Phosphate-buffered saline
siRNA - Short interfering RNA
SPE - Solid phase extraction
TFA - Trifluoroacetic acid
THF - Tetrahydrofuran
TMSI - lodotrimethylsilane
96
Appendix 2. Materials and Methods
Appendix 2.1 List of Materials
Materials
All Fluorous materials (Fluorous CbZ-OSu, Fluorous silica, Fluorous 96-well SPE plates):
Fluorous Technologies
Multiarm PEGs (4 arm PEG NHS ester, 4 arm PEG azide): Creative PEGWorks
RNA sequences were generously provided by Alnylam Pharmaceuticals
Cellular assays: Promega Dual Glo, Promega CellTiter-Glo
Transfection reagents: Lipofectamine 2000, Lipofectamine RNAiMax (Life Technologies)
Cell culture materials: Life Technologies
All other reagents were purchased from Sigma Aldrich
Instrumentation
Flash chromatography: Teledyne ISCO CombiFlash
LCMS : Waters Acquity UPLC with a Xevo QToF LC/MS
NMR - VARIAN Inova-500 NMR Spectrometer with an Oxford Instruments Ltd.
superconducting magnet
H PLC: Agilent HPLC with a Waters OST C18 column
Dynamic Light Scattering: Malavern ZetaSizer
Liquid Handling robot, chemical synthesis: Freeslate CM3
Liquid Handling robot, cellular assays: Tecan Freedom Evo
Microplate reader: Tecan Infinite
Flow cytometry: BD LSR II HTS, BD FACSDIVA
Appendix 2.2 Methods
I.
Synthesis and Purification of monomers
Materials
a. Solid phase extraction through fluorous silica (FSPE).
containing the Fluorous purification tag were routinely purified by solid
In a typical bench scale
phase extraction through Fluorous silica.
purification for 40 g silica (or 2-3 g reaction mixture), the silica was first
washed with 50 mL DMF and then preconditioned with 120 mL 85:15
methanol:water. The reaction mixture was loaded into the silica in up to 10
mL DMF. Non-Fluorous impurities were removed by washing with 150-200
mL 85:15 methanol :water. The fluorous product was then collected by
washing the silica with 200-250 mL THF or acetone. The columns were then
washed with 200 - 300 mL of 1:1:0.01 methanol:THF:TFA; this fraction often
97
b.
contained additional product that could be recovered.
This general
procedure was scaled to accommodate the amount of material to be purified.
Fluorous-CbZ-protected monomers
i. Fluorous-CbZ-protected hydroxyproline. N-[4-(1H, 1H, 2H, 2Hperfluorodecyl) benzyloxy-carbonyloxy] succinimide (16 mmol,
11.12 g) was combined with 1.5 eq. (24 mmol, 3.14 g) trans-4hydroxy-L-proline in 65 mL anhydrous DMF. Triethylamine (2 eq,
3.9 mL) was added and the mixture was stirred at room temperature
for 2 days or until LCMS analysis of the mixture showed that the NHS
ester starting material was completely reacted. The product was
purified by solid phase extraction on Fluorous silica. The reaction
mixture was split into two batches and loaded on to 60 g of fluorous
silica as described. Batches that required further purification after
solid phase extraction were re-purified by normal phase flash
chromatography on a silica column using a gradient of hexane to
ethyl acetate. 90-95% yield.
ii. Fluorous-CbZ-protected monomer 1.
Fluorous-CbZ-protected
hydroxyproline (2 mmol, 1.42 g) was dissolved in 4 mL DMF. Amine
side chain (N,N,N' trimethylethylenediamine, 1.5 eq, 3 mmol, 389
microliters) and DIPEA (2 eq, 4 mmol, 695 pL) were added to the
mixture. PyBOP (1.2 eq, 2.4 mmol, 1.25 g) was dissolved in 4 mL
DMF and added to the reaction. The mixture was stirred at 500 C
overnight. Product was purified by solid phase extraction through
40g of fluorous silica. If further purification was required, the
product was purified by normal phase flash chromatography on a
25-gram silica column using a gradient of dichloromethane to Ultra
solvent (88% dichloromethane, 20% methanol, 2% ammonium
hydroxide). 60% yield.
iii. Fluorous-CbZ-protected monomer 2. The same procedure was
followed as for Fluorous-CbZ-protected monomer 1, using 2methylaminomethyl-1,3-dioxolane (1.5 eq, 3 mmol, 341 ptL) as the
amine side chain. Batches that required further purification after
solid phase extraction were re-purified by reverse-phase HPLC
(Agilent column info here) using a gradient of 5-95% acetonitrile.
60% yield.
iv. Fluorous-CbZ-protected monomer 3. The same procedure was
followed as for Fluorous-CbZ-protected monomer 1, using Nbenzylmethylamine (1.5 eq, 3 mmol, 385 iiL) as the amine side chain.
Batches that required further purification after solid phase
extraction were re-purified by normal phase flash chromatography
on a 25-gram silica column using a gradient of dichloromethane to
80:20 dichloromethane:methanol. 80% yield.
98
v.
Fluorous-CbZ-protected monomer 4. The same procedure was
followed as for Fluorous-CbZ-protected monomer 1, using Nmethyldodecylamine (1.5 eq, 3 mmol, 751 ptL) as the amine side
chain. Batches that required further purification after solid phase
extraction were re-purified by normal phase flash chromatography
on a 25-gram silica column using a gradient of hexane to ethyl
acetate. 80% yield.
vi. Fluorous-CbZ-protected monomer S. The same procedure was
followed as for Fluorous-CbZ-protected monomer 1, using 4morpholinopiperidine (1.5 eq, 3 mmol, 504 mg) as the amine side
chain. Batches that required further purification after solid phase
extraction were re-purified by normal phase flash chromatography
on a 25-gram silica column using a gradient of dichloromethane to
Ultra solvent (88% dichloromethane, 20% methanol, 2% ammonium
hydroxide). 60% yield.
vii. Fluorous-CbZ-protected monomer 6. The same procedure was
followed as for Fluorous-CbZ-protected monomer 1, using 2-[2-(1HImidazol-1-yl) ethyl] piperidine dihydrochloride (1.2 eq, 2.4 mmol,
605 mg) as the amine side chain, with an additional 2.4 eq of DIPEA
to solubilize the amine in a portion of the DMF prior to addition to
the reaction. This reaction was run at room temperature overnight.
Batches that required further purification after solid phase
extraction were re-purified by reverse phase HPLC using a gradient
of 20 - 95% acetonitrile. 60% yield.
viii. Fluorous-CbZ-protected monomer 7. The same procedure was
followed as for Fluorous-CbZ-protected monomer 1, using PEG500amine (1.5 eq, 3 mmol, 1500 mg) as the amine side chain. Batches
that required further purification after solid phase extraction were
re-purified by reversed phase flash chromatography. 60% yield.
c.
CbZ-protected monomers
i. CbZ-protected monomer 1. Z-Hyp-OH (Aldrich) (13.2 mmol, 3.5 g)
Amine side chain (N,N,N'
was dissolved in 20 mL DMF.
trimethylethylenediamine, 1.5 eq, 19.8 mmol, 2.57 mL) and DIPEA (2
eq, 26.4 mmol, 4.59 mL) were added to the mixture. PyBOP (1.2 eq,
15.8 mmol, 8.24 g) was dissolved in 30 mL DMF and added to the
reaction. The mixture was stirred at 50*C overnight. Product was
purified by reverse phase flash chromatography on a C18 column
using a slow gradient from 100% water to 95% acetonitrile. 90%
yield.
ii. CbZ-protected monomer 2. The same procedure was followed as
for CbZ-protected monomer 1, using 2-methylaminomethyl-1,3dioxolane (1.5 eq, 19.8 mmol, 2.25 mL) as the amine side chain.
Product was purified by reverse phase flash chromatography on a
99
d.
C18 column using a slow gradient from 100% water to 95%
acetonitrile. 88% yield.
iii. CbZ-protected monomer 3. The same procedure was followed as
for CbZ-protected monomer 1, using N-benzylmethylamine (1.5 eq,
19.8 mmol, 4.96 mL) as the amine side chain. Product was purified
by normal phase flash chromatography on a silca column using a
gradient of hexane to ethyl acetate (5-95%). 95% yield.
iv. CbZ-protected monomer 4. The same procedure was followed as
for CbZ-protected monomer 1, using N-methyldodecylamine (1.5 eq,
19.8 mmol, 2.56 mL) as the amine side chain. Product was purified
by normal phase flash chromatography on a silca column using a
gradient of hexane to ethyl acetate (5-95%). 95% yield.
v. CbZ-protected monomer 5. The same procedure was followed as
for CbZ-protected monomer 1, using 4-morpholinopiperidine (1.5
eq, 19.8 mmol, 3.33 g) as the amine side chain. Product was purified
by normal phase flash chromatography on a silica column using a
gradient dichloromethane to Ultra solvent (88% dichloromethane,
20% methanol, 2% ammonium hydroxide). 80% yield.
vi. CbZ-protected monomer 6. Z-Hyp-OH (Aldrich) (15.1 mmol, 4 g)
was dissolved in 10 mL DMF. Amine side chain (2-[2-(1H-Imidazol1-yl) ethyl] piperidine dihydrochloride, 1.2 eq, 18.1 mmol, 4.56 g)
was combined with DIPEA (3.5 eq, 52.8 mmol,9.2 mL) in 15 mL DMF
and added to the reaction mixture. PyBOP (1.2 eq, 18.1 mmol, 9.42
g) was dissolved in 35 mL DMF and added to the reaction. The
mixture was stirred at room temperature overnight. Product was
purified by reverse phase flash chromatography on a C18 column
using a slow gradient from 100% water to 95% acetonitrile. 55%
yield.
Removal of CbZ protecting group to yield monomers 1-6.
i. Monomer 1. CbZ-protected monomer 1 was dissolved in DMF to a
concentration of 0.25 M and added to palladium catalyst (30 wt% on
activated carbon) at a ratio of roughly 3:1 CbZ-protected monomer:
palladium on carbon. A hydrogen balloon was attached to the
reaction flask, which was stirred for at least 3 h at room
temperature. After reaction, the hydrogen balloon was removed and
-5 mL of triethylamine was added to the mixture, which was
allowed to stir for another 10 min. The catalyst was removed by
filtration and rinsed thoroughly with methanol and water. The
eluent was concentrated and purified by reverse phase flash
chromatography on a C18 column using several column volumes of
100% water followed by a slow gradient up to 20% acetonitrile.
53% yield.
100
II.
ii. Monomer 2. The same reaction and purification methods were used
as for monomer 1. 74% yield.
iii. Monomer 3. The same reaction procedure used as for monomer 1.
Product was purified by reverse phase flash chromatography on a
C18 column using a gradient of 5-95% acetonitrile. 90% yield.
iv. Monomer 4. The same reaction procedure used as for monomer 1.
Product was purified by reverse phase flash chromatography on a
C18 column using a gradient of 5-95% acetonitrile. 80% yield
v. Monomer 5. The same reaction procedure used as for monomer 1.
Product was purified by reverse phase flash chromatography on a
C18 column using several column volumes of 100% water followed
by a slow gradient up to 50% acetonitrile. 70% yield.
vi. Monomer 6. The same reaction procedure used as for monomer 1.
Product was purified by reverse phase flash chromatography on a
C18 column using several column volumes of 100% water with 1%
triethylamine, followed by a slow gradient up to 20% acetonitrile
with 1% triethylamine. 65% yield.
Oligomer synthesis and purification
a. Bench top synthesis and purification for individual batches. Fluorous
CbZ-protected monomer was weighed in an oven-dried glass vial which was
fitted with a rubber septum and fitted with a line flowing dry nitrogen. Ten
to twelve molar equivalents of carbonyldiimidazole were added in DMF and
the mixture was stirred to dissolved the starting monomer to a final
concentration of 0.1 M. After stirring for one hour at room temperature, the
mixture was cooled to 0' C before quenching. Excess CDI was quenched
with a 1:1 mixture of water and acetonitrile such that a minimum amount of
water was added to quench excess CDI (1 molar equivalent of water: excess
CDI). The mixture was then evaporated to dryness, then resuspended in
benzene and evaporated twice to remove residual water. The material was
placed back under nitrogen flow and the second monomer (6 equivalents,
deprotected) was added in DMF to a final concentration of 0.1 M starting
monomer. The mixture was stirred at room temperature overnight, then
purified by solid phase extraction through fluorous silica, as described
previously. Yields ranged from 85 - 100%
Synthesis steps were
b. High-throughput synthesis and purification.
carried out by a Freeslate CM3 robot, using 1-mL glass vials in 96-well plate
format. All vials, stir bars, and reaction blocks were dried overnight at 1500
C before use. The interior of the robot was purged with nitrogen such that
moisture levels were below 150 ppm for the duration of the synthesis.
Reagent addition scripts were written in Automation Studio and executed
using the Library Studio software.
i. Stock solutions. Monomers were dissolved in anhydrous DMF to a
concentration of 0.24M for fluorous CbZ-protected monomers and
101
0.58M for deprotected monomers, with the exception of Monomer 3,
which was dissolved in 1:1 DMF:THF to 0.12M, and Monomer 5,
which was dissolved in 1:1 DMF: MeCN to 0.12M. CDI was dissolved
in anhydrous DMF until the solution was saturated. Solutions were
stored in the dry environment of the robot.
ii. Monomer coupling. Using the single tip lookahead feature of the
Freeslate CM3, 12 umol of starting (fluorous CbZ-protected)
monomer was added to each well (100 tL of stock solution for
monomers 3 and 5, 50 iL for other monomers). Wells were stirred
at room temperature while 125 ptL of saturated CDI solution was
added to each well using the 6-channel tip or single tip lookahead
feature. After addition, the wells were stirred for 1 h. Plates were
then cooled to 0 0 C and 5 pL of a 1:1 mixture of acetonitrile: water
was added to each well. Alternatively, quenching of CDI could be
done outside the robot by placing the plates on ice and adding MeCN:
water using a multichannel pipette. After quenching, plates were
removed from the robot and dried on a Genevac EZ-2 for -1h.
Benzene (50-100 ptL) was added to each well using a multichannel
pipette and the plates were stirred briefly before being dried again
for -45 min. Benzene was added one more time and the plates were
dried thoroughly before being replaced in the robot's glove box.
Deprotected monomer stock solution (125 iiL) was added to each
well using the robot's 6-tip pipette or single-tip lookahead feature.
The wells were stirred overnight at 22'C.
iii. Oligomer purification. Oligomers were purified by solid phase
extraction through fluorous silica in 96-well plate format. Before
use, fluorous silica plates were washed with 1 mL THF per well,
followed by 3 mL DMF, using a multichannel pipette to transfer
solvents and vacuum to pull solvents through the silica. Each well
was then conditioned with 3 mL of 85:15 DMF:water before sample
loading. The reaction mixture was then loaded onto the silica in 125
iL DMF. The silica washed with 3 mL 85:15 DMF:water to remove
excess reagents, and then the product was eluted with 4 mL of THF.
Wells were then washed with 5 mL of 1:1:0.01 THF: MeOH:TFA.
This fraction was collected and a selection of wells was analyzed by
LCMS as it often contained additional product that could be
combined with the previous fraction. Wells were then washed with
an additional 2 mL of DMF before their next use. Product fractions
were concentrated and transferred to glass 1-mL vials in 96-well
plates for the next synthesis step. Yields ranged from 70 - 99%.
iv. Coupling to Azido-PEG-amine. CDI coupling was carried out in the
same manner as for oligomer synthesis, with the exception that only
102
v.
vi.
2 eq of amine (11-Azido-3,6,9-trioxaundecan-1-amine ) were added
to the mixed imidazolide intermediate.
The same purification
Purification of Oligomer-PEG-azides.
method was used as for oligomer purification, but with additional
washing: after loading the crude
Removal of fluorous purification tag. Protected oligomers were
dissolved in 80 [tL of 1:1 acetonitrile:benzene using anhydrous
solvents. In a dry nitrogen environment, the reaction vessels were
cooled to 0"C before the addition of 25 piL of a 1M solution of
iodotrimethylsilane (TMSI) in dichloromethane. After stirring for 45
min, the reactions were quenched by the addition of 50 iL of
methanol. Thiosulfate bound to polystyrene beads (Sigma) was
added to each well and the solutions were incubated overnight to
remove residual iodine. The solution of deprotected product was
then removed from the well with a pipette, concentrated, and
purified by FSPE. Deprotected product was collected in the aqueous
fraction.
trifluoroacetic acid for 3h.
siRNA conjugation, purification, and characterization
Purified oligomer-PEG-azides were
a. siRNA conjugation conditions.
dissolved in 500 ptL MeCN. The DBCO-modified siRNA sequence was
dissolved in PBS and stored at a concentration of 1 mg/mL. Oligomer-PEGazide stock solution was added to siRNA stock solution, PBS, and acetonitrile
such that the final concentration of siRNA was 0.25 mg/mL in 30% MeCN.
For high-throughput synthesis, reactions were prepared in PCR plates and
allowed to stir on an orbital shaker overnight. Plates were then frozen at
-
III.
Deprotection was also achieved by heating to 50'C in
b.
c.
20'C for 1-3h and allowed to thaw in a refrigerator.
Purification of siRNA conjugates. The reaction mixture was dried on a
Genevac EZ-2 and then redissolved in 30% MeCN/PBS to a concentration of
1 mg/mL. One tenth volume of 3M sodium acetate buffer solution and then
5 volumes of 200 proof ethanol were added to this. The plates were covered
and frozen at 20'C overnight, then moved to a -80'C freezer for 1-3 h. For
high-throughput synthesis, the plates were then centrifuged at 3,200 rcf for
90 minutes, carefully decanted, and dried. For single-batch synthesis in
microcentrifuge tubes, tubes were centrifuged at 15,000 rcf for 20 min
before decanting.
HPLC characterization of conjugates. Samples were diluted in 0.1M
triethylamino acetate (TEAA) buffer to a concentration of 1 pg/50 [tL. Five
hundred ng - 1 sg of siRNA was injected for each run on a Waters XBridge
OST C18 (2.5 m, 2.1 x 50 mm) column and run on a gradient of 10-95% B
(A = 0.1M TEAA in water, B = 0.1M TEAA in MeCN) in 60 minutes at a flow
rate of 0.2 mL/min.
103
Appendix 3. Supplementary Data
Appendix 3.1 H NMR and "C NMR shifts
Monomer 1
Fluorous protected:
0
IH NMR (500 MHz, Chloroform-d) 6 7.37 - 7.27 (m, 2H), 7.19 (m, 2H),
5.18 - 5.01 (m, 2H), 4.88 - 4.75 (m, I H), 4.56 (d, J= 17.5 Hz, 1 H), 3.80
3.71 (m, 1H), 3.70 -- 3.52 (m, 2H), 3.47 - 3.23 (m, IH), 3.15 (d, J= 0.9 Hz,
1H), 2.99 (d, J= 0.9 Hz, 1H), 2.90 (m, 3H), 2.50 - 2.30 (m, 2H), 2.29 - 2.16
(m, 7H), 2.16 - 2.03 (m, 3H).
OA
-
CBF17
N
fO
NMR (126 MH z, CDC13) 6 173.09, 155.47, 139.58, t35.64, 129.70, 128.97, 120.66, 118.77, 116.76,
113.84, 111.52, 109.11, 70.20, 67.49, 57.79, 56.15, 50.85, 48.66, 46.94, 46.04, 39.38, 35.82, 33.51, 26.75.
13C
Deprotected:
HN
""'OH
N
'H NMR (500 MHz, Methanol-d 4 ) 6 4.41 (m, I H), 4.19 (m, 1H), 3.60 - 3.51 (m,
I H), 3.55 - 3.41 (m, 1H), 3.32 - 3.21 (m, 1H), 3.09 (s, 2H), 2.97 (s, IH), 2.78
(m, 1H), 2.64 - 2.41 (m, 2H), 2.30 (d, J= 10.6 Hz, 6H), 2.20 - 2.06 (m, IH),
1.82 (m, IH).
N
3
C NMR (126 MHz, Methanol-d 4) 6 173.89, 72.65, 57.22, 56.15, 55.08, 46.13, 45.17, 44.96, 40.01,
34.40.
Monomer 2
Fluorous protected:
/
-/
N
0
N
'OH
'H NMR (500 MHz,
2H), 5.19 - 4.76 (m,
(m, 211), 3.51 - 3.26
2.86 (m, 3H), 2.44 -
Chloroform-d) 6 7.35 - 7.24 (m, 2H), 7.24 - 7.17 (m,
4H), 4.67 - 4.55 (m, 1H), 4.03 - 3.72 (m, 4H), 3.71 - 3.53
(m, 1H), 3.24 (s, IH), 3.07 (s, IH), 3.02 (s, 1H), 2.96
2.34 (m, 2H), 2.34 - 2.00 (m, 2H).
-
C8F- 7 -
NMR. (126 M Hz, CDC13) 6 173.41,1,54.87, 139.00,1.35.05, 128.97, 128.36, 120.22, 11.8.10, 116.01,
113.19, 110.77, 108.57, 102.26, 70.18, 69.41, 66.89, 65.03, 55.36, 51.89, 50.79, 39.01, 36.58, 32.88, 26.16.
13C
Deprotected:
104
'H NMR (500 MHz, Methanol-d 4 ) 6 5.08 - 4.93 (m, 1H), 4.49 - 4.40 (m, 1H), 4.31 (m,
1H), 4.03 - 3.91 (m, 2H), 3.91 - 3.80 (m, 2H), 3.71 - 3.46 (m, 2H), 3.33 - 3.23 (m, 1H),
3.16 (s, 1H), 3.03 (s, 1H), 2.80 (m, 1H), 2.22 - 2.06 (m, 1H), 1.80 (m, 1H).
0
-
N
o
13
HN
C NMR (126 MHz, cd 3od) 6 175.27, 103.14, 73.25, 66.23, 65.89, 57.61, 55.76, 51.90,
40.61, 36.70.
'OH
Monomer 3
Fluorous protected:
/
C 8F1 7
0
S
H NMR (500 MHz, Chloroform-d) 6 7.54 - 6.91 (m, 9H), 5.24 - 4.99 (m,
2H), 4.99 - 4.70 (m, 2H), 4.68 - 4.38 (m, 2H), 3.86 - 3.53 (m, 2H), 3.04 (m,
2H), 2.91 (m, 2H), 2.85 (d, J = 19.5 Hz, 1H), 2.45 - 2.21 (m, 3H), 2.16 (m,
I1H).
N
"C NMR (126 MHz, CDC1 3) 6 172.71, 155.04, 139.05, 136.71, 135.18,
129.07, 128.49, 128.18, 127.91, 126.54, 119.86, 118.35, 116.38, 113.19,
110.86, 108.71, 70.19, 66.91, 55.48, 53.21, 51.55, 38.93, 34.67, 31.87, 26.06.
Deprotected:
HN
'H NMR (500 MHz, Methanol-d4 ) 6 7.35 - 7.13 (m, 5H), 4.77 (m, I H), 4.66 - 4.41 (m,
3H), 3.39 (m, 1H), 3.35 - 3.22 (m, IH), 2.88 (d, J = 4.1 Hz, 3H), 2.49 (m, IH), 1.99 (m,
0
N-
I H).
13
-
C NMR (126 MHz, cd 3od) 6 174.95, 137.99, 129.95, 129.68, 128.75, 128.49, 127.67,
73.41, 57.80, 55.82, 52.78, 40.90, 34.65.
Monomer 4
Fluorous protected:
/
\
0 =
o=K
1H NMR (500 MHz, Chloroform-d) 6 7.35 - 7.28 (m, 1H), 7.27 - 7.23
(m, 1H), 7.23 - 7.14 (m, 2H), 5.28 - 4.93 (m, 2H), 4.91 - 4.67 (m,
1H), 4.59 (d, J= 18.2 Hz, 1H), 3.92 - 3.74 (m, 1H), 3.74 - 3.57 (m,
1H), 3.57 - 3.42 (m, 1H), 3.36 - 3.19 (m, 1H), 3.12 (s, 1H), 3.04
-
C8 F 17
S..OH
C 12H 2 5 -N
105
2.79 (m, 4H), 2.47 - 2.30 (m, 2H), 2.28 - 2.06 (m, 211), 1.63 - 1.49 (m, 2H), 1.44 - 1.03 (m, 18H), 0.89 (t,
J= 6.9, 2.6 Hz, 3H).
' 3 C NMR (126 MHz, CDC1 3) 6 172.32, 154.89, 138.91, 135.05, 128.91, 128.31, 120.08, 118.12, 116.01,
113.17, 110.80, 108.51, 69.68, 66.90, 55.46, 50.03, 48.55, 38.79, 35.03, 34.15, 32.87, 31.94, 29.66, 29.37,
28.79, 28.45, 26.92, 26.17, 22.68, 13.97.
Deprotected:
'H NMR (500 MHz, Methanol-d4) 6 4.81 - 4.70 (m, 1 H), 4.63 - 4.57 (m, I H), 3.52 - 3.25
(m, 4H), 3.06 (s, 2H), 2.98 (s, 1H), 2.59 - 2.43 (m, 1H), 2.05 - 1.90 (m, 1H), 1.72 - 1.52
(m, 2H), 1.36 - 1.27 (m, 18H), 0.94 - 0.86 (m, 3H).
OH
HN
"C NMR (126 MHz, cd 3od) 6 168.97, 71.30, 58.65, 54.95, 50.24, 38.99, 34.47, 32.89,
30.63, 30.60, 30.55, 30.53, 30.50, 30.31, 28.89, 27.57, 23.56, 14.40.
N
0
k2H"25
Monomer 5
Fluorous protected:
'H NMR (500 MHz, Chloroform-d) 6 7.37 - 7.25 (m, 2H), 7.25 - 7.15 (m, 2H),
-
5.30 - 4.96 (m, 2H),
1H), 3.82 - 3.64 (m,
(m, 2H), 2.75 - 2.48
1.69 (m, 2H), 1.62 -
0
O=<
N
0
"OH
N-
4.95 - 4.78 (m, 1H), 4.70 - 4.42 (m, 2H), 4.18 - 3.83 (in,
5H), 3.61 - 3.45 (m, 1H), 3.31 - 2.96 (m, 1H), 2.95 - 2.76
(m, 4H), 2.47 - 2.30 (m, 3H), 2.29 - 1.99 (m, 3H), 1.97
1.05 (m, 2H).
-
C 8F1
3
C NMR (126 MHz, cd3od) 6 171.29, 155.67, 139.55, 135.66, 128.96, 128.51,
121.17, 118.75, 116.61, 113.33, 111.55, 108.84, 69.86, 67.59, 62.33, 55.78,
50.11, 45.20, 42.29, 39.17, 33.51, 29.68, 28.88, 26.77.
NN
0
Deprotected:
HN
N
N
00
OH
'H NMR (500 MHz,
- 4.20 (n, IH), 4.06
- 3.05 (m, IH), 2.88
- 2.42 (m, I H), 2.18
- 1.25 (m, 211).
Methanol-d 4) 6
- 3.96 (m, IH),
- 2.78 (m, 1H),
- 2.07 (m, 1 H),
4.57
3.72
2.77
2.08
-
4.48 (m, IH), 4.46 3.66 (m, 4H), 3.33 2.64 (m, I H), 2.65 1.92 (m, I H), 1.92 -
4.36
3.24
2.55
1.72
(m,
(m,
(m,
(m,
1H), 4.29
2H), 3.18
4H), 2.54
I H), 1.53
3
1C
NMR (126 MHz, cd 3od) 6 172.51, 73.23, 67.88, 62.95, 62.88, 57.56, 55.66,
50.97, 50.85, 45.24, 42.55, 40.80, 29.75, 28.97.
Monomer 6
106
Fluorous protected:
'H NMR (500 MHz, Chloroform-d) 6 7.98 (d, J= 33.5 Hz, 1H), 7.39 - 7.27
(m, 2H), 7.20-7.13 (m, 2H), 7.10 (s, 1H), 7.05 (d,J= 1.7 Hz, 1H), 5.11
5.02 (m, 2H), 4.92 - 4.79 (m, 2H), 4.58 (dq, J= 4.9, 2.3 Hz, 1H), 4.11
3.91 (m, 2H), 3.84 (d, J= 14.0 Hz, 1H), 3.81 - 3.73 (m, 1H), 3.63 (dt, J
11.5, 1.8 Hz, IH), 3.21 - 3.10 (m, IH), 2.95 - 2.80 (m, 2H), 2.42 - 2.25 (m,
4H), 2.25 - 2.18 (m, 1H), 2.17 - 2.03 (m, 1H), 1.81 - 1.69 (m, 2H), 1.69
1.50 (m, 3H), 1.49 - 1.27 (m, 1H).
-
C 8F 17
-
N
*"OH
-
N
( N/
3
1C
NMR (126 MHz, CDCI3) 6 171.50, 155.07, 139.04, 136.89, 135.14, 128.95, 128.49, 128.25, 125.55,
119.62, 118.23, 116.07, 113.26, 110.85, 108.63, 69.91, 66.92, 60.53, 55.63, 50.12, 45.48, 40.96, 37.86,
32.89, 31.47, 29.33, 26.07, 21.16, 19.18.
Deprotected:
'OH
-
N
'H NMR (500 MHz, Methanol-d4 ) 6 7.79 - 7.65 (m, IH), 7.27 - 7.11 (m, 1H), 7.01
6.89 (m, 1H), 4.41 (ddq, J= 11.0, 5.5, 2.7 Hz, 2H), 4.25 - 4.01 (m, 2H), 3.95 (dt, J=
12.6, 7.3 Hz, 2H), 3.80 - 3.73 (m, 1H), 3.33 - 3.17 (m, 2H), 2.85 - 2.72 (m, 1H), 2.43
- 2.27 (m, 1H), 2.25 - 2.08 (m, LH), 2.07 - 1.87 (m, 1H), 1.87 - 1.68 (m, 2H), 1.67
1.37 (m, 4H).
-
HN
CN
3
1C
NMR (126 MHz, CD30D) 6 172.80, 138.56, 128.96, 120.66, 72.97, 58.02, 55.58,
49.85, 48.08, 45.16, 41.15, 32.15, 29.81, 27.12, 19.96.
Trimers
F122
-
-
Chloroform-d) 6 7.37 - 7.27 (m, 2H), 7.23 - 7.13 (m, 2H), 5.41 - 4.94 (m, 5H), 4.94
4.37 (m, IH), 4.20 - 4.00 (m, lH), 4.00 - 3.82 (m, 7H), 3.82 - 3.57 (m, 7H), 3.57
3.19 (m, 2H), 3.19 - 3.09 (m, 3 H), 3.09 - 2.77 (m, IOH), 2.57 - 2.42 (m, 2H), 2.42
1.99 (m, 4H).
-
'H NMR (500 MHz,
4.55 (m, 4H), 4.56 3.36 (m, 3H), 3.36 2.22 (m, 8H), 2.22 -
F144
-
-
Chloroform-d) 6 7.35 - 7.24 (m, 2H), 7.22 - 7.15 (m, 2H), 5.25 - 4.93 (m, 4H), 4.92
3.57 (m, 811), 3.54 - 3.35 (m, 2H), 3.30 - 3.04 (m, 5H), 3.00 - 2.95 (m, 2H), 2.94
2.68 (m, 2H), 2.55 - 2.42 (m., 2H), 2.40 - 2.19 (m, 9H), 2.14 - 1.94 (m, 3H), 1.75
1.00 (m, 36H), 0.86 (td, J= 7.0, 2.3 Hz, 611).
-
'H NMR (500 MHz,
4.42 (m, 6H), 3.88 2.86 (m, 4H), 2.85 1.44 (m, 4H), 1.44 -
3
1C
NMR (126 MHz, CDC13) 6 172.35, 154.24, 139.69, 135.50, 129.09, 120.88, 118.79, 116.51, 113.65,
111.45, 108.94, 78.00, 77.94, 77.74, 77.49, 74.32, 73.03, 70.17, 67.59, 57.60, 56.62, 55.68, 53.28, 51.33,
50.50, 48.90, 46.38, 40.79, 40.15, 39.61, 37.79, 36.31, 35.24, 34.49, 33.54, 32.60, 30.32, 30.04, 29.46,
29.12, 27.75, 27.47, 26.86, 23.36, 14.78.
F146
107
'H NMR (500 MHz, Chloroform-d) 6 7.50
2H), 7.07 - 7.01 (m, I H), 6.95 (d, J= 11.4
(m, 2H), 3.94 - 3.44 (m, 1OH), 3.22 - 3.10
(m, IOH), 2.15 - 1.83 (m, 6H), 1.75 - 1.61
3H).
(d, J= 16.9 Hz, IH), 7.37 - 7.28 (m, 2H), 7.19 (d, J= 7.6 Hz,
Hz, I H), 5.26 - 4.94 (m, 4H), 4.87 - 4.35 (m, 5 H), 4.17 - 4.00
(m, 3H), 3.02 - 2.82 (m, 6H), 2.76 - 2.45 (m, 5H), 2.38 - 2.20
(m, 4H), 1.46 - 1.37 (m, 2H), 1.25 (s, 18H), 0.91 - 0.83 (m,
3
C NMR (126 MHz, CDC1 3) 6 172.32, 154.92, 139.80, 137.99, 135.14, 129.10, 129.05, 119.51, 116.45,
113.23, 110.97, 108.87, 106.88, 103.84, 70.94, 69.75, 67.56, 64.46, 56.77, 55.80, 54.71, 53.26, 46.35,
46.21, 43.82, 41.56, 36.96, 32.61, 31.09, 30.35, 30.06, 26.88, 24.55, 23.89, 23.39, 23.37, 19.49, 14.81.
F214
-
'H NMR (500 MHz, Chloroforin-d) 6 7.31 - 7.22 (m, 2H), 7.19 - 7.14 (m, 2H), 5.26 - 5.00 (m, 4H), 4.98
4.60 (m, 4H), 4.42 (s, IH), 3.94 - 3.55 (m, 1 H), 3.39 (s, 4H), 3.25 - 3.15 (m, 2H), 3.11 - 3.00 (m, 3 H),
2.98 - 2.78 (m, 9H), 2.48 - 2.39 (m, 1H), 2.38 - 2.07 (m, 12H), 2.08 - 1.85 (m, IH), 1.47 (s, 2H), 1.24 (d,
J= 31.8 Hz, 18H), 0.90 - 0.78 (m, 3 H).
3
C NMR (126 MHz, CDC 3) 6 173.04, 154.70, 139.67, 135.41, 129.13, 120.59, 118.78, 116.63, 113.69,
111.38, 109.18, 102.86, 74.48, 70.59, 69.81, 67.65, 65.68, 57.82, 56.93, 55.83, 53.06, 51.02, 49.34, 46.31,
39.84, 38.42, 37.08, 35.66, 34.67, 33.51, 32.56, 30.16, 27.34, 26.83, 23.33, 14.74.
F215
Chloroform-d) 6 7.35 - 7.26 (m, 2H), 7.22 - 7.18 (m, 2H), 5.41 - 5.17 (m, 2H), 5.16
4.61 (m, 4H), 4.60 - 4.37 (i, 2H), 4.28 - 4.02 (m, IH), 4.01 - 3.55 (m, 15 H), 3.55 (s,
3H), 3.11 - 2.77 (m, 1OH), 2.76 - 2.62 (in, I H), 2.54 (s, 4H), 2.48 - 2.14 (m, I1 H),
1.77 - 1.58 (m, 1H).
-
H NMR (500 MHz,
5.02 (m, 2H), 5.01 3H), 3.32 - 3.11 (m,
2.13 - 1.78 (m, 5H),
13
C NMR (126 MHz, CDC 3) 6 173.01, 155.41, 139.63, 135.90, 129.10, 120.98, 118.80, 116.66, 114.02,
111.45, 109.27, 102.88, 74.91, 74.12, 70.36, 67.77, 65.70, 62.36, 57.62, 55.79, 53.46, 52.51, 51.44, 50.51,
45.91, 42.37, 38.45, 37.16, 33.52, 30.39, 29.99, 29.39, 28.84, 28.52, 26.87, 23.39, 22.15.
F341
-
'H NMR (500 MHz, Chloroform-d) 6 7.38 - 7.28 (m, 3H), 7.26 - 6.95 (m, 6H), 5.32 - 4.96 (m, 4H), 4.92
4.60 (m, 3H), 4.58 - 4.39 (m, 3H), 3.83 - 3.51 (m, 6H), 3.41 (s, 4H), 3.28 - 3.17 (m, IH), 3.10 - 2.58 (m,
12H), 2.49 - 2.20 (m, 11 H), 2.16 - 2.03 (m, 2H), 1.99 - 1.87 (m, 1H), 1.70 - 1.38 (m, 2H), 1.38 - 0.94 (m,
18H), 0.86 (t, J= 8.3, 4.7, 1.4 Hz, 3H).
3
C NMR (126 MHz, CDC1 3) 6 172.47, 154.63, 139.65, 137.35, 135.49, 129.65, 129.31, 129.07, 128.47,
127.42, 120.81, 118.71, 116.50, 113.91, 111.55, 109.10, 74.40, 70.81, 69.77, 67.52, 57.94, 56.69, 55.84,
53.36, 52.00, 51.19, 50.23, 48.59, 46.29, 40.19, 39.53, 38.99, 37.67, 36.61, 34.74, 33.59, 32.58, 30.31,
30.03, 29.47, 27.51, 26.95, 23.34, 14.75.
F365
108
-
H NMR (500 MHz, Chloroform-d) 6 7.49 - 6.63 (m, 12H), 5.39 - 5.03 (m, 4H), 5.01 - 4.64 (m, 3H), 4.62
- 4.37 (m, 4H), 4.23 - 3.88 (m, 3H), 3.88 - 3.46 (m, I1 H), 3.45 - 3.38 (m, 3H), 3.19 - 2.95 (m, 3H), 2.95
2.73 (m, 4H), 2.71 - 2.20 (m, 11H), 2.18 - 1.76 (m, 6H), 1.74 - 1.42 (m, 4H), 1.43 - 1.16 (m, 3H).
13C NMR (126 MHz, CDC13) 6 171.78, 154.63, 140.19, 137.44, 136.86, 135.46, 130.05, 129.59, 129.32,
129.08, 128.33, 127.22, 120.45, 120.04, 118.92, 116.63, 113.80, 111.43, 109.22, 74.35, 70.67, 67.81,
62.50, 55.99, 53.35, 52.11, 51.23, 50.42, 45.29, 44.30, 42.51, 42.14, 39.37, 37.87, 34.90, 33.40, 32.30,
30.09, 28.87, 26.96, 25.31, 19.26.
F425
-
'H NMR (500 MHz, Chloroform-d) 6 7.33 - 7.20 (m, 2H), 7.20 - 7.12 (m, 2H), 5.29 - 5.00 (m., 4H), 4.97
4.65 (m, 311), 4.63 - 4.35 (m, 3H), 4.06 - 3.79 (m, 5H), 3.77 - 3.50 (m, IOH), 3.39 (d, J= 1.0 H1z, 5H),
3.17 - 2.94 (m, 4H), 2.93 - 2.85 (m, 4H), 2.84 - 2.73 (m, 2H), 2.61 - 2.12 (m, 11 H), 2.11 - 1.71 (m, 5H),
1.67 - 1.38 (m, 2H), 1.36 - 1.03 (m, 18H), 0.83 (t, J= 6.3, 1.8 Hz, 3H).
"C NMR (126 MHz, CDC1 3) 6 171.92, 154.47, 139.80, 135.48, 129.68, 129.07, 120.87, 118.74, 116.60,
113.40, 111.29, 109.19, 102.50, 74.12, 70.66, 69.80, 67.65, 65.69, 62.48, 55.90, 53.03, 51.30, 50.45, 49.05,
44.52, 42.18, 39.92, 37.55, 36.21, 34.57, 33.47, 32.55, 30.25, 30.16, 30.00, 29.02, 27.48, 26.82, 23.31,
14.72.
F443
-
'H NMR (500 MHz, Chloroform-d) 6 7.40 - 7.02 (m, 9H), 5.26 - 4.96 (m, 4H), 4.95 - 4.72 (m, 2H), 4.70
4.35 (m, 4H), 3.87 - 3.57 (m, 6H), 3.40 (s, 6H), 3.13 - 2.97 (m, 3H), 2.93 - 2.82 (m, 6H), 2.82 - 2.68 (m,
2H), 2.39 - 2.13 (m, 6H), 1.75 - 1.34 (m, 4H), 1.34 - 0.89 (m, 36H), 0.85 (t, J= 6.9, 2.1 Hz, 6H).
1
3C
NMR (126 MHz, CDC1 3) 6 172.61, 154.63, 139.61, 136.74, 135.41, 129.90, 129.47, 129.02, 128.28,
127.00, 120.44, 118.75, 116.64, 114.01, 111.56, 109.45, 74.51, 70.21, 67.52, 55.95, 53.39, 50.98, 48.83,
39.75, 37.98, 34.71, 33.51, 32.57, 30.30, 30.14, 30.11, 30.02, 29.45, 27.53, 26.83, 23.33, 14.73.
F465
-
-
'H NMR (500 MHz, Chloroform-d) 6 7.42 - 7.32 (m, IH), 7.33 - 7.22 (m, 2H), 7.23 - 7.13 (m, 2H), 7.12
6.90 (m, 2H), 5.41 - 4.96 (m, 5H), 4.97 - 4.68 (m, 2H), 4.68 - 4.36 (m, 3H), 4.27 - 3.89 (m, 3H), 3.89
3.38 (, i5H), 3.37 - 2.98 (m, 4H), 2.98 - 2.75 (m, 5H), 2.53 (d, J= 7.8 Hz, 6H), 2.43 - 2.15 (m, 6H),
2.14 - 1.78 (m, 5H), 1.78 - 1.44 (m, 5H), 1.46 - 1.26 (m, 2H), 1.27 - 1.07 (m, 18H), 0.89 - 0.76 (m, 3H).
13
C NMR (126 MHz, CDC1 3) 6 172.11, 170.75, 154.66, 139.95, 136.94, 135.29, 129.97, 129.07, 120.46,
120.07, 118.92, 116.50, 113.55, 111.43, 109.20, 74.55, 70.43, 67.80, 62.54, 55.91, 53.20, 51.21, 50.45,
48.89, 45.28, 44.30, 42.32, 39.40, 37.77, 35.47, 33.55, 32.57, 30.21, 29.00, 27.55, 26.85, 25.72, 23.34,
19.39, 14.75.
F534
109
-
'H NMR (500 MHz, Chloroform-d) 6 7.55 - 6.78 (m, 9H), 5.41 - 4.93 (m, 4H), 4.93 - 4.30 (m, 6H), 4.24
3.91 (m, I H), 3.91 - 3.51 (m, 9H), 3.42 (d, J= 0.6 Hz, 4H), 3.14 - 2.76 (m, 8H), 2.74 - 2.45 (m, 7H), 2.25
(s, 6H), 2.15 - 1.59 (m, 6H), 1.28 (s, 20H), 0.91 - 0.76 (m, 311).
13
C NMR (126 MHz, CDC1 3) 6 172.66, 170.53, 154.52, 139.77, 137.29, 135.67, 130.09, 129.39, 129.05,
128.24, 127.18, 120.63, 118.77, 116.63, 113.58, 111.27, 109.19, 74.71, 71.43, 69.96, 67.67, 62.28, 55.90,
53.28, 51.36, 50.33, 45.07, 42.30, 39.77, 37.24, 35.37, 34.37, 33.48, 32.59, 30.32, 30.03, 29.05, 27.17,
23.62, 14.78.
F542
Chloroform-d) 6 7.37 - 7.22 (m, 2H), 7.20 - 7.11 (m, 214), 5.23 - 5.05 (m, 311), 5.04
4.34 (m, 4H), 4.18 - 3.81 (m, 5H), 3.81 - 3.49 (m, 1IH), 3.38 (s, 6H), 3.19 - 2.92 (m,
4H), 2.77 - 2.53 (m, 3H), 2.52 - 2.22 (m, 9H), 2.05 - 1.80 (m, 4H), 1.66 - 1.37 (m,
18H), 0.83 (t, J= 7.1, 3.6, 1.7 Hz, 3H).
-
'H NMR (500 MHz,
4.71 (m, 3H), 4.68 4H), 2.91 - 2.80 (m,
2H), 1.37 - 0.94 (m,
13
C NMR (126 MHz, CDCI 3) 6 172.93, 170.53, 154.93, 154.19, 139.57, 135.54, 128.98, 120.65, 118.70,
116.61, 113.66, 111.32, 109.17, 102.81, 74.34, 70.17, 67.68, 65.72, 62.35, 55.86, 52.88, 51.17, 50.42,
48.45, 45.12, 42.24, 39.77, 37.37, 34.84, 33.59, 32.55, 30.05, 28.92, 27.19, 23.31, 14.71.
F632
-
H NMR (500 MHz, Chloroform-d) 6 7.60 - 7.39 (m, IH), 7.38 - 6.67 (m, 1IH), 5.37 - 4.88 (m, 5H), 4.86
- 4.59 (m, 4H), 4.58 - 4.48 (m, IH), 4.47 - 4.31 (m, IH), 3.98 - 3.65 (m, 9H), 3.65 - 3.52 (m, 3H), 3.41 (d,
J= 0.5 Hz, 8H), 3.23 - 3.12 (m, IH), 3.10 - 2.73 (m, 8H), 2.73 - 2.47 (m, 3H), 2.42 - 2.17 (m, 5H), 2.15
2.10 (m, 1H), 2.06 - 1.89 (m, 2H), 1.77 (s, 111).
13
C NMR (126 MHz, CDC1 3) 6 171.54, 154.31, 139.73, 137.67, 137.23, 135.53, 129.63, 129.33, 128.95,
128.34, 127.09, 119.70, 118.90, 116.63, 114.00, 111.38, 109.47, 107.03, 102.71, 74.55, 69.60, 67.55,
65.71, 56.04, 52.76, 51.31, 47.16, 44.83, 41.40, 40.17, 37.79, 35.16, 33.05, 30.14, 27.02, 19.84.
Appendix 3.2 LCMS analysis of trimer and hexamer synthesis
Trimers synthesized in Chapter 2: LC traces @ 214 nm. Mass spectra show ions
contributing to peak in MS trace of trimer.
110
F122
I
0.50
1:
1KF122_fresh_1CuL 176 (1 596) Crm (172 181)
1308.3196
100-i
2.50
225
200
1.2575
1.00
75
L1309.3483
2175
3.00
p Time
3.25
TOF MIS ES+
1 14e6
654.6742
250
........ .. ,.
500
. . . . . . . .1.
1
000
.
.
H-
1000
750
1
1250
1500
1500
0.75
&SO
1M
125
150
1
1 75
.
200
....
F144
, .2.2 . .
225
RKF144_dil 284 (2.547) Cm (22:288)
1472.6122
100-
m/z
.1
F14
Fl
F144
11750
2.50
2.7
2.75
Time
3.00
32
1: TOF MS ES+
9.28e
.1473.6342
1474.6632
736.8069
0250
500
750
1000
1250
1500
1750
111
Fl
F146
F146
F14
----------------0
050
1
125
175
150
2.00
2.25
250
2.75
3.00
3.25
1: TOF MS ES+
1.96e6
RK_F146fresh_3uL 194 (1.741) Cm (189:200)
726.7626
100
Iol
-727.7617
1452.5959
I
I1-1
UL
750
500
250
II
m/Z
1000
1250
1500
1750
F214
1.00
0 50
2.00
1 0
F21
3.00
2.50
F214 F2
PKF214_fresh-dil_2 242 (2.175) Cm (240:245)
1390.5074
100-
1: TOF MS ES+
4.92e5
1 390.4775-J1 392.5259
695.7391
0-
250
-%-
'I
500
750
100I
1250
-pP-FR-;
m/Z
1500
112
F215
0,50
I1.00
200
1 .so
F215
250
F2
F21
RKF215_fresh_dil 152 (1,368) Cm (150 157)
1 TOF MS ES+
8 07eD
681.1881
100,1
,,681.6757
L1362.4178
1361.4189
.,682.1821
0-
500
250
I7
750
F253
,....,I..
1250
1000
m/Z
250
300
1760
F2
F25 F253
050
,. .. . , .. ,
Timem
075
100
150
125
175
200
225
275
PKF253 _dil_5uL 195 (1 .749) Cm (195:198)
1380.5576
100,
325
1: TOF MS ES+
5.25e5
1381.5792
,1382.5966
690.7449
0
m/Z
260
500
750
1000
1250
1500
1750
113
F341
F341
F3
F34
[341
[341
F3
F34
0.O
PR
05
10
2A1.''2
12
.5 0 ,5
200
2.50
2.25
2.75
F341 BuL 259 (2.317) Cm (256:261)
.
1: TCF MS ES+
1 03e6
1394.6890
100-
1Tlime
3.25
1
3,00
l1395.7103
,1396.7494
I
0
250O
5
0
750
1000
1250
1750
1500
,I
M/Z
F365
L;
1.5k
F365
1.00
0.50
2.00
2.50
F36
3.00
F3
RK_F365_5uL 161 (1.454) Cm (159:164)
721.7708
100]
1: TOF MS ES+
5 92e5
L722.2648
1442.6532
Li 443.6576
,722.7689
I
0 -I
250
500
L
750
1
m/Z
1000
1250
1500
1750
114
F425
0 50
0.75
1.00
1.50
15
1.75
200
2
2 25
. 275
3.00
8.56e5
1458.7059
100
,1459.7343
j+
1460.7714
I I
i-i
500
750
1500
1250
1750
F4
1.50
1.00
M/Z
_________________I
1000
F443
0.50
3.2f
1: TOF MS ES+
RK_F425_8uL 252 (2.258) Cm (250:254)
250
F4
F42
F425
2.00
2.50
3.50
3.00
RK F443 long_dil 380 (3.392) Cm (378:386)
1491.8455
100-
F44, F443
Time
4.00
1: TOF MS ES+
4.40e5
,1492.8517
02.0
A
m/z
A
600
750
1000
1250
1500
1750
115
F465
a
1 00
.
0.50
1.50
I
F4
F46
F465
RkF465
3.00
2.50
2.00
1 TOF M-AS ES+
7 39e5
uL 205 (1 .844) Cm (202 207)
760.8513
L761 3452
1520.8215
-------
0
250
500
750
1500
1250
1000
F5 F53
F534
.
..
., 1.00
,0*
...
1',,......,.
, ,...0.75
0.50
1.25
1.50
m/z
1760
F534
1.75
2.00
2.25
2.50
2.75
3.00
.2i Time
3.25
1: TOF MS ES+
RKF534_5uL 251 (2.250) Cm (248:253)
8.18e5
1462.5201
100-
!146 4.5765
731.7589
0
500
750
1000
1250
1500
1760
I
m/z
116
F54/F542
F5
FS 42
Time
0.75
0.50
1.50
1.25
1.00
200
1,75
225
250
100-
.4
F6
F632
0.75
0.50
rri/
. .I
1060
750
500
1.25
1.00
325
579e5
1458.7223
250
300
1; TOF MS ES+
RKF542_6uL 241 (2.166) Cm (240:243)
0
2,75
1250
1500
1750
2.
.0
F63/F632
1.5
.75
20
RKF632_fresh 201 (1.810) Cm (194:203)
389.3875
100-,
27
.0
.71 Tim2
3.25
1: TOF MS ES+
7 .21e6
1390.3912
695.1772
0
250
500
750
1000
1250
1500
1750
117
F25326
F253262
F2532,F253262
214 0,50Da
Range 3 1440-1
3.0e-1
2.001.0e-1--
0.60
0.00
1.00
1.20
1.40
160
1.80
200
220
2.40
2.60
2.80
3.00
3.20
RKF2532_Ac_6uL 192 (1.724) Cm (189:195)
1636.7245 6.93e5
100-
RKF26326_Ac _1uL 150 1.416) Cm (156 1162)
3 24e5
978.4323
100
PKF253264_Ac 201 (1 810) Cm (199:203)
4 33e5
1147.5789
100-
0
500
M/Z
1000
1500
118
F4433 F443341, F44334
F443341
214 O.5ODa
342e-1
6.0e-14.0e-1 7
2 50 3009
2.0e-1-
0.00.50
i
1.00 ,.5 ,2.00
i , ' ' 1.50'
1.00
Time
1 '2.0
Ib I
2.50
3.00
3.50
4.00
4.50Ti e
RK_F4433_Ac 384 (3.437) Cm (379:387) 1: TOF MS ES+
3,29e5
1751.9911,
1773.9697
10
-1775.9764
875.3495 1302.7145
I
-
[1-
"
.
.
RK F44334_Ac 417 (3.724) Cm (408:417)1 TOF MS ES+
4.29e4
875.3499
100-
K876.3543
.1046.0966
-Pp-L
k
.
A.A
RKF443341_Ac_dil 322 (2.887) Cm (319:327)
3 44e5
1166.6953
100-
1166.1857,
,1167.1918
,1167.6902
0
600
L1000
.,.
1500
i
m/z
119
F54236, F542363
F542363
F5423
2'14 U.bUUa
Range: 3 12 7e-1
3.0 e-1
2.0 e-1
1 .0 e-1
0.0
0.
0.50
0.75
.
1.00
1.25
1.50
1.75
2.00
2.26
2.50
2.75
3.00
T me
3.25
RKF5423_Ac_diI3 246 (2 208) Cm (244:248)
6.83e5
1718.8962,
,1720.9270
859.8965
0
;,.
...-.
., .- ..y 4v, 1
. '
11
0- 1
RKF54236_Acdil 202 (1 818) Cm (201 205)
1019.5101
5.986e5
10U-U
,1020.0222
,1020.5206
i
------,I; -----nIIII' ' II
I4
I
-
OL
I
RK F542363_Ac dil 206 (1 652) Cm (202:210)
1149.5846
5 08e
1001149.0811,
,1150.0858
11
I..
-
II
,1150.5851
50 I ll ' 100101
15 0
m/z
120
High throughput synthesis (Chapter 3):
F122
RKFMmerl
]
26-Nov-2014
1: TOF MS ES+
RKFMmerl 175 (1.571) Cm (171:140)
2.3996
796.0820
4 Diode Array
214
100
Range 1 320
1.0
[.1139
nfl
0
40
1 .
120
1.0
1.40
2.0
220
2.60
2.40
2.80
3.40400
ri-q
3.20
RKan. 1.8. Pmmc_0iay
Range.
:2
RK L2_1AB dimer di2 182 (1.641) Cm (178:186)
1212 3157-e
1052.3145
100
1054.4021
9.230e-1
7 04
F.6-
50.0e-1
00
1. TOF MS ES+
a0 :796.2473
.~__________
0.80
1.20
1.0
141
40
2.0
14
2.20
240
2.40
4: Diode Array
214
Range: 5,8999-1
RK L2_1AB _nimer
5.0e-1
040
0.
1.4
140
0-14
140
0
3.20
3.0
2.40
1 TOF MS ES+
183 (1 649) Cm (179.188)
1 12e6
1052.3486
1308.4878
RKL21A8_utriumer
100
654.7222
2.40
220
2.40
2.0
2.60
300
RK_L2_1A0_tinmerPEG-1OuL
214
Range
'1310.5492
3.20
5 292e-1
4: Diode Array
R) L2 1A8 trimerPEG
10---
tOuL 194
(1.741) Cm (109 201)
1296.5067 1552,6423
0.0
.0
040O
.
.
.2.
1.40
120
100
8 27e0
1553.0747
O~e-I
140
1460
...
2400
220O
240
240O
.....
3.00
2.40
7673
645.7134
Time
320
20
4
1554.3144
112343
76.793
1123 4324
700 103
1250
1750
1500
mnz
27,jun,2013
HT synthesis: Removal of purification tag using TMSI
STOF MS
1,
an
s
1579.45591
DenprTeV
4: Dod1 May
8%0+1
SM
Ix
6,0-
50-U
10
6 1
R.ng,
151048
Mass: 1578.5
1
23021
797.1402
8".1704
302
100+1
1
4.
IO
1,9
2w
211)
2,A
i
RE~pprohtA34
2
4 3
DA-)M 3 20
Z
2w
7
3,I0
"14.3
637.43330","1
"u'4N28w
e
.Eer
1084.3104
M + PEG + Na
[,..3i.
2.5-1
Mass: 998.5
7T
21
27'
57 76 21
23
170
2
2110
30
3.10
187,058
low
3104
30.^''^+^20..1
13 35230
32
ol 369 1071370 2Q23 'I.....6W560264
260
'
4Do
'
VA421Q
sOJ
107314
%, 34116
144416
60
'
1653 011727 21301
,
RK
10bD
121
120D
'
nb
1600
Appendix 3.3 High-throughputpurificationof siRNA conjugates
HT purification of siRNA conjugates: Yield
Large, hydrophobic
carbamate (F333-azide)
+
siLuc-DBCO
GalNAc-siTTR-DBCO
Concdntration
7.00
I
0
1.00
2.00
"k
1.00i
-
2211
-
0.00-230
2
250 260
270 280 20 3Ui
Wavelength mn
310
320
330
340
Sample
(% of control)
Al (siLuc-F333)
96.2
B2 (siTTR-F333)
92.6
C3 (siLuc-F333)
103.7
D4 (siTTR-F333)
89.1
E5 (siLuc-F333)
93.4
F6 (siTTR-F333)
100.1
G7 (siLuc-F333)
96.5
H8 (siTTR-F333)
92.1
Average % yield
95.5
3g0
Carbamate alone
(5-10x the concentration in the
conjugation reaction)
Absorbance spectra taken by NanoDrop spectrophotometry of recovered siRNA conjugates
after ethanol precipitation. Eight test samples and three controls, which were made up to a
concentration corresponding to 100% theoretical yield. Carbamate alone refers to F333
azide modification.
122
Appendix 3.4 HPLC analysis of siRNA conjugates
Conjugation of Fluorous-protected carbamates to siRNA
140
E
120
W 100
80
e 60
-513
S13-F425
S13-F542
40
0
20
03
13
23
43
33
53
Time (min)
Conjugation of deprotected carbamates to siRNA
90
-Unmodified
E 80
siRNA
70
60
50
C 40
. 30
020
--
425
--
542
S10
0
5
10
15
35
30
Time (min)
25
20
40
45
50
55
60
Conjugation of deprotected carbamates to siRNA
140
-Unmodified
siRNA
-341
E 120
2100
N
%
-443
80
C 60
o 40
S20
0
5
10
15
20
25
35
30
45
40
50
60
55
Time (min
Comparison of Fluorous-protected and deprotected conjugates for the same trimer:
siRNA conjugation by HPLC
160
E 140
N
@J
W
0U
.0
0
120
-Unmodified
siRNA
100
-214-siRNA
80
60
40
20
0
-S13-F214
5
10
15
20
25
30
35
40
45
50
55
60
Time (min)
123
siRNA conjugation by HPLC
180
E 160
140
Unmodified siRNA
-
120
--
365-siRNA
S100
80
m
F365-siRNA
-
060
40
20
0
5
10
15
20
25
35
40
45
50
55
60
40
45
50
55
60
40
45
50
55
60
Time (min)
siRNA conjugation by HPLC
100
E
80
-
Unmodified siRNA
% 60
40
--
215-siRNA
--
F215-siRNA
0
0
10
5
15
20
25
30
35
Time (min)
siRNA conjugation by HPLC
100
E
80
60
-Unmodified siRNA
i
40
-
542-siRNA
--
F542-siRNA
.0
4
20
0
5
10
15
20
25
30
35
Time (min)
124
Appendix 3.5 Synthesis of multivalentDBCO and BCN linkers
Starting material 4-arm PEG2000-NHS, analyzed by LCMS, was observed to separate
into two fractions of observed ionized masses 1255 and 1326. Assuming this is (m+2)/2,
the average mass of these two fractions would be 2508 and 2650, respectively. Mass
spectra are shown below:
1: TOF MS ES+
4.64e3
RK_4PEGNHS-std 202 (1.819) Cm (199:204)
100-
526.2375 614.2972
482.2097,2
,746.3916
,943.1784
1326.2477
436.8330
OL
b-#NNYFq -lwl1-I
1: TOF MiS ES+
4.45e3
- K_4PEG_NHS_std 193 (1.732) C:rri (189:194)
852.1204
100896.1545
793.0830
1255.7 155
482.2162,70.2737
,1 343.7802
1 409.8275
1111.
0
The expected mass of fully substituted 4-arm PEG BCN (Figure 5.2, bottom) is then 3344
and 3486 for these two fractions, respectively. (m+2)/2 = 1673 and 1744. We isolated two
fractions with approximately these masses. Mass spectra are shown below:
125
1: TOF MS ES+
1 48e4
RKN4165 _BCN_t15 259 (2.316) Cm (251:26;).
409.2156
100-
541.3085
I I
629.3728
673.4063
691.4463
,,746.4561
.779.5088
1612.5868
824.5240
Ii.
0"
I Ii
IMJ11M~I
1 TOF MS ES+
1 2Fe4
RkN4165_BCN_t14 259 (2 316) Crm (250-262)
409.2083 541.3003
1u62.34
629.3648
673.3978
717,4310
..735.4689
779.5023
,823.5262
824.51 76
I24. 911.5886
1148.7468
1634.5918
1722.6543
2
.1810.7227
AJ911-588
200
400
,181
600
800
1000
1200
1400
1600
/.7227
1800
The expected mass of fully substituted 4-arm PEG DBCO (Figure 5.2, top) is then 3152
and 3294 for these two fractions, respectively. (m+2)/2 = 1577 and 1648. We isolated
three fractions with approximately these masses. Mass spectra are shown below. The
middle fraction (ionized mass = 1648) was used in siRNA conjugation experiments.
126
1: TOF MS ES+
1.25e4
RK_N4165_DBCOt13 297 (2.656) Cm (291305)
981.8841
100-1
111
1040.6005
952.5297
0
1494. 3894
,1560.4440
1604.4799
-I .
"I Imyv-
1 TOF MS
1. 53e4
P1N4165 DBCO_t12 296 (2.639) Crri (285:308)
835.7128 1084.6349
1001128.6672
813.6952 ,
.1143.3467
802.6871,,
'1158.0253
791.6752,
1648.5165
1 692.5463
1715.0632
1759.1033
S1173.0353
,1781.115 4
769.6570,
I- L
RKN4165_['BCOt1l1 290 (259 7) Cm (282:298)
100-
857 .7286
846. 7208
835. 7080
820.44
400
600
946.0546
1172.7037
\/
.957.0656
1
92 i
U
200
1: TOF MS ES+
&52e3
901.7662
879
.7455, ,923.7844
-
8-
,1055.2777
69
I 1 0 .9573
,A084.6360
,1099.6469
689.4061
583.3304
800
.1
Ii
. . . . . ,1. .. , ,.. .1. , . .
1000
1781.1196
1200
m/z
1400
1600
1800
1H NMR comparisons of the starting 4-arm PEG -NHS and the two cyclooctyne products
confirmed that these fractions were fully substituted, based on the peak integral ratio of
cyclooctyne signatures to PEG protons.
4 arm PEG NHS: 'H NMR (500 MHz, Methanol-d 4) 6 4.6 (s, 2H), 3.6 - 3.7 (in, 58H), 3.45 (s, 2H),
2.8 - 2.9 (m, 4H).
4 arm PEG DBCO: 'H NMR (500 MHz, Methanol-d4) Phenyl protons: 6 7.2 - 7.7 (m, 9H), PEG
protons: 3.55 - 3.7 (m, 58H), 3.45 (s, 2H)
4 arm PEG BCN: 'H NMR (500 MHz, Methanol-cd 4) PEG protons: 3.55 - 3.8 (in, 68H), 3.4-3.45 (i,
4H), BCN protons: 2.1 - 2.4 (m, 7H), 1.56 - 1.7 (m, 2H), 1.3 - 1.5 (m, I H), 0.9 - 1.05 (m, 2H).
127
Appendix 4. References
1.
Kanasty, R. L., Whitehead, K. a, Vegas, A. J. & Anderson, D. G. Action and reaction: the
biological response to siRNA and its delivery vehicles. Mol. Ther. 20, 513-24 (2012).
2.
Kanasty, R., Dorkin, J. R., Vegas, A. & Anderson, D. Delivery materials for siRNA
therapeutics. Nat. Mater. 12, 967-77 (2013).
3.
Yin, H. et al. Non-viral vectors for gene-based therapy. Nat. Rev. Genet. 15, 541-555
(2014).
4.
Morrissey, D. V et al. Potent and persistent in vivo anti-HBV activity of chemically
modified siRNAs. Nat. Biotechnol. 23, 1002-7 (2005).
5.
Okumura, A., Pitha, P. M. & Harty, R. N. ISG15 inhibits Ebola VP40 VLP budding in an
L-domain-dependent manner by blocking Nedd4 ligase activity. Proc. Natl. Acad. Sci.
U. S. A. 105, 3974-9 (2008).
6.
Alnylam RNAi Roundtable: Conjugate Delivery. (2012). at
<http://www.alnylam.com/capella/wp-content/uploads/2012/12/ALNYRNAiRoundtable-ConjugateDelivery-Dec-14-20 12.pdf>
7.
Shen, H., Sun, T. & Ferrari, M. Nanovector delivery of siRNA for cancer therapy.
Cancer Gene Ther. 19, 367-73 (2012).
8.
Davis, M. E. et al. Evidence of RNAi in humans from systemically administered siRNA
via targeted nanoparticles. Nature 464, 1067-70 (2010).
9.
Whitehead, K. a, Langer, R. & Anderson, D. G. Knocking down barriers: advances in
siRNA delivery. Nat. Rev. Drug Discov. 8, 129-38 (2009).
10.
Pei, Y. & Tuschl, T. On the art of identifying effective and specific siRNAs. Nat.
Methods 3, 670-676 (2006).
11.
Reynolds, A. et al. Rational siRNA design for RNA interference. Nat. Biotechnol. 22,
326-30 (2004).
12.
Lu, Z. J. & Mathews, D. H. OligoWalk: an online siRNA design tool utilizing
hybridization thermodynamics. Nucleic Acids Res. 36, W104-8 (2008).
13.
Schwarz, D. S. et al. Asymmetry in the assembly of the RNAi enzyme complex. Cell
115, 199-208 (2003).
14.
Tafer, H. et al. The impact of target site accessibility on the design of effective siRNAs.
Nat. Biotechnol. 26, 578-83 (2008).
15.
Nykanen, a, Haley, B. & Zamore, P. D. ATP requirements and small interfering RNA
structure in the RNA interference pathway. Cell 107, 309-21 (2001).
128
16.
Vaish, N. et al. Improved specificity of gene silencing by siRNAs containing unlocked
nucleobase analogs. Nucleic Acids Res. 39, 1823-1832 (2010).
17.
Bramsen, J. B. et al. A screen of chemical modifications identifies position-specific
modification by UNA to most potently reduce siRNA off-target effects. Nucleic Acids
Res. 38, 5761-73 (2010).
18.
Laursen, M. B. et al. Utilization of unlocked nucleic acid (UNA) to enhance siRNA
performance in vitro and in vivo. MoL. Biosyst. 6, 862 (2010).
19.
Weitzer, S. & Martinez, J. The human RNA kinase hClp1 is active on 3' transfer RNA
exons and short interfering RNAs. Nature 447, 222-6 (2007).
20.
Chen, P. Y. et al. Strand-specific 5'-O-methylation of siRNA duplexes controls guide
strand selection and targeting specificity. RNA 14, 263-74 (2008).
21.
Zuhorn, I. S., Engberts, J. B. F. N. & Hoekstra, D. Gene delivery by cationic lipid
vectors: overcoming cellular barriers. Eur. Biophys. J. 36, 349-62 (2007).
22.
Chiu, Y.-L. siRNA function in RNAi: A chemical modification analysis. Rna 9, 10341048 (2003).
23.
Khvorova, A., Reynolds, A. & Jayasena, S. D. Functional siRNAs and miRNAs exhibit
strand bias. Cell 115, 209-16 (2003).
24.
Ameres, S. L., Martinez, J. & Schroeder, R. Molecular basis for target RNA recognition
and cleavage by human RISC. Cell 130, 101-12 (2007).
25.
Brown, K. M., Chu, C.-Y. & Rana, T. M. Target accessibility dictates the potency of
human RISC. Nat. Struct. Mol. Biol. 12, 469-70 (2005).
26.
Koberle, C., Kaufmann, S. H. E. & Patzel, V. Selecting effective siRNAs based on guide
RNA structure. Nat. Protoc. 1, 1832-9 (2006).
27.
Patzel, V. et al. Design of siRNAs producing unstructured guide-RNAs results in
improved RNA interference efficiency. Nat. Biotechnol. 23, 1440-4 (2005).
28.
Huesken, D. et aL. Design of a genome-wide siRNA library using an artificial neural
network. Nat. Biotechnol. 23, 995-1001 (2005).
29.
Jackson, A. L. & Linsley, P. S. Recognizing and avoiding siRNA off-target effects for
target identification and therapeutic application. Nat. Rev. Drug Discov. 9, 57-67
(2010).
30.
Whitehead, K. A., Dahlman, J. E., Langer, R. S. & Anderson, D. G. Silencing or
Stimulation? siRNA Delivery and the Immune System. Annu. Rev. Chem. Biomol. Eng.
2, 77-96 (2011).
129
31.
Feve, B. & Bastard, J.-P. The role of interleukins in insulin resistance and type 2
diabetes mellitus. Nat. Rev. Endocrinol. 5, 305-11 (2009).
32.
Garcia-Sastre, A. & Biron, C. a. Type 1 interferons and the virus-host relationship: a
lesson in detente . Science 312, 879-82 (2006).
33.
Locksley, R. M., Killeen, N. & Lenardo, M. J. The TNF and TNF receptor superfamilies:
integrating mammalian biology. Cell 104, 487-501 (2001).
34.
Akira, S., Uematsu, S. & Takeuchi, 0. Pathogen recognition and innate immunity. Cell
124, 783-801 (2006).
35.
Marques, J. T. et al. A structural basis for discriminating between self and nonself
double-stranded RNAs in mammalian cells. Nat. Biotechnol. 24, 559-65 (2006).
36.
Takeuchi, 0. & Akira, S. Pattern recognition receptors and inflammation. Cell 140,
805-20 (2010).
37.
Akira, S., Takeda, K. & Kaisho, T. Toll-like receptors: critical proteins linking innate
and acquired immunity. Nat. Immunol. 2, 675-80 (2001).
38.
Kawai, T. & Akira, S. The role of pattern-recognition receptors in innate immunity:
update on Toll-like receptors. Nat. Immunol. 11, 373-84 (2010).
39.
Robbins, M., Judge, A. & MacLachlan, I. siRNA and innate immunity. Oligonucleotides
19, 89-102 (2009).
40.
Meurs, E. et al. Molecular Cloning and Characterization Human Double-Stranded
Kinase Induced by Interferon of the Protein. Cell 62, 379-390 (1990).
41.
Kato, H. et al. Cell type-specific involvement of RIG-I in antiviral response. Immunity
23, 19-28 (2005).
42.
Judge, A. D. et al. Sequence-dependent stimulation of the mammalian innate immune
response by synthetic siRNA. Nat. Biotechnol. 23, 457-62 (2005).
43.
Heil, F. et al. Species-specific recognition of single-stranded RNA via toll-like receptor
7 and 8. Science 303, 1526-9 (2004).
44.
Hornung, V. et al. 5'-Triphosphate RNA is the ligand for RIG-I. Science 314, 994-7
(2006).
45.
Pichlmair, A. et al. RIG-1-mediated antiviral responses to single-stranded RNA
bearing 5'-phosphates. Science 314, 997-1001 (2006).
46.
Kim, D.-H. et al. Interferon induction by siRNAs and ssRNAs synthesized by phage
polymerase. Nat. Biotechnol. 22, 321-5 (2004).
130
47.
Elm6n, J. et al. Locked nucleic acid (LNA) mediated improvements in siRNA stability
and functionality. Nucleic Acids Res. 33, 439-47 (2005).
48.
Langkjaer, N., Pasternak, A. & Wengel, J. UNA (unlocked nucleic acid): a flexible RNA
mimic that allows engineering of nucleic acid duplex stability. Bioorg. Med. Chem. 17,
5420-5 (2009).
49.
Heidel, J. D., Hu, S., Liu, X. F., Triche, T. J. & Davis, M. E. Lack of interferon response in
animals to naked siRNAs. Nat. Biotechnol. 22, 1579-82 (2004).
50.
Kornek, M. et al. 1,2-dioleoyl-3-trimethylammonium-propane (DOTAP)-formulated,
immune-stimulatory vascular endothelial growth factor a small interfering RNA
(siRNA) increases antitumoral efficacy in murine orthotopic hepatocellular
carcinoma with liver fibrosis. Mol. Med. 14, 365-73 (2008).
51.
Cubillos-ruiz, J. R. et al. Polyethylenimine-based siRNA nanocomplexes reprogram
tumor-associated dendritic cells via TLR5 to elicit therapeutic antitumor immunity.
119, (2009).
52.
Akinc, A., Thomas, M., Klibanov, A. M. & Langer, R. Exploring polyethyleniminemediated DNA transfection and the proton sponge hypothesis.J. Gene Med. 7, 657-63
(2005).
53.
Nguyen, D. N. et al. Drug delivery-mediated control of RNA immunostimulation. Mol.
Ther. 17, 1555-62 (2009).
54.
Elbashir, S. M. et al. Duplexes of 21-nucleotide RNAs mediate RNA interference in
cultured mammalian cells. Nature 411, 494-8 (2001).
55.
Holen, T. et al. Tolerated wobble mutations in siRNAs decrease specificity, but can
enhance activity in vivo. Nucleic Acids Res. 33, 4704-10 (2005).
56.
Jackson, A. L. et al. Expression profiling reveals off-target gene regulation by RNAi.
Nat. Biotechnol. 21, 635-637 (2003).
57.
Fedorov, Y. Off-target effects by siRNA can induce toxic phenotype. Rna 12, 11881196 (2006).
58.
Lai, E. C. Micro RNAs are complementary to 3' UTR sequence motifs that mediate
negative post-transcriptional regulation. Nat. Genet. 30, 363-4 (2002).
59.
Doench, J. G. & Sharp, P. a. Specificity of microRNA target selection in translational
repression. Genes Dev. 18, 504-11 (2004).
60.
Wu, L., Fan, J. & Belasco, J. G. MicroRNAs direct rapid deadenylation of mRNA. Proc.
Nat. Acad. Sci. U. S. A. 103, 4034-9 (2006).
131
61.
Farh, K. K.-H. et al. The widespread impact of mammalian MicroRNAs on mRNA
repression and evolution. Science 310, 1817-21 (2005).
62.
Krek, A. et al. Combinatorial microRNA target predictions. Nat. Genet. 37, 495-500
(2005).
63.
Lewis, B. P., Shih, I., Jones-Rhoades, M. W., Bartel, D. P. & Burge, C. B. Prediction of
mammalian microRNA targets. Cell 115, 787-98 (2003).
64.
Lim, L. P. et al. Microarray analysis shows that some microRNAs downregulate large
numbers of target mRNAs. Nature 433, 769-73 (2005).
65.
Xie, X. et al. Systematic discovery of regulatory motifs in human promoters and 3'
UTRs by comparison of several mammals. Nature 434, 338-45 (2005).
66.
Semizarov, D. et al. Specificity of short interfering RNA determined through gene
expression signatures. Proc. NatL. Acad Sci. U. S. A. 100, 6347-52 (2003).
67.
Jackson, A. L. et al. Position-specific chemical modification of siRNAs reduces 'offtarget' transcript silencing. RNA 12, 1197-205 (2006).
68.
Birmingham, A. et al. 3' UTR seed matches, but not overall identity, are associated
with RNAi off-targets. Nat. Methods 3, 199-204 (2006).
69.
Boese, Q. et al. Mechanistic insights aid computational short interfering RNA design.
Methods Enzymol. 392, 73-96 (2005).
70.
Naito, Y., Yoshimura, J., Morishita, S. & Ui-Tei, K. siDirect 2.0: updated software for
designing functional siRNA with reduced seed-dependent off-target effect. BMC
Bioinformatics10, 392 (2009).
71.
Petri, S. et al. Increased siRNA duplex stability correlates with reduced off-target and
elevated on-target effects. Rna 737-749 (2011). doi:10.1261/rna.2348111
72.
Grimm, D. et al. Fatality in mice due to oversaturation of cellular microRNA / short
hairpin RNA pathways. Nature 441, 537-541 (2006).
73.
Castanotto, D. et al. Combinatorial delivery of small interfering RNAs reduces RNAi
efficacy by selective incorporation into RISC. Nucleic Acids Res. 35, 5154-64 (2007).
74.
Vickers, T. a, Lima, W. F., Nichols, J. G. & Crooke, S. T. Reduced levels of Ago2
expression result in increased siRNA competition in mammalian cells. Nucleic Acids
Res. 35, 6598-610 (2007).
75.
Koller, E. et al. Competition for RISC binding predicts in vitro potency of siRNA.
Nucleic Acids Res. 34, 4467-76 (2006).
132
76.
Khan, A. A. et al. Transfection of small RNAs globally perturbs gene regulation by
endogenous microRNAs. Nat. Biotechnol. 27, 549-55 (2009).
77.
Yi, R., Doehle, B. P., Qin, Y., Macara, I. G. & Cullen, B. R. Overexpression of exportin 5
enhances RNA interference mediated by short hairpin RNAs and microRNAs. RNA 11,
220-6 (2005).
78.
Hutvaigner, G., Simard, M. J., Mello, C. C. & Zamore, P. D. Sequence-specific inhibition
of small RNA function. PLoS Biol. 2, E98 (2004).
79.
Grimm, D. et al. Argonaute proteins are key determinants of RNAi efficacy, toxicity,
and persistence in the adult mouse liver.J. Clin. Invest. 120, 3106-19 (2010).
80.
Giering, J. C., Grimm, D., Storm, T. A. & Kay, M. A. Expression of shRNA from a tissuespecific pol 11 promoter is an effective and safe RNAi therapeutic. MoL. Ther. 16,
1630-6 (2008).
81.
Akinc, A. et al. A combinatorial library of lipid-like materials for delivery of RNAi
therapeutics. Nat. Biotechnol. 26, 561-9 (2008).
82.
Zimmermann, T. S. et al. RNAi-mediated gene silencing in non-human primates.
Nature 441, 111-4 (2006).
83.
Love, K. T. et al. Lipid-like materials for low-dose, in vivo gene silencing. Proc. Nat/.
Acad. Sc. U. S. A. 107, 1864-9 (2010).
84.
Lv, H., Zhang, S., Wang, B., Cui, S. & Yan, J. Toxicity of cationic lipids and cationic
polymers in gene delivery.j. Control. Release 114, 100-9 (2006).
85.
Kedmi, R., Ben-Arie, N. & Peer, D. The systemic toxicity of positively charged lipid
nanoparticles and the role of Toll-like receptor 4 in immune activation. Biomaterials
31, 6867-75 (2010).
86.
Soenen, S. J. H., Brisson, A. R. & De Cuyper, M. Addressing the problem of cationic
lipid-mediated toxicity: the magnetoliposome model. Biomaterials30, 3691-701
(2009).
87.
Dokka, S., Toledo, D., Shi, X., Castranova, V. & Rojanasakul, Y. Oxygen radicalmediated pulmonary toxicity induced by some cationic liposomes. Pharm. Res. 17,
521-5 (2000).
88.
Bottega, R. & Epand, R. M. Inhibition of protein kinase C by cationic amphiphiles.
Biochemistry 31, 9025-30 (1992).
89.
Solodin, I. et al. A novel series of amphiphilic imidazolinium compounds for in vitro
and in vivo gene delivery. Biochemistry 34, 13537-44 (1995).
133
90.
Van der Woude, I. et al. Novel pyridinium surfactants for efficient, nontoxic in vitro
gene delivery. Proc. NatL. Acad. Sci. U. S. A. 94, 1160-5 (1997).
91.
Leventis, R. & Silvius, J. R. Interactions of mammalian cells with lipid dispersions
containing novel metabolizable cationic amphiphiles. Biochim. Biophys. Acta 1023,
124-32 (1990).
92.
Semple, S. C. et al. Rational design of cationic lipids for siRNA delivery. Nat.
Biotechnol. 28, 172-6 (2010).
93.
Xu, Y., Hui, S. W., Frederik, P. & Szoka, F. C. Physicochemical characterization and
purification of cationic lipoplexes. Biophys. J. 77, 341-53 (1999).
94.
Moghimi, S. M. et aL. A two-stage poly(ethylenimine)-mediated cytotoxicity:
implications for gene transfer/therapy. Mol. Ther. 11, 990-5 (2005).
95.
Malek, A. et al. In vivo pharmacokinetics, tissue distribution and underlying
mechanisms of various PEI(-PEG)/siRNA complexes. Toxicol. Appl. Pharmacol. 236,
97-108 (2009).
96.
Beyerle, A., Irmler, M., Beckers, J., Kissel, T. & Stoeger, T. Toxicity pathway focused
gene expression profiling of PEI-based polymers for pulmonary applications. Mol.
Pharm. 7, 727-37 (2010).
97.
Merkel, 0. M. et aL. Polymer-related off-target effects in non-viral siRNA delivery.
Biomaterials32, 2388-98 (2011).
98.
Omidi, Y. et al. Toxicogenomics of non-viral vectors for gene therapy: a microarray
study of lipofectin- and oligofectamine-induced gene expression changes in human
epithelial cells.J. Drug Target. 11, 311-23 (2003).
99.
Akhtar, S. & Benter, I. Toxicogenomics of non-viral drug delivery systems for RNAi:
potential impact on siRNA-mediated gene silencing activity and specificity. Adv. Drug
Deliv. Rev. 59, 164-82 (2007).
100.
Kircheis, R., Wightman, L. & Wagner, E. Design and gene delivery activity of modified
polyethylenimines. Adv. Drug Deliv. Rev. 53, 341-58 (2001).
101.
Soutschek, J. et al. Therapeutic silencing of an endogenous gene by systemic
administration of modified siRNAs. Nature 432, 173-8 (2004).
102.
LAYZER, J. M. In vivo activity of nuclease-resistant siRNAs. RNA 10, 766-771 (2004).
103.
Buyens, K., Demeester, J., De Smedt, S. S. & Sanders, N. N. Elucidating the
encapsulation of short interfering RNA in PEGylated cationic liposomes. Langmuir
25, 4886-91 (2009).
134
104.
Akinc, A. et al. Development of lipidoid-siRNA formulations for systemic delivery to
the liver. Mol. Ther. 17, 872-9 (2009).
105.
Deleavey, G. F., Watts, J. K. & Damha, M. J. in Currentprotocols in nucleic acid
chemistry / edited by Serge L. Beaucage ... [et al.] Chapter 16, Unit 16.3 (2009).
106.
Turner, J. J., Jones, S. W., Moschos, S. a, Lindsay, M. a & Gait, M. J. MALDI-TOF mass
spectral analysis of siRNA degradation in serum confirms an RNAse A-like activity.
Mol. Biosyst. 3, 43-50 (2007).
107.
Raines, R. T. Ribonuclease A. Chem. Rev. 98, 1045-1066 (1998).
108.
Mook, 0. R., Baas, F., de Wissel, M. B. & Fluiter, K. Evaluation of locked nucleic acidmodified small interfering RNA in vitro and in vivo. Mol. Cancer Ther. 6, 833-43
(2007).
109.
Braasch, D. a et al. Biodistribution of phosphodiester and phosphorothioate siRNA.
Bioorg. Med. Chem. Lett. 14, 1139-43 (2004).
110.
Gao, S. et al. The effect of chemical modification and nanoparticle formulation on
stability and biodistribution of siRNA in mice. Mol. Ther. 17, 1225-33 (2009).
111.
Wolfrum, C. et al. Mechanisms and optimization of in vivo delivery of lipophilic
siRNAs. Nat. Biotechnol. 25, 1149-57 (2007).
112.
Akinc, A. et al. Targeted delivery of RNAi therapeutics with endogenous and
exogenous ligand-based mechanisms. Mol. Ther. 18, 1357-64 (2010).
113.
Alexis, F., Pridgen, E., Molnar, L. K. & Farokhzad, 0. C. Factors affecting the clearance
and biodistribution of polymeric nanoparticles. Mol. Pharm. 5, 505-15 (2008).
114.
Romberg, B., Hennink, W. E. & Storm, G. Sheddable coatings for long-circulating
nanoparticles. Pharm. Res. 25, 55-71 (2008).
115.
Bazile, D. et al. Stealth Me.PEG-PLA nanoparticles avoid uptake by the mononuclear
phagocytes system.J. Pharm. Sci. 84, 493-8 (1995).
116.
Peracchia, M. T. et al. Visualization of in vitro protein-rejecting properties of
PEGylated stealth polycyanoacrylate nanoparticles. Biomaterials20, 1269-75
(1999).
117.
Aird, W. C. Phenotypic heterogeneity of the endothelium: 1. Structure, function, and
mechanisms. Circ. Res. 100, 158-73 (2007).
118.
Minshall, R. D., Tiruppathi, C., Vogel, S. M. & Malik, A. B. Vesicle formation and
trafficking in endothelial cells and regulation of endothelial barrier function.
Histochem. Cell Biol. 117, 105-12 (2002).
135
119.
Jain, R. K. & Stylianopoulos, T. Delivering nanomedicine to solid tumors. Nat. Rev.
Clin. Oncol. 7, 653-64 (2010).
120.
Aird, W. C. Phenotypic heterogeneity of the endothelium: II. Representative vascular
beds. Circ. Res. 100, 174-90 (2007).
121.
Wisse, E., Jacobs, F., Topal, B., Frederik, P. & De Geest, B. The size of endothelial
fenestrae in human liver sinusoids: implications for hepatocyte-directed gene
transfer. Gene Ther. 15, 1193-9 (2008).
122.
Merkel, 0. M. et al. Stability of siRNA polyplexes from poly(ethylenimine) and
poly(ethylenimine)-g-poly(ethylene glycol) under in vivo conditions: effects on
pharmacokinetics and biodistribution measured by Fluorescence Fluctuation
Spectroscopy and Single Photon Emission Com.J. Control. Release 138, 148-59
(2009).
123.
Frank-Kamenetsky, M. et al. Therapeutic RNAi targeting PCSK9 acutely lowers
plasma cholesterol in rodents and LDL cholesterol in nonhuman primates. Proc. Natl.
Acad. Sci. U.S.A. 105, 11915-20 (2008).
124.
Shi, B. et al. Biodistribution of Small Interfering RNA at the Organ and Cellular Levels
after Lipid Nanoparticle-mediated Delivery.j. Histochem. Cytochem. 59, 727-40
(2011).
125.
Schroeder, A., Levins, C. G., Cortez, C., Langer, R. & Anderson, D. G. Lipid-based
nanotherapeutics for siRNA delivery.J. Intern. Med. 267, 9-21 (2010).
126.
Fukumura, D., Duda, D. G., Munn, L. L. & Jain, R. K. Tumor microvasculature and
microenvironment: novel insights through intravital imaging in pre-clinical models.
Microcirculation 17, 206-25 (2010).
127.
Jain, R. K. Normalization of tumor vasculature: an emerging concept in antiangiogenic
therapy. Science 307, 58-62 (2005).
128.
Vakoc, B. J. et al. Three-dimensional microscopy of the tumor microenvironment in
vivo using optical frequency domain imaging. Nat. Med. 15, 1219-23 (2009).
129.
Roberts, W. G. & Palade, G. E. Neovasculature induced by vascular endothelial growth
factor is fenestrated. Cancer Res. 57, 765 (1997).
130.
Hashizume, H. et al. Openings between defective endothelial cells explain tumor
vessel leakiness. Am. J. Pathol. 156, 1363-80 (2000).
131.
Hobbs, S. K. et al. Regulation of transport pathways in tumor vessels: role of tumor
type and microenvironment. Proc. Natl. Acad. Sci. U. S. A. 95, 4607-12 (1998).
136
132.
Yuan, F. et al. Vascular permeability and microcirculation of gliomas and mammary
carcinomas transplanted in rat and mouse cranial windows. CancerRes. 54, 4564-8
(1994).
133.
Padera, T. P. et al. Lymphatic metastasis in the absence of functional intratumor
lymphatics. Science 296, 1883-6 (2002).
134.
Leu, A. J., Berk, D. A., Lymboussaki, A., Alitalo, K. & Jain, R. K. Absence of functional
lymphatics within a murine sarcoma: a molecular and functional evaluation. Cancer
Res. 60, 4324-7 (2000).
135.
Maeda, H., Wu, J., Sawa, T., Matsumura, Y. & Hori, K. Tumor vascular permeability and
the EPR effect in macromolecular therapeutics: a review.j. Control. Release 65, 27184 (2000).
136.
Maeda, H. Tumor-selective delivery of macromolecular drugs via the EPR effect:
background and future prospects. Bioconjug. Chem. 21, 797-802 (2010).
137.
Wong, C. et al. Multistage nanoparticle delivery system for deep penetration into
tumor tissue. Proc. Natl. Acad. Sci. U. S. A. 108, (2011).
138.
Leuschner, F. et al. Therapeutic siRNA silencing in inflammatory monocytes in mice.
Nat. Biotechnol. 29, 1005-1010 (2011).
139.
Caron, E. & Hall, A. Identification of two distinct mechanisms of phagocytosis
controlled by different Rho GTPases. Science 282, 1717-21 (1998).
140.
Groves, E., Dart, A. E., Covarelli, V. & Caron, E. Molecular mechanisms of phagocytic
uptake in mammalian cells. Cell. Mol. Life Sci. 65, 1957-76 (2008).
141.
Swanson, J. A. & Baer, S. C. Phagocytosis by zippers and triggers. Trends Cell Biol. 5,
89-93 (1995).
142.
Owens, D. E. & Peppas, N. a. Opsonization, biodistribution, and pharmacokinetics of
polymeric nanoparticles. Int.J. Pharm. 307, 93-102 (2006).
143.
Vonarbourg, A., Passirani, C., Saulnier, P. & Benoit, J.-P. Parameters influencing the
stealthiness of colloidal drug delivery systems. Biomaterials27, 4356-73 (2006).
144.
Vachon, E. et al. CD44 is a phagocytic receptor. Blood 107, 4149-58 (2006).
145.
Gold, E. S. et al. Dynamin 2 is required for phagocytosis in macrophages.J. Exp. Med.
190, 1849-56 (1999).
146.
Claus, V. et al. Lysosomal enzyme trafficking between phagosomes, endosomes, and
lysosomes in J774 macrophages. Enrichment of cathepsin H in early endosomes.j.
Biol. Chem. 273, 9842-51 (1998).
137
147.
Racoosin, E. L. & Swanson, J. a. M-CSF-induced macropinocytosis increases solute
endocytosis but not receptor-mediated endocytosis in mouse macrophages.J. Cell Sci.
102 (Pt 4, 867-80 (1992).
148.
Stromhaug, P. E., Berg, T. 0., Gjoen, T. & Seglen, P. 0. Differences between fluid-phase
endocytosis (pinocytosis) and receptor-mediated endocytosis in isolated rat
hepatocytes. Eur. J. Cell Biol. 73, 28-39 (1997).
149.
Swanson,
150.
Conner, S. D. & Schmid, S. L. Regulated portals of entry into the cell. Nature 422, 3744 (2003).
151.
Mukherjee, S., Ghosh, R. N. & Maxfield, F. R. Endocytosis. Physiol. Rev. 77, 759-803
(1997).
152.
Hillaireau, H. & Couvreur, P. Nanocarriers' entry into the cell: relevance to drug
delivery. Cell. Mol. Life Sci. 66, 2873-96 (2009).
153.
Jones, A. R. & Shusta, E. V. Blood-brain barrier transport of therapeutics via receptormediation. Pharm. Res. 24, 1759-71 (2007).
154.
Bareford, L. M. & Swaan, P. W. Endocytic mechanisms for targeted drug delivery. Adv.
Drug Deliv. Rev. 59, 748-58 (2007).
155.
Medina-Kauwe, L. K. 'Alternative' endocytic mechanisms exploited by pathogens:
new avenues for therapeutic delivery?Adv. Drug Deliv. Rev. 59, 798-809 (2007).
156.
Mayor, S. & Pagano, R. E. Pathways of clathrin-independent endocytosis. Nat. Rev.
Mol. Cell Biol. 8, 603-12 (2007).
157.
Marsh, M. & Helenius, A. Virus entry: open sesame. Cell 124, 729-40 (2006).
158.
Yu, B., Zhao, X., Lee, L. J. & Lee, R. J. Targeted delivery systems for oligonucleotide
therapeutics. AAPSJ. 11, 195-203 (2009).
159.
Lu, J. J., Langer, R. & Chen, J. A novel mechanism is involved in cationic lipid-mediated
functional siRNA delivery. Mol. Pharm. 6, 763-71 (2009).
160.
Boussif, 0. et al. A versatile vector for gene and oligonucleotide transfer into cells in
culture and in vivo: polyethylenimine. Proc. Natl. Acad. Sci. U. S. A. 92, 7297-301
(1995).
161.
Hafez, I. M., Maurer, N. & Cullis, P. R. On the mechanism whereby cationic lipids
promote intracellular delivery of polynucleic acids. Gene Ther. 8, 1188-96 (2001).
J. A. & Watts, C. Macropinocytosis. Trends Cell Biol. 5, 424-8 (1995).
138
162.
Guo, X., MacKay, J. A. & Szoka, F. C. Mechanism of pH-triggered collapse of
phosphatidylethanolamine liposomes stabilized by an ortho ester polyethyleneglycol
lipid. Biophys.J. 84, 1784-95 (2003).
163.
Endoh, T. & Ohtsuki, T. Cellular siRNA delivery using cell-penetrating peptides
modified for endosomal escape. Adv. Drug Deliv. Rev. 61, 704-9 (2009).
164.
Dominska, M. & Dykxhoorn, D. M. Breaking down the barriers: siRNA delivery and
endosome escape.J. Cell Sci. 123, 1183-9 (2010).
165.
Jarad, G. & Miner, J. H. Update on the glomerular filtration barrier. Curr. Opin. Nephrol.
Hypertens. 18, 226 (2009).
166.
Fogo, A. B. & Kon, V. The glomerulus - a view from the inside - the endothelial cell. Int.
J. Biochem. Cell Biol. 42, 1388-1397 (2010).
167.
Miner, J. H. Glomerular basement membrane composition and the filtration barrier.
Pediatr. Nephrol. (2011). doi:10.1007/s00467-011-1785-1
168.
Patrakka, J. & Tryggvason, K. New insights into the role of podocytes in proteinuria.
Nat. Rev. Nephrol. 5, 463-8 (2009).
169.
Wartiovaara, J. et al. Nephrin strands contribute to a porous slit diaphragm scaffold
as revealed by electron tomography. Filtration114, 1475-1483 (2004).
170.
Rodewald, R. & Karnovsky, M. J. Porous substructure of the glomerular slit
diaphragm in the rat and mouse.j. Cell Biol. 60, 423-33 (1974).
171.
He, X. M. & Carter, D. C. Atomic structure and chemistry of human serum albumin.
Nature 358, 209 - 215 (1992).
172.
Patrakka, J. & Tryggvason, K. Molecular make-up of the glomerular filtration barrier.
Biochem. Biophys. Res. Commun. 396, 164-9 (2010).
173.
Kawachi, H. et al. Role of podocyte slit diaphragm as a filtration barrier. Nephrology
(Carlton). 11, 274-81 (2006).
174.
Ghibellini, G., Leslie, E. M. & Brouwer, K. L. R. Methods to evaluate biliary excretion of
drugs in humans: an updated review. Mol. Pharm. 3, 198-211 (2006).
175.
Fleck, C. & Brdunlich, H. Factors determining the relationship between renal and
hepatic excretion of xenobiotics. Arzneimittelforschung. 40, 942-6 (1990).
176.
Fagerholm, U. Prediction of human pharmacokinetics-biliary and intestinal clearance
and enterohepatic circulation.j. Pharm. Pharmacol.60, 535-42 (2008).
177.
Huang, Y. et al. Elimination pathways of systemically delivered siRNA. Mol. Ther. 19,
381-5 (2011).
139
178.
Singha, K., Namgung, R. & Kim, W. J. Polymers in small-interfering RNA delivery.
Nucleic Acid Ther. 21, 133-47 (2011).
179.
Mo, R. H., Zaro, J. L. & Shen, W.-C. Comparison of cationic and amphipathic cell
penetrating peptides for siRNA delivery and efficacy. MoL. Pharm. 9, 299-309 (2012).
180.
Yao, Y. et al. Targeted delivery of PLK1-siRNA by ScFv suppresses Her2+ breast
cancer growth and metastasis. Sci. TransL. Med. 4, 130ra48 (2012).
181.
Dassie, J. P. et aL. Systemic administration of optimized aptamer-siRNA chimeras
promotes regression of PSMA-expressing tumors. Nat. Biotechnol. 27, 839-49
(2009).
182.
Neff, C. P. et aL. An aptamer-siRNA chimera suppresses HIV-1 viral loads and protects
from helper CD4(+) T cell decline in humanized mice. Sci. Transi. Med. 3, 66ra6
(2011).
183.
Thomas, M. et aL. Ligand-targeted delivery of small interfering RNAs to malignant
cells and tissues. Ann. N. Y. Acad. Sci. 1175, 32-9 (2009).
184.
Davis, M. E. The first targeted delivery of siRNA in humans via a self-assembling,
cyclodextrin polymer-based nanoparticle: from concept to clinic. Mol. Pharm. 6, 65968 (2009).
185.
Gonzalez, H., Hwang, S. J. & Davis, M. E. New class of polymers for the delivery of
macromolecular therapeutics. Bioconjug. Chem. 10, 1068-74 (1999).
186.
Hwang, S. J., Bellocq, N. C. & Davis, M. E. Effects of structure of beta-cyclodextrincontaining polymers on gene delivery. Bioconjug. Chem. 12, 280-90 (2001).
187.
Pun, S. H. et aL. Cyclodextrin-modified polyethylenimine polymers for gene delivery.
Bioconjug. Chem. 15, 831-40 (2004).
188.
Pun, S. H. & Davis, M. E. Development of a nonviral gene delivery vehicle for systemic
application. Bioconjug. Chem. 13, 630-9 (2002).
189.
Reineke, T. M. & Davis, M. E. Structural effects of carbohydrate-containing
polycations on gene delivery. 1. Carbohydrate size and its distance from charge
centers. Bioconjug. Chem. 14, 247-54 (2003).
190.
Popielarski, S. R., Mishra, S. & Davis, M. E. Structural effects of carbohydratecontaining polycations on gene delivery. 3. Cyclodextrin type and functionalization.
Bioconjug. Chem. 14, 672-8 (2003).
191.
Davis, M. et aL. Self-Assembling Nucleic Acid Delivery Vehicles via Linear, WaterSoluble, Cyclodextrin-Containing Polymers. Curr. Med. Chem. 11, 179-197 (2004).
140
192.
Hu-Lieskovan, S., Heidel, J. D., Bartlett, D. W., Davis, M. E. & Triche, T. J. Sequencespecific knockdown of EWS-FLII by targeted, nonviral delivery of small interfering
RNA inhibits tumor growth in a murine model of metastatic Ewing's sarcoma. Cancer
Res. 65, 8984-92 (2005).
193.
Mishra, S., Heidel, J. D., Webster, P. & Davis, M. E. Imidazole groups on a linear,
cyclodextrin-containing polycation produce enhanced gene delivery via multiple
processes. j. Control. Release 116, 179-91 (2006).
194.
Bartlett, D. W. & Davis, M. E. Physicochemical and biological characterization of
targeted, nucleic acid-containing nanoparticles. Bioconjug. Chem. 18, 456-68 (2007).
195.
Mishra, S., Webster, P. & Davis, M. E. PEGylation significantly affects cellular uptake
and intracellular trafficking of non-viral gene delivery particles. Eur. J. Cell Biol. 83,
97-111 (2004).
196.
Bellocq, N. C., Pun, S. H., Jensen, G. S. & Davis, M. E. Transferrin-containing,
cyclodextrin polymer-based particles for tumor-targeted gene delivery. Bioconjug.
Chem. 14, 1122-32 (2003).
197.
Bartlett, D. W. & Davis, M. E. Insights into the kinetics of siRNA-mediated gene
silencing from live-cell and live-animal bioluminescent imaging. Nucleic Acids Res. 34,
322-33 (2006).
198.
Bartlett, D. W., Su, H., Hildebrandt, I. J., Weber, W. a & Davis, M. E. Impact of tumorspecific targeting on the biodistribution and efficacy of siRNA nanoparticles
measured by multimodality in vivo imaging. Proc. Natl. Acad. Sci. U. S. A. 104, 15 54954 (2007).
199.
Bartlett, D. W. & Davis, M. E. Impact of tumor-specific targeting and dosing schedule
on tumor growth inhibition after intravenous administration of siRNA-containing
nanoparticles. Biotechnol. Bioeng. 99, 975-85 (2008).
200.
Heidel, J. D. et al. Administration in non-human primates of escalating intravenous
doses of targeted nanoparticles containing ribonucleotide reductase subunit M2
siRNA. Proc. Natl. Acad. Sci. U. S.A. 104, 5715-21 (2007).
201.
Alabi, C., Vegas, A. & Anderson, D. Attacking the genome: emerging siRNA
nanocarriers from concept to clinic. Curr. Opin. Pharmacol. 12, 427-33 (2012).
202.
Burnett, J. C., Rossi, J. J. & Tiemann, K. Current progress of siRNA/shRNA therapeutics
in clinical trials. Biotechnol. J. 6, 1130-46 (2011).
203.
Xu, Y. & Szoka, F. C. Mechanism of DNA release from cationic liposome/DNA
complexes used in cell transfection. Biochemistry 35, 5616-23 (1996).
204.
Zhang, S., Zhi, D. & Huang, L. Lipid-based vectors for siRNA delivery.j. Drug Target.
20, 724-35 (2012).
141
205.
Kesharwani, P., Gajbhiye, V. & Jain, N. K. A review of nanocarriers for the delivery of
small interfering RNA. Biomaterials33, 7138-50 (2012).
206.
Huang, L. & Liu, Y. In vivo delivery of RNAi with lipid-based nanoparticles. Annu. Rev.
Biomed. Eng. 13, 507-30 (2011).
207.
Sato, Y. et al. A pH-sensitive cationic lipid facilitates the delivery of liposomal siRNA
and gene silencing activity in vitro and in vivo.J. Control. Release 163, 267-76
(2012).
208.
Mahon, K. P. et al. Combinatorial approach to determine functional group effects on
lipidoid-mediated siRNA delivery. Bioconjug. Chem. 21, 1448-54 (2010).
209.
Zhang, J., Fan, H., Levorse, D. a & Crocker, L. S. Ionization behavior of amino lipids for
siRNA delivery: determination of ionization constants, SAR, and the impact of lipid
pKa on cationic lipid-biomembrane interactions. Langmuir 27, 1907-14 (2011).
210.
Hafez, I. M., Ansell, S. & Cullis, P. R. Tunable pH-sensitive liposomes composed of
mixtures of cationic and anionic lipids. Biophys.J. 79, 1438-46 (2000).
211.
Jayaraman, M. et al. Maximizing the potency of siRNA lipid nanoparticles for hepatic
gene silencing in vivo. Angew. Chem. Int. Ed. Enyl. 51, 8529-33 (2012).
212.
Zuhorn, I. S. et al. Nonbilayer phase of lipoplex-membrane mixture determines
endosomal escape of genetic cargo and transfection efficiency. Mol. Ther. 11, 801-10
(2005).
213.
Zhigaltsev, I. V, Maurer, N., Wong, K. F. & Cullis, P. R. Triggered release of doxorubicin
following mixing of cationic and anionic liposomes. Biochim. Biophys. Acta 1565,
129-35 (2002).
214.
Koltover, I. An Inverted Hexagonal Phase of Cationic Liposome-DNA Complexes
Related to DNA Release and Delivery. Science (80-.). 281, 78-81 (1998).
215.
Belliveau, N. M. et al. Microfluidic Synthesis of Highly Potent Limit-size Lipid
Nanoparticles for In Vivo Delivery of siRNA. Mol. Ther. Nucleic Acids 1, e37 (2012).
216.
Bao, Y. et al. Effect of PEGylation on Biodistribution and Gene Silencing of
siRNA/Lipid Nanoparticle Complexes. Pharm. Res. 30, 342-51 (2013).
217.
Kolli, S. et al. pH-Triggered nanoparticle mediated delivery of siRNA to liver cells in
vitro and in vivo. Bioconjug. Chem. (2013). doi:10.1021/bc3004099
218.
Lin, S.-Y. et al. Sterically polymer-based liposomal complexes with dual-shell
structure for enhancing the siRNA delivery. Biomacromolecules13, 664-75 (2012).
219.
Virtanen, J. A., Ruonala, M., Vauhkonen, M. & Somerharju, P. Lateral organization of
liquid-crystalline cholesterol-dimyristoylphosphatidylcholine bilayers. Evidence for
142
domains with hexagonal and centered rectangular cholesterol superlattices.
Biochemistry 34, 11568-81 (1995).
220.
Takahashi, H., Sinoda, K. & Hatta, I. Effects of cholesterol on the lamellar and the
inverted hexagonal phases of dielaidoylphosphatidylethanolamine. Biochim. Biophys.
Acta 1289, 209-16 (1996).
221.
Sato, Y. et al. Resolution of liver cirrhosis using vitamin A-coupled liposomes to
deliver siRNA against a collagen-specific chaperone. Nat. Biotechnol. 26, 431-42
(2008).
222.
Ishiwatari, H. et al. Treatment of pancreatic fibrosis with siRNA against a collagenspecific chaperone in vitamin A-coupled liposomes. Gut 1-12 (2012).
doi:10.1136/gutjnl-2011-301746
223.
Yoshizawa, T., Hattori, Y., Hakoshima, M., Koga, K. & Maitani, Y. Folate-linked lipidbased nanoparticles for synthetic siRNA delivery in KB tumor xenografts. Eur. J.
Pharm. Biopharm. 70, 718-25 (2008).
224.
Feng, C. et al. Silencing of the MYCN gene by siRNA delivered by folate receptortargeted liposomes in LA-N-5 cells. Pediatr. Surg. Int. 26, 1185-91 (2010).
225.
Wan, K. et al. In vivo tumor imaging using a novel RNAi-based detection mechanism.
Nanomedicine 8, 393-8 (2012).
226.
Tabernero, J. et al. First-in-Man Trial of an RNA Interference Therapeutic Targeting
VEGF and KSP in Cancer Patients with Liver Involvement. Cancer Discov. 406-417
(2013). doi:10.1158/2159-8290.CD-12-0429
227.
Alsina, M. et al. Open-Label Extension Study of the RNAi Therapeutic ALN-VSPO2 in
Cancer Patients Responding to Therapy. American Society of Clinical Oncology
Meeting (2012). at <http://www.alnylam.com/capella/wpcontent/uploads/2012/06/ALN-VSP-ExtensionStudyPoster-ASCO-June2012panel.pdf>
228.
Jeong, J. H., Mok, H., Oh, Y.-K. & Park, T. G. siRNA conjugate delivery systems.
Bioconjug. Chem. 20, 5-14 (2009).
229.
Rozema, D. B. et al. Dynamic PolyConjugates for targeted in vivo delivery of siRNA to
hepatocytes. Proc. Natl. Acad. Sci. U. S. A. 104, 12982-7 (2007).
230.
Wakefield, D. H., Klein, J. J., Wolff, J. a & Rozema, D. B. Membrane activity and
transfection ability of amphipathic polycations as a function of alkyl group size.
Bioconjug. Chem. 16, 1204-8 (2005).
231.
Lewis, D. Dynamic Polyconjugates ( DPC ) Technology: An Elegant Solution to the
siRNA Delivery Problem. (2006). at
<http://www.arrowheadresearch.com/sites/default/files/udocs/ArrowheadResear
chCorporation-DPCTechnologyWhitePaper.pdf>
143
232.
Wong, S. C. et al. Co-injection of a targeted, reversibly masked endosomolytic
polymer dramatically improves the efficacy of cholesterol-conjugated small
interfering RNAs in vivo. Nucleic Acid Ther. 22, 380-90 (2012).
233.
Macron, D. Arrowhead Presents Preclinical Data on HBV Candidate, Subcutaneous
Delivery Tech. Gene Silencing News (2012). at
<http://www.arrowheadresearch.com/sites/default/files/news/GeneSilenNews_1
11612_reprint.pdf>
234.
Rozema, D. B., Ekena, K., Lewis, D. L., Loomis, A. G. & Wolff, J. A. Endosomolysis by
masking of a membrane-active agent (EMMA) for cytoplasmic release of
macromolecules. Bioconjug. Chem. 14, 51-7 (2003).
235.
Wooddell, C. I. et al. Hepatocyte-targeted RNAi Therapeutics for the Treatment of
Chronic Hepatitis B Virus Infection. Mol. Ther. (2013). doi:10.1038/mt.2013.31
236.
Biessen, E. a et al. Synthesis of cluster galactosides with high affinity for the hepatic
asialoglycoprotein receptor.]. Med. Chem. 38, 1538-46 (1995).
237.
Rensen, P. C. N., van Leeuwen, S. H., Sliedregt, L. a J. M., van Berkel, T. J. C. & Biessen,
E. a L. Design and synthesis of novel N-acetylgalactosamine-terminated glycolipids
for targeting of lipoproteins to the hepatic asialoglycoprotein receptor. J. Med. Chem.
47, 5798-808 (2004).
238.
Baenziger, J. U. & Fiete, D. Galactose and N-acetylgalactosamine-specific endocytosis
of glycopeptides by isolated rat hepatocytes. Cell 22, 611-20 (1980).
239.
Rensen, P. C. et al. Determination of the upper size limit for uptake and processing of
ligands by the asialoglycoprotein receptor on hepatocytes in vitro and in vivo.J. Biol.
Chem. 276,37577-84 (2001).
240.
Kallanthottathil, R. Conjugation Strategies for In Vivo siRNA Delivery. (2012). at
<http://www.alnylam.com/capella/wp-content/uploads/2012/11/ALNY-OTSConjugate-Oct2012.pdf>
241.
Smith, D., Schtller, V., Engst, C., Raidler, J. & Liedl, T. Nucleic acid nanostructures for
biomedical applications. Nanomedicine (Lond). 8, 105-21 (2013).
242.
Lee, H. et al. Molecularly self-assembled nucleic acid nanoparticles for targeted in
vivo siRNA delivery. Nat. Nanotechnol. 7, 389-93 (2012).
243.
Shu, D., Shu, Y., Haque, F., Abdelmawla, S. & Guo, P. Thermodynamically stable RNA
three-way junction for constructing multifunctional nanoparticles for delivery of
therapeutics. Nat. Nanotechnol. 6, 658-67 (2011).
244.
Goodman, R. P. et al. Rapid chiral assembly of rigid DNA building blocks for molecular
nanofabrication. Science 310, 1661-5 (2005).
144
245.
Xia, W. & Low, P. S. Folate-targeted therapies for cancer. J. Med. Chem. 53, 6811-24
(2010).
246.
Petros, R. a & DeSimone, J. M. Strategies in the design of nanoparticles for therapeutic
applications. Nat. Rev. Drug Discov. 9, 615-27 (2010).
247.
Gindy, M. E., Leone, A. M. & Cunningham, J. J. Challenges in the pharmaceutical
development of lipid-based short interfering ribonucleic acid therapeutics. Expert
Opin. Drug Defiv. 9, 171-82 (2012).
248.
Chen, D. et al. Rapid discovery of potent siRNA-containing lipid nanoparticles
enabled by controlled microfluidic formulation. J. Am. Chem. Soc. 134, 6948-51
(2012).
249.
Alabi, C. a et al. Multiparametric approach for the evaluation of lipid nanoparticles for
siRNA delivery. Proc. NatL. Acad. Sci. U. S. A. 110, 12881-6 (2013).
250.
Boisgu6rin, P. et al. Delivery of therapeutic oligonucleotides with cell penetrating
peptides. Adv. Drug Deliv. Rev. (2015). doi:10.1016/j.addr.2015.02.008
251.
Whitehead, K. a et al. Degradable lipid nanoparticles with predictable in vivo siRNA
delivery activity. Nat. Commun. 5, 4277 (2014).
145