-
10.1261/rna.459807Access the most recent version at doi: 2007
13: 431-456 originally published online February 28, 2007RNA
Daniel De Paula, M. Vitória L.B. Bentley and Ram I. Mahato and
targetingHydrophobization and bioconjugation for enhanced siRNA
delivery
References
http://rnajournal.cshlp.org/content/13/4/431.full.html#related-urlsArticle
cited in:
http://rnajournal.cshlp.org/content/13/4/431.full.html#ref-list-1This
article cites 151 articles, 46 of which can be accessed free
at:
serviceEmail alerting
click heretop right corner of the article orReceive free email
alerts when new articles cite this article - sign up in the box at
the
http://rnajournal.cshlp.org/subscriptions go to: RNATo subscribe
to
Copyright © 2007 RNA Society
Cold Spring Harbor Laboratory Press on February 12, 2009 -
Published by rnajournal.cshlp.orgDownloaded from
http://rnajournal.cshlp.org/lookup/doi/10.1261/rna.459807http://rnajournal.cshlp.org/content/13/4/431.full.html#ref-list-1http://rnajournal.cshlp.org/content/13/4/431.full.html#related-urlshttp://rnajournal.cshlp.org/cgi/alerts/ctalert?alertType=citedby&addAlert=cited_by&saveAlert=no&cited_by_criteria_resid=rna;13/4/431&return_type=article&return_url=http://rnajournal.cshlp.org/content/13/4/431.full.pdfhttp://rnajournal.cshlp.org/subscriptionshttp://rnajournal.cshlp.org/http://www.cshlpress.com
-
REVIEW
Hydrophobization and bioconjugation for enhanced
siRNA delivery and targeting
DANIEL DE PAULA,1,2 M. VITÓRIA L.B. BENTLEY,1 and RAM I.
MAHATO2
1Faculdade de Ciências Farmacêuticas de Ribeirão Preto,
Universidade de São Paulo, Ribeirão Preto, SP, Brazil2Department
of Pharmaceutical Sciences, University of Tennessee Health Science
Center, Memphis, Tennessee 38163, USA
ABSTRACT
RNA interference (RNAi) is an evolutionarily conserved process
by which double-stranded small interfering RNA (siRNA)induces
sequence-specific, post-transcriptional gene silencing. Unlike
other mRNA targeting strategies, RNAi takes advantage ofthe
physiological gene silencing machinery. The potential use of siRNA
as therapeutic agents has attracted great attention as anovel
approach for treating severe and chronic diseases. RNAi can be
achieved by either delivery of chemically synthesizedsiRNAs or
endogenous expression of small hairpin RNA, siRNA, and microRNA
(miRNA). However, the relatively high dose ofsiRNA required for
gene silencing limits its therapeutic applications. This review
discusses several strategies to improvetherapeutic efficacy as well
as to abrogate off-target effects and immunostimulation caused by
siRNAs. There is an in-depthdiscussion on various issues related to
the (1) mechanisms of RNAi, (2) methods of siRNA production, (3)
barriers to RNAi-basedtherapies, (4) biodistribution, (5) design of
siRNA molecules, (6) chemical modification and bioconjugation, (7)
complexformation with lipids and polymers, (8) encapsulation into
lipid particles, and (9) target specificity for enhanced
therapeuticeffectiveness.
Keywords: RNA interference; small interfering RNA;
bioconjugation; chemical modification; complex formation
INTRODUCTION
RNA interference (RNAi) is an evolutionarily conservedprocess by
which double-stranded small interfering RNA(siRNA) induces
sequence-specific, post-transcriptionalgene silencing (Hannon
2002). The process of RNAiconsists of an initiator step, in which
long double-strandedRNA (dsRNA) is cleaved into siRNA fragments,
and aneffector step, in which these fragments are incorporatedinto
a protein complex, dissociated, and used as a guidingsequence to
recognize homologous mRNA that is sub-sequently cleaved. RNAi is
considered a self-defense mech-anism of eukaryotic cells to combat
infection by RNAviruses and transposons. It is also assumed to
tightlyregulate protein levels in response to various
environmentalstimuli (Meister and Tuschl 2004).
The revolutionary finding of RNAi resulted from thework of
Andrew Fire and coworkers at the CarnegieInstitute in Washington,
D.C., who demonstrated in 1998that injection of dsRNA into
Caenorhabditis elegans leads toefficient sequence-specific gene
silencing (Fire et al. 1998).At that time, the state of the art in
gene silencing was theuse of antisense oligodeoxynucleotide (ODNs),
whichcomprise single strands of short DNA or RNA comple-mentary
sequences that hybridize with the targeted mRNA(Mahato et al.
2005). However, the dsRNA seemed toinduce silencing through a
pathway distinct from classicalantisense therapies due to the
catalytic nature of RNAi, inwhich one siRNA can be used over and
over to guide thecleavage of many mRNA molecules (Dykxhoorn et
al.2003). Bertrand et al. (2002) have compared the effects
ofantisense ODNs and siRNAs targeting green fluorescentprotein
(GFP) in vitro and in vivo. siRNA was quantita-tively more
efficient, and its effect lasted for a longer timein cell culture.
In mice, siRNAs were able to silence geneexpression, whereas no
effect was observed in the presenceof antisense ODNs.
The first evidence that siRNAs can mediate sequence-specific
gene silencing in mammalian cells was provided in
rna4598 De Paula et al. REVIEW RA
Abbreviations: See list at end of paper.Reprint requests to: Ram
I. Mahato, Department of Pharmaceutical
Sciences, Health Science Center, University of Tennessee, 26
South DunlapStreet, Feurt Building, Room 413, Memphis, TN 38163,
USA; e-mail:[email protected]; fax: (901) 448-6092.
Article published online ahead of print. Article and publication
date areat
http://www.rnajournal.org/cgi/doi/10.1261/rna.459807.
RNA (2007), 13:431–456. Published by Cold Spring Harbor
Laboratory Press. Copyright � 2007 RNA Society. 431
JOBNAME: RNA 13#4 2007 PAGE: 1 OUTPUT: Tuesday March 6 18:39:02
2007
csh/RNA/131733/rna4598
Cold Spring Harbor Laboratory Press on February 12, 2009 -
Published by rnajournal.cshlp.orgDownloaded from
http://rnajournal.cshlp.org/http://www.cshlpress.com
-
2001 when the conversion of dsRNA into short RNAfragments was
shown to be bypassed by the transfectionof siRNA molecules into
cells (Elbashir et al. 2001a). Afterthat, various in vivo effects
of siRNA and short hairpinRNA (shRNA) have been reported (Lewis et
al. 2002;McCaffrey et al. 2002; Xia et al. 2002; Song et al.
2003b).For example, McCaffrey et al. (2002) showed that siRNAand
shRNA reduce luciferase (Luc) expression in the liverin a
sequence-specific manner. Song et al. (2003b) foundthat siRNA
targeted to the Fas receptor protects mice fromliver fibrosis.
Human clinical trials of RNAi-based drugs are currentlyunder way
by Acuity Pharmaceuticals and Sirna Therapeu-tics. Both companies
are working on intravitreal adminis-tration of siRNA targeting
vascular endothelial growthfactor (VEGF), whose overexpression is
the primary causeof age-related macular degeneration (AMD). The
firstresults are encouraging in terms of tolerability of
siRNAcompounds. Other clinical trials involving siRNA fortreating
chronic myeloid leukemia and respiratory syncytialvirus infection
are also being carried out by the HadassahMedical Organization and
Alnylam Pharmaceutics, Inc.,respectively.
Despite early excitement, siRNAs have shown to activateimmune
response in a sequence- and concentration-depen-dent manner,
leading to nonspecific gene silencing (Jacksonet al. 2003; Sioud
2005). In this respect, the introduction ofchemical modifications
and generation of designed siRNAshave become essential for
achieving high gene silencingwith a low degree of undesired
effects. Chemical modifica-tions of siRNAs also appear to stabilize
these molecules inserum and show enhanced gene silencing (Braasch
et al.2003; Chiu and Rana 2003; Layzer et al. 2004). A
break-through in this area has recently been achieved by
linkingcholesterol (Chol) to siRNAs (Soutschek et al. 2004), andby
encapsulating them into stable nucleic acid lipidparticles
(Zimmermann et al. 2006). Undoubtedly, effectivein vivo delivery of
siRNAs will be a key factor in turningsiRNA into a new class of
therapeutics. In this review, wediscuss several strategies to
abrogate undesired effects andimprove the efficiency of siRNA,
including chemical modifi-cations, siRNA sequence design,
bioconjugation, complexformation, and encapsulation into lipid
particles.
MECHANISM OF RNA INTERFERENCE
The initial studies that unraveled the RNAi machinery
wereobtained from experiments using Drosophila extracts.Figure 1
illustrates the mechanism of RNAi. It involvesprocessing of long
dsRNA of 500–1000 nucleotides (nt),which triggers into fragments of
siRNA. Generally, siRNAis 21–23-nt dsRNA segments with 39-overhangs
of 2 nt onboth strands, and unphosphorylated hydroxyl groups at
the29 and 39 sites. The siRNA structure is characteristic of
anRNase III-like cleavage pattern and plays a crucial role in
its
recognition by other RNAi components. Dicer is theenzyme known
to process dsRNA into siRNAs, and itcontains dual catalytic domains
and additional helicase andPAZ domains (Elbashir et al. 2001b;
Zamore and Haley2005). Processed siRNA is incorporated into a
proteincomplex, termed RNA-induced silencing complex (RISC).Dicer
is also involved in the early steps of RISC formationand may be
required for siRNA entry into RISC (Lee et al.2004). In mammalian
cells, the protein Argonaute 2 (Ago2)is the catalytic component of
RISC that cleaves targetmRNAs (Liu et al. 2004b; Meister et al.
2004). In the RISCassembly process, siRNAs are initially loaded
into Ago2 asduplexes, and then Ago2 cleaves the passenger
strand,thereby liberating the guide strand from the siRNA duplexand
producing active RISC capable of cleaving targetmRNAs (Matranga et
al. 2005; Leuschner et al. 2006).The guide strand serves as a
template for the recognitionof homologous mRNA, which upon binding
to RISC iscleaved between bases 10 and 11 relative to the 59 base
ofthe guide siRNA strand by the catalytic activity of Ago2(Elbashir
et al. 2001c). The template siRNA is not affectedby this reaction,
so the RISC can undergo numerous cyclesof mRNA cleavage that
comprise the high efficiency ofRNAi (Elbashir et al. 2001b; Meister
and Tuschl 2004). Themechanism by which passenger strands are
cleaved by Ago2follows the same rules established for the
siRNA-guidedcleavage of a target mRNA (Leuschner et al. 2006).
RNAi can also be induced by endogenous expression ofsmall
regulatory RNAs known as microRNAs (miRNAs)(Fig. 1). The generation
of miRNAs occurs via sequentialprocessing and maturation of long
primary transcripts (pri-miRNA). Pre-microRNA (pre-miRNA) exits in
the nucleusupon cleavage by Drosha and is recognized by the
endo-nuclease Dicer, which processes the pre-miRNA into a22-nt
mature miRNA (Bartel 2004). The mature miRNA issubsequently
incorporated into the silencing complex(RISC), where it mediates
translational repression. DuringRISC assembly, the Ago2-mediated
cleavage of the passen-ger strand is blocked by mismatches between
the guideand passenger strands of miRNAs. In this case, a
slower‘‘backup’’ pathway dissociates and destroys the
passengerstrand, liberating active RISC (Matranga et al.
2005;Leuschner et al. 2006). In contrast to siRNA, miRNAs medi-ate
RNAi through a translational inhibition mechanisminvolving
imperfect complementarity to target mRNAs(Jackson et al. 2003;
Bartel 2004).
One major difference between mammalian RNAi andRNAi in other
eukaryotes is the lack of an amplificationsystem for long-term
persistence of RNAi in mammaliancells. For example, in Drosophila,
z35 molecules of dsRNAcan silence z1000 copies of a target mRNA per
cell and canpersist over the course of many generations (Nykanen et
al.2001). In contrast, in mammalian cells RNAi persistseffectively
only for an average of 66 h due to its dilutionduring cell
divisions (Chiu and Rana 2002).
De Paula et al.
432 RNA, Vol. 13, No. 4
JOBNAME: RNA 13#4 2007 PAGE: 2 OUTPUT: Tuesday March 6 18:39:02
2007
csh/RNA/131733/rna4598
Cold Spring Harbor Laboratory Press on February 12, 2009 -
Published by rnajournal.cshlp.orgDownloaded from
http://rnajournal.cshlp.org/http://www.cshlpress.com
-
METHODS OF siRNA PRODUCTION
The methods of siRNA production can be classified intothree
different categories: (1) chemical synthesis, (2) invitro
transcription, and (3) endogenous expression. Thechoice of
production method depends on the target gene,cell type, in vitro or
in vivo use, and desired extent of genesilencing. Table 1 describes
the advantages and disadvan-tages of each method of siRNA
production.
Chemical synthesis
Chemical synthesis is the most direct means of generatingsiRNAs
and has several advantages, including precisecontrol of the amount
and purity of siRNA, ease incharacterization and scale-up, and ease
in chemical mod-ifications for enhanced stability and target
specificity(Elbashir et al. 2001a). The chemical synthesis of
siRNArequires several steps including the generation of
twohomologous strands, annealing of the strands, addition
ofchemical entities to increase stability, and ensuring that
2-nt overhangs are present. The siRNA duplex requires
a39-hydroxyl group and a 59-phosphate group for functionalactivity
(Dykxhoorn et al. 2003). Unlike DNA, RNA pos-sesses an additional
hydroxyl group at the 29 position ofeach ribose building block,
which destabilizes RNA underthe basic conditions generally present
in DNA synthesisreactions. Hence, the most difficult step in RNA
synthesisis the simultaneous protection of the 59- and
29-hydroxylgroups during solid-phase chemistry.
One drawback of using chemically synthesized siRNA isthat the
most effective target sequence is unpredictablesince gene silencing
efficiency may vary depending on thesegments of the transcripts
that are targeted. For example,Holen et al. (2002) observed that
only a few siRNAsresulted in a significant reduction of human
tissue factor(HTF) expression after targeting its mRNA with
severalsiRNAs synthesized against different sites of the samemRNA.
The results suggested that accessible siRNA targetsites may be rare
in some human mRNAs. However, moreeffective gene silencing can be
achieved by targetingdifferent segments of the same transcript
simultaneously.
FIGURE 1. Mechanisms of RNA interference (RNAi). RNAi is
triggered by long double-stranded RNA (dsRNA), small interfering
RNA (siRNA),plasmid or virus-based short hairpin RNA (shRNA), or
microRNA (miRNA), and siRNA. Long dsRNA, shRNA, and pre-miRNA are
processedby Dicer into 21–23-nt siRNA duplexes with symmetric 2-nt
39 overhangs and 59-phosphate groups. Processed siRNAs assemble
with cellularproteins to form an RNA-induced silencing complex
(RISC). During RISC assembly, one strand (passenger) is eliminated,
while the other strand(guide) produces an active RISC, which
eventually triggers the sequence-specific mRNA degradation (in case
of total complementarity) ortranslational repression (in case of
incorporation of partially complementary miRNA).
siRNA delivery and targeting
www.rnajournal.org 433
JOBNAME: RNA 13#4 2007 PAGE: 3 OUTPUT: Tuesday March 6 18:39:02
2007
csh/RNA/131733/rna4598
Fig. 1 live 4/C
Cold Spring Harbor Laboratory Press on February 12, 2009 -
Published by rnajournal.cshlp.orgDownloaded from
http://rnajournal.cshlp.org/http://www.cshlpress.com
-
Significant enhancement in gene silencing was achieved
bytargeting the HIV-1 coreceptor CXCR4 and the apoptosis-inducing
Fas ligand with two or more siRNAs againstdifferent sites of the
same mRNA (Ji et al. 2003).
In vitro transcription and enzymatic digestion
Since chemically synthesized siRNAs are expensive,attempts are
being made to produce siRNAs by in vitrotranscription using T7 RNA
polymerase. Synthetic DNAtemplates containing the T7 RNA polymerase
promoterregion followed by desired RNA sequence can be
producedusing an automated DNA synthesizer and then amplifiedusing
PCR. The T7 polymerase binds to the promotersequence, initiates
transcription, and then moves along thetemplate strand toward its
59 end, elongating the RNAtranscript as it goes. Although a
termination region on theDNA may be used to halt transcription, the
runofftranscription, which stops when the 59 end of the DNAtemplate
strand is reached, is also commonly used. Tran-scription of the PCR
fragment by the T7 RNA polymeraseproduces both sense and antisense
RNAs, which spontane-ously anneal, forming full-length dsRNA
(Milligan andUhlenbeck 1989; Donze and Picard 2002).
The in vitro transcription approach is limited by
specificsequence requirements related to T7 polymerase. The
lastguanosine of the T7 promoter is invariably the
firstribonucleotide that is incorporated into the RNA by theT7
polymerase during transcription. Therefore, all siRNAsproduced by
this method start with a 59-G residue andrequire a C-39 residue at
position 19 (i.e., 59-G-N17-C-39)to allow annealing with the
complementary RNA, whichalso has to start with a 59-G residue.
Furthermore, since theefficacy of siRNAs targeted to different
regions of a genevaries dramatically, the number of sequences that
can betargeted using siRNAs generated by this method is limited
(Donze and Picard 2002). Kim et al. (2004) demonstratedthat
siRNAs synthesized from the T7 polymerase systemcan trigger a
potent induction of interferon a and b in anumber of cell lines.
The induction of these interferons wasalso seen by short
single-stranded RNAs transcribed withT3, T7, and Sp6 RNA
polymerases. These investigatorsfurther demonstrated that the
presence of triphosphateson the 59 end of T7 transcripts is
required for interferoninduction, since the treatment of T7
transcripts withphosphatase could abrogate this effect.
Sohail et al. (2003) reported an alternative approach
forproducing a desired siRNA sequence using T7 polymerasein vitro
transcription followed by transcript digestion bydeoxyribozymes,
which are known as ‘‘DNAzymes’’ orcatalytic DNA. The cleavage of
RNA by DNAzymes occursbetween two specific nucleotides, and the
requirement forthis dinucleotide sequence is different for
different DNA-zyme groups, making them flexible tools for digesting
avariety of sequences (Breaker and Joyce 1994; Feldman andSen
2001). The siRNAs produced by this method causeddose-dependent
inhibition of type 1 insulin-like growthfactor receptor in human
breast cancer cells comparable tothat induced by chemically
synthesized siRNAs of the samesequence (Sohail et al. 2003).
Both recombinant human Dicer (re-hDicer) and Escher-ichia coli
RNAse III are used to cleave long dsRNAproduced by in vitro
transcription. Treatment with E. coliRNase III or re-hDicer
randomly cleaves the RNA into apopulation of siRNA molecules that
are effective mediatorsof gene silencing in a manner similar to
that observed whenusing synthetic siRNA. Usually, the cleavage
products ofE. coli enzyme RNase III range from 12 to 15 base
pairs(bp) in length with termini identical to those produced
byDicer (Amarasinghe et al. 2001). Although these shortdsRNAs are
not long enough to trigger an RNAi inmammalian cells, the average
product length generated by
TABLE 1. Methods of producing siRNA
Methods of production Characteristics Advantages
Disadvantages
Chemical synthesis d 21–23-bp siRNA d Ease in sequence design d
High costd Long dicer substrate
RNA (disRNA) ;27 bpd Chemical modifications d Transient gene
silencing
d Quick method to silencea specific gene
In vitro transcription+enzymaticdigestion
d Pool of siRNAs d Potent silencing d Higher probability of
inducingoff-target effects andnonspecific silencing
d Low cost d Triphosphate induction of interferonby T7
polymerase transcripts
Endogenous expression(plasmid or viral vectors)
d siRNA d Allows both long-term andtransient gene silencing
d Difficult to transfect in caseof plasmid
d shRNA d Inducible gene silencing d Vector construction is
oftentime-consuming
d pre-miRNA
De Paula et al.
434 RNA, Vol. 13, No. 4
JOBNAME: RNA 13#4 2007 PAGE: 4 OUTPUT: Tuesday March 6 18:39:21
2007
csh/RNA/131733/rna4598
Cold Spring Harbor Laboratory Press on February 12, 2009 -
Published by rnajournal.cshlp.orgDownloaded from
http://rnajournal.cshlp.org/http://www.cshlpress.com
-
RNase III digestion can be increased by altering
digestionconditions. Yang et al. (2002) have shown that
20–25-bpsiRNAs produced by RNase III digestion efficientlyinhibited
various endogenous genes in different mamma-lian cell lines without
nonspecific effects. In addition,RNase III can digest dsRNA faster
than re-hDicer andcan be easily overexpressed and purified
(Amarasinghe et al.2001; Bernstein et al. 2001). The siRNAs
produced byre-hDicer are 21–23 bp in length and contain 2-nt 39
over-hangs with 59-phosphate and 39-hydroxyl termini, whichare
essential for RNAi activity. Dicer-generated siRNAs havebeen shown
to be effective in silencing exogenous andendogenous gene
expression in mammalian cells (Kawasakiet al. 2003; Myers et al.
2003).
The enzymatic synthesis of siRNA may provide moreeffective gene
silencing than chemically synthesized siRNA,as the enzymatically
generated siRNA can correspond tosequences overlapping the entire
gene. However, complexsiRNA populations may have a higher
probability oftargeting other genes, and therefore promoting
nonspecificeffects, than do discrete siRNAs. Additionally,
enzymaticsynthesis of siRNA requires separation of siRNA
fromuncleaved RNA duplexes and residual nucleic acids (Yanget al.
2002; Myers et al. 2003).
Endogenous expression
The application of synthetic siRNAs is restricted by
bothlow-to-moderate transfection efficiencyand the short-term
persistence of tran-sient gene expression. A single trans-fection
of siRNA may not provide asufficient window of functional
deple-tion for proteins with long half-lives.Another potential
problem inherent inchemically synthesized siRNA is vari-ability in
transfection efficiency, espe-cially in difficult-to-transfect
cells. Tocircumvent these limitations, expressionvectors currently
in use employ siRNAor shRNA expression cassettes thatresemble
pre-miRNAs and undergoprocessing by Dicer. Like syntheticsiRNAs,
they are designed to pair per-fectly with the target mRNA to
induceRNAi. These shRNAs are designed foreither transient or
long-term genesilencing and can be produced fromplasmid or viral
expression vectors(Amarzguioui et al. 2005).
Plasmid vectors
shRNA, siRNA, and miRNA can beproduced from plasmid vectors
con-taining promoters that are dependent
on either RNA polymerase (Pol) II or Pol III. Among them,Pol III
promoters are used most frequently because it ispossible to express
small RNAs that carry the structuralfeature of siRNA. Figure 2 is a
schematic representation ofdifferent strategies used to create
expression cassettes usingRNA polymerase promoters for generation
of siRNA,shRNA, and miRNA. In the first strategy, the sense
andantisense strands are expressed as two independent tran-scripts
that hybridize within the cells to form functionalsiRNA duplexes
(Fig. 2A). In the second strategy, the senseand antisense strands
are expressed as a single transcriptseparated by a short loop of
4–10 nt of sequence. Thetranscript forms a hairpin structure that
can be processedby Dicer into functional siRNA (Fig. 2B). In the
thirdstrategy, miRNAs that are complementary to the targetgene are
expressed using the Pol II promoter (Fig. 2C;Dykxhoorn et al. 2003;
Sioud 2004).
Two different Pol III promoters predominantly beingused for
shRNA expression are the U6 promoter and theH1 promoter
(Brummelkamp et al. 2002; Paddison et al.2002). In addition to
these two promoters, other Pol IIIpromoters such as 5S, 7SK, and
tRNA promoters have alsobeen used for expressing siRNAs
endogenously (Paul et al.2002; Czauderna et al. 2003b; Kawasaki and
Taira 2003).The activities of these promoters vary from cell type
to celltype. To optimize shRNA expression, it is beneficialto
create expression vectors with at least two differentpromoters and
transfect them into the cells being targeted
FIGURE 2. Schematic representation of expression cassettes using
RNA polymerase pro-moters for generation of small-interfering RNAs
(siRNA). (A) Tandem-type promoters expresssense and antisense
strands individually. After transcription, both strands hybridize
forming aduplex siRNA. (B) Short hairpin RNAs (shRNA) are expressed
as a single transcript separatedby a short loop of 4–10 nt of
sequence. The transcripts form a hairpin structure that can
beprocessed by Dicer into functional siRNAs. (C) Expression of
imperfect duplex hairpinstructures that is based on pre-microRNA
(pre-miRNA) structures. pre-miRNAs are processedby Dicer into a
mature miRNA, which can direct gene silencing.
siRNA delivery and targeting
www.rnajournal.org 435
JOBNAME: RNA 13#4 2007 PAGE: 5 OUTPUT: Tuesday March 6 18:39:21
2007
csh/RNA/131733/rna4598
Fig. 2 live 4/C
Cold Spring Harbor Laboratory Press on February 12, 2009 -
Published by rnajournal.cshlp.orgDownloaded from
http://rnajournal.cshlp.org/http://www.cshlpress.com
-
for gene knockdown. Construction of shRNA expressionvectors
poses two serious technical challenges. First, it isdifficult to
sequence constructs that contain a hairpinregion, probably because
of the tight palindromic structure.Second, z20%–40% of constructs
get mutated within thehairpin region (Brummelkamp et al. 2002;
Paddison et al.2002; Paul et al. 2002).
Viral vectors
The introduction of siRNA expression plasmids into cellsoften
requires electroporation, microinjection, or complexformation with
synthetic carriers (lipids, polymers, orpeptides). While most
rapidly dividing cell lines are easilytransfected using shRNA
expression plasmids, these plas-mid vectors are not easily
transfected into primary cells,stem cells, and nondividing cells.
In the absence of celldivision, the siRNA expression plasmids
cannot be intro-duced into the nucleus, where the DNA is
transcribed. Toovercome this limitation, different viral vectors
encodingshRNA including retroviral, adenoviral, and
adeno-associ-ated viral (AAV) are being developed. Typically,
thesevectors use a Pol III promoter, such as U6, H1, or transferRNA
promoters (Dykxhoorn et al. 2003).
Retroviral vectors have been reported to mediate anefficient and
stable siRNA expression (Rubinson et al. 2003;Liu et al. 2004a).
Unlike Moloney murine leukemia virus(MoMLV), lentiviral vectors
efficiently integrate into thegenome of nondividing cells, such as
pancreatic islets,hematopoietic stem cells, or terminally
differentiated cells.A lentiviral vector encoding shRNA has been
shown toeffectively silence GFP, BCL-2, and Interleukin (IL)
12receptor (CD25) genes (Tiscornia et al. 2003; Schomberet al.
2004; Wong et al. 2005). Lentiviral vectors encodingshRNA have also
been shown to inhibit HIV-1 infection inhematopoietic stem cells
and human CD4+ T-cells (Li et al.2003; Nishitsuji et al. 2006).
However, lentiviral vectors areassociated with infrequent
insertional mutagenesis. Conse-quently, the use of adenoviral
vectors is being exploredsince these vectors do not integrate into
the host genomeand efficiently transduce both dividing and
nondividingcells (Wu et al. 2003). Adenoviruses contain a
lineardouble-stranded DNA genome that remains episomal
afterinfection. Recombinant adenoviral vectors containingexpression
cassettes of interest are readily generated andcan be purified to
very high titers (up to 1013 infectionunits/mL) (Huang and Kochanek
2005).
A variety of properties make AAV vectors an interestingtool for
organ-directed shRNA expression, including thelack of
pathogenicity, and ease of vector production. TheAAV-2 type is
highly prevalent in the human populationand frequently neutralized
by antibodies. Therefore, it isimportant to evaluate different AAV
serotypes for organ-directed shRNA expression. AAV-8 vectors
expressingshRNA are reported to transduce almost 100% of
hepato-cytes after intravenous injection into mice (Grimm and
Kay
2006). However, overexpression of shRNA caused length-and
dose-dependent liver injury and ultimate death.Morbidity was
associated with the down-regulation ofliver-derived miRNA,
indicating possible competitionbetween miRNA and shRNA for nuclear
Krayopherinexportin 5 (Grimm et al. 2006). Therefore, it is
importantto optimize siRNA dose, length, and sequence to avoid
over-saturation of endogenous small RNA pathways.
Inducible expression vectors
Plasmid- and viral-vector-based constitutive expressions
ofshRNAs by RNA Pol III U6 and H1 promoters often fail toadjust the
levels of gene silencing necessary for cell survivaland growth.
Besides, gene silencing for longer periods mayresult in
nonphysiological responses. This problem can becircumvented by
generating inducible regulation of RNAi(Czauderna et al. 2003b;
Wiznerowicz and Trono 2003;Gupta et al. 2004). Ohkawa and Taira
(2000) described thesuccessful regulation of gene silencing by the
integration ofthe bacterial tetracycline operon sequence (tetO)
into theU6 promoter. Matsukura et al. (2003) have applied
thistetracycline-inducible U6 promoter for in vivo transcrip-tion
of shRNA, which enables stable transfection followedby conditional
expression of shRNA. The main drawback ofthe tetracycline-inducible
system is a relatively high back-ground expression in the uninduced
state in certain celllines (No et al. 1996; Van Craenenbroeck et
al. 2001). Incontrast, the ecdysone-inducible system is tightly
regulated,with no expression in the uninduced state and a
rapidinductive response (Gupta et al. 2004).
Owing to the time-consuming process of cloning siRNAinto plasmid
constructs and the need for verification of thecloned sequence, an
easier approach of screening sequencesinvolving the production of
siRNA expression cassettes(SECs) was developed (Castanotto et al.
2002). SECs arePCR-derived siRNA expression templates
introduceddirectly into the cells. SECs consist of a Pol III
promoter,a sequence encoding an shRNA, and an RNA polymeraseIII
termination site. The final result is a PCR product thatcontains a
Pol III promoter, a DNA sequence that, oncetranscribed, forms
shRNA, and a terminator sequence.After transcription, the shRNA is
intracellularly cleavedinto siRNA by the RNAi machinery. By
incorporatingrestriction sites at their ends, SECs effectively
elicit genesilencing and can be cloned into a plasmid to create
ansiRNA expression vector. The expediency and low cost ofthis
procedure lends itself to mass screening of siRNAlibraries as well
as identification of siRNA target sites(Chang 2004; Wooddell et al.
2005).
BARRIERS TO RNAi-BASED THERAPIES
RNAi intersects a number of pathways and siRNA trans-fection
often activates immune response leading to unin-tended and
off-target effects. Therefore, the potential
De Paula et al.
436 RNA, Vol. 13, No. 4
JOBNAME: RNA 13#4 2007 PAGE: 6 OUTPUT: Tuesday March 6 18:39:31
2007
csh/RNA/131733/rna4598
Cold Spring Harbor Laboratory Press on February 12, 2009 -
Published by rnajournal.cshlp.orgDownloaded from
http://rnajournal.cshlp.org/http://www.cshlpress.com
-
induction of inflammatory cytokines and interferon
(IFN)responses by siRNAs represent a major obstacle for theiruse as
inhibitors of gene expression. Therapeutic applica-tions of siRNAs
require a better understanding of themechanisms that trigger such
unwanted effects.
Immune response activation
Figure 3 is a schematic representation of the immuneresponse
activation by siRNAs. Non-immune and immunecells get activated in
the presence of long dsRNA, leading tothe activation of cytoplasmic
receptors such as the dsRNA-dependent protein kinase R (PKR) and
the retinoic acid-inducible gene-I (RIG-I) (Saunders and Barber
2003;Yoneyama et al. 2004). PKR is activated by
autophosphor-ylation following binding to dsRNA. Once activated,
itphosphorylates the eukaryotic translation initiation
factor(EIF-2)-a, leading to the global suppression of
proteinbiosynthesis and subsequent programmed cell death. PKRcan
also activate nuclear factor kB (NF-kB) with conse-
quent induction of type-I IFN production. A family of
29-59-oligoadenylate synthetases (2959-OAS) is also activatedby
dsRNA. This leads to the activation of RNase L, whicheventually
triggers the nonspecific degradation of mRNA(Marques and Williams
2005).
RIG-I is an intracellular dsRNA sensor capable oftriggering IFN
production. Diverse cell types derived fromRIG-I knockout mice have
impaired responses to viraldsRNA, establishing the essential role
of RIG-I in themammalian antiviral response (Yoneyama et al.
2004;Marques and Williams 2005). The helicase RIG-I isnecessary for
recognizing blunt-ended siRNAs, leading tothe activation of dsRNA
signaling. RIG-I unwinds siRNAscontaining blunt ends more
efficiently than siRNAs con-taining 39 overhangs. The efficiency in
duplex unwinding isthen translated into signaling downstream to
interferonregulatory factor 3 (IRF-3) and NF-kB activation
(Marqueset al. 2006).
Mammalian immune cells express a family of toll-likereceptors
(TLRs), which recognize pathogen-associated
molecular patterns including unmethy-lated cytosine-guanine
motifs (com-monly known as CpG motifs) and viraldsRNA. TLR3, the
receptor for dsRNA,was a logical candidate for recognizingsiRNA in
the context of immunostimu-lation. Indeed, TLR3 overexpressedin
cultured human embryonic kidney(HEK) 293 cells was capable
ofrecognizing siRNAs (Marques andWilliams 2005). However, the
activationof immune cells by siRNAs is sequencedependent, and
either sense or anti-sense strands individually can
induceinflammatory cytokine production asefficiently as duplex
siRNAs (Judge et al.2005; Sioud 2005, 2006). Thus, as TLR3neither
recognizes single-strandedRNA (ssRNA) nor requires
sequencespecificity, it is unlikely involved inactivation of the
immune system bysiRNAs (Marques and Williams 2005;Sioud 2005).
TLR7 and TLR8 were initially shownto mediate the recognition of
RNAviruses and small synthetic antiviralcompounds referred to as
imidazoqui-nolines. TLR7, TLR8, and TLR9 areexpressed in endosomes
and requireendosomal maturation for efficient sig-naling (Marques
and Williams 2005).siRNA recognition by TLR7, TLR8, andTLR9 results
in activation of NF-kBand IRFs, which induce inflammatorycytokines
and IFNs, respectively (Fig. 3;
FIGURE 3. Activation of immune response by siRNAs. siRNA/shRNA
can be delivered bytwo different ways: (1) Delivery of viral vector
followed by endogenous expression of shRNA inthe nucleus, which is
transported to the cytoplasm and processed by Dicer into siRNAs.The
siRNA causes sequence-specific mRNA degradation and does not induce
an interferon(IFN) response. (2) Synthetic delivery systems forming
an siRNA complex particle, forexample, liposome. In this way, the
siRNA is taken up by the cell via endocytosis,
causingsequence-specific mRNA cleavage through RNAi and activation
of the immune response. Theimmune response can be activated by
different pathways: (a) recognition of siRNAs by Toll-like
receptors (TLRs); (b) activation of retinoic acid-inducible gene I
(RIG-I) by blunt-endsiRNAs; (c) activation of dsRNA-dependent
protein kinase R (PKR); (d) activation of 29-59-oligoadenylate
synthetases (2959-OAS). Generally, activation of these pathways
leads to theinduction of several cellular factors, including
nuclear factor kB (NF-kB), interferon regulatoryfactors (IRFs),
eukaryotic translation initiation factor 2a (EIF-2a), and RNase L.
Induction ofinterferons (IFNs), inflammatory cytokines, nonspecific
mRNA degradation, and generalinhibition of protein synthesis are
some undesired effects of the immunostimulation caused
bysiRNAs.
siRNA delivery and targeting
www.rnajournal.org 437
JOBNAME: RNA 13#4 2007 PAGE: 7 OUTPUT: Tuesday March 6 18:39:32
2007
csh/RNA/131733/rna4598
Fig. 3 live 4/C
Cold Spring Harbor Laboratory Press on February 12, 2009 -
Published by rnajournal.cshlp.orgDownloaded from
http://rnajournal.cshlp.org/http://www.cshlpress.com
-
Marques and Williams 2005). Interestingly, not all
ssRNAmolecules are capable of activating TLR7 and TLR8, but,rather,
U- and G-rich ssRNAs seem to be preferentiallyrecognized (Sioud
2006). It has now become clear thatTLR7 and TLR8 mediate the
recognition of siRNAs in asequence-dependent manner (Heil et al.
2004; Hornunget al. 2005; Judge et al. 2005).
Recognition of siRNAs by TLRs takes place in theendosome, before
the siRNAs enter the cytoplasm. There-fore, it is expected that if
siRNAs could enter the cytoplasmavoiding the endosome, they should
bypass the activationof immune systems but still mediate gene
silencing. Robbinset al. (2006) tested the immunostimulatory
effects of lipid-delivered siRNAs versus Pol III promoter-expressed
shRNAsin primary CD34+ progenitor-derived hematopoietic cells.They
observed that in this system, cationic lipid/siRNA com-plexes are
potent inducers of IFN-a and type-I IFN geneexpression, whereas the
same sequences when expressedendogenously are
nonimmunostimulatory.
The method of siRNA delivery also has a great influenceon the
immunostimulatory activity of siRNAs (Fig. 3).Injection of siRNAs
after complex formation with cationicliposomes into mice induces
the release of inflammatorycytokines, including IL-6, tumor
necrosis factor-a (TNF-a),and IFN-a (Judge et al. 2005; Sioud
2005). In contrast,injection of naked, unmodified siRNAs or siRNAs
conju-gated to cholesterol has no significant effect on
immuno-stimulation (Heidel et al. 2004; Marques and Williams2005).
There are two possible explanations for the absenceof
immunostimulation by naked siRNAs. First, unmodifiedsiRNAs have a
short half-life in serum and may be degradedbefore being recognized
by specific receptors. Second,cationic lipid/siRNA complexes are
more readily recog-nized by immune cells than are naked siRNAs. In
fact,Sioud (2005) reported that naked siRNAs containing
aphosphorothioate (PS) backbone were not immunostimu-latory.
However, these results should be interpreted care-fully considering
that naked CpG ODNs are known to beimmunostimulatory independent of
backbone modifica-tions (Marques and Williams 2005).
Off-target effects and nonspecific gene silencing
For its therapeutic applications, siRNA must not cause
anyunintended effect other than sequence-specific gene silenc-ing.
However, recent studies indicate that there are off-target effects
associated with the use of siRNA (Jacksonet al. 2003). Off-target
effects consist of any gene silencingeffect caused by siRNAs in
nontarget mRNAs through theRNAi mechanism. Generally, it occurs due
to lack ofcomplementarity between siRNAs and target mRNAs.
Anexplanation for the fact that siRNAs can induce silencing
ofnontarget genes can be found in the RNAi machinery.Although the
actual substrate specificity of individualsiRNAs appears to be very
high, RNAi machinery can
tolerate single mutations located in the center of the
siRNAmolecule without losing the ability to induce gene silenc-ing.
It makes some siRNAs able to silencing other thanthose genes
related to their homologous mRNA. However,complete inactivation of
the RNAi mechanism can occur iffour or more mutations are
introduced in the guide strand,making the active RISC unable to
recognize the targetmRNA (Jackson et al. 2003; Persengiev et al.
2004; Sioud2004).
siRNAs and miRNAs were found to be functionallyinterchangeable.
siRNAs can act as miRNAs dependingon the degree of complementarity
with the target mRNA. Ifsynthetic siRNAs bear a sufficiently low
degree of comple-mentary bases, target mRNA translation occurs
instead ofmRNA degradation, whereas miRNAs will lead to
mRNAdegradation if perfect complementarity with target mRNAexists.
This process is one of the reasons for the off-targeteffects of
siRNAs (Jackson et al. 2003; Bartel 2004). Anotherreason for the
occurrence of off-target effects is that notonly the antisense but
also the sense strand of an siRNA candirect gene silencing of
nontarget genes, and it has beendocumented to occur when as few as
15 bp of comple-mentarity exists between the siRNA and mRNA
(Jacksonet al. 2003). Birmingham et al. (2006) also demonstrated
thatthe presence of one or more perfect matches between thehexamer
or heptamer seed (positions 2–7 or 2–8 of theantisense strand) of
an siRNA and the 39-untranslatedregion (UTR), but not the 59-UTR or
open reading frame,is associated with off-targeting. A high
proportion of ‘‘off-target’’ transcripts silenced by siRNAs have
been shown tohave 39-UTR sequence complementarity to the seed
regionof the siRNA, since the base mismatches within the siRNAseed
region reduced the set of original off-target transcripts.Since
there is no algorithm that can eliminate significantnumbers of
7–8-base matches of siRNAs to the tran-scriptome, it will be
difficult to achieve perfect specificity(Jackson et al. 2006).
These findings suggest a strongmechanistic parallel between siRNA
off-targeting andmiRNA-mediated gene regulation.
Nonspecific gene silencing is those effects caused bysiRNAs
through any pathway other than RNAi. Semizarovet al. (2003)
investigated the overall cellular effects of siRNAson transcription
levels. In a concentration-dependentmanner, siRNAs nonspecifically
interfered in the expressionof a significant number of genes, many
of which are knownto be involved in apoptosis and stress response.
Persengievet al. (2004) reported that treatment with siRNA may
resultin diverse cellular activities, such as cell signaling,
cytoskel-etal organization, gene expression, metabolism, and
celladhesion. Jackson et al. (2003) attributed the
nonspecificeffects to cross-hybridization of transcripts
containingregions of partial homology with the siRNA
sequence.However, the nonspecific effects observed by Persengievet
al. (2004) cannot be explained by off-target regulationbecause the
siRNAs tested lacked significant sequence
De Paula et al.
438 RNA, Vol. 13, No. 4
JOBNAME: RNA 13#4 2007 PAGE: 8 OUTPUT: Tuesday March 6 18:39:43
2007
csh/RNA/131733/rna4598
Cold Spring Harbor Laboratory Press on February 12, 2009 -
Published by rnajournal.cshlp.orgDownloaded from
http://rnajournal.cshlp.org/http://www.cshlpress.com
-
similarity to any human gene. Moreover, for all of the
genestested, expression was affected by two siRNAs withsequences
that are completely unrelated. Bridge et al.(2003) also reported
that shRNAs can affect the expressionof many genes, including
several IFN targets. The activationof sequence-independent
inhibition of gene expression bysiRNA or shRNA-expression vectors
seems to be, at least inpart, due to the induction of type-I IFN,
PKR, andsignaling through TLR3 (Kariko et al. 2004; Marques
andWilliams 2005).
BIODISTRIBUTION OF siRNA
To develop effective delivery systems of chemically synthe-sized
siRNA, it is important to gain in-depth understandingof its
biodistribution and pharmacokinetic properties atwhole-body, organ,
and cellular levels.
siRNA duplexes are much more stable than ssRNAs.However, since
they are similar to single-stranded antisenseoligonucleotides, they
still tend to degrade on incubationwith serum, contributing to
their short half-lives in vivo.125I-labeled siRNAs are widely
distributed in the body afterintravenous injection into mice, with
preferential accumu-lation in the liver and kidney (Braasch et al.
2004).Radiolabeled siRNA is also distributed to the heart,
spleen,and lung. However, very little siRNA is detected in thebrain
(Braasch et al. 2004; Soutschek et al. 2004; Santelet al. 2006).
Intravenous administration of naked siRNA inrats showed a short
half-life of 6 min and a clearance of17.6 mL/min. The poor
pharmacokinetic properties ofsiRNAs are related to their low in
vivo stability, since theycan be degraded by endogenous RNases.
Another reason istheir fast elimination by kidney filtration
because of theirsmall molecular mass (z7 kDa) (Soutschek et al.
2004).Consequently, improving the pharmacokinetic propertiesof
siRNA duplexes is an important goal for developingsiRNA-mediated
gene silencing.
STRATEGIES FOR IMPROVING RNAi-BASEDTHERAPIES
To develop strategies for improving RNAi, we need toaddress
several problems, including the enzymatic stability,cellular
uptake, pharmacokinetic properties, and potentialfor off-target
effects and immunoactivation of siRNAs. Thesequence design of siRNA
molecules is extremely importantto improve their efficacy as well
as to reduce the potentialfor off-target effects and activation of
the immune response(Reynolds et al. 2004). Backbone modifications
and bio-conjugation with lipids and peptides are known to
improvethe stability and cellular uptake of siRNAs (Chiu and
Rana2003; Lorenz et al. 2004). Several reagents have also
beenemployed for systemic administration of siRNAs as analternative
to the covalent modifications of the siRNAmolecule (Simeoni et al.
2003; Morrissey et al. 2005b;
Urban-Klein et al. 2005; Santel et al. 2006). Cationicliposomes
are routinely used for transfection of nucleicacids into mammalian
cells. Positively charged carriers,such as protamine-antibody
fusion proteins, polyethylenei-mine (PEI), and
cyclodextrin-containing polycation, havebeen tested for complex
formation with siRNAs. To achievecell-specific delivery of siRNAs,
several targeting ligands,including cell membrane receptors and
antibodies, havebeen explored (Song et al. 2005).
Design of siRNAs
A thorough understanding of the sequence, size, andstructural
requirements of siRNAs is essential to effectivelymediate RNAi.
Table 2 summarizes the most importantsiRNA design rules to improve
gene silencing and avoidundesired effects.
siRNA sequence
Some research groups have elaborated several guidelines
fordesigning siRNAs that can potentially silence gene expres-sion
(Elbashir et al. 2002; Paddison et al. 2002; Reynoldset al. 2004).
Several sequence motifs are consistent witheffective siRNAs,
including AAN19TT, NAN19NN, NAR-N17YNN, and NANN17YNN (where N is
any nucleotide, Ris a purine, and Y is a pyrimidine). In principle,
any regionof the mRNA can be targeted. However, it should bepointed
out that complementarity is often found in the 39-UTR of the
off-target sequences. Furthermore, sequencecomplementarity between
the 59 end of the guide strandand the mRNA is known to be the key
to off-targetsilencing. Additionally, intronic sequences must be
avoidedsince RNAi is a cytoplasmic process (Dykxhoorn et al.
2003).
Ui-Tei et al. (2004) proposed that highly effective RNAican be
achieved if siRNA satisfying the following foursequence conditions
at the same time is used: (1) AUrichness in the 59-terminal,
7-bp-long region of theantisense strand; (2) G/C at the 59 end of
the sense strand;and (3) the absence of any long GC stretch of
>9 bp inlength. Even though Elbashir et al. (2002) reported that
thetarget region should be at least 50–100 nt downstream fromthe
start codon, Dykxhoorn et al. (2003) suggest that thereis a
predisposition for effective siRNA-directed silencingtoward the 39
portion of the gene. Moreover, sequenceswith even representation of
all nucleotides on the antisensestrand are favored, and those
regions with stretches ofa single nucleotide, especially G, should
be avoided asG-rich oligonucleotides have a tendency to form
quartets.Reynolds et al. (2004) analyzed 180 siRNAs targeting
everyother position of two 197-base regions of firefly
luciferaseand human cyclophilin B mRNA in a total of 90 siRNAsper
gene. A two-base shift in the target position wassufficient to
significantly alter siRNA activity, suggestingthat its
functionality is determined by the siRNA-specificproperties. Most
potent siRNA has a G/C content ranging
siRNA delivery and targeting
www.rnajournal.org 439
JOBNAME: RNA 13#4 2007 PAGE: 9 OUTPUT: Tuesday March 6 18:39:44
2007
csh/RNA/131733/rna4598
Cold Spring Harbor Laboratory Press on February 12, 2009 -
Published by rnajournal.cshlp.orgDownloaded from
http://rnajournal.cshlp.org/http://www.cshlpress.com
-
from 36% to 52%. siRNA duplexes with a GC content>52% may
have difficulty in dissociating into passengerand guide strands,
while siRNAs with a GC content 19 nt of sequence
25–30-nt disRNAs are more potent than conventional
21–23-ntsiRNAs
Long dsRNA (;30 nt) can activate dsRNA-dependent PKRresponse
Structure The A-form helix of siRNA-mRNA is required for
themechanism of RNAi
29-OH groups of uridines are associated with immunerecognition
of siRNAs
Modifications that do not completely hinder the unwinding ofthe
siRNA duplex are allowed since it is not a requirement
forhybridization of siRNA with mRNA and induction of RNAi
Single-strand siRNAs are more immunostimulatory than
theirrespective double-strand siRNAs
29-OH groups are related to enzymatic stability but are
notrequired to mediate RNAi
Blunt structure at the 39 end is the strongest terminal
structurefor promoting activation of dsRNA-dependent PKR
Asymmetrical 25–30-nt siRNAs with 59 blunt ends and 2-nt39 ends
are more potent than symmetrical 2-nt overhangscontaining
siRNAs
(disRNA) Dicer substrate siRNA.
De Paula et al.
440 RNA, Vol. 13, No. 4
JOBNAME: RNA 13#4 2007 PAGE: 10 OUTPUT: Tuesday March 6 18:39:44
2007
csh/RNA/131733/rna4598
Cold Spring Harbor Laboratory Press on February 12, 2009 -
Published by rnajournal.cshlp.orgDownloaded from
http://rnajournal.cshlp.org/http://www.cshlpress.com
-
effectively activate innate immunity in the context
ofdouble-strand siRNA to a level comparable to that obtainedwith
free single-strand RNA. siRNA duplexes seem to beless effective
than siRNA in activating innate immunity.The intracellular
receptors for single-strand siRNA, inparticular TLR8 and TLR7, do
not effectively recognizemost immunostimulatory RNA motifs in the
context ofsiRNA duplexes (Sioud 2006).
A number of academic and commercially affiliated Web-based
softwares have been developed to assist researchers inthe
identification of efficient siRNA sequences (Naito et al.2004; Yuan
et al. 2004). Since the rules that govern efficientsiRNA-directed
silencing are still unknown, researchersseeking to silence gene
expression should synthesize severalsiRNAs to a gene and validate
empirically the efficiency ofeach one. To ensure that the chosen
siRNA sequence targetsa single gene, a search of the selected
sequence should becarried out against sequence databases. This can
be doneusing the Smith–Waterman algorithm or the Basic
LocalAlignment Search Tool (BLAST) located at the NationalCenter of
Biotechnology Information Web site (http://www.ncbi.nlm.gov).
Sequences in these databases thatshare partial homology with siRNAs
might be targetedfor silencing by the siRNAs. Potential off-target
effects ofthe siRNAs might be minimized by choosing an siRNAwith
maximal sequence divergence from the list of geneswith partial
sequence identity to the intended mRNA target.
siRNA size
Synthetic RNA duplexes of 25–30 nt in length (morespecifically,
27 nt), which are Dicer substrates, have beenshown to be up to
100-fold more potent than correspond-ing conventional 21-nt siRNAs
(Kim et al. 2005a). The useof 27-nt dsRNAs, also called Dicer
substrate dsRNA(disRNA), allows targeting of some sites within a
givensequence that are refractory to suppression with
traditional21-nt siRNAs. The use of disRNAs to trigger RNAi
canresult in enhanced and prolonged gene silencing at
lowerconcentrations than those required for conventional
21-ntsiRNAs applications (Kim et al. 2005a). It is important tonote
that in vitro Dicer cleavage of 27-nt disRNAs beforetransfection
does not significantly enhance gene silencingefficiency.
Additionally, it has been shown that chemicallysynthesized shRNAs
that are Dicer substrates are morepotent inducers of RNAi than
conventional siRNAs. More-over, a two-base 39 overhang directs
Dicer cleavage (Siolaset al. 2005). The enhanced potency of the
longer duplexes isattributed to the fact that they are substrates
of the Dicerendonuclease, directly linking the production of siRNAs
forincorporation into the RISC (Kim et al. 2005a).
The size of siRNA also plays an important role in theactivation
of immune response in a sequence-independentmanner. Although it was
initially postulated that aminimum of 30 nt was necessary to
activate dsRNAsignaling through PKR activation, shorter siRNAs, as
few
as 21–23 nt, are capable of binding and activating PKR invitro
(Marques et al. 2006). In addition, Hornung et al.(2005) observed
that 12-nt ssRNAs containing the immu-nostimulatory motif
(GUCCUUCAA) were poor inducersof IFN-a in PDCs. However, the
investigators observedthat the motif identified needs to be part of
a longer ODNsequence, at least 19 bases, to induce IFN-a
activation. Theapproach reported by Kim et al. (2005a), which
consists ofDicer substrate 27-nt dsRNA with increased RNAi
potencyrelative to conventional 21-nt siRNAs, may facilitate theuse
of lower concentrations of duplex RNA. Therefore,since the
undesired effects can occur in an siRNA dose-dependent manner, the
use of Dicer substrate 27-nt dsRNAmay be an alternative to overcome
this problem withoutlosing efficacy.
siRNA structure
According to Chiu and Rana (2002), the A-form helix of theguide
strand-mRNA duplex is required for the mechanism ofRNAi. Chemical
modifications disrupting the functionalgroups of the major groove
of the A-form helix formed bythe guide strand and its mRNA
specifically at the cleavagesite inhibit RNAi. The major groove is
also required forpromoting RNA–RISC interactions that subsequently
lead tomRNA cleavage (Chiu and Rana 2002, 2003).
The 29-OH of the ribonucleotide of RNAs is required forthe
nucleophilic attack during the hydrolysis of the RNAbackbone, the
reaction catalyzed by degradative RNases(Chiu and Rana 2003).
However, 29-OH is not required forRNAi-mediated degradation, and,
even more specifically, itis not required for nucleotide
base-pairing with nucleotideslining the mRNA cleavage site (Chiu
and Rana 2003).Additionally, Sioud (2006) has shown that the 29-OH
ofuridines is required for immune recognition and signalingof
siRNAs, being responsible, at least in part, for activationof the
immune response.
New design approaches that improve the performance oflong dsRNAs
as Dicer substrates have been reported (Kimet al. 2005a). These new
approaches include 25–30-ntasymmetric dsRNAs with a 59 blunt end
and a 2-nt 39overhang on the other end. The improved efficacy
ofdisRNAs is postulated to result from their recognitionand
cleavage by Dicer, followed by their subsequentincorporation into
the RISC complex. This interpretationis supported by observations
that Drosophila Dicer is notonly instrumental in handing over siRNA
to nascent RISC,but it is itself a component of the latter (Lee et
al. 2004).Providing the RNAi machinery with a Dicer
substratepresumably results in more efficient incorporation of
theactive 21-nt siRNA strand into RISC. These modificationsalso
direct the way Dicer processes the dsRNA substrate,resulting in a
more homogeneous pool of siRNA products.
The extremities of the siRNA duplex are critical deter-minants
of its capacity to trigger immune response activa-tion. A blunt
structure at the 39end is the strongest terminal
siRNA delivery and targeting
www.rnajournal.org 441
JOBNAME: RNA 13#4 2007 PAGE: 11 OUTPUT: Tuesday March 6 18:39:45
2007
csh/RNA/131733/rna4598
Cold Spring Harbor Laboratory Press on February 12, 2009 -
Published by rnajournal.cshlp.orgDownloaded from
http://rnajournal.cshlp.org/http://www.cshlpress.com
-
structure for promoting activation of dsRNA signalingthrough the
PKR pathway, followed by a 59overhang(Marques et al. 2006). In
contrast, 39overhangs, normallypresent in endogenous Dicer products
such as miRNAs,allow RNAi to proceed without activation of
dsRNA-dependent PKR (Marques et al. 2006). The presence ofthe 39
overhangs also precludes activation of dsRNAsignaling by siRNAs and
reveals an important basis fordiscrimination between self and
nonself dsRNAs. Interest-ingly, this type of discrimination appears
to take place afterRIG-I binds to dsRNAs (Marques et al. 2006).
Chemical modifications
Chemical modifications to sugars, backbones, or bases
ofoligonucleotides can be used to control their pharmacoki-netic
profiles and reduce nonspecific effects without affect-ing their
biological activity. The valuable know-how ofbackbone modifications
of antisense ODNs can be readilyadapted to develop new siRNA
technologies (Marshall andKaiser 2004). Although siRNA duplexes
used for silencingare inherently more stable than ssRNAs, there are
reasonsto chemically modify one or both strands. Apart
fromimproving stability and reducing off-target effects,
chemicalmodifications can aid in broadly targeting siRNAs into
cellsand tissues, and certain conjugates can enhance uptake
inspecific cell types. Multiple chemical modifications can
beintroduced at various positions with the siRNA duplexes.Figure 4
shows the most common chemical modificationsintroduced in
siRNAs.
Phosphodiester modifications
One of the simplest modifications is the introduction
ofphosphorothioate (PS) linkages, which can be prepared byreplacing
one of the two nonbridging oxygen atoms in theinternucleotide
linkage of RNA by a sulfur atom. PSlinkages are known to reduce
cleavage by nucleases andincrease the half-life of ODNs in vivo.
However, oligonu-cleotides with extensive PS linkages are also
known toincrease binding to serum proteins and can be toxic in
vivo(Manoharan 2004; Mahato et al. 2005).
Braasch et al. (2003) reported that PS siRNAs containing12–21 PS
substitutions per strand were stable duringextended incubation in
fetal calf serum media, but theirstability was not higher than
unmodified siRNA duplexes.The investigators also monitored the
inhibition of humancaveolin-1 (hCav) expression by PS siRNAs. The
modifiedsiRNAs were able to silence the hCav gene, but
lessinhibition was observed by fully modified siRNAs. Thenuclear
uptake of PS siRNAs was greater than that ofunmodified siRNAs,
which can explain their reducedactivity. Although backbone
modifications did not reducethe silencing activity of siRNAs
targeting the HTF gene inhuman keratinocytes, most extensively PS
siRNAs provedto be cytotoxic, resulting in z70% cell death
compared
with mock-transfected cells (Amarzguioui et al. 2003).
Theintroduction of PS linkages to siRNA yields mixed resultsfor
siRNA distribution to various organs. In another study,Braasch et
al. (2004) observed an increase in the distribu-tion of siRNAs to
spleen, heart, and lung, while thedistribution to kidney and liver
decreased. However, theeffects were modest, suggesting that the
introduction of PSlinkages may not play a major role in determining
thedistribution of siRNA.
An alternative backbone modification that confersincreased
biological stability to nucleic acids is the bor-anophosphonate
linkage. In boranophosphonate ODNs,the nonbridging phosphodiester
oxygen is replaced withan isoelectronic borane (–BH3) moiety. The
charge distri-bution of boranophosphonate ODNs also differs from
thatof normal phosphate and phosphorothioate ODNs. Thus, itchanges
the polarity and increases the hydrophobicity ofthe molecule. While
boranophosphonate modification hasbeen less widely studied, it has
shown more advantagesthan PSs. Targeting GFP in HeLa cells, Hall et
al. (2004)observed that the activity of boranophosphonate
siRNAsexceeds that of PS siRNAs, and it was often greater thanthat
of unmodified siRNAs, irrespective of the base orstrand modified.
However, boranophosphonate modifica-tions placed at the center of
the antisense strand reducedRNAi activity. Boranophosphonate siRNAs
modified atminimal levels also showed improved stability over
unmod-ified siRNAs against nuclease degradation. Additionally,
FIGURE 4. Most common chemical modifications introduced
insiRNAs. (A) Phosphodiester modifications: unmodified siRNA
(phos-phodiester RNA); phosphorothioate RNA; boranophosphonate.
(B)29-Sugar modifications: 29-O-Methyl RNA; 29-deoxy-29-fluoro
RNA;locked nucleic acid (LNA).
De Paula et al.
442 RNA, Vol. 13, No. 4
JOBNAME: RNA 13#4 2007 PAGE: 12 OUTPUT: Tuesday March 6 18:39:45
2007
csh/RNA/131733/rna4598
Cold Spring Harbor Laboratory Press on February 12, 2009 -
Published by rnajournal.cshlp.orgDownloaded from
http://rnajournal.cshlp.org/http://www.cshlpress.com
-
increasing the number of modifications yielded more
stablesiRNAs.
29-Sugar modifications
Modifications of RNA at the 29-position of the ribose ringhave
been shown to increase siRNA stability againstendonucleases and
reduce immune response activation.These modifications include
29-O-methyl (29-OMe), 29-deoxy-29-fluoro modifications, and locked
nucleic acid.
Fluoro and methyl linkages. The siRNA motif consistingof 29-OMe
and 29-fluoro nucleotides has enhanced plasmastability and
increased in vivo potency. The 29-OMe sugarmodification retains the
canonical right-handed A-formhelical geometry, which is required
for siRNA activity. Thismodification has also been shown to
increase the nucleaseresistance of ODNs and siRNA duplexes (Chiu
and Rana2003). The effect of 29-OMe modification has been foundto
be dependent on both position and extent of incorpo-ration. siRNAs
with fully 29-OMe-substituted sense strandswere functional when the
duplexes were of 20-bp bluntconstruction, but not for canonical
duplexes of 19-bpconstructs with 39 overhangs (Kraynack and Baker
2006).In contrast, siRNAs with alternating 29-OMe and unmod-ified
nucleotides (Czauderna et al. 2003a), or alternating29-OMe or
29-O-fluoro nucleotides (Allerson et al. 2005),had activity similar
to that of the unmodified duplexes.This suggests that minimal
chemical modifications arecompatible with siRNA function. Jackson
et al. (2006)reported that 29-OMe modifications to specific
positionswithin the siRNA seed region reduce both the number
ofoff-target transcripts and the magnitude of their
regulation,without significantly affecting silencing of the
intendedtargets. Current evidence suggests that 29-OMe
sugarmodifications decrease the free energy of hybridization,which
would tend to compensate for, rather than impair,weaker
base-pairing (Inoue et al. 1987; Lesnik et al. 1993).Additionally,
the effects on off-target silencing of the 29-OMe modification show
a sharp position dependence, incontrast to the broader position
dependence of basesubstitutions (Jackson et al. 2006). It is
important to notethat 29-OMe modifications impairing the cleavage
of thepassenger strand also impair target RNA cleavage. How-ever,
29-O-methyl ribose groups positioned 1 or 2 ntdownstream from the
cleavage site did not impair targetRNA cleavage (Matranga et al.
2005; Leuschner et al. 2006).
Chiu and Rana (2003) targeted the EGFP gene in HeLacells using
different siRNAs modified at the 29-OH site. Theeffects of 29-OH
modifications on RNAi were studied byreplacing uridine and cytidine
in the antisense strand ofsiRNA by 29-fluoro-uridine (29-FU) and
29-fluoro-cytidine(29-FC), respectively. These modifications
increased thesiRNA stability upon exposure to HeLa cell
extracts,without losing gene silencing activity. On the other
hand,modifying the 29-OH to a bulky methyl group to create
29-OMe nucleotides in the sense or antisense strand
greatlydiminished EGFP gene silencing, whereas
double-stranded29-OMe-modified siRNAs completely abolished RNAi.
Onereasonable explanation is that the methyl group, as a
bulkygroup, may severely limit the interactions among siRNA,target
mRNA, and the RNAi machinery that are required tosuccessfully
mediate gene silencing.
Layzer et al. (2004) observed that modification of siRNAwith
29-fluoro pyrimidines does not have an adverse effecton gene
silencing and target specificity, as seen in its abilityto
distinguish between the GL2 and GL3 luciferase targetmRNAs. In
addition, 29-F-modified siRNA showed adramatic increase in the
stability over 29-OH siRNAs inhuman plasma. More than 50% of the
29-OH siRNA wasdegraded within a minute of exposure to plasma,
andpractically all molecules were degraded after 4 h. Incontrast,
>50% of the 29-F siRNA molecules remained fulllength after 24 h
of exposure to plasma. However, aftertransfecting 29-F GL2 siRNA
into HeLa R19-Luc cells,which stably express the GL2 luciferase
gene, little differ-ence was observed in both the duration and
magnitude ofsuppression between 29-F and 29-OH siRNAs. These
resultsare consistent with those of Song et al. (2003a),
whodemonstrated that 29-OH siRNA activity is limited bydilution in
cell culture, but not by intracellular stability.Additionally, both
29-OH and 29-F siRNAs reduced lucif-erase expression in vivo by
>85% when injected in rodents,but 29-F modification of the siRNA
did not grant anyadvantage to either the magnitude or the interval
ofresponse. Although 29-F-modified siRNAs have muchgreater
stability in plasma than 29-OH siRNAs, these datasuggest that once
siRNAs are localized within the cell,modifications that enhance
nuclease stability do not influ-ence activity or persistence.
Locked nucleic acid. Locked nucleic acid (LNA), alsoreferred to
as inaccessible RNA, is a family of conforma-tionally locked
nucleotide analogs that displays unprece-dented hybridization
affinity toward complementary DNAand RNA. Commonly used LNA
contains a methylenebridge connecting the 29 oxygen with the 49
carbon of theribose ring. This bridge locks the ribose ring in the
39-endoconformation characteristic of RNA (Bondensgaard et al.2000;
Braasch and Corey 2001). LNA has been shown to becompatible with
siRNA intracellular machinery, preservingmolecule integrity while
offering several improvements thatare relevant to the development
of siRNA technology(Braasch et al. 2003; Elmen et al. 2005).
LNA may be used to increase the functional half-life ofsiRNA in
vivo by two different mechanisms: (1) enhancingthe resistance of
the constituent RNA strands againstdegradation by ssRNases, and (2)
stabilizing the siRNAduplex structure that is crucial for silencing
activity. Elmenet al. (2005) reported that the stability of siRNA
can beenhanced by conjugating LNA at the 39 ends of the sense
siRNA delivery and targeting
www.rnajournal.org 443
JOBNAME: RNA 13#4 2007 PAGE: 13 OUTPUT: Tuesday March 6 18:39:51
2007
csh/RNA/131733/rna4598
Cold Spring Harbor Laboratory Press on February 12, 2009 -
Published by rnajournal.cshlp.orgDownloaded from
http://rnajournal.cshlp.org/http://www.cshlpress.com
-
strand of siRNA. Introduction of LNA modifications at the39
overhangs in either one or both strands of siRNA againstfirefly
luciferase revealed no loss of silencing effect of siRNAin cultured
cells. Additionally, one LNA at the 59 end of thesense strand was
shown to be fully compatible withsilencing activity, while an LNA
at the 59 end of theantisense strand dramatically impaired the
silencing effect.Since the strand that displays the weakest binding
energy atits closing 59 base pair is incorporated preferentially
intothe RISC (Khvorova et al. 2003; Schwarz et al. 2003),
theconjugation of LNA at the 59 sense strand might be used toalter
strand bias in favor of incorporation of the antisensestrand. It
may help in reducing off-target effects andincreasing potency.
Additionally, the potency of siRNAsmay be improved by simply
lowering the number of RISCsloaded with the sense strand. LNA
modifications of the 39or 59 termini of the sense strand of siRNAs
can abrogatetheir immunostimulatory activity. However, depending
onthe extent and location of LNA modifications, they canblock not
only the induction of IFNs but also the genesilencing activity of
the siRNA (Hornung et al. 2005).
Bioconjugation
Bioconjugation of one or both strands of siRNAs withlipids and
polymers is often desirable to (1) further increasetheir
thermodynamic and nuclease stability, (2) improve
thebiodistribution and pharmacokinetic profiles of siRNAs,and (3)
target them to specific cell types.
Lipid conjugation
Conjugation with lipids may enhance siRNA uptake
viareceptor-mediated endocytosis or by an increased mem-brane
permeability of the otherwise negatively chargedRNA. Conjugation of
nucleic acids with cholesterol hasbeen demonstrated to enhance
cellular uptake in cellculture and hepatic deposition after
systemic administra-tion (Cheng et al. 2006). This is because
cholesterolconjugation increases the hydrophobicity and
cellularassociation of nucleic acids. Lorenz et al. (2004)
haveconjugated siRNAs with cholesterol derivatives like
lith-ocholic and lauric acids at the 59 end of the sense
strand.Cholesterol conjugation was shown to increase the
cellularuptake of siRNAs in human liver cells without use ofany
transfection reagent. Incubation of cells during 4 h with50 nM of
siRNAs with a modified sense strand down-regulated b-galactosidase
expression to a higher extent thansiRNAs with a modified antisense
strand or two modifiedstrands.
Apolipoprotein B (ApoB) is expressed predominantly inthe liver
and jejunum, and is an essential protein for theassembly and
secretion of very-low-density lipoprotein(VLDL) and low-density
lipoprotein (LDL), which arerequired for the transport and
metabolism of cholesterol.Elevated ApoB and LDL levels are
correlated with increased
risk of coronary artery disease, and inadequate control
ofLDL-cholesterol after acute coronary syndromes resultsin
increased risk of recurrent cardiac events or death.Soutschek et
al. (2004) reported the conjugation of cho-lesterol to the 39 end
of the sense strand of an siRNAmolecule targeting ApoB. The
concentration of this proteinin human blood samples correlates with
those of choles-terol, and higher levels of both compounds are
associatedwith an increased risk of coronary heart disease.
Intrave-nous administration of radiolabeled siRNA-chol
conjugatesinto rats at 50 mg/kg has shown improved
pharmacokineticproperties compared to unconjugated siRNAs.
siRNA-cholconjugates showed an elimination half-life (t1/2) of 95
minand a plasma clearance (CL) of 0.5 mL/min, whereasunconjugated
siRNAs have a t1/2 of 6 min and a CL of17.6 mL/min. Significant
levels of this conjugate werefound in the liver, heart, kidney,
lung, and adipose tissue.Most importantly, siRNAs efficiently
reduced the levels ofApoB mRNA by >50% in the liver and by 70%
in thejejunum, which resulted in a lowering of the bloodcholesterol
levels comparable to that observed in mice inwhich the ApoB gene
had been deleted.
Peptide conjugation
Protein transduction domains (PTDs) offer an alternativeto the
traditional methods of siRNA delivery. PTDs areshort amino acid
sequences that are able to interact withthe plasma membrane in a
way that leads to a highly effi-cient uptake into the cytoplasm.
Only short stretches,mainly consisting of positively charged amino
acids, suchas arginine and lysine, are responsible for
translocating thePTDs through the plasma membrane—hence the name
cell-penetrating peptides (CPPs) or membrane permeant
peptides(MPPs). Stretches of these amino acids help to maintainthe
helical structure of the PTD, which is the structuralprerequisite
for membrane penetration. Many natural andsynthetic MPPs are known
to contain this structural pro-perty of PTDs (Mahat et al. 1999;
Schepers 2005).
Cellular uptake of MPPs takes place in a receptor-independent
fashion. MPPs are mostly amphipathic mol-ecules that interact with
the negatively charged head groupsof the plasma membrane via their
positive amino acidresidues. MPPs can transfect different cell
types with veryhigh efficiency and rapid uptake kinetics. Their
greatversatility with respect to cargo and cell type, as well
astheir high efficiency in delivering cargo molecules into
cells,makes the MPPs a valuable tool for transfecting siRNAsinto
mammalian cells and even fully grown organisms.Some cargo molecules
may only be coupled at one definedsite, where the chemical
modification does not alterfunctionality. This can be achieved by
making use of singlefunctional groups that naturally occur in the
cargo mole-cule or may be introduced by the utilization of
chemicallymodified building block in the synthesis of the MPPs.This
system can be exploited for efficient delivery of
De Paula et al.
444 RNA, Vol. 13, No. 4
JOBNAME: RNA 13#4 2007 PAGE: 14 OUTPUT: Tuesday March 6 18:39:51
2007
csh/RNA/131733/rna4598
Cold Spring Harbor Laboratory Press on February 12, 2009 -
Published by rnajournal.cshlp.orgDownloaded from
http://rnajournal.cshlp.org/http://www.cshlpress.com
-
siRNAs into cells and into animals. For coupling of thesiRNA,
both the MPP and the siRNA must be modified byfunctional groups
(Hallbrink et al. 2001; Schepers 2005).
Muratovska and Eccles (2004) have shown that siRNAscan be
delivered directly into the cytoplasm when conju-gated with MPPs,
such as penetratin or transportan (Fig.5A). Constitutively
expressed luciferase and GFP geneswere successfully silenced in a
high proportion of cells ofdifferent types with penetratin– or
transportan–siRNAconjugates. For example, in CHO-AA8-Luc Tet Off
cells,the percentage of luciferase activity silenced on each of
the3 d following incubation of the cells with MPP–siRNA con-jugates
was markedly superior to the silencing following
transfection with Lipofectamine 2000 (Fig. 5B). The
salientfeatures of this system are as follows: (1) the
peptidefacilitates transport across the plasma membrane, and
theMPP–siRNA conjugates are freely translocated into thecytoplasm;
(2) the disulfide bond is reduced in the cyto-plasm, releasing the
bioactive siRNA to cause sequence-specific mRNA degradation; and
(3) the uptake of MPP–siRNA conjugates is rapid and occurs directly
through themembrane with no need for classical
receptor-mediateduptake or endo- or pinocytosis.
Simeoni et al. (2003) described a new peptide-based genedelivery
system, MPG, which forms stable noncovalentcomplexes with nucleic
acids. MPG is a bipartite amphi-
pathic peptide derived from both thefusion peptide domain of
HIV-1 gp41protein and the nuclear localizationsignal (NLS) of
Simian virus 40 largeT antigen. The investigators showedthat cell
entry is independent of theendosomal pathway and that the NLS ofMPG
is involved in both electrostaticinteractions with nucleic acid and
nu-clear targeting. MPG/nucleic acid par-ticles interact with the
nuclear importmachinery; however, a mutation thataffects the NLS of
MPG disrupts theseinteractions and prevents nuclear deliv-ery of
nucleic acids. Therefore, thismutated-MPG system was used for
siRNAdelivery into the cytoplasm. Theinvestigators demonstrated
that thismutation yields a variant of MPG, whichis a powerful tool
for siRNA deliveryinto mammalian cells, enabling rapidrelease of
the siRNA into the cyto-plasm and promoting robust down-regulation
of target mRNA.
Complex formation
One of the major challenges of trans-forming siRNAs from
laboratoryreagents to therapeutics is to developan effective drug
delivery system. Pre-vious efforts to deliver siRNAs to theliver
involved high-pressure tail-veininjections, resulting in silencing
of areporter gene in the liver (Morrisseyet al. 2005a). However,
this harsh treat-ment requires rapid injection of solu-tions
two-and-a-half times the totalblood volume of the animal.
Althoughtransfection efficiency of siRNAs intocultured cells is
satisfactory for mostin vitro applications, its therapeutic
FIGURE 5. Gene silencing effect of siRNA conjugated with
membrane permeant peptides(MPPs). (A) Conjugation of siRNA with two
different MPPs: penetratin and transportan.MPPs and siRNAs were
synthesized with a thiol group attached. The reaction between the
thiolgroups on the siRNAs and the MPPs was catalyzed by the oxidant
diamide. (B) Silencing effectof siRNA on luciferase activity.
CHO-AA8-Luc Tet-Off cells stably expressing luciferase
weretransfected with 25 nM siRNA conjugated with MPPs (penetratin
and transportan) orcomplexed with Lipofectamine 2000. Luciferase
activity was measured at days 1, 2, and 3 post-transfection.
Reproduced with permission from Muratovska and Eccles (2004).
siRNA delivery and targeting
www.rnajournal.org 445
JOBNAME: RNA 13#4 2007 PAGE: 15 OUTPUT: Tuesday March 6 18:39:52
2007
csh/RNA/131733/rna4598
Cold Spring Harbor Laboratory Press on February 12, 2009 -
Published by rnajournal.cshlp.orgDownloaded from
http://rnajournal.cshlp.org/http://www.cshlpress.com
-
applications in vivo present an altogether more
dauntingchallenge. Ideally, a delivery mechanism would (1)
becapable of binding siRNAs in a reversible manner to
ensuresubsequent release of the siRNAs in target cells; (2)
protectsiRNAs from nuclease degradation during transit throughthe
circulation; (3) escape from endosomal compartment;(4) be
biocompatible; and (5) avoid rapid clearance by theliver and
kidney.
Cationic lipids
Since the introduction in 1987 of the transfection
reagentLipofectin, an equimolar mixture of the cationic
lipidN-[1-(2,3-dioleyloxy)propyl]-N,N,N-trimethylammoniumchloride
(DOTMA) and the colipid dioleoylphosphatidy-lethanolamine (DOPE),
many cationic liposomes have beentested for in vitro and in vivo
transfection of nucleic acids.The flexibility in the design of
cationic lipid structure andliposome composition combined with the
diversity ofmethods for their preparation and in vivo efficiency
havepromoted the notion that cationic lipids can be efficientlyused
for human gene transfer (Mahato et al. 1997). Thephysicochemical
properties of lipid/nucleic acid complexesare strongly influenced
by the (1) relative proportions ofeach component; (2) structure of
the headgroup of thecationic lipids; (3) cationic lipid: colipid
molar and chargeratio; (4) particle size of lipid/nucleic
complexes; and (5)liposome size (Mahato et al. 1998; Spagnou et al.
2004).Many effective cationic lipids contain protonable poly-amines
linked to dialkyl or cholesterol anchors. Thecationic lipids in the
liposomal formulation form com-plexes with DNA or RNA, leading to
their enhancedcellular association. In general, cationic lipids
with multi-valent headgroups have been shown to be more effective
ingene transfer compared to their monovalent counterparts(Mahato et
al. 1998, 1999; Keller 2005).
siRNAs interact electrostatically with cationic liposomesto form
particles that are able to transfer them into cells.Since the
nature of cationic lipid interactions with DNAand RNA is identical,
it seems obvious to adapt pDNAformulations for siRNA delivery.
However, pDNA andsiRNA are otherwise very different from each other
inmolecular weight and topography with potentially impor-tant
consequences. In contrast to pDNA, siRNA cannotcondense into
particles of nanometric dimensions, beingalready a small
subnanometric nucleic acid (Spagnou et al.2004). Additionally,
electrostatic interactions betweensiRNA and cationic liposomes pose
two potential problems:(1) a relatively uncontrolled interaction
process leading tolipid/siRNA complexes of excessive size and poor
stabilityand (2) incomplete encapsulation of siRNA molecules,which
thereby exposes siRNA to potential enzymatic orphysical degradation
prior to delivery to cells. Such con-siderations should make clear
the fact that pDNA andsiRNA are completely different kinds of
nucleic acids andthat a lipid/siRNA complex should be regarded as
different
from lipid/pDNA complex formulations (Spagnou et al.2004; Keller
2005).
Torchilin and colleagues (Zhang et al. 2006)
successfullydelivered siRNA into lung tumor cells by loading
siRNAinto liposomes bearing arginine octamer (R8), a type ofMPP,
attached to the liposome surface. siRNA-loadedR8 liposomes were
formulated using 1,2-dioleoyl-3-trimethylammonium-propane
(DOTAP)/siRNA with theaddition of the PEG-phosphatidylethanolamine.
The R8liposomes loaded with HDM2–siRNA demonstrated ahigh stability
and protected the incorporated siRNA fromdegradation by blood serum
even after 24 h of incubation.This siRNA formulation showed very
high transfectionefficiency in lung tumor cells. The
R8-liposome-mediatedtransfection takes place even in the presence
of plasmaproteins.
Cationic liposomes are known to interact with serumproteins,
lipoproteins, heparin, and glycosaminoglycansin the extracellular
matrix, leading to aggregation or releaseof nucleic acids from the
complexes even before reachingthe target cells. Cationic lipids can
also activate the com-plement system, leading to their rapid
clearance by macro-phages of the reticuloendothelial systems
(Mahato et al.1997). To circumvent these problems, many
liposomalcarriers have been coated with PEG to minimize
theirinteraction with serum proteins or complement system andto
improve the circulation time. In addition, PEGylationmay help to
stabilize the lipid/nucleic acid complexes,leading to the reduction
in macrophage clearance. Santelet al. (2006) developed a new
liposomal system for siRNAdelivery. This system consists of 50 mol%
of b-L-arginyl-2,3-L-diaminopropionic
acid-N-palmityl-N-oleyl-amidetrihydrochloride (AtuFECT01), 49 mol%
of neutral/helperlipid
1,2-diphytanoyl-sn-glycero-3-phosphoethanolamine(DPhyPE), and 1
mol% PEGylated lipid
N-(carbonyl-methoxypolyethyleneglycol-2000)-1,2-distearoyl-sn-glycero-3-phosphoethanol-amine
sodium salt (DSPE-PEG) (Fig.6A). The new liposomal system was able
to deliver Cy3–siRNA into different cells lines (HeLa, HUVEC) in a
verylow concentration (10 nM). siRNA delivery by this lipo-some was
able to silence protein kinase N3 (PKN3)expression in a
concentration-dependent manner as shownin Figure 6B. The liposomal
system also showed improvedsiRNA cellular uptake and, more
importantly, escape fromthe endocytotic/endosomal pathway into the
cytoplasm(Fig. 6C,D). Data from IL-12 ELISA demonstrated that1 mol%
of DSPE-PEG in the liposome/siRNA complexwas sufficient to reduce
unspecific toxic side effects in vivo.The researchers also injected
1.88 mg/kg of complexedCy3–siRNA intravenously in mice to evaluate
the effect ofthe liposome on siRNA biodistribution. Complexed
siR-NAs were taken up by the vasculature endotheliumin different
organs, mainly liver, heart, lung, spleen,and kidney, showing a
profound delay in the clearancerate.
De Paula et al.
446 RNA, Vol. 13, No. 4
JOBNAME: RNA 13#4 2007 PAGE: 16 OUTPUT: Tuesday March 6 18:40:03
2007
csh/RNA/131733/rna4598
Cold Spring Harbor Laboratory Press on February 12, 2009 -
Published by rnajournal.cshlp.orgDownloaded from
http://rnajournal.cshlp.org/http://www.cshlpress.com
-
Cationic polymers
Due to high cytotoxicity on repeated use and potent
anti-inflammatory activity in vivo (Mahato et al. 1998), interestin
polymeric gene carriers is growing. Like cationic lipids,cationic
polymers spontaneously form complexes withnucleic acids because of
electrostatic interactions betweenpositively charged amine groups
of the polycations and
negatively charged phosphate groupsof the nucleic acids. The
interactionbetween cationic polymers/nucleic acidcomplex and
negatively charged cellmembranes can enhance its uptake bycells and
thus increase transfection effi-ciency (Han et al. 2000). Various
cat-ionic polymers have been studied aspotential vectors for gene
delivery.These include proteins such as histonesand cationized
human serum albumin,polypeptides such as poly-L-lysine(PLL) and
poly-L-ornithine, and poly-amines such as PEI and starburst
poly-amidoamine (pAMAM) dendrimers.Copolymers containing
hydrophilic seg-ments such as PEG and dextran havealso been
synthesized.
pAMAM dendrimers are a class ofhighly branched spherical
polymerswhose surface charge and diameter aredetermined by the
number of syntheticsteps. The three-dimensional sphericalstructure
of dendrimers offers synthesiscontrol in terms of degree and
genera-tion of branching. This control allowsthe production of
polymer particleswith a very low degree of polydispersity,which is
a significant advantage overother polymers such as PLL. The
pri-mary amino groups present on thesurface of pAMAM dendrimers
arepositively charged at physiological pH.These surface amino
groups provideuseful moieties for the functionalmodification of
dendrimers since theyallow covalent coupling reactions undermild
conditions and in a controlledfashion (Mahato et al. 1999).
AlthoughpAMAM dendrimers have been largelyapplied for delivery of
plasmids andODNs (Santhakumaran 2004; Hollins2004; Choi 2005), this
strategy remainsto be explored for siRNA delivery.Generally, it is
expected that the advan-tages observed for ODNs may be adap-ted for
siRNA. However, Kang et al.
(2005) reported that pAMAM dendrimers were not effec-tive for
delivery of siRNA against multidrug resistance 1protein (MDR-1) in
NIH 3T3 MDR cells.
Among the cationic polymers employed for gene deliv-ery, PEI has
been the most widely used polymer for siRNAdelivery. PEIs with
various molecular weights, degrees ofbranching, and other
modifications have been largely usedfor transfection of siRNAs in
different cell lines and live
FIGURE 6. In vitro evaluation of lipid/siRNA complexes. (A)
Chemical structures of lipidsused for preparation of cationic
liposomes and complex formation with siRNAs.
(AtuFECT01)b-L-Arginyl-2,3-L-diaminopropionic
acid-N-palmityl-N-oleyl-amide trihydrochoride; (helperlipid)
1,2-diphytanoyl-sn-glycero-3-phosphoethanolamine; (PEG-lipid)
distearoyl-sn-glycero-3-phospho-ethanolamine sodium salt. (B)
Concentration-dependent inhibition of PKN3 proteinexpression with
lipid/siRNA complex, but not with naked siRNA in HeLa cells as
assessed byimmunoblot. PTEN served as the loading control. (C)
Intracellular distribution of lipid/Cy3–siRNA complexes in HeLa
cells analyzed by confocal microscopy. (D) Merged
high-magnifi-cation pictures of a cytoplasmic area from HUVECs
showing endosomal/lysosomal localizationof siRNAs. Naked siRNAs
remain trapped in the endosomal pathway (upper panel),
whereasformulated siRNAs escape the late endosomal/lysosomal
vesicles (lower panel). Reprinted bypermission from Macmillan
Publishers Ltd.: [Gene Therapy] (Santel et al.) � (2006).
siRNA delivery and targeting
www.rnajournal.org 447
JOBNAME: RNA 13#4 2007 PAGE: 17 OUTPUT: Tuesday March 6 18:40:03
2007
csh/RNA/131733/rna4598
Fig. 6 live 4/C
Cold Spring Harbor Laboratory Press on February 12, 2009 -
Published by rnajournal.cshlp.orgDownloaded from
http://rnajournal.cshlp.org/http://www.cshlpress.com
-
animals (Urban-Klein et al. 2005; Grzelinski et al. 2006).The
high transfection efficiency of PEI can be attributed toits
buffering effect or the ‘‘proton sponge effect’’ due to
itssecondary and tertiary amines. The cytotoxicity and
trans-fection efficiency of PEI are directly proportional to
itsmolecular weight. PEI is often grafted with PEG orcholesterol to
reduce its cytotoxicity (Wang et al. 2002;Patil et al. 2005). Using
atomic force microscopy, Grzelinskiet al. (2006) showed that
complexation of unmodifiedsiRNAs with PEI leads to the formation of
complexes thatcondense and completely cover siRNAs. The
investigatorsshowed that the delivery of siRNAs against growth
factorpleiotrophin complexed with PEI was able to
generateantitumoral effects in an orthotropic mouse
glioblastomamodel with U87 cells growing intracranially.
Urban-Kleinet al. (2005) showed that noncovalent complexation
ofsynthetic siRNAs with low-molecular-weight PEI
efficientlystabilizes siRNAs and delivers siRNAs into cells
wherethey display full bioactivity at nontoxic concentrations.In a
subcutaneous mouse tumor model, intraperitonealadministration of
PEI/siRNA complexes, but not ofnaked siRNAs, was delivered into the
tumors, resulting ina marked reduction of tumor growth through
siRNA-mediated HER-2 receptor down-regulation.
The presence of readily accessible amine groups on thecationic
polymers provides a convenient method forintroducing ligands for
receptor-mediated gene delivery.A common approach is to modify the
amine using theN-hydroxyl succinimide (NHS) moiety, which forms
amideand imide bonds with primary and secondary
amines,respectively. The carboxyl groups of PEG can be coupledto
the amine in the presence of NHS and carbodiimidereagents. In many
cases, a hetero-bifunctional linker is usedfor conjugating the
ligand to