On the art of identifying effective and specific siRNAs
Yi Pei & Thomas Tuschl
Small interfering RNAs (siRNAs) have been widely exploited for quence-specific gene knockdown, predominantly to investigate gene function in cultured vertebrate cells, and also hold promi as therapeutic agents. Becau not all siRNAs that are cognate to a given target mRNA are equally effective, computational tools have been developed bad on experimental data to increa the likelihood of lecting effective siRNAs. Furthermore, becau target-complementary siRNAs can also target other mRNAs containing quence gments that are partially complementary to the siRNA, most computational tools include ways to reduce potential off-target effects in the siRNA lection process. Though the methods facilitate lection of functional siRNAs, they do not yet alleviate the need for experimental validation. This perspective provides a practical guide bad on current wisdom for lecting siRNAs.
The evolutionarily conrved process whereby small double-stranded (ds)RNAs of distinct size and structure quence-specifically suppress the expression of their target genes are referred to as RNA silencing or RNA interference (RNAi)1. Among the repertoire of known small RNAs, siRNAs mediate ge
ne-specific silencing pri-marily via recognizing and inducing degradation of the mRNAs of targeted genes. Conquently, siRNAs have become one of the most valuable reagents to function-ally annotate genomes and posss great potential as therapeutics 2–4.
Shortly after the discovery that siRNA duplexes can specifically silence mammalian genes, it was thought that almost any target-complementary siRNA effectively and specifically silences its cognate target gene 5. In practice, however, different siRNAs often manifest a spectrum of potency, and only a fraction of them are highly effective 6. Small positional shifts along the target mRNA were suffi-cient to alter siRNA function in an apparently unpredict-able manner 6–8. Moreover, siRNAs may nonspecifically target unrelated genes with only partial quence-com-plementarity (off-target effects)9–13. Hence, it is critical to identify effective and specific siRNA quences to per-form reliable gene-knockdown experiments.
Initially, empirical rules had been propod for siRNA lection, some of which were bad on the first identified functional siRNAs 5. The evolving understanding of the RNAi mechanism, together with statistical analys of libraries of siRNAs with experi-mentally determined efficiency, led to computer-bad approaches that incread the likelihood of identifying effective and specific siRNAs 6,14,15. The tools, however, are not perfect. (i) Not every lected siRNA meets the desired thres
holds of potency and specificity, so that experimental proof of downregu-lation of targeted mRNA or protein remains impor-tant, not even considering the evaluation of potential off-target effects. (ii) A substantial fraction of active siRNAs may be dismisd becau the weighing of fac-tors influencing activity is complex and partly unde-fined 6,9,16. Not surprisingly, experimental approaches to generate and identify effective siRNAs have been developed to complement rule-driven siRNA lec-tion strategies 16–18.
小学生近视率There are many excellent recent reviews covering the mechanism of RNAi 19–22. Elements of this mechanism that are important for the lection of siRNA are sum-marized in Box 1.
Howard Hughes Medical Institute, Laboratory of RNA Molecular Biology, The Rockefeller University, 1230 York Avenue, Box 186, New York, New York 10021, USA. Correspondence should be addresd to T.T. (ttuschl@rockefeller.edu).
PUBLISHED ONLINE 23 AUGUST 2006; DOI:10.1038/NMETH911
670 | VOL.3 NO.9 | SEPTEMBER 2006 | NATURE METHODS
PERSPECTIVE
©2006 N a t u r e P u b l i s h i n g G r o u p h t t p ://w w w .n a t u r e .c o m /n a t u r e m e t h o d s
Here we provide a practical guide and an overview of the theo-retical basis for identification and lection of effective and specific siRNAs.
A GUIDE FOR siRNA SELECTION Target mRNA analysis
The lection of siRNAs against a gene of interest starts with an annotated target mRNA quence, including its 5′ and 3′ untrans-lated regions (UTRs), splice, polymorphic and allelic variants. Becau the coding quence is the most reliable mRNA quence information available, it is commonly targeted. The UTRs are gen-erally less well characterized, but can also be targeted with similar gene-knockdown efficiency 8,23,24. Though it has often been recom-mended to avoid targeting quences that contain known binding sites for mRNA-binding proteins, such as the exon-exon junction complex, there is no detailed experimental study available to asss the importance of this guideline.
For practical reasons, lection of siRNAs is often carried out with additional constraints, for example identifying siRNAs that target (i) orthologs in more than one species or (ii) all possible splice variants of a gene.
Databa arch for published and validated siRNAs
Several databas archive experimentally tested siRNA quences from the literature 25–27. Additionally, validated siRNAs can be acquired from commercial resources (for example, the Silencer vali-dated siRNAs from Ambion and HP validated siRNAs from Qiagen). Some vendors, su
ch as Ambion, Qiagen and Dharmacon also pro-vide predesigned siRNAs or custom siRNA design rvice. Though prevalidated reagents provide an excellent starting point, the ur
still has to examine whether the siRNAs are potent and specific to meet the needs 28.
If there are no matches to the target gene of interest in any of the databas or in the literature, it is advisable to lect 3–5 candidate siRNAs using available guidelines and tools, and subquently to validate the reagents.
Selected algorithms and siRNA quence lection tools
Several siRNA quence lection algorithms have been developed in recent years that rely on intrinsic quence and stability features of functional siRNAs 6,14,15,23,29–35. A smaller number of algorithms consider the condary structure and accessibility of the targeted mRNA 36–38. The approaches underlying the algorithms range from empirical obrvations to sophisticated machine learning. After the siRNA quence lection from the target mRNA quence, each candidate siRNA is examined for similarity to all other mRNA transcripts that might unintentionally be targeted at a genome-wide level. Most of the siRNA lection algorithms have been combined with a variant of such programs, and the more ur-friendly tools are listed in Table 1 (for a more complete list, e
ref. 28). The lected siRNAs can be custom synthesized from four siRNA-licend reagent suppliers: Ambion, Dharmacon, Qiagen and Sigma Proligo.Prevalidation of siRNAs
Becau the determination of the preci level of gene knock-down for each siRNA is a demanding process, and the assays need to be adapted for newly targeted genes, reporter-bad assays have been developed to accelerate the identification of potent siRNAs among various synthesized siRNAs. In the systems, plasmids, which carry the target quence fud to a reporter
NATURE METHODS | VOL.3 NO.9 | SEPTEMBER 2006 | 671
PERSPECTIVE
四年级作文书
©2006 N a t u r e P u b l i s h i n g G r o u p h t t p ://w w w .n a t u r e .c o m /n a t u r e m e t h o d s
gene and a control gene for normalization, are cointroduced into cells together with the target-specific siRNAs 14,39,40. The dual-lucifera–bad siCHECK system from Promega is widely ud and provides a ranking for siRNA activity within 24 h. The reporter-bad activity generally correlates well with the efficacy of depleting the endogenous target (our unpublished obrva-tions). The preval
idated siRNAs can then be ud to validate the depletion of the endogenous target mRNA, which is discusd in detail in an accompanying review 41.
CONSIDERATIONS FOR SELECTING EFFECTIVE AND SPECIFIC siRNAs
Sequence asymmetry of siRNA duplexes
It has been demonstrated that structurally symmetric (duplexes with symmetric 2-nucleotide (nt) 3′ overhangs) but primary quence–asymmetric (different nucleotides on each end) siRNAs, from which the target-mRNA complementary guide strand has greater propensity to be asmbled into the RNA-induced silenc-ing complex (RISC) than the pasnger strand, show improved effi-cacy and specificity 42,43 (Fig. 1). The same finding emerges from quence analysis of miRNA precursors and largely explains the asymmetric accumulation of the majority of miRNAs 42. The asym-metry is determined by the different quence composition, and the conquent differences in thermodynamic stability and molecular dynamic behavior of the two ba-paired ends of an siRNA duplex: the strand with the less stable 5′ end, owing to either weaker ba-pairing or introduction of mismatches, is favorably or exclusively loaded into RISC 44. The asymmetry rule has been implemented in many siRNA design algorithms by computing either the A ⋅U ba pair content or loc
al free energy at both ends of an siRNA, followed by lection of the duplexes with less stable, (A+U)-enriched 5′ end on the guide strand 20.
一桥大学
Becau the majority of miRNAs start with a 5′ uridine, it is also conceivable that 5′ uridine–specific interaction contributes to more effective RISC asmbly and function beyond the thermodynamic contributions discusd here. Furthermore, miRNA duplexes con-tain an average of six non–Watson-Crick ba pairs distributed over the entire miRNA length, who contribution to RISC asmbly and asymmetry has not been evaluated.
siRNA duplex stability
Most analyzed functional siRNAs had a low-to-medium G+C content ranging between 30% and 52% (refs. 6,31). It has been argued that too low G+C content may destabilize siRNA duplexes and reduce the affinity for target mRNA binding, whereas too high G+C content may impede RISC loading and/or cleavage-product relea. Additionally, surveys of functional siRNAs revealed that sta-ble duplexes devoid of internal repeats or palindromes, which may form intrastrand condary structures, were better silencers 6,31,45. An equally likely explanation is that the condary structure of the target mRNA, which mirrors the predicted guide siRNA condary structure, interferes with targeting.
Although the overall duplex stability is important, the center of the duplex (positions 9–14 on the guide strand) appears to prefer-entially have low internal stability 31,42,46. It has recently been noticed that miRNAs and siRNAs asmble into RISC by different mecha-nisms; siRNAs require cleavage of the pasnger strand for effective RISC asmbly, whereas a mismatched RNa III–procesd miRNA duplex does not require pasnger strand cleavage 47. It is conceiv-able that the central-duplex instability may influence how effectively and to what ratios the RISC complexes with different core compo-nents are loaded. Alteration of the structure and stability of siRNA duplexes can also be controlled by incorporation of chemically modified nucleotide analogs. The effects of modifications, however,
siRNA design software
www.cs.hku.hk/~sirna
Candidate siRNAs propod by various previously developed quence lection tools are classified bad on target accessibility.
38
672 | VOL.3 NO.9 | SEPTEMBER 2006 | NATURE METHODS
PERSPECTIVE
2006 N a t u r e P u b l i s h i n g G r o u p h t t p ://w w w .n a t u r e .c o m /n a t u r e m e t h o d s
three quence features correlate with the rule of thermodynamic asymmetry, and the preferred nucleotides on indicated positions may contribute to the bias for lection of
antin strand. The A or U at position 10 is at the cleavage site and may promote catalytic RISC-mediated pasnger strand and substrate cleavage. Other quence determinants may be involved in steps along the RNAi pathway, such as RISC loading 54.
In addition to the positional nucleotide preference, certain motifs are commonly avoided in chemically synthesized siRNA duplexes that could affect the synthesis yield, purification or the annealing of siRNA strands. Extended runs of altering G ⋅C pairs (more than 7)32 or runs of more than three guanines are sometimes avoided.
Moreover, in light of the reports that certain siRNAs can activate immune respon in a cell- and quence-dependent manner 55–57, it is a prudent measure to filter out siRNA quences containing putative immunostimulatory motifs in either strand to minimize toxicities and nonspecific silencing effects, especially when siRNAs are lected for in vivo and therapeutic u. Alternatively, immuno-Figure 1 | A scheme for siRNA-mediated gene silencing. The primary quence asymmetry of duplex determines which strand is preferentially The cleavage site is indicated by scissors in the t
arget mRNA. Target recognition and off-target activity can occur in two modes, the catalytic siRNA-guided cleavage reaction requiring extensive
complementarity in the region surrounding the cleavage site (blue) and the miRNA-like destabilization of mRNAs requiring pairing of the siRNA 5′ end (green).
Guide strand (preferentially enters RISC)
Target recognition
RISC proteins
Pasnger strand destroyed
Pasnger strand (preferentially destroyed)3′OH 3′′3′ OH
P 5′Guide strand
3′′Cleavage-bad recognition
Guide strand
OH
OH 3P 5′
Guide strand
OH
miRNA-like recognition
P 5′
212019181716151413121110987654321
212019181716151413121110987654321
NATURE METHODS | VOL.3 NO.9 | SEPTEMBER 2006 | 673
PERSPECTIVE
o m /n a t u r e m e t h o d s
饮料简笔画
interaction with the RNAi machinery may contribute to asymmet-ric RISC asmbly. However, the comparison with conventional siRNAs is again complicated by the differences in length of the guide strand and the differences in strength of 5′ terminal guide strand ba-pairing when the paired region is longer than in conventional siRNA duplexes.
Specificity
Each strand of an siRNA duplex, once asmbled into RISC, can guide recognition of fully and partially complementary target mRNAs, referred to as on- and off-targets, respectively. Though quence asymmetry can be ud to bias pasnger strand exclu-sion, chemical methods of preventing pasnger-strand u have also been introduced (for example, Dharmacon’s ON-TARGET siRNA). For the purpo of this discussion, we will distinguish off-targets into two class (Fig. 2): (i) tho that share contiguous and centrally located quence complementarity over more than half of the siRNA quence somewhere within the mRNA quence 71, and (ii) tho that show solely 6 or 7 nucleotides of perfect match preferentially in the 3′ UTRs with positions 2–7 or 2–8 (ed region) of the guide siRNA 9,11,12. The latter interaction is the major driving force behind endogenous miRNA–target mRNA recognition 20,72. Although the off-targets of the latter class are predominant, their actual number identified in microarray analys was significantly smaller than the
number of computationally predicted targets with quence com-plementary to the ed region of the guide strand, suggesting that additional specificity determinants remain to be identified 9,12.
Furthermore, structural and biochemical studies showed that guide-strand position 1 and the nucleotides at the 3′ overhang (posi-tions 20 and 21) have little, if any, contribution to the specificity of target recognition, and that mismatches near the 5′ and 3′ ends can be tolerated for RISC-guided cleavage if the remaining pairing to the target was unperturbed 73,74.
To enforce specificity, the current strategy is to lect siRNAs in which the strand(s) entering RISC has some mismatches to all unde-sired target mRNAs, especially their 3′ UTRs. Typically at least three mismatches are recommended between positions 2 and 19 and the mismatches near the 5′ end and in the center of the examined strand should be assigned higher significance 11,71,75,76. In addition to the position, the identity of the quence mismatches also influence specificity to a certain extent 75,77,78.
Prently most tools u blastn or Smith-Waterman algo-rithm to remove potential off-targeting siRNAs during the siRNA quence lection process 79. In addition to the arch method, the quality and completeness of the lected genome-wide mRNA quence databa is also of high im
portance 79–81. The current tools, however, cannot eliminate all the potential off-targets, espe-cially tho that contain the short quence gments comple-mentary to the ed region of the guide strand, and likely discard many potentially functional siRNAs 9. While improved algorithms are awaited, position-specific chemical modification of the ed-quence of the guide siRNA can be ud to reduce off-tar-get effects 82. It is therefore important to experimentally control off-target effects or to dilute the off-target effects beyond the detection limit by codelivering veral different target-specific siRNAs 10,41.Allele-specific gene silencing
To take advantage of the quence specificity of RNAi, a prerequi-site to achieve allele-specific gene silencing is to identify the most
significant difference between two alleles, which may be as little as a single nucleotide change stemming from mutation or poly-morphism 83. Placing this quence discrepancy in the center of an siRNA, at or near the RISC cleavage site ems to be best for discrim-inating between alleles 8,76–78,83. In some cas, introducing an addi-tional mismatch at other positions in the siRNA may improve the allele specificity, as long as the mismatch is tolerated for cleavage 83. A limitation of this approach is that the choice of siRNA is restricted, and the siRNA may not be sufficiently effective. In this respect, it is interesting to note that the introduction of a G ⋅U wobble mismatch in the 5′ terminal
siRNA-mRNA interaction incread the potency of some siRNAs 75. The efficacy of silencing may also be incread by destabilizing ba-pairing at the 5′ end of the guide strand following the asymmetry rule 78.
Alternatively, both alleles can be nondiscriminately silenced by an effective siRNA distant from the polymorphic site, accompanied by ectopic expression of the desired quence-modified allele refractory to the siRNA 84. Vectors that simultaneously express transgene and short hairpin RNAs have been developed 85.
OUTLOOK
In summary, guidelines are available that increa the likelihood of identifying effective and specific siRNAs at the expen of elimi-nating many potentially functional and specific siRNAs. The guidelines assist in reducing the numbers of siRNAs that need to be experimentally validated to identify potent and specific siRNAs for a given target gene. As reagent manufacturers have recognized the need for constant validation of siRNA knockdown experiments and developed promising lines of reagents, effective siRNAs can be identified at a rapid pace and will soon lead to the ultimate goal of production of validated genome-wide siRNA libraries needed for high-throughput or individual gene silencing experiments.
ACKNOWLEDGMENTS
We apologize to authors who works are not cited owing to space limitations. We thank C. Echeverri at Cenix Bioscience for valuable discussion. We also thank M. Landthaler, P. Landgraf, J. Brennecke and C. Rogler for critical reading of the manuscript. Y.P. is supported by the Ruth L. Kirschstein Fellowship from the US National Institutes of Health–National Institute of General Medical SciencesPETING INTERESTS STATEMENT
The authors declare competing financial interests (e the Nature Methods website for details).
Published online at /naturemethods/
Reprints and permissions information is available online at /reprintsandpermissions/
1. Meister, G. & Tuschl, T. Mechanisms of gene silencing by double-stranded
RNA. Nature 431, 343–349 (2004).
2. Dortt, Y. & Tuschl, T. siRNAs: applications in functional genomics and
potential as therapeutics. Nat. Rev. Drug Discov. 3, 318–329 (2004).
泽泻功效与作用3. Echeverri, C.J. & Perrimon, N. High-throughput RNAi screening in cultured
cells: a ur’s guide. Nat. Rev. Genet. 7, 373–384 (2006).
4. Dykxhoorn, D.M., Pallir, D. & Lieberman, J. The silent treatment: siRNAs as
small molecule drugs. Gene Ther. 13, 541–552 (2006).
5. Elbashir, S.M., Harborth, J., Weber, K. & Tuschl, T. Analysis of gene function
in somatic mammalian cells using small interfering RNAs. Methods 26, 199–213 (2002).6. Reynolds, A. et al. Rational siRNA design for RNA interference. Nat.
Biotechnol. 22, 326–330 (2004).
固态硬盘格式化7. Holen, T., Amarzguioui, M., Wiiger, M.T., Babaie, E. & Prydz, H. Positional
effects of short interfering RNAs targeting the human coagulation trigger Tissue Factor. Nucleic Acids Res. 30, 1757–1766 (2002).8. Harborth, J. et al. Sequence, chemical, and structural variation
of small
interfering RNAs and short hairpin RNAs and the effect on mammalian gene
674 | VOL.3 NO.9 | SEPTEMBER 2006 | NATURE METHODS
PERSPECTIVE
©2006 N a t u r e P u b l i s h i n g G r o u p h t t p ://w w w .n a t u r e .c o m /n a t u r e m e t h o d s
silencing. Antin Nucleic Acid Drug Dev. 13, 83–105 (2003).9. Birmingham, A. et al. 3′ UTR ed matches, but not overall identity, are
associated with RNAi off-targets. Nat. Methods 3, 199–204 (2006).
10. Jackson, A.L. & Linsley, P.S. Noi amidst the silence: off-target effects of
siRNAs? Trends Genet. 20, 521–524 (2004).11. Lin, X. et al. siRNA-mediated off-target gene silencing triggered by a 7 nt
complementation. Nucleic Acids Res. 33, 4527–4535 (2005).12. Jackson, A.L. et al. Widespread siRNA “off-target” transcript silencing
mediated by ed region quence complementarity. RNA 12, 1179–1187 (2006).13. Fedorov, Y. et al. Off-target effects by siRNA can induce toxic phenotype. RNA
12, 1188–1196 (2006).14. Huesken, D. et al. Design of a genome-wide siRNA library using an artificial
neural network. Nat. Biotechnol. 23, 995–1001 (2005).
15. Saetrom, P. & Snove, O., Jr. A comparison of siRNA efficacy predictors.
Biochem. Biophys. Res. Commun. 321, 247–253 (2004).16. Zhao, H.F. et al. High-throughput screening of effective siRNAs from RNAi
libraries delivered via bacterial invasion. Nat. Methods 2, 967–973 (2005).17. Ito, M., Kawano, K., Miyagishi, M. & Taira, K. Genome-wide application of
RNAi to the discovery of potential drug targets. FEBS Lett. 579, 5988–5995 (2005).
18. Kasim, V., Taira, K. & Miyagishi, M. Screening of siRNA target quences by
using fragmentized DNA. J. Gene Med. 8, 782–791 (2006).
19. Sontheimer, E.J. Asmbly and function of RNA silencing complexes. Nat.
Rev. Mol. Cell Biol. 6, 127–138 (2005).
20. Tomari, Y. & Zamore, P.D. Perspective: machines for RNAi. Genes Dev. 19,
517–529 (2005).
21. Filipowicz, W., Jaskiewicz, L., Kolb, F.A. & Pillai, R.S. Post-transcriptional gene
silencing by siRNAs and miRNAs. Curr. Opin. Struct. Biol. 15, 331–341 (2005).22. Valencia-Sanchez, M.A., Liu, J., Hannon, G.J. & Parker, R. Control of
translation and mRNA degradation by miRNAs and siRNAs. Genes Dev. 20, 515–524 (2006).23. Hsieh, A.C. et al. A library of siRNA duplexes targeting the phosphoinositide
3-kina pathway: determinants of gene silencing for u in cell-bad screens. Nucleic Acids Res.
32, 893–901 (2004).24. McManus, M.T. et al. Small interfering RNA-mediated gene silencing in
T lymphocytes. J. Immunol. 169, 5754–5760 (2002).25. Ren, Y. et al. siRecords: an extensive databa of mammalian siRNAs with
efficacy ratings. Bioinformatics 22, 1027–1028 (2006).
26. Chalk, A.M., Warfinge, R.E., Georgii-Hemming, P. & Sonnhammer, E.L.
siRNAdb: a databa of siRNA quences. Nucleic Acids Res. 33, D131–D134 (2005).27. Truss, M. et al. HuSiDa–the human siRNA databa: an open-access databa
for published functional siRNA quences and technical details of efficient transfer into recipient cells. Nucleic Acids Res. 33, D108–D111 (2005).
28. Smith, C. Sharpening the tools of RNA interference. Nat. Methods 3, 475–486
(2006).
29. Shabalina, S.A., Spiridonov, A.N. & Ogurtsov, A.Y. Computational models
with thermodynamic and composition features improve siRNA design. BMC Bioinformatics 7, 65 (2006).
30. Yuan, B., Latek, R., Hossbach, M., Tuschl, T. & Lewitter, F. siRNA Selection
Server: an automated siRNA oligonucleotide prediction rver. Nucleic Acids Res. 32, W130–W134 (2004).
31. Chalk, A.M., Wahlestedt, C. & Sonnhammer, E.L. Improved and automated
prediction of effective siRNA. Biochem. Biophys. Res. Commun. 319, 264–274 (2004).32. Ui-Tei, K . et al. Guidelines for the lection of highly effective siRNA
quences for mammalian and chick RNA interference. Nucleic Acids Res. 32, 936–948 (2004).
33. Takasaki, S., Kotani, S. & Konagaya, A. An effective method for lecting
siRNA target quences in mammalian cells. Cell Cycle 3, 790–795 (2004).34. Jagla, B. et al. Sequence characteristics of functional siRNAs. RNA 11, 864–
872 (2005).
春节的抒情散文
35. Amarzguioui, M. & Prydz, H. An algorithm for lection of functional siRNA
香蕉粥的做法
quences. Biochem. Biophys. Res. Commun. 316, 1050–1058 (2004).
36. Ding, Y., Chan, C.Y. & Lawrence, C.E. Sfold web rver for statistical folding
and rational design of nucleic acids. Nucleic Acids Res. 32, W135–W141 (2004).
37. Luo, K.Q. & Chang, D.C. The gene-silencing efficiency of siRNA is strongly
dependent on the local structure of mRNA at the targeted region. Biochem. Biophys. Res. Commun. 318, 303–310 (2004).38. Yiu, S.M. et al. Filtering of ineffective siRNAs and improved siRNA design
tool. Bioinformatics 21, 144–151 (2005).
39. Kumar, R., Conklin, D.S. & Mittal, V. High-throughput lection of effective
RNAi probes for gene silencing. Genome Res. 13, 2333–2340 (2003).
40. Malik, I., Garrido, M., Bahr, M., Kugler, S. & Michel, U. Comparison of test
systems for RNAinterference. Biochem. Biophys. Res. Commun. 341, 245–253 (2006).
41. Cullen, B.R. Enhancing and confirming the specificity of RNAi experiments.
Nat. Methods 3, 677–681 (2006).
42. Khvorova, A., Reynolds, A. & Jayana, S.D. Functional siRNAs and miRNAs
exhibit strand bias. Cell 115, 209–216 (2003).43. Schwarz, D.S. et al. Asymmetry in the asmbly of the RNAi enzyme complex.
Cell 115, 199–208 (2003).
44. Hutvagner, G. Small RNA asymmetry in RNAi: function in RISC asmbly and
gene regulation. FEBS Lett. 579, 5850–5857 (2005).45. Patzel, V. et al. Design of siRNAs producing unstructured guide-RNAs results in
improved RNA interference efficiency. Nat. Biotechnol. 23, 1440–1444 (2005).46. Taxman, D.J. et al. Criteria for effective design, construction, and gene
knockdown by shRNA vectors. BMC Biotechnol. 6, 7 (2006).
47. Matranga, C., Tomari, Y., Shin, C., Bartel, D.P. & Zamore, P.D. Pasnger-strand cleavage facilitates asmbly of siRNA into Ago2-containing RNAi enzyme complexes. Cell 123, 607–620 (2005).
48. Manoharan, M. RNA interference and chemically modified small interfering
RNAs. Curr. Opin. Chem. Biol. 8, 570–579 (2004).
49. Hossbach, M., Gruber, J., Osborn, M., Weber, K. & Tuschl, T. Gene silencing
with siRNA duplexes compod of target-mRNA-complementary and partially palindromic or partially complementary single-stranded siRNAs. RNA Biology 3 (2006).
50. Cullen, B.R. Induction of stable RNA interference in mammalian cells. Gene
Ther. 13, 503–508 (2006).
51. Heale, B.S., Soifer, H.S., Bowers, C. & Rossi, J.J. siRNA target site condary
structure predictions using local stable substructures. Nucleic Acids Res. 33, e30 (2005).
52. Schubert, S., Grunweller, A., Erdmann, V.A. & Kurreck, J. Local RNA target
structure influences siRNA efficacy: systematic analysis of intentionally designed binding regions. J. Mol. Biol. 348, 883–893 (2005).53. Overhoff, M. et al. Local RNA target structure influences siRNA efficacy: a
systematic global analysis. J. Mol. Biol. 348, 871–881 (2005).
54. Mittal, V. Improving the efficiency of RNA interference in mammals. Nat. Rev.
Genet. 5, 355–365 (2004).55. Judge, A.D. et al. Sequence-dependent stimulation of the mammalian innate
immune respon by synthetic siRNA. Nat. Biotechnol. 23, 457–462 (2005).56. Hornung, V. et al. Sequence-specific potent induction of IFN-α by short
interfering RNA in plasmacytoid dendritic cells through TLR7. Nat. Med. 11, 263–270 (2005).
57. Marques, J.T. & Williams, B.R. Activation of the mammalian immune system
by siRNAs. Nat. Biotechnol. 23, 1399–1405 (2005).
58. Judge, A.D., Bola, G., Lee, A.C. & MacLachlan, I. Design of noninflammatory
synthetic siRNA mediating potent gene silencing in vivo . Mol. Ther. 13, 494–505 (2006).
59. Snøve, O. Jr. & Rossi, J.J. Expressing short hairpin RNAs in vivo. Nat. Methods
3, 689–695 (2006).60. Elbashir, S.M. et al. Duplexes of 21-nucleotide RNAs mediate RNA
interference in cultured mammalian cells. Nature 411, 494–498 (2001).
61. Elbashir, S.M., Lendeckel, W. & Tuschl, T. RNA interference is mediated by 21-
and 22-nucleotide RNAs. Genes Dev. 15, 188–200 (2001).
62. Caplen, N.J., Parrish, S., Imani, F., Fire, A. & Morgan, R.A. Specific inhibition
of gene expression by small double-stranded RNAs in invertebrate and vertebrate systems. Proc. Natl. Acad. Sci. USA 98, 9742–9747 (2001).
63. Paddison, P.J., Caudy, A.A., Bernstein, E., Hannon, G.J. & Conklin, D.S. Short
hairpin RNAs (shRNAs) induce quence-specific silencing in mammalian cells. Genes Dev. 16, 948–958 (2002).64. K im, D.H. et al. Synthetic dsRNA Dicer substrates enhance RNAi potency and
efficacy. Nat. Biotechnol. 23, 222–226 (2005).
65. Amarzguioui, M., Rossi, J.J. & Kim, D. Approaches for chemically synthesized
siRNA and vector-mediated RNAi. FEBS Lett. 579, 5974–5981 (2005).66. Ro, S.D. et al. Functional polarity is introduced by Dicer processing of short
substrate RNAs. Nucleic Acids Res. 33, 4140–4156 (2005).67. Reynolds, A. et al. Induction of the interferon respon by siRNA is cell type-
and duplex length-dependent. RNA 12, 988–993 (2006).68. Marques, J.T. et al. A structural basis for discriminating between lf and
nonlf double-stranded RNAs in mammalian cells. Nat. Biotechnol. 24, 559–565 (2006).69. Vermeulen, A. et al. The contributions of dsRNA structure to Dicer specificity
and efficiency. RNA 11, 674–682 (2005).70. Soutschek, J. et al. Therapeutic silencing of an endogenous gene by systemic
administration of modified siRNAs. Nature 432, 173–178 (2004).71. Jackson, A.L. et al. Expression profiling reveals off-target gene regulation by
RNAi. Nat. Biotechnol. 21, 635–637 (2003).
NATURE METHODS | VOL.3 NO.9 | SEPTEMBER 2006 | 675
PERSPECTIVE
©2006 N a t u r e P u b l i s h i n g G r o u p h t t p ://w w w .n a t u r e .c o m /n a t u r e m e t h o d s
