Introduction

RNA interference (RNAi) is an evolutionarily conserved mechanism that permits the selective post-transcriptional downregulation of target genes in the cells of metazoan eukaryotes, including humans.¹ As such, RNAi has enormous potential as not only an invaluable tool in biological research and drug development but also as a possible approach to the in vivo inactivation of gene products linked to human disease progression and pathology.

RNAi is mediated by 22 bp double-stranded RNA (dsRNA) molecules, bearing 2 nucleotide (nt) 3' overhangs, termed short interfering RNA (siRNA) duplexes.¹ These siRNA duplexes can be generated in situ by cleavage of long dsRNAs, or artificial short hairpin RNAs (shRNAs), by the cytoplasmic RNase III enzyme Dicer^{2, 3, 4, 5} or can be directly introduced into cells by transfection.⁶ One strand of the siRNA duplex, termed the guide strand, is then incorporated into the RNA-induced silencing complex (RISC),^{7, 8, 9} while the second, passenger strand, is released and degraded. Once incorporated into RISC, the siRNA guides this protein complex to mRNAs bearing highly complementary target sites, which can then be cleaved by the action of the RISC-component Argonaut 2, leading to their degradation.^{8, 9, 10, 11}

siRNAs appear functionally identical, at least in mammalian cells, to endogenous 22 nt single-stranded noncoding RNAs termed micro RNAs (miRNAs).^{12, 13, 14} miRNAs are encoded within the host cell genome and over 200 are now known to exist in humans and mice.^{1, 15} Initially, miRNAs are transcribed by RNA polymerase II (polII) as part of a long primary miRNA transcript, which may be several thousand nucleotides in length, that is capped and polyadenylated^{16, 17} (Figure 1). The mature miRNA forms part of one arm of an 80 nt imperfect RNA stem–loop structure located within the primary miRNA. This stem–loop structure is recognized by the nuclear microprocessor complex, consisting minimally of the RNase III enzyme Drosha and a cofactor termed DGCR8, which cleaves the stem 22 bp away from the terminal loop to release an 60 nt RNA hairpin intermediate called a pre-miRNA.^{18, 19, 20, 21} Pre-miRNAs are closely similar in structure to shRNAs, both of which are short RNA hairpins bearing an 2 nt 3' overhang that, in the case of the pre-miRNA, results from Drosha cleavage of the primary miRNA precursor (Figure 1).¹⁸ Artificial shRNAs, in contrast, are normally directly transcribed by RNA polymerase III (polIII) and the 3' overhang results from polIII transcription termination after the second T in a stretch of 'T' residues in the shRNA expression cassette (see below).^{2, 3}

Figure 1.

Short hairpin RNAs utilize the micro RNA (miRNA) processing machinery. miRNAs are initially transcribed by polII as long pre-miRNAs and then processed by the nuclear RNase III enzyme Drosha to give the 60 nt pre-miRNA hairpin intermediate. In contrast, shRNAs, which are functionally and structurally equivalent to pre-miRNAs, are directly transcribed by polIII. The subsequent nuclear export and cytoplasmic processing steps are indistinguishable for pre-miRNAs and shRNAs in human cells. See text for detailed discussion.

Full figure and legend (66K)

Both pre-miRNAs and shRNAs are then bound by a complex consisting of the nuclear export factor Exportin 5 (Exp5) and the GTP-bound form of its cofactor Ran.^{22, 23} The Exp5/Ran–GTP complex, which recognizes small RNA stem–loops bearing short, 2 nt 3' overhangs, transports pre-miRNAs and shRNAs to the cytoplasm where hydrolysis of the Ran-bound GTP moiety induces release of the RNA cargo. The pre-miRNAs and shRNAs now recruit the cytoplasmic RNase III enzyme Dicer, which recognizes and binds short (2 nt) 3' overhangs present at the termini of dsRNA molecules.^{4, 5} Dicer then cleaves 21–22 nt away from the base of the RNA hairpin, on both RNA strands, to remove the terminal loop and generate the siRNA or miRNA duplex intermediate²⁴ (Figure 1).

Recent research has clarified the mechanism that governs which strand of the duplex intermediate is then incorporated into RISC.^{25, 26} Specifically, RISC incorporates the RNA strand whose 5'-end is less tightly base-paired. Thus, if one strand has stable G:C base pairs at its 5'-end, while the second has, for example, less stable A:U or G:U base pairs, or even a mismatch, then the latter strand will be incorporated into RISC, while the former is degraded. However, if the stability of base-pairing at both ends of the duplex is similar, then selection of the incorporated RNA strand becomes stochastic and fairly equivalent levels of each strand may enter RISC. For most, but not all, natural miRNAs only one strand of the duplex intermediate is generally detected in RISC. In the case of artificial shRNAs, or siRNA duplexes, ensuring that the correct (i.e. antisense) strand enters RISC is a major design criterion, as described in more detail below.

shRNA design and target selection

Optimization of the design of an shRNA to increase the likelihood that it will effectively degrade its target mRNA can be effectively subdivided into three steps. The first is ensuring efficient loading of RISC by the antisense siRNA strand, the second is ensuring that the loaded siRNA has the potential to induce efficient target mRNA degradation, while the third is to select an mRNA target that is available for RISC binding. In addition, it is important to avoid 'off target' effects of the shRNA chosen by minimizing the potential of the shRNA hairpin to activate host cell defense pathways, such as the interferon response, that can be induced by dsRNAs²⁷ and by avoiding shRNAs that bear significant homology to other, irrelevant host mRNAs. Moreover, shRNAs should not contain extraneous inverted repeats that would interfere with the appropriate folding of the shRNA hairpin.

As shRNAs are effectively artificial orthologs of endogenous pre-miRNAs (Figure 1), which are presumably evolutionarily selected for efficiency and selectivity, it is worthwhile to consider which attributes of pre-miRNAs might be usefully incorporated into shRNAs.^{28, 29} These considerations, together with extensive analyses of artificial shRNA and siRNA function reported by several groups, suggest a number of design criteria, as listed below. Throughout, I will use an shRNA design in which the 5' arm of the shRNA hairpin forms the sense (passenger) strand, while the 3' arm forms the antisense (guide) strand of the siRNA duplex intermediate.

shRNA transcription and processing

Transcription of shRNAs is normally performed using an RNA polIII promoter such as the H1 or U6 promoter.^{2, 3} Unlike many polIII-dependent promoters, H1 and U6 are compact promoters located 5' to the transcribed sequence. However, the U6 promoter strongly favors a 'G' residue at the first position of the transcribed sequence, while the H1 promoter has a weak preference for an 'A' residue. PolIII-mediated transcription terminates after the second or, less commonly, third residue of a 'TTTTT' stretch, and the 3'-terminal 'UU' of polIII-derived shRNAs is used to form the 3' 2 nt overhang that is characteristic of pre-miRNAs (Figures 1 and 2). Both the preference for a specific residue at the +1 position in the transcript, and the requirement for 2 'U' residues at the 3'-end of artificial shRNA transcripts, influence both shRNA expression vector design and mRNA target site selection (Figure 2).

Figure 2.

Incorporating pre-miRNA structural elements into shRNA designs facilitates their function. Shown is the predicted structure of the human pre-miR-1 precursor with the mature miRNA/siRNA shown in red and the passenger strand in blue. Dicer cleavage sites are indicated by arrows. The proposed shRNA precursor shown incorporates aspects of the pre-miR-1 structure as well as other advantageous design features, as discussed in detail in the text. Bases labeled as N are determined by the mRNA target site chosen, which is schematically represented at the bottom.

Full figure and legend (75K)

As noted above, artificial shRNAs are transcribed in the nucleus and processed in the cytoplasm by Dicer (Figure 1). Nuclear export of shRNAs and pre-miRNAs by Exp5 requires efficient recognition by the Exp5/Ran:GTP heterodimer. Fortunately, this heterodimer requires only an RNA stem of 16 bp, a short 3' overhang and a terminal loop of >6 nt.^{22, 30} Similarly, efficient cleavage by Dicer appears to only require an RNA stem–loop of sufficient size (i.e. a stem of 19 bp)^{2, 3} and a short 3' overhang.⁵ These characteristics are therefore very easily incorporated into the shRNA design (Figure 2) and are a defining characteristic of pre-miRNAs.

Strand selection

As noted above, RISC selects the strand of the siRNA duplex whose 5'-end is less tightly base-paired.^{25, 26} In the miRNA example shown in Figure 2, the 3' arm of the pre-miR-1 precursor is incorporated into RISC as mature miR-1, while the 5' arm is excluded. Inspection shows that the 5'-end of the excluded, 5' strand features two terminal G:C base-pairs and is tightly base-paired. In contrast, the 5'-end of the included, 3' strand consists of a single, less stable A:U base-pair next to an A to G mismatch. Therefore, the artificial shRNA you design could contain two G:C base-pairs at the 5'-end of the 5' strand and a weak base-pair (A:U or G:U) next to a mismatch at the 5'-end of the 3', guide strand. This constrains the mRNA target site selected, as it will need to have G or C residues at the positions equivalent to the two 5' nucleotides of the passenger (sense) strand of the shRNA (Figure 2). The mismatch at the 5'-end of the 3', guide (antisense) shRNA strand is introduced by changing one position in the 5' passenger strand so it no longer matches the actual mRNA target sequence and is therefore no longer complementary to the guide strand. Note that there are several ways to introduce a discrepancy in the base-pairing stability of the 5'-ends of the siRNA duplex derived from the shRNA, and this is just one possible design approach.

The approach suggested above for forcing incorporation of the antisense shRNA strand into RISC introduces a 1 nt bulge into the shRNA. Indeed, bulges are always present in natural pre-miRNAs and these mismatches may fine-tune the cleavage sites used by Drosha and Dicer and/or may preclude activation of dsRNA-responsive cellular signaling pathways, such as the interferon response, by cellular pre-miRNAs. Indeed, recent preliminary data analyzing RNAi induced by siRNA duplexes that differ only in whether they form a perfect duplex or instead contain a central, single-nucleotide mismatch shows that, while both knock down their specific mRNA target with equal efficiency, the siRNA duplexes bearing the mismatch cause far fewer nonspecific effects (F Neipel, personal communication). Therefore, although it has been previously reported that efficient activation of the interferon response requires perfect dsRNAs of 30 bp in length,³¹ I strongly recommend the inclusion of at least one additional bulge within the shRNA design, as seen in native miRNAs, instead of designing a perfect RNA helix. This is, of course, again achieved by introducing a mutation into the passenger (sense) strand of the shRNA (Figure 2), not into the guide strand.

siRNA function

While RISC loading of the shRNA antisense strand should be efficient if the above rules are followed, how is RISC cleavage efficiency ensured? Relatively little is known about whether, or even if, siRNA primary sequence, in and of itself, influences RISC function once loading has occurred, but evidence from computational analyses suggests an at least modest effect. Specifically, analyses reported by Reynolds et al.³² suggest that an 'A' residue at position 3 of the sense strand and a U at position 10 of the sense strand enhance siRNA function significantly (Figure 2). In addition, avoiding a G at position 13 of the sense strand appears to be helpful. Finally, there is evidence that runs of G's and C's may be deleterious to function, while an excess of A's and U's in, particularly, the 5' 8 nt or so of the antisense (guide) strand may facilitate siRNA function. These latter design criteria may actually relate more to the next point.

mRNA target site accessibility

Unfortunately, even if your siRNA is optimally designed and efficiently incorporated into RISC, there is still no guarantee that it will be effective. Current evidence suggests that this is due to the fact that RISC complexes are not able to access mRNA target sites that are occluded by protein binding or by even relatively weak RNA secondary structure. Thus, Overhoff et al.³³ have presented compelling data demonstrating that local mRNA target secondary structure strongly inhibits siRNA efficiency, while Brown et al.³⁴ used an elegant approach to modulate artificially the folding of an mRNA target site and thereby demonstrate that reduced mRNA folding leads to enhanced mRNA cleavage. Perhaps the most compelling evidence showing that mRNA secondary structure inhibits RNAi comes from Westerhout et al.,³⁵ who selected a series of human immunodeficiency virus type 1 (HIV-1) variants that had become resistant to an anti-viral siRNA upon serial passage in its presence. Remarkably, one HIV-1 variant proved to be highly resistant to a previously very effective siRNA, yet had no mutations within the chosen mRNA target site. The resistant HIV-1 variant did, however, contain a point mutation 7 nt 5' to the target site that, upon closer analysis, proved able to stabilize a secondary structure that sequestered the viral mRNA target. Therefore, an otherwise fully functional siRNA can be entirely ineffective when the mRNA target is sequestered by another macromolecular interaction.

While it remains quite difficult to predict whether a given mRNA sequence is part of an RNA secondary structure, there has recently been significant progress in developing computer algorithms that address this problem.^{33, 36} For example, a potentially very useful program for the identification of accessible target sites within mRNA molecules can be accessed at http://sfold.wadsworth.org.³⁷ Beyond these more sophisticated computational approaches, it is evident that G:C-rich target sequences will be more likely to be sequestered into RNA secondary structures than would A:U-rich sequences.³² While the ongoing difficulty in accurately predicting whether a given mRNA target sequence is accessible means that even well-designed siRNAs fail at a significant rate; our experience suggests that following the shRNA design criteria listed above will give >80% knock-down of target mRNA-mediated protein expression by 1 out of three of the shRNAs tested.

Testing an shRNA for effectiveness

Having selected an mRNA target sequence and designed an shRNA that is predicted to load RISC efficiently with an siRNA specific for that sequence, how do you determine whether you do get effective, that is, >80%, knock-down of the target mRNA and its encoded gene product? Clearly, the first task is to construct an expression vector, based on the U6 or H1 promoter, which will transcribe the shRNA at high levels in transfected or transduced cells. A number of vectors have been reported, but we favor the expression vector pSUPER, based on the H1 promoter, because of its convenient design.³ pSUPER contains convenient unique BglII and HindIII sites, the former located immediately 5' to the transcription start site, that allow synthetic DNA oligonucleotides encoding the 60 nt shRNA sequence, together with the 3'-flanking 'TTTTT' transcription termination signal, to be readily inserted. Moreover, the resultant expression cassette is then flanked by several unique restriction enzyme sites that allow it to be readily moved into viral vectors (see below).

Once you have made a set of, say, five or more pSuper vectors targeted to your gene of interest, you could simply transfect each of these into an expressing cell line and use Western analysis, or RT-PCR or Northern analysis, to see if gene knock-down occurs. This is, in my view, a poor strategy to validate the effectiveness of the encoded siRNAs. One problem is that you may well be unsure as to the transfection efficiency of the cells you have analyzed. If the target cells are primary cells, or immortalized but differentiated cells such as B or T cells, the transfection efficiency you obtain may be fairly low. Of note, a respectable 30% transfection efficiency could give you no more than a 30% drop in target mRNA or protein expression levels even if the siRNA was 100% effective! Secondly, the protein encoded by the targeted mRNA may have a substantial half-life, so that only a modest drop in protein expression would be observed by 3 days post-transfection even if the siRNA was 100% effective and your transfection efficiency was also 100%! Conversely, because RNAi in human cells is predominantly, or perhaps even entirely, a cytoplasmic process,^{38, 39} quantitation of the mRNA encoded by the targeted gene might tend to underestimate the degree of knock-down obtained with the shRNA tested, because the nuclear pool of the targeted mRNA would be largely unaffected even if the cytoplasmic mRNA pool was severely reduced.

To address these issues, we use a standard approach to assess shRNA efficacy, which requires the construction of an expression plasmid that encodes an epitope-tagged form of the protein produced by the targeted gene. We then cotransfect 293T cells with this expression plasmid and one of the pSuper derivatives, wait 2 days, lyse the transfected cells and perform a Western analysis using an antiserum specific for the epitope tag. This approach avoids the problems listed above because: (1) Cotransfection is essentially 100% efficient, so that almost every cell that takes up the protein expression plasmid also takes up the pSuper-based shRNA vector, even if only 30% of the cells are actually transfected. (2) As there is no pre-existing pool of the epitope-tagged protein, protein half-life is not an issue. (3) As target protein expression is used as a read-out, the issue of a resistant, nuclear pool of the targeted mRNA is not a concern. Using this validation approach, it is therefore very simple to identify the shRNAs that are most effective at knocking down the targeted gene.

Establishing stable RNAi in cultured cells

Having identified one or more highly effective shRNA expression cassettes, the next question is how to produce a stable knock-down of the target gene in the relevant cells. One possibility is to move the small H1-based shRNA expression cassette into a plasmid bearing a eukaryotic selectable marker and to then transfect and subsequently select cells that express the marker. While feasible, this approach is limited to cells that are fairly easy to stably transfect, that is, this approach would tend to work poorly in most primary cells.

A better approach is to express the shRNA in the relevant cells using a retroviral^{40, 41} or lentiviral vector.^{42, 43, 44} These vectors can be readily generated at quite high titers (10⁶/ml) by cotransfection into 293T cells together with the relevant packaging constructs. Moreover, they will readily infect most cells, including primary cells and (in the case of lentiviral vectors) nondividing cells, and will integrate into the target cell genome in a predictable manner. Finally, both retroviral and, particularly, lentiviral vectors will generally give you long-term, stable expression of the encoded shRNA.^{42, 43, 44}

A range of retroviral or lentiviral vectors, bearing markers that allow drug selection or selection by fluorescence-activated cell sorting, have been described and many of these appear to work well.^{40, 41, 42, 43, 44} One criterion that we have noticed is important is that the vector should be self-inactivating (SIN), that is, the U3 region of the 3'-long terminal repeat (LTR) in the vector plasmid should be largely deleted so that the 5'-LTR promoter generated during reverse transcription is inactive in the transduced cells. Moreover, the shRNA expression cassette should be located 5' to any polII-dependent promoter that is used to drive expression of the mRNA encoding the selectable marker. The reason for this is that we have noted that the polIII-dependent H1 promoter is subject to significant transcriptional interference from polII-dependent promoters that are located 5'.

An effective lentiviral shRNA expression vector that was developed in my laboratory,⁴² termed pNL-SIN-CMV, is shown in Figure 3 in a form expressing the blastocidin resistance marker (forms expressing green fluorescent protein are also available). This vector contains unique ClaI and XbaI restriction sites, present in the 3'-LTR U3 region, that allow the insertion of the entire shRNA expression cassette from pSuper derivatives. This not only results in deletion of much of the lentiviral U3 region in transductants but also generates two copies of the H1-based shRNA expression cassette, one of which is located 5' to the CMV promoter used to drive the selectable marker. This vector has been shown to give stable, readily detectable levels of siRNA expression in transduced cells and can be used to knock-down target genes for periods of at least several months.⁴²

Figure 3.

Schematic of the NL-SIN-CMV lentiviral shRNA expression vector. This HIV-1-based lentiviral vector, shown here with the blastocidin resistance (BLR) marker, lacks the viral tat, rev, nef, env and vpu genes, but retains the viral gag, pol, vif and vpr genes as well as the Rev response element (RRE). In addition, almost all the 3'-LTR U3 region has been deleted and replaced with unique ClaI and XbaI sites, which allow insertion of H1-promoter-based shRNA expression cassettes derived from pSuper. This expression cassette is duplicated during reverse transcription in susceptible infected cells. As shown previously,⁴² this vector is able to direct stable shRNA expression in transduced cells.

Full figure and legend (11K)

Although the focus of this review is on shRNA expression in cultured cells, we note that others have used lentiviral shRNA expression vectors to knock-down gene expression in vivo in mice.^{43, 44} This knock-down was shown to be stable and ectopic expression of the encoded shRNAs did not have any nonspecific effects in these experimental animals.

Confirming the specificity of RNAi phenotypes

Once you have transduced cells with a retroviral or lentiviral expression vector encoding an shRNA specific for the gene of interest, you should see a substantial reduction in the level of expression of even stable proteins encoded by the targeted mRNA, if you wait a few days. If you do not see such a reduction, you should reconfirm the integrity of the lentiviral vector sequence and also confirm that you are indeed getting siRNA expression in the transduced cells, using Northern or primer extension analysis. We have not had a problem obtaining a good knock-down in transduced cells if good expression of a previously validated siRNA is achieved.

Assuming that you do get a good, stable knock-down of your target gene, and an interesting phenotype, how do you then show that this effect is specific, that is, not due to the fortuitous knock-down of some other gene product whose mRNA bears a related sequence or due to nonspecific activation of the host cell interferon response? By far, the best way to validate the specificity of the observed phenotype is to reintroduce the targeted gene in a form that is resistant to the siRNA used.⁴⁵ That is, you can mutate the targeted region, if it is in the mRNA open-reading frame, by taking advantage of the redundancy of the genetic code to introduce mutations that do not alter the predicted protein product yet render the encoded mRNA resistant to the shRNA used. Alternately, if you have used a target sequence in the 3'-untranslated region (3'-UTR) of the mRNA of interest, you can simply delete the natural 3'-UTR in the cDNA expression vector you design. Once you reintroduce the RNAi-resistant form of the targeted gene, by transfection or transduction into the knock-down cells, the observed phenotype should return to the previous, wild-type state despite ongoing shRNA expression.

If the above rescue approach is not feasible, then you could knock-down the gene using one or, preferably, two additional shRNAs that target entirely distinct sequences on the mRNA and show that you observe the same phenotype. This control is based on the premise that all three shRNAs would not activate the same nonspecific response or target the same unknown but unrelevant host mRNA transcript. To further support this hypothesis, you will need to show that expression of an irrelevant but functional shRNA (e.g. one targeted to green fluorescent protein, if this is not your selectable marker) has no phenotype. These control experiments, while not as definitive a demonstration of siRNA selectivity as the rescue experiment delineated above, are nevertheless generally viewed as an acceptable alternate approach. Importantly, simply showing that another, irrelevant or scrambled shRNA does not exert the same phenotype as the experimental shRNA is not an adequate control, as the irrelevant shRNA would not affect expression of a hypothetical mRNA that happened to bear a sequence homologous to the one targeted by the experimental shRNA.

Conclusion

Although RNAi is becoming an increasingly valuable tool in both basic and applied biological research, and also has the potential to be used in vivo in the treatment of disease, it remains difficult for researchers to design shRNAs or siRNA duplexes that consistently knock-down target gene expression to <20% of the level seen in wild-type cells. In this review, I have attempted to delineate a series of design criteria and technologies that should allow readers to construct effective shRNA expression vectors specific for essentially any human mRNA. Unfortunately, even armed with this information, knock-down of a target gene by >80% will still only be achieved with about one-third of the shRNAs tested. While it would clearly be preferable to obtain efficient knock-down with every shRNA tested, a success rate of one in three is acceptable in most experimental settings, although it does greatly complicate efforts to use viral shRNA expression libraries as a way of performing genetic screens. It remains to be seen if novel principles and insights will lead to future shRNA design criteria that give efficient target gene knock-down with every shRNA analyzed. Regardless, RNAi in general, and shRNAs in particular, promise to lead to a revolution in our understanding of the biological roles of vertebrate genes in health and disease. The reagents described in this review, including pSuper and the pNL-SIN-CMV lentiviral shRNA expression vector, are available upon request.

References

1. Tomari Y, Zamore PD. Perspective: machines for RNAi. Genes Dev 2005; 19: 517–529. | Article | PubMed | ISI | ChemPort |
2. Paddison PJ, Caudy AA, Bernstein E, Hannon GJ, Conklin DS. Short hairpin RNAs (shRNAs) induce sequence-specific silencing in mammalian cells. Genes Dev 2002; 16: 948–958. | Article | PubMed | ISI | ChemPort |
3. Brummelkamp TR, Bernards R, Agami R. A system for stable expression of short interfering RNAs in mammalian cells. Science 2002; 296: 550–553. | Article | PubMed | ISI | ChemPort |
4. Bernstein E, Caudy AA, Hammond SM, Hannon GJ. Role for a bidentate ribonuclease in the initiation step of RNA interference. Nature 2001; 409: 295–296. | Article |
5. Zhang H, Kolb FA, Brondani V, Billy E, Filipowicz W. Human Dicer preferentially cleaves dsRNAs at their termini without a requirement for ATP. EMBO J 2002; 21: 5875–5885. | Article | PubMed | ISI | ChemPort |
6. Elbashir SM, Harborth J, Lendeckel W, Yalcin A, Weber K, Tuschl T. Duplexes of 21-nucleotide RNAs mediate RNA interference in cultured mammalian cells. Nature 2001; 411: 494–498. | Article | PubMed | ISI | ChemPort |
7. Hammond SM, Bernstein E, Beach D, Hannon GJ. An RNA-directed nuclease mediates post-transcriptional gene silencing in Drosophila cells. Nature 2000; 404: 293–295. | Article | PubMed | ISI | ChemPort |
8. Schwarz DS, Hutvágner G, Haley B, Zamore PD. Evidence that siRNAs function as guides, not primers, in the Drosophila and human RNAi pathways. Mol Cell 2002; 10: 537–548. | Article | PubMed | ISI | ChemPort |
9. Martinez J, Patkaniowska A, Urlaub H, Lührmann R, Tuschl T. Single-stranded antisense siRNAs guide target RNA cleavage in RNAi. Cell 2002; 110: 563–574. | Article | PubMed | ISI | ChemPort |
10. Liu J, Carmell MA, Rivas FV, Marsden CG, Thomson JM, Song JJ et al. Argonaute2 is the catalytic engine of mammalian RNAi. Science 2004; 305: 1437–1441. | Article | PubMed | ISI | ChemPort |
11. Meister G, Landthaler M, Patkaniowska A, Dorsett Y, Teng G, Tuschl T. Human Argonaute2 mediates RNA cleavage targeted by miRNAs and siRNAs. Mol Cell 2004; 15: 185–197. | Article | PubMed | ISI | ChemPort |
12. Doench JG, Petersen CP, Sharp PA. siRNAs can function as miRNAs. Genes Dev 2003; 17: 438–442. | Article | PubMed | ISI | ChemPort |
13. Zeng Y, Yi R, Cullen BR. MicroRNAs and small interfering RNAs can inhibit mRNA expression by similar mechanisms. Proc Natl Acad Sci USA 2003; 100: 9779–9784. | Article | PubMed | ChemPort |
14. Hutvágner G, Zamore PD. A microRNA in a multiple-turnover RNAi enzyme complex. Science 2002; 297: 2056–2060. | Article | PubMed | ISI | ChemPort |
15. Lagos-Quintana M, Rauhut R, Lendeckel W, Tuschl T. Identification of novel genes coding for small expressed RNAs. Science 2001; 294: 853–858. | Article | PubMed | ISI | ChemPort |
16. Cai X, Hagedorn CH, Cullen BR. Human microRNAs are processed from capped, polyadenylated transcripts that can also function as mRNAs. RNA 2004; 10: 1957–1966. | Article | PubMed | ISI | ChemPort |
17. Lee Y, Kim M, Han J, Yeom KH, Lee S, Baek SH et al. MicroRNA genes are transcribed by RNA polymerase II. EMBO J 2004; 23: 4051–4060. | Article | PubMed | ISI | ChemPort |
18. Lee Y, Ahn C, Han J, Choi H, Kim J, Yim J et al. The nuclear RNase III Drosha initiates microRNA processing. Nature 2003; 425: 415–419. | Article | PubMed | ISI | ChemPort |
19. Han J, Lee Y, Yeom KH, Kim YK, Jin H, Kim VN. The Drosha–DGCR8 complex in primary microRNA processing. Genes Dev 2004; 18: 3016–3027. | Article | PubMed | ISI | ChemPort |
20. Gregory RI, Yan KP, Amuthan G, Chendrimada T, Doratotaj B, Cooch N et al. The microprocessor complex mediates the genesis of microRNAs. Nature 2004; 432: 235–240. | Article | PubMed | ISI | ChemPort |
21. Denli AM, Tops BB, Plasterk RH, Ketting RF, Hannon GJ. Processing of primary microRNAs by the microprocessor complex. Nature 2004; 432: 231–235. | Article | PubMed | ISI | ChemPort |
22. Lund E, Guttinger S, Calado A, Dahlberg JE, Kutay U. Nuclear export of microRNA precursors. Science 2004; 303: 95–98. | Article | PubMed | ISI | ChemPort |
23. Yi R, Qin Y, Macara IG, Cullen BR. Exportin-5 mediates the nuclear export of pre-microRNAs and short hairpin RNAs. Genes Dev 2003; 17: 3011–3016. | Article | PubMed | ISI | ChemPort |
24. Zeng Y, Yi R, Cullen BR. Recognition and cleavage of primary microRNA precursors by the nuclear processing enzyme Drosha. EMBO J 2005; 24: 138–148. | Article | PubMed | ISI | ChemPort |
25. Khvorova A, Reynolds A, Jayasena SD. Functional siRNAs and miRNAs exhibit strand bias. Cell 2003; 115: 209–216. | Article | PubMed | ISI | ChemPort |
26. Schwarz DS, Hutvagner G, Du T, Xu Z, Aronin N, Zamore PD. Asymmetry in the assembly of the RNAi enzyme complex. Cell 2003; 115: 199–208. | Article | PubMed | ISI | ChemPort |
27. Bridge AJ, Pebernard S, Ducraux A, Nicoulaz AL, Iggo R. Induction of an interferon response by RNAi vectors in mammalian cells. Nat Genet 2003; 34: 263–264. | Article | PubMed | ISI | ChemPort |
28. Zeng Y, Wagner EJ, Cullen BR. Both natural and designed micro RNAs can inhibit the expression of cognate mRNAs when expressed in human cells. Mol Cell 2002; 9: 1327–1333. | Article | PubMed | ISI | ChemPort |
29. Boden D, Pusch O, Silbermann R, Lee F, Tucker L, Ramratnam B. Enhanced gene silencing of HIV-1 specific siRNA using microRNA designed hairpins. Nucleic Acid Res 2004; 32: 1154–1158. | Article | PubMed | ISI | ChemPort |
30. Zeng Y, Cullen BR. Structural requirements for pre-microRNA binding and nuclear export by Exportin 5. Nucleic Acid Res 2004; 32: 4776–4785. | Article | PubMed | ISI | ChemPort |
31. Manche L, Green SR, Schmedt C, Matthews MB. Interactions between double-stranded RNA regulators and the protein kinase. Mol Cell Biol 1992; 12: 5238–5248. | PubMed | ISI | ChemPort |
32. Reynolds A, Leake D, Boese Q, Scaringe S, Marshall WS, Khvorova A. Rational siRNA design for RNA interference. Nat Biotech 2004; 22: 326–330. | Article | PubMed | ISI | ChemPort |
33. Overhoff M, Alken M, Far RK, Lemaitre M, Lebleu B, Sczakiel G et al. Local RNA target structure influences siRNA efficacy: a systematic global analysis. J Mol Biol 2005; 348: 871–881. | Article | PubMed | ISI | ChemPort |
34. Brown KM, Chu C, Rana TM. Target accessibility dictates the potency of human RISC. Nat Struct Mol Biol 2005; 12: 469–470. | Article | PubMed | ISI | ChemPort |
35. Westerhout EM, Ooms M, Vink M, Das AT, Berkhout B. HIV-1 can escape from RNA interference by evolving an alternative structure in its RNA genome. Nucleic Acid Res 2005; 33: 796–804. | Article | PubMed | ISI | ChemPort |
36. Ding Y, Lawrence CE. A statistical sampling algorithm for RNA secondary structure prediction. Nucleic Acid Res 2003; 31: 7280–7301. | Article | PubMed | ISI | ChemPort |
37. Ding Y, Chan CY, Lawrence CE. Sfold web server for statistical folding and rational design of nucleic acids. Nucleic Acid Res 2004; 32: W135–W141. | PubMed | ISI | ChemPort |
38. Zeng Y, Cullen BR. RNA interference in human cells is restricted to the cytoplasm. RNA 2002; 8: 855–860. | Article | PubMed | ISI | ChemPort |
39. Robb GB, Brown KM, Khurana J, Rana TM. Specific and potent RNAi in the nucleus of human cells. Nat Struct Mol Biol 2005; 12: 133–137. | Article | PubMed | ISI | ChemPort |
40. Barton GM, Medzhitov R. Retroviral delivery of small interfering RNA into primary cells. Proc Natl Acad Sci USA 2002; 99: 14943–14945. | Article | PubMed | ChemPort |
41. Brummelkamp TR, Bernards R, Agami R. Stable suppression of tumorigenicity by virus-mediated RNA interference. Cancer Cell 2002; 2: 243–247. | Article | PubMed | ISI | ChemPort |
42. Lee M-TM, Coburn GA, McClure MO, Cullen BR. Inhibition of human immunodeficiency virus type 1 replication in primary macrophages by using Tat- or CCR5-specific small interfering RNAs expressed from a lentivirus vector. J Virol 2003; 77: 11964–11972. | Article | PubMed | ISI | ChemPort |
43. Rubinson DA, Dillon CP, Kwiatkowski AV, Sievers C, Yang L, Kopinja J et al. A lentivirus-based system to functionally silence genes in primary mammalian cells, stem cells and transgenic mice by RNA interference. Nat Genet 2003; 33: 401–406. | Article | PubMed | ISI | ChemPort |
44. Tiscornia G, Singer O, Ikawa M, Verma IM. A general method for gene knockdown in mice by using lentiviral vectors expressing small interfereing RNA. Proc Natl Acad Sci USA 2003; 100: 1844–1848. | Article | PubMed | ChemPort |
45. Whither RNAi? Nat Cell Biol 2003; 5: 489–490. | Article |

Acknowledgements

I thank Dr Alexandra Schäfer for assistance with the figures used in this manuscript. This work was supported by Grant GM071408 from the National Institutes of Health.