COMMENTARY

The specifics of small interfering RNA specificity

Andrew Dillin* Molecular and Cell Biology Laboratory, The Salk Institute for Biological Studies, 10010 North Torrey Pines Road, La Jolla, CA 92037

he discovery of transgene silenc- ing in plants and double- stranded RNA (dsRNA) inter- ference in the worm TCaenorhabditis elegans has led to the latest revolution in molecular biology, RNA interference (RNAi). Over 10 years ago it was noted that several trans- genic plant lines each containing the same ectopic transgene not only failed to be expressed but also inhibited the expression of the endogenous gene (1). Similarly, a determined and Andy Fire (2), attempting to reduce gene function using antisense RNA in the worm, discovered a minor contami- nant in their antisense RNA prepara- tions effectively and repeatedly reduced expression of the endogenous gene. In both cases, dsRNA homologous to the gene of interest was responsible for these observations. In the last 4 years, these discoveries have been extended to include protozoa, fungi, and mammals. What is RNAi? RNAi is a highly con- served mechanism found in almost all and believed to serve as an antiviral defense mechanism. The mo- lecular details are becoming clear from combined genetic and biochemical ap- proaches (reviewed in refs. 3 and 4). On entry into the cell, the dsRNA is cleaved by an RNase III like , , into small interfering (21- to 23- nt) (siRNAs) (5–8) (Fig. 1). Bio- chemical evidence indicates the siRNAs are incorporated into a multisubunit complex, the RNAi-induced si- Fig. 1. Overview of dsRNA-mediated mRNA degradation. On entry into cells, dsRNA is cleaved by Dicer lencing complex (RISC), which directs into 21- to 23-nt siRNAs. siRNAs are complexed with a large multiprotein complex, the RISC. RISC is thought the siRNA to the appropriate mRNA. to unwind the siRNA to help target the appropriate mRNA (shown in green). The siRNA͞mRNA hybrid is New data suggest that the RISC com- cleaved, releasing the siRNA, and the mRNA is degraded by endo and . In worms and plants, plex may unwind the siRNA to help in- the siRNA can also serve as a template for RdRP using the mRNA as a template. Elongation of the siRNA can lead to the production of more siRNAs that could share homology to other genes (shown in orange), teractions with the target mRNA (9). causing their degradation, known as transitive RNAi. Mismatches Ͼ1–2 bp within the 21- to 23-nt siRNA effectively disrupt proper degradation of the target mRNA plays an important role in RNAi. One, kingdom (14). Mainly, siRNA targeting (10, 11). worms with mutations in novel RdRPs is very specific; however, elongation of In worms, interaction between the are not able to perform RNAi (12, 13). the siRNA using the mRNA as a tem- siRNA and mRNA can lead to immedi- Two, elegant studies using fusion genes plate could lead to nonspecific interfer- ate cleavage by Dicer, liberating a new of unc-22 and GFP showed that dsRNA ence by sequences homologous to other siRNA, and degradation of the mRNA directed against GFP could effectively genes, known as transitive RNAi. by endo- and exonucleases (Fig. 1). Al- At present, this phenomenon is partic- ternatively, the siRNA can serve as a lead to the degradation of the endoge- nous unc-22 (13) (Fig. 2). These experi- ular to plants and nematodes and does primer for an RNA-dependent RNA not appear to be a concern for mamma- ments also showed the polarity of the polymerase (RdRP), creating many Ј Ј lian systems, because there is no easily more siRNAs. The action of an RdRP RdRP must be 5 to 3 on the antisense Ј identifiable mammalian RdRP homolog. could explain the catalytic mechanism of strand because unc-22 fused to the 5 Additionally, in vitro reconstitution of RNAi, because only a few dsRNA mole- end of GFP caused degradation of the mammalian RNAi activity does not re- cules are required to degrade a much endogenous unc-22, but fusion to the larger population of mRNAs, and RNAi 3Ј end did not. These experiments dis- is inheritable (2). Two important pieces played a new level of specificity not pre- See companion articles on pages 6343 and 6347. of data support the idea that an RdRP viously appreciated outside the plant *E-mail: [email protected].

www.pnas.org͞cgi͞doi͞10.1073͞pnas.1232238100 PNAS ͉ May 27, 2003 ͉ vol. 100 ͉ no. 11 ͉ 6289–6291 Downloaded by guest on October 1, 2021 process of RNAi directed against an dsRNAs induce the IFN response, exogenous gene, GFP, in human embry- thereby shutting down , in- onic kidney cells (293 cells). Remark- ducing RNaseL, and apoptosis. Lower ably, in each of three experimental set- eukaryotes lack this response. One pos- tings, there was no statistically sibility may be a competition between consistent change. Therefore, two con- those mechanisms required for transitive clusions can be drawn from this experi- RNAi, mainly an RdRP, and the IFN ment. One, if transitive RNAi does exist response. It is quite possible that exten- in mammals, it is not robust enough in sion of a siRNA, using the mRNA as a this experimental setting to cause non- template, could create a long dsRNA specific degradation of any mRNA de- recognized by the IFN response. tectable on their DNA microarray. How- Worms, which lack an IFN response, are ever, it should be noted that although capable of transitive RNAi and can take there was no clear decrease of mRNAs up kilobase-sized dsRNA, but mammals found after 48 and 72 h of siRNA expo- cannot. The evolution of the IFN re- sure, it is not clear how long it might sponse possibly came at the expense of take a cell to mount such a response. the RdRP. Two, there appears to be no active up- Are we ready for the first siRNA regulation of cellular genes in response therapy? No. Although we are closer to to RNAi. However, it is possible that the clinical setting than previously antic- the RNAi machinery is acutely up-regu- ipated, several key questions need atten- lated in response to siRNA, but not tion. It is becoming clear that siRNAs maintained after 48 h. can target the appropriate transcript Chi et al. (16) also recreate the ele- with great fidelity, and this holds prom- gant worm experiments of Sijen et al. ise to inactive mRNAs of disease alleles (13) to formally test transitive RNAi in without affecting the normal allele. If 293 cells. In this experiment, Chi et al. indeed mammalian cells lack an RdRP, created a fusion gene of actin DNA se- then the issue of transitive RNAi is no quences with GFP DNA sequences (Fig. longer a problem. Although the articles 2). Within the same cell, they engi- described here indicate that kidney and neered another fusion gene between lung cells are not capable of transitive actin and luciferase. If the siRNA were RNAi, we do not know whether other able to spread to nearby sequences, it cell types (postmitotic for instance) will would create new siRNAs homologous have the same transcriptional response to actin, causing the subsequent degra- to RNAi or possess an RdRP that could Fig. 2. Formal test of transitive RNAi in worms dation of the actin–luciferase fusion be missing in these cell types. What is and human 293 cells. (A) In worms, Sijen et al. (13) gene. This did not happen. What did used GFP–unc-22 fusion genes to test whether happen is that the actin–GFP fusion clear, however, is that we must hold all mRNA from the endogenous unc-22 gene (shown gene was degraded, but the actin-lucif- siRNA-based therapies to the same set in red) could be degraded. When dsRNA directed erase gene remained intact. Even more of criteria used in these studies before toward GFP was used, the endogenous unc-22 embarking on clinical trials. Ј compelling, siRNA directed to within 20 could be degraded only when unc-22 was fused 5 Of greater concern is that we do not to the GFP DNA sequences. Fusion at the 3Ј end did nt of the actin–GFP fusion junction did not cause degradation of the endogenous unc-22 not lead to degradation of the actin– know whether cells undergoing RNAi mRNA. (B) Chi et al. (16) used a similar approach to luciferase fusion. This result clearly indi- are as healthy as their unaffected coun- create human kidney cells (293 cells) expressing cates that 293 cells lack transitive RNAi. terparts. A few pieces of data suggest gene fusions of actin–GFP and actin–luciferase. In a tour de force, Semizarov et al. that they should not. One, siRNAs and siRNA directed against GFP caused degradation of (17) reasoned that if siRNA caused non- members of the RNAi machinery play a the actin–GFP fusion but did not lead to elongation specific effects, then comparison of role in chromosome architecture in sev- of the siRNA into the actin sequences and subse- three different siRNA experiments tar- eral organisms (reviewed in refs. 18 and quent degradation of the actin–luciferase gene 19). Two, the RNAi machinery can be fusion. Two siRNAs were used, E1 and E3. E1 is geting three different genes should within 20 bp of the fusion junction. share some commonality within their saturated (20). Therefore, it is conceiv- transcriptional profiles. They did not. able that a cell devoting much of its The expression profiles of lung carci- RNAi machinery to the process of quire an RNA polymerase (8, 9) and noma cells treated with siRNA directed mRNA degradation could become de- RNAi in mouse oocytes is not blocked toward Rb, AKT1, or Plk1 were unre- fective in chromosome function or be by drugs that interfere with RNA poly- lated. Therefore, similar to the Chi et al. more susceptible to viral infection. Fi- merases (15). Lastly, two reports in this study, nonspecific degradation of cellu- nally, worms and green algae with de- issue of PNAS, using DNA microarray lar mRNAs by transitive RNAi was un- fects in the RNAi machinery have a analysis, indicate that there is no signifi- detectable, and there appears to be no higher incidence of transposon hopping, cant decline of endogenous mRNAs in transcriptional regulation of cellular which in turn could lead to increased mammalian cells subjected to siRNA gene in response to siRNAs. Equally mutagenesis (21–23). It is unknown (16, 17). important, Semizarov et al. discovered whether this function is conserved in In one set of experiments, Chi et al. that high doses of siRNAs did indeed humans, but, if so, siRNA could lead to (16) used a DNA microarray containing induce many nonspecific genes, but genome instability and cancer later in 43,000 DNA elements (36,000 genes) to lower doses did not. Therefore, more is life. address the specificity of siRNA. They not always better. A competitive growth experiment will asked whether global changes in mRNA Why do worms have transitive RNAi be necessary to clarify this issue. In this levels could occur because of the active and mammals do not? In mammals, long experiment, a cell culture expressing a

6290 ͉ www.pnas.org͞cgi͞doi͞10.1073͞pnas.1232238100 Dillin Downloaded by guest on October 1, 2021 reporter gene is divided in half. One siRNA in one cell type can spread to Although these questions need to be half is treated with siRNA directed to- neighboring cells not intended to receive addressed, the therapeutic use for ward the reporter and the other half the siRNA. siRNA is not actively taken siRNA looks promising. With the rich sham treated. The two halves are cocul- up in cultured cells; however, it is not resource of experiments used to deter- tured for several divisions (n ϭ 10–20). clear what happens in an intake animal. mine the efficacy of antisense RNA, If there is indeed a negative impact on Precedence in worms and plants indi- viral vector delivery systems (24–27) cell physiology by RNAi, then those cates that this type of mechanism does and the stability of dsRNA, it will not cells treated with siRNA against the re- exist and may serve as a systematic be too long in the future that we see the porter will be under represented after method to ‘‘immunize’’ the entire organ- first therapy using this revolutionary technology. 10–20 divisions. If there is no impact, isms against potential viral infection. then an equal number of siRNA treated Therefore, siRNA directed against your I thank Matthew Weitzman, Jan Karlseder, and sham treated cells will be present. favorite oncogene could reduce tumor and Tony Hunter for insightful discussions Also of concern is the question of growth, but could also affect normal cell and Jamie Simon for design and construction confinement. We do not know whether division in the body. of the figures.

1. Napoli, C., Lemieux, C. & Jorgensen, R. (1990) Lendeckel, W. & Tuschl, T. (2001) EMBO J. 20, 19. Dernburg, A. F. & Karpen, G. H. (2002) Cell 111, Plant Cell 2, 279–289. 6877–6888. 159–162. 2. Fire, A., Xu, S., Montgomery, M. K., Kostas, S. A., 11. Elbashir, S. M., Harborth, J., Lendeckel, W., Yalcin, A., 20. Kamath, R. S., Martinez-Campos, M., Zipperlen, Driver, S. E. & Mello, C. C. (1998) Nature 391, Weber, K. & Tuschl, T. (2001) Nature 411, 494–498. P., Fraser, A. G. & Ahringer, J. (2000) Genome 806–811. 12. Smardon, A., Spoerke, J. M., Stacey, S. C., Klein, Biol. 2, 231–237. 3. Tijsterman, M., Ketting, R. F. & Plasterk, R. H. M. E., Mackin, N. & Maine, E. M. (2000) Curr. 21. Ketting, R. F., Haverkamp, T. H., van Luenen, (2002) Annu. Rev. Genet. 36, 489–519. Biol. 10, 169–178. H. G. & Plasterk, R. H. (1999) Cell 99, 133–141. 4. Tabara, H., Yigit, E., Siomi, H. & Mello, C. C. 13. Sijen, T., Fleenor, J., Simmer, F., Thijssen, K. L., 22. Ketting, R. F. & Plasterk, R. H. (2000) Nature 404, (2002) Cell 109, 861–871. Parrish, S., Timmons, L., Plasterk, R. H. & Fire, A. 296–298. 5. Bernstein, E., Caudy, A. A., Hammond, S. M. & (2001) Cell 107, 465–476. 23. Wu-Scharf, D., Jeong, B., Zhang, C. & Cerutti, H. Hannon, G. J. (2001) Nature 409, 363–366. 14. Voinnet, O., Vain, P., Angell, S. & Baulcombe, (2000) Science 290, 1159–1162. 6. Ketting, R. F., Fischer, S. E., Bernstein, E., Sijen, D. C. (1998) Cell 95, 177–187. 24. Kootstra, N. A. & Verma, I. M. (2003) Annu. Rev. T., Hannon, G. J. & Plasterk, R. H. (2001) Genes 15. Stein, P., Svoboda, P., Anger, M. & Schultz, R. M. Pharmacol. Toxicol. 43, 413–439. Dev. 15, 2654–2659. (2003) RNA 9, 187–192. 25. Tiscornia, G., Singer, O., Ikawa, M. & Verma, 7. Knight, S. W. & Bass, B. L. (2001) Science 293, 16. Chi, J.-T., Chang, H. Y., Wang, N. N., Chang, I. M. (2003) Proc. Natl. Acad. Sci. USA 100, 2269–2271. D. S., Dunphy, N. & Brown, P. O. (2003) Proc. 1844–1848. 8. Zamore, P. D., Tuschl, T., Sharp, P. A. & Bartel, Natl. Acad. Sci. USA 100, 6343–6346. 26. Stewart, S. A., Dykxhoorn, D. M., Palliser, D., D. P. (2000) Cell 101, 25–33. 17. Semizarov, D., Frost, L., Sarthy, A., Kroeger, P., Mizuno, H., Yu, E. Y., An, D. S., Sabatini, D. M., 9. Martinez, J., Patkaniowska, A., Urlaub, H., Halbert, D. N. & Fesik, S. W. (2003) Proc. Natl. Chen, I. S., Hahn, W. C., Sharp, P. A., et al. (2003) Luhrmann, R. & Tuschl, T. (2002) Cell 110, Acad. Sci. USA 100, 6347–6352. RNA 9, 493–501. 563–574. 18. Bailis, J. M. & Forsburg, S. L. (2002) Genome Biol. 27. Xia, H., Mao, Q., Paulson, H. L. & Davidson, B. L. 10. Elbashir, S. M., Martinez, J., Patkaniowska, A., 3, 1035.1–1035.4. (2002) Nat. Biotechnol. 20, 1006–1010.

Dillin PNAS ͉ May 27, 2003 ͉ vol. 100 ͉ no. 11 ͉ 6291 Downloaded by guest on October 1, 2021