Logo of pnasPNASInfo for AuthorsSubscriptionsAboutThis Article
Proc Natl Acad Sci U S A. Feb 17, 2004; 101(7): 1892–1897.
Published online Feb 9, 2004. doi:  10.1073/pnas.0308698100
PMCID: PMC357023
Cell Biology

Short interfering RNAs can induce unexpected and divergent changes in the levels of untargeted proteins in mammalian cells


RNA interference (RNAi) mediated by short interfering RNAs (siRNAs) is a widely used method to analyze gene function. To use RNAi knockdown accurately to infer gene function, it is essential to determine the specificity of siRNA-mediated RNAi. We have assessed the specificity of 10 different siRNAs corresponding to the MEN1 gene by examining the expression of two additional genes, TP53 (p53) and CDKN1A (p21), which are considered functionally unrelated to menin but are sensitive markers of cell state. MEN1 RNA and corresponding protein levels were all reduced after siRNA transfection of HeLa cells, although the degree of inhibition mediated by individual siRNAs varied. Unexpectedly, we observed dramatic and significant changes in protein levels of p53 and p21 that were unrelated to silencing of the target gene. The modulations in p53 and p21 levels were not abolished on titration of the siRNAs, and similar results were obtained in three other cell lines; in none of the cell lines tested did we see an effect on the protein levels of actin. These data suggest that siRNAs can induce nonspecific effects on protein levels that are siRNA sequence dependent but that these effects may be difficult to detect until genes central to a pivotal cellular response, such as p53 and p21, are studied. We find no evidence that activation of the double-stranded RNA-triggered IFN-associated antiviral pathways accounts for these effects, but we speculate that partial complementary sequence matches to off-target genes may result in a micro-RNA-like inhibition of translation.

The discovery that double-stranded RNA (dsRNA) molecules of 21–23 nt (short interfering RNAs, siRNAs) can silence targeted genes through sequence-specific cleavage of the cognate RNA transcript has led to the rapid adoption of technologies based on this RNA interference (RNAi) mechanism for analysis of gene function. The specificity of the effect of siRNAs in mammalian cells has not been comprehensively studied, but the few studies that have involved genomewide expression profiles of cells transfected with different siRNAs against a single gene have arrived at somewhat contradictory conclusions (13).

Although DNA microarrays are ideally suited to measure siRNA specificity at the mRNA level, they may not provide an accurate representation of potential effects at the protein level, including off-target effects due to the action of siRNAs as micro-RNAs (miRNAs). miRNAs, a naturally occurring species of noncoding RNA, interact with the same protein complex involved in RNAi but mediate inhibition of gene expression through translational blockade (49). Sequence fidelity between these small RNA species and the RNA transcript is thought to be the critical parameter determining whether RNA cleavage or translational blockade is induced. RNA cleavage can occur when an exact sequence match between the antisense siRNA strand and the cognate target transcript is present, and translational repression occurs when there is less homology between the two species (10).

A vital assumption in the application of siRNA-mediated RNAi as a functional genomics tool is that the knockdown of a targeted gene is specific at both the RNA and protein level. In this study, we investigated the specificity of siRNA-mediated gene silencing by transfecting 10 different siRNAs corresponding to a single gene [multiple endocrine neoplasia, type I (MEN1)] and assayed for expression of TP53 (p53) and CDKN1A (p21) in four different human cell lines. Because p53 and p21 function in a vast number of related and unrelated cellular pathways, we hypothesized that these proteins would serve as highly sensitive detectors for direct and indirect off-target effects mediated by the siRNAs. If siRNAs are indeed highly specific for their cognate targets, then siRNAs designed against the same gene, but with different nucleotide sequences, would have similar effects on the levels of p53 and p21; or if there is no direct functional link between menin and p21 and p53, then the mRNA and protein levels for these genes should remain unchanged relative to controls. This assessment of siRNA specificity differs from that of previous tests in that it accounts for instances in which siRNAs could alter protein levels without concomitantly affecting mRNA levels.


siRNA Design. Ten different siRNAs were designed corresponding to the MEN1 gene (GenBank accession no. U93236). Sequences were chosen on the basis of early studies of RNAi in Drosophila and mammalian cells describing intracellularly generated siRNAs and effective synthetic siRNAs, respectively (1113). siRNAs against the bacterial chloramphenicol acetyl transferase (CAT) and β-galactosidase (lacZ) genes, with no significant sequence homology to any known human gene, were synthesized as negative controls. siRNA corresponding to p53 was used as a positive control for silencing. All siRNA sequences were blast searched in the National Center for Biotechnology Information's (NCBI) “search for short nearly exact matches” mode against all human sequences deposited in the GenBank and RefSeq databases and were not found to have significant homology (>17 contiguous nucleotides of identity) to genes other than the targets. siRNAs were synthesized and annealed by either Dharmacon (MEN1 6-10 and p53) or Qiagen (MEN1 1-5, lacZ, and CAT). MEN1-5 siRNA was fluorescently labeled at the 5′ end of the sense strand for visualization of transfection efficiency. All siRNA sequences are listed in Table 1.

Table 1.
siRNAs used for menin knockdown

Cell Culture, Transfections, and Western Blot. All cell lines were purchased from the American Type Culture Collection and maintained in DMEM supplemented with 10% vol/vol FBS. For HeLa cells, 50 nM siRNA was transfected at 70–80% confluency. Forty nM siRNA was used for CaSKi, SiHa, and MCF7 cells. The ratio of siRNA to lipid was 1:2.5. Cells were harvested 72 h posttransfection and assayed by Western blot for expression of menin (14), p53 (sc-126, Santa Cruz Biotechnology), p21 (AB-11, Lab Vision, Fremont, CA), and actin (sc-1616, Santa Cruz Biotechnology). Transfections were carried out by using Lipofectamine 2000 (Invitrogen). Mock transfected cells were treated with Lipofectamine 2000, but no siRNA. To test the effect of different lipid-based transfection agents, MEN-1 siRNAs were also transfected into HeLa cells by using Oligofectamine (MEN1-1-10) and Lipofectamine (MEN1-1-5) (Invitrogen). To test whether contaminants in the siRNA prep could be inducing the nonspecific effects, MEN1-2, MEN1-3, LacZ, and CAT siRNAs were resynthesized and assayed.

Quantitative RT-PCR. Total RNA extracted from siRNA-transfected HeLa cells was DNase-treated (RQ1, Promega) and reverse transcribed (High-Capacity cDNA Archive Kit, Applied Biosystems). Quantitative PCR was performed on a Perkin–Elmer 7700 by using TaqMan PCR master mix and commercially available primers and FAM-labeled TaqMan probes (Assays-on-Demand, Applied Biosystems). Genes assayed included MEN1, p21, p53, 2′-5′-oligoadenylate synthetase (OAS1), and GAPD (glyceraldehyde-3-phosphate dehydrogenase). Each sample was run in duplicate (Exp. 1) and then repeated in triplicate (Exp. 2). Two negative controls consisting of mock reverse-transcribed cDNA (–RT) were included with each set of PCRs to detect the possibility of genomic DNA contamination. Average Ct values for MEN1, p53, p21, and OAS1 were calculated and normalized to Ct values for GAPD. The average ΔCt value from each two experiments was calculated, and the results were graphed with the corresponding standard deviation indicated with error bars in the figures.

Computational Analyses. Four different computer algorithms were designed to search for homologous targets in the 3′ UTRs of mRNA transcripts. As a source of mRNA sequences, we used the 19,508 NCBI human mRNA Reference Sequences (RefSeqs) available on July 24, 2003 (accession nos. beginning with NM_). We extracted the 3′ UTR of each RefSeq by taking all sequence downstream of the annotated protein coding sequence (CDS). Algorithm 1 was designed to search for targets with a minimum of 8 complementary nucleotides at both ends of the siRNA strand, allowing for a nonhomologous central bulge. The number of bases in the bulge varied from zero to the maximum length of the sequence minus 16, the combined length of the 8-bp anchors. Algorithm 2 was identical but allowed for G:U “wobble” base pairing between the siRNA strand and the 3′ UTR target. Algorithm 3 searched for matches between 3′ UTRs and the first eight nucleotides of the 5′ ends of each siRNA strand. Algorithm 4 searched for identity between 3′ UTRs and the first 10 nucleotides of the 5′ ends of each siRNA strand, allowing for G:U interactions. Additional details and algorithms are available from the authors on request.


siRNAs Targeted to a Single Gene Can Induce Significant and Divergent Changes in the Levels of p53 and p21. To investigate the specificity of siRNA-mediated RNAi, we transfected 10 different siRNAs corresponding to the MEN1 gene into HeLa cells and assayed the levels of menin, p21 and p53, and actin by Western blot analysis. Controls included cells either mock transfected or transfected with siRNAs containing sequences with no significant homology to any known human gene (either CAT-siRNA or LacZ-siRNA). The transfection efficiency was >90% as judged by fluorescence using the FITC-labeled MEN1-5 siRNA. Relative to mock, all 10 MEN1-siRNAs appeared to reduce menin levels, albeit to different levels (Fig. 1a); MEN1-siRNAs 1, 5, 7, and 10 showed the greatest level of inhibition at a protein level. Surprisingly, we detected dramatic differences in the levels of p21 and p53 that were not related to the inhibition of menin. Of the four siRNAs that reduced menin levels the most, one up-regulated p21 and p53 (MEN1-7), one down-regulated p21 (MEN1-10), and two showed no significant change relative to mock transfected (MEN1-1 and MEN1-5). Variations in menin and p21 levels were detected between siRNA's MEN1-1 and MEN1-2, which target the same region of MEN1 but differ by only one additional nucleotide on the 5′ end of the antisense strand. Similarly, variations in menin, p53, and p21 were detected between siRNAs MEN1-3 and MEN1-4, which differ by one additional nucleotide on the 5′ end of the sense strand. Cells transfected with the six remaining MEN1-siRNAs also showed significant variations in p21 and p53 levels. The effects of MEN1-siRNAs on p21 mRNA and protein levels could not be fully explained on the basis of the changes in p53. The p53 siRNA consistently reduced the expression of its direct target p53 and its down stream target p21 relative to mock transfected cells but had no effect on menin levels. None of the siRNAs tested affected actin levels. The CAT control siRNA showed no or minimal effect on the levels of menin, p53, or p21 relative to mock transfection; however, we saw variable results using the LacZ siRNA, in this instance both a decrease in menin and p21 levels, but this was inconsistent between experiments and this variability was not seen in other cell lines (Fig. 1b). Similar results were observed when the assay was repeated with re-synthesized oligonucleotides (MEN1-2, MEN1-3, LacZ, and CAT; data not shown).

Fig. 1.
Western blot and quantitative RT-PCR analyses of siRNA transfected cells. (a) Western blot analysis of HeLa cells transfected with 10 different MEN1-siRNAs show significant differences in the levels of p21 and p53 that are independent of menin levels. ...

RNAi is often very effective at minimal concentrations, and using the lowest possible concentration of siRNA has been suggested to prevent saturation of the RNAi machinery and unwanted side effects (15). In this case, titration of the siRNA down to 10 nM reduced the silencing of the MEN1 target but did not completely abolish the effects on p53/p21 induced by some of the siRNAs (Fig. 1a). These results suggest that the effects we observe are not likely to be caused by using siRNAs at too high a concentration. To ensure that the effects seen were independent of the lipid agent used we compared the results obtained with two additional cationic lipids and saw the same results for all of the siRNAs tested (data not shown).

To determine whether the nonspecific effects we observed were limited to HeLa cells, we transfected all 13 siRNAs (MEN1-10, LacZ, CAT, and p53) into CaSki, SiHa, and MCF7 cells. Despite relatively poor silencing of menin in these cell lines, we still detected significant changes in p53 and p21 levels that were broadly consistent with that seen in HeLa cells for each of the siRNAs studied (Fig. 1b).

Quantitative RT-PCR Analyses Show Relatively Minor Differences in p21 and p53 Expression. Quantitative PCR analysis of mRNA extracted from siRNA-transfected HeLa cells was used to compare mRNA levels of MEN1, p53, and p21 with the corresponding protein data. All 10 siRNAs to MEN1 were associated with a decrease in MEN1 transcript levels compared to mock transfected cells. The percent of activity remaining for menin varied from ≈14% for MEN1-5 to 55% for MEN1-3 (Fig. 1c). In general, MEN1 RNA levels correlated with the Western blot data. In particular those siRNAs that induced an 80% or greater inhibition of MEN1 RNA generated the greatest decrease in protein levels, and the two siRNAs that showed the least inhibition of expression at the RNA level showed the highest levels of protein (MEN1-3 and MEN1-9).

The results of quantitative RT-PCR analyses on p53 and p21 revealed variations in expression that generally correlated with the Western blot data, but the magnitude of the difference observed between mRNA and protein were not always equal (Fig. 1c). For example, despite significant and consistent increases in p53/p21 levels observed on Western blot for MEN1-siRNAs 7 and 8, we detected a <2-fold change at the mRNA level. In another example, MEN1-6 reduced mRNA levels by nearly 5-fold, but the reduction at the protein level was relatively minor. Qualitative differences were detected as well. MEN1-8, for example, had little or no effect on p21 mRNA levels, but the corresponding Western blots showed an increase in p21 protein (compare Fig. 1 a and b).

MEN1-siRNAs Do Not Activate an IFN Response. siRNAs are believed to be too short to activate the nonspecific antiviral responses against dsRNA molecules, but a recent report showing that short hairpin RNAs (shRNAs) can activate at least one of these responses (16) prompted us to test whether chemically synthesized siRNAs could also induce this response. By quantitative RT-PCR analysis, we measured the expression of OAS1, a classic IFN target gene that is induced >50-fold on activation of an IFN response (16). Relative to mock transfected cells, the mRNA levels of OAS1 in siRNA-transfected cells varied between 0.5- and 3-fold, far below those reported on activation of an IFN response (Fig. 2).

Fig. 2.
Quantitative RT-PCR analysis of OAS1. OAS1 induction by all 13 siRNAs is within 3-fold, which is far less than that reported on activation of an IFN response.

Computational Analyses of Men-1 siRNAs Reveal a Multitude of Potential Partial Sequence Homologies for Other Targets. The sequence requirements of a miRNA are less stringent than those for siRNAs, as exemplified by the Caenorhabditis elegans miRNAs, let-7 and lin-4, where the alignment of the miRNA with its target sequence is inexact, resulting in small areas of mismatch that should generate structural perturbations (e.g., “bubbles” or “bulges”) of both the miRNA and the transcript, so as to induce a repression of protein translation. In a computational approach analyzing known miRNAs in Drosophila, it was found that the 5′ ends of many miRNAs are perfectly complementary to an 8-nt core sequence motif found in the 3′ UTR region of putative target transcripts. In some instances, complementarity did not extend beyond the core motif, suggesting that perfect homology between the first eight nucleotides at the 5′ end of an miRNA and its 3′ UTR target may be all that is necessary for translational repression (17). Such interactions as G:U “wobble” base pairing between RNAs may reduce stringency even further. To investigate whether the off-target effects on p53 and p21 we observe could be due to siRNAs functioning as miRNAs, we designed four different algorithms to search for sequence interactions between siRNAs and the 3′ UTR of mRNAs cataloged within the RefSeq database of human mRNAs.

Our most stringent algorithm (A1) was designed to search for targets with a minimum of 8 complementary nucleotides at both ends of either siRNA strand, allowing for a nonhomologous central bulge. Using these parameters, only one of our oligos had a match to a 3′ UTR (Table 1, A1). This match was between the antisense strand of MEN1-siRNA 7 (one of the siRNAs consistently associated with up-regulation of both p53 and p21) and a predicted ORF of an unknown gene on chromosome 1 (NM_030806.1). Reducing the stringency to allow for G:U “wobble” base pairing in the second algorithm (A2) identified many more targets for each strand of each siRNA (Table 1, A2). Neither algorithm identified p53 or p21 as potential targets of our siRNAs, suggesting that the variations in p53 and p21 expression are not due to direct interactions with the siRNAs. Some of the candidates we identified encode genes of unknown function, making it difficult to link them to p53 and/or p21 pathways. Others are reported to function in a variety of different signaling pathways that, if silenced, could indirectly mediate the observed effects on p53 and/or p21 (Table 2, which is published as supporting information on the PNAS web site). Virtually all of the candidate proteins lack antibodies, thereby making it difficult to validate their expression status at the protein level.

Based on the miRNA findings in Drosophila described above, we designed a third algorithm (A3) to find matches between 3′ UTRs and the first eight nucleotides of the 5′ ends of either siRNA strand. Our final algorithm (A4) was similar but searched for 10 complementary nucleotide matches at the 5′ ends, allowing for G:U base pairing. Given these low degrees of sequence stringency, it was perhaps not surprising that all 10 MEN1-siRNAs had target-binding sites in the 3′ UTRs of multiple mRNA transcripts (Table 1, A3 and A4). siRNAs MEN1-1 and MEN1-2 differ by only one additional nucleotide on the 5′ end of the antisense strand. siRNA MEN1-9 differs from MEN1-1 by one nucleotide at the 5′ end and two nucleotides at the 3′ end (plus a noncognate dTdT overhang). Despite these relatively minor sequence differences, we saw dramatic differences in the number of hits these siRNAs generated with our algorithms, illustrating the degree of care required in analyzing siRNAs with even minor sequence differences.


The gene for multiple endocrine neoplasia, type I (MEN1) encodes a tumor suppressor of unknown function called menin. Menin is reported to interact with a variety of different proteins (1824), but to date no direct functional relationship has been established between menin and p53 or p21. We investigated the specificity of siRNA-mediated RNAi after independent efforts to explore the function of the gene for multiple endocrine neoplasia, type 1 (MEN1) showed inconsistent changes in the levels of p53 and p21. We rationalized that different siRNAs targeted to the same gene, but with different sequences, would yield identical effects on p53 and p21 if siRNA-mediated RNAi is truly specific. Ten different siRNAs corresponding to MEN1, all capable of inhibiting MEN1 gene expression at the RNA and protein level, were tested for their effects on p21 and p53 by Western blot analysis. Unexpectedly, we detected significant and divergent changes in the levels of p53 and p21 when the 10 different MEN1 siRNAs were assessed in parallel. Protein and mRNA levels generally correlated, although the magnitude of the changes was not always equal. The observed effects on p53 and p21 were not related to menin silencing and therefore do not define a functional relationship between menin and either p53 or p21. Titration of the siRNA reduced the silencing of the MEN1 target but did not completely abolish the effects on p53 and p21 induced by some of the siRNAs. Resynthesis of several of the siRNAs reproduced the same effects, indicating that this was not due to a synthesis error or contaminant. These effects were also independent of the lipid agent used. Significant changes in p53/p21 expression were also observed in CaSki, SiHa, and MCF7 cells that were broadly similar for each of the siRNAs studied, suggesting that the effects were not limited to HeLa cells and were siRNA sequence-dependent. On the basis of analysis of the expression levels of the 2′-5′ oligoadenylate synthetase gene, we saw no indication of an induction of an IFN-associated response to any of the siRNAs under study. Together, our data suggest that siRNAs can induce nonspecific but sequence-dependent effects, presumably by acting on other unknown targets. As p53 and p21 are activated in response to a variety of cellular stresses including DNA damage, oxidative stress, excessive mitogenic stimuli, and activation of interferons, it is difficult to identify the interactions underlying the effects observed.

Studies have shown that a single mismatch between an siRNA and its target mRNA abrogates silencing (12). However, studies that have used microarray technologies to investigate the specificity of siRNAs have arrived at somewhat contradictory conclusions (13), and there is no consensus in the RNAi field regarding whether or not an siRNA can silence a partially matched target in mammalian cells. In one study (3), global gene expression profiles were compared between HEK293 cells transfected with two different siRNAs directed against GFP that was transiently or stably expressed. Expression profiles of cells transfected with two control siRNAs with scrambled sequences were also compared. Variations in gene expression were detected, but the magnitude of these changes was small (<2-fold) and attributed to experimental noise rather than off-target effects from the siRNA. In a separate, perhaps more comprehensive study, expression profiles were examined in cells transfected with 16 siRNAs to one gene and 8 siRNAs to another gene. Expression profiles varied widely, to the extent that unique expression signatures could be distinguished for each individual siRNA. Furthermore, the authors reported reductions in the levels of mRNA transcripts containing as few as 11 contiguous nucleotide matches to the siRNA (1).

Mature miRNAs are single-stranded RNA molecules of 21–22 nt that function in translational repression by binding to the 3′ UTR of a mRNA. More than 100 miRNAs have now been identified in worms, flies, and humans (49), but the mechanism of their action is poorly understood. Mounting evidence suggests that miRNAs, like siRNAs, are processed by Dicer and are incorporated into a protein complex known as the RNA-induced silencing complex, or RISC (2528). It is believed that the extent of sequence homology between the antisense strand of the siRNA and the single strand or the mature miRNA and its target determine whether the RNA target is degraded or translational repression occurs. In model systems, RNA species with perfect matches to a target will behave like a siRNA and degrade the cognate message, whereas RNAs with partial matches will function as miRNAs and repress translation (10). Recent data showing that human miRNAs with fully complementary mRNA targets can induce mRNA degradation further support this hypothesis (29). The sequence requirements for an miRNA to function in translational repression are thus far less stringent than those for siRNAs. In fact, the 5′ ends of some naturally occurring miRNAs contain as few as seven nucleotides of sequence that are complementary to the 3′ UTR of target mRNAs (17). Such interactions as G:U “wobble” base pairing between RNAs may reduce stringency even further. Based on these predictions, we designed a series of algorithms designed to search for low-stringency sequence interactions between siRNAs and the 3′ UTR of known human mRNAs. We identified a multitude of potential targets that may be translationally repressed because of an off-target miRNA effect by these siRNAs. Although a comprehensive analysis of the mRNA and corresponding protein levels of these targets is necessary to functionally prove a miRNA effect, the overwhelmingly large number of candidate targets that we detected invites speculation that the effects observed here are due to off-target translational repression.

Whatever mechanism is underlying the effects we observe, these results further emphasize the need for careful validation of downstream effects mediated by siRNAs. Studying RNA expression alone may seriously underestimate off-target effects. Once the rules for siRNA and miRNA sequence context are better defined experimentally, improved computational resources will be needed to aid in the design of siRNAs, to minimize the potential for off-target interactions. These issues should be more fully addressed before the uncritical use of RNAi technology on a large scale for functional genomics studies and for therapeutic purposes.

Supplementary Material

Supporting Table:


We thank Greg Crawford and Mike Erdos for experimental advice over the course of this project.


Abbreviations: miRNA, micro-RNA; RNAi, RNA interference; siRNA, short interfering RNA.


1. Jackson, A. L., Bartz, S. R., Schelter, J., Kobayashi, S. V., Burchard, J., Mao, M., Li, B., Cavet, G. & Linsley, P. S. (2003) Nat. Biotechnol. 21, 635–637. [PubMed]
2. Semizarov, D., Frost, L., Sarthy, A., Kroeger, P., Halbert, D. N. & Fesik, S. W. (2003) Proc. Natl. Acad. Sci. USA 100, 6347–6352. [PMC free article] [PubMed]
3. Chi, J. T., Chang, H. Y., Wang, N. N., Chang, D. S., Dunphy, N. & Brown, P. O. (2003) Proc. Natl. Acad. Sci. USA 100, 6343–6346. [PMC free article] [PubMed]
4. Lagos-Quintana, M., Rauhut, R., Meyer, J., Borkhardt, A. & Tuschl, T. (2003) RNA 9, 175–179. [PMC free article] [PubMed]
5. Lagos-Quintana, M., Rauhut, R., Yalcin, A., Meyer, J., Lendeckel, W. & Tuschl, T. (2002) Curr. Biol. 12, 735–739. [PubMed]
6. Lagos-Quintana, M., Rauhut, R., Lendeckel, W. & Tuschl, T. (2001) Science 294, 853–858. [PubMed]
7. Lim, L. P., Lau, N. C., Weinstein, E. G., Abdelhakim, A., Yekta, S., Rhoades, M. W., Burge, C. B. & Bartel, D. P. (2003) Genes Dev. 17, 991–1008. [PMC free article] [PubMed]
8. Lau, N. C., Lim, L. P., Weinstein, E. G. & Bartel, D. P. (2001) Science 294, 858–862. [PubMed]
9. Lee, R. C. & Ambros, V. (2001) Science 294, 862–864. [PubMed]
10. Doench, J. G., Petersen, C. P. & Sharp, P. A. (2003) Genes Dev. 17, 438–442. [PMC free article] [PubMed]
11. Elbashir, S. M., Harborth, J., Lendeckel, W., Yalcin, A., Weber, K. & Tuschl, T. (2001) Nature 411, 494–498. [PubMed]
12. Elbashir, S. M., Martinez, J., Patkaniowska, A., Lendeckel, W. & Tuschl, T. (2001) EMBO J. 20, 6877–6888. [PMC free article] [PubMed]
13. Caplen, N. J., Parrish, S., Imani, F., Fire, A. & Morgan, R. A. (2001) Proc. Natl. Acad. Sci. USA 98, 9742–9747. [PMC free article] [PubMed]
14. Guru, S. C., Goldsmith, P. K., Burns, A. L., Marx, S. J., Spiegel, A. M., Collins, F. S. & Chandrasekharappa, S. C. (1998) Proc. Natl. Acad. Sci. USA 95, 1630–1634. [PMC free article] [PubMed]
15. Anonymous (2003) Nat. Cell Biol. 5, 489–490. [PubMed]
16. Bridge, A. J., Pebernard, S., Ducraux, A., Nicoulaz, A. L. & Iggo, R. (2003) Nat. Genet. 34, 263–264. [PubMed]
17. Lai, E. C. (2002) Nat. Genet. 30, 363–364. [PubMed]
18. Lemmens, I. H., Forsberg, L., Pannett, A. A., Meyen, E., Piehl, F., Turner, J. J., Van de Ven, W. J., Thakker, R. V., Larsson, C. & Kas, K. (2001) Biochem. Biophys. Res. Commun. 286, 426–431. [PubMed]
19. Kaji, H., Canaff, L., Lebrun, J. J., Goltzman, D. & Hendy, G. N. (2001) Proc. Natl. Acad. Sci. USA 98, 3837–3842. [PMC free article] [PubMed]
20. Agarwal, S. K., Guru, S. C., Heppner, C., Erdos, M. R., Collins, R. M., Park, S. Y., Saggar, S., Chandrasekharappa, S. C., Collins, F. S., Spiegel, A. M., et al. (1999) Cell 96, 143–152. [PubMed]
21. Ohkura, N., Kishi, M., Tsukada, T. & Yamaguchi, K. (2001) Biochem. Biophys. Res. Commun. 282, 1206–1210. [PubMed]
22. Heppner, C., Bilimoria, K. Y., Agarwal, S. K., Kester, M., Whitty, L. J., Guru, S. C., Chandrasekharappa, S. C., Collins, F. S., Spiegel, A. M., Marx, S. J. & Burns, A. L. (2001) Oncogene 20, 4917–4925. [PubMed]
23. Sukhodolets, K. E., Hickman, A. B., Agarwal, S. K., Sukhodolets, M. V., Obungu, V. H., Novotny, E. A., Crabtree, J. S., Chandrasekharappa, S. C., Collins, F. S., Spiegel, A. M., et al. (2003) Mol. Cell. Biol. 23, 493–509. [PMC free article] [PubMed]
24. Obungu, V. H., Lee Burns, A., Agarwal, S. K., Chandrasekharapa, S. C., Adelstein, R. S. & Marx, S. J. (2003) Oncogene 22, 6347–6358. [PubMed]
25. Mourelatos, Z., Dostie, J., Paushkin, S., Sharma, A., Charroux, B., Abel, L., Rappsilber, J., Mann, M. & Dreyfuss, G. (2002) Genes Dev. 16, 720–728. [PMC free article] [PubMed]
26. Ketting, R. F., Fischer, S. E., Bernstein, E., Sijen, T., Hannon, G. J. & Plasterk, R. H. (2001) Genes Dev. 15, 2654–2659. [PMC free article] [PubMed]
27. Grishok, A., Pasquinelli, A. E., Conte, D., Li, N., Parrish, S., Ha, I., Baillie, D. L., Fire, A., Ruvkun, G. & Mello, C. C. (2001) Cell 106, 23–34. [PubMed]
28. Hutvagner, G., McLachlan, J., Pasquinelli, A. E., Balint, E., Tuschl, T. & Zamore, P. D. (2001) Science 293, 834–838. [PubMed]
29. Zeng, Y., Yi, R. & Cullen, B. R. (2003) Proc. Natl. Acad. Sci. USA 100, 9779–9784. [PMC free article] [PubMed]

Articles from Proceedings of the National Academy of Sciences of the United States of America are provided here courtesy of National Academy of Sciences
PubReader format: click here to try


Related citations in PubMed

See reviews...See all...

Cited by other articles in PMC

See all...


Recent Activity

Your browsing activity is empty.

Activity recording is turned off.

Turn recording back on

See more...