Abstract: MicroRNAs (miRNAs) play a pivotal role in the regulation of genes involved in diverse processes such as development, differentiation, and cellular growth control. Recently, many viral-encoded miRNAs have been discovered, for the most part in viruses transcribed from double-stranded DNA genomes. As with their cellular counterparts, the functions of most viral-derived miRNAs are unknown; however, functions have been documented or proposed for viral miRNAs from three different viral families—herpesviruses, polyomaviruses, and retroviruses. Several virus-encoded miRNAs have unique aspects to their biogenesis, such as the polymerase that transcribes them or their location within the precursor transcript. Additionally, viral interactions with cellular miRNAs have also been identified, and these have substantially expanded our appreciation of miRNA functions. MicroRNAs (miRNAs) play a pivotal role in the regulation of genes involved in diverse processes such as development, differentiation, and cellular growth control. Recently, many viral-encoded miRNAs have been discovered, for the most part in viruses transcribed from double-stranded DNA genomes. As with their cellular counterparts, the functions of most viral-derived miRNAs are unknown; however, functions have been documented or proposed for viral miRNAs from three different viral families—herpesviruses, polyomaviruses, and retroviruses. Several virus-encoded miRNAs have unique aspects to their biogenesis, such as the polymerase that transcribes them or their location within the precursor transcript. Additionally, viral interactions with cellular miRNAs have also been identified, and these have substantially expanded our appreciation of miRNA functions. MicroRNAs (miRNAs) are small (approximately 22 nt) RNAs that regulate gene expression by a variety of mechanisms. Initially discovered in C. elegans, they are now known to be widespread in nature. Thus far, over 1000 miRNAs have been identified, with greater than 320 found in humans (MiRBase) (Griffiths-Jones, 2004Griffiths-Jones S. Nucleic Acids Res. 2004; 32: D109-D111Crossref PubMed Google Scholar). Recently, based on the technique of phylogenetic shadowing, it has been suggested that the actual number of human miRNAs could be as high as 1000 (Berezikov et al., 2005Berezikov E. Guryev V. van de Belt J. Wienholds E. Plasterk R.H. Cuppen E. Cell. 2005; 120: 21-24Abstract Full Text Full Text PDF PubMed Scopus (984) Google Scholar). miRNAs are predicted to play a pivotal role in the regulation of many genes, especially those at nodes of signaling pathways involved in such processes as development and growth control. It therefore comes as no surprise that viruses, which typically employ many components of the host gene expression machinery, also encode miRNAs. Potential functions have been suggested for several viral miRNAs; in addition, host-encoded miRNAs have recently been described that can modulate viral replication via interaction with target sites in viral transcripts. It seems certain that many new virus-encoded miRNAs will be identified in the coming years, and equally certain that with them will come new insights into miRNA biogenesis and function. Here we review what has been learned to date from the study of such miRNAs. miRNAs are derived from larger RNA precursors by the sequential action of two ribonucleases of the RNase III family, known as Drosha and Dicer (Kim, 2005Kim V.N. Nat. Rev. Mol. Cell Biol. 2005; 6: 376-385Crossref PubMed Scopus (1872) Google Scholar). The latter enzyme was discovered as part of the RNA interference (RNAi) machinery, which generates small interfering RNAs (siRNAs) from cytoplasmic RNA duplexes. In fact, the late steps in miRNA generation share many commonalties with siRNA biogenesis, but the nature and location of the early steps differ substantially. miRNAs are generated by excision from a hairpin structure which is, in turn, typically derived from a longer Pol II transcript (typically hundreds to several thousands of nucleotides) called a pri-miRNA (see Figure 1). In the nucleus, the pri-miRNA is cleaved by Drosha into an approximately 60–80 nt imperfect hairpin called a pre-miRNA, which in turn is bound and exported from the nucleus by the complex of the karyopherin exportin 5 and Ran (Bohnsack et al., 2004Bohnsack M.T. Czaplinski K. Gorlich D. RNA. 2004; 10: 185-191Crossref PubMed Scopus (978) Google Scholar, Lund et al., 2004Lund E. Guttinger S. Calado A. Dahlberg J.E. Kutay U. Science. 2004; 303 (Published online November 20, 2003): 95-98https://doi.org/10.1126/science.1090599Crossref PubMed Scopus (1939) Google Scholar, Yi et al., 2003Yi R. Qin Y. Macara I.G. Cullen B.R. Genes Dev. 2003; 17 (Published online December 17, 2003): 3011-3016https://doi.org/10.1101/gad.1158803Crossref PubMed Scopus (2058) Google Scholar). In the cytoplasm, the pre-miRNA is then recognized and cleaved by Dicer into a (presumably transient) partially dsRNA structure that is then unwound, leaving one strand energetically favored to enter the multiprotein RNA-induced silencing complex (RISC). RISC then directs either cleavage or translational inhibition of its target mRNA, depending on the degree of complementarity between the RISC bound miRNA and target mRNA. Perfect complementarity generally results in cleavage; imperfect complementarity usually leads to impaired translation, most often when the target sequences are located in the 3′UTR of the mRNA. Diagram of miRNA biogenesis and activity. The outer rectangular boxes highlight some miRNA properties that have been identified in viral systems. Prediction of miRNA targets is still an imperfect science. It is known that pairing of nucleotides 2–8 from the 5′ end of the miRNA (the so-called "seed" region) with the target is especially important for target recognition; mutational disruption of this region can ablate miRNA activity. But there is more to target recognition than just seed complementarity, since any given 7-mer would have thousands of complements in the mammalian genome. Progress has been made by using informatic approaches that impose additional requirements for evolutionary conservation and localization of the target in the 3′UTR, but these still indicate that thousands of human genes are potential miRNA targets. Related to this, it was recently shown that, with overexpression of individual miRNAs, the levels of 100–200 transcripts promptly declined, as judged by array-based expression profiling (Lim et al., 2005Lim L.P. Lau N.C. Garrett-Engele P. Grimson A. Schelter J.M. Castle J. Bartel D.P. Linsley P.S. Johnson J.M. Nature. 2005; 433 (Published online January 30, 2005): 769-773https://doi.org/10.1038/nature03315Crossref PubMed Scopus (3786) Google Scholar). Since many of the downregulated mRNAs had seed complementarity at their 3′UTRs, it is unlikely that this result was due solely to secondary effects of inhibition of a small number of directly targeted transcription factors. However, details of the actual mechanisms underlying this regulation remain unknown. Clearly, miRNA-based regulation has the potential to be staggeringly complex. Given the important role that miRNAs play in gene regulation, it was a safe bet that viruses would employ them to modulate both their own gene expression and that of their host cells. There are several ways in which virus-infected cells have been screened for miRNAs. Most studies begin with bioinformatic analyses aimed at identifying stem-loop structures compatible with pre-miRNAs. Experimental screening for miRNAs usually proceeds via cDNA cloning strategies originally developed for cellular miRNAs, with high-throughput sequencing of large numbers of the resulting clones. Northern blotting of total cellular RNA with the appropriate sequence from the clones provides additional confirmation. Extensive cDNA cloning studies by the Tuschl group across many families of RNA viruses have failed to identify miRNAs from viruses with RNA genomes (Pfeffer et al., 2005Pfeffer S. Sewer A. Lagos-Quintana M. Sheridan R. Sander C. Grasser F.A. van Dyk L.F. Ho C.K. Shuman S. Chien M. et al.Nat. Methods. 2005; 2 (Published online February 16, 2005): 269-276https://doi.org/10.1038/nmeth746Crossref PubMed Scopus (929) Google Scholar). This negative finding is subject to all the usual interpretive caveats of negative results but is consistent with the prominent role of the DNA-dependent RNA polymerase II in the biogenesis of pri-miRNAs. Of course, it remains formally possible to envision Drosha acting on RNA templates produced by viral RNA-dependent RNA polymerases, especially for virus families (e.g., myxoviruses) in which genomic replication or transcription occurs in the nucleus (where Drosha resides). For viruses whose RNA-based replication cycle is cytoplasmic (e.g., cornonaviruses, picornaviruses), however, segregation away from the Drosha machinery provides an additional theoretical barrier to miRNA biogenesis. To date, miRNAs have been identified in the following classes of viruses: Herpesviruses. Herpesviruses are large, enveloped viruses with dsDNA genomes that typically encode 100–200 genes. There are eight human herpesviruses, all of them linked to important disease syndromes in man. A defining feature of herpesviral biology is the presence of two alternative genetic lifestyles—a cryptic or latent infection and a patent, or lytic, replicative cycle. During latency, only a handful of viral genes are expressed, and no viral progeny are produced; the viral genome is persistently maintained over many cell generations as a low copy number nuclear plasmid. Lytic replication is characterized by the temporally regulated expression of most of the viral genes, with extensive viral DNA replication, cell death, and release of infectious progeny. Lytic replication can be evoked from latency, a phenomenon that accounts for the relapsing nature of many herpesviral disorders. Herpesviruses can be classified into three subfamilies (α, β, or γ) based on sequence relatedness and virus biology. To date, three members of the γ (lymphotropic) subfamily—EBV and KSHV of humans and MHV68 of mice—and one β-herpesvirus (HCMV) have been shown to encode miRNAs; additionally, two α viruses, HSV-1 and -2, have been predicted but not experimentally proven to encode miRNAs (Cai et al., 2005Cai X. Lu S. Zhang Z. Gonzalez C.M. Damania B. Cullen B.R. Proc. Natl. Acad. Sci. USA. 2005; 102 (Published online March 30, 2005): 5570-5575https://doi.org/10.1073/pnas.0408192102Crossref PubMed Scopus (484) Google Scholar, Pfeffer et al., 2004Pfeffer S. Zavolan M. Grasser F.A. Chien M. Russo J.J. Ju J. John B. Enright A.J. Marks D. Sander C. Tuschl T. Science. 2004; 304: 734-736Crossref PubMed Scopus (1226) Google Scholar, Pfeffer et al., 2005Pfeffer S. Sewer A. Lagos-Quintana M. Sheridan R. Sander C. Grasser F.A. van Dyk L.F. Ho C.K. Shuman S. Chien M. et al.Nat. Methods. 2005; 2 (Published online February 16, 2005): 269-276https://doi.org/10.1038/nmeth746Crossref PubMed Scopus (929) Google Scholar, Samols et al., 2005Samols M.A. Hu J. Skalsky R.L. Renne R. J. Virol. 2005; 79: 9301-9305Crossref PubMed Scopus (342) Google Scholar). None of these miRNAs appear to be conserved among the different viruses (Pfeffer et al., 2005Pfeffer S. Sewer A. Lagos-Quintana M. Sheridan R. Sander C. Grasser F.A. van Dyk L.F. Ho C.K. Shuman S. Chien M. et al.Nat. Methods. 2005; 2 (Published online February 16, 2005): 269-276https://doi.org/10.1038/nmeth746Crossref PubMed Scopus (929) Google Scholar). EBV, causative agent of infectious mononucleosis and etiologically linked to several malignant lymphomas, was the first virus demonstrated to encode miRNAs (Pfeffer et al., 2004Pfeffer S. Zavolan M. Grasser F.A. Chien M. Russo J.J. Ju J. John B. Enright A.J. Marks D. Sander C. Tuschl T. Science. 2004; 304: 734-736Crossref PubMed Scopus (1226) Google Scholar). Cloning from a B cell line latently infected with EBV identified five miRNAs. The function of the EBV miRNAs has not been established, but computational predictions suggest that some of these miRNAs may target chemokines, cytokines, and apoptotic and cell growth control genes such as p53. Additionally, one of the miRNAs (mir-BART2) apparently has a viral target (Table 1): it is antisense to a region in a lytic mRNA (BALF5) encoding the viral DNA polymerase. Because it is fully complementary to that mRNA, it would be predicted to cleave the transcript. Since the experiment that identified the miRNA was conducted in a latently infected cell, such a cleavage product would not be expected to be found in that context. But, strikingly, earlier cDNA cloning of lytic-cycle transcripts for the EBV DNA polymerase detected an "aberrant" mRNA whose sequence had undergone a rearrangement precisely at the site of complementarity to the miRNA (Furnari et al., 1993Furnari F.B. Adams M.D. Pagano J.S. Proc. Natl. Acad. Sci. USA. 1993; 90: 378-382Crossref PubMed Scopus (46) Google Scholar)! It seems highly likely that cleavage of the transcript did indeed occur in vivo—with subsequent repair/rescue of the cleaved transcript by an as-yet-uncertain mechanism. Less clear is what the biological role of this cleavage might be. The cleaved and modified RNA remains fairly abundant, is polyadenylated, and retains its full coding potential. Moreover, infected cells also produce a second mRNA for polymerase that is unaffected by this miRNA. So early notions that the miRNA might extinguish polymerase expression and thereby block lytic replication seem unlikely, and its biological role remains something of an enigma. KSHV is the causative agent of Kaposi's sarcoma (an inflammatory and proliferative lesion of the endothelium) and is also linked to several rare lymphoproliferative syndromes. Several groups, using traditional RACE-like cloning (Cai et al., 2005Cai X. Lu S. Zhang Z. Gonzalez C.M. Damania B. Cullen B.R. Proc. Natl. Acad. Sci. USA. 2005; 102 (Published online March 30, 2005): 5570-5575https://doi.org/10.1073/pnas.0408192102Crossref PubMed Scopus (484) Google Scholar, Pfeffer et al., 2005Pfeffer S. Sewer A. Lagos-Quintana M. Sheridan R. Sander C. Grasser F.A. van Dyk L.F. Ho C.K. Shuman S. Chien M. et al.Nat. Methods. 2005; 2 (Published online February 16, 2005): 269-276https://doi.org/10.1038/nmeth746Crossref PubMed Scopus (929) Google Scholar, Samols et al., 2005Samols M.A. Hu J. Skalsky R.L. Renne R. J. Virol. 2005; 79: 9301-9305Crossref PubMed Scopus (342) Google Scholar) have identified miRNAs in KSHV. There is broad agreement among these studies that a majority of the KSHV miRNAs are encoded in a single 4.5 kb region of the genome. Most derive from a noncoding region that is represented in an intron of a latent mRNA (Li et al., 2002Li H. Komatsu T. Dezube B.J. Kaye K.M. J. Virol. 2002; 76: 11880-11888Crossref PubMed Scopus (60) Google Scholar). Computational predictions suggest that some of these miRNAs may have cellular targets involved in such activities as apoptosis, signaling, and B cell regulation (Cai et al., 2005Cai X. Lu S. Zhang Z. Gonzalez C.M. Damania B. Cullen B.R. Proc. Natl. Acad. Sci. USA. 2005; 102 (Published online March 30, 2005): 5570-5575https://doi.org/10.1073/pnas.0408192102Crossref PubMed Scopus (484) Google Scholar), which, if proven, would have important implications for KSHV-associated disease. Two of the KSHV miRNAs map within the body of a latent transcript that is strongly upregulated during lytic replication. This mRNA encodes a family of proteins (kaposins A, B, and C) with important roles in latency: kaposin A affects cell growth control, and kaposin B enhances cytokine production by infected cells (Kliche et al., 2001Kliche S. Nagel W. Kremmer E. Atzler C. Ege A. Knorr T. Koszinowski U. Kolanus W. Haas J. Mol. Cell. 2001; 7: 833-843Abstract Full Text Full Text PDF PubMed Scopus (117) Google Scholar, McCormick and Ganem, 2005McCormick C. Ganem D. Science. 2005; 307: 739-741Crossref PubMed Scopus (185) Google Scholar). Both proteins are tightly regulated in latency, as overexpression of either is deleterious to cells. Is it possible that Drosha-mediated cis cleavage of the transcript during pri-miRNA formation serves to downregulate the steady-state level of kaposin mRNA during latency? If so, this regulation would have to be nullified during lytic infection, since levels of kaposin mRNA and protein become very high as lytic replication proceeds. MHV68 is a murine γ herpes virus that is much simpler to grow and to genetically manipulate than either EBV or KSHV; however, it is less pathogenic in its native host and does not recapitulate many of the pathologic features of infection by its human counterparts. MHV68 encodes at least nine miRNAs as demonstrated by molecular cloning (Pfeffer et al., 2005Pfeffer S. Sewer A. Lagos-Quintana M. Sheridan R. Sander C. Grasser F.A. van Dyk L.F. Ho C.K. Shuman S. Chien M. et al.Nat. Methods. 2005; 2 (Published online February 16, 2005): 269-276https://doi.org/10.1038/nmeth746Crossref PubMed Scopus (929) Google Scholar). All cluster within a 6 kb region near the M1 terminus of the linear genome. A mutant that deletes a region that includes miR-M1-1, miR-M1-2, miR-M1-3, and miR-M1-4 was completely viable and displayed no demonstrable phenotype in vivo (Simas et al., 1998Simas J.P. Bowden R.J. Paige V. Efstathiou S. J. Gen. Virol. 1998; 79: 149-153PubMed Google Scholar). Thus, any function supplied by these miRNAs must be nonessential or redundant. Perhaps the most noteworthy aspect of the MHV68 miRNAs concerns their biogenesis: at least some of them are thought to be derived from RNA Pol III transcripts (Pfeffer et al., 2005Pfeffer S. Sewer A. Lagos-Quintana M. Sheridan R. Sander C. Grasser F.A. van Dyk L.F. Ho C.K. Shuman S. Chien M. et al.Nat. Methods. 2005; 2 (Published online February 16, 2005): 269-276https://doi.org/10.1038/nmeth746Crossref PubMed Scopus (929) Google Scholar). This suggests that not all mammalian pre-miRNAs need to be processed from Pol II transcripts, and it serves as a reminder that the pathways to miRNAs may be more varied than presently assumed. Human cytomegalovirus (HCMV) infection is often subclinical in normal hosts but is more serious in immunocompromised patients, in whom it can precipitate retinitis, pneumonitis, hepatitis, and colitis; additionally, intrauterine infection can result in birth defects. HCMV encodes for at least nine microRNAs (Pfeffer et al., 2005Pfeffer S. Sewer A. Lagos-Quintana M. Sheridan R. Sander C. Grasser F.A. van Dyk L.F. Ho C.K. Shuman S. Chien M. et al.Nat. Methods. 2005; 2 (Published online February 16, 2005): 269-276https://doi.org/10.1038/nmeth746Crossref PubMed Scopus (929) Google Scholar), three of which are complementary to regions encoding ORFs, suggesting the possibility of miRNA-induced viral gene autoregulation. However, no direct functional studies of these miRNAs have yet been reported. Importantly, all of the HCMV miRNAs were cloned from lytically infected cells, indicating that herpesviral utilization of miRNAs is not solely associated with the latency program. Thus, many more miRNAs may await discovery, since, for the other herpesviral family members where miRNA cloning was attempted, only cells that were predominantly latently infected (and therefore transcribing only restricted portions of the genome) were used as a source for RNA. Polyomaviruses. The polyomavirus family is composed of nonenveloped viruses with small dsDNA genomes. The best-known polyomavirus is simian virus 40 (SV40), which replicates in simian cells and produces a subclinical but persistent infection of monkeys. SV40 has recently been found to encode a set of miRNAs whose in vivo target has been clearly defined (Sullivan et al., 2005Sullivan C.S. Grundhoff A.T. Tevethia S. Pipas J.M. Ganem D. Nature. 2005; 435: 682-686Crossref PubMed Scopus (515) Google Scholar). The miRNAs originate from a single pre-miRNA that maps to the late strand of the viral genome (the strand transcribed after the onset of viral DNA replication) and is found downstream of the late polyadenylation site. Since no other promoter is evident in this region, the pri-miRNA is thought to emanate from the viral late pre-mRNA (the primary transcript produced prior to cleavage and polyadenylation). Processing of the pri-miRNA generates a 57 nt pre-miRNA that displays two unusual features. First, multiple miRNAs are generated from it, deriving from both arms of the pre-miRNA hairpin. Second, the processing of this miRNA appears to be remarkably inefficient, resulting in large amounts of pre-miRNA accumulating in the cell relative to the processed miRNAs. The SV40 miRNAs are the first viral miRNAs whose target is known with certainty (Table 1). The SV40 genome is circular, and the region encoding the miRNAs on the late strand overlaps the body of the viral early mRNAs produced from the opposite strand. The product miRNAs are, therefore, completely complementary to early mRNA and thus would be predicted to cleave that transcript. In fact, the 3′ (polyadenylated) product fragment predicted from such cleavage is readily demonstrable in cells, and mutational disruption of the pre-miRNA results in the disappearance of these fragments, confirming that they result from the action of the miRNAs. The net result is that the expression of the miRNAs reduces the expression of early mRNAs (and their major translation product, the viral large T antigen) at late times in the replicative cycle. T antigen is a dominant target of host cytotoxic T lymphocyte (CTL) responses, and miRNA-mediated downregulation of T antigen synthesis diminishes the susceptibility of infected cells to CTL-mediated lysis in vitro. This suggests that one function of the SV40 miRNAs may be to help evade immune detection during its long persistence in the host; additional functions are certainly conceivable. Whatever the in vivo function of this pre-miRNA, it is presumably important, as it is conserved with three other polyomaviruses: the baboon virus SA12 and the human viruses, BK and JC (Sullivan et al., 2005Sullivan C.S. Grundhoff A.T. Tevethia S. Pipas J.M. Ganem D. Nature. 2005; 435: 682-686Crossref PubMed Scopus (515) Google Scholar). The finding that the miRNAs target a viral transcript does not, in principle, exclude the possibility that it might also target host RNAs as well. However, studies of the related mouse polyoma virus (Py) bear importantly on this issue. No homology exists between Py and SV40 in the region of the SV40 pre-miRNA, and Py is predicted to encode no miRNAs from this region. However, we have recently identified a Py miRNA that emanates from a different region of the late pre-mRNA. Like the SV40 miRNA, it, too, is complementary to early viral mRNA, resulting in cleavage of the mRNAs for both middle and large T antigens (C.S.S., A. Grundhoff, R. Treisman, C.K. Sung, T. Benjamin, and D.G., unpublished data). Thus, these two completely unrelated miRNAs are expressed with the same kinetics and target the same viral mRNA but would be expected to share no cellular targets; this strongly suggests that the principal function of these miRNAs is indeed to downregulate T antigen mRNA. Retroviruses. Retroviruses are small, enveloped RNA viruses that replicate by reverse transcription, depositing a dsDNA copy of their genome into the host chromosome. This integrated provirus serves as the template for viral gene expression via host Pol II-mediated transcription. This makes retroviruses the likeliest of all RNA viruses to encode miRNAs, since all retroviral transcription emanates from host machinery similar to that directing expression of cellular miRNAs. Nonetheless, Tuschl and colleagues did not identify any virally encoded miRNAs from HIV-infected cells by molecular cloning approaches, even after screening over 1500 amplicons (Pfeffer et al., 2005Pfeffer S. Sewer A. Lagos-Quintana M. Sheridan R. Sander C. Grasser F.A. van Dyk L.F. Ho C.K. Shuman S. Chien M. et al.Nat. Methods. 2005; 2 (Published online February 16, 2005): 269-276https://doi.org/10.1038/nmeth746Crossref PubMed Scopus (929) Google Scholar). Fuji and colleagues, however, have reported evidence for a miRNA within the nef region of the genome, as judged by cloning and Northern blot analysis, and proposed a role for it in downregulation of viral transcription based on preliminary studies with reporter genes (Omoto and Fujii, 2005Omoto S. Fujii Y.R. J. Gen. Virol. 2005; 86: 751-755Crossref PubMed Scopus (154) Google Scholar, Omoto et al., 2004Omoto S. Ito M. Tsutsumi Y. Ichikawa Y. Okuyama H. Brisibe E.A. Saksena N.K. Fujii Y.R. Retrovirology. 2004; 1: 44Crossref PubMed Scopus (210) Google Scholar). But the mechanism that could account for such a phenotype is unclear, as is the reason for the discrepancy between the two different cloning studies. Finally, it was recently reported that HIV encodes small RNAs derived from hairpin structures composed of a 19 base pair perfectly complimentary stem and a small loop—a structure somewhat similar to endogenous miRNAs but most reminiscent of exogenous engineered small hairpin RNAs (shRNAs) (Bennasser et al., 2005Bennasser Y. Le S.Y. Benkirane M. Jeang K.T. Immunity. 2005; 22: 607-619Abstract Full Text Full Text PDF PubMed Scopus (370) Google Scholar). It is currently unknown whether or not these hairpins occur endogenously in other organisms, as well as what their the function (if any) is, nor is it understood why they have not been cloned by either of the above-mentioned HIV studies. Further studies will be needed to resolve these issues. It has been less than 2 years since Tuschl and colleagues reported the discovery of the first viral-encoded miRNAs, and it is likely that many others await discovery. Since virus families as diverse as Herpesviridae and Polyomaviridae encode miRNAs, it is likely that many other DNA viruses utilize this strategy. Indeed, computational approaches have already suggested the existence of candidate miRNAs in adenoviruses and poxviruses (Pfeffer et al., 2005Pfeffer S. Sewer A. Lagos-Quintana M. Sheridan R. Sander C. Grasser F.A. van Dyk L.F. Ho C.K. Shuman S. Chien M. et al.Nat. Methods. 2005; 2 (Published online February 16, 2005): 269-276https://doi.org/10.1038/nmeth746Crossref PubMed Scopus (929) Google Scholar). The poxvirus case merits particular scrutiny, since poxviruses replicate in the cytoplasm using their own transcriptional machinery; if these can indeed engender miRNAs, the mechanisms by which they do so would likely present interesting variations on currently known themes. Finally, as noted earlier, most RNA viruses are not predicted (or experimentally found) to encode miRNAs; however, computational studies of yellow fever virus show an attractive potential candidate miRNA sequence in that genome (Pfeffer et al., 2005Pfeffer S. Sewer A. Lagos-Quintana M. Sheridan R. Sander C. Grasser F.A. van Dyk L.F. Ho C.K. Shuman S. Chien M. et al.Nat. Methods. 2005; 2 (Published online February 16, 2005): 269-276https://doi.org/10.1038/nmeth746Crossref PubMed Scopus (929) Google Scholar). Clearly, such candidates will require experimental validation. Since miRNAs play such pivotal roles in the regulation of host gene expression, it would be surprising if viral transcripts were not subjected to regulation by at least some host-encoded miRNAs. Indeed, a striking example of such regulation has recently been uncovered by Sarnow and colleagues (Jopling et al., 2005Jopling C.L. Yi M. Lancaster A.M. Lemon S.M. Sarnow P. Science. 2005; 309: 1577-1581Crossref PubMed Scopus (2010) Google Scholar). Hepatitis C virus (HCV) is a plus-strand RNA virus that is an important cause of chronic liver injury in humans. HCV is extraordinarily difficult to grow in culture, but HCV replicons from selected strains are capable of persistent replication in the liver-derived cell line HuH7. These cells express the liver-specific host miRNA miR-122, and partial complements to the miR-122 seed sequence exist in both the 5′ and 3′ UTRs of HCV. So one might have thought that miR-122 would function as an antiviral factor, targeting HCV RNA for cleavage. But just the reverse turns out to be true—sequestration of miR-122 with sequence-specific 2′-O-methyl antisense oligonucleotides inhibits viral replication, suggesting that miR-122 is a positive regulator of viral replication! Moreover, genetic studies indicate that the effect maps to the target sequence in the 5′ (and not the 3′) UTR. How all this comes to pass is deeply mysterious at present, but miR-122 is not alone in its ability to regulate viral replication. Voinnet et al. recently described another host miRNA, miR-32, that can impair translation of mRNAs bearing target sequences from the primate foamy virus 1 (PFV), a retrovirus of the nonpathogenic spumavirus subfamily (Lecellier et al., 2005Lecellier C.H. Dunoyer P. Arar K. Lehmann-Che J. Eyquem S. Himber C. Saib A. Voinnet O. Science. 2005; 308: 557-560Crossref PubMed Scopus (749) Google Scholar). If miRNAs can modulate viral replication, it might also be expected that viruses would encode factors that modulate miRNA biogenesis or function. In fact, PFV has also been shown to encode a protein, Tas, that can partially relieve the translational inhibition induced by miR-32 on PRV mRNA (Lecellier et al., 2005Lecellier C.H. Dunoyer P. Arar K. Lehmann-Che J. Eyquem S. Himber C. Saib A. Voinnet O. Science. 2005; 308: 557-560Crossref PubMed Scopus (749) Google Scholar). This effect does not appear to be specific for miR-32, as it blocks other miRNA-mediated events; therefore, Tas may be one of a growing family of viral proteins involved in blockage of other small-RNA-induced inhibitory pathways. Gene products (either proteins or small RNAs) that inhibit both RNAi pathways and miRNA generation have also been identified in several other animal virus families (Lu and Cullen, 2004Lu S. Cullen B.R. J. Virol. 2004; 78: 12868-12876Crossref PubMed Scopus (291) Google Scholar, Sullivan and Ganem, 2005Sullivan C.S. Ganem D. J. Virol. 2005; : 7371-7379Crossref PubMed Scopus (134) Google Scholar) and in plant viruses (Voinnet, 2005Voinnet O. Nat. Rev. Genet. 2005; 6: 206-220Crossref PubMed Scopus (629) Google Scholar), though the specific contribution of miRNA (as opposed to siRNA) inhibition to the viral life cycle remains to be established in these cases. Finally, we note that two proteins commonly associated with RISC were originally cloned as binding proteins of the EBV nuclear protein EBNA-2 (GEM3/DP103 [Grundhoff et al., 1999Grundhoff A.T. Kremmer E. Tureci O. Glieden A. Gindorf C. Atz J. Mueller-Lantzsch N. Schubach W.H. Grasser F.A. J. Biol. Chem. 1999; 274: 19136-19144Crossref PubMed Scopus (97) Google Scholar] and p100 SM Tudor [Tong et al., 1995Tong X. Drapkin R. Yalamanchili R. Mosialos G. Kieff E. Mol. Cell. Biol. 1995; 15: 4735-4744Crossref PubMed Scopus (206) Google Scholar]). While both these host proteins have putative nuclear roles other than those involved in cytoplasmic RISC action, and while interaction cloning often produces chaff in equal proportion to wheat, these are clues probably worth revisiting in the contemporary era. In sum, in a very brief period, viruses have already taught us many things about miRNA biogenesis and function. Among these are that mammalian miRNAs can be derived from Pol III transcripts, can emerge from regions distal to polyA sites, can modulate CTL susceptibility, can target 5′ UTR sequences, and can even positively regulate RNA accumulation. Viruses have historically been valued by molecular biologists for their ability to furnish key clues to important pathways of cellular gene expression. With the miRNA and RNAi pathways, history appears to be repeating itself—again.