A Gammaherpesvirus MicroRNA Targets EWSR1 (Ewing Sarcoma Breakpoint Region 1) In Vivo To Promote Latent Infection of Germinal Center B Cells

Gammaherpesviruses, including the human pathogens Epstein-Barr virus (EBV) and Kaposi's sarcoma-associated herpesvirus (KSHV), directly contribute to the genesis of multiple types of malignancies. In vivo, these viruses infect B cells and manipulate B cell biology to establish lifelong infection. To accomplish this, gammaherpesviruses employ an array of gene products, including miRNAs, short noncoding RNAs that bind to and repress protein synthesis from specific target mRNAs. The in vivo relevance of repression of targets of gammaherpesvirus miRNAs remains highly elusive. Here, we identified a murine gammaherpesvirus miRNA as critical for in vivo infection and validated the host mRNA EWSR1 (Ewing sarcoma breakpoint region 1) as the predominant target for this miRNA. Using a novel technology, we demonstrated that repression of EWSR1 was essential for in vivo infection of the critical B cell reservoir. These findings provide the first in vivo demonstration of the significance of repression of a specific host mRNA by a gammaherpesvirus miRNA.

robust replication of the viral DNA genome, abundant expression of a majority of viral genes, and the production of infectious viral particles, latency is characterized by the restricted expression of a small subset of viral genes and the stable maintenance of viral genome as an episome (4)(5)(6). During chronic infection, gammaherpesviruses establish latency in the circulating reservoir of resting memory B cells (3,7). The specific mechanisms by which these viruses establish infection in the memory B cell population in vivo remains poorly understood; however, several key findings have led to a well-accepted model of gammaherpesvirus-driven B cell maturation in which the virus initially infects naive B cells and then, independent of antigen, drives infected cells through germinal center reactions into the memory B cell compartment (3,7). As such, germinal center B cells represent an essential stage of B cell differentiation and are a major target for transformation in gammaherpesvirus-associated lymphomagenesis.
Gammaherpesviruses employ multiple molecular mechanisms to restrict viral gene expression and evade the host immune system, thereby ensuring efficient establishment of latency for the life of the host. One of these strategies is the transcription of noncoding RNAs (ncRNAs) such as microRNAs (miRNAs). miRNAs are a class of small, evolutionarily conserved regulatory RNA molecules approximately 22 nucleotides (nt) in length. Typically, miRNAs posttranscriptionally regulate gene expression by binding with imperfect complementarity to cognate sequences within mRNA target transcripts (8). As a result of miRNA binding within the context of the RNA-induced silencing complex (RISC), mRNA targets are silenced through mechanisms involving mRNA degradation, destabilization, and translational repression (8,9). Since their discovery, a wide range of studies have revealed that mammalian miRNAs are involved in multiple host biological processes, including development and differentiation, cell survival, apoptosis, and immune system regulation (10). Thus, it is not surprising that many viruses, and in particular herpesviruses, have evolved to utilize miRNAs to fine-tune host and viral gene expression, contributing to immune evasion and promoting persistent virus infection (11)(12)(13)(14).
The human gammaherpesviruses EBV and KSHV encode 25 and 12 pre-miRNAs, which generate at least 44 and 25 mature miRNAs, respectively (12,13,(15)(16)(17)(18). Numerous putative targets for EBV and KSHV miRNAs have been identified through bioinformatic analyses of high-throughput sequencing data from RNA isolated by cross-linking immunoprecipitation assays, and more recently, through cross-linking, ligation, and sequencing of hybrids (CLASH) (19)(20)(21). Repression of some of these putative targets during virus infection has been shown to modulate essential cellular pathways in vitro, promoting critical functions such as cell survival and proliferation and dampening host innate immune responses (reviewed in references 14, 22, 23, and 24). Despite this growing knowledge base, the specific in vivo functions of these miRNAs and their targets in the natural host are largely unknown due to the strict species specificity of these viruses.
Murine gammaherpesvirus 68 (MHV68) is a natural pathogen of rodents that is genetically and pathogenically related to EBV and KSHV (25,26). Thus, MHV68 infection of mice provides a robust small animal model for investigating specific molecular and cellular aspects of gammaherpesvirus latency and pathogenesis in vivo (25). Like the human gammaherpesviruses, MHV68 establishes a stable and lifelong latency in B cells and is highly associated with the development of lymphoproliferative disorders and B cell lymphomas (27)(28)(29). MHV68 encodes a cluster of 14 pre-miRNAs, which generate up to 28 mature miRNAs (13,(30)(31)(32)(33). Interestingly, the MHV68 pre-miRNAs are all generated from unique transcripts that encode tRNA-like elements (vtRNAs) with embedded RNA polymerase III promoters upstream of one or two pre-miRNA stemloops (34,35). MHV68 encodes eight of these tRNA-miRNA-encoding RNA (TMER) transcripts (30,36), which range in size from 200 to 250 nt and differ significantly in sequence despite their similar structural features. The TMER transcripts are processed using a noncanonical miRNA biogenesis pathway that utilizes tRNase Z instead of Drosha to yield pre-miRNAs that are subsequently processed to mature miRNAs by Dicer (34,35).
Notably, the MHV68 TMERs, like the EBV EBERs, are among the only transcripts expressed in vivo in latently infected B cells and in infected B cells associated with lymphoproliferative disease and lymphoma (37)(38)(39). Likewise, we and others have demonstrated that MHV68 mature miRNAs are extensively expressed during latent infection (13,(30)(31)(32)(33)(34)(35), suggesting that these TMER-encoded miRNAs may hold essential biological functions during chronic infection. Consistent with this possibility, we have previously demonstrated that an MHV68 mutant deficient in all 14 pre-miRNAs is attenuated for latency establishment in vivo (30). However, which miRNAs and miRNA targets mediate this phenotype remain unclear.
We describe here our efforts to use the MHV68 system to define for the first time the in vivo significance of miRNA-mediated repression of a specific host mRNA target. Using a panel of MHV68 TMER mutants, we determined that TMER5-encoded MHV68 miR-7 is required for efficient establishment of MHV68 latency in vivo and sought to determine the target that mediates this requirement. Using our recently defined set of qCLASHidentified host mRNA targets of MHV68 miRNAs (72) as a starting point, we validated molecular repression of the top miR-7 targets. Among these, the most highly repressed host mRNA transcript was EWSR1, which encodes the RNA binding protein and transcriptional repressor/activator EWSR1/EWS (40)(41)(42). Notably, shRNA-mediated selective repression of EWSR1 in miR-7-deficient viruses fully restored latent infection in vivo, demonstrating a key biological role for miR-7-5p repression of EWSR1. Moreover, miR-7-5p repression of EWSR1 specifically promoted latent infection of germinal center B cells, suggesting that EWSR1 repression may be a key regulatory step for gammaherpesvirus-driven B cell differentiation. This is to our knowledge the first identification of a biological role for a specific gammaherpesvirus miRNA-host mRNA interaction in vivo.

RESULTS
MHV68 miR-7 is required for the establishment of splenic latency in vivo. We have previously determined that the TMER5-derived miRNA mghv-miR-M1-7-5p is expressed in vivo during latency (30); however, whether this or any other TMER5-derived miRNA play a biologically significant role during infection is unknown. To assess this possibility, we generated a mutant virus, MHV68.ΔmiR7.12, which is deficient in both miR-7 and miR-12 pre-miRNA stem-loops ( Fig. 1A; see Tables S1 and S2 in the supplemental material). Mutants were generated on the backbone of MHV68.ORF73␤la, a ␤-lactamase-expressing marker virus that displays wild-type (WT) phenotypes during both lytic and latent infection (43). fibroblasts were infected with MHV68.WT or MHV68.ΔmiR7.12 at an MOI of 0.05 and then overlaid with methylcellulose. At the indicated time points, cells and supernatants were harvested, and viral titers were determined by plaque assay. Values represent the means Ϯ the standard errors of the mean (SEM). Each growth curve was performed in duplicate in each of two independent experiments. Significance was determined using a two-tailed, unpaired t test (*, P Ͻ 0.05). (C) Presence of viral genome in latently infected splenocytes harvested from in vivo samples. Three wild-type B6 mice per sample group per experiment were infected i.n. with 10 4 PFU of MHV68.WT or MHV68.ΔmiR7.12. After 16 days, splenocytes were harvested, pooled, and then subjected to limiting-dilution nested PCR to detect the presence of viral genome. The frequencies of cells harboring viral genome were determined using a Poisson distribution, as indicated by the line at 63.2%. Values represent means Ϯ the SEM of three independent experiments. Significance was determined using a two-tailed, unpaired t test (*, P Ͻ 0.05).
Gammaherpesvirus miRNA Targets EWSR1 In Vivo ® To first determine whether miR-7 or miR-12 contributed to lytic virus infection, we quantified the replication capacity of MHV68.ΔmiR7.12 versus parental WT MHV68.ORF73␤la (MHV68.WT) virus in both in vitro and in vivo assays. For in vitro multistep growth curve assays, NIH 3T12 fibroblasts were infected at a multiplicity of infection (MOI) of 0.05. Samples were then harvested at time points from 0 to 120 h, and virus titers were determined by plaque assay (Fig. 1B). At 48 h postinfection, MHV68.ΔmiR7.12 titers were significantly reduced compared to WT virus. However, titers recovered to nearly WT by later time points, indicating a slight delay in MHV68.ΔmiR7.12 replication. To determine whether this mutant virus displayed a similar replication delay in vivo, wild-type C57BL/6J (B6) mice were infected intranasally (i.n.) with 10 3 PFU of MHV68 or MHV68.ΔmiR7.12, and virus titers were then quantified in lungs at 4 and 8 days postinfection (dpi) (see Fig. S1 in the supplemental material). Although the mean titer of MHV68.ΔmiR7.12 (250 PFU) at 4 dpi was slightly reduced compared to the WT virus titer (420 PFU), this difference was not significant, indicating that any delay in mutant virus replication was minimal in vivo. Together, these results demonstrated that MHV68 TMER5 miRNAs were largely dispensable for lytic virus replication.
To determine whether TMER5-derived miRNAs are required for latent infection, wild-type B6 mice were inoculated i.n. with 10 4 PFU of MHV68.WT or MHV68.ΔmiR7.12. At 16 dpi, the frequency of latently infected cells in the spleen was quantified using a limiting dilution nested PCR assay ( Fig. 1C; see Table S3) (43)(44)(45). In accordance with our previous findings (30), infection with WT virus resulted in the establishment of latency in approximately 1 in 400 splenocytes. In contrast, virus lacking TMER5 miRNAs displayed a significant 3.4-fold reduction in latently infected splenocytes to a frequency of 1 in 1,370. These results demonstrated that the TMER5-derived miRNAs were required for the efficient establishment of latency in vivo.
To define whether the critical function of TMER5-derived miRNAs in latency was mediated by miRNAs derived from pre-miR-7 or pre-miR-12, we generated two other mutant viruses, MHV68.ΔmiR7 and MHV68.ΔmiR12, which carry deletions of single pre-miR-7 or pre-miR-12 stem-loops, respectively ( Fig. 2A; see Tables S1 and S2). Interestingly, in multistep lytic growth curves, both single stem-loop mutant viruses displayed slightly delayed replication kinetics (Fig. 2B) similar to the double stem-loop mutant MHV68.ΔmiR7.12 (Fig. 1B). However, any replication defects were again relatively minor as the delay was quickly overcome at later times postinfection.
In contrast to the comparable replication delays observed for both mutant viruses during lytic infection, the two mutant viruses demonstrated disparate phenotypes Values represent means Ϯ the SEM. Each growth curve was performed in duplicate in each of two independent experiments. Significance was determined using a two-tailed, unpaired t test (*, P Ͻ 0.05). (C) Presence of viral genome in latently infected splenocytes harvested from in vivo samples. Three wild-type B6 mice per sample group per experiment were infected i.n. with 10 4 PFU of MHV68.WT, MHV68.ΔmiR7, or MHV68.ΔmiR12. After 16 days, splenocytes were harvested, pooled, and then subjected to limiting-dilution nested PCR to detect the presence of viral genome. The frequencies of cells harboring viral genome were determined using a Poisson distribution, as indicated by the line at 63.2%. Values represent means Ϯ the SEM of three independent experiments. Significance was determined using a two-tailed, unpaired t test (*, P Ͻ 0.05). during latency. MHV68.ΔmiR7 displayed a reproducible 6.1-fold attenuation to a frequency of 1 in 2,300 ( Fig. 2C; see Table S3), which is similar to the defect observed during MHV68.ΔmiR7.12 infection. Conversely, the ability of MHV68.ΔmiR12 to establish latency (1 in 270) was similar to that of wild-type virus (1 in 380), demonstrating that miRNAs derived from miR-12 are completely dispensable for latent infection. Taken together, these data demonstrate that MHV68 pre-miR-7 is the critical determinant of TMER5 required for the efficient establishment of splenic latency in vivo.
mghv-miR-M1-7-5p targets and represses EWSR1 in vitro and in vivo. The regulatory function of miRNAs is typically mediated through sequence-specific binding to cognate mRNA targets (8). Because the pre-miR-7-deficient virus was attenuated for latency in vivo, we hypothesized that repression of at least one specific mRNA target of a pre-miR-7-derived miRNA must promote latent infection. To molecularly define specific targets of all MHV68 miRNAs, we have previously performed qCLASH on B cells latently infected with MHV68 (72). Within this study, we identified 10 and 15 targets of miR-7-5p and miR-7-3p, respectively (Table S4). However, because miRNA binding to mRNA targets does not necessarily equate to repression of the target transcript, we sought to molecularly validate these miR-7 targets in work described here. To focus these additional studies, we selected the top four most prevalent targets for each miRNA for molecular validation: miR-7-5p targets RGS16 (regulator of G-protein signaling 16), EWSR1, ARHGEF18 (rho/rac guanine nucleotide exchange factor 18), and BIRC5 (baculoviral IAP repeat containing 5), whereas miR-7-3p targets TMEM38B (transmembrane protein 38B), LARS2 (leucyl-tRNA synthetase 2 mitochondrial), RNF138 (ring finger protein 138), and ANAPC7 (anaphase promoting complex subunit 7).
To determine whether the binding of these specific miRNAs to their cognate qCLASH-identified target transcripts resulted in target repression, we quantified endogenous mRNA levels in the presence of individual miRNAs. NIH 3T12 cells were mock transfected or were transfected with miR-7-5p mimic (Fig. 3A) or miR-7-3p mimic (Fig. 3B); 24 h later, the levels of specific host transcripts were quantified by quantitative reverse transcription-PCR (qRT-PCR). Endogenous EWSR1 and ARHGEF18 transcripts were significantly reduced (61 and 40% reductions, respectively) in the presence of miR-7-5p, whereas the expression of BIRC5 transcript was not affected. Unexpectedly, RGS16 expression was increased 145% in the presence of miR-7-5p, suggesting that miR-7-5p may stabilize rather than repress this host transcript. In contrast, of the four miR-7-3p targets tested, only TMEM38B expression was significantly repressed (36% reduction). Thus, of the top miRNA-mRNA target interactions tested, miR-7-5p binding to EWSR1 resulted in the highest degree of target repression.
These results indicated that EWSR1 may be an important target for repression by miR-7-derived miRNAs. This level of repression was mediated by strong sequence complementarity between miR-7-5p and EWSR1, as defined by alignment of the short Ago-associated hybrid sequences recovered in qCLASH experiments (Fig. 3C). Interestingly, as opposed to miRNA binding interactions which frequently occur within the 3= untranslated region (3=UTR) of the target transcript, miR-7-5p bound to the coding sequence of EWSR1. Supporting our observations of strong EWSR1 repression by miR-7-5p, this interaction displayed a high degree of complementarity throughout, including 6 of 7 nt within the seed sequence (nt 2 to 8) of the miRNA. Consistent with the strong repression of endogenous EWSR1 mRNA (Fig. 3A) mediated by this miRNA, miR-7-5p also resulted in a striking repression of both luciferase target transcripts carrying the EWSR1 miRNA binding sequence (Fig. 3D) and endogenous levels of EWSR1 protein (Fig. 3E).
Like EBV and KSHV, MHV68 predominantly establishes latent infection in circulating mature B cells (27,43,(46)(47)(48)(49)(50)(51). To determine whether EWSR1 is specifically repressed in latently infected B cells in vivo, we infected mice with MHV68.H2bYFP, an enhanced yellow fluorescent protein (eYFP)-marked recombinant MHV68 that is phenotypically wild type (46). At 16 dpi, splenocytes were harvested and sorted into infected (CD4 - 4A). After RNA extraction, EWSR1 and control BIRC5 mRNA expression levels were determined by qRT-PCR (Fig. 4B). Compared to YFPnoninfected B cells, the level of EWSR1 transcript was significantly decreased in YFP ϩ latently infected B cells. In contrast, the expression of control BIRC5 transcript, which was not altered by miR-7-5p in vitro (Fig. 3A), was not inhibited in vivo and instead was increased during in vivo virus infection (Fig. 4B). Together, these data demonstrated that EWSR1 is strongly repressed by miR-7-5p both in vitro and in vivo.

Specific repression of EWSR1 promotes MHV68 splenic latency in vivo.
The reduced EWSR1 transcripts in infected B cells in vivo suggested that repression of EWSR1 protein may play an important role in regulating MHV68 latency. Because miR-7 was required for efficient latency establishment in vivo (Fig. 2C), we hypothesized that specific repression of EWSR1 by miR-7-5p promotes MHV68 latency. To test this hypothesis, we sought to provide a mechanistic link between specific targeting of EWSR1 and in vivo latency. Fortunately, the MHV68 TMERs offer a unique means to selectively target individual transcripts in infected cells: it has been previously demonstrated (i) that non-MHV68 miRNAs are processed normally and are fully functional in the context of a TMER unit (34) and (ii) that short hairpin RNAs (shRNAs), which resemble pre-miRNA stem-loops, can generate functional siRNAs when fused downstream of a tRNA (52). Thus, we speculated that direct replacement of the MHV68 pre-miR-7 stem-loop with a sequence-specific shRNA directed against EWSR1 should generate a functional and highly specific anti-EWSR1 siRNA.
To test this possibility, we first designed four EWSR1-specific small interfering RNAs (siRNAs) (EW-siR-1, -2, -3, and -4), and four control siRNAs with scrambled sequences (SC-siR-1, -2, -3, and -4) (Fig. S2A) and then tested their ability to suppress EWSR1. NIH 3T12 cells were transfected with EWSR1-specific or scrambled siRNAs, and Western blotting for EWSR1 protein was performed 48 h after transfection (Fig. S2B). Compared to their respective scrambled siRNAs, EWSR1-specific siR-2, -3, and -4 exhibited potent inhibitory effects on target gene expression. Based on these results, EWSR1-specific siR-2 and -3, and their respective scrambled siRNAs, were chosen for generation of shRNAs for replacement of the pre-miR-7 stem-loop. shRNAs were generated from the siRNA sequence, using the loop sequence "TCAAGAG" to link the siRNA with complementary base pairs on the stem-loop (Table S2 and Fig. S3).
To test whether the EWSR1-specific shRNAs could functionally restore miR-7 activity, we generated viruses carrying the anti-EWSR1 or scrambled shRNAs in place of the TMER5-derived miRNA stem-loops in the MHV68.ΔmiR7.12 virus. As described above (Fig. 2C), TMER5-encoded pre-miR-7 was crucial for the efficient establishment of latency in vivo, but TMER5-encoded pre-miR-12 was completely dispensable for infection. Likewise, the TMER5 double stem-loop mutant virus MHV68.ΔmiR7.12, lacking both pre-miR-7 and pre-miR-12, displayed an attenuated phenotype equivalent to that of the single pre-miR-7 stem-loop mutant virus (Fig. 1C). Thus, to maximize shRNA efficacy in the context of virus infection, we inserted anti-EWSR1 or scrambled shRNAs in place of both TMER5-derived miRNA stem-loops (Table S2). Importantly, in silico Mfold (53) predictions of TMER5 folding indicated that the secondary structure of the TMER5 vtRNA was not affected by shRNA insertions (Fig. 5A).
To verify that TMER5-derived shRNAs retained functional activity, we examined repression of EWSR1 in luciferase assays and in assays for endogenous mRNA and endogenous protein. NIH 3T12 cells were transfected with recombinant pUC57 vectors carrying recombinant shRNA-expressing TMER5s. Notably, TMER5-encoded anti-EWSR1 shRNAs significantly repressed the activity of luciferase transcripts containing a portion of the EWSR1 coding region (nt 1000 to 1539) (Fig. 5B), demonstrating that the shRNAs are functionally active when encoded within TMER5 transcriptional units. Likewise, TMER5-derived anti-EWSR1 shRNAs suppressed endogenous EWSR1 transcripts (Fig. 5C) and endogenous EWSR1 protein (Fig. 5D). As expected, the repression of endogenous EWSR1 protein levels was not as robust from plasmid-expressed, TMER5-derived transcripts, which require expression and processing, as it was for direct transfection of siRNAs. Cumulatively, these findings demonstrated that anti-EWSR1 shRNAs maintained functionality when expressed downstream of vtRNA5 in the TMER5 transcript.
Based on these results, we incorporated the anti-EWSR1 shRNAs or scrambled shRNAs into the MHV68 genome in place of the pre-miR-7 and pre-miR-12 stem-loops to generate the recombinant viruses MHV68.EW.shR and MHV68.SC.shR, respectively (Fig. 5E). To verify that TMER5-derived shRNAs could be functionally expressed in the context of virus infection, NIH 3T12 cells were infected with MHV68.EW.shR or MHV68.SC.shR, and the levels of endogenous EWSR1 mRNA and EWSR1 protein were quantified using qRT-PCR and Western blotting. Importantly, recombinant virus expressing anti-EWSR1 shRNAs significantly repressed endogenous mRNA (Fig. 5F) and protein (Fig. 5G) levels, demonstrating functionality of shRNAs expressed from the virus itself. To determine whether recombinant virus expressing anti-EWSR1 shRNAs specifically repressed EWSR1 transcript in latently infected B cells in vivo, we infected mice with wild-type MHV68, MHV68.EW.shR, or MHV68.SC.shR for 16 days and quantified EWSR1 expression in infected B cells. The phenotypically wild-type parental virus used in these studies (MHV68.ORF73␤la) expresses ␤-lactamase as a C-terminal fusion to the latencyassociated nuclear antigen (mLANA); the use of this marker in combination with the fluorescent ␤-lactamase substrate CCF4-AM facilitates the identification of infected cells from in vivo samples (43). Thus, following spleen harvest, infected B cells were flow cytometrically sorted based on CCF4-AM ϩ CD19 ϩ gating (Fig. 5H). After RNA extraction from sorted cells, EWSR1 mRNA expression was determined by qRT-PCR (Fig. 5I). Notably, B cells infected with mutant virus carrying control scrambled shRNAs in place of pre-miRNA stem-loops displayed increased levels of EWSR1 transcript, providing additional evidence that TMER5-derived miRNAs suppress EWSR1 in vivo. In contrast, the level of EWSR1 in B cells infected with mutant virus carrying anti-EWSR1 shRNAs was reduced to a level below that of wild-type virus, demonstrating effective shRNAmediated repression of EWSR1 in vivo. Together, these results demonstrated that anti-EWSR1 shRNAs expressed downstream of vtRNA5 in the TMER5 transcriptional unit are fully functional during MHV68 infection.
To determine the role of EWSR1 repression during latent infection, we inoculated wild-type B6 mice with MHV68.EW.shR or MHV68.SC.shR, or with parental wild-type virus MHV68.WT, and then quantified features of chronic infection, including spleen weights and the frequency of latently infected cells. As expected, the weights of spleens from mice infected with wild-type virus (137 mg) were significantly more than those of mock-infected mice (88 mg) (Fig. 6A), consistent with lymphocyte expansion in response to virus infection (54). Notably though, while the splenomegaly induced by virus recombinants expressing scrambled shRNAs in place of pre-miR-7 and pre-miR-12 stem-loops was substantially reduced (117 mg) compared to wild-type virus, the expression of anti-EWSR1 shRNAs restored this attenuation (159 mg). Similarly, the virus expressing scrambled shRNAs was significantly attenuated for latency (1 in 1,180) compared to wild-type virus (1 in 470) ( Fig. 6B; see Table S3), a defect that was similar to that observed in mice infected with pre-miR-7-deficient virus ( Fig. 1C; see Table S3). On the other hand, latent infection was fully restored to wild-type levels (1 in 320) in mice infected with viruses expressing anti-EWSR1 shRNAs. This finding was in contrast to the slight delay in lytic replication observed with viruses deficient in pre-miR-7 and Wild-type B6 mice (three per sample group per experiment) were infected i.n. with 10 4 PFU of indicated viruses. After 16 days, splenocytes were harvested, pooled, and then subjected to limiting-dilution nested PCR to detect the presence of viral genome. The frequencies of cells harboring viral genome were determined, exactly as described for Fig. 1C. Values represent the means Ϯ the SEM of three independent experiments. Significance was determined using a two-tailed, unpaired t test (**, P Ͻ 0.01; *, P Ͻ 0.05).
Gammaherpesvirus miRNA Targets EWSR1 In Vivo ® pre-miR-12 (Fig. 1B), which was not rescued by insertion of anti-EWSR1 shRNAs (Fig. S4). Cumulatively, these data demonstrated that repression of the single miRNA target EWSR1 could functionally restore the attenuated latency phenotype of viruses deficient in pre-miR-7. Moreover, these findings demonstrate an essential role for viral miRNAmediated repression of EWSR1 in promoting the establishment of MHV68 latency.

Repression of EWSR1 modulates latent infection of germinal center B cells.
Gammaherpesviruses have evolved multiple mechanisms to manipulate infected B cells, allowing the virus to drive naive B cells, independent of antigen, through the germinal center B cell stage into the memory B cell compartment that serves as the long-term latency reservoir (3,7,25,55). To determine whether EWSR1 repression was required for any specific stage of mature B cell differentiation, we performed multiparametric flow cytometric analysis using staining for virus-positive cells in combination with staining for specific B cell surface markers. Thus, 16 days after infection of wild-type B6 mice with control wild-type virus or viruses expressing scrambled or anti-EWSR1 shRNAs, splenocytes were harvested for B cell subset staining with CCF4-AM and with antibodies directed against the B cell surface markers CD19, GL7, and IgM (Fig. 7A). Staining for virus-positive cells indicated that, similar to results from limiting dilution PCR analyses (Fig. 6B), the overall percentage of virus-positive cells was significantly reduced in mice infected with miR-7-5p-deficient virus expressing scrambled shRNAs (0.23% WT versus 0.03% SC.shR) (Fig. 7B) but was recovered to nearly wild-type levels with virus expressing anti-EWSR1 shRNAs (0.20% EW.shR).
Among cells staining positive for virus, B cell subsets were identified as naive (CD19 ϩ GL7 -IgM ϩ ), germinal center (CD19 ϩ GL7 ϩ ), and memory (CD19 ϩ GL7 -IgM -) based on well-established cell surface markers (30,(56)(57)(58). Although infection of all three subsets of B cells was apparent with each virus (Fig. 7C to E), the vast majority of cells resided in the germinal center B cell compartment. Notably though, while germi-  Fig. 7D) with miR-7-5p-deficient virus expressing scrambled shRNAs was reduced to 57% of the total. Interestingly, the percentage of naive B cells (Fig. 7C) infected with the scrambled shRNA virus was significantly increased compared to wild-type virus (9% WT versus 23% SC.shR), suggesting a compensatory increase in infected naive B cells as a result of decreased germinal center B cell infection. Although infection of memory B cells with the scrambled shRNA virus was also slightly increased, these results were not statistically significant. Together, these data suggested that miR-7-5p may directly contribute to the virus-mediated differentiation of naive B cells into germinal center B cells. Consistent with this possibility, incorporation of anti-EWSR1 shRNAs into the miR-7-5p-deficient virus fully restored both naive B cell (9% WT versus 8% EW.shR) and germinal center B cell (79% WT versus 83% EW.shR) infection to levels equivalent to that of wild-type virus. Collectively, these data demonstrate that the TMER5-encoded mghv-miR-7 promotes the infection of germinal center B cells through targeted repression of the host transcript EWSR1.

DISCUSSION
With the development of new methodologies such as qCLASH, an increasing number of bona fide host mRNA targets for virus miRNAs have been recently defined. However, to date, the biological functions of these miRNA-mRNA interactions in in vivo host-virus infection systems remain almost completely unknown. We present here, for the first time, clear evidence of an in vivo role for a specific gammaherpesvirus miRNA-host mRNA interaction in the context of chronic infection. Through the systematic application of increasingly refined MHV68 mutant viruses, we determined that MHV68 TMER5-encoded pre-miR-7 was essential for establishing latency in vivo. Through in vitro validation of qCLASH-identified targets, we defined EWSR1 as a likely important host mRNA target of the pre-miR-7-derived miRNAs mghv-miR-M1-7-5p or mghv-miR-M1-7-3p. Subsequently, through the use of recombinant viruses expressing scrambled or target-specific shRNAs in place of TMER5-encoded miRNAs, we demonstrated that in vivo repression of EWSR1 completely restored the attenuation of MHV68 TMER5 mutant viruses. Notably, the use of a unique marker for identification of infected cells in vivo revealed a specific defect of miR-7-deficient viruses in germinal center B cells, and repression of EWSR1 completely rescued this defect. Cumulatively, these findings directly implicate mghv-miR-M1-7-5p-mediated repression of the host mRNA target EWSR1 as a key regulatory step in promoting MHV68 germinal center B cell latency in vivo. Establishment of a system for in vivo validation of gammaherpesvirus miRNAmediated repression of host mRNA targets. miRNAs function via sequence-specific but nonperfect binding to complementary sequences on RNA targets. The critical nucleotides for target binding typically lie within the miRNA seed sequence (nt 2 to 7 of the miRNA) (8); however, CLASH-based studies have recently revealed a wide range of other types of complementation between Ago-associated miRNAs and their cognate binding partners (19)(20)(21)72). Moreover, while single miRNAs can bind to multiple target RNAs, the in vivo consequence of repression of individual target RNAs is a function of both the biological relevance of the target and the in situ stoichiometry of the interactions between the miRNA and its particular target RNAs. Thus, defining the biologically relevant targets of individual miRNAs in an in vivo system a priori remains a difficult problem.
Here, we have combined the unique properties of the MHV68 TMERs with highly specific shRNAs to define the biological function of viral miRNA-mediated repression of a host target transcript. Based on previous reports demonstrating effective processing of heterologous miRNAs from the MHV68 TMERs (34) and functional expression and processing of shRNAs placed downstream of tRNAs (52), we speculated that direct replacement of an MHV68 pre-miRNA stem-loop with an shRNA would generate a functional and highly specific siRNA directed against the target transcript of interest. Indeed, substitution of the TMER5-encoded pre-miRNA stem-loops with shRNAs di-rected against EWSR1, the most highly repressed target of a TMER5-encoded miRNA, fully restored attenuation of the TMER5 pre-miRNA mutant virus. These findings demonstrate the utility of this approach, and demonstrate for the first time the in vivo biological relevance of miRNA-mediated repression of a specific host mRNA target.
A biological function for EWSR1 in germinal center B cell latency. EWSR1/EWS is a multifunctional protein that is mostly known for its central role as a fusion protein in the genesis of Ewing's sarcoma. In that scenario, fusion of the N-terminal transcription-activating domain of EWSR1 to the C-terminal DNA-binding domain of the transcription factor FLI-1 (Friend Leukemia Integration 1) generates an oncogenic transcriptional activator, EWS/FLI-1, that drives cell growth, proliferation, and transformation (59,60).
In contrast, the functions of full-length EWSR1 are less fully understood. Native EWSR1 is a ubiquitous transcriptional repressor and RNA binding protein that is expressed in numerous tissues and cell types (61), including pro-/pre-B cells from the bone marrow and mature B cells from the spleen (41). In the context of cellular stress responses, EWSR1 is involved in the regulation of gene expression and alternative splicing of DNA damage response-associated genes (40,42,62). Although its function during virus infections remains unknown, a role is likely dependent upon cell type and virus genera. For example, transcriptomic analyses revealed EWSR1 repression in brain tissue of grouper persistently infected with nervous necrosis virus (63), but EWSR1 expression is required for efficient replication of hepatitis C virus (64).
It is noteworthy that previous work has demonstrated that EWS-deficient mice display a marked defect in pre-B cell development (41); however, a specific function for EWSR1 in circulating mature B cells has not been previously reported. We identified here a particularly prominent defect in germinal center B cell infection in the absence of pre-miR-7, which was rescued by expression of anti-EWSR1 shRNAs, demonstrating the biological significance of EWSR1 repression in this cell type during virus infection. Notably, germinal center B cells are central players in chronic gammaherpesvirus infection, likely playing a critical role in the transition between initial infection of naive B cells and the establishment of long-term infection in the circulating memory B cell compartment. While the transition from naive B cells to memory B cells normally requires specific antigen stimulation of the B cell receptor plus T cell-mediated activation of the costimulatory molecule CD40, gammaherpesviruses have evolved highly regulated mechanisms to drive this process independent of antigen. For example, it is thought that the expression of a restricted subset of EBV genes in naive B cells drives these cells through the germinal center stage before a more restricted program propels their ultimate differentiation to memory B cells (3,7,55). In support of this concept, MHV68 is preferentially detected in germinal center B cells during the establishment phase of latency (43, 46-48, 51, 65). Consistent with the importance of germinal center B cells during gammaherpesvirus infection, Hodgkin's lymphoma arises from EBVinfected B cells that are blocked at the germinal center stage, and Burkitt's lymphoma arises from a germinal center B cell that is entering the memory compartment but is stuck at the point of proliferation (3,66).
The findings described here strongly suggest that repression of EWSR1 is critical for MHV68 infection of, or transition through, the germinal center B cell stage. Whether the restriction of germinal center B cell infection by EWSR1 relates to its activity as an RNA binding protein and/or its function as a transcriptional repressor remains to be tested. It is also plausible that EWSR1 functions to directly restrict germinal center B cell infection in some way. In support of this possibility, it is noteworthy that the EWSR1 family protein FUS has recently been demonstrated to regulate KSHV infection through suppression of lytic gene expression (67). Thus, the function of EWSR1 family proteins in gammaherpesvirus infection warrants significant further investigation. Moreover, the finding of a specific function for EWSR1 in germinal center B cells suggests that this protein may play a previously unappreciated role in B cell biology. proteinase K at 56°C for 8 h, followed by inactivation at 95°C for 15 min. Two rounds of nested PCR were then performed using primers (Table S5) specific for MHV68 ORF72, as previously described (30,45). The resulting 195-bp amplicons were visualized on a 3% agarose gel. The data are expressed as the means and standard errors of the percentages of PCRs that were positive for viral genome at each cell dilution.
Plasmid construction. The pUC57-Amp plasmid-based TMER5 mutants T5.ΔmiR7.12-EW.shR and T5.ΔmiR7.12-SC.shR, which carry anti-EWSR1 shRNAs (EW-shR-3 and EW-shR-2) or scrambled sequence shRNAs (SC-shR-3 and SC-shR-2) in place of pre-miR-7 and pre-miR-12 stem-loops, were synthesized and cloned at Genewiz. A 40-bp sequence upstream of TMER5 and a 20-bp sequence downstream of TMER5 were included in these constructs in order to ensure efficient expression of TMER5-encoded shRNAs in the context of the plasmids. The resulting recombinant plasmids pUC57-EW.shR and pUC57-SC.shR, expressing anti-EWSR1 shRNAs or scrambled shRNAs downstream of vtRNA5, were verified by Sanger sequencing. For luciferase assays, the murine EWSR1 coding region (nt 303 to 848 or nt 1000 to 1539) was cloned into the PmeI site of the pmirGLO dual-luciferase miRNA target expression vector (Promega, E1330), and the resulting recombinant plasmids were verified by Sanger sequencing.
Quantitative reverse transcription-PCR. At the indicated time points, the cells were harvested, and the total cellular RNA was extracted using the TRIzol reagent (Ambion, catalog no. 15596018). After DNase I (Ambion, AM1907) treatment, RNA was reverse transcribed into cDNA using ProtoScript II reverse transcriptase (New England Biolabs, M0368S) with random primer mix (New England Biolabs, S1330S). All qPCRs were performed in triplicate on the Bio-Rad CFX96 Touch real-time PCR detection system using Maxima SYBR green/fluorescein qPCR master mix (Thermo Scientific, K0243), with the primers listed in Table S5. The results were analyzed using Bio-Rad CFX Manager 3.1 software. Glyceraldehyde-3phosphate dehydrogenase (GAPDH) was used as an internal control, and the expression levels of other genes were normalized to that of GAPDH. Relative changes in expression were determined by the comparative C T method (71).
Western blots. Cells were harvested at the indicated time points, washed with cold phosphatebuffered saline (PBS), and then lysed with protein lysis buffer (150 mM NaCl, 1% NP-40, 50 mM Tris [pH 8.0], and 1ϫ cOmplete protease inhibitor cocktail). Cell lysates were then quantified for total protein content using a Pierce BCA protein assay kit (Thermo Fisher Scientific, catalog no. 23225). Equivalent amounts of protein were mixed with an equal volume of 6ϫ SDS protein loading buffer (Morganville Scientific, LB0100) containing 5% ␤-mercaptoethanol and then boiled at 100°C for 10 min. Denatured proteins were separated by using SDS-PAGE on 10% gels and then transferred onto polyvinylidene difluoride membranes (Millipore, IPVH00010). After transfer, membranes were blocked in TBST buffer (5% nonfat dry milk and 0.1% Tween 20 in Tris-borate-EDTA) for 1 h and then incubated with mouse anti-␤-actin at 1:1,000 (Cell Signaling Technology, 3700S) or rabbit anti-EWSR1 at 1:10,000 (Abcam, ab133288) at 4°C overnight. After three washes with TBST, the membranes were incubated with horseradish peroxidase-conjugated goat anti-mouse IgG at 1:5,000 (Southern Biotech, 1010-05) or goat anti-rabbit IgG at 1:5,000 (Southern Biotech, 4050-05) for 1 h at room temperature. After three washes with TBST, the protein bands were visualized by chemiluminescence using Amersham ECL Prime Western blotting detection reagents (GE Healthcare, RPN2232). The band intensities were quantified using ImageJ software (National Institutes of Health [NIH], Bethesda, MD).
For sorting latently infected B cells expressing anti-EWSR1 or scrambled shRNAs, mice were infected i.n. with 10 4 PFU of MHV68.EW.shR or MHV68.SC.shR, or with wild-type control virus. At 16 dpi, splenocytes were harvested, and B cells were isolated by immunomagnetic negative selection using an EasySep mouse B cell isolation kit (Stemcell Technologies, catalog no. 19854). The cells were blocked and then stained with APC-Cy7-conjugated rat anti-mouse CD19 (BD Biosciences, catalog no. 557655) and with CCF4-AM as described above. Infected B cells (CCF4-AM ϩ CD19 ϩ ) were sorted using a BD FACSAria II flow cytometer (BD Biosciences). RNAs from sorted cells were immediately extracted by using an RNAqueous-Micro kit (Ambion, AM1931) before qRT-PCR analyses.
Statistical analyses. All data were analyzed using GraphPad Prism 5 software (GraphPad, San Diego, CA). Statistical significance was determined using a two-tailed, unpaired Student t test. For limitingdilution nested PCR assays, the frequencies of cells positive for the MHV68 viral genome were determined by the intersection of nonlinear regression curves with the line at 63.2% using Poisson distribution analysis, which indicates the point at which one viral genome positive cell is present in a given population.