Herpes Simplex Virus 1 Lytic Infection Blocks MicroRNA (miRNA) Biogenesis at the Stage of Nuclear Export of Pre-miRNAs

Various mechanisms have been identified by which viruses target host small RNA biogenesis pathways to achieve optimal infection outcomes. Herpes simplex virus 1 (HSV-1) is a ubiquitous human pathogen whose successful persistence in the host entails both productive (“lytic”) and latent infection. Although many HSV-1 miRNAs have been discovered and some are thought to help control the lytic/latent switch, little is known about regulation of their biogenesis. By characterizing expression of both pre-miRNAs and mature miRNAs under various conditions, this study revealed striking differences in miRNA biogenesis between lytic and latent infection and uncovered a regulatory mechanism that blocks pre-miRNA nuclear export and is dependent on viral protein ICP27 and viral DNA synthesis. This mechanism represents a new virus-host interaction that could limit the repressive effects of HSV-1 miRNAs hypothesized to promote latency and may shed light on the regulation of miRNA nuclear export, which has been relatively unexplored.

promoters (deletion of the LAT promoter does not affect their expression during lytic infection as it does during latency) (12). In addition to species detected by previous Northern analyses in Vero cells, including pre-miR-H1 to pre-miR-H6 and mature miR-H1, miR-H2, and miR-H6, we additionally detected mature miR-H4-3p in both 293T and Neuro-2a cells and miR-H5 in Neuro-2a cells (Fig. 1). All 6 pre-miRNAs became detectable by 5 h postinfection (hpi) and abundant at 10 and 18 hpi in both cell lines. However, mature miR-H3 was not detected in either cell line, consistent with its very low level of expression (Ͻ1 molecule/cell) as measured by qRT-PCR of Ͻ40-nucleotide (nt) RNA in Vero cells (12). Mature miR-H1, miR-H2, miR-H5, and miR-H6 were detectable at late times and at much lower abundances than their precursors. miR-H4 was detectable as early as 2 hpi, as previously observed by qRT-PCR in Vero cells (12), and eventually reached a level that was similar to that seen with pre-miR-H4 in Neuro-2a cells but still substantially less than that of pre-miR-H4 in 293T cells. Taking the results together, despite slight differences observed among different miRNAs and cell lines, there were generally high pre-miRNA/mature miRNA (pre/mature) ratios in both 293T and Neuro-2a cells, as shown here, and in Vero cells, as shown previously (2,4,12). Thus, this phenomenon was observed in multiple species and cell types.
Differential miRNA biogenesis in acutely and latently infected mouse trigeminal ganglia. To investigate whether high pre/mature ratios also occur in vivo, we performed two independent experiments ( Fig. 2; see also Fig. S1 in the supplemental material) in which we infected mice on the cornea with 2 ϫ 10 5 PFU/eye of HSV-1 and analyzed miRNA expression in trigeminal ganglia on various days postinfection (dpi) by Northern blotting. We also explanted some ganglia at 30 dpi and cultivated them for 1 to 3 days to reactivate the virus before collecting the tissues for RNA analysis. miR-H2, miR-H4, and miR-H5 showed one expression pattern: Their mature miRNAs were readily detected after 6 dpi, with levels steadily increasing through 30 dpi and then decreasing during reactivation ( Fig. 2; see also Fig. S1). This is consistent with previous qRT-PCR data (12,22,26). However, during establishment and maintenance of latency (6 to 30 dpi), pre-miR-H4 was undetectable and pre-miR-H2 and H5 were much less abun- ). The miRNAs that are most abundant during lytic infection, miR-H1 and miR-H6, showed a different pattern. At the height of acute ganglionic infection (3 dpi), pre-miR-H1, but not miR-H1, was detected weakly, and neither was detected subsequently ( Fig. 2; see also Fig. S1), consistent with previous qRT-PCR results (4,12,22). Pre-miR-H6, but not mature miR-H6, was clearly detectable at 3 and 6 dpi, but neither pre-miR-H6 nor mature miR-H6 was detected later ( Fig. 2; see also Fig. S1). Our inability to detect miR-H6 was not consistent with previous qRT-PCR results that showed high miR-H6 expression levels in both acutely infected ganglia and latently infected ganglia (3,12,22). Insufficient sensitivity of our Northern analyses might explain this discrepancy. Finally, we did not reproducibly detect bands corresponding to pre-miR-H3 or mature miR-H3 that were clearly distinct from background in mouse ganglia, which was consistent with undetectable levels in previous qRT-PCR results obtained from Ͻ40-nt RNA (12) but not with reported high levels in unfractionated RNA (22). We suspect that the latter result was due to the presence of longer species contributing to miR-H3 qRT-PCR signals. Taking the results together, during establishment and maintenance of latency, mature miRNAs, whenever detectable, were always more abundant than pre-miRNAs (i.e., low pre/mature ratios), whereas at the height of acute infection (3 dpi), no mature miRNA was consistently detected despite multiple pre-miRNAs being detected (high pre/mature ratios), highlighting the contrast between acute infection and latent infection in miRNA biogenesis.
Lower pre/mature ratios in lentivirus-transduced cells than during HSV-1 lytic infection. We next asked whether the high pre/mature ratios seen during lytic infection might have been due simply to a lack of sufficient time to complete biogenesis. To determine how quickly mature miRNAs could be expressed in cells, we employed an inducible lentiviral transduction system. We cloned sequences encoding miR-H2 or FIG 2 HSV-1 pre-miRNA and mature miRNA expression in trigeminal ganglia at different times after infection or after reactivation from latency. Mice were mock infected or infected on the cornea with 2 ϫ 10 5 PFU of HSV-1 strain KOS per eye. At the various days postinfection (dpi) indicated, trigeminal ganglia were collected for immediate RNA isolation. Additionally, at 30 dpi, some ganglia were explanted and cultivated for various days postexplantation (dpe) before RNA was isolated. Sixteen ganglia were pooled for each condition, and small RNAs (Ͻ200 bases) were isolated and analyzed by Northern blot hybridization. RNAs purified from Neuro-2a cells that had been mock infected or infected with HSV-1 strain KOS for 10 h were run alongside (first two lanes) the RNAs from ganglia to provide markers. The sources of RNAs and time points are indicated at the tops of each series (left and right, respectively) of Northern images. Positions of miRNA species are indicated to the right of the images. Ethidium bromide signals at ϳ80 bases from the gels used for each series of Northern images are displayed at the bottom. Results of an independent experiment are shown in Fig. S1. miR-H3/H4 pre-miRNAs along with flanking sequences into the pTRIPZ vector, in which inducible gene expression is driven by a tetracycline-responsive promoter (sequences encoding pre-miR-H3 and -H4 were cloned as one fragment as they are separated by only 129 bp). We then produced lentiviruses and used them to stably transduce 293T and Neuro-2a cells. We monitored miRNA expression following doxycycline (Dox) addition to the transduced cells. The three miRNAs differed greatly in their final pre/mature ratios, with miR-H4 having the lowest and miR-H3 the highest in both the 293T-derived and Neuro-2a-derived cell lines (Fig. 3), suggesting important inherent differences in biogenesis efficiency among them. Nevertheless, the transduced cells always exhibited more rapid and efficient miRNA biogenesis than that seen in HSV-1infected cells (Fig. 1). For example, the miR-H2 and miR-H4 pre/mature ratios seen at 5 h after addition of Dox were already lower than those seen at 18 h after HSV-1 infection, and mature miR-H3 was detectable in transduced cells but undetectable in infected cells despite the high pre-miR-H3 levels in both settings. These results argue for a regulatory mechanism impeding accumulation of mature miRNAs during lytic infection.
High pre/mature ratios are due to inefficient pre-miRNA conversion to mature miRNA rather than to mature miRNA destabilization. Two hypotheses could explain the high pre/mature miRNA ratios: (i) block of pre-to-mature conversion and (ii) expedited degradation of mature miRNA. To investigate these hypotheses, we analyzed miRNA stability following addition of a transcription inhibitor, actinomycin D (ActD). In The experiment using 293TH2 cells was performed twice; the experiment using Neuro-2aH2 cells was performed once. (B) As described for panel A, but the 293TH3/H4 and Neuro-2aH3/H4 cell lines were created by transduction using a lentivirus expressing miR-H3 and miR-H4, and miR-H3 and H4 expression levels were analyzed. These experiments were repeated with cells constructed using different lentivirus constructs expressing miR-H3 and/or miR-H4.
HSV Impairs Nuclear Export of miRNAs ® January/February 2019 Volume 10 Issue 1 e02856-18 mbio.asm.org 5 transduced 293T cells, at 3 days following Dox addition, mature miR-H2, miR-H3, and miR-H4 were all stable for 8 h following ActD addition, but the corresponding pre-miRNAs were all much less stable ( Fig. 4A and B). In HSV-1-infected cells, when ActD was added at 2 hpi, pre-miR-H2 and pre-miR-H4 levels decreased substantially over the next 8 h ( Fig. 4C; see also Fig. S2). However, when ActD was added at 8 hpi, pre-miR-H2 levels did not decrease over the 8 h that followed, and pre-miR-H4 and pre-miR-H1 levels decreased by only ϳ2-fold during this time frame ( Fig. 4D; see also Fig. S2). Following ActD addition at either 2 or 8 hpi, the levels of the corresponding mature miRNAs remained stable or even increased, with the possible exception of the level of miR-H4, which appeared to decrease and then become stable following ActD addition at 8 hpi (Fig. S2). These results suggest that inefficient conversion of pre-miRNA to mature miRNA, rather than mature miRNA instability, accounts for high pre/mature ratios and that pre-miRNAs become more stable over the course of infection, consistent with interference with pre-miRNA conversion to miRNA late in infection. A host miRNA expressed from a recombinant HSV-1 exhibited inefficient pre-miRNA processing during lytic infection. We wondered whether this impairment miR-H2 expression from 293TH2 cells was induced by adding 1 g/ml of Dox for 3 days. Then, 1 g/ml of ActD was added to inhibit transcription. Cells were harvested at the times indicated at the top of the image and were analyzed by Northern blot hybridization for miR-H2 expression. Uniform loading on the gel used for the Northern blot hybridization was demonstrated by ethidium bromide staining (shown at the bottom). This experiment was performed once. (B) As described for panel A, but 293TH3/H4 cells were used and the levels of expression of miR-H3 (upper panel) and miR-H4 (lower panel) were analyzed. This experiment was repeated with cells constructed using different lentivirus vectors expressing miR-H3 and/or miR-H4. (C) 293T cells were infected with HSV-1 strain KOS (MOI ϭ 10). At 2 hpi, 1 g/ml ActD was added, and cells were harvested at the times indicated at the top of the images and analyzed by Northern blot hybridization for miR-H2 expression. The upper image shows pre-miR-H2 expression (mature miR-H2 was not detectable at that time point). The bottom images show the ethidium bromide-stained gel at ϳ80 bases. (D) As described for panel C, but ActD was added at 8 hpi, and miR-H1 of biogenesis is specific to viral miRNAs. The effects of HSV infection on endogenous or transduced miRNAs are difficult to study due to the already high levels of most mature host miRNAs and the shutoff of most host transcription (reviewed in reference 27). To avoid these issues, we expressed miR-138 as an HSV-1 gene product. This miRNA was chosen because its low level of endogenous expression in 293T cells (28) allows easy detection of its exogenous expression from a virus. In an initial experiment, we cloned pre-miR-138-1 and flanking sequences into the inducible lentiviral transduction system described above to assess the intrinsic biogenesis efficiency of this miRNA. Upon Dox addition to miR-138-expressing 293T cells, pre-miR-138-1 levels were nearly constant from 2 to 48 h, but mature miR-138 levels increased steadily, resulting in pre/mature ratios of about 0.25 by 24 h (Fig. 5A), indicating that the biogenesis of miR-138 is intrinsically efficient. We then modified the HSV-1 genome to insert the same pre-miR138-1 and flanking sequences between the Us11 and Us12 open reading frames such that miR-138 would be encoded within the Us12 3= untranslated region (UTR) and Us11 5= UTR and thus would be expressed from both the Us12 immediate early and Us11 late promoters (Fig. 5B). The resulting recombinant HSV-1, designated WTLyt138, showed replication kinetics in Vero cells indistinguishable from those of the corresponding bacterial artificial chromosome (BAC)-derived WT parental virus (WT-BAC) ( Fig. 5C) and showed no more than a modest defect in Us11 expression (data not shown) which should not affect miRNA biogenesis, because deletion of Us11 has little if any impact on miRNA biogenesis (see below). As expected for transcription from both immediate early and late promoters, pre-miR-138-1 expression from WTLyt138 virus appeared to be biphasic, with initial expression at 2 hpi followed by a decrease in the level of expression at 5 hpi and then an increase again after 5 hpi (Fig. 5D). Pre-miRNA conversion to mature miRNA was efficient during the first phase, with a pre/mature ratio of 0.47 at 5 hpi. After 5 hpi, however, mature miR-138 levels increased only slightly despite dramatic increases in pre-miR-138-1 levels, resulting in a 4-fold-higher pre/ mature ratio at 16 hpi than at 5 hpi, arguing that biogenesis of a host miRNA was also impeded late during lytic infection.

Pre-miRNA nuclear export is blocked during HSV-1 lytic infection.
To understand the mechanism of this inefficient biogenesis, we first asked whether infection changed levels of key factors involved in miRNA biogenesis and found that Drosha, Exportin 5, and Dicer were not downregulated during HSV-1 infection (Fig. S3), indicating that changes in the expression levels of these proteins did not explain inefficient miRNA biogenesis.
We then analyzed miRNA expression in nuclear and cytoplasmic fractions of WTLyt138-infected cells. The biphasic expression of miR-138 from this virus facilitated analysis of this experiment. Successful fractionation was achieved and resulted in clean separation of U1 small nuclear RNA from lysine-tRNA (Fig. 6). At 5 hpi, pre-miR-138-1 was found to be present at equivalent levels in the two fractions. However, at 12 hpi, there was much more pre-miR-138-1 in the nuclear fraction than in the cytoplasmic fraction, indicating inefficient nuclear export after 5 hpi. Also, much larger fractions of pre-miR-H1 and -H6 were found in the nucleus than in the cytoplasm at 12 hpi (Fig. 6). Similar results were observed with pre-miR-H6 and mature miR-H6 in WT-infected cells (Fig. S4). We found no evidence of impaired Dicer cleavage as judged by the relatively low pre/mature ratios in the cytoplasmic fractions at 5 hpi for miR-138 and at 12 hpi for ). Therefore, we conclude that HSV-1 infection impedes miRNA biogenesis by blocking nuclear export. ICP27 and viral DNA synthesis but not Us11 are necessary for blocking pre-miRNA processing. Hypothesizing that viral gene products might be involved in impeding miRNA biogenesis, we first tested the idea that Us11, an RNA binding protein with affinity for double-stranded RNA and even higher affinity for imperfect stem-loops (29,30), might sequester pre-miRNAs, especially as Us11 has been suggested to block RNA silencing (31). However, pAUs11 mutant virus derived from strain Patton (32), which is Us11-defective due to a polyadenylation signal 5= to the Us11 start codon, showed pre/mature ratios for miR-H1 and miR-H2 similar to those of the corresponding rescued virus (data not shown). Because we detected a weak immunoreactive band that comigrated with strain Patton Us11 in pAUs11-infected cells on a Western blot (Fig. 7A), we wondered whether leaky expression might be sufficient to impede miRNA biogenesis. We therefore constructed two additional Us11 mutant viruses on the basis of our KOS-derived WT-BAC virus (33). One virus, named MUs11a, removes an A from the Us11 start codon, which should result in translation starting from an out-of-frame AUG. The other virus, named MUs11ab, further replaces the second and third in-frame ATG with TGA stop codons in the MUs11a virus. Ablation of Us11 immunoreactivity comigrating with KOS Us11 was confirmed for both viruses (Fig. 7A). Both viruses exhibited levels of pre-miR-H1 and pre-miR-H2 and mature miR-H1 and miR-H2 similar to those seen with the WT (Fig. 7B), suggesting that Us11 is dispensable for regulation of miRNA biogenesis.
Another candidate for impeding miRNA biogenesis is ICP27, which plays a role in mRNA nuclear export (reviewed in reference 34). A homolog of ICP27 has been reported to block RNA silencing in plants (35). To test its involvement in miRNA biogenesis, we utilized an ICP27-null virus, d27 derived from HSV-1 strain KOS1.1 (36). Following infection of Vero cells, d27 expressed similar or possibly greater amounts of mature miR-H2 but substantially lower amounts of pre-miR-H2 than did both its parental WT virus and the rescued derivative, R27 (Fig. 7C). In another experiment, d27 expressed amounts of mature miR-H6 similar to those seen with the WT virus but much less pre-miR-H6 than the WT virus in infected Vero cells, and yet the d27 and WT viruses expressed similar amounts of pre-miR-H6 and mature miR-H6 in E11 cells, which ectopically express ICP4 and ICP27 (37) (Fig. 7D). In a third experiment, total RNA or RNA from nuclear and cytoplasmic fractions was isolated from WT and d27-infected 293T cells at 12 hpi. At that time point, similar to the results seen at the 12-hpi point (Fig. 6), WT virus expressed readily detectable pre-miR-H6 but much less mature miR-H6 (Fig. 7E, lane 2). The pre-miR-H6 was primarily found in the nuclear fraction (lane 3), with substantially less seen in the cytoplasmic fraction, and that small amount could be accounted for by nuclear contamination as evidenced by detection of U1 snRNA (lane 4). In contrast, the ratio of pre-miR-H6 to mature miR-H6 was lower in d27-infected cells than in WT-infected cells at 12 hpi (compare lanes 5 and 1), and the levels of pre-miR-H6 were similar in the nuclear and cytoplasmic fractions, which were cleanly separated based on the U1 and lys-tRNA markers. Taken together, these results show that ICP27 contributes to high pre/mature ratios and to retention of pre-miRNA in the nucleus.
Finally, we tested whether the impairment in miRNA biogenesis is dependent on viral DNA synthesis. 293T cells were infected with WTLyt138 in the absence or presence of acyclovir (ACV), an inhibitor of HSV DNA synthesis. At 8 hpi, infected cells were either mock treated or treated with ActD, and RNA was harvested immediately or at 14 hpi. In the absence of ACV and ActD, the miR-138 pre/mature ratios increased between 8 and 14 hpi (Fig. 7F, lanes 2 and 4), similarly to what we observed in Fig. 5D. Adding ActD in the absence of ACV at 8 hpi resulted in little change in pre-miR138-1 and mature miR-138 levels between 8 hpi and 14 hpi (lanes 2 and 3), consistent with the data in Fig. 4D indicating inefficient conversion of pre-miRNA to mature miRNA. However, in the presence of ACV, at both 8 and 14 hpi, the levels of mature miR-138 were similar to those seen in the absence of ACV at the same time points, but the levels of pre-miR-138-1 were considerably lower (compare lanes 5 and 7 with lanes 2 and 4), similarly to what we observed in the absence of ICP27 (Fig. 7C to E). Addition of ActD to ACV-treated infected cells at 8 hpi resulted in a decrease in pre-miR-138 and an increase in mature miR-138 (compare lanes 5 and 6), indicating conversion of pre-miRNA to mature miRNA. Thus, high pre/mature ratios are dependent on both ICP27 and viral DNA synthesis but not on Us11.

DISCUSSION
By characterizing HSV-1 pre-miRNA and miRNA expression in transduced and infected cells and in mouse ganglia, we found that although these miRNAs differed in inherent biogenesis efficiencies, conversion of pre-miRNA to miRNA was considerably less efficient during lytic infection than during latent infection or following lentiviral transduction, regardless of the miRNAs or cell types tested. Much if not all of this At 12 hpi, total RNA was isolated from infected cells (T; lanes 1, 2, and 5) or from nuclear (N; lanes 3 and 6) or cytoplasmic (C; lanes 4 and 7) fractions and analyzed as described for Fig. 6 for miR-H6, U1, and lys-tRNA (as indicated to the right of the panel). This experiment was performed twice. (F) Effects of ACV and ActD on miRNA biogenesis. 293T cells were infected (MOI ϭ 5) with WT-BAC (lane 1) as a negative control for miR-138 expression or with WTLyt138 (lanes 2 to 7), as indicated in the top two lines of the panel. Plus signs (ϩ) on the next two lines indicate treatment with ACV at the time of infection (lanes 5 to 7) or with ActD at 8 hpi (lanes 3 and 6), respectively. ACV was added to block viral DNA synthesis, and ActD was added to block RNA synthesis to help assess the stability of pre-miR-138 and mature miR-138. Plus signs on the next two lines indicate whether infected cells were harvested for RNA isolation at 8 hpi (lanes 2 and 5) or 14 hpi (lanes 1, 3, 4, 6, and 7). The image below shows the results of Northern analysis of expression of pre-miR-138-1 and miR-138, as indicated to the left. Pre/mature ratios calculated following band quantification are displayed below the image. let-7a levels as loading controls are shown at the bottom. This experiment was performed twice. impairment of miRNA biogenesis was due to a block in pre-miRNA export from the nucleus to the cytoplasm. High pre/mature ratios occurred late in infection and were dependent on ICP27 and viral DNA synthesis but not on Us11. Below, we discuss these results and their implications for HSV-1 infection.

Regulation of miRNA biogenesis during HSV-1 lytic infection and variability in intrinsic biogenesis efficiency both contribute to miRNA expression.
Our Northern results show that all HSV-1 miRNAs tested exhibited high pre/mature ratios during lytic infection of multiple cell lines. Consistent with this, most exhibited high pre/mature ratios in ganglia at 3 dpi, arguing for a regulatory mechanism that applies to both acute ganglionic infection and lytic infection of cultured cells. During latency, pre/mature ratios were considerably reduced, arguing that miRNA biogenesis became more efficient, presumably due to reduced expression of the viral function(s) that impaired miRNA biogenesis. This mechanism then likely contributed to high expression of these miRNAs during latency.
Based on our results from lentivirus-transduced cells, the order of intrinsic efficiencies of conversion of pre-miRNA to miRNA is miR-H4 Ͼ miR-H2 Ͼ miR-H3, consistent with the relative pre/mature ratios for these miRNAs in infected cells and ganglia and suggesting that intrinsic biogenesis efficiencies contribute to determining expression levels. The finding of different intrinsic biogenesis efficiencies explains the inability to detect mature miR-H3 during lytic or latent infection by Northern analysis and the inability to detect pre-miR-H4 in infected ganglia. Variability in intrinsic biogenesis efficiency may reflect sequence-specific differences in pre-miRNA export and/or Dicer cleavage (38,39). Regardless, for each miRNA tested, including the host miR-138, pre-miRNA conversion to miRNA was more efficient in lentivirus-transduced cells than in HSV-1 lytically infected cells, consistent with impairment of miRNA biogenesis during lytic infection.
Mechanism of impairment of miRNA biogenesis during lytic infection. miRNA biogenesis during lytic infection of cells is temporally regulated. When detectable at early times of infection (e.g., in the case of ectopically expressed miR-138), pre-miRNAs convert efficiently to miRNAs, but the level of conversion is reduced later in infection. Moreover, pre-miRNAs exhibit short half-lives in the presence of ActD at 2 hpi but are much more stable at 8 hpi. Thus, impairment of miRNA biogenesis is a late infection event. Consistent with this, ACV treatment resulted in both reduced pre/mature ratios and reduced stability of pre-miRNA in the presence of ActD. Similarly, our finding that pre-miRNAs were depleted in the cytoplasm and enriched in the nucleus at 12 hpi, but not at 5 hpi, indicates that inhibition of nuclear export of pre-miRNAs occurs late. Thus, the results strongly suggest that inhibition of nuclear export of pre-miRNAs explains the impairment of miRNA biogenesis late during lytic infection. However, we cannot exclude the possibility of effects on other steps. Blocking nuclear export of pre-miRNAs is likely not unique to HSV-1 infection. Baculovirus encodes a miRNA that was reported to target the Exportin-5 cofactor Ran (40). Accumulation of pre-miR-31 in the nucleus with undetectable mature miR-31 has been reported in certain cancer cell lines (41), while activation of the oncogenic protein kinase extracellular signal-regulated kinase (ERK) has been reported to downregulate nuclear export of pre-miRNAs in hepatocellular carcinoma cells (42). Understanding how HSV blocks nuclear export of pre-miRNAs may reveal mechanisms that regulate miRNA biogenesis in other viral and cellular systems.
Deletion of ICP27 also greatly reduced pre/mature ratios for HSV-1 miRNAs. Interestingly, this deletion reduced the expression of the pre-miRNAs with no reduction or possibly an increase in the levels of the mature miRNAs, which we suggest was due to these pre-miRNAs being late viral gene products (12) and thus likely to be dependent on ICP27 for their full expression. Nevertheless, the simplest interpretation is that ICP27, one of whose many functions (34) is to promote export of certain mRNAs (43)(44)(45), is required for impairment of miRNA biogenesis. Assuming that this is correct, the issue becomes whether ICP27 itself acts to impair miRNA biogenesis or whether ICP27 is required for synthesis of one or more viral gene products that do so. Supporting the former possibility, the turkey herpesvirus (HVT) homolog of ICP27 has been reported to be sufficient for blocking RNA silencing in tobacco (35), and it has been speculated that ICP27 may block nuclear export of certain RNAs (46). Supporting the latter possibility are the immediate early expression of ICP27 versus the late impairment of miRNA biogenesis and the similar phenotypes of ACV treatment and ICP27 deletion, especially given that most viral late and certain DNA replication proteins require ICP27 for their efficient synthesis (47). A third possibility is that HSV-1 induces expression of a host factor that impairs miRNA biogenesis in an ICP27-dependent fashion. Experiments are under way to test these alternatives.
Implications for HSV-1 infection. Our results lead to a model (Fig. 8) in which HSV-1 lytic infection blocks pre-miRNA conversion to miRNAs at the stage of nuclear export whereas lytic gene products responsible for this block are depleted during latent infection, allowing more efficient biogenesis. Certain HSV miRNAs that are expressed abundantly during latency have been shown to be able to repress expression of important lytic proteins such as ICP0, ICP4, and ICP34.5 in transfection experiments (3,9,10,(13)(14)(15). miRNAs that are abundant during latency might also inhibit the expression of host proteins that would otherwise promote lytic rather than latent infection. Thus, latent HSV miRNAs have been hypothesized to promote latency. Although some in vivo mutational studies of HSV-1 and HSV-2 miRNAs have yielded negative results (15,25), interpretation of these results is complicated by the possibility of functional redundancy and the limitations of using small animal models to recapitulate latency in humans using small-animal models. Blocking miRNA biogenesis during lytic infection may serve to limit the extent to which expression of their targets is repressed, allowing less-restricted viral replication. This proposal is supported by a study that showed that HSV-1 infection suppresses RNA-induced silencing and that knockdown of Ago2 expression increases HSV-1 replication (48). During latent infection, when repression of lytic gene expression (or certain host factors) is beneficial to the virus, removal of the block to miRNA biogenesis combined with high transcription of the LAT locus permits strong expression of latent miRNAs (Fig. 8).
In plants and invertebrates, viruses rely on virus-encoded suppressors of RNA interference (RNAi) to combat an RNAi-mediated antiviral response, although the functional relevance of this suppressive activity in mammals remains to be established (49). Mammalian viruses, including some herpesviruses, have displayed various ways of regulating miRNA biogenesis (reviewed in references 19 and 50). However, we are unaware of any previous examples of regulation of pre-miRNA nuclear export by mammalian viruses. Effects of viral infection on expression of regulators of miRNA biogenesis have been observed with gammaherpesviruses (see, e.g., references 51 and 52), and there is evidence that levels of certain mature viral miRNAs are regulated posttranscriptionally (see, e.g., references 53 and 54). Specific downregulation of a host miRNA by overexpression of a complementary target has been reported previously for both betaherpesviruses and gammaherpesviruses (55)(56)(57)(58). However, although the high pre/mature ratios of most viral miRNAs that we document here can also be found during lytic infection with HSV-2 and pseudorabies (9,10,24,25,59), they have not been generally observed during the lytic cycles of the avian alphaherpesvirus, Marek's disease virus, betaherpesviruses, or gammaherpesviruses (60)(61)(62)(63)(64)(65)(66). Therefore, general impairment of conversion of pre-miRNA to mature miRNA appears to be found mainly among mammalian alphaherpesviruses. We speculate that this might relate to their neurotropism. Regardless, our findings identify a novel mechanism of differential regulation of miRNA expression that may ensure optimal outcomes for lytic and latent infection.

MATERIALS AND METHODS
Cells and viruses. Vero, 293T, and Neuro-2A cells were maintained as previously described (28). Vero and Neuro-2A cells were obtained from ATCC and used following less than 6 months of passage. 293T cells were authenticated by small-tandem-repeat profiling at the Dana-Farber Cancer Institute Molecular Diagnostics Laboratory in December 2018. Viruses were propagated in Vero cells and titrated by plaque assays following standard protocols (67). WTLyt138, MUs11a, and MUs11b viruses were constructed based on a BAC-containing strain, KOS (33), by two-step red recombination (68). To construct WTLyt138 virus, the following human pre-miR-138-1 and flanking sequences were synthesized as a gBlocks gene fragment (Integrated DNA Technologies): CTTAGAATTCGGATCATCAATCATCTGCCTTCCCAGGACCACCCA GCAGTGTTGCACGTAGAGCAGAGGTGCAGTGCTGCACGTACAGCAGAGGCGACCCCTTCTCCATACTTCAGAGA CCTCTAGCATCGTGTTGTGGGACAGCTGGTGTTGTGAATCAGGCCGTTGCCAATCAGAGAACGGCTACTTCACAA CACCAGGGCCACACTGCACTGCAAGCAGCGAGCAGCAAGCAACAAGCATCCAGACACCTGAGCACCAGGCTGC CAGCAAGAGCACCTGCCTGTGAGGAAACCTCCCAGCCCACCCCATACCCCACCCCATTATCCCACCCTACAAGC TTGGTA. The sequences were cloned between the EcoRI and HindIII sites of FLAG-HA-pcDNA3.1(-) plasmid (AddGene). A kanamycin resistance gene (Kan) flanked by one I-SceI site was then cloned into the EcoRI site of the resulting plasmid using the following two primers: primer CTTAGAATTCGGA TCATCAATCATCTGCCTTCCCAGGACCACCCAGCAGTGTTGCACGTATAGGGATAACAGGGTAATCGATTT and primer CTTAGAATTCGCCAGTGTTACAACCAATTAACC. The resulting plasmid was used as a template to amplify a cassette to be recombined with the HSV-1 genome using the following two primers: primer AATCCGGTAACCCGTTGAGTCCCGGGTACGACCATCACCCGAGTCTCTGGAATTCGGATCATCAATCATCTGCC and primer GGGTTGGGTCTGGCTCATCTCGAGAGACACGGGGGGGAACCACCCTCCGCGTAGGGTGGGATAAT GGGGTGG. The subsequent recombination of this cassette with the viral genome and removal of Kan were conducted as described previously (68). The two-step red recombination protocol was also used to construct MUs11a BAC using primer TACGACCATCACCCGAGTCTCTGGGCGGAGGGTGGTTCCCCCCCGTGT CTCTCGAGTGAGCCAGACCCAAcCCCCGTAGGGATAACAGGGTAATCGATTT and primer TAAGTAAACATCTG GGTCGCCCGGCCCAACTGGGGCCGGGGGTTGGGTCTGGCTCACTCGAGAGACACGGGGGGGAGCCAGTGTTAC AACCAATTAACC as well as to construct MUs11b based on MUs11a using primer TGTTTACTTAAAAGGC GTGCCGTCCGCCGGCTGACACCCCAGAGGTGTTCACGCACCTCGAGGACACCCGCGCTGATAGGGATAACAGG GTAATCGATTT and primer TCATTATCACCCCGTTGCGGGGGTCCGgAGATTCAGCGCGGGTGTCCTCGAGGTG CGTGAACACCTCTGGGGTGTCAGCCAGTGTTACAACCAATTAACC.
Viruses were produced by transfecting BAC DNAs into Vero cells using Lipofectamine 3000 following the instructions of the manufacturer (Thermo Fisher). Cells were infected by adding virus at the indicated MOI, incubating for 1 h at 37°C with gentle shaking every 15 min for adsorption, and replacing the supernatant with fresh media and incubating at 37°C for the indicated times.
Northern blot hybridization. RNAs of Ͻ200 nucleotides were purified from cultured cells or pooled mouse ganglia using a miRNeasy minikit (Qiagen) following the manufacturer's instructions and were resolved on Novex 15% Tris-borate-EDTA-urea polyacrylamide gels (Thermo Fisher). We used radioactive and nonradioactive methods interchangeably for Northern blot hybridization. The nonradioactive method followed a previously described protocol (15). LNA probes were synthesized and 3= end labeled with digoxigenin by Exiqon. The radioactive method followed the same protocol before hybridization. During the prehybridization step, 10 pmol of probe was incubated with 1 l of T4 polynucleotide kinase (NEB), 2.5 l of the supplied buffer, and 1 to 5 l of [␥-32 P]ATP (Perkin Elmer) in a total volume of 25 l for 1 h at 37°C before this mixture was added into the hybridization tube. The subsequent hybridization and washing conditions were the same as those used with the nonradioactive method. After the last wash, the membrane was exposed to a phosphor screen and visualized by a phosphorimager. Band intensities were quantified by Quantity One 1-D Analysis Software (Bio-Rad). The probe sequences and optimized hybridization and washing temperatures for different RNAs are listed in Table 1.
Animal procedures. The animal housing and experimental procedures were approved by the Institutional Animal Care and Use Committee of Harvard Medical School in accordance with federal guidelines. Male CD-1 mice (Charles River Laboratories) (7 weeks old) were infected with HSV-1 strain KOS as described previously (15). At the indicated times, trigeminal ganglia were removed and placed on dry ice before being stored at Ϫ80°C. For reactivation, trigeminal ganglia removed from mice at 30 dpi were held in Dulbecco's modified Eagle medium supplemented with 10% fetal bovine serum on ice during the dissection procedure and were then incubated at 37°C for 1 to 3 days before the ganglia were transferred to Eppendorf tubes. Sixteen ganglia were pooled for each condition before RNA purification.
Construction of lentivirus-transduced cell lines. The synthesized gBlocks gene fragment described above was used as a miR-138-expressing sequence. miR-H2-expressing and miR-H3/H4-expressing sequences were amplified from HSV-1 WT BAC using primers TGACCTCGAGTTTTCCTGGCCCGACCCGCG CCTCTT and CTGTGAATTCGACCCCATAGTGATCAGCGACTCC for miR-H2 and primers TGACCTCGAGGAC GCGGACTCGGGAACGTGGA and CTGTGAATTCTCGCGCGCGGCCCTTTAAAGGCG for miR-H3/H4. Each miRNA-expressing sequence was inserted between the XhoI and EcoRI restriction sites of pTRIPZ (69) (a generous gift from Bryan Cullen). A 4-g amount of each of the resulting plasmids was transfected together with 7.1 g of psPAX2 plasmid (Addgene) and 3.9 g of pMD2.G plasmid (Addgene) into 293T cells in a 100-mm-diameter plate using Lipofectamine 3000 following the instructions of the manufacturer (Thermo Fisher). Lentiviruses were harvested from the supernatants 3 days posttransfection and added to 50% confluent 293T or Neuro-2a cells in the presence of 8 g/ml of hexadimethrine bromide (Sigma). Two days later, the supernatant was removed and replaced with fresh medium containing 1 g/ml of puromycin. Surviving cells were expanded in the presence of puromycin until the cells were stably maintained.
Cell fractionation. Cell fractionation was performed using a tissue grinder (Wheaton) (1 ml) as previously described (71). RNA was purified from total, cytoplasmic, and nuclear fractions using a mirVana miRNA isolation kit following the instructions of the manufacturer (Ambion). a The probes are labeled with digoxigenin at the 3' end, for nonradioactive detection. ϩN, locked nucleic acid-modified nucleotide. b The washing temperature is typically the same as hybridization temperature, but during the last two washes, raising the temperature by 5°C helps further remove nonspecific bands.