MicroRNAs (miRNAs) are ~22-nucleotide (nt) regulatory RNAs that can posttranscriptionally inhibit the translation of target mRNAs, typically by binding to complementary sites located in the 3′ untranslated region (3′UTR) (1
). miRNAs are expressed by all plants and animals, as well as by several DNA viruses. In particular, numerous members of the herpesvirus family have been found to express multiple viral miRNAs in infected cells (10
Humans are the only natural host of herpes simplex virus 2 (HSV-2). HSV-2 typically replicates productively in the mucosal epithelium of the genitalia and establishes latent infection in neurons of the sacral ganglia (7
). During latency, HSV-2 gene expression is restricted, except for that of the noncoding latency-associated transcript (LAT). Three HSV-2 miRNAs that were initially identified by sequencing small RNAs expressed in cells transfected with an HSV-2 LAT expression vector have been reported, and expression of these miRNAs in HSV-2-infected cells was also confirmed (8
). While these miRNAs were initially named miR-I, miR-II, and miR-III, their sequence similarities to the HSV-1 miRNAs miR-H3, miR-H4, and miR-H2, respectively (11
), combined with their shared locations within the genomes of these closely related pathogens, led the authors of the miRBase database (4
) to rename them HSV-2 miR-H3, miR-H4, and miR-H2. This revised nomenclature is used throughout this paper. No HSV-2 homologs of the other five known HSV-1 miRNAs, miR-H1, miR-H5, miR-H6, miR-H7, and miR-H8, have been reported.
To determine if additional HSV-2 miRNAs are expressed during latent infection in humans, we performed deep sequencing of miRNAs expressed in HSV-2-infected human sacral dorsal root ganglia. Three ganglia were collected at autopsy less than 24 h after death, frozen on dry ice, and stored at −80°C. HSV-2 genomic DNA levels were quantified using real-time PCR with primers and a probe specific for the HSV-2 glycoprotein G gene by using total DNA isolated from frozen ganglia. For subject 1, we detected 1,616 copies of the HSV-2 genome per μg total DNA; for subject 2, we detected 328 copies per μg; and for subject 3, we detected 4,567 copies per μg.
Total RNA prepared from all three sacral ganglia was used for small (18- to 24-nt) RNA isolation, cDNA synthesis, and Solexa deep sequencing, as previously described (12
). Sequencing returned 1,125,318 useable reads from subject 1, 5,417,868 reads from subject 2, and 6,191,093 reads from subject 3. Since only a very small percentage of neurons within the sacral ganglia were latently infected with HSV-2, the vast majority of sequences were, as expected, found to represent cellular miRNAs (data not shown).
Analysis of the sequence data corresponding to subject 1 revealed the presence of HSV-2 miRNA miR-H3, previously termed miR-I (8
), as well as a novel HSV-2 miRNA, designated miR-H10 (Table ). Analysis of data from subject 2 identified HSV-2 miR-H3 and also miR-H4, previously termed miR-II (9
), as well as miR-H10 and another novel HSV-2 miRNA, designated HSV-2 miR-H7. Sequences obtained from subject 3 also included miR-H3 and miR-H4, as well as an additional novel HSV-2 miRNA, designated miR-H9. Overall, the viral miRNA sequences recovered from these 3 samples were surprisingly heterogeneous, with no HSV-2 miRNA other than miR-H3 being recovered from all three subjects, although HSV-2 miR-H4 and miR-H10 were each recovered from two subjects. Surprisingly, we did not recover the previously described HSV-2 miRNA miR-H2, previously termed miR-III (9
), from any of the subjects analyzed. Of note, we observed little evidence that HSV-2 miRNA expression correlated with HSV-2 genome copy number, a result which is consistent with a previous report indicating that the HSV-2 genome copy number also does not correlate with LAT expression in cells with latent HSV-2 infection (13
HSV-2 miRNA expression profile in human sacral ganglia latently infected with HSV-2
The genomic loci encompassing the novel HSV-2 miRNAs miR-H7, miR-H9, and miR-H10 were each found to fold into the hairpin structures predicted for true primary miRNA (pri-miRNA) transcripts, with each mature miRNA, as expected, forming the upper ~22 nt of one arm of the stem (Fig. ). This strongly suggests that these are indeed legitimate miRNAs. The genomic locations of miR-H7 and miR-H9 indicate that they, like HSV-2 miR-H3 and miR-H4, are transcribed as part of the second exon of LAT and place both miRNAs antisense to ICP0
(Fig. ), an immediate-early gene that plays an important role in HSV-2 lytic replication (7
). The genomic location of miR-H10 is currently unique in that it does not map to the exonic regions of LAT or indeed anywhere near the LAT gene (Fig. ), as is the case with all the other known HSV-2 and HSV-1 miRNAs (2
). The genomic location of miR-H10, distal to the LAT locus, explains why this miRNA was not identified in a previous HSV-2 deep-sequencing effort which analyzed RNA samples derived from cells transfected with an HSV-2 LAT expression plasmid (9
). Instead, miR-H10 must be excised from a novel latency transcript derived from the unique long (UL
) region, just 3′ to the UL15
polyadenylation site (Fig. ). At this time, it is unclear if miR-H10 is derived from a longer form of the UL15
transcript that results from read-through of the known UL15 polyadenylation site or if it is derived from a novel, latency-specific HSV-2 transcript.
FIG. 1. Three novel HSV-2 miRNAs recovered from latently infected human sacral ganglia. (A) Predicted secondary structures adopted by the pri-miRNA precursors of the novel HSV-2 miRNAs miR-H7, miR-H9, and miR-H10. Triangles indicate the boundaries of each predominant (more ...)
To confirm the expression of the novel HSV-2 miRNAs miR-H7 and miR-H9, we used quantitative reverse transcription-PCR (qRT-PCR) to analyze total RNA samples derived from the three original HSV-2-infected sacral ganglia (Fig. ). Viral miRNA expression levels were normalized to levels of a cellular miRNA, miR-16, and values are expressed relative to a negative control sample consisting of total RNA derived from an HSV-2 DNA-negative sacral ganglion. miR-H7 was found to be present at differing but readily detectable levels in all three of the HSV-2-infected sacral ganglia examined (Fig. ), while miR-H9 was expressed at very low levels, barely above background. qRT-PCR was not able to detect significant levels of the previously reported HSV-2 miR-H2 (9
), which was also not detected by deep sequencing (Table ). In contrast, in productively HSV-2-infected Vero cells, qRT-PCR revealed that miR-H2, miR-H7, and miR-H9 all accumulated to readily detectable levels as infection progressed, with miR-H9 being expressed at levels significantly higher than those of either miR-H2 or miR-H7 at 20 h postinfection (Fig. ). This pattern parallels the known expression kinetics of LAT, which behaves as a viral late gene during productive replication (5
FIG. 2. HSV-2 miRNA expression analyzed by qRT-PCR. All values are given as fold expression over a noninfected control and are normalized to a cellular miRNA, miR-16. Samples were assayed in triplicate, and average values are displayed, with standard deviations (more ...)
The exceptionally high G·C content of miR-H10 (Fig. ) prevented the design of primers able to specifically detect this HSV-2 miRNA by qRT-PCR. Therefore, we were unable to verify the expression of miR-H10 by any technique other than deep sequencing. However, because we were able to recover this miRNA independently from two different latently infected human samples (Table ), and because pri-miR-H10 is predicted to fold into an RNA hairpin structure closely similar to those observed for other, authentic miRNA precursors (Fig. ), we believe miR-H10 to be a genuine HSV-2 miRNA.
To demonstrate that these HSV-2 miRNAs are indeed functional, Renilla luciferase-based reporter plasmids, bearing a single, artificial target site perfectly complementary to each miRNA inserted into the 3′UTR, were cotransfected into 293T cells along with synthetic RNA duplexes that mimic the predicted viral miRNA duplex intermediates (Fig. ). miR-H7 and miR-H9 were able to inhibit luciferase reporter expression by ~5-fold relative to a negative control reporter plasmid bearing an unrelated target site, while miR-H10 exerted a more modest, ~2-fold inhibitory effect (Fig. ).
FIG. 3. Biological activity of HSV-2 miRNAs. Renilla luciferase reporter plasmids bearing a single 3′UTR target site perfectly complementary to each HSV-2 miRNA were cotransfected into 293T cells along with miRNA duplexes bearing the indicated mature (more ...)
Given the similarity between the HSV-2 and HSV-1 genomes, it is interesting to compare the miRNAs encoded by these two viruses. The previously identified HSV-2 and HSV-1 miR-H4 miRNAs (9
) show only a moderate, ~52% level of sequence identity. However, there is a higher level of sequence identity (~74%) between the two viral miR-H3 miRNAs (Fig. ). Importantly, however, nucleotides 2 to 7 of miR-H3 and miR-H4 miRNAs are not fully conserved between HSV-1 and HSV-2. This region, referred to as the miRNA seed region, is particularly important for miRNA recognition of target transcripts (1
). Of the three novel HSV-2 miRNAs identified in this study, only miR-H7 shows homology to a known HSV-1 miRNA; the two miR-H7 miRNAs are remarkably well conserved, being ~91% identical, including the entire miRNA seed region (Fig. ).
FIG. 4. Comparison of conserved HSV-2 and HSV-1 miRNAs. The HSV-2 miRNAs miR-H3, miR-H4, and miR-H7 are aligned with their HSV-1 homologs, with conserved nucleotides indicated in gray. HSV-2 miR-H9 and miR-H10 do not have any currently known HSV-1 equivalents. (more ...)
Experimental evidence indicates that HSV-1 and HSV-2 miR-H3 and miR-H4 miRNAs, which are all located antisense to the viral ICP34.5
mRNA (Fig. ), downregulate ICP34.5 expression (8
). As mutations in these miRNAs cannot perturb their total complementarity to the antisense ICP34.5 mRNA target, there should be relatively little evolutionary pressure to conserve the miRNA seed regions intact. Conversely, if viral miRNAs downregulate important human mRNA targets, which presumably evolve more slowly than viral sequences, then there might be evolutionary pressure to maintain the integrity of the viral miRNA seed sequences. If this is indeed the case, then it suggests that the HSV-1 and HSV-2 miR-H3 and miR-H4 miRNAs may exclusively function to downregulate the viral ICP34.5 transcript, while the HSV-1 and HSV-2 miR-H7 miRNAs appear more likely to target the same set of currently unknown cellular mRNAs. Importantly, while the miR-H7 miRNAs in both HSV-1 and HSV-2 are transcribed antisense to ICP0
, they share a conserved genomic location opposite the first intron of ICP0
(Fig. ). As miRNAs are believed to function exclusively in the cytoplasm (14
), they are therefore not predicted to downregulate ICP0 protein expression. Interestingly, it has been reported that unspliced forms of ICP0 mRNA can accumulate in the nuclei of neurons with latent HSV-1 infection (6
), and it is therefore possible that miR-H7 functions to prevent the accumulation of these unspliced transcripts in the cytoplasm by inducing their degradation via RNA interference.
Unlike miR-H7, HSV-2 miR-H9 is located antisense to an exonic region of the ICP0 transcript and is therefore predicted to have the ability to inhibit ICP0 expression. This would be similar to the previously reported abilities of both HSV-1 and HSV-2 miR-H2 miRNAs, which are also located antisense to an ICP0 exon, to reduce ICP0 protein expression (9
). It has previously been proposed that downregulation of ICP0 by the viral miR-H2 miRNAs may act to stabilize viral latency (11
), and miR-H9 may also limit the activation of HSV-2 in latently infected neurons in vivo.
Finally, miR-H10 is not located antisense to any known HSV-2 mRNA and is instead located in a short noncoding region between the viral UL15 and UL18 transcripts (Fig. ). While the role of miR-H10 in the HSV-2 life cycle is therefore currently unclear, this may again suggest that this miRNA targets cellular mRNAs for inhibition. It is interesting that HSV-2 miR-H10, which has no HSV-1 equivalent in terms of primary sequence, is currently unique in that it is transcribed from the genomic UL
region, well away from the viral genomic repeats, where all other known HSV-2 and HSV-1 miRNAs originate (2
). It will therefore be of interest to identify the primary miRNA that is processed to give rise to miR-H10 and to determine whether this is indeed a latent transcript.