Epigenetic Regulation of Kaposi's Sarcoma-Associated Herpesvirus Latency by Virus-Encoded MicroRNAs That Target Rta and the Cellular Rbl2-DNMT Pathway ^▿

Cells and plasmids.

293T cells were maintained in Dulbecco's modified Eagle's medium (DMEM) supplemented with 10% fetal bovine serum (FBS) and penicillin-streptomycin. Dermal microvascular endothelial cells (DMVECs) were maintained in EGM-MV microvascular endothelial cell growth medium (cc-3125; Lonza). 293-derived cell lines were all cultured identically to 293 cells, except with the addition of 100 μg/ml hygromycin B for the selection of the wild-type (wt) and recombinant viral genomes. The KSHV BAC36 ΔmiR mutant was generated using recombineering methods, essentially as described previously (36). The primers for PCR generation of the Kan-LoxP cassette for site-specific insertion were CCAGTAGAGTGACCCAGCTGGTTTCCATAAATGGATATACTTCCGGAAAAGAATTCCTGCAGCCCAATTCCG (forward [F]) and GGGAGGAGGAAAAAGTACGCGGTTGTTTACGCAGGGTGCGGTGCTGCCCACTAGTGGATCCCCTCGAGGGACCTA (reverse [R]). For ectopic expression of the KSHV miRNA cluster, a region from nucleotide (nt) 118747 to nt 122305 within the KSHV genome was amplified from BCBL-1 genomic DNA and inserted into pcDNA3. The pcDNA3 plasmids expressing individual KSHV miRNAs and the luciferase reporter plasmids as KSHV miRNA sensors were generated previously (29). The 3′ untranslated region (UTR) of ORF50 was amplified by PCR by using primers GCGAGATCTCATTTGGCAGGGTTACACGTTTAC (F) and GATGGCCGGCCCCTCTAGGGCCGTCATATTCTC (R), and the open reading frame (ORF) of ORF50 was amplified by PCR using primers GCGTCTAGAATGAAAGAATGTTCCAAGCTTG (F) and GATGGCCGGCCCATTTGGCAGGGTTACACGTTTA (R). The PCR products were then digested with BglII/FseI (ORF50 3′ UTR) or XbaI/FseI (ORF50) and inserted into the 3′ UTR of pGL3-Promoter. The 3′ UTR of human Rbl2 was amplified by PCR by using primers GCGGCTAGCGGTTAGTCTCTTGTATTAAAC (F) and GATGGCCGGCCTAATTTATAAATGCTTTATTG (R). The PCR product was then digested with NheI/FseI and inserted into the 3′ UTR of pGL3-Promoter. Site-directed mutation of the Rbl2 3′ UTR was generated by using the QuikChange method (Stratagene, Inc.). The 6-mer miR-K12-4-5p seed match site (TTTAGC) was changed to an XhoI restriction enzyme site (CTCGAG). The Renilla internal control vector pGL4.74[hRluc/TK] was purchased from Promega.

Reconstitution of recombinant virus.

Freshly prepared BAC DNA was introduced into 293T cells by using a Qiagen Effectene transfection kit. Briefly, 1 μg of BAC DNA was used to transfect 293T cells of 40 to 60% confluence in a 60-mm dish. Twenty-four hours after transfection, the cells were subcultured in a 100-mm plate with fresh medium containing 100 μg/ml hygromycin B. When the colonies were visible (usually 10 to 14 days after transfection), the cell colonies were dislodged and seeded into new 10-mm plates. When the monolayer reached 80 to 90% confluence, cells from each flask were split into new 150-mm plates. When the monolayer reached 40 to 60% confluence, cells were induced with 20 ng/ml 12-O-tetradecanoylphorbol-13-acetate (TPA) and 1 mM sodium butyrate for 4 days. The induced culture media were collected and filtered through 0.45-μm filters. The virions were then pelleted by ultracentrifugation on a 25% sucrose cushion at 100,000 × g for 1 h with a Beckman SW28 rotor. The pellets were dissolved in 1% of the original volume of DMEM with 10% FBS and stored at −80°C.

Infection.

293T cells or DMVECs plated in 6-well plates were incubated with concentrated virus plus polybrene (4 μg/ml) for 3 to 4 h at 37°C. The inocula were then removed and replaced with fresh media. For reverse transcriptase PCR (RT-PCR), infected cells were harvested at 72 h postinfection. To make stable 293 cell pools, 293T cells were split into new 100-mm plates at 24 h postinfection with fresh medium containing 100 μg/ml hygromycin B. Green fluorescent protein (GFP) expression was used to monitor infection 2 days after infection. When the colonies were visible (usually 14 to 18 days after infection), the cell colonies were dislodged and seeded into new 100-mm plates. Then, the stable cell pools were ready to be used for genome copy number, luciferase, chromatin immunoprecipitation (ChIP), and methyl-DNA immunoprecipitation (MeDIP) assays.

Quantitative RT-PCR.

Briefly, RNA was isolated from 10⁶ cells by using an RNeasy kit (Qiagen) and then further treated with DNase I. RT-PCR was done as previously described. Real-time PCR was performed using a SYBR green probe with an ABI Prism 7000 according to the manufacturer's specified parameters. The primer sequences for RT-PCR were as follows: for ORF50, CCTTCGGCCCGGAGTCT (F) and CGGTTGCAGTTGCGTATACTCT (R); for LANA, TTACCTCCACCGGCACTCTT (F) and GGATGGGATGGAGGGATTG (R); for vFLIP, GGAGGAGGGCAGGTTAACG (F) and TCTGCGACCTGCACGAAA (R); for vCyclin, CATTGCCCGCCTCTATTATCA (F) and TGACGTTGGCAGGAACCA (R); for viral interleukin-6 (vIL-6), ATGCCGGTACGGTAACAGAG (F) and GGTCGGTTCACTGCTGGTAT (R); for PAN, CGGTGTTTTGGCTGGGTTT (F) and AAACCTTGCCGTCTGGTCACT (R); for DNMT1, GCACCTCATTTGCCGAATACA (F) and TCTCCTGCATCAGCCCAAATA (R); for DNMT3a, GCTGAGAGTGGGATTCATCCA (F) and CCGAGGGAGTCTCCTTTTAGAAA (R); for DNMT3b, TGATGACCGGCCGTTCTT (F) and TCTTGTCGCCAACCTTCATG (R); for β-actin, AACCCAGCCACACCACAAAG (F) and CACTGACTTGAGACCAGTTGAATAAAA (R); and for GFP, AGCAAAGACCCCAACGAGAA (F) and GGCGGCGGTCACGAA (R).

Luciferase assay.

293T or 293-derived cells were transfected with Lipofectamine 2000 (Invitrogen) according to the manufacturer's specifications. 293T cells (2 × 10⁵) were introduced into 800 μg of KSHV miRNA expression vector and 40 ng of reporter DNA with 10 ng Renilla internal control vector pGL4.74 (Promega) and then harvested at 48 h posttransfection. For the luciferase assay of stable 293 cell pools containing BAC, cells were transfected with 200 ng reporter DNA and 10 ng pGL4.74 and then harvested at 48 h posttransfection. Relative luciferase activity was assayed using the Promega dual-luciferase reporter assay system. All data points were averages of results from at least three independent transfections.

Genome copy number assay.

Cells (use approximately 1 × 10 ⁶ cells per sample) were collected and resuspended in 100 μl SDS lysis buffer (1% SDS, 10 mM EDTA, 50 mM Tris, pH 8.0.). After brief sonication, immunoprecipitation (IP) dilution buffer (0.01% SDS, 1.1% Triton X-100, 1.2 mM EDTA, 16.7 mM Tris, pH 8.0, 167 mM NaCl) was added to 1 ml and then incubated with proteinase K for 2 to 3 h at 50°C. Cell lysate (300 μl) was removed and subjected to phenol-chloroform extraction and ethanol precipitation. Precipitated DNA was then assayed by real-time PCR using primers for the ORF50 region of KSHV and normalized by the cellular DNA signal at the actin gene locus. The primers for the genome copy number assay were as follows: for ORF50, CCTTCGGCCCGGAGTCT (F) and CGGTTGCAGTTGCGTATACTCT (R), and for β-actin, AACCCAGCCACACCACAAAG (F) and CACTGACTTGAGACCAGTTGAATAAAA (R).

Antibodies.

The following rabbit polyclonal antibodies were used for ChIP assays: anti-IgG (Santa Cruz Biotechnology), anti-acetylated histone H3 and H4 (Millipore), and anti-trimethyl histone H3K9 and H3K4 (Millipore). Rabbit polyclonal anti-Rbl2 (Abcam) and anti-DNMT1 (New England Biolabs) and mouse monoclonal anti-actin (Sigma) were used for Western blotting.

ChIP and KSHV genome array.

ChIP assays were performed as described previously. For the KSHV genome array, the entire DNA from one IP (10⁶ cells per IP) was mixed with Power SYBR green PCR master mix and aliquoted completely into one 384-well plate for a total of 12 μl per well (9). Quantification of precipitated DNA was determined using real-time PCR and the Absolute Quantification program (ABI 7900HT fast real-time PCR system; Applied Biosystems). Primers were designed using Primer Express, version 2.0 (Applied Biosystems), in conjunction with a batch algorithm that generated 379 real-time primer sets across the KSHV genome, yielding amplification regions spaced, on average, every 370 bp. PCR data were normalized to input values that were quantified in parallel for each experiment. IPs were performed in triplicate for each antibody, and the PCRs were repeated at least three times. The mean for each identical set of IPs was determined, and values that were greater or less than the mean by more than 10% were excluded.

MeDIP.

Methyl-DNA immunoprecipitation (MeDIP) methods have been described previously (16). Briefly, genomic DNA was purified from 5 × 10⁶ cells by using a Promega Wizard genomic DNA purification kit (A1125). The purified genomic DNA was sonicated to produce DNA fragments between 200 and 800 bp. The sonicated DNA was denatured at 95°C for 10 min. Denatured DNA (1 μg) was saved for the total input control. For each IP, 10 μg denatured DNA was diluted in 500 μl IP buffer (0.1% SDS, 1% Triton X-100, 2 mM EDTA, 20 mM Tris, pH 8.0, 150 mM NaCl) with 12.5 μg bovine serum albumin (BSA) and 5 μl anti-methyl cytidine (ab10805-50; Abcam) and incubated overnight at 4°C. To collect the immune complex, 50 μl of a protein G mixture was added to each IP reaction mixture and incubated at 4°C for 2 to 3 h. Beads were centrifuged and washed for 10 min at 4°C with each of the following: 2× IP buffer, 1× IP buffer plus 500 mM NaCl, 1× LiCl wash (250 mM LiCl, 0.5% NP-40, 0.1% Na deoxycholate, 5 mM EDTA, 10 mM Tris-Cl [pH 8.0]), 2× TE buffer (10 mM Tris-Cl [pH 8.0], 1 mM EDTA). The immune complexes were eluted by incubation in 200 μl of TE-1% SDS at 65°C for 30 min, and 200 μl of TE and proteinase K (final concentration, 250 μg/ml) was added to the mixture and incubated overnight at 37°C. DNA was purified by phenol-chloroform extraction and ethanol precipitation in the presence of 20 μg of glycogen and then washed in 70% ethanol, dried, and resuspended in 100 μl of TE. The MeDIP DNA was then used for real-time PCR analysis or the KSHV genome array. The primers for MeDIP were as follows: for KSHV 69580, AGGCCGTCAAACTTCGTTGT (F) and GGCGTCGCAGGAAAACC (R); for the ORF50 promoter, CCCGCCCAGAAACCAGTAG (F) and TGCGGAGTAAGGTTGACTTTTTAA (R); for KSHV 124269, TGGCTTTTTTGGGTGGGTAA (F) and CACCCCACAGACCTGGAGTT (R); for α-globin-2, GACAAGTTCCTGGCTTCTGTGA (F) and GCTACCGAGGCTCCAGCTTA (R); and for β-actin, AACCCAGCCACACCACAAAG (F) and CACTGACTTGAGACCAGTTGAATAAAA (R).

Deletion within KSHV miRNA cluster causes derepression of ORF50 transcription.

The KSHV miRNA consists of a cluster of 12 miRNAs that are contained within the major latency transcripts and adjacent to Ori-Lyt (R) (Fig. 1A). Two v-miRNAs (miR-K12-10 and -12) are embedded with the K12 ORF, while the others (miR-K12-1 through -9 and miR-K12-11) are downstream of the ORF71 gene. A deletion of miR-K12-1 through -9 and -11 was generated by recombineering of KSHV bacmid (BAC36) (36, 44). The deletion was confirmed by restriction enzyme analysis (Fig. 1B), and Southern blotting was used to demonstrate that a single-site Kan-LoxP cassette was inserted during the recombineering process (Fig. 1C). The final construct was further validated by PCR and sequencing across the deleted region (data not shown). The wild-type (wt) and mutant (ΔmiR) bacmids were transfected into 293 cells and assessed for viral gene expression patterns and production of progeny virus. We found that in transfected 293 cells, wt and ΔmiR bacmids expressed similar levels of LANA, vCyclin, K12, and ORF74 mRNA relative to GFP, which is expressed from the bacmid genome and serves as an internal control for bacmid copy number and expression (data not shown). In general, we observed only minor differences in LANA, vCyclin, K12, and ORF74 viral gene expression but found that ΔmiR-transfected cells expressed consistently higher levels of ORF50 mRNA than wt cells (data not shown). To determine if this difference was physiologically relevant, we generated infectious virus from 293 cells transfected with wt or ΔmiR bacmids. Virus particles were normalized by quantification of GFP-producing units and by real-time PCR analysis of virion DNA (data not shown). Similar numbers of GFP positive cells were formed 72 h after infection of 293 cells (7 to 8% GFP⁺) (Fig. 1D and E) or DMVECs (3 to 4% GFP⁺) (Fig. 1F and G). RNA expression was measured by RT-PCR for ORF50, ORF73, and K14 and quantified relative to the level for GFP. As was observed in transfected 293 cells, we found that ORF50 mRNA expression was elevated 2- to 4-fold in ΔmiR bacmid-infected cells relative to the level for cells infected with the wt (Fig. 1D and F). Expression levels of ORF73 (LANA) and K14 were not significantly different. To determine if ORF50 activation corresponded to activation of any other early lytic genes, we assayed mRNA expression of vIL-6 and PAN during infection of 293T cells (Fig. 1E) or DMVECs (Fig. 1G). We found that vIL-6 mRNA levels were increased ∼3- to 4-fold and that PAN mRNA levels were increased ∼6- to 11-fold in BAC36 ΔmiR relative to the levels for the BAC36 wt virus (Fig. 1E and G). In contrast, no change was observed in the mRNA levels of two other latency-associated genes, encoding vFLIP and vCyclin (Fig. 1E and G). To determine if elevated levels of ORF50 correlated with an increase in viral reactivation, we analyzed the average viral DNA copy number in 293 cell pools carrying either the BAC36 wt or BAC36 ΔmiR (Fig. 1H). Real-time PCR analysis of KSHV genome DNA relative to cellular actin DNA did not reveal any significant increase in genome copy number in stable cell lines carrying the ΔmiR mutant relative to the level for cells infected with wt BAC36 virus. This suggests that miRNA deletion (ΔmiR) does not cause robust lytic infection, despite the fact that ORF50 mRNA levels are elevated.

Transcriptional silencing and PTS of ORF50 by v-miRNA.

Most miRNAs are thought to function through posttranscriptional silencing (PTS) of target genes mediated by sequence hybridization with 3′ UTRs. To test whether ORF50 was subject to PTS by v-miRNA, we fused the ORF50 3′ UTR and ORF50 coding sequence to a luciferase reporter plasmid. These reporter plasmids were then tested for their relative levels of expression of luciferase in 293 cells stably infected with wt or ΔmiR viral DNA (Fig. 2A). The luciferase control plasmid (pGL3 Ctrl) expressed near-identical levels in 293 cells containing the wt and ΔmiR. In contrast, luciferase fused to the ORF50 3′ UTR or to the ORF50 coding sequence expressed ∼2-fold-higher levels of luciferase in ΔmiR-containing cells (Fig. 2A). This suggests that KSHV v-miRNA is capable of PTS directed against the ORF50 3′ UTR and coding sequence.

ORF50 mRNA expression levels were elevated in the ΔmiR mutant relative to the levels for cells containing the wt bacmid (Fig. 1). To determine if a single miRNA was responsible for ORF50 repression, we examined a panel of KSHV miRNA expression constructs for their ability to repress ORF50. We found that ectopic expression of miR-K12-5 caused an ∼2-fold reduction in ORF50 mRNA relative to the level for vector control-transfected cells (Fig. 2B). This indicates that miR-K12-5 is capable of repressing ORF50 mRNA when introduced in ΔmiR BAC36-containing cells. To determine if miR-K12-5 was capable of inhibiting translation of the ORF50 3′ UTR or the ORF50 coding sequence, we tested its ability to inhibit the luciferase fusion constructs. We found that miR-K12-5 could inhibit luciferase fused to the ORF50 3′ UTR but could not control the vector or ORF50 coding sequence (Fig. 2C). An analysis of the seed sequence of miR-K12-5 revealed several low-probability alignments with the ORF50 3′ UTR, leaving open the possibility that its mechanism of PTS is indirect. Taken together, these results lead us to conclude that KSHV miRNA can suppress ORF50 through posttranslational mechanisms mediated by the ORF and the 3′ UTR and that miR-K12-5 can partially reconstitute this activity.

ΔmiR genomes have altered histone modifications.

KSHV miRNA may also control ORF50 expression by regulating gene expression and chromatin organization. To investigate whether viral miRNAs have any effect on the viral chromatin structure, we first examined the histone modification patterns of wt and ΔmiR genomes in stable infected 293 cells (Fig. 3). Genome-wide histone modifications were examined by a ChIP assay using a 384-well real-time PCR array covering the genome at a density of ∼400-bp intervals. We assayed histone H3 K4 and K9 methylation patterns, which have provided highly reliable ChIP data in our hands and reflect a general difference in euchromatic (H3 K4 methylation) and heterochromatic (H3 K9 methylation) chromatin domains. Interestingly, we found that the patterns of H3 K4 and K9 were mostly nonoverlapping in cells containing wt BAC36 (compare Fig. 3A and C). A cluster of H3 K4 methylation was found downstream of the major latency transcripts (genome coordinates ∼99,721 to 114,840 in KSHV), and a second cluster was observed at positions ∼15481 to 22320. This second cluster may reflect the expression of vIL-6, which has been shown to be expressed in most latently infected cells. In contrast, H3 K9 methylation was most elevated in a large cluster spanning the regions between kb ∼40 and 80, and additional peaks were detected at positions ∼2521 to 14400 and ∼127876 to 132120. Comparison with ΔmiR genomes revealed a very similar pattern of H3 K4 methylation, although a general reduction in ChIP efficiency was observed across the entire genome. In contrast, the H3 K9 methylation pattern showed a striking loss of signal in the region between kb 2 and 14 in the ΔmiR genomes relative to the level for the wt genomes. This region is in close proximity to the TRs, which have been implicated in heterochromatin formation. We also asked whether histone acetylation was altered significantly at regions where histone methylation was different (Fig. 4). We found that histone H3 and H4 acetylation levels were elevated at genome positions 6592 (near TR), 71486 (ORF50 promoter region), and 124269 (miRNA flanking sequence) but not at position 64417 (Fig. 4A and B). The increase in histone acetylation at these regions mirrors the general loss of histone H3 K9 methylation in the ΔmiR genomes. These findings suggest that miRNA deletion causes an altered pattern of histone H3 K9 methylation and histone H3 acetylation at multiple regions throughout the viral genome.

ΔmiR genomes have reduced DNA methylation.

Histone modifications can be coupled with DNA methylation, and some microRNAs have been implicated in regulation of DNA methylation. We therefore investigated whether DNA methylation patterns were altered in ΔmiR genomes relative to the patterns in wt genomes (Fig. 5A). The genome-wide DNA methylation pattern was assessed using a methyl-cytosine DNA immunoprecipitation (MeDIP) assay followed by real-time PCR analysis using the KSHV genome array. We found that cells containing wt BAC36 had specific patterns of DNA methylation, with peak clusters occurring at kb 2 to 6 and at kb 124 to 127. Interestingly, the peak region comprising kb 124 to 127 overlaps the open reading frames of the latency transcripts for ORF73 and ORF72 but does not include the promoter of this gene. Remarkably, almost all of the DNA methylation peaks corresponded to histone H3 K9 methylation peaks. When CpG methylation levels were compared between wt and ΔmiR genomes, we found that miR deletion caused a striking loss of DNA methylation (Fig. 5B). Loss of the MeDIP signal was reexamined by sequence-specific PCR using different primer sets and different 293 cell clones of cells containing the wt and ΔmiR (Fig. 5C). Consistently, the MeDIP signal was reduced ∼4- to 8-fold in ΔmiR genomes relative to the wt level. We also found that a cellular gene locus within the α-globin gene, which is known to be highly methylated in nonerythroid cells, was reduced ∼6-fold for MeDIP signal in cells containing the ΔmiR mutant relative to the level for wt genomes (Fig. 5D). An ∼2-fold loss of DNA methylation was observed at the cellular actin locus (Fig. 5D, right). These data suggest that KSHV miRNAs are required for efficient DNA methylation of both viral and cellular DNA in latently infected cells.

miR-K12-4-5p represses Rbl2 and upregulates DNA methyltransferases.

Previous studies with mice have identified a miRNA cluster that affects global DNA methylation patterns (3). In that study, they found that the miRNA targeted a cellular transcriptional repressor, Rbl2, that regulates DNMT3a and -3b mRNA expression. To explore the possibility that KSHV miRNA may function through a similar mechanism, we used bioinformatic methods to search for a KSHV miRNA seed sequence that might target the 3′ UTR of human Rbl2 mRNA (29). We found a high-probability hit for KSHV miR-K12-4-5p and the 3′ UTR of human Rbl2 mRNA (Fig. 6A). To test whether miR-K12-4-5p had any effect on Rbl2 protein levels, we examined the ability of the miR-K12-4-5p expression vector to reduce Rbl2 in ΔmiR-containing 293 cells (Fig. 6B). We found that ectopic expression of miR-K12-4-5p showed a small (∼2-fold) decrease in Rbl2 relative to the level for the control expression vector. Interestingly, we observed a corresponding increase (∼1.4-fold) in DNMT1 protein levels and no significant change in actin protein levels (Fig. 6B, lower panels). To determine if miR-K12-4-5p targeted Rbl2 mRNA directly, we assayed the effect of miR-K12-4-5p expression on luciferase reporter constructs containing either a fusion of the wt Rbl2 3′ UTR or a mutant Rbl2 3′ UTR with a 5-bp substitution mutation in the putative seed target sequence (Fig. 6C). We found that ectopic expression of miR-K12-4-5p reduced the luciferase activity of the wt 3′ UTR fusion but not the mutant fusion or the pGL3 control. This suggests that the 3′ UTR of Rbl2 is a direct target of miR-K12-4-5p. To determine if miR-K12-5p depletion of Rbl2 affected Rbl2-regulated genes involved in DNA methylation, we assayed the mRNA expression of DNMT1, DNMT3a, and DNMT3b in cells transfected with miR-K2-4-5p (Fig. 6D). Examination of mRNA levels by RT-PCR revealed that miR-K12-4-5p-transfected cells resulted in an ∼4-fold increase in DNMT3a and modest increases in DNMT1 and DNMT3b mRNA levels (Fig. 6D). To determine if the changes in DNA methyltransferase levels were functionally significant, we examined the effects of miR-K12-4-5p expression on MeDIP at several viral and cellular loci (Fig. 6E). We observed that ectopic expression of miR-K12-4-5p in ΔmiR-containing cells caused a 2- to 5-fold increase in MeDIP signal relative to the level for the control expression vector at the KSHV loci comprising position 124269 and cellular α-globin-2. Taken together, these results indicate miR12-4-5p can reduce Rbl2 protein levels, increase DNA methyltransferase mRNA levels, and increase cytosine DNA methylation at a several viral and cellular loci.