Herpes simplex virus 1 (HSV-1) infection and stress responses disrupt transcription termination by RNA Polymerase II (Pol II). In HSV-1 infection but not upon salt or heat stress, this is accompanied by a dramatic increase in chromatin accessibility downstream of genes. Here, we show that the HSV-1 immediate-early protein ICP22 is both necessary and sufficient to induce downstream open chromatin regions (dOCRs) upon disrupted transcription termination. This was matched by a marked ICP22-dependent loss of histones downstream of affected genes consistent with impaired histone repositioning in the wake of Pol II. Efficient knock-down of the ICP22-interacting histone chaperon FACT, though insufficient to induce dOCRs in ΔICP22 infection, increased dOCR induction upon wild-type HSV-1 infection. Interestingly, this was accompanied by a marked increase in chromatin accessibility within gene bodies. We propose a model in which allosteric changes in Pol II composition downstream of genes and ICP22-mediated interference with FACT activity explain the differential impairment of histone re-positioning in the wake of Pol II observed in HSV-1 infection.
Article
The HSV-1 ICP22 protein selectively impairs histone repositioning upon Pol II transcription downstream of genes
https://doi.org/10.21203/rs.3.rs-998249/v1
This work is licensed under a CC BY 4.0 License
Journal Publication
published 31 Jul, 2023
Version 1
posted
You are reading this latest preprint version
Herpes simplex virus 1
ICP22
transcription termination
chromatin conformation
histone repositioning
Productive Herpes Simplex Virus type 1 (HSV-1) infection induces a profound shut-off of host gene expression by targeting multiple steps of RNA metabolism (1). This curtails antiviral host responses and facilitates efficient productive infection (2). Disruption of transcription termination (DoTT) of cellular genes significantly contributes to HSV-1-induced host cell shut-off (3). DoTT commonly results in read-through transcription extending for tens to hundreds of thousands of nucleotides beyond poly(A) sites and into downstream genes. The viral immediate early protein ICP27 plays a direct, bimodal role in HSV-1-induced DoTT (4). On the one hand, ICP27 interacts with and disrupts the essential mRNA 3’-end processing factor CPSF thereby inducing the assembly of a dead-end 3’ processing complex that is unable to cleave mRNA 3’ ends. On the other hand, ICP27 acts as a sequence-dependent activator of mRNA 3’ processing by binding to viral (and some host) transcripts thereby restoring CPSF activity and 3’ mRNA cleavage. Interestingly, poly(A) read-through transcription is accompanied by a dramatic increase in chromatin accessibility downstream of the affected poly(A) sites (5). Normally, open chromatin is only observed around gene promoters as well as in the gene bodies of highly expressed genes. During HSV-1 infection, chromatin accessibility selectively increases downstream of genes with strong read-through transcription indicative of impaired histone repositioning in the wake of RNA polymerase II (Pol II) downstream of affected poly(A) sites. Disrupted transcription termination with Pol II transcription extending far downstream of gene 3’ends (DoGs) is also observed upon cellular stress responses (6, 7), influenza virus infection (8, 9) and cancer (10). However, downstream open chromatin regions (dOCRs) do not arise upon salt or heat stress (5), indicating that transcription read-through is either not sufficiently strong for dOCR induction in stress responses or that specific viral factors are involved.
During the process of transcription, Pol II removes and subsequently repositions histones to facilitate efficient transcription elongation while maintaining chromatin architecture. The two transcription elongation factors and histone chaperons FACT (comprised of SSRP1 and SPT16) and SPT6 play a key role in this process (reviewed in (11)). The histone-chaperoning activity of the FACT complex facilitates both removal and reassembly of histones in the wake of actively transcribing Pol II. Distinct regions of FACT interact with both the H2A-H2B dimer and the (H3-H4)2 tetramer, and can promote displacement of H2A-H2B from nucleosomes (12). In particular, FACT mediates both histone eviction and repositioning at the early stages of transcription elongation (13). SPT6 directly interacts with histones and assembles nucleosomes in vitro (14). This activity is required for the maintenance of a chromatin structure that prevents improper usage of cryptic promoter elements, suggesting that SPT6 also reassembles nucleosomes in the wake of Pol II (14, 15). Interestingly, the viral ICP22 protein interacts with the FACT complex and relocalizes it to viral replication compartments (16). ICP22 (encoded by the HSV-1 US1 gene) is an important, but non-essential, multifunctional viral 63-kDa polypeptide conserved in Alphaherpesviruses that is expressed with immediate early kinetics. HSV-1 mutants lacking ICP22 do not form plaques in skin and lung fibroblasts but replicate in Vero or BHK cells (17). ICP22 features a structurally-conserved core domain flanked by poorly-conserved, intrinsically disordered regions that are extensively phosphorylated by the viral UL13 protein kinase, and, to a lesser extent, by the US3 protein kinase (18, 19). ICP22 is also phosphorylated by yet unidentified cellular kinases and nucleotidylated by casein kinase II (20, 21). It is thus very likely that ICP22 exerts multiple different functions throughout productive virus infection that are governed by post-translational modifications. Accordingly, ICP22 has been shown to interact with a variety of different cellular proteins including cyclin-dependent kinase 9 (CDK9; pTEFβ) (22–24), cell division cycle 25C (CDC25C) (25) as well as FACT (16). Notably, ICP22 facilitates the recruitment of FACT and SPT6 to replication compartments (RCs) and to the transcribing Pol II complex (16). Very early during productive infection, ICP22 induces the formation of virus-induced chaperone-enriched (VICE) domains, which may aid proper folding of newly synthesized viral proteins (26). It may by itself also function as a virally encoded co-chaperone (J-protein/Hsp40) by interacting with heat shock protein 70 (Hsc70) to prevent aggregation of misfolded proteins (27). During productive infection, ICP22 and the viral UL13 protein kinase mediate an intermediate phosphorylation of the carboxyl terminal domain (CTD) of Pol II, termed Pol IIi (11, 12). ICP22 also alters the activity of the cell cycle component CDK1 to enhance viral late (L) gene expression (13). This effect of ICP22 on key cell cycle regulators may explain the cell type-dependent growth defect of ICP22 null mutants (28). Finally, it also plays an important role in HSV-1 nuclear egress by interacting with the viral nuclear egress complex (UL31 and UL34 proteins) (29).
Here, we show that the viral ICP22 protein is necessary to induce dOCRs in HSV-1 infection. Furthermore, ectopic expression of ICP22 is sufficient for induction of dOCRs upon disruption of transcription termination and poly(A) read-through transcription induced by either ectopic expression of ICP27 or salt stress. Read-through transcription and induction of dOCRs downstream of genes was associated with a notable histone depletion downstream of gene 3’ends. Efficient knock-down of the histone chaperones FACT not only enhanced dOCR induction but also increased chromatin accessibility within gene bodies in an ICP22-dependent manner. It thereby alleviated the selective, DoTT-dependent increase in chromatin accessibility downstream of genes. We propose a model in which functional inhibition of FACT by the viral ICP22 protein and allosteric changes in Pol II composition downstream of genes result in selectively impaired histone repositioning in the wake of Pol II.
Viral late gene expression is not required for dOCR induction
To identify the viral gene responsible for induction of dOCRs during HSV-1 infection, we first assessed whether viral genome replication and thus viral late gene expression was required for the induction of dOCRs. We infected primary human fibroblasts (HFFFs) with HSV-1 strain 17 for 8 h and strain F for 8 and 12 h in presence or absence of the viral DNA polymerase inhibitor phosphonoacetic acid (PAA) and performed ATAC-seq (n=2). The length of dOCRs was quantified for 4,162 genes that exhibited no read-in transcription originating from read-through transcription of an upstream gene. In uninfected cells, the general absence of downstream open chromatin is reflected by only short (if any) dOCRs for the vast majority of cellular genes (Fig. 1a). Both wild-type (WT) strains 17 and F induced dOCRs. However, this was less prominent for strain F (Fig. 1a, example in Sup. Fig. 1a), consistent with a lower extent of read-through transcription upon infection with this strain (4). Strikingly, inhibition of viral DNA replication by PAA substantially increased dOCR length for both virus strains while PAA treatment had no effect on uninfected cells. Of note, PAA treatment of HSV-1-infected cells resulted in a much higher percentage of cellular reads in the ATAC-seq data due to the inhibition of viral DNA replication. To exclude that the increase in dOCR induction by PAA was simply due to a greater sequencing depth on the cellular genome, we performed down-sampling of the respective sequencing libraries so that all samples had approximately the same number of reads mapping to the human genome. Even after down-sampling, significantly longer dOCRs were observed upon PAA treatment for both strains (Sup. Fig. 1b). We thus hypothesize that enhanced induction of dOCRs upon inhibition of viral late gene expression by PAA treatment resulted from the reduced host shut-off and substantially higher host transcriptional activity both within and downstream of genes (30). Furthermore, we conclude that viral late gene expression is not required for dOCR induction.
dOCRs arise upon strong transcriptional activity downstream of genes
To quantify the role of individual virus gene null mutants regarding the induction of dOCRs, we defined a subset of cellular genes that showed strong and consistent dOCR induction upon infection with both virus strains. To this end, we performed an unsupervised clustering analysis of dOCR lengths in mock, WT and WT+PAA infection (Fig. 1b). This identified a total of nine different gene clusters; three of these (orange to red bars) showed dOCR induction while six (blue and green bars in Fig. 1b) exhibited no induction of dOCRs during infection. Cluster 5 (305 genes, dark red bar in Fig. 1b) exhibited the strongest induction of dOCRs independent of virus strain. In contrast, Clusters 2 (290 genes, orange) and 6 (701 genes, red) showed weaker dOCR induction, which was nevertheless clearly visible upon PAA treatment for both strains. Notably, these three clusters differed in the presence of dOCRs prior to infection. Here, Clusters 5 and 6 already exhibited short dOCRs prior to infection that strongly increased in length and thus extended significantly further downstream upon infection.
To investigate the cause of these differences in the induction of dOCRs between the different gene clusters, we made use of our previously published 4sU-seq time-course of the first 8 h of WT strain 17 infection (3). We analyzed (i) the percentage of read-through transcription (difference between infected and mock in the following ratio: FPKM within 5 kb downstream of the gene 3’end / gene FPKM x 100) and (ii) the absolute extent of transcriptional read-through activity (FPKM within 5 kb downstream of the gene 3’end, denoted as ‘downstream FPKM’). Interestingly, we only observed minor differences in the percentage of read-through transcription between clusters, with Cluster 2 (orange, weak dOCR induction) showing the highest percentage of read-through transcription of all clusters (Fig. 1c). In contrast, downstream transcriptional activity was substantially greater in Cluster 5 (dark red, highest dOCR induction) than in all other clusters (Fig. 1d). This was matched by higher gene expression levels (gene FPKM) in Cluster 5 already prior to infection (Sup. Fig. 1c). Cluster 5 thus comprises the most strongly expressed cellular genes with the highest absolute levels of downstream transcription due to DoTT. Clusters 2 and 6 also showed slightly elevated downstream transcriptional activity compared to clusters without apparent induction of dOCRs (Fig. 1d). Accordingly, the absolute extent of transcriptional activity downstream of the respective genes rather than the percentage of read-through transcription determines the extent of dOCR induction. This also explains why genes with a high percentage of read-through transcription but relatively low gene expression (Cluster 2) as well as genes with a moderate percentage of read-through transcription, but higher gene expression (Cluster 6) exhibit some induction of dOCRs. Strikingly, induction of open chromatin predominantly occurred downstream of affected gene 3’ ends although transcription of the respective gene bodies was as least as high as downstream transcription. Notably, dOCRs can also extend into the gene body of downstream genes upon read-in transcription (as exemplified by read-in transcription from the SRSF3 gene into the CDKN1A gene (see Suppl. Fig. 1a). We conclude that induction of dOCRs selectively arises as a consequence of HSV‑1-induced DoTT when a high level of transcriptional activity extends downstream of genes beyond affected poly(A) sites.
ICP22 is required for the induction of dOCRs
To identify the viral gene responsible for dOCR induction, we infected HFFF with a range of single gene deletion mutants (n=2). This included null mutants of the immediate early genes ICP0, ICP22 (R325) and ICP27 as well as of the virion host shut-off protein (vhs). Due to the attenuated progression of DICP0 and DICP22 infections, ATAC-seq was performed at 12 h p.i. for these mutants. For WT and the other mutants, infection was performed at 8 h p.i. For DICP22 infection, we also included PAA treatment as we hypothesized that the loss of serine 2 phosphorylation of RNA polymerase II (Ser-2P) during HSV-1 infection might play a role in induction of dOCRs. This loss of Ser-2P results from the combined effects of both an ICP22-dependent and a unknown ICP22-independent, viral late gene-dependent mechanism (31). To analyze the induction of dOCRs in the single-gene deletion mutants, we focused on genes in Cluster 5, which showed the most and most consistent dOCRs. Infection with DICP0 and Dvhs still led to the induction of dOCRs at levels comparable to WT strain 17 (Fig. 2a). Furthermore, DICP27 infection still induced dOCRs albeit to a lesser extent, consistent with the residual, presumably stress-induced read-through transcription observed in DICP27 infection (4). Strikingly, infection with the ICP22 deletion mutant completely abolished the induction of dOCRs. Here, even PAA treatment did not result in any detectable dOCR induction. This was confirmed by additional ATAC-seq experiments with 8 h and 12 h p.i. ± PAA treatment both for the DICP22 mutant and its parental WT strain F (n=2) (Fig. 2b) as well as using DICP22 mutants derived from KOS1.1 (32) and BAC-derived WT strain 17 (n=2, Supp. Fig. 2a, b). Total RNA-seq (n=2) performed in parallel to ATAC-seq for mock, WT strain F and DICP22 infection (8 and 12 h p.i.) confirmed the presence of extensive DoTT and strong downstream transcriptional activity in DICP22 infection both with and without PAA treatment. These data also confirmed the correlation of dOCR induction with downstream transcriptional activity for strain F, which increased upon PAA treatment (Sup. Fig. 2c-e). In contrast, no induction or dOCRs was observed in DICP22 infection even for genes with high downstream transcriptional activity (Fig. 2d, Sup. Fig. 2f-h). We conclude that ICP22 is required for induction of dOCRs and that inhibition of viral late gene expression enhances the induction of dOCRs due to a reduced shut-off of host transcription.
To clarify the role of ICP27 in induction of dOCRs, we correlated downstream transcriptional activity and dOCR lengths of all analyzed genes under study using our previously published 4sU-seq data for all mutant viruses (4). Infections with WT strain 17, DICP0 and Dvhs all showed a positive correlation (Fig. 2e, Sup. Fig. 2i, j). This was also observed for DICP27 infection, however less pronounced (Fig. 2f). Although ICP27 is responsible for the majority of HSV-1-induced DoTT, residual, presumably stress-induced transcriptional activity downstream of genes is thus sufficient for the induction of dOCRs. In contrast, this was absent for DICP22 infection irrespective of PAA treatment (Sup. Fig. 2k, l), which confirmed the above results in independent ATAC- and RNA-seq experiments. We conclude that neither ICP0, vhs nor ICP27 are required for induction of dOCRs during HSV-1 infection.
ICP22 is sufficient for induction of dOCRs upon read-through transcription
Our analyses so far indicate that both ICP22 expression and transcription downstream of genes are necessary for induction of dOCRs. To test whether ectopic ICP22 expression was sufficient to induce dOCRs in the presence of downstream transcription, we generated telomerase-immortalized human foreskin fibroblasts (T-HFs) that express either ICP22 in isolation (T-HFs-ICP22 cells) or in combination with ICP27 (T-HFs-ICP22/ICP27 cells) upon doxycycline (Dox) exposure (Sup. Fig. 3a-d). While co-induction of ICP27 served to induce downstream transcriptional activity in the T-HFs-ICP22/ICP27 cells, we exposed the T-HFs-ICP22 cells (with and without Dox-exposure) to 2 h of salt stress to induce transcription downstream of genes (DoGs). We subsequently analyzed induction of dOCRs by Omni-ATAC-seq and in parallel quantified downstream transcriptional activity by total RNA-seq for the same samples (n=2). Omni-ATAC-seq represents a recent improvement in the ATAC-seq protocol published during the course of this study, which improves signal-to-background ratios and reduces the amount of contaminating mitochondrial reads (33, 34). Expectedly, in the absence of read-through transcription, Dox-induced ICP22 expression did not induce dOCRs. In contrast, Dox-induced co-expression of ICP27 and ICP22 resulted in extensive induction of dOCRs (Fig. 3a, b, Sup. Fig. 3e), which correlated with the extent of transcription downstream of genes (Fig. 3c). This also confirmed previous findings from HeLa cells which demonstrated that ectopic expression of ICP27 is sufficient to disrupt transcription termination (4). Salt stress alone induced read-through transcription, albeit to a lower level than ICP27, and was not sufficient for induction of dOCRs (Fig. 3a, b, Sup. Fig. 3e, f), thereby confirming our previous results obtained in primary HFFF cells (5). Consistent with the observed induction of dOCRs in cells infected with an ICP27-null mutant, salt stress also triggered induction of dOCRs in presence of ICP22. While this did not reach the same degree as ICP27 and ICP22 co-expression, it is consistent with the lower downstream transcriptional activity upon salt stress (Fig. 3d). The slightly higher dOCR lengths in the unstimulated T-HF-ICP22/ICP27 cells compared to unstimulated T-HFs-ICP22 cells with salt stress presumably results from weak leaky ICP22 and ICP27 expression. We conclude that ICP22 is sufficient for induction of dOCRs upon either ICP27- or salt stress-induced read-through transcription.
Induction of dOCRs is associated with a loss of histones downstream of genes
We hypothesized that induction of dOCRs was due to impaired histone repositioning in the wake of Pol II transcription downstream of genes. To test this hypothesis, we first analyzed changes in genome-wide occupancy of histone H3, as well as the two major histone marks associated with heterochromatic regions (H3K27me3) or active transcription (H3K36me3) in uninfected and WT strain 17 infected (8h p.i.) cells by ChIPmentation (n=2 or 3; see methods for details). Interestingly, a metagene analysis for genes with strong induction of dOCRs (Cluster 5) did not reveal any significant virus-induced changes in occupancy of H3 or H3K27me3 up- or downstream of the transcription termination site (TTS) (Fig. 4a,b; Sup. Fig. 4a,b). For H3K36me3, a reduction during HSV-1 infection was observed in the 5 kb upstream of the TTS (Fig. 4c), which was partly statistically significant (Wilcoxon test, p<0.05) and much more pronounced for genes with strong dOCR induction than for genes without dOCR induction (= all genes except for Clusters 2, 5 and 6, Sup. Fig. 4c).
While an increase in chromatin accessibility should still be readily detectable by ATAC-seq even when only a very small percentage of cells is still transcribing the genes by 8 h p.i. (and thus induce dOCRs), a loss in histone occupancy would likely be masked by the cells not transcribing the respective genes anymore. To prevent the virus-induced sequestration of Pol II to replication compartments and alleviate the reduction in host transcriptional activity (30), we repeated the ChIPmentation experiments upon PAA treatment (mock and WT strain 17 infection at 8 h p.i.; both including 8 h of PAA treatment). This time, we also analyzed histones H1 and H4 in addition to H3, H3K27me3 and H3K36me3. Quality of the histone ChIPmentation data was confirmed by metagene analyses showing the expected occupancy profiles with a strong depletion of H1, H3 and H4 at transcription start sites (TSSs) compared to gene bodies and downstream regions (Fig. 4d,e, Supp. Fig. 5a-d). Strikingly, starting around or slightly upstream of the TTS and extending for about 25 kb downstream of the TTS, all three histones showed a reduction in coverage in WT strain 17 infection with PAA treatment compared to the uninfected cells for genes with strong induction of dOCRs (Cluster 5) (Fig. 4d,e, Supp. Fig. 5a). While this was only significant (Wilcoxon test, p<0.05) for some of these positions, it was observed consistently for all three histones. Furthermore, no such reduction was observed for genes without induction of dOCRs (Supp. Fig. 5b-d). A combined metagene analysis of H1, H3 and H4 showed the same trend with statistical significance observed for a larger number of positions in the region downstream of the TTS (Supp. Fig. 5g,h). An even more downstream (>50 kb) increase in H1, H3, and H4 observed in WT strain 17 infection is likely due to the normalization procedure for individual gene curves in the metagene analysis. Here, occupancy profiles for each gene and replicate were first normalized to a sum of 1 before averaging across genes and replicates to avoid biases due to differences in gene expression (see Materials and Methods). Thus, these profiles represent relative distributions of histones across the gene and downstream regions. The histone modification marks exhibited distributions consistent with their association with heterochromatic or transcribed regions, with H3K27me3 being depleted on gene bodies (Fig. 4f, Supp. Fig. 5e) and H3K36me3 (Fig. 4g, Supp. Fig. 5f) being enriched. In both cases, a reduction of histone modification marks was observed for genes with strong induction of dOCRs selectively in the region from the TTS or upstream of the TTS to around 25 kb downstream of the TTS (Fig. 4f,g). This is consistent with the similar reduction for histone H3 and was not observed for genes without induction of dOCRs (Supp. Fig. 5e,f). Interestingly, PAA treatment even enhanced the HSV-1-induced reduction in H3K36me3 within the 5 kb upstream of the TTS selectively for genes with dOCRs (Cluster 5). This was unexpected as PAA treatment efficiently prevents the global virus-induced loss of Pol II transcription of cellular genes. As no such loss was observed for H3K27me3 this does not result from a global loss of H3 in this region. The underlying molecular mechanism remains unclear.
To confirm that the alterations in histone abundance downstream of the TTS were specifically associated with induction of dOCRs, we performed ChIPmentation for H1 in mock, WT and DICP22 infection (both in strain F) at 8 h p.i. (Fig. 4h). This confirmed the observations for WT strain 17 infection with a selective reduction in H1 occupancy within the 25 kb downstream of the TTS in WT infection compared to both mock and DICP22 infection. In contrast, H1 occupancy profiles for mock and DICP22 infection were highly similar. Again, no such effect was observed for genes without induction of dOCRs (Suppl. Fig. 5i). Statistical analysis confirmed the differences downstream of the TTS to be significant (Wilcoxon test, p<0.05) at several positions when comparing WT and mock as well as WT and DICP22 infection (Supp. Fig. 5j, k). No significant differences were observed between mock and DICP22 infection (Supp. Fig. 5l). We conclude that induction of dOCRs in WT HSV-1 infection is associated with alterations of histone occupancy downstream of affected genes, which are absent in DICP22 infection. This is consistent with a model in which ICP22 interferes with histone repositioning in the wake of Pol II passage.
Depletion of FACT increases chromatin accessibility in an ICP22-dependent manner
The two histone chaperones FACT (SPT16/SSRP1) and SPT6 play a key role in the re-assembly of nucleosomes after the passing of Pol II (11). During the course of this study, both histone chaperones were shown to be recruited into viral replication compartments by the viral ICP22 protein (16). As their functional inhibition by ICP22 might explain the induction of dOCRs, we investigated whether depletion of either of the two factors would restore the induction of dOCRs upon infection with an ICP22-null mutant. For this purpose, we generated T-HF cells with doxycycline-inducible shRNA knock-down of SSRP1 and SPT6 (35). For both cellular proteins, efficient knock-down was achieved with two different shRNAs after three days of doxycycline treatment (1 µg/ml) (Sup. Fig. 6a, b). Knock-down of SSRP1 resulted in a concomitant loss of its interaction partner SPT16, consistent with previous reports (36). Furthermore, knock-down of SPT6 not only significantly reduced SSRP1, but also resulted in a modest reduction of Pol II levels which further declined upon HSV-1 infection independently of ICP22. Nevertheless, knock-down of neither of the two cellular proteins had any discernable effect on viral gene expression upon high MOI infection (Sup. Fig. 6c, d). This implies that both proteins are not required for productive HSV-1 infection. To assess whether the respective histone chaperons played any role in HSV-1-mediated induction of dOCRs, we performed Omni-ATAC-seq on WT strain F-, DICP22- and mock-infected cells (with PAA treatment in all cases). Omni-ATAC-seq was performed on cells from the same experiment as utilized for the Western blots in Sup. Fig. 6a-d. Neither depletion of SPT6 nor FACT (SSRP1) resulted in significant dOCR induction in DICP22 infection or mock (Fig. 5a-c). While knock-down of SPT6 had no effect on the extent of dOCR induction by WT HSV-1, depletion of FACT significantly increased dOCR induction (Fig. 5a). Down-sampling of the ATAC-seq data to the same number of cellular reads per sample confirmed that this did not result from differences in virus replication (Fig. 5b, Suppl. Fig. 6e). Strikingly, knock-down of FACT also led to an increase in chromatin accessibility within gene bodies for several hundred genes (examples in Fig. 5d and Suppl. Fig. 7a-c; 282 genes with ≥2-fold increase, Fig. 5e). Notably, this preferentially affected genes within Cluster 5 (strong dOCR induction), with 22% of genes showing an increase in chromatin accessibility within gene bodies compared to ≈8.5% for Clusters 2 and 6 (weak dOCR induction) and 4.5% for genes without dOCR induction. Notably, increased chromatin accessibility within gene bodies was now also observed for genes without DoTT and corresponding dOCRs. Interestingly, these genes often already showed a slight increase in gene body chromatin accessibility in WT infection without FACT depletion (Suppl. Fig. 7c). The HSV-1-induced increase in chromatin accessibility is thus not fully restricted to transcribed regions downstream of genes. Knock-down of FACT in mock- and DICP22-infected cells also resulted in a modest increase in chromatin accessibility within gene bodies for 202 and 164 genes, respectively (Fig. 5f,g). However, this was much less prominent than observed in WT infection of FACT knock-down cells. Interestingly, for both mock and DICP22-infected cells, this was more commonly observed for Cluster 5 genes than for genes without dOCR induction (mock: 13.1% vs. 4.2% of genes; DICP22: 23.3% vs 1.7%). We conclude that transcription of Cluster 5 genes may be more susceptible to FACT depletion and thus dOCR induction. In summary, FACT knock-down not only enhanced the induction of dOCRs in WT HSV-1 infection but also alleviated its restriction to regions downstream of genes with read-through transcription (Fig. 5e). These data directly implicate FACT in the ICP22-mediated impairment of histone repositioning in the wake of Pol II.
Both HSV-1 infection and various stressors trigger extensive transcription downstream of genes but dOCRs only arise in HSV-1 infection. Our study explains this difference by demonstrating that the viral ICP22 protein is necessary for the induction of dOCRs in HSV-1 infection. Importantly, dOCR formation critically depended on the extent of transcriptional activity downstream of genes. This results in increased dOCR induction upon inhibition of viral DNA synthesis (and consequently of viral late gene expression), which preserves host transcriptional activity. Strong dOCR induction in absence of viral DNA replication also implies that deprivation of cellular factors by their recruitment to viral replication compartments is not responsible for induction of dOCRs.
The ICP22 mutants, which we employed in this study, either lack the ICP22 protein completely (KOS1.1 and strain 17) or just its C-terminal 200 aa (strain F, R325). Concordant data between the different mutants implies that the N-terminus of ICP22, which is important for VICE domain formation, is not required for the induction of dOCRs (37). The deleted region includes the core sequence (motif 1) present in all α-herpesvirus US1 homologs, which harbors the CDK9-binding site and secondary structure elements important for folding and functioning of this region (38). Thus, we cannot exclude that the deletion alters the conformation, localization, or association of the ICP22 N-terminus with other cellular factors. Mutational work and interaction studies on ICP22 are ongoing to identify the underlying molecular mechanism. Furthermore, as ectopic expression of ICP22 was sufficient to induce strong dOCRs upon co-expression with ICP27, phosphorylation of ICP22 by the viral UL13 or US3 kinases is not required for the induction of dOCRs. Even in absence of UL13 and US3, ectopically expressed ICP22 is still extensively modified by cellular factors (22). Post-translational modifications of ICP22 may thus still be important for dOCR induction. Interestingly, infection with an ICP27-null mutant still induced a weak but nevertheless significant increase in dOCRs. This is consistent with the reduced but nevertheless detectable levels of transcription downstream of genes in infection with an ICP27-null mutant (3). Accordingly, cellular stress responses, and thus DoG transcription (6, 39) were sufficient to induce dOCRs in presence of ICP22. It will be interesting to see whether ICP22 also triggers dOCRs in other models of impaired transcription termination.
ChIPmentation for major histones and common histone marks revealed a selective decrease in histone occupancy within the first ≈25 kb downstream of genes with dOCR induction. This is consistent with the increased chromatin accessibility observed in ATAC-seq experiments and was dependent on the presence of ICP22. The concordant loss of total H1, H3 and H4 as well as of both the activating and inhibitory histone marks H3K36me3 and H3K27me, respectively, indicates that impaired histone repositioning is not restricted to histones with specific histone marks but affects histone repositioning in general. These results support a model in which HSV-1 impairs histone repositioning in the wake of Pol II downstream of genes. It is important to note that the global loss of histone occupancy downstream of genes was only detectable when viral genome replication, and thus viral late gene expression, was inhibited by PAA. Inhibition of viral late gene expression efficiently prevents the dramatic HSV-1-induced loss of Pol II activity on the host genome. Our findings thus stress the importance of studying the function of viral immediate early gene products on the host transcriptional machinery in absence of the HSV-1 induced shut-down of host transcription.
During Pol II transcription, nucleosomes are first destabilized and momentarily removed to be reassembled once again at the same position in the wake of Pol II (reviewed in (11)). This is facilitated by Pol II-associated histone chaperones including SPT6 and FACT. During the course of this study, ICP22 was found to directly interact with FACT and recruit both FACT and SPT6 to the viral replication compartments (16). Interestingly, knock-down of FACT but not SPT6 significantly increased dOCR induction. The requirement for FACT, but not for SPT6, in nucleosome reassembly may be explained by a model in which SPT6-mediated histone chaperoning can be compensated by FACT, which can chaperone all four core histones onto DNA. In contrast, loss of FACT activity cannot be compensated by SPT6, which only chaperones histones H3 and H4 (11). Importantly, FACT knock-down also resulted in a small, nevertheless significant increase in chromatin accessibility within gene bodies of several hundred genes in uninfected and ΔICP22-infected cells. This was substantially increased upon WT but not ΔICP22 infection and often reached similar levels as observed downstream of the respective genes. Of note, a slight increase in chromatin accessibility within gene bodies was also observable for many of these genes in WT HSV-1 infection (without FACT depletion), thereby demonstrating that the increased chromatin accessibility observed in HSV-1 infection is not completely restricted to genomic regions downstream of genes. These findings indicate that increased chromatin accessibility both within and downstream of genes results from a common mechanism, namely functional impairment of FACT by the combined effects of FACT knock-down and the viral ICP22 protein. More direct proof of the involvement of FACT in the increased chromatin accessibility would come from FACT chromatin immunoprecipitation (ChIP-seq or ChIPmentation) experiments. While SPT6 ChIPmentation experiments we performed revealed that SPT6 travels with Pol II far downstream of genes upon DoTT (data not shown), we were unable to obtain data of sufficient quality for FACT despite multiple attempts. We nevertheless would like to propose a model in which the viral ICP22 protein, via its interaction with FACT, interferes with the function of FACT to reposition histones in the wake of Pol II. The FACT complex is thought to associate indirectly with Pol II by either HP1 or the PAF1 complex bridging the interaction (reviewed in (11, 40, 41). Furthermore, chromatin remodelers can interact with histones and FACT and likely promote FACT association with chromatin. It is thus probably not surprising that the resistance of FACT to viral inhibition by ICP22 changes during the course of transcription and drops downstream of genes where allosteric changes in Pol II precede transcription termination. The selective induction of dOCRs in HSV-1 infection thereby also supports the allosteric model of transcription termination, which proposes allosteric changes in Pol II composition at the end of genes (42). In the setting of both efficient FACT knock-down and functional impairment of FACT by the viral ICP22 protein, these differences are alleviated often resulting in a concordant increase in chromatin accessibility both within gene bodies and downstream thereof for genes with strong dOCR induction. Multiple different functions have been attributed to ICP22. One of the most striking is the loss of serine 2 phosphorylation (Ser-2P) of the Pol II CTD, which governs the recruitment of other cellular proteins to Pol II. We thus cannot exclude that viral interference with Ser-2P and resulting changes in Pol II composition, in addition to the direct interaction of ICP22 with FACT, contribute to the observed effects.
At present, we can only speculate about the functional importance of this viral interference with histone repositioning. Interestingly, efficient knock-down of neither SPT6 nor FACT did significantly impair viral protein expression during productive infection arguing against an important role of the two factors in viral transcription. This may reflect the fact that the viral genome remains largely unchromatinized during productive infection (43, 44) thereby alleviating the need for histone removal and repositioning during Pol II transcription of the viral genome. Importantly, both FACT and SPT6 represent important transcription elongation factors. Viral interference with their activities is thus likely to contribute to the selective virus-induced shut-off of host transcription. Due to the multiple functions of ICP22, the relative contribution of FACT manipulation will be difficult to decipher. It is important to note that FACT has also been shown to play an important role during early steps of transcription. In Drosophila, it alleviates transcription inhibition by DSIF (DRB sensitivity-inducing factor) and NELF (negative elongation factor) (45). The ICP22-mediated induction of dOCR may thus only represent a bystander effect of more important viral interference with key mechanism of transcription elongation upstream. However, it may also help increase the pool of free histones (46, 47) to aid the chromatinization of incoming viral genomes at early times of infection when both ICP22 and ICP27 are expressed which may also play a role during the establishment of viral latency. In summary, our findings highlight HSV-1 as an exciting model to study fundamental aspects of the transcriptional machinery in human cells.
Cell culture and infections
Human Fetal Foreskin Fibroblasts (HFFF, purchased from ECACC), Telomerase-Immortalized Human Foreskin Fibroblasts (T-HF) (48), Baby Hamster Kidney cell line (BHK, obtained from ATCC), Human Bone Osteosarcoma Epithelial Cells (U2OS, kindly provided by Stacey Efstathiou), ICP27-complementing Vero 2-2 cell line (kindly provided by Prof. Dr. Beate Sodeik) and Human embryonic kidney 293T cells (HEK-293T, obtained from ATCC) were cultured in Dulbecco’s Modified Eagle Medium (DMEM, ThermoFisher #41966052) supplemented with 10% (v/v) Fetal Bovine Serum (FBS, Biochrom #S0115), 1x MEM Non-Essential Amino Acids (ThermoFisher #11140050) and 1% penicillin/streptomycin. All cells were incubated at 37 °C in a 5% (v/v) CO2-enriched incubator. HFFFs were utilized from passage 11 to 17 for all high-throughput experiments.
This study was performed using wild-type (WT) HSV-1 strain 17, BAC-derived HSV-1 strain 17 (kindly provided by Beate Sodeik) (49), wild-type HSV-1 strain F, and mutant viruses R325 (ΔICP22 C-terminal 220 amino acids, strain F (50)), wild-type KOS1.1 (kindly provided by Steven Rice) (51), vhs-inactivated mutant (Δvhs, strain 17, (52)), ICP27-null mutant (ΔICP27, strain KOS, (53)) and ICP0-null mutant (ΔICP0, strain 17, (54)). Virus stocks were produced in BHK cells as described (3) except for viruses mentioned below. Stocks of the ICP27-null mutant were produced on complementing Vero 2–2 cells (55) and ICP0-null mutant in U2OS cells. All produced viruses were Ficoll-gradient purified.
For all experiments, cells were infected using a multiplicity of infection of 10 (MOI) in fresh media 24 h after the last split. Subsequently, the inoculum was removed, and conditioned media was applied to the cells. The time at which inoculum was replaced with growth media was marked as the 0h time point. To block viral DNA replication, phosphonoacetic acid (PAA, 350 μg/ml) was added in conditioned media to cultured cells after the inoculum was removed. The number of biological replicates that were performed for each experiment is indicated in the main text.
Cell line manipulation and generation
Artificial miRNAs (amiRNAs) against SSRP1 and SPT6 were selected as described (35), cloned into a doxycycline-inducible lentiviral vector (see below) and utilized to generate T-HF cells that enable efficient, dox-inducible knock-down of the respective host proteins. Primer sequences are listed in Supplementary table 1. Transduced T-HF cells were maintained in 5 μg/ml Puromycin. Knock-down was induced by 1 μg/ml doxycycline for 72h with fresh Dox added at 48 h.
HA-ICP22 and HA-ICP22 + V5-ICP27 cells were generated as follows. Lentiviral vectors encoding N-terminal 3xFLAG and V5-tagged (tandem tag) UL54 ORF under control of the doxycycline-inducible pTRE-Tight promoter were produced by cloning the corresponding ORF from the HSV-1 genome (strain 17) via intermediate vectors into pW-TH3. The pW-TH3 vector was derived from pCW57.1 by sequential insertion of a synthetic multi-cloning site (prW64/65) and three stop codons (prW110/111) between the NheI and AgeI restriction sites. pCW57.1 was a gift from David Root (Addgene plasmid #41393; http://n2t.net/addgene:41393). pW-TH7 (3xFlag-V5-NT1) was created by amplifying the N-terminal part of NT1 from the V5-NT1 vector (56) by using primers prW196/197 and inserted back between the BamHI and EcoRI of the same V5-NT1 vector. The 3xFlag-V5-NT1 ORF was excised with EcoRI and XbaI and inserted between the EcoRI and NheI sites of pW-TH3 (now designated pW-TH9). The UL54 ORF was amplified from the HSV-1 genome by PCR using primers prW365/366. The PCR product was digested with BamHI and BglII and inserted into BamHI cut pW-TH9 (now designated pW-TH57). To generate the doxycycline-inducible vector with HA-tagged US1 ORF (designated as LDJ5), the vector YC1 was used as backbone. YC1 was generated to carry blasticidin resistance gene instead of puromycin by restriction digestion of pW-TH3 vector with XbaI and AgeI and insertion of a hPGK.blast construct (purchased from GeneArt) via infusion cloning. The US1 ORF was amplified from the HSV-1 genome by PCR using primers prW1656/1657 and the extracted band was cloned via infusion cloning into YC1, linearized by restriction digest with MluI and NheI. Primer sequences used for generating HA-ICP22 and V5-ICP27 cell lines are listed in Supplementary table 2.
To generate V5-ICP27 and HA-ICP22 doxycycline-inducible cell lines, HEK-293T cells were transfected with pW-TH57 and LDJ5, respectively. Transduced T-HF cells were kept in selection with 5 µg/ml puromycin and 5 µg/ml blasticidin, respectively. To generate V5-ICP27 + HA-ICP22 doxycycline-inducible cell line, V5-ICP27 cells were lentivirally transduced with LDJ5. Transduced T-HF cells were kept in selection with both 5 µg/ml puromycin and 5 µg/ml blasticidin. Expression of proteins was induced by 5 μg/ml doxycycline for 48h.
Western blots
Samples were harvested at the indicated time points by removal of growth media, followed by 1x wash with phosphate buffered saline (PBS, Sigma-Aldrich #D8537) and lysis in 1x Laemmli buffer containing 5% (v/v) β-mercaptoethanol. Samples were sonicated and heated for 5 min at 95 °C before loading onto a Novex WedgeWell 4–20% Tris-Glycine Gel (ThermoFisher #XP04200BOX). Proteins were transferred to 0.2 µm nitrocellulose membranes (Sigma-Aldrich #GE10600001 ), blocked for 1 h at room temperature (RT) in 1x PBS with 0.2% Tween (PBS-T) containing 5% (w/v) milk (Carl Roth #T145.3), and probed using anti-V5 (Cell Signaling #13202), anti-HA clone 11 (Biolegend #16B12), anti-FLAG (Sigma-Aldrich # F3165) anti-α-Tubulin (Cell Signaling #2144), anti-β-Actin clone C-4 (Santa Cruz Biotechnology #sc-47778), anti-GAPDH (Cell Signaling #2118), anti-SPT6 (Novus Biologicals #NB100-2582), anti-SSRP1 (Biolegend #609710), anti-ICP8 clone 11E2 (Santa Cruz Biotechnology #sc-53330), anti-gD clone DL6 (Santa Cruz Biotechnology #sc-21719) overnight at 4 °C at a 1:1,000 diluation unless otherwise specified. Before addition of each antibody, blots were washed with 3x PBS-T. After incubation with either anti-rabbit–horseradish peroxidase (HRP, Sigma-Aldrich #A0545), anti-mouse–HRP (Sigma-Aldrich #A9044) at a 1:10,000 dilution, anti-rat-HRP (Sigma-Aldrich #SAB3700539) at a 1:5,000 diluation, or IRDye 680RD goat-anti-rabbit IgG (Licor #926-68071) and IRDye 800CW donkey anti-mouse IgG secondary antibody (Licor #926-32212) at 1:5,000 dilution, bands were visualized using the LI-COR Odyssey FC Imaging System.
Immunoflourescence analyses
105 HA-ICP22 and HA-ICP22+V5-ICP27 cells were plated in 12 well-dishes with addition of 5 μg/mL of doxycycline (Merck #AMBH2D6FB132). At 48h post-induction, cells were fixed with 4% formaldehyde (PFA) in PBS for 15 min at RT, washed three times in PBS, and either stored at 4 
°C overnight in PBS or processed immediately as follows. Cells were permeabilized in 0.5% Triton X-100 in PBS for 5-10 minutes and blocked in blocking buffer (10% FBS, 0.25M glycine, 1x PBS) for 1h at RT. Anti-HA antibody clone F-7 (Santa Cruz Biotechnology #sc-7392) or anti-V5 antibody (Cell Signaling #13202) were incubated in 10% FBS and 1x PBS for 1h at RT at dilution of 1:500. The secondary anti-mouse IgG, Alexa Fluor 488 (ThermoFisher #A11017) or anti-rabbit IgG, Alexa Flour 568 (abcam #ab175471) were incubated in 10% FBS in 1x PBS for 1h at RT with 0.5μg/mL 4’,6-diamidino-2-phenylindole (DAPI). All steps were followed by three 5 min washes in 1x PBS after which the images were taken on Leica DMi8 fluorescence microscope.
ATAC-seq and Omni-ATAC-seq
ATAC-seq was performed according to the original protocol starting with 1x105 cells per condition (57). An improved ATAC-seq protocol, Omni-ATAC-seq was performed according to the original protocol starting with 1x105 cells per condition (33). For each experiment biological duplicates were carried out.
Sequencing libraries for ATAC samples were prepared as specified using the Nextera DNA Library prep kit (Illumina #15028212) or in combination with primers made by IDT based on Illumina primers with unique dual (UD) index adapters for both i_5 and i_7.
Sequencing libraries for Omni-ATAC samples were prepared as 50 µL reactions containing: 12,5 µL DNA, 6,25 µL i_5_x and i_7_x (10 µM) and 25 µL 2x NEBNext Ultra II Q5 Master Mix (NEB #M0544). The number of cycles necessary for the library amplification was determined from the pre-amplification of transposed fragments.
Both ATAC-seq and Omni-ATAC-seq libraries were quantified by Agilent Bioanalyser and sequenced by NextSeq 500 (Ilumina) at the Core Unit Systemmedizin, Würzburg, Germany (35 bp paired-end reads). All samples were sequenced at equimolar ratios.
RNA-seq controls
To confirm the presence of read-through transcription, on the day of ATAC/Omni-ATAC-seq experiment, total RNA was collected. Biological duplicates were carried out. For total RNA, cells were collected in 500 μl TRI reagent and total RNA was isolated with Quick-RNA Microprep Kit (Zymo Research #R1050) according to manufacturer’s instructions. Following steps were performed by the Core Unit Systemmedizin, Würzburg, Germany. For the total RNA libraries, both cytoplasmic and mitochondrial rRNA species were depleted. No rRNA depletion was performed for 4sU-RNA samples as rRNA only contributes about 40–50% of reads in 4sU-RNA samples. Library preparation for sequencing was performed using the stranded TruSeq RNA-seq protocol (Illumina, San Diego, USA). Libraries were sequenced on NextSeq 500 (Ilumina). 4sU-seq data for Figure 1c-d and Supp. Fig. 1c were taken from previous work (3, 5) and data for Figure 2e-f and Supp. Fig. 2i-l from (4).
ChIPmentation, library preparation and sequencing
Two days prior to infection, two million HFFF cells were seeded in 15 cm dishes. At the day of infection, cells had expanded to ~80% confluency. Cells were infected with the respective viruses as described in results section (n=2 for all conditions except for H3K36me3 in WT strain 17 without PAA (n=3)). PAA (350 μg/mL) was added to the conditioned cell culture media that was supplied to the cells after removal of virus inoculum. At 8 p.i., cells were fixed by adding ChIP Cross-link Gold according to manufacturer’s instructions (Diagenode #C01019027) and subsequently with 1% PBS-buffered formaldehyde. Cells were scraped in 1mL of ice-cold 1x PBS containing protease inhibitor cocktail (1x) (Roche #11836153001) with an additional 1mM phenylmethylsulfonyl fluoride (PMSF). Cells were pelleted at 1500 rpm for 20 min at 4 °C. Supernatant was aspirated and cell pellets were frozen in liquid N2.
Cell pellets were resuspended in 1.5 mL 0.25% [w/v] SDS sonication buffer (10 mM Tris pH=8.0, 0.25% [w/v] SDS, 2 mM EDTA) with 1x protease inhibitors and 1 mM additional PMSF and incubated on ice for 10 minutes. Cells were sonicated in fifteen 1 minute intervals, 25% amplitude, with Branson Ultrasonics SonifierTM S-450 until most fragments were in the range of 200-700 bp as determined by agarose gel electrophoresis. Two million cells used for the preparation of the ChIPmentation libraries were diluted 1:1.5 with equilibration buffer (10 mM Tris, 233 mM NaCl, 1.66% [v/v] Triton X-100, 0.166% [w/v] sodium deoxycholate, 1 mM EDTA, protease inhibitors) and spun at 14,000x g for 10 minutes at 4 °C to pellet insoluble material. Supernatant was transferred to a new 1.5 mL screw-cap tube and topped up with RIPA-LS (10 mM Tris-HCl pH 8.0, 140 mM NaCl, 1 mM EDTA pH 8.0, 0.1% [w/v] SDS, 0.1% [w/v] sodium deoxycholate, 1% [v/v] Triton X-100, protease inhibitors) to 200 μL. Input and gel samples were preserved. Lysates were incubated with 1:100/IP of anti-H1 antibody (Invitrogen #PA5-30055), 1:50/IP of anti-H3 antibody (Invitrogen, PA5-16183), 1:50/IP of anti-H4 antibody (Cell Signaling, #14149S), 1 μg/IP of anti-H3K27me3 (Diagenode, #C15410195) and 1 μg/IP of anti-H3K36me3 (Diagenode, #C15410192) on a rotator overnight at 4 °C. Dependent on the added amount of antibody, the amount of Protein A magnetic beads (ThermoFisher Scientific #10001D) was adjusted (e.g. for 1-2 μg of antibody/IP = 15 μL of beads) and blocked overnight with 0.1% [w/v] bovine serum albumin in RIPA buffer. On the following day, beads were added to the IP samples for 2 h on a rotator at 4 °C to capture the antibody-bound fragments. The immunoprecipitated chromatin was subsequently washed twice with 150 μL each of ice-cold buffers RIPA-LS, RIPA-HS (10 mM Tris-HCl pH 8.0, 50 0mM NaCl, 1 mM EDTA pH 8.0, 0.1% [w/v] SDS, 0.1% [v/v] sodium deoxycholate, 1% [v/v] Triton X-100), RIPA-LiCl (10 mM Tris-HCl pH 8.0, 250 mM LiCl, 1 mM EDTA pH 8.0, 0.5% [w/v] sodium deoxycholate, 0.5% [v/v] Nonidet P-40) and 10 mM Tris pH 8.0 containing protease inhibitors. Beads were washed once more with ice-cold 10 mM Tris pH 8.0 lacking inhibitors and transferred into new tubes.
Beads were resuspended in 25 μL of the tagmentation reaction mix (Nextera DNA Sample Prep Kit, Illumina) containing 5 μL of 5x Tagmentation buffer, 1 μL of Tagment DNA enzyme, topped up with H2O to the final volume and incubated at 37 °C for 10 minutes in a thermocycler. Beads were mixed after 5 minutes by gentle pipetting. To inactivate the Tn5 enzyme, 150 μL of ice cold RIPA-LS was added to the tagmentation reaction. Beads were washed twice with 150 μL of RIPA-LS and 1x Tris-EDTA and subjected to de-crosslinking by adding 100 uL ChIPmentation elution buffer (160 mM NaCl, 40 μg/mL RNase A (Sigma-Aldrich #R4642), 1x Tris-EDTA (Sigma #T9285) and incubating for 1h at 37 °C followed by overnight shaking at 65 °C. The next day, 4 mM EDTA and 200 μg/mL Proteinase K (Roche, #03115828001) were added, and samples incubated for another 2h at 45 °C with 1000 rpm shaking. Supernatant was transferred into a new tube and another 100 μL of ChIPmentation elution buffer was added for another hour at 45 °C with 1000 rpm shaking. DNA was isolated with MinElute PCR Purification Kit (Qiagen #28004) and eluted in 21 μL of H2O.
DNA for the final library was prepared with 25 μL NEBNext Ultra II Q5 Master Mix, 3.75 μL IDT custom primer i5_n_x (10 μM); 3.75 μl IDT custom primer i7_n_x (10 μM) (see Supplementary Table 3); 3.75 μL H20 and 13.75 μL ChIPmentation DNA. The Cq value obtained from the library quantification, rounded up to the nearest integer plus one additional cycle, was used to amplify the rest of the ChIPmentation DNA. Library qualities were verified by High Sensitivity DNA Analysis on the Bioanalyzer 2100 (Agilent) before performing sequencing on NextSeq 500 (paired-end 35bp reads) at the Core Unit Systemmedizin, Würzburg, Germany (samples without PAA) or DNBSEQ-G400 2x100bp in BGI, Hong Kong, China (samples with PAA). All samples were sequenced at equimolar ratios.
Read alignment
Sequencing reads for ATAC-seq, RNA-seq, 4sU-seq and ChIPmentation were mapped against (i) the human genome (GRCh37/hg19), (ii) human rRNA sequences and (iii) the HSV-1 genome (HSV-1 strain 17, GenBank accession code: JN555585, only for HSV-1 infection data) using ContextMap v2.7.9 (58) (using BWA as short read aligner (59) and allowing a maximum indel size of 3 and at most 5 mismatches). For the two repeat regions in the HSV-1 genome, only one copy each was retained, excluding nucleotides 1–9,213 and 145,590–152,222.
Analysis of open chromatin regions
For ATAC-seq data, BAM files with mapped reads were converted to BED format using BEDTools (60) and OCRs were determined from these BED files using F-Seq with default parameters (61). No filtering of OCRs was performed. dOCR length for a gene was calculated as previously described (5). In brief, downstream OCRs were assigned to each gene in the following way. First, all OCRs overlapping with the 10kb downstream of a gene were assigned to this gene. Second, OCRs starting at most 5kb downstream of the so far most downstream OCR of a gene were also assigned to this gene. This was performed iteratively, until no more OCRs could be assigned. Here, individual OCRs could be assigned to multiple genes. dOCR length of a gene was then calculated as the total genomic length downstream of this gene covered by OCRs assigned to the gene. Similarly, OCR length in gene bodies was calculated as the total genomic length of the gene bodies covered by OCRs.
Quantification of downstream transcriptional activity and read-through
Number of read fragments per gene or in downstream regions were determined from the mapped RNA-seq or 4sU-seq reads in a strand-specific manner using featureCounts (62) and gene annotations from Ensembl (version 87 for GRCh37). For genes, all read pairs (= fragments) overlapping exonic regions on the corresponding strand by ≥25bp were counted for the corresponding gene. For downstream regions, all fragments overlapping the 5kb downstream of the gene 3’end were counted. Gene expression and downstream transcriptional activity were quantified in terms of fragments per kilobase of exons per million mapped reads (FPKM) and averaged between replicates. Only reads mapped to the human genome were counted for the total number of mapped reads for FPKM calculation. Percentage of read-through was calculated as previously described (5). In brief: First, the percentage of transcription downstream a gene was calculated separately for each replicate as: Percentage of downstream transcription = 100 x (FPKM in 5kb downstream of gene)/(gene FPKM). Percentage of downstream was averaged between replicates and transcription percentage of read-through was then calculated as percentage of downstream transcription in infected– percentage of downstream transcription in uninfected or untreated cells. Negative values were set to 0.
Metagene analyses
Metagene analyses were performed as previously described (63) using the software developed for this publication. For each gene, the regions −3 kb to +3 kb of the TSS were divided into 250 bp bins, the regions -5 kb to +100 kb of the TTS into 500bp bins and the remainder of the gene body (+3 kb of TSS to −5 kb of TTS) into 100 bins of variable length in order to compare genes with different lengths. For each bin, the average coverage per genome position was calculated and normalized to a total sum of 1. Metagene curves for each replicate were created by averaging results for corresponding bins across all genes considered and metagene plots show the average metagene curves across replicates. To determine statistical significance of differences between average metagene curves for two conditions, paired Wilcoxon signed rank tests were performed for each bin comparing normalized coverage values for each gene for this bin between the two conditions. P-values were adjusted for multiple testing with the Bonferroni method across all bins within each subfigure and are color-coded in the bottom track of subfigures: red = adj. P-value ≤ 10-5; orange = adj. P-value ≤ 10-3; yellow: adj. P-value ≤ 0.05.
Acknowledgments
This work was supported by the European Research Council (ERC-2016-CoG 721016 – HERPES to L.D.), DFG (1275/6-1 to L.D. and FR2938/9-1 to C.C.F.). A.W.W. was the recipient of a generous grant from the Alexander von Humboldt Foundation and the German Federal Foreign Office.
Authors’ contributions
L.Dj., T.H., C.C.F. and L.D. designed the experiments. L.Dj., T.H., A.M. and A.W.W. performed the experiments and L.Dj., K.R., T.H. and C.C.F. analyzed the data. K.R, E.W., M.K., F.E. and C.C.F. performed the computational analyses. L.Dj., T.H., A.W.W., C.C.F. and L.D. drafted the article. All authors contributed to revising the article draft.
Data availability
All sequencing data have been submitted to the Gene Expression Omnibus (GEO) under accession GSE185241 (secure token: cfileamkphijvqj).
- T. Hennig, L. Djakovic, L. Dölken, A. W. Whisnant, A Review of the Multipronged Attack of Herpes Simplex Virus 1 on the Host Transcriptional Machinery. Viruses 2021, Vol. 13, Page 1836. 13, 1836 (2021).
- H. G. Rivas, S. K. Schmaling, M. M. Gaglia, Shutoff of Host Gene Expression in Influenza A Virus and Herpesviruses: Similar Mechanisms and Common Themes. Viruses. 8, 102 (2016).
- A. J. Rutkowski et al., Widespread disruption of host transcription termination in HSV-1 infection. Nat. Commun. 6, 7126 (2015).
- X. Wang et al., Herpes simplex virus blocks host transcription termination via the bimodal activities of ICP27. Nat. Commun. 11 (2020), doi:10.1038/s41467-019-14109-x.
- T. Hennig et al., HSV-1-induced disruption of transcription termination resembles a cellular stress response but selectively increases chromatin accessibility downstream of genes. PLOS Pathog. 14, e1006954 (2018).
- A. Vilborg, M. C. Passarelli, T. A. Yario, K. T. Tycowski, J. A. Steitz, Widespread Inducible Transcription Downstream of Human Genes. Mol Cell. 59, 449–461 (2015).
- A. Vilborg, J. A. Steitz, Readthrough transcription: How are DoGs made and what do they do? RNA Biol. 14, 632–636 (2017).
- M. E. Nemeroff, S. M. Barabino, Y. Li, W. Keller, R. M. Krug, Influenza virus NS1 protein interacts with the cellular 30 kDa subunit of CPSF and inhibits 3’end formation of cellular pre-mRNAs. Mol Cell. 1, 991–1000 (1998).
- D. L. V. Bauer et al., Influenza Virus Mounts a Two-Pronged Attack on Host RNA Polymerase II Transcription. Cell Rep. 23, 2119-2129.e3 (2018).
- A. R. Grosso et al., Pervasive transcription read-through promotes aberrant expression of oncogenes and RNA chimeras in renal carcinoma. Elife. 4 (2015), doi:10.7554/eLife.09214.
- A. A. Duina, Histone Chaperones Spt6 and FACT: Similarities and Differences in Modes of Action at Transcribed Genes. Genet Res Int. 2011, 625210 (2011).
- R. Belotserkovskaya et al., FACT facilitates transcription-dependent nucleosome alteration. Science (80-.). 301, 1090–1093 (2003).
- A. Bortvin, F. Winston, Evidence that Spt6p controls chromatin structure by a direct interaction with histones. Science (80-.). 272, 1473–1476 (1996).
- C. D. Kaplan, L. Laprade, F. Winston, Transcription elongation factors repress transcription initiation from cryptic sites. Science (80-.). 301, 1096–1099 (2003).
- A. Narain et al., Targeted protein degradation reveals a direct role of SPT6 in RNAPII elongation and termination. Mol. Cell. 81, 3110-3127.e14 (2021).
- H. L. Fox, J. A. Dembowski, N. A. DeLuca, A Herpesviral Immediate Early Protein Promotes Transcription Elongation of Viral Transcripts. MBio. 8 (2017), doi:10.1128/mBio.00745-17.
- A. E. Sears, I. W. Halliburton, B. Meignier, S. Silver, B. Roizman, Herpes simplex virus 1 mutant deleted in the alpha 22 gene: growth and gene expression in permissive and restrictive cells and establishment of latency in mice. J. Virol. 55, 338–346 (1985).
- A. Kato et al., Identification of Proteins Phosphorylated Directly by the Us3 Protein Kinase Encoded by Herpes Simplex Virus 1. J. Virol. 79, 9325–9331 (2005).
- F. C. Purves, B. Roizman, The UL13 gene of herpes simplex virus 1 encodes the functions for posttranslational processing associated with phosphorylation of the regulatory protein α22. Proc. Natl. Acad. Sci. U. S. A. 89, 7310–7314 (1992).
- J. A. Blaho, C. S. Zong, K. A. Mortimer, Tyrosine phosphorylation of the herpes simplex virus type 1 regulatory protein ICP22 and a cellular protein which shares antigenic determinants with ICP22. J. Virol. 71, 9828–9832 (1997).
- C. Mitchell, J. A. Blaho, A. L. McCormick, B. Roizman, The nucleotidylylation of herpes simplex virus 1 regulatory protein α22 by human casein kinase II. J. Biol. Chem. 272, 25394–25400 (1997).
- J. Zaborowska et al., Herpes simplex virus 1 (HSV-1) ICP22 protein directly interacts with cyclin-Dependent kinase (CDK)9 to Inhibit RNA polymerase II transcription elongation. PLoS One. 9 (2014), doi:10.1371/journal.pone.0107654.
- M. Ou, R. M. Sandri-Goldin, Inhibition of cdk9 during herpes simplex virus 1 infection impedes viral transcription. PLoS One. 8, e79007 (2013).
- L. O. Durand, S. J. Advani, A. P. W. Poon, B. Roizman, The Carboxyl-Terminal Domain of RNA Polymerase II Is Phosphorylated by a Complex Containing cdk9 and Infected-Cell Protein 22 of Herpes Simplex Virus 1. J. Virol. 79, 6757–6762 (2005).
- B. A. Smith-Donald, B. Roizman, The Interaction of Herpes Simplex Virus 1 Regulatory Protein ICP22 with the cdc25C Phosphatase Is Enabled In Vitro by Viral Protein Kinases US3 and UL13. J. Virol. 82, 4533–4543 (2008).
- C. Su Hui Teo, R. A. Serwa, P. O’Hare, Spatial and Temporal Resolution of Global Protein Synthesis during HSV Infection Using Bioorthogonal Precursors and Click Chemistry. PLOS Pathog. 12, e1005927 (2016).
- M. Adlakha, C. M. Livingston, I. Bezsonova, S. K. Weller, The HSV-1 immediate early protein ICP22 is a functional mimic of a cellular J-protein. J. Virol. (2019), doi:10.1128/JVI.01564-19.
- J. S. Orlando, T. L. Astor, S. A. Rundle, P. A. Schaffer, The Products of the Herpes Simplex Virus Type 1 Immediate-Early US1/US1.5 Genes Downregulate Levels of S-Phase-Specific Cyclins and Facilitate Virus Replication in S-Phase Vero Cells. J. Virol. 80, 4005–4016 (2006).
- Y. Maruzuru et al., Role of Herpes Simplex Virus 1 Immediate Early Protein ICP22 in Viral Nuclear Egress. J. Virol. 88, 7445–7454 (2014).
- D. T. McSwiggen et al., Evidence for DNA-mediated nuclear compartmentalization distinct from phase separation. Elife. 8 (2019), doi:10.7554/eLife.47098.
- K. A. Fraser, S. A. Rice, Herpes simplex virus immediate-early protein ICP22 triggers loss of serine 2-phosphorylated RNA polymerase II. J Virol. 81, 5091–5101 (2007).
- S. A. Rice, M. C. Long, V. Lam, P. A. Schaffer, C. A. Spencer, Herpes simplex virus immediate-early protein ICP22 is required for viral modification of host RNA polymerase II and establishment of the normal viral transcription program. J. Virol. 69, 5550–5559 (1995).
- M. R. Corces et al., An improved ATAC-seq protocol reduces background and enables interrogation of frozen tissues. Nat. Methods. 14, 959–962 (2017).
- M. R. Corces et al., Omni-ATAC-seq: Improved ATAC-seq protocol. Protoc. Exch. (2017), doi:10.1038/protex.2017.096.
- C. Fellmann et al., An optimized microRNA backbone for effective single-copy RNAi. Cell Rep. 5, 1704–1713 (2013).
- A. Safina et al., Complex mutual regulation of facilitates chromatin transcription (FACT) subunits on both mRNA and protein levels in human cells. Cell Cycle. 12, 2423–2434 (2013).
- T. W. Bastian, C. M. Livingston, S. K. Weller, S. A. Rice, Herpes Simplex Virus Type 1 Immediate-Early Protein ICP22 Is Required for VICE Domain Formation during Productive Viral Infection. J. Virol. 84, 2384–2394 (2010).
- A. W. Kolb, T. R. Schmidt, D. W. Dyer, C. R. Brandt, Sequence variation in the herpes simplex virus U S1 ocular virulence determinant. Investig. Ophthalmol. Vis. Sci. 52, 4630–4638 (2011).
- A. Vilborg et al., Comparative analysis reveals genomic features of stress-induced transcriptional readthrough. Proc Natl Acad Sci U S A. 114, E8362–E8371 (2017).
- K. Gurova, H. W. Chang, M. E. Valieva, P. Sandlesh, V. M. Studitsky, Structure and function of the histone chaperone FACT – Resolving FACTual issues. Biochim. Biophys. Acta - Gene Regul. Mech. 1861 (2018), pp. 892–904.
- T. Formosa, F. Winston, The role of FACT in managing chromatin: Disruption, assembly, or repair? Nucleic Acids Res. 48 (2021), pp. 11929–11941.
- N. J. Proudfoot, Ending the message: poly(A) signals then and now. Genes Dev. 25, 1770–1782 (2011).
- S. S. Leinbach, W. C. Summers, The Structure of Herpes Simplex Virus Type 1 DNA as Probed by Micrococcal Nuclease Digestion. J. Gen. Virol. 51, 45–59 (1980).
- J. A. Dembowski, N. A. DeLuca, Selective Recruitment of Nuclear Factors to Productively Replicating Herpes Simplex Virus Genomes. PLOS Pathog. 11, e1004939 (2015).
- T. T. Tettey et al., A Role for FACT in RNA Polymerase II Promoter-Proximal Pausing. Cell Rep. 27, 3770-3779.e7 (2019).
- K. L. Conn, M. J. Hendzel, L. M. Schang, Linker Histones Are Mobilized during Infection with Herpes Simplex Virus Type 1. J. Virol. 82, 8629–8646 (2008).
- L. M. Schang, M. Y. Hu, E. F. Cortes, K. Sun, Chromatin-mediated epigenetic regulation of HSV-1 transcription as a potential target in antiviral therapy. Antiviral Res. 192, 105103 (2021).
- Y. Xu, S. A. Cei, A. Rodriguez Huete, K. S. Colletti, G. S. Pari, Human Cytomegalovirus DNA Replication Requires Transcriptional Activation via an IE2- and UL84-Responsive Bidirectional Promoter Element within ori Lyt. J. Virol. 78, 11664–11677 (2004).
- M. Sandbaumhüter et al., Cytosolic herpes simplex virus capsids not only require binding inner tegument protein pUL36 but also pUL37 for active transport prior to secondary envelopment. Cell. Microbiol. 15, 248–269 (2013).
- L. E. Post, B. Roizman, A generalized technique for deletion of specific genes in large genomes: a gene 22 of herpes simplex virus 1 is not essential for growth. Cell. 25, 227–232 (1981).
- M. C. Long, V. Leong, P. A. Schaffer, C. A. Spencer, S. A. Rice, ICP22 and the UL13 Protein Kinase Are both Required for Herpes Simplex Virus-Induced Modification of the Large Subunit of RNA Polymerase II. J. Virol. 73, 5593–5604 (1999).
- M. L. Fenwick, R. D. Everett, Inactivation of the shutoff gene (UL41) of herpes simplex virus types 1 and 2. J Gen Virol. 71 (Pt 12, 2961–2967 (1990).
- I. L. Smith, M. A. Hardwicke, R. M. Sandri-Goldin, Evidence that the herpes simplex virus immediate early protein ICP27 acts post-transcriptionally during infection to regulate gene expression. Virology. 186, 74–86 (1992).
- N. D. Stow, E. C. Stow, Isolation and characterization of a herpes simplex virus type 1 mutant containing a deletion within the gene encoding the immediate early polypeptide Vmw110. J. Gen. Virol. 67, 2571–2585 (1986).
- R. E. Sekulovich, K. Leary, R. M. Sandri-Goldin, The herpes simplex virus type 1 alpha protein ICP27 can act as a trans-repressor or a trans-activator in combination with ICP4 and ICP0. J Virol. 62, 4510–4522 (1988).
- M. Bolstad, F. Abaitua, C. M. Crump, P. O’Hare, Autocatalytic activity of the ubiquitin-specific protease domain of herpes simplex virus 1 VP1-2. J Virol. 85, 8738–8751 (2011).
- J. D. Buenrostro, P. G. Giresi, L. C. Zaba, H. Y. Chang, W. J. Greenleaf, Transposition of native chromatin for fast and sensitive epigenomic profiling of open chromatin, DNA-binding proteins and nucleosome position. Nat Methods. 10, 1213–1218 (2013).
- T. Bonfert, E. Kirner, G. Csaba, R. Zimmer, C. C. Friedel, ContextMap 2: fast and accurate context-based RNA-seq mapping. BMC Bioinformatics. 16, 122 (2015).
- H. Li, R. Durbin, Fast and accurate short read alignment with Burrows-Wheeler transform. Bioinformatics. 25, 1754–1760 (2009).
- A. R. Quinlan, I. M. Hall, BEDTools: a flexible suite of utilities for comparing genomic features. Bioinformatics. 26, 841–842 (2010).
- A. P. Boyle, J. Guinney, G. E. Crawford, T. S. Furey, F-Seq: a feature density estimator for high-throughput sequence tags. Bioinformatics. 24, 2537–2538 (2008).
- Y. Liao, G. K. Smyth, W. Shi, featureCounts: an efficient general purpose program for assigning sequence reads to genomic features. Bioinformatics. 30, 923–930 (2014).
- A. P. Chirackal Manavalan et al., CDK12 controls G1/S progression by regulating RNAPII processivity at core DNA replication genes. EMBO Rep. 20 (2019), doi:10.15252/embr.201847592.
There is NO Competing Interest.
- S1Filefinal.pdf
Supplementary Figures