RNA decay during gammaherpesvirus infection reduces RNA polymerase II occupancy of host promoters but spares viral promoters

In mammalian cells, widespread acceleration of cytoplasmic mRNA degradation is linked to impaired RNA polymerase II (Pol II) transcription. This mRNA decay-induced transcriptional repression occurs during infection with gammaherpesviruses including Kaposi’s sarcoma-associated herpesvirus (KSHV) and murine gammaherpesvirus 68 (MHV68), which encode an mRNA endonuclease that initiates widespread RNA decay. Here, we show that MHV68-induced mRNA decay leads to a genome-wide reduction of Pol II occupancy at mammalian promoters. This reduced Pol II occupancy is accompanied by down-regulation of multiple Pol II subunits and TFIIB in the nucleus of infected cells, as revealed by mass spectrometry-based global measurements of protein abundance. Viral genes, despite the fact that they require Pol II for transcription, escape transcriptional repression. Protection is not governed by viral promoter sequences; instead, location on the viral genome is both necessary and sufficient to escape the transcriptional repression effects of mRNA decay. We propose a model in which the ability to escape from transcriptional repression is linked to the localization of viral DNA within replication compartments, providing a means for these viruses to counteract decay-induced transcript loss.


Introduction
Regulating messenger RNA (mRNA) abundance is of central importance during both cellular homeostasis and disease. While it is intuitive that transcriptional changes in the nucleus impact RNA levels in the cytoplasm, evidence in both yeast and mammalian cells now indicates that the converse is also true: altered cytoplasmic mRNA decay rates can broadly impact transcription by RNA polymerase II (Pol II) [1]. In yeast, stabilizing the cytoplasmic mRNA pool by removing mRNA decay factors such as Xrn1 causes a compensatory decrease in Pol II transcription [2][3][4][5][6]. Furthermore, reducing or slowing Pol II transcription leads to an increase in overall mRNA stability [2], supporting a "buffering" model in which yeast can compensate for widespread perturbations to mRNA abundance by alternatively changing rates of decay or transcription.
Yeast and mammals share key proteins involved in cytoplasmic mRNA decay and transcription. However, studies with mammalian RNA-decay inducing viruses and viral nucleases have revealed that accelerated mRNA decay is not broadly counteracted by increased transcription-based repopulation, but instead leads to a more extensive shutdown of cellular gene expression. Infection with the gammaherpesviruses KSHV or MHV68 causes reduced Pol II occupancy at individually tested host promoters in a manner dependent on mRNA cleavage by the viral endonucleases SOX or muSOX [22,23]. In uninfected cells expressing muSOX, this Pol II repression phenotype is linked to differential trafficking of RNA binding proteins (RBP) from the cytoplasm to the nucleus [23]. RBP trafficking is thought to occur upon their release from mRNA during the act of transcript degradation, which is initiated by muSOX and completed by cellular exonucleases including Xrn1. It is possible that this altered RBP distribution in cells is interpreted as a stress signal related to pathogenesis, leading to transcriptional dampening rather than the compensatory increase that occurs in other gene regulatory contexts [1].
MHV68 mRNAs are also susceptible to cleavage by muSOX during infection [24]. However, measurements of nascent transcript production from a subset of viral promoters suggested that viral genes are robustly transcribed during the stage of infection when host transcription is reduced [25]. An outstanding question is how these viral promoters escape mRNA decay-induced transcriptional repression despite being transcribed by host Pol II. Here, we measured Pol II occupancy across the mouse and MHV68 genomes in infected cells to more comprehensively define how accelerated mRNA decay impacts polymerase occupancy. Our data demonstrate a genome-wide reduction in Pol II occupancy at host promoters under conditions of increased mRNA decay. However, this phenotype was not reversed upon depletion of cellular exonucleases, suggesting that additional mechanisms contribute to transcriptional repression during infection. Indeed, measurements of global proteome changes by mass spectrometry revealed an mRNA decay-linked decrease in nuclear levels of several transcription factors including key Pol II subunits, which likely contributes to the impact of MHV68 infection on host transcription. Notably, Pol II occupancy of the viral genome appears broadly resistant to the effects of mRNA degradation, despite the change in nuclear levels of Pol II. This protection is not conferred by viral promoter sequences; instead, location on the replicating viral genome is both necessary and sufficient to escape the transcriptional effects of mRNA decay. We propose a model in which DNA amplified in viral replication compartments, unlike the cellular chromatin, is not subject to transcriptional repression during accelerated cytoplasmic mRNA decay. Thus, while both viral and cellular mRNA pools are susceptible to cytoplasmic degradation, transcriptional repopulation selectively counteracts this loss for viral genes.

Accelerated RNA decay broadly reduces Pol II occupancy
We previously reported that Pol II occupancy at several individual mammalian promoters was significantly reduced during accelerated cytoplasmic mRNA decay [23,25]. To more comprehensively assess the extent of mRNA decay-induced transcriptional repression, we evaluated global Pol II occupancy by chromatin immunoprecipitation and deep sequencing (ChIP-seq) in mock versus MHV68 infected MC57G mouse fibroblasts at 24 hours post infection (hpi), which is near the end of the infectious cycle [26]. As a control, we also infected cells with a version of MHV68 containing the point mutation R443I in the viral muSOX endonuclease gene, which reduces its mRNA cleavage activity. MHV68 R443I replicates similarly to WT MHV68 in cultured mouse fibroblasts and in the lungs of infected mice, although it displays in vivo defects in viral trafficking to distal sites and latency establishment [7]. MHV68-infected cells showed promoter-proximal loss of Pol II at 86% of loci compared to mock-infected cells, averaged across two biological replicates and subtracting input signal from Pol II occupancy counts from -50 to +200 around the transcription start site (TSS) (Fig 1A and S1A Fig). Loss of Pol II occupancy was primarily due to muSOX-induced mRNA decay, as this signal loss either did not occur or was reduced at 67% of these loci during infection with R443I MHV68 (Fig 1B and S1B Fig). Sixty-three genes, or 1.3% of loci, showed higher Pol II occupancy in WT infection as compared to mock infection, although these did not share any common gene ontology (GO) terms. A representative genome browser view shows reduced Pol II signal throughout Srsf2 in the MHV68 condition of two replicate ChIP experiments (Fig 1C). The variability between replicates in R443I recovery (Fig 1C) is likely due to differences in the extent of host shutoff reduction between replicates, as R443I reduces muSOX activity but it is not a catalytic mutant. We validated the ChIP-seq results by ChIP-qPCR at the Rplp0 and Fus promoters, both of which showed a significant reduction of Pol II occupancy in cells infected with WT but not R443I MHV68 (Fig 1D). We used a second antibody that recognizes the N-terminus of the major subunit of Pol II, Rpb1, to control for potential bias in the phosphorylation state of the CTD recognized by the ChIP-seq antibody, 8WG16 [27]. We saw a similar decrease in Pol II levels at the promoter with both antibodies as well as a decrease in the Rplp0 gene body Pol II occupancy with the antibody recognizing the Rpb1 N-terminus upon MHV68 infection (Fig 1D, S1C Fig).
To assess the stage of transcription impacted by RNA decay, we compared the amount of Pol II in promoter regions (-50 to +200) and the amount of Pol II in gene bodies (+200 to +5000) between mock and infected samples. Consistent with the above analyses, MHV68 infected cells contained less promoter proximal Pol II than uninfected cells, and this signal was partially recovered in the R443I infection ( Fig 1E). Genes with reduced Pol II promoter binding during MHV68 infection also had less Pol II within the gene body compared to mock infected cells, which again was not observed during R443I infection ( Fig 1F). Notably, the difference in regression line slopes between MHV68 and mock infected cells was more marked when comparing Pol II promoter occupancy, suggesting that the primary transcriptional defect is at the stages of recruitment and initiation. This is consistent with data from previous studies that found no defect in serine 2 phosphorylation of elongating Pol II and no consistent change to Pol II elongation during MHV68 infection [23,25].
Finally, to assess if RNA decay affected transcription termination, we plotted sequencing reads between 1kb upstream and 1kb downstream of the transcription termination site (TTS) as a histogram (Fig 1G and S1D Fig). TTS-proximal Pol II shows no defect in termination in either infection condition, suggesting Pol II removal from host transcripts is not impacted by MHV68 infection or by RNA decay.
Depletion of host RNA decay factors is not sufficient to abrogate feedback between mRNA decay and Pol II occupancy muSOX expression alone is sufficient to reduce cytoplasmic populations of mRNA by 60-80% [28][29][30], and in this non-infectious setting the loss of promoter-proximal Pol II is dependent on the presence of the host RNA decay enzymes Xrn1 and Dis3L2 [25]. To determine if this mechanism is also sufficient to explain the loss of Pol II occupancy during MHV68 infection, we depleted Xrn1 and Dis3L2 individually or in tandem using siRNAs (Fig 2A). Unexpectedly, depletion of these cellular decay factors did not reproducibly rescue Pol II occupancy at the Fus and Rplp0 promoters during MHV68 infection compared to mock infected cells ( Fig 2B). These data suggest that additional factors contribute to the mRNA decay-dependent decrease in Pol II from cellular promoters during MHV68 infection.

General transcription factors and Pol II subunits are depleted in the nucleus in an RNA decay dependent manner
To identify other proteins that might contribute to the reduced Pol II occupancy at host promoters during infection, we applied an unbiased mass spectrometry-based approach to measure protein abundance and localization differences in 3T3 mouse fibroblast cells that were Promoter-proximal Pol II recruitment to mammalian genes is RNA-decay dependent. (A) Pol II ChIP-seq signal profiles of host genes in mock-infected, MHV68 WT-infected and R443I-infected MC57G cells. Each row of the heat map displays Pol II occupancy of one gene from -1000 to +1000 in 25 bp bins. Genes are ranked by the Pol II-transcription start site (TSS) proximal signal in mock infected cells. (B) Sequence tags were plotted as a histogram with 25 bp bins for -2000 to +4000 around the TSS. Mock (red), MHV68 (blue) and MHV68 R443I (green) traces are shown. (C) Pol II ChIP-seq coverage across the Srsf2 gene for ChIP-seq replicate 1 and 2 with mock, WT MHV68 and R443I shown. Alignment files were converted to the tiled data file (tdf) format and visualized in the Integrative Genome Viewer. (D) ChIP-qPCR validation of Pol II occupancy at the Rplp0 and Fus promoters plotted with standard deviation. Pol II ChIP was performed on mock, MHV68 WT or MHV68 R443I infected MC57G cells and Pol II levels were assayed near the TSS of two repressed host genes during MHV68 infection from the ChIP-seq data. IgG is from the MHV68 infection condition. ( � p < 0.05, students paired ttest on raw % input values) (E) Scatterplot of Pol II occupancy of promoters, averaging the sum of ChIP-seq tags from -50 to +200 across two replicate experiments with inputs subtracted out. Mock is compared to WT MHV68 infection on the left and to MHV68 R443I infection on the right. Linear regression lines are plotted in blue and equations are provided. A y = x line is plotted for reference in red. (F) Scatterplot of Pol II occupancy of gene bodies, averaging the sum of ChIP-seq tags from -50 to +200 across two replicate experiments with inputs subtracted out. Data as described for (E). (G) Pol II transcription termination is not dependent on RNA decay. Sequence tags were plotted as a histogram in 25 bp bins for transcription termination sequence (TTS) proximal Pol II for -1000 to +1000 around the TTS with the same color scheme as (B).
https://doi.org/10.1371/journal.ppat.1008269.g001 mock infected or infected with WT or R443I MHV68 at 24 hpi. We previously showed that accelerated mRNA decay by the coordinated activity of muSOX and Xrn1 causes RNA binding protein relocalization from the cytoplasm to the nucleus and that this relocalization plays a role in transcriptional repression in 293T cells [23]. We therefore performed a similar experiment in infected cells, in which each sample was separated into nuclear and cytoplasmic fractions and proteins from each fraction were labeled with isobaric tandem mass tags (TMT). We then multiplexed these TMT labeled fractions and subjected them to quantitative liquid chromatography/tandem mass spectrometry (LC/MS-MS) (Figs 3A and S2A, S1 Table). In agreement with our previous findings [23], we observed the expected increase in nuclear levels of the RNA binding proteins PABPC1 and LARP4 in cells infected with WT MHV68 compared to mock or R443I infected cells (Fig 3B).
We then searched for additional proteomic differences specific to WT MHV68 infected cells that might contribute to decreased Pol II occupancy. We examined classes of proteins that increased or decreased in a host shutoff dependent manner in the nucleus and the cytoplasm (S2 B-D). Notably, RNA Polymerase II activity-associated proteins were substantially decreased in WT MHV68 infection compared to R443I or mock cells (Fig 3C). The nucleusspecific changes included decreased levels of 7 subunits of Pol II, the general transcription factor (GTF) TFIIB, and the large subunit of the GTF TFIIA ( Fig 3D). The lack of a corresponding increase in the cytoplasmic levels of Pol II subunits suggests that muSOX activity decreased their overall abundance rather than changing their localization (Figs 3E and S2D).  Table). R443I virus infection showed relatively constant levels of Rpb1, Rpb2, TFIIB and TBP across the time course ( Fig 3F). Collectively, these findings demonstrate that muSOX-induced mRNA degradation during MHV68 infection causes both RNA binding protein relocalization and depletion of multiple components of the cellular transcriptional machinery. It is likely that this combination of phenotypes contributes to the broad reduction in Pol II occupancy at host promoters.

Pol II occupancy at viral genes is not affected by RNA decay
Herpesviral genes are transcribed in the nucleus by mammalian Pol II, and thus could be similarly subjected to the mRNA decay-induced transcriptional repression observed across the host genome. We therefore analyzed Pol II occupancy of the viral genome in WT MHV68 and R443I infected cells to determine how viral genes respond to RNA decay in light of the decrease in Pol II subunits. The average Pol II signature across 82 viral ORFs between -500 and +500 around the transcription start site was comparable in cells infected with WT or R443I MHV68, with a slightly higher signal in the R443I infection ( Fig 4A). Indeed, as shown in the genome browser of a representative genomic region, there was a robust and widespread Pol II signal in both infection conditions ( Fig 4B). To determine if transcription from the viral genome was affected by host shutoff, we pulse labeled WT MHV68 and R443I infected cells with 4-thiouridine (4SU) for 10 min to measure nascent transcript levels. 4SU gets incorporated into actively transcribing mRNAs and can be subsequently coupled to MTSEA -biotin and purified over streptavidin beads [59]. For the 7 transcripts we measured, we saw no statistically significant difference in nascent transcription comparing WT to R443I (S3 Fig), consistent with our Pol II ChIP-seq results. Thus, herpesviral genes appear to escape mRNA decayinduced transcriptional repression.
In contrast to host genes, we did not observe peaks of promoter-proximal Pol II on the viral genome in our ChIP-seq data (Fig 4A and 4B). Pol II ChIP qPCR across various regions of ORF54 and ORF37 also showed a similar Pol II signal throughout each gene body and at their promoter. We did see a decrease in signal near the TTS of ORF54, however this was true for both WT and R443I infections ( Fig 4C). We hypothesize that pervasive transcription of the viral genome at 24 hpi, compounded by the presence of TSSs on both Watson and Crick strands, likely contributes to this broad Pol II signal.

PABPC is not excluded from viral replication compartments
We next considered how the viral genome maintains high levels of Pol II occupancy during accelerated mRNA decay. During replication, viral DNA is transcribed within viral replication compartments (RCs) in the nucleus, which exclude host DNA and thus stain poorly with DAPI [31][32][33]. Although they are not membrane bound, these compartments selectively enrich for factors required for viral replication (such as the viral processivity factor ORF59) and gene expression (such as Pol II) [34][35][36]. Virus-induced mRNA decay causes increased trafficking of cytoplasmic poly(A) binding protein (PABPC) from the cytoplasm to the nucleus (see Fig 3B), which has been linked to transcriptional repression [23,37]. To determine whether replication compartments selectively exclude nuclear PABPC as a means of avoiding transcriptional repression, we monitored its localization in cells lytically infected with MHV68 or KSHV. As expected, the majority of infected cells displayed increased nuclear PABPC compared to unreactivated or uninfected cells as measured by immunofluorescence (IF). However, PABPC did not appear to be excluded from replication compartments (Fig 5A and 5B and S4  Fig). To confirm PABPC localization relative to RCs, we first identified cells with RCs by looking for nuclear regions containing Pol II or ORF59 that also showed reduced DAPI staining. We then quantified nuclear PABPC pixel intensity, and counted cells as having an overlapping RC/PABPC signal when the PABPC pixel intensity was at least 2x that of cells without nuclear PABPC from the same image. Indeed, 85 percent of MHV68 infected cells showed overlapping PABPC/Pol II signal and 90 percent of reactivated iSLKs showed an overlapping PABPC/ ORF59 signal (Fig 5C). Thus, PABPC exclusion from RCs is unlikely to underlie viral promoter escape.

Viral promoters fail to escape transcriptional repression outside of the viral genome
We next tested whether viral promoter sequence elements were sufficient to enable escape by querying whether they conferred resistance to transcriptional repression upon relocation from the viral genome into the host DNA. We inserted the MHV68 late promoter M7 or the KSHV LANA promoter upstream of a puromycin resistance cassette into 293T cell chromatin using either lentiviral transduction or random plasmid DNA integration, respectively. We induced  mRNA decay and host transcriptional repression in these cells by expression of MHV68 muSOX or the herpes simplex virus type 1 (HSV-1) endonuclease vhs. Similar to the host promoters Gapdh and Rplp0, Pol II occupancy of both integrated viral promoters was repressed upon transfection with muSOX and vhs (Fig 6A and 6B).
To confirm that viral promoters integrated into the host genome were also repressed in the context of infection, we infected MC57G cells containing an integrated MHV68 M4 promoter-driven puromycin cassette with MHV68. Indeed, the M4 promoter was repressed to a similar degree as host Fus and Rplp0 promoters during infection with WT MHV68 but not the mRNA decay-deficient MHV68 R443I at 24hpi (Fig 6C). In summary, these data argue against the presence of protective elements within viral promoters, and instead suggest that the chromatin context or location on the viral genome underlies escape from mRNA decay-induced transcriptional repression.

A non-MHV68 promoter escapes repression on the viral genome
A prediction from the above results is that a non-MHV68 promoter would gain protection from transcriptional repression when present on the replicating viral genome. To test this, we measured the Pol II occupancy and transcriptional potential of the cytomegalovirus (CMV) promoter, which drives green fluorescent protein (GFP) expression on the MHV68 genome. Unlike alpha-and gammaherpesviruses, the betaherpesvirus CMV does not accelerate mRNA decay. Furthermore, in either a plasmid context or when integrated into 293T cells, the CMV promoter is strongly repressed during muSOX-induced mRNA decay, similar to host genes. This was seen upon measurement of nascent RNA production by 4SU pulse labeling in muSOX-expressing cells (Fig 7A, ActB shown for reference) [23]. We also observed repression of the CMV promoter driving puromycin resistance integrated into the host chromatin during infection by measuring Pol II occupancy using ChIP (Fig 7B). However, in the context of the replicating MHV68 genome, the CMV promoter driving GFP had a Pol II ChIP-seq signal during both MHV68 WT and R443I infection that was not markedly different from several other viral genes (ORFs 4, 25, 48, 52, 72, M11) ( Fig 7C).
Finally, we evaluated whether the lytic stage of infection was necessary to confer escape from transcriptional repression, or whether some other feature of the non-replicating viral genome was important. To this end, we measured Pol II occupancy at promoters of several genes expressed from the KSHV genome during latent infection in 293T cells, when the viral DNA is maintained as an episome but not amplified in replication compartments. Latent infection does not promote mRNA turnover, and thus we transfected muSOX into these cells to stimulate mRNA turnover in the absence of lytic replication. Similar to the host genes Gapdh and Rplp0, we observed an RNA decay-induced repression of Pol II occupancy at the latent viral promoter LANA while the viral K1 promoter was repressed in 4 out of 5 experiments (Fig 7D). We also observed repression of the human Ef1α promoter, which is present on the KSHV viral episome as a driver of the GFP reporter gene (Fig 7D). In sum, these observations indicate that the viral genome per se does not confer protection from transcriptional repression; instead, features linked to lytic genome amplification in replication compartments facilitate robust Pol II recruitment under conditions of accelerated mRNA turnover. outlined in white and were identified as nuclei with ORF59 staining that overlaps with regions that stain poorly with DAPI. The inset shows a merge of ORF59 and PABPC for several cells that co-stain for both proteins in RCs. (

Discussion
Here, we demonstrate that virus-induced mRNA decay broadly decreases Pol II occupancy of mammalian promoters, extending prior observations made with individual cellular genes to the genomic scale. Using an unbiased TMT-based proteomics approach, we reveal that infection with MHV68 causes RNA decay-dependent changes in the abundance of numerous cellular proteins with roles in Pol II transcription, including depletion of 7 of the 12 Pol II subunits, all of which are specific to Pol II and not shared by Pol I or III. The TMT findings underscore Pol II-ChIP qPCR was used to measure Pol II levels 24 h post SOX or vhs transfection as described in (A) and significance as assigned in (A). (C) The MHV68 M4 viral promoter driving puromycin resistance was lentivirally integrated into MC57G cells. The cells were then infected with WT MHV68 or R443I at a MOI of 5 and Pol II levels at the indicated promoters were assayed by Pol II ChIP-qPCR. Significance as assigned in (A).
https://doi.org/10.1371/journal.ppat.1008269.g006  [23], whereupon 500 μM of 4SU was added for 10 min and labeled RNA was isolated by biotin-streptavidin pull down. Levels of newly transcribed RNA from the indicated genes were measured by RT-qPCR. All samples were normalized to 18S and levels of RNA from control GFP expressing cells were set to 1. (B) The CMV promoter driving puromycin resistance was lentivirally integrated into MC57G cells. The cells were then infected with WT MHV68 or R443I at an MOI of 5 and Pol II levels at the indicated promoters were assayed by Pol II ChIP-qPCR. (C) ChIP-seq traces comparing WT MHV68 to R443I coverage from -300 to +200 around the TSS in 25 bp bins, averaged from two biological replicates of infected MC57G cells. Six representative viral genes and CMV-GFP are shown. (D) 293T cells were infected with KSHV and then transfected with muSOX twice over a 24 h period to induce accelerated RNA turnover. Pol II levels at the indicated KSHV and human promoters were measured by ChIP-qPCR and plotted with standard deviation. ( � p < 0.05, students t-test on raw % input values).
https://doi.org/10.1371/journal.ppat.1008269.g007 how a multitude of alterations to protein localization and abundance likely shape the gene expression environment during infection, and further highlight the ripple effects of accelerating mRNA decay. Notably, promoters on the replicating viral genome recruit Pol II with similar efficiency during normal or accelerated mRNA decay, indicating that they escape transcriptional inhibition and overcome the decrease in Pol II and transcription factor subunits. This protection is lost upon relocation of the viral promoters into the host genome or when the viral genome is in a latent state, emphasizing the importance of DNA context and replication state, rather than promoter sequence, in determining the transcriptional response to infection.
We expect that the reduction in Pol II and general transcription factor protein levels during WT MHV68 infection at least partially underlies the ensuing Pol II occupancy defect on the host genome. Thus, at least two events contribute to mRNA decay-linked transcriptional repression: relocalization of RNA binding proteins such as PABPC to the nucleus upon degradation of muSOX-cleaved RNAs [23,25], and depletion of key transcriptional components during infection. The fact that knockdown of Xrn1 or Dis3L2 does not rescue Pol II occupancy during WT MHV68 infection indicates that either these mechanisms are redundant, or that the Pol II depletion phenotype is dominant. The mechanism underlying reduced Pol II occupancy upon protein relocalization remains unknown, but one hypothesis is that aberrant accumulation of RNA binding proteins in the nucleus is sensed as a cellular stress signal that causes the cell to reduce its gene expression [1,23]. While both phenotypes are observed during MHV68 infection, it is interesting that GTF and Pol II subunit depletion does not occur during muSOX-induced mRNA decay in the absence of infection [23]. Instead, other downstream effects of host shutoff, such as infection-specific relocalization of RBPs, or host shutoff dependent changes to viral protein levels, could result in these changes to a key category of proteins.
Pol II activity is also impeded during infection with some RNA viruses. For example, Influenza A virus reduces Pol II subunit levels through direct binding of Rpb1 by the viral polymerase, leading to decreased Pol II occupancy at host genes [38][39][40][41]. However, unlike herpesviruses, RNA viruses do not depend on Pol II for viral transcription. Efficient recruitment of Pol II to the MHV68 genome is notable given the decreased abundance of multiple Pol II subunits, which suggests that the viral genome out competes the host chromatin to attract the remaining pool of Pol II.
Across the host genome, Pol II appears to be most strongly affected at the stage of recruitment. This is supported by the TMT-MS data, which shows a decrease in the GTFs TFIIB and the large subunit of TFIIA that are key to formation of the preinitiation complex. The reduction in promoter-proximal Pol II is propagated into the gene body, although the differences there were less pronounced, suggesting that once transcription of a gene initiates there are not subsequent blocks to elongation. This agrees with prior studies of individual host loci in MHV68 infected or muSOX expressing cells [23,25]. The population of promoter-proximal Pol II detected by ChIP comprises a combination of actively engaged Pol II as well as Pol II transiently binding to the DNA. Thus, a large reduction in promoter-proximal Pol II may not be expected to be fully borne out in the gene body [42]. That said, we did note a robust decrease of Pol II in the Rplp0 gene body of MHV68 infected cells when we used an antibody recognizing the N-terminus of Rpb1. It is therefore possible that the modest reduction in Pol II across gene bodies observed in the ChIP-seq using the 8WG16 antibody from MHV68 infected cells may underestimate the magnitude of the gene body decreases. We did not observe a defect in transcription termination, or evidence of polymerases on the DNA after TTSs, suggesting that MHV68 does not induce the termination defect and run-on transcription of host genes that occurs during infection with HSV-1 [43].
Reduced Pol II occupancy of the host genome during MHV68 infection parallels infection data from the alphaherpesvirus HSV-1. HSV-1 infection results in a genome-wide reduction in Pol II occupancy of host chromatin and reduced nascent RNA production of the majority of host transcripts [43][44][45]. Furthermore, early data showed that while the endogenous mouse β-globin locus was transcriptionally repressed during HSV-1 infection, its expression was rescued upon integration of the β-globin gene into the HSV-1 genome [46,47]. Experiments with gene-specific null viruses have implicated immediate early (IE) proteins in the HSV-1 transcriptional repression phenotype, including ICP27, which can accelerate mRNA turnover [45,48]. Thus, in addition to targeting host and viral mRNAs for cytoplasmic decay, both alpha and gammaherpesviruses restrict Pol II occupancy of the host genome, although additional work is needed to establish whether the underlying mechanism is similar.
During the lytic stage of herpesvirus infection, viral DNA is localized to replication compartments. These non-membrane bound structures are sites of viral genome replication, transcription, and packaging [49]. They are enriched for many host and viral proteins involved in these processes, including chromatin modifying factors and Pol II [34][35][36]. Replication compartments grow and coalesce during infection, excluding host chromatin, which becomes pushed to the nuclear periphery [31][32][33]. Previous analyses of cellular protein distribution by mass spectrometry revealed that muSOX and Xrn1-coordinated mRNA decay causes differential trafficking of many cellular RNA binding proteins between the cytoplasm and nucleus. At least one of these proteins, PABPC, was shown to be associated with transcriptional repression [23]. Here, in addition to observing increased PABPC in the nucleus of infected cells, we also detected PABPC in viral replication compartments. This indicates that while PABPC relocalizes to infected nuclei in a RNA decay-dependent manner, selective exclusion of PABPC from these compartments is not the mechanism by which the viral genome escapes repression. Instead, we hypothesize that either the state of viral DNA or other protective features of the replication compartment confers escape from transcriptional repression.
Protection from mRNA decay-induced transcriptional repression was not conferred to a non-MHV68 or viral promoters present on the latent viral episome, indicating that lytic phase viral DNA replication or replication compartment formation is key to this phenotype. The latent genomes of gammaherpesviruses including KSHV and Epstein Bar Virus (EBV) are heavily methylated and histone-rich, with the exception of promoters of genes expressed during latency, the latent origin of replication, and CTCF binding sites [50][51][52]. Notably, during latency key lytic promoters like RTA contain both the activating H3K4me3 and the repressive H3K27me3 histone marks, representing a 'poised' state [53]. Rapid changes shortly after lytic cycle induction result in hypomethylation and removal of repressive marks at early gene promoters [50,53]. At later stages of infection, KSHV, EBV and HSV-1 genomes have been shown to be largely nucleosome free, consistent with histones not being packaged in virions [33,35,54,55]. For gammaherpesviruses, this is the stage at which mRNA degradation is most pronounced. We therefore hypothesize that the 'open' state of viral DNA during lytic replication plays an important role in facilitating Pol II recruitment under conditions of accelerated cytoplasmic mRNA decay and may in fact be sufficient to explain the robust recruitment of Pol II to replication compartments. It is possible that the open state of the DNA, particularly late in infection, also contributes to the relatively uniform occupancy of Pol II across MHV68 genes and absence of clear promoter-proximal peaks. It will be interesting to investigate if the general chromatin state of the host genome also changes in infection or if is largely the lack of chromatinization of the viral genome that is driving these changes.
In conclusion, decay-coupled transcriptional repression is a novel facet of viral manipulation of the cellular gene expression landscape. We hypothesize that the ability to escape from transcriptional repression is linked to the open conformation of viral DNA in replication compartments, and that this helps these viruses counteract decay-induced viral transcript loss and overcome depletion of Pol II and general transcription factor protein subunits. Future work is geared towards exploring whether widespread mRNA decay alters cellular chromatin states and DNA accessibility and, if so, how this influences transcription factor recruitment.
Primers used for qPCR and for cloning are listed in S2 Table. M7, M4 and CMV-puro were made by cloning the respective promoter sequence upstream of the puromycin cassette in a pLKO-sgRNA vector (Addgene 131697). The promoter of M4 was defined as the 1131 bp region upstream of the M4 coding region not overlapping with other viral ORFs and the promoter sequence of M7 was previously published [56]. pGL3-kLANA-puromycin (Addgene 131699) was made by introducing the puromycin gene into the I-CeuI and XbaI sites of pGL3-kLANA via InFusion (Clonetech) cloning. pGL3-kLANA was made by amplifying the KSHV LANA promoter sequence [57] and using InFusion cloning to introduce it into the XhoI and HindIII sites of pGL3.
DNA transfections were carried out in 293T cells at 70-90% confluency in 10 cm plates with 10 μg host-shutoff factor plasmid or vector control using PolyJet (SignaGen) 24 hours before harvesting and again at 18 hours before harvesting.
Lentiviral transduction was carried out by spinfecting 1x10 6 293T or MC57G cells with lentivirus made from 2 nd generation plasmids at a MOI <1 for 2 hours with 4 μg/mL polybrene (Fisher). Twenty-four hours later puromycin (1 μg/ml for 293T or 3 μg/ml for MC57G) was added to select for transduced cells. 293T cells stably expressing the kLANA-puromycin cassette were made by transfecting the cells with 1 μg pGL3-kLANA-puromycin using PolyJet, then selecting with puromycin for random integration events.
For the CMV-luciferase promoter assay, populations of 293T cells expressing muSOX or control GFP and the cell surface marker Thy1.1 were selected 24 hours post-transfection using the Miltenyi Biotec MACS cell separation system according to the manufacturer's instructions as previously described [23]. Briefly, muSOX and GFP were each expressed from a vector containing an upstream Thy1.1 sequence separated from muSOX or GFP by a P2A ribosome skipping site, which results in separation of Thy1.1 from the downstream gene. Transfected cells were then passed over a column to enrich for Thy1.1-expressing cells, which were subsequently eluted and subjected to 4SU pulse labeling prior to harvesting.
293T cells were infected with KSHV by reactivating iSLK-Bac16 cells for 48 h and then spinfecting cell-free virus onto a monolayer of 293T cells, as described for lentiviral production but in the absence of polybrene.
3 million MC57G cells were nucleofected with a 100uL Neon tip (1400 Voltage/20 Width/2 Pulse) to knockdown Xrn1 or Dis3L2 using a pool of 4 siRNAs or a non-targeting control using 10 μL of 20 μM siRNA.

4SU labeling
Adapted from [59]. Cells were incubated in suspension with 500 μM 4SU (Sigma T4509) in DMEM supplemented with 10% FBS for 10 min then washed with PBS. RNA was extracted with TRIzol followed by isopropanol purification. RNA (300 μg) was incubated in biotinylation buffer (50 mM HEPES [pH 7.5], 5 mM EDTA) and 5 μg MTSEA-biotin (Biotium #90066), rotating at room temperature in the dark for 30 min. A no biotin control was made from an equal amount of total RNA and incubated as above but without the addition of MTSEA-biotin. RNA was then phenol:chloroform extracted and precipitated with isopropanol. The pellet was resuspended in DEPC-treated water and mixed with 50 μL Dynabeads MyOne streptavidin C1 (Invitrogen) that had been pre-washed twice with wash buffer (100 mM Tris [pH 7.5], 10 mM EDTA, 1 M NaCl, 0.1% Tween 20). Samples were rotated in the dark for 1 hour at room temperature, then washed twice with 65˚C wash buffer and twice with room temperature wash buffer. Samples were eluted twice with 100 μL 5% Beta-mercaptoethanol (BME) in DEPC H 2 0 for 5 min then RNA was precipitated with isopropanol and quantified by RT-qPCR.

Viral mutagenesis, propagation and infections
The WT and R443I MHV68 bacterial artificial chromosome (BAC) were previously described [7,60]. MHV68 was produced by first making p0 virus by transfecting NIH 3T3 cells in 6-well plates with 2.5 μg BAC DNA using TransIT-X2 (Mirus Bio) for 24h. Five to seven days later, the cells were split into a 10 cm dish, then harvested and frozen. After 5-7 days the majority of cells showed cytopathic effect (CPE). 30 μL of p0 was then added per confluent 10 cm dish of NIH 3T12 cells, split 2 days later into four 10 cm dishes and harvested 4-6 days later when all cells were infected. Cells were collected by centrifugation (5 min, 1500 rpm) and the pellet was dounced 10 times. The cell pellet and supernatant were then ultracentrifuged for 2 h at 30,000 rpm. The resulting pellet was resuspended and titered by plaque assay. Cells were infected with MHV68 at an MOI of 5 for 24 hours.

RT-qPCR
RNA was reverse transcribed using AMV RT (Promega) with random 9-mer primers. Total RNA was DNase treated with TURBO DNase (ThermoFisher) or DNase treated on column with Zymo Direct-zol RNA MiniPrep Plus DNase. cDNA was quantified using iTaq Universal SYBR Mastermix (BioRad) and transcript-specific primers. All qPCR results are normalized to 18S levels and WT or vector control set to 1.

Immunofluoresence
NIH 3T3 cells were plated on coverslips (7.5 x 10 4 cells/well of a 12-well), infected the following day for 25-27 h, and then fixed in 4% formaldehyde for 10 min. iSLK-Bac16 cells were plated on coverslips (5 x 10 4 cells/well of a 12-well), reactivated 24h later with doxycycline and sodium butyrate for 48 hours and then fixed as above. Cells were permeabilized with ice-cold methanol at -20˚C for at least 20 min and incubated with anti-Pol II antibody (Biolegend, 8WG16 at 1:200), anti-PABPC antibody (Abcam ab21060 at 1:200 for mouse cells and Santa Cruz sc32318 at 1:25 for human cells) or anti-ORF59 antibody (Advanced Biotechnologies 13-211-100 at 1:200) in 5% BSA overnight at 4˚C. Secondary antibodies were added (1:1000) for 1 h at 37˚C. Coverslips were mounted in DAPI-containing Vectashield (VectorLabs). We identified RCs by Pol II recruitment or ORF59 staining corresponding with DAPI-poor regions [34,36,61]. Images were collected on a Zeiss LSM 710 AxioObserver with a 40x oil objective. Cells with RCs were first identified, then PABPC co-localization was determined by counting cells with at least 2x nuclear pixel intensity relative to non-RC containing cells in the same image (ImageJ). , and Tris-EDTA for 5 min each at 4˚C with rotation. DNA was eluted from the beads using 100 μl of elution buffer (150 mM NaCl, 50 μg/ml Proteinase K) and incubated at 55˚C for 2 hours, then 65˚C for 12 hours. DNA was purified using a Zymo Oligo Clean & Concentrator kit, and quantified by qPCR using primers to the promoter regions of indicated genes for 50 cycles. qPCR was simultaneously performed on input chromatin and normalized to input. Raw percent input values are normalized by setting mock infected or a transfection control to 1 and are plotted as relative % input. IgG pull downs were performed with the MHV68 infected sample under scramble conditions, or the muSOX or SOX transfected sample for transfection experiments.

ChIP Sequencing and data analysis
Libraries were prepared for sequencing (Kapa Hyper Prep) using equal amounts of chromatin within each experiment, amplified for 10-12 cycles based on Kapa's recommendations and sequenced on a HiSeq4000 with 100bp single end reads. Sequencing quality was assessed with FastQC and trimmed with Sickle. A custom index was made of the mm10 and MHV68 genomes with Bowtie2 build. Reads were mapped to that index using Bowtie2 (2.3.0).
To visualize reads in the Integrative Genomics Viewer (IGV), tdf files were made from bam files and indexes made from those bam files. To make tdf files for the viral genome, a chrom. size file was created for MHV68 and then the count command was run. Analysis of mapped reads was performed with HOMER (Hypergeometric Optimization of Motif EnRichment) [62]. Tag directories were created from the bam files generated from Bowtie with the make-TagDirectory command. This was done separately for the mouse and MHV68 genomes. For mouse genes, the analyzeRepeats command was used to look at tag density in the body of genes and the-c pausing option was used to compare pausing ratios. annotatePeaks was run in both the-hist and-ghist mode to assess composite Pol II occupancy across the mouse genome as well as individual gene occupancy. Viral analyses were done by first creating a custom genome annotation in HOMER using the loadGenome command. The TSS file was created from data generously shared by Scott Tibbetts (University of Florida) including portions of the viral BAC provided by Laurie Krug (Stony Brook University). annotatePeaks was run in the-hist mode to assess composite Pol II occupancy across the viral genome as well as individual gene occupancy. Sequencing data are available on GEO repository (accession number GSE132574).

Fractionation
NIH3T3 cells were fractionated into nuclear and cytoplasmic fractions using membrane filters as previously described [63]. Briefly, cells were washed twice with ice-cold PBS and the cell pellet was resuspended in 0.25 M sucrose, 25 mM KCl, 5 mM MgCl2, 20 mM HEPES-KOH, pH7.4 and incubated on ice for 10 min. The cells were lysed by passing them through a 14 μm polycarbonate membrane (Sterlitech Corporation). The nuclei were then isolated by centrifugation at 1,400 x g for 10 min and the supernatant was retained as the cytoplasmic fraction.

Cell lysis, protein digestion and TMT labeling
Lysis, digestion, and TMT labeling were performed as previously described [23] with a few modifications. Briefly, nuclei were lysed in 100 mM Tris-HCl, pH 8.0, 4% SDS, 1 mM EDTA preheated to 70˚C and cytoplasmic fractions were adjusted to 1% SDS. Complete lysis was achieved by successive rounds of heating and sonicating. After protein concentration assessment by BCA assay (Pierce), 100μg of protein from each sample was reduced and alkylated with 25 mM tris(2-carboxyethyl)phosphine (Pierce) and 50 mM chloroacetamide respectively and then the protein was precipitated via methanol-chloroform cleanup [64]. Samples were digested overnight at 37˚C in 50 mM HEPES pH 8.3 at a 1:50 trypsin:protein ratio. Peptides from all three biological replicates were labeled with a 6-plex TMT kit following the scheme in Fig 3A and pooled in equal peptide amounts, resulting in three individual 6-plex experiments. Pooled peptides were fractionated by 2D StageTip as described [23] with the following changes: from the SCX fractionation, fractions 1 and 4 were pooled together and fractions 2 and 3 were pooled together prior to fractionation by SDB-RPS StageTips.

LC-MS/MS and informatic analysis
Peptides were analyzed via LC-MS/MS using a Dionex Ultimate 3000 UPLC coupled online to an EASYSpray ion source and Q Exactive HF. A linear gradient of 5% ACN to 42% ACN in 0.1% FA over 150 min on an EASYSpray C18 column (75 μm x 50 cm) heated to 50˚C was used to separate the peptides. Following ionization at 1.7kV, an MS1 scan was performed from 400 to 1800 m/z at 120,000 resolution with an automatic gain control (AGC) setting of 3e6 and a maximum injection time (MIT) of 30 ms recorded in profile. The top 15 precursors were then selected for fragmentation and MS2 scans were acquired at a resolution of 30,000 with an AGC setting of 1e5, a MIT of 42 ms, an isolation window of 0.8 m/z, a fixed first mass of 100 m/z, normalized collision energy of 33, intensity threshold of 2e5, peptide match set to preferred, and a dynamic exclusion of 45 s recorded in profile.
MS/MS data were analyzed by Proteome Discoverer (Thermo Fisher Scientific, v2.2.0.388). The nuclear channels (126 -128) and cytoplasmic channels (129-131) were analyzed in separate Proteome Discoverer studies as described before [23]. A fully tryptic search against a mouse Uniprot database appended with MHV68 and common contaminant sequences (downloaded 10/2016-17,149 sequences) requiring 5 ppm mass accuracy on the precursor ions and 0.02 Da accuracy on the fragment ions was performed. Static carbamidomethyl modifications to cysteine, static TMT additions to peptide N-termini and lysine residues, dynamic oxidation of methionine, dynamic deamidation of asparagine, dynamic methionine loss and acetylation of protein n-termini, and dynamic phosphorylation of serine, threonine, and tyrosine were allowed as modifications in the search. Matched spectra were scored by Percolator and reporter ion signal-to-noise values were extracted. Following parsimonious protein assembly at a 1% FDR for proteins and peptides, reporter ion quantitation was performed for unique and razor peptides with an average S/N of at least 10 and a precursor co-isolation threshold of less than 30% which did not contain a variable modification and normalized to the total detected signal in each TMT channel. Protein abundances were calculated as the sum of all reporter ion values in a particular channel for each protein. Imputation of missing values was performed by low abundance resampling. The data were scaled based on the Mock infection samples (channels 126 and 129). Statistically differential proteins were assessed via the background based t-test analysis implemented in Proteome Discoverer. The resulting data was exported to Excel for further analysis. The mass spectrometry proteomics data reported in this paper have been deposited at the ProteomeXchange Consortium via the PRIDE partner repository [65]. The PRIDE accession number is PXD015786.
Supporting information S1 Fig. ChIP-seq replicate with WT and R443I MHV68. A) Pol II ChIP-seq signal profiles of host genes are shown in mock-infected, MHV68-infected and R443I-infected cells. Each row of the heat map displays Pol II occupancy of one gene from -1000 to +1000 in 25 bp bins. Genes are ranked by the Pol II-transcription start site (TSS) proximal signal in mock infected cells. (B) Sequence tags were plotted as a histogram in 25 bp bins for -2000 to +4000 around the TSS. Mock, MHV68 (red) and MHV68 (blue) and R443I (green) traces are shown along with their input controls. (C) ChIP-qPCR validation of Pol II occupancy at the Rplp0 and Fus promoters using an antibody specific for the N-terminus of Rpb1 plotted with standard deviation. Pol II ChIP was performed on mock, MHV68 WT or MHV68 R443I infected MC57G cells and Pol II levels were assayed near the TSS of two repressed host genes during MHV68 infection from the ChIP-seq data. IgG is from the MHV68 infection condition. ( � p < 0.05, �� p < 0.001, students paired t-test on raw % input values) (D) Pol II transcription termination is not dependent on RNA decay. Sequence tags were plotted as a histogram in 25 bp bins for transcription termination sequence (TTS) proximal Pol II for -1000 to +1000 around the TTS with the same color scheme as (B). (TIF)

S2 Fig. TMT-MS fractionation validation and Panther DB terms broken down by cellular compartment and condition.
A) Reporter ion abundance from the TMT-MS data showing that the nuclear and cytoplasmic distribution of the nuclear protein H4 and the cytoplasmic protein GAPDH are primarily detected in their correct compartments, demonstrating successful fractionation. Graphs display the mean with standard deviation of 9 biological replicates including mock, MHV68 and R443I infection conditions. (B-D) Gene ontology terms for proteins increased and decreased in each compartment in a host shutoff dependent manner. Lists were generated by taking all proteins with a log 2 fold change greater than 0.2 comparing WT MHV68 to R443I and looking at the molecular function enrichment in Panther DB [66]. Terms with fold enrichment greater than 6 were included for the cytoplasm and greater than 5 for the nucleus.  Table. Reporter ion abundances from TMT-MS of NIH3T3 mouse fibroblasts infected with WT MHV68 or R443I MHV68. NIH3T3 mouse fibroblasts were infected with WT MHV68 or R443I MHV68 then fractionated into nucleus and cytoplasm, labeled with tandem mass tags and analyzed by quantitative liquid chromatography/mass spectrometry. This table provides protein identification information, normalized and scaled reporter ion abundances for each compartment. (XLSX) S2 Table. List of all DNA sequences used in this study. (XLSX)