Merkel Cell Polyomavirus DNA Replication Induces Senescence in Human Dermal Fibroblasts in a Kap1/Trim28-Dependent Manner.

We here describe Kap1 as a restriction factor in MCPyV infection. We report a novel, indirect mechanism by which Kap1 affects MCPyV replication. In contrast with from other DNA viruses, Kap1 does not associate with the viral genome in MCPyV infection and has no impact on viral gene expression. In MCPyV-infected nHDF cells, Kap1 phosphorylation (pKap1 S824) accumulates because of genomic stress mainly induced by viral DNA replication. In contrast, ectopic expression of LT or LT MCPyV mutants, previously shown to be important for induction of genotoxic stress, does not result in a similar extent of pKap1 accumulation. We show that cells actively replicating MCPyV accumulate pKap1 (in a manner dependent on the presence of ATM) and display a senescence phenotype reflected by G2 arrest. These results are supported by transcriptome analyses showing that LT antigen, in a manner dependent on the presence of Kap1, induces expression of secreted factors, which is known as the senescence-associated secretory phenotype (SASP).

Here, we show that Kap1 is in a complex with MCPyV T-Ags sT and LT. Using Kap1 knockout cells, we showed that MCPyV DNA replication levels are increased in the absence of Kap1 and that partial complementation can be performed by reintroducing Kap1. In contrast to other DNA viruses, Kap1 is not recruited to MCPyV DNA and viral gene expression remains unaffected. Instead, our data obtained by analyses of normal human dermal fibroblast (nHDF) cells transfected with religated MCPyV genome or infected with MCPyV suggest that MCPyV DNA replication induces genotoxic stress followed by ATM-mediated phosphorylation of Kap1, a phenotype that is not induced by LT overexpression alone. Consequently, MCPyV-replicating nHDF cells arrest in G 2 and undergo senescence. We hypothesize that Kap1-induced senescence is a host restriction mechanism against MCPyV replication in nHDF cells.

RESULTS
Kap1 interacts with the early gene products of MCPyV. We recently identified chromatin-associated protein Kap1 as a putative cellular binding partner of the MCPyV early gene products by mass spectrometry (22). To confirm this initial observation and to characterize which early gene product interacts with Kap1, we performed coimmunoprecipitation (Co-IP) experiments in H1299 cells ectopically expressing the FLAGtagged early region (ER) protein, FLAG-tagged LT protein, or FLAG-tagged sT protein (Fig. 1A). Furthermore, we show that this coprecipitation was independent of protein tags (Fig. 1B). To specify the protein regions required for Kap1 LT interaction, we performed Co-IPs in H1299 cells overexpressing FLAG-tagged Kap1 protein and yellow fluorescent protein (YFP)-tagged MCPyV full-length early gene region (ER) protein or ER protein with the truncating mutations that follow within the LT-Ag open reading frame (ORF) and lead to early termination after the Zinc finger domain, after the origin binding domain (OBD), or upstream of the OBD (as can be found in MCC cells). Co-IPs performed with an anti-FLAG antibody (Ab) showed that the full-length as well as all the C-terminally truncated MCPyV ER products interacted with the FLAG-tagged Kap1 protein, suggesting that residues within the N-terminal regions that are shared between LT and sT are important for Kap1 binding of T-Ags (Fig. 1C). To clarify whether the DNAJ domain (amino acids [aa] 1 to 70) in the N-terminal regions of both LT and sT (11) confers Kap1 binding, we performed glutathione S-transferase (GST) pulldown experiments with HEK293 cell lysates and bacterially expressed LT deletion mutants N-terminally fused to GST encompassing aa 1 to 258, aa 171 to 258, or aa 79 to 170 (20). While we observed a Kap1 binding in the case of the deletion mutant encompassing aa 1 to 258 and, to a lesser extent, in that encompassing aa 79 to 170, we did not observe binding to endogenous Kap1 in that encompassing aa 171 to 258 (Fig. 1D). Reciprocally, Co-IPs addressing which Kap1 domain is important for LT binding revealed that the Kap1 N-terminal RBCC domain is essential for LT precipitation (Fig. 1E).
Kap1 depletion increases MCPyV DNA replication. To evaluate the consequences of Kap1/T protein interaction, we performed MCPyV DNA replication assays using a semipermissive system (12,15,16,35,36) in cells positive or negative for Kap1 expression. We generated HEK293 cells ( Fig. 2A and B) or H1299 cells ( Fig. 2C and D) in which Kap1 was deleted by CRISPR/Cas9 ( Fig. 2A and C). The levels of proliferation of Kap1 knockout cells did not significantly differ from the levels seen with the control (CON) cells (Fig. 2G). We then transfected Kap1 knockout and parental control cells with recircularized MCPyV genomes, isolated the genomic DNA at the indicated time points, and performed DpnI digestion and quantitative PCR (qPCR) analysis of replicated DNA. Intriguingly, we observed a significant increase in MCPyV DNA replication in Kap1 knockout cells at 4 and 8 days posttranslation (p.t.) ( Fig. 2B and D). This phenotype of increased MCPyV DNA replication could be partially reversed by lentivirus-mediated reexpression of full-length Kap1 in Kap1 knockout cells but not by reexpression of a Kap1 protein missing the RBCC domain, which is responsible for LT binding (Fig. 2E to G). At 48 h p.t., cell lysate experiments were performed, FLAG-tagged T proteins precipitated with M2 beads (upper blot, IP), and coprecipitated endogenous Kap1 was detected with an anti-Kap1 antibody (lower blot, Co-IP). (B) Endogenous Kap1 coprecipitated with untagged MCPyV T antigens. Cells were transiently transfected with untagged LT or sT; using methods similar to those used in the experiments whose results are shown in panel A, cell lysates were prepared, endogenous Kap1 was precipitated with an anti-Kap1 antibody, and coprecipitating T-Ags were detected with Cm2B4 Ab (LT) or 2T2 Ab (sT). (C) The N-terminal region of MCPyV LT-Ag interacts with Kap1. H1299 cells were transiently transfected with FLAG-tagged Kap1 in the presence of YFP-tagged MCPyV early region constructs. Kap1 was precipitated with an anti-FLAG antibody (IP), and coprecipitated T antigens were detected using a GFP antibody (Co-IP). trunc., truncated. (D) N-terminal region aa 1 to 258 is essential for Kap1 binding. GST pulldown experiments were performed with GST-tagged LT deletion constructs (20). Proteins extracts from HEK293 cells were incubated with GST-tagged proteins, and bound Kap1 protein was visualized by Western blotting using Kap1-specific antibody (upper panel). The middle panel illustrates the amount of GST fusion proteins used in the experiment as illustrated by Coomassie staining. GST pulldown assay data show that the DNAJ domain contributed to LT binding; however, aa 171 to 258 also mediated Kap1 binding. (E) The RBCC domain of Kap1 interacts with MCPyV LT-Ag. H1299 cells were transiently transfected with FLAG-tagged Kap1 expression constructs in the presence of the YFP-tagged MCPyV early region. Ectopically expressed Kap1 proteins were precipitated with an anti-FLAG antibody (IP), and precipitating T proteins were detected by the use of Cm2B4 antibody (Co-IP).

Subcellular Kap1 localization does not change in cells replicating MCPyV.
To address whether LT and Kap1 colocalize in cells actively replicating MCPyV, we performed immunofluorescence (IF) staining in nHDF cells transfected with MCPyV genome. While nHDF cells supported MCPyV DNA replication to a higher extent than PFSK-1 cells (see Fig. S1A to C in the supplemental material), we did not observe significant differences in Kap1 localization between cells positive or negative for LT expression (Fig. 3). In both groups, the calculation of the Pearson correlation coefficient indicated low to moderate colocalization, with means of 0.56675 in cells with diffuse LT staining and 0.56675 in cells clearly showing an LT replication center. Kap1 localization was characterized by diffuse and slightly granular nuclear staining results.
Kap1 is not recruited to viral DNA and has no direct influence of LT binding to the viral ori. Since previous studies had demonstrated Kap1 recruitment to herpesviral control regions (CRs) such as the KSHV replication and transcription activator (RTA) promoter (32), we examined the binding of Kap1 at the MCPyV origin of replication (ori). We performed chromatin immunoprecipitation-qPCR (ChIP-qPCR) and ChIP sequencing (ChIP-seq) with an anti-Kap1 antibody and a control antibody in HEK293 cells transiently transfected with an MCPyV ER expression construct and an additional plasmid carrying the MCPyV ori region (Fig. 4). The precipitated DNA was quantified by qPCR with primers specific for the ori region, cellular promoter regions known to be bound by Kap1 in HEK293 cells (positive controls; ZNF180 and ZNF274), or regions in the cellular GAPDH (glyceraldehyde-3-phosphate dehydrogenase) and ACHE genes (negative controls) (37). As shown in Fig. 4, Kap1 levels were significantly enriched at the ZNF180 and ZNF274 promoter regions compared to the negative controls. In contrast, we observed no enrichment at the viral ori, regardless of whether LT was present or absent (Fig. 4). Interestingly, we observed decreased Kap1 binding at the cellular control regions in cells overexpressing the ER protein ( Fig. 4A and C). We included LT ChIP PCR as a control to enrich for DNA regions located on relatively small, circular DNAs (Fig. 4B). We confirmed the enrichment of Kap1 at cellular promoter regions by ChIP-seq and observed statistically significant enrichment of Kap1 at host  promoter regions compared to the viral ori (Fig. 4C). These results demonstrate that Kap1 is not recruited to the viral noncoding control region (NCCR). In line with this finding, we did not observe an effect of Kap1 on the expression of early or late viral transcripts as indicated by qPCR performed in viral replication assays in cells positive or negative for Kap1 expression ( Fig. S1E and F).
To investigate whether the restriction in MCPyV DNA replication was due to modulation of LT binding to the viral ori region, we performed ChIP-qPCR experiments with an anti-LT Ab (Cm2B4) in HEK293 cells and HEK293 Kap1 knockout cells transiently transfected with LT and the MCPyV genome. As expected, we observed LT enrichment at the viral ori (Fig. S2A); however, this enrichment remained unchanged in the presence or absence of Kap1 ( Fig. S2A and B). We obtained similar results by performing electrophoretic mobility shift assays (EMSAs) in HEK293 cells and HEK293 Kap1 knockout cells (Fig. S2C) and DNA-protein interaction-enzyme-linked immunosorbent assays (DPI-ELISAs) using streptavidin-binding-peptide (SBP) tag-purified LT from HEK293 cells and HEK293 Kap1 knockout cells (Fig. S2D to F).
MCPyV DNA replication induces Kap1 phosphorylation in cells transfected with MCPyV genome and cells infected with MCPyV. Kap1 abundance and functions are regulated by posttranslational modifications. Kap1 SUMOylation and subsequent CHD3 and SETDB1 recruitment result in a repressive function of Kap1. Kap1 can also be phosphorylated, mainly at serine 824 and serine 473 in the context of DNA damage responses (Fig. 5A). Phosphorylation at these residues counteracts Kap1 SUMOylation, resulting in remodeling of heterochromatic regions and increased accessibility to proteins of the DNA damage response pathway.
To analyze whether the phosphorylation levels of Kap1 change in cells actively replicating MCPyV, we transfected primary nHDF cells with religated MCPyV genomes. At 8 days p.t., we determined pKap1 S824 levels by immunofluorescence. We detected that cells positive for LT showed a significant increase in the levels of pKap1 S824 ( Fig. 5B and C). Additionally, upon closer inspection of the LT-positive fraction, we found the levels of pKap1 S824 to be significantly higher in cells with distinct and large MCPyV replication centers than in those with a more uniform staining pattern, suggesting that ongoing DNA replication rather than LT expression alone induces Kap1 phosphorylation (Fig. 5D). This notion is furthermore supported by the fact that Kap1 phosphorylation was also observed in nHDF cells infected with a replication-competent MCPyV ( Fig. 6A and B) but not in nHDF cultures transduced with LT or sT expression constructs alone ( Fig. S3C and D).
Since Kap1 phosphorylation can be induced by ATM and ATM is substantially activated in cells replicating MCPyV (38), we analyzed Kap1 phosphorylation in the presence or absence of an ATM inhibitor, KU55933. Results of experiments in nHDF cells transfected with replication-competent MCPyV genomes ( Fig. 5D and E) or infected with MCPyV particles (Fig. 6C) indeed indicate that phosphorylation of Kap1 on serine 824 is dependent on ATM. Furthermore, when we treated cells with an ATM inhibitor, shown are ChIP-qPCR results representing positive-control (ZNF180 and ZNF274) and negative-control (ACHE and GAPDH) regions for Kap1 binding. (B) ChIP PCR experiment performed similarly to the experiments whose results are shown in panel A, with an additional ChIP performed using the anti-LT Ab Cm2B4. A positive-control region of LT binding is represented by the origin of replication (Ori), while VP1 represents a negative-control region. (C) ChIP-seq analysis (n ϭ 1) of Kap1 binding to cellular and ori-containing plasmid DNA in either control cells (-) or ER-expressing HEK293 cells (ϩ). Kap1 peaks (n ϭ 456) were detected on the host genome using MACS2. The cellular ChIP background (bg) level was determined using the same amount of matched control sites generated by EaSeq. All data representing host and ori results are represented as relative levels of read count enrichment in the respective region windows over the input sample. Host background levels in both samples were set to a value of 1 to enable comparability of data from ER-expressing and control cells as well as host loci and plasmids. The statistical significance of data representing differential Kap1 binding to host cell loci was determined using a two-tailed t test. Three background regions on the ori plasmid (scatterplot; data represent means and standard errors of the means [SEM]) and the ori region itself were used to analyze the relative levels of Kap1 binding to the ori. Siebels et al. ® we observed an increase in the abundance of cells with MCPyV DNA replication centers together with an increase of MCPyV copy numbers, indicating that ATM-dependent phosphorylation of Kap1 is critical for MCPyV replication ( Fig. 5F to H).
On the basis of our observations that the viral early proteins coprecipitated Kap1 and that cells transduced with replication-competent MCPyV genomes induced pKap1, we subsequently analyzed whether LT and sT also coprecipitate with phosphorylated Kap1. Due to insufficient numbers of cells being positive for MCPyV in in vitro replication and MCPyV infection, we performed these experiments in HEK293 cells overexpressing the early region (ER), LT, or sT protein (Fig. 6D). Interestingly, the HEK293 cells, in contrast to the nHDF cells, showed pKap1S824 induction in response to viral ER expression (Fig. 6D, lane 2; see also Fig. S6E). Immunoprecipitation of FLAG-tagged early viral proteins showed that pKap1S824 was coprecipitated only in the case of viral ER expression, which also results in significantly higher LT expression levels and pKap1S824 induction.
Since Kap1 function is regulated by phosphorylation and SUMOylation, which counteract each other, we next analyzed potential changes of Kap1 SUMOylation levels. For this purpose, we used HeLa cells stably expressing His-SUMO1 or His-SUMO2 (39). Western blotting of nickel-nitrilotriacetic acid (Ni-NTA)-purified His-SUMO1/2 conjugates demonstrated clear conjugation of the endogenous Kap1 protein to SUMO2, while no Kap1 complexes were retrieved from His-SUMO1 cells (Fig. S4). Transfection of HeLa His-SUMO2 cells or the corresponding parental cells with MCPyV ER or vector control (mock) resulted in decreased SUMO2-Kap1 levels (Fig. S4, lower panel, lanes 4 and 8), suggesting that expression of LT or sT or both is sufficient to reduce Kap1 sumoylation.
MCPyV replication induces Kap1-dependent senescence in nHDF cells. Kap1 has recently been shown to confer senescence in cells responding to DNA damage (40). To investigate whether MCPyV replication, subsequent ATM activation, and phosphorylation of Kap1 S824 would result in a senescent phenotype, we transfected nHDF cells with replication-competent MCPyV genomes and stained for expression of ␤-galactosidase, a protein that functions as a marker of senescent cells, at day 10 p.t. (Fig. 7). Indeed, cells positive for LT expression showed highly increased ␤-galactosidase staining compared to LT-negative cells (Fig. 7A). We did not succeed in quantification of sufficient numbers of these cells for statistical analysis, since only a minor fraction of the cells supported MCPyV replication as reflected by results of analyses of replication compartments. However, as the senescent phenotype is generally accompanied by the activation of p21, we chose to investigate p21 transcript levels instead. As shown in Fig. 7, we found p21 transcript levels to be significantly increased in nHDF cells replicating MCPyV (Fig. 7B). Cells ectopically expressing LT, sT, or ER ( Fig. 7C) did not exhibit p21 upregulation, again suggesting that active DNA replication induces the phenotype and not early gene expression alone.
Since we did not find a direct repressive role of Kap1 with regard to viral gene expression, we aimed at elucidating the role of Kap1 by performing host transcriptome analysis. Accordingly, HEK293 cells and HEK293 Kap1 knockout cells were transiently transfected with an ER construct and at 48 h p.t., mRNA was isolated and subjected to transcriptome analyses (Fig. S6). We first focused on significant changes (log 2 fold change of Ͼ1; false-discovery rate [FDR] of Ͻ0.1) dependent on the presence or absence of Kap1 in either ER-negative or ER-positive cells by comparing transcriptome data from HEK293 Kap1 knockout cells with data from HEK293 cells or HEK293 Kap1 knockout ER cells with HEK293 ER cells, respectively (Fig. 8A, comparisons [columns] 1 and 2). In line with previous reports (27), we found that Kap1 was involved in the repression of transcription, mainly of that of ZNF proteins (see Table S2 in the supplemental material). To look for significant changes induced by the ER proteins in either the presence or absence of Kap1, we compared HEK293 ER with HEK293 cells or HEK293 Kap1 knockout ER with HEK293 Kap1 knockout cells, respectively (comparisons 3 and 4). Gene ontology (GO) term analysis of differentially expressed genes (DEGs) in comparisons 3 and 4 (see Table S2) suggested that ER expression regulates a number  Table S2). GO analysis revealed that genes involved in organization of the extracellular matrix, cell adhesion, inflammatory responses, and cell proliferation are differentially expressed in a manner dependent on the presence of both Kap1 and MCPyV ER. In senescent cells, upregulation of inflammatory cytokines, chemokines, extracellular matrix remodeling factors, and growth factors contributes to the so-called senescenceassociated secretory phenotype (SASP) (41)(42)(43). As shown in Fig. 9 (see also Fig. S7), we found several of the functional classes associated with SASP, in particular, cytokines and growth factors, to be upregulated in cells expressing the ER and Kap1.
To further substantiate the hypothesis that LT-dependent viral DNA replication, but not LT expression alone, induces Kap1 phosphorylation in nHDF cells, we generated full-length viral genomes expressing an LT protein with a mutation in the DNA-binding domain (K331A) (44). Immunoblotting, DNA-protein interaction-ELISA (DPI-ELISA), and in vitro replication assays verified that the mutant LT-Ag was efficiently expressed but was unable to bind to the viral origin and unable to support DNA replication (Fig. S5). As shown in Fig. 10, the viral mutant was unable to induce Kap1 S824 phosphorylation (Fig. 10A, C, and D) or upregulation of p21 transcription (Fig. 10B and E). In contrast, the wild-type (wt) virus and a control virus expressing a replication-competent LT mutant (S861A) that was previously shown to exhibit lower levels of genotoxic stress in C33A cells (38) were able to efficiently induce phosphorylation of Kap1 S824 and p21 activation. Given previous observations of a growth-inhibiting phenotype mediated by expression of full-length LT in different cell types (18,36,38), we evaluated changes in the proliferation rates of nHDF cells transfected with wt or mutant MCPyV genomes. As shown in Fig. 11, cultures transfected with wt genomes or the S8631 mutant, but not those transfected with the K331A mutant, exhibited significantly lower proliferation rates than were seen with mock-transfected fibroblasts (Fig. 11). These data are furthermore supported by cell cycle analyses (Fig. 12). While we observed a clear accumulation of cells in G 2 in nHDF cells transfected with replication-competent MCPyV genomes (Fig. 12A to C), this cell cycle arrest phenotype was less prominent in cells receiving the replication-incompetent MCPyV K331A genome. Interestingly, the replication-competent MCPyV S816A genome, which was previously shown to be less genotoxic than the wt genome (38), showed a significantly reduced proliferation rate and a slight increase in cell numbers in G 2 (Fig. 12B).

DISCUSSION
Here, we demonstrate that chromatin-associated factor Kap1/TRIM28 serves a restriction factor for MCPyV in primary human dermal fibroblasts, a cell type previously identified as a putative primary target of MCPyV infection in vivo (14). We identified Kap1 as a protein coprecipitating with the early gene products LT and sT. We show that  (Table S3). To evaluate enrichment of these genes in Kap1-dependent or ER-dependent DEGs, a hypergeometric test was performed for each of comparisons 1 to 6, with 20,200 protein coding genes representing the total population and 202 senescence genes and a SASP-associated gene representing successes. The graph shows negative logarithms of the resulting P values. (B) The genes associated with the SASP of comparison 6 (LT versus Kap1 knockout) were ranked, and the 20 most highly differentially expressed genes are presented according to their log 2 fold change values. Before the ranking was performed, genes with a baseMean value of Ͻ10 were rejected. Siebels et al. Kap1 Restricts MCPyV DNA Replication ® cells devoid of Kap1 showed increased MCPyV DNA replication. This phenotype is independent of cell proliferation, since we did not observe a significant difference in the levels of proliferation in HEK293 or H1299 cells devoid of Kap1 compared to control cells. Increased MCPyV DNA replication in the absence of Kap1 was able to be partially reversed by overexpressing Kap1 in Kap1 knockout cells. However, a mutant that is devoid of the RBCC domain in Kap1, which is responsible for the interaction with the T-Ags, no longer rescued the observed phenotype.
Kap1 has previously been shown to repress viral transcription of several herpesviruses (KSHV, EBV, and CMV) but also that of parvoviruses, e.g., adeno-associated virus (AAV) (28)(29)(30)(31)(32)(33)45). For KSHV and CMV, Kap1 regulates latency by inducing the repressive histone modification H3K9me3 at lytic genes, while latent genes stay free of repressive marks (31,32). In the early phase of KSHV infection, LANA (latency-associated nuclear antigen) recruits Kap1 to the KSHV genome to shut down lytic gene expression. Similarly, Kap1 restricts AAV replication by recruiting histone methyltransferases and depositing H3K9 marks at viral promoters. When Kap1 is depleted, these viruses show increased viral replication and viral gene expression. Differently from the results described above, we identified an indirect mechanism by which Kap1 protects the cell against damage induced by MCPyV DNA replication. We have not observed Kap1 recruitment to the MCPyV ori, which contains the regulatory elements of early and late gene expression. By ChIP-qPCR, we have not observed an enrichment of Kap1 at viral regulatory regions, which is in line with the unchanged gene expression of MCPyV early or late region in cells with or without Kap1 expression. The fact that our ChIP-qPCR, EMSA, and DNA-protein interaction-ELISA results invariably showed no significant differences with respect to LT binding to viral DNA furthermore suggests that Kap1 does not simply interfere with LT recruitment to the viral ori to repress replication. Although we observed coprecipitation of LT with Kap1 and, to lesser extent, also with phosphorylated Kap1 in established cell lines and primary cells, our ChIP experiments did not show an enrichment of Kap1 at the origin of replication bound by LT. Whereas LT, when bound to the viral ori, forms a dodecamer, consisting of two hexamers in head-to-head orientation, LT in complexes with Kap1 might be present as lower-molecular-weight complexes, e.g., monomeric or dimeric complexes. Our observation that Kap1 is not enriched at the viral ori or in MCPyV genomes in general is supported by the finding that Kap1 is not recruited to LT replication centers in immunofluorescence stainings performed in primary cells transduced with replication-competent MCPyV genomes or infected with MCPyV. In contrast, although we have observed reduced Kap1 binding at cellular promoter regions of the zinc finger proteins ZNF180 and ZNF274 upon expression of the MCPyV ER, our transcriptome analysis did not show significant overlapping of DEGs in cells devoid of Kap1 or in cells overexpressing LT (Fig. 8) such as would be expected if LT were to generally interfere with Kap1 recruitment to cellular promoters. More recently, phosphorylation of Kap1 serine 824 was shown to represent an important posttranslational modification in regulating Kap1 function, with phosphorylation at this site resulting in an abrogation of SETDB1 binding and subsequent derepression of Kap1 targets. Rauwel and colleagues demonstrated that phosphorylation in CD34 ϩ cells at that site represents a switch factor for the latent and lytic cycles of human-pathogenic CMV (hCMV) (31). Interestingly, that study also demonstrated that hCMV can be reactivated by activation of ATM.
Our findings indicate that MCPyV DNA replication results in the ATM-dependent phosphorylation of Kap1 S824 and in subsequent cell cycle arrest in primary nHDF cells. Interestingly, our experiments demonstrated that ATM-mediated Kap1 phosphorylation is a critical step for virus replication. Treatment of cells with an ATM inhibitor restored MCPyV replication in nHDF cells. We show that phosphorylation of Kap1 S824 is dependent on viral DNA replication, since a replication-defective viral mutant, MCPyV (K331A), which expresses LT at nearly wt levels at early time points, does not induce pKap S824. We also included a MCPyV mutant in which we replaced MCPyV LT S816A, representing a phosphorylation site in LT previously described to be important for restriction of cell proliferation and cell cycle arrest (38). This serine residue has been shown to be phosphorylated by ATM, and an ectopically expressed LT S816A mutant was reported to show a less severe form of restriction of C33A cells. In contrast, the S816A mutant behaved similarly to the wt virus in our experiments. This fact might be explained by the different cellular background and experimental system used in our study: we mutated LT in the context of the virus, and our experiments were performed in the presence of sT and other proteins of the early region.
In line with our hypothesis that DNA damage induced by MCPyV DNA replication restricts cell proliferation, we found that ectopic expression of T-Ag or ER in primary cells did not result in pKap S824 or in induction of p21 transcription. However, we observed that experiments performed with HEK293 cells, which contain parts of the adenovirus genome (46) and express adenoviral proteins E1A and E1B, resulted in pKap1S824 induction in response to MCPyV ER expression. A possible explanation for this observed difference between primary cells and HEK293 cells might be that HEK293 cells, due to their adenoviral protein expression pattern (and overlapping functions with respect to PyV early protein expression), are already preactivated with regard to DNA damage response and induction of Kap1 phosphorylation. Interestingly, our results indicate that nHDF cells harboring replicating MCPyV genomes arrest in G 2 and undergo senescence, which we demonstrated by detection of increased levels of p21 transcript and ␤-galactosidase staining and an increase of in the number of transcripts coding for cytokines and growth factors known to contribute to a senescence phenotype-associated protein (SASP) complex (40,42,43).
Previously, human dermal fibroblasts were proposed as primary target cells of MCPyV infection. However, our results indicate that primary human dermal fibroblasts transfected with religated MCPyV genome undergo senescence. In line with efficient restriction of viral replication in these cells, we did not observe an increase of the levels of MCPyV-positive cells over time and no infectious virus was observed in the supernatant of the cultures (see Fig. S1D in the supplemental material). We hypothesize that the senescence phenotype in nHDF cells represents an efficient host defense mechanism against viral replication. At present, we do not know what the functional role of the observed interaction between sT/LT and Kap1 is. We suspect that it might serve to counteract the activity mediated by this pathway, but, if so, it evidently failed to rescue viral replication in our in vitro system. While we infected or transfected commercial nHDF cells from a single donor, it seems possible that this cell type can support MCPyV infection in vivo or in the previously described ex vivo model (14), due, for example, to the presence of paracrine factors provided by other cell types.
MCPyV replication assay. In vitro replication assays were performed as previously described (15,16,35). For nHDF cells, 1 ϫ 10 6 cells were transfected with 2 g MCVSyn or pUC18 plasmid (mock) using electroporation. Cells were resuspended in 100 l buffer R and mixed with DNA (in double-distilled water [ddH 2 O]), and 100-l Golden tips were used (1 pulse at 1,700 V for 20 ms). Cells were immediately transferred into 1 ml FGM-2 medium (Promo Cell) without antibiotics.
Senescence-associated ␤-galactosidase staining. Early passages (P2 to P7) of nHDF cells were transfected with MCPyV genome by electroporation. ␤-Galactosidase staining was combined with an immunofluorescence assay (IFA), and the corresponding experiments were performed as previously described (51).
Gene expression analysis by reverse transcription-quantitative PCR (RT-qPCR). RNA was isolated by the use of RNA Bee (Amsbio) followed by DNase I digestion (Invitrogen). A 1-g volume of RNA was used for random cDNA synthesis (Superscript III; Invitrogen).
Transcriptome analysis. Library preparation was carried out using a NEBNext Ultra RNA library preparation kit (Illumina) and 1 g RNA. Libraries were sequenced on an Illumina HiSeq 2500 platform (SR50). Reads were mapped and counted on the hg38 human genome using Star v2.5 (52). DEGs were identified using DEseq2 (53). Differentially regulated genes were selected by calculation of a falsediscovery rate below 0.1 (FDR Ͻ 0.1) and log 2 fold change values greater than or equal to 1 and less than or equal to Ϫ1. Gene Ontology was performed using DAVID (54,55) and Ingenuity pathway analysis (Qiagen).
Western blotting. Cells were washed in Tris-buffered saline (TBS), cooled on dry ice, and resuspended in radioimmunoprecipitation assay (RIPA) buffer ( Coimmunoprecipitation and GST pulldown experiments. Total cell extracts and Co-IPs were generated as described previously (17).
Sequencing libraries were prepared from 2 to 10 ng DNA using a NEXTflex Illumina ChIP-seq library prep kit (Bio Scientific) and were sequenced on an Illumina NextSeq (SR50) system. Sequencing reads were aligned to the human reference genome (hg19) and pCR2.1-MCPyV-NCCR using Bowtie version 1.2.2 (58). Sites with enriched Kap1 levels were detected by the use of MACS version 2.1.2 (59), and matched negative-control region sets were generated with EaSeq (60).
Cell cycle analysis. A total of 1 ϫ 10 6 nHDF cells were transfected with 2 g MCPyV or control DNA using electroporation. At 2 and 4 days p.t., cells were fixed (70% ethanol), washed in phosphate-buffered saline (PBS), and resuspended in 200 l FACS buffer (PBS, 1% FBS, 5 mM EDTA, 25 mM HEPES, 100 g/ml RNase A) followed by propidium iodide staining and flow cytometry. FlowJo software and the Dean-Jett-Fox algorithm were used.
Statistics. Statistical analysis was performed in GraphPad Prism using t tests, including one-sample t tests, for samples following a normal distribution; the Mann-Whitney test and the Wilcoxon signed rank test were applied for nonparametric samples. For p21 transcripts, a mixed random intercept model was computed by the use of IBM SPSS software, taking the different starting point of p21 transcription into account. A two-tailed P value of Ͻ0.05 was considered significant.
Data availability. Transcriptome sequencing (RNA-Seq) data have been submitted to ENA under accession number PRJEB30502.

SUPPLEMENTAL MATERIAL
Supplemental material is available online only.