Kaposi’s Sarcoma-Associated Herpesvirus Lytic Replication Is Independent of Anaphase-Promoting Complex Activity

DNA viruses have evolved complex strategies to gain control over the cell cycle. Several of them target APC/C, a key cellular machinery that controls the timely progression of the cell cycle, by either blocking or enhancing its activity. Here, we investigated the activity of APC/C during the lytic replication cycle of KSHV and found that, in contrast to that of KSHV's close relatives EBV and HCMV, KSHV lytic replication occurs while the APC/C is active. Perturbing APC/C activity by depleting a core protein or the adaptor proteins of the catalytic domain, and hence interfering with normal cell-cycle progression, did not affect virus replication. This suggests that KSHV has evolved to replicate independently of the activity of APC/C and in various cell cycle conditions.

lymphatic endothelial environment is uniquely permissive to spontaneous KSHV reactivation (33,34). To enrich the population of infected cells undergoing lytic replication, we used a KSHV-BAC16 mutant engineered to undergo lytic replication upon infection.
The first lytic KSHV mutants were generated by M. Budt et al. (35), utilizing the KSHV-BAC36 backbone, and A. Gallo et al. (36), utilizing the KSHV-BAC16 backbone, by inserting a constitutively active promoter upstream of ORF50. We constructed a similar KSHV mutant utilizing the KSHV-BAC16, hereafter referred to as KSHV-Lyt in order to preserve the nomenclature suggested in the earlier studies (35,36) (Fig. 1A). However, instead of reconstituting the virus using RPE-1 cells as done previously (35,36), we reconstituted the KSHV-Lyt in LECs. Two weeks after the KSHV-Lyt DNA transfection, we detected the appearance of the first plaques, which continued to expand until all cells were infected (Fig. 1B). The virus released in the supernatant was used to infect LECs, human blood cells (BECs), and umbilical vein endothelial cells (HuVECs). At 72 h postinfection (p.i.), we quantified the KSHV-Lyt-infected cells using enhanced green fluorescent protein (EGFP), expressed by KSHV-Lyt, as a marker of infection. Strikingly, HuVECs were the most resistant, whereas LECs were the most susceptible to infection and BEC infection rates were three times lower than LECs (Fig. 1C). These results confirmed that LECs are a suitable, primary host cell model to study KSHV-Lyt replication.
To assess the expression of KSHV lytic proteins at the single-cell level, we infected LECs with four infectious units (IU)/cell. Using this infection dose, at 48 h p.i., the morphology of all cells changed from cobblestone to spindle cells (Fig. 1D), and at 96 h p.i., practically all cells rounded and detached from the bottom of the plate, indicating a cytopathic effect. We analyzed the expression of ORF50, K8, ORF45, ORF57, and K8.1 by immunofluorescence in the KSHV-Lyt-infected LECs at 48 h p.i., together with LANA, which is constitutively expressed during both viral latency and the lytic cycle. While LANA was expressed in almost all infected cells, the KSHV lytic proteins ORF50, K8, ORF57, ORF45, and K8. 1 were expressed in about one third of the analyzed cells (Fig.  1E). This is in line with observations made in other experimental models such as BCBL1-TREX-RTA and iSLK.219 cells, where KSHV lytic replication (triggered by ORF50, expressed from a doxycycline-inducible promoter) occurs asynchronously and only in a subset of cells (32,37,38).
Overall, these findings show that LECs provide a highly permissive cellular environment for the KSHV-Lyt replication and are a suitable model to study the effects of KSHV lytic replication in primary cells.
KSHV lytic replication occurs in G 1 while APC/C is active. To test whether KSHV interferes with APC/C, we infected LECs with KSHV-Lyt using four IU/cell. Latent and lytic viral protein expression was confirmed by immunoblotting for LANA, ORF50, ORF57, ORF45, and K8.1 ( Fig. 2A). Additionally, we noted that infection of LECs by KSHV-Lyt led to a marked increase in the levels of phosphorylated CHK2 at threonine 68 (pCHK2-T68) and p53 at serine 15 (pp53-S15) as early as 24 h p.i., marking the induction of the DNA damage response (DDR). The lytic cells, identified by high ORF50 expression, also showed higher levels of phosphorylated H2AX at S139 (pH2AX-S139), indicating the DDR was stronger in the lytic cells ( Fig. 2A and B), as has been previously shown (39)(40)(41).
Next, we tested whether the APC/C was active during KSHV-Lyt infection by monitoring the levels of the APC/C adaptor proteins (CDH1 and CDC20), the APC/C inhibitor (EMI1), and two APC/C substrates (CCNB1 and GMNN), as compared to their levels in uninfected cells. Throughout the 72-h time course, we observed that both CDH1 and CDC20 were expressed in KSHV-Lyt-infected LECs, suggesting that both proteins are available as APC/C adaptors. The levels of CCNB1 did not increase markedly beyond the range that was observed in uninfected cells (Ͻ2-fold), while the GMNN levels fluctuated in a similar fashion as EMI1 ( Fig. 2A), indicating that APC/C is functional during the KSHV lytic cycle. At the mRNA level, EMI1 and CCNB1 levels were downregulated at 24 h and later upregulated during the KSHV infection. In contrast, GMNN mRNA levels were either similar to uninfected or increased in the late stages of infection by 2-fold (Fig. 2C). Therefore, GMNN protein represents a suitable sensor of APC/C activity during the KSHV lytic cycle. In particular, accumulation of high GMNN levels in the nucleus indicates that the APC/C is inactive while lower levels of GMNN indicate that the APC/C is active.
Further analysis of GMNN expression at the single-cell level showed that GMNN was expressed only in a small fraction of infected cells (2%). Among the lytic, ORF57-positive cells, only 1% of cells showed GMNN accumulation, indicating an inactive APC/C complex in these cells (Fig. 2D). Next, we analyzed the cell cycle profiles after staining the cellular DNA with propidium iodide (PI) and found that infected LECs accumulated mostly in G 1 , similarly to uninfected cells (Fig. 2E) and where APC/C-CDH1 is in an active state. The accumulation of cells in G 1 was also observed when only the lytic cells (ORF57-positive) were analyzed after treatment with phosphonoacetic acid (PAA) to specifically block the KSHV genome replication and thus the accumulation of viral DNA in the nucleus (Fig. 2F). Altogether, the analysis of the APC/C substrate levels showed that APC/C activity in the KSHV-Lyt LECs (both in the total population and the ORF57-positive subset) is similar to the uninfected counterparts and the cell cycle profiles indicate that both of these KSHV-Lyt LEC populations accumulate in G 1 where APC/C is active.
Subsequently, to validate our findings in another infection model, we repeated experiments using the cancer-derived iSLK.219 cell line stably infected with rKSHV.219, which constitutively expresses GFP under the control of the cellular EF1a promoter and RFP from the PAN promoter, an ORF50-responsive viral lytic promoter (42). In these cells the virus is in a latent state and lytic reactivation is induced through doxycycline (Dox)-inducible ORF50 expression (32). To induce the lytic cycle, we used, besides Dox, sodium butyrate (NaB), a histone deacetylase inhibitor commonly used to enhance KSHV reactivation (32). Here, we found that during KSHV reactivation, the total levels of APC/C substrates (GMNN and CCNB1) did not accumulate through the 32-h time course of the experiment but rather oscillated similarly to EMI1 levels (Fig. 2G). The expression patterns of ORF50, ORF45, and ORF57 in the lytic iSLK.219 cells were similar to KSHV-Lyt-infected LECs. We further investigated the oscillation of APC/C activity in lytic cells at the single-cell level using GMNN as a sensor. One day after inducing the lytic cycle, we found that the percentage of high-GMNN cells in the lytic (RFP-positive) subset of cells did not differ significantly from the whole-cell population (Fig. 2H). This shows that the GMNN levels oscillate in the same way in both the RFP-positive (lytic) and RFP-negative (latent) cells, and, consequently, suggests that APC/C functions are unaffected by the lytic replication cycle in iSLK.219 cells.
Taken together, these results show that during the lytic cycle in both the KSHV-Lytinfected LECs and iSLK.219 cells, the APC/C remains under the control of its inhibitor EMI1 and retains the ability to degrade target substrates, such as GMNN.
KSHV lytic replication cycle is independent from the APC/C activity. Sequence analysis of the KSHV proteins (JSC1 strain; GenBank sequence GQ994935.1) using GPS-ARM software (43) revealed that several KSHV proteins (some of them essential for the completion of the lytic cycle) contain D-box motifs that can be recognized by the APC/C as a substrate for ubiquitination. D-box motifs were identified with high confidence in both viral kinases ORF21 (amino acids [aa] 51 to 54) and ORF36 (aa 9 to 12); in the early KSHV protein ORF45 (aa 289 to 291); and several components of the KSHV virion such as ORF19 (aa 284 to 287), ORF25 (aa 1242 to 1245 and aa 1335 to 1338), ORF26 (aa 234 to 237), and ORF65 (aa 71 to 74), therefore representing putative APC/C targets. Following this observation, we asked whether interfering with APC/C function to affect the ability to degrade targets would affect the KSHV lytic replication cycle. To this end, we assessed the levels of KSHV lytic proteins and infectious virus titers when APC/C-CDH1 was constitutively active (by EMI1 depletion) or inactive (by CDH1 depletion).
Depletion of CDH1 and EMI1, as well as the expression levels of KSHV lytic proteins, was confirmed by immunoblot in the KSHV-Lyt infected LECs (Fig. 3A). CDH1 depletion led to only slightly higher titers of KSHV-Lyt, while EMI1 depletion had no effect on the KSHV-Lyt titers in LECs (Fig. 3B). Next, we tested the effects of EMI1 and CDH1 depletion in the lytic iSLK.219 cells and found that the viral titers were comparable to the cells treated with the control small interfering RNA (siRNA) (Fig. 3C and D). This suggests that altering the APC/C-CDH1 activity does not affect the KSHV lytic cycle in either LECs or iSLK.219 cells.
Previous studies have shown that in actively dividing cancer cell lines, simultaneous depletion of CDH1 and CDC20 adaptor proteins, together with EMI1, i.e., three subunits of the APC/C, leads to a completely dysfunctional APC/C and a prolonged S-phase (19,22,44). We therefore measured the KSHV virus production from the reactivated iSLK.219 cells where CDH1, CDC20, and EMI1 had been depleted. Efficient silencing was verified by immunoblotting ( Fig. 3E). Notably, upon EMI1 depletion, CDC20 levels also decreased dramatically, in line with CDC20 being both an adaptor protein and a substrate of the APC/C-CDH1. As an indication of efficient APC/C inactivation, we also monitored the levels of GMNN by immunoblotting and at the single-cell level. Immunoblot analysis of GMNN showed that in the absence of APC/C-CDH1, APC/C-CDC20 can degrade GMNN, although more inefficiently than the APC/C-CDH1 complex (Fig. 3E). Therefore, depletion of both CDH1 and CDC20 was required to completely inactivate the APC/C function. In cells where EMI1, CDC20, and CDH1 were depleted, the percentage of cells with high nuclear GMNN signal increased significantly compared to the siNeg-treated cells, whereas the siEMI1-treated cells, here used as a control, showed almost no cells with high nuclear GMNN levels (Fig. 3F). Next, upon EMI1, CDH1, and CDC20 depletion in iSLK.219 cells, we observed a 3-fold increase in the number of 5-ethynyl-2'-deoxyuridine (EdU)-positive cells (i.e., cells going/gone through S-phase) (Fig. 3G), suggesting a profound alteration of cell cycle regulation, as already reported for other cancer-derived cell lines (19,44). However, when the KSHV titers in the reactivated iSLK.219 cells depleted of EMI1, CDH1, and CDC20 were measured, no significant differences were found compared to the control (siNeg-treated) cells (Fig. 3H). This suggests that the KSHV lytic cycle can occur efficiently in a cellular environment where the APC/C is inactive and encompassing a prolonged S-phase.
Similarly, depletion of APC3, an essential core protein of APC/C, increased the APC/C substrate levels (AURKA, AURKB, CCNB1, CCNA2, GMNN, securin, and PLK1) in both the KSHV-Lyt-infected LECs and iSLK.219 cells (Fig. 4A and B). Yet this did not influence the KSHV lytic cycle and virus titers ( Fig. 4C and D). To obtain a better view of the effects of APC3 depletion in the KSHV-Lyt-infected LECs, we coinfected the siAPC3-and siNeg-treated KSHV-Lyt-infected LECs with a lentiviral vector expressing GMNN from the hCMV promoter, a constitutive strong promoter to enrich the population of GMNN-expressing cells. As expected, upon APC3 depletion, GMNN expression in the lytic cells (ORF57-positive) increased by more than 2-fold (Fig. 4E). Of note, we observed a cytoplasmic localization of GMNN in agreement with the study of M. Dimaki et al. (45), showing that when expressed by a constitutive promoter, GMNN is excluded from the nucleus and resides in the cytoplasm of G 1 cells. Together these results show that disruption of the APC/C by depletion of one of its core proteins (APC3) does not affect the KSHV lytic replication cycle in either iSLK.219 cells or KSHV-Lyt LECs.
Unscheduled APC/C activity-induced rereplication stress affects the integrity of cellular chromosomes, but not KSHV episomes. In actively dividing cancer cells, in contrast to primary cells, one consequence of the unscheduled APC/C activation (due to EMI1 depletion) is the induction of rereplication stress (22,23). Accordingly, when we compared the cell cycle profiles of primary LECs and the cancer-derived iSLK.219 cells after EMI1 depletion, we found rereplication occurring only in iSLK.219 cells (about 45% of the cells were ϾG 2 at 48 h after EMI1 depletion), while most of the LECs remained in G 1 with less than 1% of EMI1-depleted LECs undergoing rereplication (Fig. 5A). As a consequence of the rereplication-induced genomic stress in iSLK.219 cells, we found that the number of EdU-positive cells was reduced by half at 24 h and to less than 5% at 48 h after EMI1 depletion (Fig. 5B). Furthermore, the signals of 53BP1 and pCHK1-S317 in the EMI1-depleted iSLK.219 cells appeared punctuated (in foci) (Fig. 5C), and the levels of pp53-S15 increased as early as 24 h after EMI1 depletion (Fig. 5D), thereby confirming an ongoing DDR due to rereplication-induced genomic stress.
Interestingly, despite these deep alterations (i.e., the dramatic decrease in the EdU-positive cells and DDR induction), EMI1-depleted iSLK.219 cells were still able to sustain the full KSHV lytic replication cycle and released infectious virus to the same extent as the control (siNeg) treated cells (see Fig. 3C). Therefore, we hypothesized that rereplication stress in iSLK.219 cells would affect the cellular chromosomes but not the viral episomes. To test this, the genomic DNA of latent iSLK.219 cells undergoing rereplication (after EMI1 depletion) was subjected to next-generation sequencing and compared to the control treated cells (siNeg; Fig. 6A). To identify the rereplicating regions, the read densities across the KSHV genome and host chromosomes were analyzed. The genomic DNA isolated from serum-starved iSLK.219 cells to block proliferation was included to normalize the densities. In Fig. 6A, we show the densities across chromosome 1, as it is the largest chromosome and therefore expected to have the highest number of DNA replication origins. Remarkably, we found that after EMI1 depletion, multiple regions in this chromosome were rereplicated at both 24 h and 48 h post silencing (indicated by arrowheads in Fig. 6A). Additionally, we found that the number of KSHV episomes per cell doubled at 24 h post EMI1 depletion (Fig. 6A, center  panel). Yet, the number of KSHV episomes per copy of the human genome was comparable to the control (siNeg) cells (at a ratio of approximately one), indicating that the KSHV episomes replicated in full synchrony with the host genome. However, there were no rereplicated regions at the latent origin of replication (K1 and TR regions, marked in red in Fig. 6B) or elsewhere across the entire KSHV genome, which explains why KSHV titers were not affected by the rereplication stress.
Altogether, these findings indicate that rereplication stress affects the integrity of the cellular genome but not the viral episomes, thus allowing the efficient production of infectious virus from the EMI1-depleted iSLK.219 cells.

DISCUSSION
Several DNA viruses replicate efficiently in specific cell cycle phases. To enable this, viruses have evolved to target key cellular machineries such as the APC/C to gain control over cell cycle progression (46). In this study, we found that APC/C activity is not affected by the KSHV lytic replication cycle and the virus can replicate in the G 1 phase where APC/C-CDH1 is active. This is in contrast to the other herpesviruses, HCMV and EBV, which have been found to block APC/C activity during their lytic cycle (25,26,(29)(30)(31). Moreover, our data show that a functionally intact APC/C complex is dispensable for efficient KSHV reactivation. It is particularly interesting that blocking the APC/C function and thus dysregulating the cell cycle did not affect the KSHV lytic cycle. So, unlike other herpesviruses, which require specific cell cycle phases for efficient lytic replication, KSHV has adapted to reactivate and replicate efficiently in various cell cycle phases.
Besides G 1 , APC/C is also activated in response to DDR through the p53 signaling pathway, which leads to EMI1 downregulation (47). This, in turn, allows APC/C-CDH1 to trigger the degradation of important cellular regulators that promote S-phase transition and thereby facilitate the cell cycle block (47). KSHV lytic replication following virus reactivation and during the lytic burst that normally occurs upon primary infection leads to DDR (39)(40)(41)48). In LECs de novo infected with KSHV-Lyt, the mRNA and protein levels of EMI1 and CCNB1 were initially downregulated but increased at later time points (48 and 72 h p.i) compared to uninfected cells. This could suggest that an intrinsic mechanism of KSHV during primary infection is to manipulate EMI1 levels in order to preserve the ability of the infected cells to proceed to S-phase while the virus establishes latency. Further supporting this hypothesis is the observation that GMNN mRNA is also upregulated during the KSHV lytic cycle (this study and reference 49), which is important for the progression of S-phase.
A recent study has shown that the KSHV-encoded K10 protein interacts with several subunits of the APC/C (50). Although we showed that KSHV lytic replication cycle does not interfere with the APC/C function, the utility of this interaction remains an open question. K10 is a late lytic protein (48 to 72 h) and localizes in the nucleus together with APC/C (38,51). Interestingly, K10 has also been shown to interact with the CSL/CBF1 complex by occupying the binding site of Notch intracellular domain (NICD) and impairing NICD-mediated gene regulation (52). Possibly, binding to APC/C subunits may serve as a docking site for K10 to sequester the CSL/CBF1 complex away from chromatin.
K. J. Neelsen et al. (23) have shown that during rereplication the newly replicated DNA at the replication forks contains single-strand DNA gaps. Therefore, they predicted that rereplication of regions with DNA gaps leads to the accumulation of short double-stranded DNA fragments particularly encompassing the refiring replication origins. In our analysis of cellular chromosomes, the rereplicated regions were detected and these were more prominent at 24 h after EMI1 depletion when the cells were still able to undergo DNA replication. At 48 h post EMI1 depletion, when the cells stopped their DNA replication, the read densities across the cellular chromosomes normalized, suggesting that the short double-stranded DNA forms, predicted by K. J. Neelsen et al. (23), are short-lived.
A recent mapping of active and dormant origins of replication in HeLa cells indicated that active origins of replication localize in regions of open chromatin enriched with H3K9ac, H3K4me3, and H4K20me1 histone marks (53). In our analysis, we observed that the rereplicated regions were localized on specific sites of the chromosomes and not evenly distributed (Fig. 6A). Therefore, it is likely that the chromatin regions undergoing rereplication in iSLK.219 cells are more accessible (i.e., open) and contain higher densities of active replication origins.
KSHV episomes have been shown to contain at least two latent replication origins, the TR region (OriP) and the AT-rich region close to the K5 gene (OriA) (11,14,54,55). Under normal conditions, KSHV replication can start at both origins and proceeds bi-directionally; however, there is a slight preference for OriP (56). Interestingly, the KSHV episomes were not affected by rereplication stress and replicated in full synchrony with the host DNA. It would be interesting in future studies to address how the KSHV origins of replication, one LANA-dependent (OriP) and the other LANAindependent (OriA), allow successful replication of KSHV episomes even during rereplication stress.
Our findings were confirmed in two different experimental systems, the iSLK.219 cells and LECs de novo infected with KSHV-Lyt, the latter providing a novel and useful model to study KSHV lytic replication in a physiologically relevant cell type. Overall, our results show that the KSHV lytic replication program occurs efficiently even when the cell cycle is strongly perturbed and is independent of the activity of APC/C.
KSHV-Lyt BAC generation and reconstitution. The design of the lytic KSHV BAC16 is based on the principle described for KSHV-Lyt BAC36 by M. Budt et al. (35). In this construct, the PGK promoter is inserted upstream of ORF50 (RTA) to enforce constitutive RTA expression. The lytic BAC16 was constructed as follows. In the first step, the shuttle vector containing the sequence of ORF49_KanR_pPGK was constructed using three fragments that were generated by PCR with overlapping primers as follows: (i) ORF49 was amplified from BAC16 (57); (ii) the KanR cassette with an I-SceI site was amplified from the pMCMV3 mRFP1 vector (kind gift of Martin Messerle); and (iii) the mPGK promoter was amplified from the pHRSIN vector. The fragments were cloned into a PstI/SalI-treated pMCMV3 mRFP1 vector by Gibson assembly. In the second step, the 2.6-kb fragment was amplified from the shuttle vector with primers containing arms of homology to BAC16, KSHV-Lyt forward primer 5=-ctccgcaaggggtagtctgttgtgagaatac tgtccaggcagccacaaaaatgacatcgagaaggcccct-3= and KSHV-Lyt reverse primer 5=-ccgagaggccgacgaagctttc cacacaggaccgccgaagcttcttacccttgtcatcttgcgccatggt-3=. The resulting fragment was gel-purified and electroporated into GS1783 cells harboring BAC16 followed by a two-step traceless recombination procedure as described in B. K. Tischer et al. (58). The modifications were confirmed by restriction analysis and sequencing.
KSHV-Lyt virus was reconstituted by transfecting purified KSHV-Lyt BAC DNA in LECs using FugeneHD (Promega) at a DNA (g) to FugeneHD (l) ratio of 1:3.5 following the manufacturers recommendations. KSHV-Lyt spread to uninfected LECs was monitored by using EGFP, expressed by KSHV-Lyt, as a marker of infection. The supernatants containing KSHV-Lyt were stored at -80°C.
Where indicated, band intensities were quantified using FIJI (59) software and shown in the figure as actin-normalized relative band intensities.
Cell cycle and proliferation analysis. Cells were fixed with 70% ethyl alcohol (EtOH) and stained with antibodies against ORF57 (sc-135746, SCBT) and anti-mouse IgG (HϩL) Alexa Fluor Plus 647 (A-32728, Thermo Scientific), where indicated. DNA was then stained with FxCycle PI/RNaseA staining solution (Thermo Scientific) for 1 h. The cells were analyzed using a BD Accuri C6 flow cytometer and data were processed using the CFlow Sampler software (BD).
For cell proliferation analysis, cells were maintained for 2 h in medium containing 10 M 5-ethynyl-2'-deoxyuridine (EdU). Subsequently, the cells were fixed in 4% paraformaldehyde in PBS and the EdU incorporated in the cell DNA was coupled to Alexa Fluor 647 according to the manufacturer's instructions provided in the Click-iT EdU Alexa Fluor 647 imaging kit (Thermo Fisher Scientific). Images were acquired using a CellInsight High Content microscope (Thermo Fisher Scientific) and analyzed using Cell Profiler 3.0 software to quantify EdU-positive cells.