The parvovirus minute virus of mice modulates the DNA damage response to facilitate viral replication and a pre-mitotic cell cycle block

Abstract

The DNA damage response (DDR) is a critical cellular network that affords cells the ability to repair DNA damage they have incurred from endogenous and exogenous sources. Recently, it has become appreciated that viruses, both DNA and RNA, can induce the DDR and have evolved the ability to interact with this ancient antiviral mechanism. Viruses can choose to inactivate or utilize this host response, which typically requires specific modulation for either case. Importantly, it has been shown that MVM utilizes this response to facilitate its replication in an ATM-dependent manner. MVM induces a DDR-dependent pre-mitotic, G2/M cell cycle block via activation of the checkpoint kinase, Chk2, and depletion of the RNA and protein of the key mitotic cyclin, cyclin B1. Unexpectedly, this cell cycle block was shown to be p21- and Chk1-independent. MVM infection results in the recruitment and activation of numerous DDR proteins including Chk2, RPA32 and p53. Upon activation, via phosphorylation, p53, a critical tumor suppressor, is known to transactivate several hundred genes including the well-characterized CDK inhibitor, p21. However, previous work demonstrated a sustained, proteasome-dependent loss of p21 during infection, which was required for efficient replication. This depletion of p21 during infection was unexpected, given the activation of p53 during infection, and because p21 is a known, potent cell cycle inhibitor. We investigated the loss of p21 during infection and found that siRNA knockdown of specific components of the CRL4Cdt2 E3 ubiquitin ligaseCul4A, DDB1 and Cdt2stabilized p21 during MVM infection. Importantly, siRNA knockdown of specific components of CRL4Cdt2 reduced viral replication. DDB1 and Cdt2, the adapter protein and substrate recognition factor of the CRL4Cdt2, respectively, were recruited to viral replication factories, termed autonomous parvovirus-associated replication (APAR) bodies, suggesting that MVM may be hijacking this important E3 ubiquitin ligase. The recruitment and utilization of this ligase is likely specific, as the APC/CCDC20 E3 ubiquitin ligase, which also targets p21 for proteasome-dependent degradation, was not recruited to APAR bodies nor did siRNA knockdown of specific components of this ligase stabilize p21 during infection. Taken together, these results suggest that MVM specifically utilizes the CRL4Cdt2 E3 ubiquitin ligase to target p21 for degradation and that the activity of this E3 ubiquitin ligase is required for efficient MVM replication. The requirements for the activity of CRL4Cdt2 gave us a hint as to why MVM would target p21 for degradation. It has been shown that CRL4Cdt2 activity requires interaction with PCNA, a DNA polymerase d cofactor, and chromatin. Importantly, p21 is an inhibitor of PCNA activity. As PCNA and DNA polymerase d are known to be required for parvoviral rolling hairpin replication, we hypothesized that MVM must target p21 for depletion to allow for PCNA activity, which is required for viral replication. p21 mutants that were unable to interact with either CRL4Cdt2 or PCNA were stable during MVM infection, suggesting that p21 needed to interact with both the E3 ubiquitin ligase and PCNA for degradation during MVM infection. We next constructed p21 mutants that were resistant to ubiquitination, by mutating the seven lysine residues in p21 to arginine, and maintained or lost their ability to interact with PCNA. Importantly, a stable p21 mutant that interacted with PCNA resulted in the significant depletion of MVM replication whereas a stable p21 mutant that no longer interacted with PCNA had no effect on replication. Taken together, this data suggests that MVM co-opts a cellular E3 ubiquitin ligase to target the CDK inhibitor p21 for degradation, which is required to allow the PCNA activity that MVM needs for efficient replication. To induce a pre-mitotic G2/M cell cycle block, MVM depletes the key mitotic cyclin, cyclin B1, which is preceded by the loss of its encoding RNA. We next sought to determine how MVM programs the depletion of cyclin B1 RNA. Initial studies indicated a loss of cyclin B1 RNA between 18 and 24 hours post-infection in a NS2-independent manner. This loss of RNA and protein was seen in both human and murine cells lines, suggesting that programming the depletion of cyclin B1 is a critical hallmark of MVM infection. Interestingly, the viral mechanism which targets cyclin B1 could overcome the effects of an exogenous DNA damaging agent known to induce cyclin B1 levels. The stability of cyclin B1 RNA during MVM infection is comparable to doxorubicin-treated cells, while the production of nascent cyclin B1 RNA is substantially depleted. Importantly, the chromatin landscape of the cyclin B1 promoter during MVM infection was consistent with an open conformation, yet significantly lower levels of RNA polymerase II (RNA pol II) were found to occupy the cyclin B1 gene. The NF-Y transcription factor and B-myb, a component of MuvB-B-Myb (MMB) complex, were both found to bind the cyclin B1 promoter during infection. However, the key G2/M transcription factor, FoxM1, was found to occupy the cyclin B1 promoter at significantly lower levels in MVM infected cells compared to mock- or doxorubicin-treated cells. FoxM1, which requires hyperphosphorylation to activate its transcriptional activity, was found to have lower levels of phosphorylation during MVM infection, compared to mock- or doxorubicin-treated cells. Reconstitution of FoxM1 to the cyclin B1 promoter, via catalytically inactive Cas9 fused to the FoxM1 transactivation domain, upregulated cyclin B1 RNA and protein during MVM infection. Taken together, these results suggest that MVM prevents the activation and binding of the critical transcription factor FoxM1 to the cyclin B1 promoter, thereby reducing RNA pol II transcriptional activity and the production of nascent cyclin B1 RNA and its encoded protein, ultimately facilitating establishment of a pre-mitotic G2/M block. An important RNA-binding protein, HuR, known to regulate two cyclins important for MVM infection, cyclin A and cyclin B1, was also investigated. We observed the predominantly nuclear HuR was heavily relocalized to the cytoplasm during MVM infection. Ectopic expression of NS1, but not NS2, resulted in the cytoplasmic relocalization of HuR in a Crm1-independent manner. Importantly, siRNA knockdown of HuR during MVM infection resulted in a further reduction of cyclin B1 protein, but not cyclin B1 RNA. This data suggested that HuR may promote the translation of specific RNAs during MVM infection. HuR was also shown to bind MVM RNA. While more work needs to be undertaken to fully understand the ramifications of this interaction, HuR is likely to be an important regulator of MVM RNA stability or translation. Taken together, these observations suggest that MVM expertly modulates the DDR and other components of the cellular machinery to ensure an environment conducive to facilitation of viral replication. To establish this environment, MVM targets several core cellular mechanisms including proteasome-mediated protein degradation, RNA transcription, RNA translation and subcellular localization of cellular proteins.