As a deubiquitinating enzyme (DUB), the physiological substrates of ataxin-3 (ATX-3) remain elusive, which limits our understanding of its normal cellular function and that of pathogenic mechanism of spinocerebellar ataxia type 3 (SCA3). Here, we identify p53 to be a novel substrate of ATX-3. ATX-3 binds to native and polyubiquitinated p53 and deubiquitinates and stabilizes p53 by repressing its degradation through the ubiquitin (Ub)-proteasome pathway. ATX-3 deletion destabilizes p53, resulting in deficiency of p53 activity and functions, whereas ectopic expression of ATX-3 induces selective transcription/expression of p53 target genes and promotes p53-dependent apoptosis in both mammalian cells and the central nervous system of zebrafish. Furthermore, the polyglutamine (polyQ)-expanded ATX-3 retains enhanced interaction and deubiquitination catalytic activity to p53 and causes more severe p53-dependent neurodegeneration in zebrafish brains and in the substantia nigra pars compacta (SNpc) or striatum of a transgenic SCA3 mouse model. Our findings identify a novel molecular link between ATX-3 and p53-mediated cell death and provide an explanation for the direct involvement of p53 in SCA3 disease pathogenesis.

REV1 is a eukaryotic member of the Y-family of DNA polymerases involved in translesion DNA synthesis and genome mutagenesis. Recently, REV1 is also found to function in homologous recombination. However, it remains unclear how REV1 is recruited to the sites where homologous recombination is processed. Here, we report that loss of mammalian REV1 results in a specific defect in replication-associated gene conversion. We found that REV1 is targeted to laser-induced DNA damage stripes in a manner dependent on its ubiquitin-binding motifs, on RAD18, and on monoubiquitinated FANCD2 (FANCD2-mUb) that associates with REV1. Expression of a FANCD2-Ub chimeric protein in RAD18-depleted cells enhances REV1 assembly at laser-damaged sites, suggesting that FANCD2-mUb functions downstream of RAD18 to recruit REV1 to DNA breaks. Consistent with this suggestion we found that REV1 and FANCD2 are epistatic with respect to sensitivity to the double-strand break-inducer camptothecin. REV1 enrichment at DNA damage stripes also partially depends on BRCA1 and BRCA2, components of the FANCD2/BRCA supercomplex. Intriguingly, analogous to FANCD2-mUb and BRCA1/BRCA2, REV1 plays an unexpected role in protecting nascent replication tracts from degradation by stabilizing RAD51 filaments. Collectively these data suggest that REV1 plays multiple roles at stalled replication forks in response to replication stress.

DNA polymerase η (Polη) facilitates translesion DNA synthesis (TLS) across ultraviolet (UV) irradiation- and cisplatin-induced DNA lesions implicated in skin carcinogenesis and chemoresistant phenotype formation, respectively. However, whether post-translational modifications of Polη are involved in these processes remains largely unknown. Here, we reported that human Polη undergoes O-GlcNAcylation at threonine 457 by O-GlcNAc transferase upon DNA damage. Abrogation of this modification results in a reduced level of CRL4CDT2-dependent Polη polyubiquitination at lysine 462, a delayed p97-dependent removal of Polη from replication forks, and significantly enhanced UV-induced mutagenesis even though Polη focus formation and its efficacy to bypass across cyclobutane pyrimidine dimers after UV irradiation are not affected. Furthermore, the O-GlcNAc-deficient T457A mutation impairs TLS to bypass across cisplatin-induced lesions, causing increased cellular sensitivity to cisplatin. Our findings demonstrate a novel role of Polη O-GlcNAcylation in TLS regulation and genome stability maintenance and establish a new rationale to improve chemotherapeutic treatment.