As Caenorhabditis elegans hermaphrodites age, sperm become depleted, ovulation arrests, and oocytes accumulate in the gonad arm. Large ribonucleoprotein (RNP) foci form in these arrested oocytes that contain RNA-binding proteins and translationally masked maternal mRNAs. Within 65 min of mating, the RNP foci dissociate and fertilization proceeds. The majority of arrested oocytes with foci result in viable embryos upon fertilization, suggesting that foci are not deleterious to oocyte function. We have determined that foci formation is not strictly a function of aging, and the somatic, ceh-18, branch of the major sperm protein pathway regulates the formation and dissociation of oocyte foci. Our hypothesis for the function of oocyte RNP foci is similar to the RNA-related functions of processing bodies (P bodies) and stress granules; here, we show three orthologs of P body proteins, DCP-2, CAR-1 and CGH-1, and two markers of stress granules, poly (A) binding protein (PABP) and TIA-1, appear to be present in the oocyte RNP foci. Our results are the first in vivo demonstration linking components of P bodies and stress granules in the germ line of a metazoan. Furthermore, our data demonstrate that formation of oocyte RNP foci is inducible in non-arrested oocytes by heat shock, osmotic stress, or anoxia, similar to the induction of stress granules in mammalian cells and P bodies in yeast. These data suggest commonalities between oocytes undergoing delayed fertilization and cells that are stressed environmentally, as to how they modulate mRNAs and regulate translation.

The central recombination enzyme RAD51 has been implicated in replication fork processing and restart in response to replication stress. Here, we use a separation-of-function allele of RAD51 that retains DNA binding, but not D-loop activity, to reveal mechanistic aspects of RAD51’s roles in the response to replication stress. Here, we find that cells lacking RAD51’s enzymatic activity protect replication forks from MRE11-dependent degradation, as expected from previous studies. Unexpectedly, we find that RAD51’s strand exchange activity is not required to convert stalled forks to a form that can be degraded by DNA2. Such conversion was shown previously to require replication fork regression, supporting a model in which fork regression depends on a non-enzymatic function of RAD51. We also show RAD51 promotes replication restart by both strand exchange-dependent and strand exchange-independent mechanisms.

The RAD54 family DNA translocases have several biochemical activities. One activity, demonstrated previously for the budding yeast translocases, is ATPase-dependent disruption of RAD51-dsDNA binding. This activity is thought to promote dissociation of RAD51 from heteroduplex DNA following strand exchange during homologous recombination. In addition, previous experiments in budding yeast have shown that the same activity of Rad54 removes Rad51 from undamaged sites on chromosomes; mutants lacking Rad54 accumulate nonrepair-associated complexes that can block growth and lead to chromosome loss. Here, we show that human RAD54 also promotes the dissociation of RAD51 from dsDNA and not ssDNA. We also show that translocase depletion in tumor cell lines leads to the accumulation of RAD51 on chromosomes, forming complexes that are not associated with markers of DNA damage. We further show that combined depletion of RAD54L and RAD54B and/or artificial induction of RAD51 overexpression blocks replication and promotes chromosome segregation defects. These results support a model in which RAD54L and RAD54B counteract genome-destabilizing effects of direct binding of RAD51 to dsDNA in human tumor cells. Thus, in addition to having genome-stabilizing DNA repair activity, human RAD51 has genome-destabilizing activity when expressed at high levels, as is the case in many human tumors.