Replication independent DNA double-strand break retention may prevent genomic instability.

BACKGROUND: Global hypomethylation and genomic instability are cardinal features of cancers. Recently, we established a method for the detection of DNA methylation levels at sites close to endogenous DNA double strand breaks (EDSBs), and found that those sites have a higher level of methylation than the rest of the genome. Interestingly, the most significant differences between EDSBs and genomes were observed when cells were cultured in the absence of serum. DNA methylation levels on each genomic location are different. Therefore, there are more replication-independent EDSBs (RIND-EDSBs) located in methylated genomic regions. Moreover, methylated and unmethylated RIND-EDSBs are differentially processed. Euchromatins respond rapidly to DSBs induced by irradiation with the phosphorylation of H2AX, gamma-H2AX, and these initiate the DSB repair process. During G0, most DSBs are repaired by non-homologous end-joining repair (NHEJ), mediated by at least two distinct pathways; the Ku-mediated and the ataxia telangiectasia-mutated (ATM)-mediated. The ATM-mediated pathway is more precise. Here we explored how cells process methylated RIND-EDSBs and if RIND-EDSBs play a role in global hypomethylation-induced genomic instability. RESULTS: We observed a significant number of methylated RIND-EDSBs that are retained within deacetylated chromatin and free from an immediate cellular response to DSBs, the gamma-H2AX. When cells were treated with tricostatin A (TSA) and the histones became hyperacetylated, the amount of gamma-H2AX-bound DNA increased and the retained RIND-EDSBs were rapidly repaired. When NHEJ was simultaneously inhibited in TSA-treated cells, more EDSBs were detected. Without TSA, a sporadic increase in unmethylated RIND-EDSBs could be observed when Ku-mediated NHEJ was inhibited. Finally, a remarkable increase in RIND-EDSB methylation levels was observed when cells were depleted of ATM, but not of Ku86 and RAD51. CONCLUSIONS: Methylated RIND-EDSBs are retained in non-acetylated heterochromatin because there is a prolonged time lag between RIND-EDSB production and repair. The rapid cellular responses to DSBs may be blocked by compact heterochromatin structure which then allows these breaks to be repaired by a more precise ATM-dependent pathway. In contrast, Ku-mediated NHEJ can repair euchromatin-associated EDSBs. Consequently, spontaneous mutations in hypomethylated genome are produced at faster rates because unmethylated EDSBs are unable to avoid the more error-prone NHEJ mechanisms.

LINE-1 methylation status of endogenous DNA double-strand breaks.

DNA methylation and the repair of DNA double-strand breaks (DSBs) are important processes for maintaining genomic integrity. Although DSBs can be produced by numerous agents, they also occur spontaneously as endogenous DSBs (EDSBs). In this study, we evaluated the methylation status of EDSBs to determine if there is a connection between DNA methylation and EDSBs. We utilized interspersed repetitive sequence polymerase chain reaction (PCR), ligation-mediated PCR and combined bisulfite restriction analysis to examine the extent of EDSBs and methylation at long interspersed nuclear element-1 (LINE-1) sequences nearby EDSBs. We tested normal white blood cells and several cell lines derived from epithelial cancers and leukemias. Significant levels of EDSBs were detectable in all cell types. EDSBs were also found in both replicating and non-replicating cells. We found that EDSBs contain higher levels of methylation than the cellular genome. This hypermethylation is replication independent and the methylation was present in the genome at the location prior to the DNA DSB. The differences in methylation levels between EDSBs and the rest of the genome suggests that EDSBs are differentially processed, by production, end-modification, or repair, depending on the DNA methylation status.

Characterization of the Cullin7 E3 ubiquitin ligase--heterodimerization of cullin substrate receptors as a novel mechanism to regulate cullin E3 ligase activity.

Cul1 and Cul7 are cullin E3 ubiquitin ligase scaffold proteins. Cul1 is known to form a complex with the RING domain protein Rbx1 and one of approximately 70 different F-box proteins. F-box proteins function as substrate receptor subunits and recruit numerous substrates for poly-ubiquitination. Similarly to Cul1, Cul7 interacts with Rbx1, however, only one F-box protein, Fbxw8, has been shown to bind to Cul7. To date only few Cul7 E3 ubiquitin ligase substrates, including cyclin D1, IRS-1 and GRASP65, have been reported, and using Fbxw8 affinity purification, we were unable to identify additional substrate proteins. Here we provide evidence for a model in which Cul7-Rbx1 can promote the ubiquitination of Cul1 substrates by forming high order complexes with Cul1-Rbx1. Binding of Cul1-Rbx1 to Cul7-Rbx1 is mediated via heterodimerization of Fbxw8 with other F-box proteins which function to recruit substrates into the E3 ligase complex. The formation of this high order complex is likely to increase polyubiquitination efficiency.

Regulation of cullin RING E3 ubiquitin ligases by CAND1 in vivo.

Cullin RING ligases are multi-subunit complexes consisting of a cullin protein which forms a scaffold onto which the RING protein Rbx1/2 and substrate receptor subunits assemble. CAND1, which binds to cullins that are not conjugated with Nedd8 and not associated with substrate receptors, has been shown to function as a positive regulator of Cullin ligases in vivo. Two models have been proposed to explain this requirement: (i) CAND1 sequesters cullin proteins and thus prevents autoubiquitination of substrate receptors, and (ii) CAND1 is required to promote the exchange of bound substrate receptors. Using mammalian cells, we show that CAND1 is predominantly cytoplasmically localized and that cullins are the major CAND1 interacting proteins. However, only small amounts of CAND1 bind to Cul1 in cells, despite low basal levels of Cul1 neddylation and approximately equal cytoplasmic endogenous protein concentrations of CAND1 and Cul1. Compared to F-box protein substrate receptors, binding of CAND1 to Cul1 in vivo is weak. Furthermore, preventing binding of F-box substrate receptors to Cul1 does not increase CAND1 binding. In conclusion, our study suggests that CAND1 does not function by sequestering cullins in vivo to prevent substrate receptor autoubiquitination and is likely to regulate cullin RING ligase activity via alternative mechanisms.