Factors regulating late stages of homologous recombination and sub-pathway choice in G2

Abstract

Cells have evolved multiple mechanisms to preserve genome integrity and restore structural and functional properties of the genome following deoxyribonucleic acid (DNA) damage. DNA double-strand breaks (DSBs) are critical lesions whose timely and faithful repair is important for cellular viability and genomic stability. Among the multiple pathways dedicated for handling DSBs, homologous recombination (HR) provides a high-fidelity mechanism for error-free repair in cycling cells. HR is also important for the faithful duplication of the genome by providing means of tolerating replication stress and overcoming a variety of lesions occurring as a result of replication errors such as single-strand breaks, gaps and one-ended DSBs that impede progressing replication forks. HR at two-ended DSBs can proceed through two known sub-pathways with distinct genetic outcomes: the double Holliday junction (dHJ) pathway and synthesis-dependent strand annealing (SDSA), which result in crossover (CO) and non-CO recombination products, respectively. Although the initial steps of HR, such as resection, have been extensively studied, less is known about late stages of HR, especially in mammalian cell, as well as the factors governing sub-pathway choice. Here, experiments were set up to specifically analyse HR-mediated repair of X-ray-induced DSBs in G2 cells, a setup that avoids the complications of S-phase induced damage and focuses on two-ended breaks. The first part of this study shows that the histone variant H3.3 is required during late stages of HR, in a step following Rad51 loading and removal, and in a manner epistatic to the chromatin remodeler ATRX. Indeed, H3.3 is needed for DNA repair synthesis in G2 cells, and subsequent formation of sister chromatid exchanges (SCEs). To monitor in vivo histone deposition, a stable cell line system expressing SNAP-H3.3 was used to visualize newly synthesized histone incorporation following laser-induced DNA damage. Consistent with previous reports, H3.3 is deposited early (up to 1 h) post IR, but through an HR-independent mechanism. However, late H3.3 incorporation (8 h) was dependent on Rad51, ATRX and the chaperon DAXX, corroborating an active deposition of H3.3 at sites of DNA damage during HR. Furthermore, H3.3 deposition was also dependent on the processivity factor PCNA and its loader RFC-1, suggesting a tight association with DNA synthesis. The results collectively inspire a model where ATRX-DAXX-mediated H3.3 deposition is tightly coupled to DNA repair synthesis and serves to facilitate progression of the displacement loop (D-loop) and repair completion in a sub-pathway of HR that leads to the formation of CO products. Page | 2 To further understand the regulation of HR sub-pathway choice, cells lacking ATRX and the SDSA factor RECQ5 were analysed for their HR repair capacity. ATRX-deficient U2OS cells with inducible ATRX expression and HeLa cells were used to establish comparisons of pathways usage. While both cell lines are dependent on Rad51 for repair of G2-induced DSBs, ATRX-deficient cells rely entirely on RECQ5-mediated SDSA for the repair of these breaks, while HeLa cells can employ both pathways. Further analysis revealed that ATRX-dependent HR dominates in cells that have both factors, resulting in elevated IR-induced SCEs in cells with induced ATRX expression. Additional factors were analysed to show that SDSA surprisingly does not require the essential HR factor Rad54 for repair, indicating the deeper inherent differences between these two sub-pathways and suggesting a more complex regulation of pathway choice. Quantitative analysis of HR events showed that approximately 50% of ATRX-dependent HR events result in a CO product, a much higher frequency than previously thought. This led to further analysis of HR intermediates formed in this pathway manifesting as IR-induced ultra-fine bridges (UFBs) visualized in anaphase cells lacking the resolvases Mus81 and Gen. IR-induced UFBs formed in an ATRX-dependent manner, suggesting that these HR intermediates are largely processed by the resolution pathway. Furthermore, Mus81 foci analysis showed that Mus81 was recruited to DSBs in mitotic HeLa cells but not in U2OS cells, suggesting the presence of distinct HR intermediates in these cells. Recruitment of Mus81 was not affected by BLM depletion, suggesting that Mus81 is occupying all suitable substrates even in the presence of BLM and that these structures are not subject to dissolution. Similarly, BLM depletion did not affect Rad51 foci formation, γH2AX foci dynamics, or the level of IR-induced SCEs, further corroborating the notion that BLM is not active in this sub-pathway of HR. Taken together, the data suggest that ATRX-dependent HR dominates over RECQ5-dependent SDSA for the repair of two-ended breaks in G2 cells. This pathway involves the formation of HR intermediates that are exclusively resolved by the structure-specific nucleases Mus81 and Gen1 while being refractory to dissolution, explaining the observed high frequency of SCEs observed. This dominance could be explained by a model whereby ATRX-dependent histone deposition inside an expanding D-loop structurally hinders both strands displacement during SDSA and the dissolution of a dHJ by BLM.