Mechanism of RAD51-dependent DNA interstrand cross-link repair

Abstract

DNA interstrand cross-links (ICLs) are toxic DNA lesions whose repair in S phase of eukaryotic cells is incompletely understood. In Xenopus egg extracts, ICL repair is initiated when two replication forks converge on the lesion. Dual incisions then create a DNA double-strand break (DSB) in one sister chromatid, whereas lesion bypass restores the other sister. We report that the broken sister chromatid is repaired via RAD51-dependent strand invasion into the regenerated sister. Recombination acts downstream of FANCI-FANCD2, yet RAD51 binds ICL-stalled replication forks independently of FANCI-FANCD2 and before DSB formation. Our results elucidate the functional link between the Fanconi anemia pathway and the recombination machinery during ICL repair. In addition, they demonstrate the complete repair of a DSB via homologous recombination in vitro.

Figure 1

The X-arc contains intermediates of ICL repair. (A) pICL schematic. (B) Model of ICL repair in Xenopus egg extracts (2, 10). (C) pControl or pICL was replicated in egg extract, digested with HincII, and analyzed by 2DGE. Arrowheads, see main text. (D) Cartoon of 2DGE patterns and relevant DNA intermediates. (E) ICL repair of samples from (C) was analyzed under normal and branch migration (+BM) conditions. “Background,” SapI fragments from contaminating uncross-linked plasmid. For primary data, see Figure S1E and F. All graphed experiments were performed at least three times, and a representative example is shown.

Figure 2

ICL repair requires RAD51-dependent HR. (A) Cross-linked sperm chromatin was replicated in extract containing buffer (Mock) or the indicated BRC peptide, and chromatin-associated proteins were analyzed by Western blotting. (B) BRC peptides immobilized on glutathione sepharose beads were incubated with extract (NPE), pulled down, and the supernatant and pellet blotted for RAD51. (C) pICL was replicated in extract containing buffer (Mock) or the indicated BRC peptide. Samples were digested with HincII and analyzed by 2DGE. Purple arrowhead, X-arc position. (D) Samples from (C) were analyzed for ICL repair as in Figure 1E.

Figure 3

RAD51-dependent HR functions at a late step in ICL repair. (A) Schematic depiction of leading strand intermediates from the rightward fork. (B) pICL was replicated in extract containing buffer (Mock) or the indicated BRC peptide. Samples were digested with AflIII and separated on a sequencing gel alongside a ladder generated with primer S [shown in (A)]. Nascent strands generated by the rightward fork are indicated at right. The following products from (B) were quantified and graphed: (C) −20 to −40, (D) −1, (E) insertion, and (F) extension.

Figure 4

Interplay between the HR and FA pathways. (A) Schematic of ChIP primer pairs. (B) pICL was replicated and analyzed by RPA and RAD51 ChIP. Controls containing BRC^WT peptide, pControl, or lacking DNA replication are shown in Figure S6A. (C) Samples from (B) were also analyzed for the timing of fork convergence, −1 product accumulation, and repair (raw data in Figure S6B–D). The data were graphed as % of peak value and compared with RPA and RAD51 ChIP at site I [from (B)]. (D) pICL was replicated in extract for 40 minutes and immunoprecipitated with RAD51 antibodies (see Methods). Recovered DNA was digested with HincII and analyzed by agarose gel electrophoresis. Repair intermediates are depicted at left for pICL (blue) and an internal control plasmid, pQuant (gray). (E) pICL was replicated in mock-depleted egg extract (Mock), FANCD2-depleted extract (FANCD2Δ), or FANCD2Δ extract supplemented with 375 nM FANCI-FANCD2 (FANCD2Δ+ID). Samples were digested with HincII and analyzed by 2DGE. Arrowheads, see text. See Figure S7E for complete 2D gel time courses. (F) Samples from (E) were analyzed by ChIP using primer pair I. Primer pairs II and III are shown in Figure S7A.