Quantitative genomic analysis of RecA protein binding during DNA double-strand break repair reveals RecBCD action in vivo

Abstract

Understanding molecular mechanisms in the context of living cells requires the development of new methods of in vivo biochemical analysis to complement established in vitro biochemistry. A critically important molecular mechanism is genetic recombination, required for the beneficial reassortment of genetic information and for DNA double-strand break repair (DSBR). Central to recombination is the RecA (Rad51) protein that assembles into a spiral filament on DNA and mediates genetic exchange. Here we have developed a method that combines chromatin immunoprecipitation with next-generation sequencing (ChIP-Seq) and mathematical modeling to quantify RecA protein binding during the active repair of a single DSB in the chromosome of Escherichia coli. We have used quantitative genomic analysis to infer the key in vivo molecular parameters governing RecA loading by the helicase/nuclease RecBCD at recombination hot-spots, known as Chi. Our genomic analysis has also revealed that DSBR at the lacZ locus causes a second RecBCD-mediated DSBR event to occur in the terminus region of the chromosome, over 1 Mb away.

Keywords: DNA repair; RecA; RecBCD; homologous recombination; mechanistic modelling.

Fig. 1.

DSBR in E. coli. (A and B) Schematic representation of DSB processing by the RecBCD complex. (A) Before Chi recognition, both the RecB and RecD motors progress along the DNA. RecD is the faster motor and as a result a loop of ssDNA (Loop 1) is formed ahead of the slower RecB motor. The 3′ ssDNA strand is scanned for the Chi sequence by the RecC protein. (B) After Chi recognition, RecBCD likely undergoes a conformational change so that only the RecB motor is engaged. The RecA protein is recruited by the RecB nuclease domain and loaded onto the ssDNA loop generated by RecB unwinding to promote RecA nucleoprotein filament formation. In this schematic representation, the Chi site is shown held in its recognition site. However, the Chi site will be released either by disassembly of the RecBCD complex or at some point before this and the second single-stranded region on the 3′ terminating strand will be converted from a loop to a tail. Therefore, this region is denoted Loop/Tail 2. The mathematical model described in SI Appendix does not depend on the ATP/magnesium dependent differential cleavage of DNA strands (7, 8), nor does it depend on the precise time that the 3′ end is released from the complex following Chi recognition. (C) The hairpin endonuclease SbcCD is used to cleave a 246-bp interrupted palindrome inserted in the lacZ gene of the E. coli chromosome. Cleavage of this DNA hairpin results in the generation of a site-specific DSB on only one of a pair of replicating sister chromosomes, thus leaving an intact sister chromosome to serve as a template for repair by homologous recombination.

Fig. 2.

RecA binding following a DSB at lacZ::pal246. ChIP in combination with qPCR revealed that RecA binding to DNA is DSB-dependent and stimulated in response to both endogenous Chi sites (A and B) and artificial Chi arrays (C and D) surrounding the DSB. ChIP-qPCR primer pairs were designed to generate amplicons that span a 40-kb region surrounding the DSB. Chi sites are represented by green circles and triple-Chi arrays by red circles on the map. All Chi sites shown are in the active orientation. The palindrome (pal246) is located at position 0 kb on the map (A) Strain DL1777 (lacZ⁺ sbcDC⁺). (B) Strain DL2859 (lacZ::pal246 sbcDC⁺). (C) Strain DL4899 (DL1777 + lacZY::χχχ mhpA::χχχ) (D) Strain DL4900 (DL2859 + lacZY::χχχ mhpA::χχχ). RecA binding at the DSB was normalized to a control site located in the hycG gene on the opposite side of the chromosome. Error bars represent the SEM, where n = 3.

Fig. 3.

High-resolution analysis of DSB-dependent RecA loading by ChIP-Seq. (A) ChIP-Seq analysis of the lacZ region in the absence of a DSB carried out in strain (DL4899) lacking the 246-bp palindrome. In A–C red circles represent triple-Chi arrays and green circles single Chi sites, all Chi sites shown are in the active orientation. The raw distributions of hits are depicted in gray. To help visualization, the data were filtered using a loess filter (bandwidth 5,700 nucleotides, span 0.02, red curve). The data shown are from a representative example of two biological replicates. (B) ChIP-Seq analysis of the lacZ region undergoing DSBR in a strain (DL4900) carrying the 246-bp palindrome located at 0 kb on the map. (C) ChIP-Seq analysis of the lacZ region of strain DL5215 demonstrates that removing the endogenous Chi sites within a 30-kb region surrounding the DSB stimulates RecA loading at more distant Chi sites during DSBR. White circles represent endogenous Chi sites that have been deleted.

Fig. S1.

Hypothetical mechanism for the conversion of a two-ended break to a one-ended break. (A) SbcCD enzyme cleaves a hairpin formed on the lagging-strand at the site of an interrupted palindrome. (B) The two ends are processed by RecBCD enzyme. (C) The origin-proximal end is processed to a Chi site and RecA protein is loaded. The origin-distal end is processed up to the replication fork avoiding recognition of an origin-distal Chi site. (D) The origin-proximal end recombines with the sister chromosome and the nick left on the origin-distal side is ligated.

Fig. S2.

High-resolution analysis shows Chi position correlated RecA loading. ChIP-Seq analysis of the lacZ region of strain DL5215 (red curve) with superimposition of loess-filtered data from strain DL4900 (dashed red curve). When Chi sites have been removed within a 30-kb region surrounding the DSB (red curve), RecA loading is stimulated at more distant Chi sites. Red circles represent triple-Chi arrays, green circles single Chi sites, and white circles represent endogenous Chi sites that have been deleted. All Chi sites shown are in the active orientation. The number of hits is depicted in gray. To help visualization, the data were filtered using a loess filter with a bandwith of 5,700 and a span of 0.02. The data shown are from a representative example of two biological replicates.

Fig. 4.

High-resolution DSB-dependent RecA loading in the presence of one to six Chi sites. (A–F) The efficiency of Chi recognition in vivo was investigated by placing Chi arrays (red circle on the map) containing either 1, 2, 3, 4, 5, or 6 Chi sites on the origin-proximal side of the DSB and using RecA binding (ChIP-Seq) as a measure of Chi recognition. Green circles represent single endogenous Chi sites that are positioned in the active orientation. The number of hits per 250-bp window normalized to the total number of reads in the analyzed region is depicted in gray. To help visualization, the data were filtered using a loess filter (bandwidth 5,700 nucleotides, span 0.057, red curve). The blue line corresponds to the prediction of our mathematical model that estimates the probability that a nucleotide in the vicinity of a DSB is coated by RecA. The data shown are from a representative example of two biological replicates. (G) The three parameters of the model (RecBCD motors’ speed ratio, its probability of Chi recognition and its processivity of RecA loading) were inferred independently on each dataset using a maximum-likelihood strategy (SI Appendix).

Fig. S3.

Comparison of RecA loading in the presence of one or six Chi sites at the array. Comparison of data (smoothed and normalized to the total number of hits in the analyzed region) for the strain containing one Chi at the array position (DL5322, blue line) and the strain containing six Chi at the same position (DL5330, pink line). The first 50 kb of the origin-proximal region are represented. When six Chi sites are present at the array, a high amount of RecA loading happens at the array and there is less loading at the subsequent Chi sites than when there is only one Chi site at the array position. Red circles represent triple-Chi arrays and green circles single Chi sites. The data were smoothed using a loess filter (bandwidth 5,700, span 1.14).

Fig. 5.

Genomic distribution of RecA binding reveals DSB-dependent RecA binding at the terminus. (A and B) Genome-wide RecA binding. DSB-independent RecA binding was observed at distinct loci across the genome, these loci predominantly correspond the ORFs of genes that encode rRNA, tRNA, and ribosomal proteins. In addition, DSB-dependent binding was observed in the terminus region. (A) Strain DL4900 (lacZ::pal246 sbcDC⁺) RecA-ChIP; (B) strain DL4899 (lacZ::pal246 sbcDC⁻) RecA-ChIP. (C and D) Zoom-in showing RecA binding in the terminus region of the chromosome. The distribution of RecA binding strongly correlates with the position of correctly oriented Chi sites in the region, indicating a DSB occurs in the terminus region that is repaired by RecBCD-mediated HR. The dif site is located at 1,588,774 bp and terC is located at 1,607,181 bp. No other ter sites are located within this region. (C) Strain DL4900 (lacZ::pal246 sbcDC⁺). (D) Strain DL4899 (lacZ::pal246 sbcDC⁻). Green circles represent single endogenous Chi sites that are positioned in the active orientation. The raw distributions of hits are depicted in blue and orange.