RecBCD coordinates repair of two ends at a DNA double-strand break, preventing aberrant chromosome amplification

Abstract DNA double-strand break (DSB) repair is critical for cell survival. A diverse range of organisms from bacteria to humans rely on homologous recombination for accurate DSB repair. This requires both coordinate action of the two ends of a DSB and stringent control of the resultant DNA replication to prevent unwarranted DNA amplification and aneuploidy. In Escherichia coli, RecBCD enzyme is responsible for the initial steps of homologous recombination. Previous work has revealed recD mutants to be nuclease defective but recombination proficient. Despite this proficiency, we show here that a recD null mutant is defective for the repair of a two-ended DSB and that this defect is associated with unregulated chromosome amplification and defective chromosome segregation. Our results demonstrate that RecBCD plays an important role in avoiding this amplification by coordinating the two recombining ends in a manner that prevents divergent replication forks progressing away from the DSB site.

. RecD is required for growth in the presence of an SbcCD-induced DNA doublestrand break. Growth curve showing the effect of SbcCD expression on population growth of strains containing (Pal + ) or not (Pal -) the DNA palindrome as measured by optical density of the culture at 600 nm (OD 600 ). SbcCD expression was either induced (SbcCD + ) or repressed (SbcCD -) at Time 0 min. Data are represented as mean ± range, n = 2.

l o w i n g R e c A C h r o m a t i n Immunoprecipitation (ChIP).
A) Normalised enrichment of DNA sequences across the E. coli chromosome following RecA ChIP. The results for two biological repeats are shown.
Due to sample variability, one of the samples is displayed at two different scales. B) Qualitatively similar results were obtained for paired-end read sequenced ΔrecD SbcCD + Palsamples irrespective of whether the sequenced reads were mapped as single, or paired-end reads. Counts were normalized using a median of ratios scaling factor.  In the presence of RecD, cleavage of the palindrome by SbcCD does not results in an ectopic origin of chromosomal DNA replication. DSB-dependent change in relative DNA abundance isolated from recD + cells. Data values for recD + SbcCD + Paland recD + SbcCD + Pal + are the average of two biological repeats normalized by the total number of mapped reads. Data was averaged using a 10 Kb window.

Bacterial cell culture and Strain Construction
All bacteria used in this study were derived from the non-pathogenic E. coli K12 strain BW27784 (1). For all experiments, cells were grown at 37 °C in LB growth media except those presented in Figures 3F, 3G, S3G and S3H, were cells were grown in M9 minimal media supplemented with glycerol (0.2%) and anhydrotetracycline (1 ng ml -1 ). Expression of SbcCD and I-SceI was repressed or induced by the addition of glucose or arabinose at the indicated concentration.

Strain Construction
Strains were created by plasmid mediated gene replacement (PMGR), a method for precise modification of E. coli chromosomes. Briefly, the parental strain was first transformed using the PMGR vector. Integration of the temperature sensitive plasmid into the host chromosome was selected for by growth at 42 °C in the presence of chloramphenicol (Cm). Subsequent excision of the plasmid was permitted by growth at 30 °C in the absence of Cm, prior to negative selection of the plasmid by growth at 37 °C in the presence of sucrose (5%). Colonies were first screened for Cm sensitivity to ensure plasmid loss, and then screened for the desired modification by Polymerase Chain Reaction (PCR), sequencing and phenotypic tests (when available).

Plasmid Construction
PMGR vectors are derivatives of the plasmid pTOF24 (3) containing two regions of homology to the E. coli chromosome, each approximately 400bp in size. Cloning was carried out using standard molecular biology techniques (PCR, restriction digest, DNA ligation, transformation, and Sanger sequencing). Unless noted otherwise, genomic DNA (gDNA) from E. coli strain DL1777 (4) was used as a template for the PCR reactions. All plasmids were maintained in the host E. coli strain XL1-Blue (Stratagene).
PMGR vector pDL4068 was created in a multi-step process. First the P ftsKi -lacIcerulean,tetR-eyfp cassette of pDL3196 was amplified using primers pGB2F and pGB2R and subsequently cloned into plasmid pGB2 (7) using SalI and HindIII restriction enzymes. Next, in order to increase the expression of the lacI-cerulean,tetR-eyfp cassette, the promoter P ftsKi was altered by two sequential rounds of site-directed mutagenesis using primer combinations Minus10F/Minus10R and Minus35F/Minus35R to give plasmid pDL4005. This newly created synthetic promoter was named P MW1 . The P MW1 -lacI-cerulean,tetR-eyfp was amplified from pDL4005 using primers FP_NotI_F and FP_NotI_R2 and subsequently cloned into PMGR vector pDL2802 using NotI restriction enzyme to give pDL4068.
In the course of this investigation it became apparent that recombination could occur due the CFP and YFP genes of the P MW1 -lacI-cerulean,tetR-eyfp cassette. To prevent this from occurring and improve the signal of tetO array bound TetR-YFP we designed an E. coli codon optimized Ypet gene that shared sequence homology with the cerulean gene of no more than 15 bp. This gene was synthesized by Eurofins Genomics and provided as plasmid pCR2.1-Ypet_coli. PMGR vector pDL4680 was created to introduce this Ypet cassette into the E. coli genome and was created in a multi-step process. First, the tetR gene of plasmid pDL3196 was amplified by PCR using primers TetR-mCherryA_F1 and TetR-mCherryA_R1. Next the Ypet gene of pCR2.1-Ypet_coli was amplified by PCR using primers tetRYpet_coliF2 and tetRYpet_coliR2. These two PCR products were fused together by crossover PCR using primers TetR-mCherryA_F1 and tetRYpet_coliR2 to create the tetR-Ypet gene. The tetR-eyfp sequence of pDL4068 was then replaced by tetR-Ypet by restriction digest and cloning using XbaI restriction enzyme. The resultant plasmid was named pDL4648. Finally, PMGR vector pDL4680 was created by amplifying tetR-Ypet-ykgC from plasmid pDL4648 by PCR using primers YkgC-F1 and pMW11-R and cloning the product into plasmid pTOF24 using PstI and SalI restriction enzymes.

Growth Curves
Overnight (stationary phase) cultures of strains DL2573 (RecBCD + Pal -), DL2006 (RecBCD + Pal + ), DL3743 (ΔrecD Pal -) and DL3391 (ΔrecD Pal + ), were diluted to an OD 600 of 0.05 in LB broth (Time -30 min) and grown at 37 °C with vigorous shaking for 30 min before splitting in two (Time 0) and adding glucose (0.5%) to one to repress SbcCD expression (SbcCD -) and arabinose (0.2%) to the other to induce SbcCD expression (SbcCD + ). Cultures were maintained with vigorous shaking at 37 °C and had their OD 600 monitored every 30 min for 4 h. Cultures were maintained in exponential growth phase by diluting upon reaching an OD 600 of 0.5. Plotted optical density values are measured OD 600 corrected for dilution. The results of two biological repeats of each strain are plotted in Figure S1A. Chromosome fragments of interest were detected using 32 P α-dATP incorporated radiolabeled DNA probes that were created using Agilent Prime-It II random labelling kit and a DNA template generated by PCR using either primers DIGperR-F and DIGperR-R (oriC proximal probe) or primers DIGmalZ-F and DIGmalZ-R (oriC distal probe) and gDNA isolated from strain DL1777 as a template. Probes were hybridized to membranes overnight at 65 °C in Radiolabeled membranes were then exposed to a GE healthcare storage phosphor screen. A Molecular Dynamics Storm 860 phosphorImager scanner was used to scan the phospho screen. The resultant image files were visualised using both ImageQuant and FIJI (8) software. Membranes were first probed using one probe and then stripped by incubating for 1 h in a solution composed of 50% formamide, 0.75 M NaCl, 50mM NaH 2 PO 4 and 5 mM EDTA at 65 °C, followed by a 30 min incubation in 200 ml of 0.075 M NaCl, 0.0075 M sodium citrate and 0.1% SDS for 30 min at 65 °C, before probing with the second probe. Stripped membranes were exposed and scanned to ensure removal of the original probe.

2D Gel Electrophoresis, Southern Blotting, Probing
Prior to restriction digest, gDNA (in plugs) was washed 6 times in appropriate restriction buffer for 1 h at room temperature with agitation, prior to overnight digestion at 37 °C in restriction buffer with 600 units of restriction enzyme. For analysis of lacI::terB ( Figure S3B, C) gDNA was digested with NdeI restriction enzyme. For analysis of ykgP::terB::eaeH ( Figure 3D, E, S3D, E, F) gDNA was digested with PvuII-HF restriction enzyme. Digested DNA was loaded onto a 0.4% agarose gel in 1x TBE (89 mM Tris-borate, 2 mM EDTA) at 1 V/cm for 24 h at 4 °C. Lanes containing the separated DNA were cut out, rotated 90°, and cast in a new gel composed of 1% agarose in 1x TBE supplemented with 0.3 µg ml -1 ethidium bromide. This gel (the 2 nd dimension) was ran at 6V/cm for 15 h in circulating 1x TBE buffer supplemented with 0.3 µg ml -1 ethidium bromide at 4 °C. DNA was then transferred to a positively charged nylon membrane by Southern blotting and cross-linked to the membrane using UV-light. Chromosome fragments of interest were detected using 32 P α-dATP incorporated radiolabeled DNA probes as described for PFGE. The probe for detecting the NdeI digested fragment containing (or not) lacI::terB was prepared using primers lacZp.F and lacZp.R. The probe for detecting the PvuII digested fragment containing (or not) ykgP::terB::eaeH was prepared using primers 3YkgM-F and ykgM-probeR.

Quantification of Blocked Replication Forks
The signal of probed gels was quantified using ImageQuant software. The percentage of DNA stalled at ykgP::terB::eaeH was calculated as 'spot' divided by 'spot' plus 'linear', where 'spot' was the signal within the visible spot corresponding to stalled replication forks (indicated by purple arrow in example images displayed in Figure S3F) minus the average background signal and 'linear' was the signal within the visible spot corresponding to linear DNA (indicated by green arrow in example images displayed in Figure S3F) minus the average background signal. Three biological repeats were quantified.

Genomic DNA Isolation and Illumina Sequencing
Cultures of strains DL2573 (recD + Pal + ), DL2006 (recD + Pal + ), DL3743 (ΔrecD Pal -) and DL3391 (ΔrecD Pal + ) were grown to exponential growth phase in LB broth at 37 °C prior to inducing SbcCD expression by the addition of arabinose (final concentration 0.2%). After 1 h of growth in SbcCD + conditions, gDNA was isolated from 20 ml samples of these cultures using a Wizard Genomic DNA Purification Kit following manufacturer's guidelines. Purified gDNA was treated with the supplied RNase for 50 min and rehydrated overnight in the supplied TE buffer at 4 °C. Three units of the RNase blend Riboshredder was then added to further destroy contaminating RNA. Samples were then purified again by phenol/chloroform extraction and ethanol precipitation. In tandem, gDNA was isolated using the same procedure form non-replicating stationary phase cultures of strain DL2573 to act as a control for sequence bias. Libraries were prepared from the gDNA by Edinburgh Genomics using an Illumina TruSeq DNA Sample Prep kit. Edinburgh Genomics subsequently obtained paired-end reads of the samples using an Illumina HiSeq 2000 platform. Two biological repeats of each strain were acquired.

Marker Frequency Analysis
Edinburgh Genomics supplied paired-end reads with adapter sequences removed. Reads were aligned to the DL2573 draft reference genome sequence (GSE107973) using the Burrows-Wheels Alignment software BWA-MEM and the number of reads mapped to each bp of the genome quantified using SAMtools (mpileup). Technical repeats (multiple sequencing runs of the same biological gDNA library) were combined by merging BAM files using SAMtools software (9) in order to improve coverage. Mapped reads was averaged over 1Kb or 10Kb bins (as indicated in Figure legends) using the software R.

Chromatin ImmunoPrecipitation
Cultures of strains DL4201 (RecBCD + Pal -), DL4184 (RecBCD + Pal + ), DL6204 (ΔrecD Pal -) and DL5699 (ΔrecD Pal + ) were grown to exponential growth phase in LB broth supplemented with glucose (final concentration of 0.5%) at 37 °C prior to inducing SbcCD expression by the addition of arabinose (final concentration 0.2%). After 1 h of growth in SbcCD + conditions, cells were fixed by the addition of formaldehyde (final concentration 1%) for 10 min at 22.5 °C to crosslink proteins to DNA. Crosslinking was quenched by the addition of 0.5 M glycine. Cells were then collected by centrifugation at 1,500 x g for 7 min before washing three times in icecold 1x PBS and re-suspending in 250 µl of ChIP buffer (10 ml ChIP buffer consists of 200 mM Tris-HCl (pH 8.0), 600 mM NaCl 4% Triton X and 1 cOmplete TM protease inhibitor cocktail EDTA-free tablet). Samples were then sonicated using a Diagenode Bioruptor ® at 30 seconds intervals for 10 minutes at high amplitude. After sonication, 350 µl of ChIP buffer was added to each sample and the samples gently mixed by pipetting. Immunoprecipitation was performed overnight at 4°C using 1/100 anti-RecA antibody (Abcam, ab63797). Immunoprecipitated samples were then incubated with Protein G Dynabeads® for 2 hours with rotation at room temperature. All samples were washed three times with 1 X PBS + 0.02% Tween-20 before resuspending the Protein G Dynabeads® in 200 µl of TE buffer + 1% SDS. 100 µl of TE buffer + 1% SDS were added to the input samples and all samples were then incubated at 65°C for 10 hours to reverse the formaldehyde cross-links. DNA was isolated using the MinElute PCR purification kit according to manufacturer's instructions. DNA was eluted in 100 µl of TE buffer using a 2-step elution. Samples were stored at -20°C. Two biological repeats of each strain were acquired.

Illumina ChIP-seq Library Preparation
Libraries of the immunoprecipitated DNA were made using NEBNext ® ChIP-Seq library preparation kit. Briefly, the samples were first subjected to end repair to fill in ssDNA overhangs, remove 3' phosphates and phosphorylate the 5' ends of DNA. Klenow exo-was used to adenylate the 3' ends of the DNA and NEBNext DNA adaptors (provided in the NEBNext Multiplex Oligos for Illumina kit) were ligated using T4 DNA ligase. After each step, the DNA was purified using the Qiagen MinElute PCR purification kit according to the manufacturer's instructions. After adaptor ligation, the adaptor-modified DNA fragments were enriched by PCR using primers (provided in the NEBNext Multiplex Oligos for Illumina kit) corresponding to the beginning of each adaptor. Finally, agarose gel electrophoresis was used to size select adaptor-ligated DNA with an average size of approximately 300 bp. All samples were quantified on a Bioanalyzer (Agilent) before being sequenced on either an Illumina® HiSeq 2500 (for DL4184, DL4201 and DL5699) or HiSeq 4000 (for DL6204) by Edinburgh Genomics.

ChIP-seq Data Analysis
50 bp single-end reads (for DL4184, DL4201 and DL5699) and 75 bp pair-end reads (for DL6204) were mapped to the DL4201 draft reference genome sequence (GSE107972) using the default parameters of software Bowtie 2 (10). As a control, the two sequenced ends for DL6204 samples were mapped individually as single-end reads ( Figure S2B). Read depths were calculated using SAMtools software (with parameter -d set to 10 6 ). A single, samplespecific scaling factor was applied to the number of mapped reads for each position of the genome to normalise for differences in sequencing depth. To calculate this scaling factor, we implemented the median of ratios normalisation of DESeq software (11) using R. Operationally this involved dividing the number of mapped reads for each genomic position of the sample of interest by the geometric mean of the number of mapped reads for the corresponding genomic position across all eight samples (two biological repeats of four strains). The scaling factor was the median value for all genomic positions. This normalisation is based on the hypothesis that the perturbations (DSB induction and absence of recD) alter RecA binding to chromosomal DNA at a minority (<50%) of gDNA positions along the chromosome relative to controls. This hypothesis was deemed reasonable as the effect of the DSB on gDNA enrichment following RecA ChIP was only seen across <3% of the genome and the region amplified (as detected by MFA) was <30% of the genome. Following this normalization, values were smoothed using the MATLAB loess local regression function as documented in the Figure legends. The conclusions that were drawn in this paper are neither dependent upon the specific normalization method or binning method used as the same qualitative effects are observed using the number of mapped reads prior to normalization and binning.

P sfiA -gfp Measurements
Cultures of strains DL4849 (RecBCD + Pal -), DL4848 (RecBCD + Pal + ), DL4851 (ΔrecD Pal -) and DL4850 (ΔrecD Pal + ) were grown to exponential growth phase in LB broth at 37 °C with vigorous shaking. Cultures were diluted to an OD 600 of 0.005 in either LB broth 0.5% glucose (SbcCD -) or LB broth 0.5% glucose (SbcCD + ). After 2 h of further growth, cultures were sampled and GFP fluorescence measured using an Apogee A50 flow cytometer. Data were saved as .csv files and analysed using MATLAB software. Either two or three biological repeats of each strain and condition were acquired.

Relative Cellular DNA content measurements
Cultures of strains DL2573 (RecBCD + Pal -), DL2006 (RecBCD + Pal + ), DL3743 (ΔrecD Pal -) and DL3391 (ΔrecD Pal + ) were grown to exponential growth phase in LB broth at 37 °C with vigorous shaking. Cultures were diluted to an OD 600 of 0.02 in either LB broth 0.5% glucose (SbcCD -) or LB broth 0.2% arabinose (SbcCD + ). After 1 h of further growth, cells were fixed by adding 1 ml of culture to 8 ml of 100% ethanol. Fixed cells were collected by centrifugation, washed twice in 1x PBS and re-suspended in 400 µl of 1x PBS. RNA was then degraded and DNA stained by the addition of 100 µl of a 1x PBS, 50 µg ml -1 propidium iodide, 500 µl ml -1 RNaseA solution. Propidium iodide fluorescence (a measure of DNA content) was measured using an Apogee A50 flow cytometer. Data were saved as .csv files and analysed using MATLAB software. Three biological repeats of each strain and condition were acquired.

Microscopy
Cultures of strains DL4696 (RecBCD + Pal -), DL4695 (RecBCD + Pal + ), DL4709 (ΔrecD Pal -) and DL4708 (ΔrecD Pal + ) were grown to exponential growth phase in M9 minimal growth media supplemented with 0.2% glycerol and 1 ng ml -1 anhydrotetracycline at 37 °C with vigorous shaking to an OD 600 of 0.2 after which either SbcCD was induced (SbcCD + ) or repressed (SbcCD -) by the addition of arabinose (0.2%) or glucose (0.2%) respectively. After 1 h of further growth, 10 µl of cell culture was mounted on a pad of 1% agarose.H 2 O, covered with #1.5 coverslip and imaged by widefield fluorescence microscopy at a resolution of 100nm X, 100nm Y, 200nm Z using a Zeiss Axiovert 200 fluorescence microscope equipped with a 100x Objective NA1.4 phase objective with a 1.6x Optivar, Photometrics Evolve 512 EMCCD camera, Xenon light source and piezo stage. The microscope was controlled using Metamorph software.
Acquired images were deconvolved using Autoquant X2 and visualized and processed using FIJI. Three biological repeats were acquired for each strain.