Bacillus subtilis PcrA Couples DNA Replication, Transcription, Recombination and Segregation

Bacillus subtilis PcrA abrogates replication-transcription conflicts in vivo and disrupts RecA nucleoprotein filaments in vitro. Inactivation of pcrA is lethal. We show that PcrA depletion lethality is suppressed by recJ (involved in end resection), recA (the recombinase), or mfd (transcription-coupled repair) inactivation, but not by inactivating end resection (addAB or recQ), positive and negative RecA modulators (rarA or recX and recU), or genes involved in the reactivation of a stalled RNA polymerase (recD2, helD, hepA, and ywqA). We also report that B. subtilis mutations previously designated as recL16 actually map to the recO locus, and confirm that PcrA depletion lethality is suppressed by recO inactivation. The pcrA gene is epistatic to recA or mfd, but it is not epistatic to addAB, recJ, recQ, recO16, rarA, recX, recU, recD2, helD, hepA, or ywqA in response to DNA damage. PcrA depletion led to the accumulation of unsegregated chromosomes, and this defect is increased by recQ, rarA, or recU inactivation. We propose that PcrA, which is crucial to maintain cell viability, is involved in different DNA transactions.


INTRODUCTION
Homologous recombination is the major pathway to circumvent a replicative stress, a replication fork collapse and for the elimination of DNA double-strand breaks (DSBs) induced by endogenous or exogenous stress. Super-family 1 (SF1) DNA helicases, which are conserved motor proteins that couple nucleoside triphosphate hydrolysis to the unwinding of duplex DNA, play crucial roles in repair-by-recombination and in coping with replication-transcription conflicts (RTCs) (Wu and Hickson, 2006;Singleton et al., 2007). The prototype of bacterial SF1 helicases that translocate with 3 ′ → 5 ′ direction is UvrD (Singleton et al., 2007;Dillingham, 2011). This enzyme shares a significant degree of structural similarity with Rep, which is restricted to the γ-Proteobacteria Class, PcrA and yeast Srs2 DNA helicases (Wu and Hickson, 2006;Marini and Krejci, 2010). Rep and PcrA play essential roles in the replication of extrachromosomal elements, whereas UvrD and PcrA participate in the resolution of RTCs by poorly understood mechanisms (Boubakri et al., 2010;Bruning et al., 2014;Epshtein et al., 2014;Merrikh et al., 2015). In vitro studies reveal that PcrA and UvrD interact with the RNA polymerase (RNAP), Rep interacts with the replicative DNA helicase (DnaB in Proteobacteria) and Srs2 physically interacts with Rad51 and with the PCNA sliding clamp (ortholog of bacterial DnaN) among other proteins (Antony et al., 2009;Guy et al., 2009;Kaniecki et al., 2017;Sanders et al., 2017). Absence of Escherichia coli Rep and UvrD renders cells inviable when grown in rich medium (Taucher-Scholtz et al., 1983), but lack of Bacillus subtilis PcrA renders cells inviable even when grown in minimal medium (Petit et al., 1998;Merrikh et al., 2015). To gain insight into the crucial steps carried on by PcrA, a comparative analysis with UvrD and Rep was undertaken. From the comparative analysis of E. coli (best-characterized representative of the Proteobacteria Phylum) and B. subtilis (best-characterized from the Firmicutes Phylum), which are evolutionarily separated by more than 2,000 million years, a genetic divergence larger than that between human and paramecium, we expect to understand the role of the SF1 UvrD-like DNA helicases.
When the DNA of a single inert mature haploid nonreplicating chromosome of a B. subtilis spore is damaged, the RTCs of the single spore genome are compounded during the rapid outgrowth. In the absence of end resection (AddAB and/or RecJ-RecQ [RecS]) cells remain recombination proficient and apparently are as capable of repairing damaged template bases as the wild type (wt) control (Vlasic et al., 2014). These spores, which lack an intact homologous template, require RecA mediators (RecO, RecR), RecA itself, and positive (RecF) and negative (RecX, RecU) RecA modulators, with PcrA facilitating DNA replication through transcription units (Vlasic et al., 2014;Merrikh et al., 2015;Raguse et al., 2017). It is likely that PcrA is implicated in the processing of damaged replication forks and/or recombination intermediates formed at damaged forks and in circumventing DNA lesions in the absence of an intact homologous template, without generating a fork breakage that should be lethal for the revival of a haploid spore (Raguse et al., 2017). Then, a nucleoprotein RecA filament aids to overcome RTCs via different DNA damage tolerance pathways and to reactivate replication by recruiting the damage checkpoint DisA and pre-primosome DnaD proteins (Million-Weaver et al., 2015;Torres et al., 2019). This is consistent with the observation that the lethality of B. subtilis pcrA cells is suppressed by recF17, recL16, recO, or recR mutations (Petit and Ehrlich, 2002).
In E. coli cells, the synthetic lethality of uvrD rep is partially suppressed by recJ, recQ, recO, recR, or recF inactivation in minimal medium, but it is only marginally suppressed by recA inactivation (Lestini and Michel, 2008;Guy et al., 2009). Inactivation of recJ or recQ provides very limited suppression of rep uvrD rich medium lethality, but recA inactivation does not (Petit and Ehrlich, 2002;Veaute et al., 2005;Lestini and Michel, 2008;Guy et al., 2009). in vivo assays reveal that PcrA expression in E. coli can substitute several functions of UvrD, but antagonizes the function of Rep, providing a heterologous dominant negative phenotype (Petit et al., 1998). These data suggest that in the absence of UvrD and Rep, toxic RecA nucleoprotein filaments and/or RecA-mediated recombination intermediates or DNA structures can accumulate. This hypothesis is supported by the following observations: (i) in budding yeast, suppressors of Srs2 mutations map in the Rad51 gene (Aboussekhra et al., 1992); and (ii) RecA-or Rad51mediated DNA strand exchange is actively prevented by the PcrA, UvrD, or Srs2 DNA helicase in vitro, considered as a paradigmatic anti-recombinase activity (Krejci et al., 2003;Veaute et al., 2003Veaute et al., , 2005Anand et al., 2007;Park et al., 2010;Fagerburg et al., 2012;Petrova et al., 2015;Kaniecki et al., 2017).
Other studies have demonstrated that the rich-medium synthetic lethality of uvrD rep cells is caused primarily by RTCs, with partial reduction of transcription or translation rates across heavily transcribed genes in the opposite orientation relative to the replication forks, to compensate the impact of transcription on DNA replication (Guy et al., 2009;Baharoglu et al., 2010;Kamarthapu et al., 2016;Myka et al., 2017). Indeed, rich-medium synthetic lethality of uvrD rep cells is fully suppressed by reducing transcription, as mutations in different RNAP subunits (rpo * [rpoB and rpoC point mutants]), or reducing translation elongation, such as by mutations in a tRNA gene (AspRS, aspT), in an aminoacyl tRNA synthetase, in a translation factor needed for efficient formation of prolineproline bonds (EF-P), and spoT1 mutation [encoding a (p)ppGpp pyrophosphorylase-defective SpoT] (Guy et al., 2009;Baharoglu et al., 2010;Kamarthapu et al., 2016;Myka et al., 2017). Transcription and translation are coupled with the leading ribosome pushing RNAP forward; however, when these two processes become uncoupled, RNAP rpo * mutants are prone to pausing with (p)ppGpp promoting UvrD-mediated RNAP backtracking (Kamarthapu et al., 2016;Myka et al., 2017). These observations altogether open the question about the primary cause of PcrA lethality. (Unless stated otherwise, indicated genes and products are of B. subtilis origin. The nomenclature used to denote the origin of proteins from other bacteria is based on the bacterial genus and species abbreviation [e.g., E. coli UvrD is referred to as UvrD Eco ]).
The physiological causes of PcrA/UvrD-Rep Eco lethality is/are poorly understood. To understand the role of PcrA in rich medium exponentially growing B. subtilis cells, we have studied the genetic linkage of PcrA depletion  with mutations in genes acting at the presynaptic ( recJ, recQ, addAB, recL16, rarA, recX, recU) and synaptic ( recA) stages, as well as in genes that contribute to bypass RTCs or facilitate RNAP backtracking or removal ( helD, recD2 [absent in E. coli], hepA [also termed yqhHor rapA], ywqA or mfd). We show that PcrA depletion reduced cell viability by >4,000-fold, and survival in the presence of limiting H 2 O 2 or methyl methanesulfonate (MMS) concentrations, suggesting that PcrA is involved in repair-by-recombination. The recL16 mutations were mapped in the recO gene and the mutation was termed recO16. The PcrA depletion lethality is suppressed by recO16, recJ, recA, or mfd inactivation, but not by recQ, rarA, recX, recU, addAB, helD, hepA, or ywqA inactivation. The pcrA gene is not epistatic to recO16, recJ, rarA, recQ, recX, recU, recD2, addAB, helD, hepA, or ywqA in response to MMS-induced DNA damage, but it is epistatic to recA or mfd. Absence of PcrA promotes a net accumulation of unsegregated nucleoids, and this defect is increased in the absence of RarA, RecQ, and RecU. We conclude that PcrA contributes to untangle branched intermediates and works at the interface of DNA replication, transcription, recombination and segregation.

Bacterial Strains and Plasmids
All B. subtilis strains used derived from BG214, and are listed in Table 1. The gene to be characterized was deleted by gene replacement with the six-cat-six (SCS) cassette flanked by appropriate homologous regions up-and downstream. The SCS cassette, composed of two directly oriented β-recombinase cognate sites (six sites) and the cat gene, confers chloramphenicol resistance (Cm R ) (Rojo and Alonso, 1995). Natural competent cells were transformed with the SCS cassette flanked by homologous regions to the gene to be deleted with selection for Cm R . Integration of the SCS cassette, through double crossover recombination, replaced the gene under characterization. The β recombinase promoter, which maps within the six site, can read the downstream gene to overcome any potential polar effect (Rojo et al., 1994). A plasmid-borne β-recombinase gene was moved into the background, and followed by β sitespecific recombinase-mediated excision between the two directly oriented six sites, leading to the deletion of the cat gene and one six site (Rojo et al., 1994). The final outcome of this strategy is that the gene to be characterized is replaced by a single six site (Sanchez et al., 2005). Accuracy of deletions was confirmed by complementation with a plasmid-borne gene (Sanchez et al., 2005). The IPTG-inducible sspB cassette encoding the SspB adaptor protein (a gift from Houra and Christopher Merrikh, Vanderbilt University, USA) was ectopically integrated into the amy locus by natural transformation (Alonso et al., 1988;Torres et al., 2017). The null hepA or ywqA mutations (a gift from Marie-Agnès Petit, Université Paris-Saclay, France) were moved into the BG214 background (Table 1) by SPP1-mediated transduction (Valero-Rello et al., 2017).
To segregate the recO mutations and to reconstruct a new BG107 strain (BG107-1), limiting concentrations of chromosomal DNA of the original BG107 (recL16) strain and 0.1 µg/ml of pHP14 DNA were added to competent BG214 cells, with selection for the plasmid-borne Cm R marker. From those transformants it was expected that by co-transformation of any unlinked markers (congression) 0.1 to 1% of the Cm R cells should receive the recL16 MMS sensitive (MMS S ) phenotype (Alonso et al., 1988). Selection of MMS S was performed by streaking colonies on agar plate containing or lacking MMS (Alonso et al., 1988). Five selected MMS S clones were sequenced by high-throughput sequence analyzer (Illumina) technology using standard sequencing libraries and filtered sequence data (Beijing Genomics Institute [BGI]), of ∼1 gigabases per sample, followed by whole-genome comparison with the nucleotide sequencing of the MMS R parental BG214 strain. The K37A, T65I, or Q254A pcrA variants were generated by means of mutation site directed mutagenesis (QuickChange Kit, Stratagene) using the pQE-1-borne wt pcrA plasmid as a template. Unexpectedly, all the Q254A mutants analyzed also contained the unselected E224V mutation. K37A, T65I, or Q254A-E224V pcrA genes were amplified by PCR and were cloned into XmaI-BamHI-cleaved pHP14. The resultant recombinant plasmids were used to transform competent B. subtilis BG214 cells.

Survival Assays and Colony Size
Plating exponentially growing pcrA-ssrA sspB (pcrA T ) cells in rich medium onto agar plates containing isopropyl-β-D thiogalactopyranoside (IPTG) induced SspB expression from a regulated promoter, which then bound the SsrA peptide tag and rapidly delivered the tagged PcrA-SsrA protein to the B. subtilis ClpXP protease for degradation (PcrA degron [pcrA T ] strain) (Keiler et al., 1996;Griffith and Grossman, 2008;Merrikh et al., 2015). PcrA degron cultures were grown to OD 560 = 0.4. The cultures were divided and aliquots plated in LB agar plates alone or with 500 µM IPTG (Calbiochem). The percentage of colony forming units (CFUs) in LB agar plates containing IPTG was measured. The mean and SEM were calculated using Prism 6 software (GraphPad), and a Student's t-test, with P < 0.01, was performed to denote the threshold of significance.
Cell sensitivity to chronic MMS (Sigma Aldrich) or H 2 O 2 (Sigma Aldrich) exposure was determined by growing cultures to OD 560 = 0.4 and plating appropriate dilutions on rich LB agar plates containing IPTG (500 µM) and MMS (1.3 mM) or H 2 O 2 (0.2 mM) as described (Sánchez et al., 2007). Cells grew in rich LB medium with a doubling time of 28-35 min. Plates were incubated overnight (16-18 h, 37 • C) and the number of CFUs determined. Experiments were conducted independently at least four times. Fractional survival data are shown as mean ± SEM. Statistical analysis was performed with a two-tailed Student's ttest. For experiments involving more than two groups, one-way analysis of variance (ANOVA) was performed. For all tests, a P < 0.1 was considered significant (Supplementary Material). All statistical analyses were performed using Prism 6 software.
B. subtilis cells form round smooth colonies that raised above the agar. Colony size on Petri dishes was calculated via the diameter of a hypothetical circular colony. After overnight incubation, Petri dishes pictures were acquired and analyzed with the aid of a BioRad ChemiDoc TM imaging system equipped with the QuantityOne software (BioRad). The relative mean colony diameter of ∼50 isolated blind scoring colonies from pcrA T vs. pcrA T cells bearing a second mutation were measured from the pictures using ImageJ software (NIH). Upon colony magnification, the relative mean colony size was calculated using the formula for the area of a circle. Average and standard deviation were calculated using Prism 6 software. The colony area was compared by analysis of variance or Student's t-test, with P < 0.01 as the threshold of significance.

Nucleotide Sequence Analysis
The samples of genomic DNA from B. subtilis wt (BG214, Reference strain) and the Test recL16 (BG107) strains were analyzed using the first step of comparative genome sequencing, a service provided by NimbleGen Systems, Inc. as described earlier (Albert et al., 2005). Briefly, the genome of B. subtilis wt was tiled on custom-designed "mapping" microarrays with 29 base oligonucleotide pieces (probes) and a 7 or 8 base spacing from the start of one probe to the start of the next. The genomic DNA from the Reference BG214 strain was labeled with Cy5 and of the Test BG107 strain with Cy3. The ratios of hybridization intensities (reference/test) were calculated and plotted against genome position. If the test sample contains no mutation, the reference/test ratio equals 1. This analysis gives a high-resolution map of possible mutation sites, in which each mutation is localized in a window of 29 bases (the length of the reporting probe). A custom algorithm, based on hybridization intensity ratios significantly above background that have good agreement between results from corresponding probes from both strands, allows the identification of likely sites of mutation. The data were analyzed graphically using the SignalMap software provided by NimbleGen. The regions around the probes with a reference/test ratio significantly >1 were tested for mutations by direct DNA sequencing. Sequencing was performed with Big Dye (from Operon Technologies Inc.) using the protocols of the University of Wisconsin Biotechnology Center, after PCR amplification of the target genomic DNA sequence.
The samples of genomic DNA from B. subtilis wt (BG214, Reference strain) and five MMS S clones from the newly constructed. Test strain were re-sequenced by high-throughput sequence analyzer (Illumina) technology using standard sequencing libraries and filtered sequence data, a service provided by BGI, of ∼1 gigabase per sample. The sequencing data were used to conduct paired-end nucleotide sequencing with the rec + BG214 reference strain and the five MMS S clones from the newly constructed BG107-1 sample as described (Quail et al., 2008).

Fluorescence Microscopy and Data Analysis
For chromosome segregation analyses (Figure 2), cells were fixed and stained as described . To obtain exponentially growing cells, overnight cultures were inoculated in LB rich medium. The recO16, recA, rarA, recJ, recQ, recX, or recU mutants in the pcrA-ssrA sspB (pcrA T ) context ( Table 1) were grown unperturbed in LB medium to OD 560 = 0.2 (37 • C). IPTG (500 µM) was added to half of the culture, and both cultures were incubated (60 min, 37 • C). Then, cells were collected, subjected to fixation with 2% formaldehyde, and finally stained with 4 ′ ,6 ′ -diamino-2-phenylindole (DAPI) (1 µg/ml). Samples were visualized and photographed by fluorescence microscopy with a Hamamatsu 3CCD Digital Camera C7780 coupled to a BX61 Olympus fluorescence microscope, equipped with an 100x immersion oil lens and a DAPI filter (U-MNU2).
The ImageJ software (NIH) was used to merge the phase contrast and DAPI-fluorescence images, which allowed us to distinguish the septum, and thus determine the filamentation event and was also used to determine the cell length. Blind scoring was performed on captured images as described .

Experimental Rationale
The phenotypes associated with the absence of PcrA are exceedingly complex and reflect an involvement in several aspects of DNA metabolism, including DNA replication, transcription, RTCs, repair-by-recombination and chromosomal segregation. The PcrA depletion lethality was attributed to the accumulation of toxic RecA-mediated intermediates or SOS toxicity. In fact, inactivation of positive RecA mediators (recO, recR), modulator (recF) (Table S1) or an unknown function (recL) suppresses the pcrA lethality of cells grown in minimal medium (Petit and Ehrlich, 2002). In addition, a mutation in recO, recR, recF, or recL reduces and delays the SOS induction (Gassel and Alonso, 1989). However, the lethality in the pcrA recA1 (a leaky recA mutation) challenged such hypothesis (Petit and Ehrlich, 2002). This observation was revisited and extended using rec-deficient strains ( Table 1).
Finally, the UvrD Eco or PcrA helicase/translocase plays a crucial role in nucleotide excision repair (Mendonca et al., 1993;Petit and Ehrlich, 2002;Epshtein, 2015), but only the former contributes to DNA mismatch repair (Lenhart et al., 2016). In this report, we show that PcrA also contributes to the repair of non-bulky lesions of oxidative nature generated upon exposure to MMS or H 2 O 2 (Figure 1). Upon PcrA depletion, cells were exposed to limiting MMS or H 2 O 2 concentrations, whose lesions are specifically removed by base excision repair (Sedgwick, 2004;Fu et al., 2012). If damaged template bases escape specialized repair, because they are in ssDNA regions, the offending lesion halts DNA polymerase (DNAP) or RNAP progression, and the lesion-containing gap is circumvented via damage avoidance pathways (template switching, fork reversal, translesion synthesis). When the damage is bypassed and the lesion is present in duplex DNA, it is removed by specialized pathways (Sedgwick, 2004;Fu et al., 2012).

PcrA ATP Binding Mutants Render Cells Unviable
PcrA3 Sau (T61I), defective in ATPase and helicase activities, and the PcrA Sau (PcrAH − ) mutant variant (K33A Q250A), which lacks ATPase and helicase activities, can displace RecA Eco from ssDNA, and inhibit RecA Eco -mediated DNA strand exchange in vitro (Anand et al., 2007), suggesting that ATP binding and hydrolysis and the DNA helicase activity of PcrA are dispensable for its anti-recombinase activity. To assess whether the ATPase and/or translocase activity of B. subtilis PcrA are essential for cell proliferation, plasmid-borne pcrA genes with T65I, K37A, or Q254A mutations, equivalent to S. aureus mutations, were constructed and used to replace the wt pcrA gene. The single pcrA Q254A mutant gene also contained the unselected E224V mutation. The crystal structure of PcrA Bst DNA helicase with ADP suggests that residue Q254 is in hydrogen bonding distance to the γ-phosphate group of ATP, the E224 and a water molecule at the nucleotide binding pocket (Subramanya et al., 1996;Dillingham et al., 1999). The Q254 residue directly contacts the γ-phosphate of ATP, and the E224 forms hydrogen bonds with two water molecules that are hydrogen bonded to γphosphate oxygen atoms (Dillingham et al., 1999). Residue E224 is in an ideal position for activating the nucleophilic water molecule during hydrolysis. Mutations in residue Q254 alter the coupling between ATPase and helicase activities, and a transient interaction of Q254 with the γ-phosphate of ATP modifies the protein DNA binding site. Same mutations in residue Q254 are toxic for E. coli cells (Dillingham et al., 1999), and the untargeted E224V mutation might overcome such toxicity. The plasmid-borne pcrA Q254A (E224V) gene was also tested.
Monomeric plasmid DNA is inactive for transformation, but if it shares homology with recipient is activated upon interaction with the homologous region in the chromosome via RecA-mediated gene conversion, and the information present in the plasmid is transferred to the host chromosome [see Canosi et al. (1981)]. Monomeric DNA of plasmidborne pcrA T65I (pCB1225), pcrA K37A (pCB1133), or pcrA Q254A (E224V) (pCB1119) was used to transform B. subtilis BG214 cells with selection for the plasmid marker (Table 1). Transformants carrying the established low copy unstable pCB1225, pCB1133, or pCB1119 plasmids were grown in the absence of selective pressure, and the plasmid-less segregants were subjected to nucleotide sequence analysis. We confirmed that in all transformants that lost pCB1225, the pcrA T65I gene had replaced the chromosomal wt pcrA gene by gene conversion (Petit et al., 1998). In only ∼20% of the transformants that lost pCB1119, pcrA Q254A (E224V) mutant gene had replaced the wt gene, but all sequenced pcrA Q254A clones also contained the unselected E224V mutation. However, in all transformants that loss pCB1133, nucleotide sequence analyses revealed the presence of the wt gene, thus we have failed in the attempt to observe the replacement of the wt gene by the pcrA K37A mutant gene, suggesting that the K37A mutation, which impairs ATP binding, renders cells non-viable. This failure, however, could also be attributed to different reasons, for example that the mutation indirectly impacts in the expression of the downstream essential ligA gene (http://www.subtiwiki.unigoettingen.de). Since the PcrAH − Sau mutant variant can promote disassembly of heterologous RecA Eco from ssDNA (Anand et al., 2007), but PcrAH − fails to overcome RTCs, and depletion of wt PcrA in a background expressing PcrAH − renders cell inviable , we have dropped the plasmid segregation approach and moved our analysis to the condition in which PcrA is selectively depleted by the pcrA-ssrA sspB degron (pcrA T ) strain  to answer these puzzling observations. PcrA Depletion Inviability Requires RecJ and RecO, but Not RecQ, RarA, RecU, RecX, or AddAB First, to confirm the reduction of cell survival following PcrA depletion, the pcrA-ssrA sspB degron (pcrA T ) strain (Table 1) was exponentially grown to an OD 560 = 0.4 (at 37 • C) in LB medium and then, appropriate dilutions were plated on LB agar plates lacking or containing 500 µM IPTG (Materials and methods). In the absence of IPTG, the viability of the pcrA-ssrA sspB (pcrA T ) degron strain was slightly compromised (1.3-fold) ( Figure 1A, gray bar [-IPTG condition]) under the experimental conditions used (see below), perhaps due to noise introduced by sspB gene expression (see below). The plating efficiency of the pcrA T strain was reduced ∼5,000-fold upon plating onto LB agar plates containing IPTG when compared to the pcrA + control (rec + ) ( Figure 1A, yellow bar [+ IPTG condition]). This is consistent with the earlier observation that pcrA T cell viability was reduced by >1,000-fold when plated onto 100 µM IPTGcontaining plates . We have observed a linear decrease in the number of viable cells with increasing IPTG concentration, but it saturates above 500 µM IPTG (data not shown). Upon IPTG addition the expression of the SspB adaptor increases, and SspB interacts with the SsrA moiety of the PcrA-SsrA protein to deliver it to the ClpXP protease for PcrA degradation [see Keiler et al. (1996), Griffith and Grossman (2008)]. It is worth mentioning that no fitness cost to B. subtilis cells was observed at IPTG concentrations as high as 5 mM IPTG. After 15 min of IPTG addition to the pcrA T degron strain, 60 to 90% of PcrA is degraded .
To gain insight into PcrA contribution to repair-byrecombination, the pcrA T degron strain was exposed to limiting  (Figures 1A,B), suggesting that depletion of PcrA renders cells sensitive to both DNA damaging agents.
B. subtilis pcrA lethality is suppressed in the recL16 background (Petit and Ehrlich, 2002). Before testing the causes of the suppression of B. subtilis pcrA lethality in the recL16 context, we must understand the function(s) impaired in the recL16 strain. As described in Supplementary Material, Annex 1, the B. subtilis mutations previously designated as recL16 actually map to the recO locus (C. M, Marielle C. Gruenig, Michael M. Cox and J.C.A., to be published elsewhere). To simplify the analysis, the MMS S phenotype was transferred by gene congression to competent BG214 cells. Five of the resulting MMS S clones were selected for whole genome sequencing along with the isogenic BG214 rec + isogenic strain ( Supplementary  Material, Annex 1). One of the MMS S clones showed a TGA Opal stop triplet at codon 37 of recO and was designated recO16 (BG107-1 strain) and selected for further analysis (Table 1).
In the presence of IPTG, the survival of the recO16, addAB, recX, or recU strain decreased ∼12-, ∼65-, ∼3-, and ∼7fold upon addition of 0.2 mM H 2 O 2 (blue bar [+ IPTG and H 2 O 2 ]), and ∼2.5 × 10 6 -, ∼6 × 10 6 -, ∼1 × 10 5 ,-and ∼1 × 10 4 -fold upon addition of 1.3 mM MMS (green bar [+ IPTG and MMS]), respectively (Figure 1). Cell survival when comparing the single recJ, recQ, and rarA mutant strains and the rec + strain was not statistically significant decreased under the H 2 O 2 and MMS concentrations used in this assay (Figure 1), although cell survival was significantly decreased in the recJ and recQ background at higher H 2 O 2 and MMS concentrations, and in the rarA context at higher H 2 O 2 concentrations (Sanchez et al., 2006;Romero et al., 2019a). No significant differences were observed when the survival of rec + and rec-deficient strains in the absence of IPTG was compared to the presence of IPTG.
Next, the contribution of the absence of the AddAB helicasenucleases complex (counterpart of RecBCD Eco ), the RecJ 5 ′ → 3 ′ ssDNA exonuclease, or the RecQ DNA helicase (Table S1) in the pcrA T context was analyzed. In E. coli cells, the RecBCD complex during the end resection process loads RecA onto naked ssDNA, to generate a 3'-overhang coated by RecA (Kowalczykowski, 2015). If AddAB works in a similar fashion it is expected that PcrA depletion inviability requires AddAB. The presence of IPTG in rich medium agar plates did not significantly affected cell viability of the recJ pcrA T (Figure 1A), and thus the lethality upon PcrA depletion was suppressed by inactivating recJ, but the colony area was ∼9-fold smaller than in the absence of IPTG. In contrast, PcrA depletion was not suppressed by addAB or recQ inactivation ( Figure 1B). It is likely that the inviability upon PcrA depletion required the RecJ, but not the AddAB or RecQ functions. Furthermore, the viability of addAB pcrA T cells was significantly decreased (∼10-fold), but not that of recQ pcrA T cells (∼3-fold) when compared with the pcrA T degron strain ( Figure 1B).
Exponentially growing recJ pcrA T , addAB pcrA T , or recQ pcrA T cells were then plated on LB agar plates containing IPTG and H 2 O 2 or MMS. The recJ mutation significantly increased the survival of pcrA T cells on plates containing IPTG and H 2 O 2 (P < 0.001). Unexpectedly, cell survival decreased by ∼3-fold on plates containing IPTG and MMS when compared with pcrA T cells (Figure 1A), suggesting that addition of IPTG and MMS renders recJ pcrA T cells extremely sensitive (P < 0.001) when compared to the only IPTG condition (Figure 1A, yellow vs. green bar). Addition of IPTG and H 2 O 2 or MMS to the recQ pcrA T strain did not significantly affect cell survival, when compared with pcrA T cells ( Figure 1B). Cell survival in the addAB pcrA T strain was significantly decreased (by ∼150and ∼50-fold) on plates containing IPTG and H 2 O 2 or MMS, respectively, when compared with pcrA T cells ( Figure 1B). It is likely that pcrA is not epistatic to recJ or addAB in response to MMS-induced DNA damage, and that PcrA is important possibly for backup pathways for single-strand gap and DSB repair.
Here, we have observed that there are different host requirements for the suppression of lethality between E. coli and B. subtilis cells impaired in end resection: first, the minimal medium synthetic lethality of E. coli rep uvrD cells is suppressed by recJ or recQ inactivation (Lestini and Michel, 2008), although lack of RecQ or RecJ provides very limited suppression of rep uvrD rich medium lethality (Guy et al., 2009). On the other hand, in B. subtilis, the lethality upon PcrA depletion is suppressed by recJ inactivation of cells grown in rich medium ( Figure 1A) and it will be of significant interest to understand the molecular basis of the small colony size of recJ pcrA T upon PcrA depletion. Second, the lethality upon PcrA depletion is not suppressed by recQ inactivation, and survival of the recQ pcrA T cells was not significantly affected and reduced in the presence of H 2 O 2and MMS-induced DNA damage, respectively ( Figure 1B). Unlike E. coli, two RecQ-like enzymes (RecQ and RecS) are present in B. subtilis (Fernández et al., 1998). RecQ, which is 591 amino acid long, shares ∼43% identity with RecS (496 amino acid in length) if the first 346 residues containing the DExH helicase domains are used for the alignment; in short, RecS lacks the zinc-finger, the winged-helix, and the RNaseD C-terminal domains (Fernández et al., 1998;Bernstein et al., 2003). We cannot rule out that the partial genetic redundancy, exerted by RecS and RecQ, might mask the phenotype. However, since the viability of the recJ pcrA T or recQ pcrA T strain in the presence of IPTG and MMS was similar (Figures 1A,B), we have to assume that PcrA depletion and recJ or recQ inactivation might have different host requirements.
Next, the contribution of the recO16 mutation in the survival of the recO16 pcrA T strain was evaluated. The presence of IPTG did not significantly affect cell viability of the recO16 pcrA T strain grown in rich medium ( Figure 1A). This is in good agreement with the observation that pcrA inactivation inviability requires RecO when grown in synthetic minimal medium (Petit and Ehrlich, 2002). To test whether PcrA works in a similar or different pathway than RecO, exponentially growing recO16 pcrA T cells were plated on LB agar plates containing 500 µM IPTG and H 2 O 2 or MMS. The survival of the recO16 pcrA T strain significantly increased (>3,500-fold), when compared to the pcrA T strain upon plating in IPTG and H 2 O 2 containing plates. Unexpectedly, the survival of the recO16 pcrA T strain was significantly decreased (>30-fold) when compared to the pcrA T strain on plates containing IPTG and MMS ( Figure 1A). Addition of IPTG and MMS rendered recO16 pcrA T cells extremely sensitive (P < 0.001) when compared to just the addition of IPTG alone (Figure 1A, yellow vs. green bar). RecO has two activities: to load RecA onto SsbA-coated ssDNA in concert with RecR and to mediate DNA strand annealing independently of RecR (Kidane et al., 2004;Manfredi et al., 2008Manfredi et al., , 2010Lenhart et al., 2014). Since pcrA inviability requires RecO and RecR (Petit and Ehrlich, 2002), we assumed that the inactivation of recO compromises RecA loading at or behind a stalled fork, and thereby avoids unscheduled or unwanted RecA-mediated recombination during DNA replication in the context of PcrA depletion.
The contribution of mutants in the modulation of RecA filament growth in the inviability of PcrA depleted cells was also assessed. RarA has at least two activities: to control the loading of pre-primosomal proteins at a stalled fork and to positively modulate RecA filament growth Romero et al., 2019aRomero et al., , 2020. PcrA depletion lethality was not suppressed by rarA inactivation (Figure 1B). After IPTG addition, the viability of PcrA depleted cells was significantly decreased (by ∼7-fold) in the rarA pcrA T context when compared with pcrA T cells ( Figure 1B).
The absence of RarA renders cells sensitive to H 2 O 2 , but not to MMS (Romero et al., 2019a). Cell survival in the rarA pcrA T background was not significantly reduced on plates containing IPTG and H 2 O 2 or MMS, when compared to the pcrA T strain ( Figure 1B). However, this observation is not consistent with the proposal that in E. coli, UvrD prevents RecA binding to ssDNA, possibly by counteracting RarA (Lestini and Michel, 2007).
RecX is a negative modulator of RecA filament growth (Cárdenas et al., 2012;Le et al., 2017), whereas RecU has two activities: to negatively modulate RecA filament growth and to cleave Holliday junctions (HJs) at a cognate site in concert with the RuvAB branch migration translocase Carrasco et al., 2005;Cárdenas et al., 2012;Cañas et al., 2014;Serrano et al., 2018). The absence of RecX or RecU did not restore viability of PcrA depleted cells grown in rich medium agar plates containing IPTG. The viability of PcrA depleted cells was significantly reduced (by ∼12-and ∼13-fold) in the recX pcrA T or recU pcrA T context, respectively, when compared to the pcrA T strain (Figure 1B). The presence of IPTG and H 2 O 2 or MMS strongly reduced cell survival in the recX pcrA T (∼30-and ∼40-fold) and recU pcrA T (∼13-and ∼40-fold) backgrounds, respectively, when compared to the pcrA T strain (Figure 1B), suggesting that the pcrA gene is likely not epistatic to recX or recU in response to non-bulky DNA lesions of oxidative nature. In E. coli, uvrD ruvC (counterpart of B. subtilis RecU) cells are inviable (Magner et al., 2007).

PcrA Depletion Inviability Requires RecA
Rich medium synthetic lethality of E. coli uvrD rep cells is not suppressed by recA inactivation (Veaute et al., 2005;Guy et al., 2009). Similarly, B. subtilis pcrA lethality is not suppressed when the leaky recA1 (formerly termed recE1) mutation is moved into the background (Petit and Ehrlich, 2002), but the RecA1 activities present in the background may mask the outcome (e.g., competent recA1 cells are marginally affected, whereas null recA cells are blocked in chromosomal transformation) (Alonso et al., 1988). To address whether recA inactivation suppresses cell inviability upon PcrA depletion, a recA null mutant allele ( recA) was moved, via SPP1-mediated generalized transduction, onto the pcrA T strain (see Table 1). Absence of RecA reduced survival of the parental strain by ∼80-fold in response to H 2 O 2 , and strongly reduced cell survival (∼6 × 10 6 -fold) in response to MMS, when compared to the wt strain ( Figure 1A).
The lethality of PcrA depleted cells was significantly suppressed (P < 0.001) by recA inactivation upon IPTG addition to rich LB agar plates (Figure 1A), but the colony area was ∼17fold smaller than in the absence of IPTG. Addition of IPTG and H 2 O 2 or MMS did not significantly reduce cell survival in the recA pcrA T background when compared to the recA strain ( Figure 1A, blue vs. green bar). To re-evaluate the results in the recA pcrA T background, the recA pcrA T strain was reconstructed, and similar results were observed. The apparent contradiction between PcrA depleted cells in the recA context ( Figure 1A) with the pcrA recA1 condition (Petit and Ehrlich, 2002), can be attributed either to the leaky recA1 mutation or to background differences, since there are extrachromosomal elements (as the conjugative element ICEBs1 and the prophage SPβ) in the pcrA recA1 context (Petit and Ehrlich, 2002) that are absent in the recA pcrA T background (see Table 1).
It is likely, therefore, that: (i) inviability upon PcrA depletion could not be associated with the inability to promote autoproteolysis of the transcriptional repressor LexA, because inactivation of recA prevents SOS induction, but suppresses pcrA T lethality in the presence of IPTG ( Figure 1A); and (ii) PcrA depletion inviability requires RecA, and pcrA is epistatic to recA in response to non-bulky DNA lesions of oxidative nature.

PcrA Depletion Leads to Unsegregated Chromosomes
As previously proposed for eukaryotic Srs2 (Marini and Krejci, 2010), the role of PcrA might be to prevent the accumulation of crossovers (CO) by promoting synthesis-dependent strand annealing (SDSA), leading to the exclusive accumulation of non-crossover (NCO) products. In circular chromosomes, the outcome of CO and NCO will be a dimeric or two monomeric chromosomes, respectively. Dimeric chromosomes are deleterious and need to be processed before cell division. To test whether PcrA prevents CO accumulation, the chromosome segregation of PcrA depleted cells was studied.
Branched DNA structures can be processed by different pathways. First, the RecU HJ resolvase, in concert with the RuvAB translocase, cleaves the outside or the inside strands of a HJ, followed by religation to produce CO and NCO products, respectively Cañas et al., 2014). Second, a HJ can be dissolved by the RecQ helicase in concert with a Type I DNA topoisomerase to produce NCO products (Kowalczykowski, 2015). Finally, PcrA may dismantle RecA nucleoprotein filaments from branched structures and may promote SDSA by unwinding the elongated invading strand, a step that is followed by annealing to the ssDNA of the other end of the break, an event associated with NCO products (Marini and Krejci, 2010).
To evaluate whether PcrA depletion provokes a chromosomal segregation defect, the nucleoid (supercoiled and compacted chromosome bound by proteins) of the pcrA T , recO16 pcrA T , recJ pcrA T , recQ pcrA T , rarA pcrA T , recX pcrA T , recU pcrA T, or recA pcrA T cells was DAPI-stained and analyzed by fluorescence microscopy. As controls we have used the recU and recA strains. In the absence of any external DNA damage and at mid-exponential phase, recU or recA mutations reduce the number of CFUs by ∼5-and ∼10-fold (Figure 1), and ∼30 and ∼40% of cells are filamented, respectively (Figure 2) , suggesting that in the absence of RecU or RecA, a cell subpopulation undergoes a death-byrecombination phenotype.
Previously it has been shown that after 15 min (37 • C) of 100 µM IPTG addition to B. subtilis cells 60 to 90% of PcrA is degraded . In this study, cells were grown in rich medium under unperturbed conditions until they reached OD 560 = 0.2 (37 • C). IPTG (500 µM) was added to half of the culture, and both cultures were incubated (60 min, 37 • C) before harvesting, fixing and staining the cells with DAPI. During vegetative growth, net accumulation of anucleated cells, unsegregated chromosomes and filamented cells was rare in rec + cells in the absence (Figure 2) or presence of IPTG (data not shown). In this scenario, cells displayed an average length of 4-6 µm, and exhibited a bimodal distribution of nucleoid positioning with ∼35% of total cells having two nucleoids, and ∼65% of total cells containing only one nucleoid with about twice the fluorescence signal as judged by eye . This suggests that the former class were replicated cells with segregated chromosomes and the latter were replicated cells with yet-unsegregated chromosomes.
In the absence of IPTG, ∼97% of pcrA-ssrA cells appeared normal compared to the rec + control (∼100%) (Figure 2). Absence of DAPI staining (anucleated cells) significantly increased (by ∼20-fold) and the fraction of cells with aberrant chromosomes by ∼6-fold when compared to the rec + control, whereas upon IPTG addition absence of DAPI staining was significantly decreased (Figure 2), suggesting that the pcrA-ssrA fusion or noise from sspB gene expression affects chromosomal segregation of unperturbed exponentially growing pcrA T cells. In the presence of IPTG, PcrA dropped, the proportion of cells with an incompletely separated nucleoid or aberrant chromosomes increased by ∼50-fold when compared to the condition without IPTG, and the average cell length was >8 µm in ∼20% of total pcrA T cells (Figure 2). Since those elongated cells contained a single nucleoid it was assumed that they were filamented cells. It is likely that, upon PcrA depletion, NCO shifted toward CO products. In other words, PcrA might suppress COs or might directly contribute to the formation of NCO products, as was shown for its yeast homolog Srs2 (Marini and Krejci, 2010).
In the absence of IPTG, the proportion of anucleated cells and cells with aberrant chromosomes was marginally affected in the recO16 pcrA T , recJ pcrA T, and recA pcrA T strains when compared to the single mutant strain, but in the recA or recA pcrA T condition 40-45% of total cells were filamented (Figure 2). In the presence of IPTG, the absence of DAPI staining and the proportion of unsegregated nucleoids were not significantly affected (<2-fold) when compared to the parental control (rec + pcrA T ) strain (Figure 2). Likewise, in the recO16 pcrA T , recJ pcrA T, and recA pcrA T conditions a significant proportion of cells formed filaments, with <2-fold increase in the proportion of filamented cells when compared to the most affected parental strain (Figure 2). It is likely that the PcrA pro-SDSA function requires RecJ, RecO, or RecA.
The remaining strains were classified into three different classes. First, for RarA and RecQ, which have two activities each FIGURE 2 | Chromosome segregation in the absence of presynaptic or synaptic functions. Cells were grown in LB medium to OD 560 = 0.2; after 60 min, cells were harvested, prepared for DAPI DNA-fluorescence microscopy, and the percentage of anucleate and unsegregated nucleoids determined (condition a). In parallel, at OD 560 = 0.2, IPTG (500 µM) was added and after 60 min, cells were harvested, DAPI-stained, and the percentage of anucleate and unsegregated nucleoids determined (condition b). Representative fluorescent images of two dividing DAPI-treated cells (DNA stain, light blue) are shown. The pictures are taken at the same amplification, two none separated cells (four nucleoids) are presented under normal conditions. The mean of at least three independent experiments is shown.
(see above): inactivation of recQ or rarA revealed a marginal chromosome segregation defect, but ∼20% of total recQ cells formed filaments (Figure 2). In the presence of IPTG, the proportion of unsegregated nucleoids significantly increased in recQ pcrA T or rarA pcrA T when compared to the pcrA T control, but absence of DAPI staining was not significantly affected (Figure 2). Upon PcrA depletion the proportion of filamentous cells was similar in pcrA T and rarA pcrA T cells, but significantly increased in the recQ pcrA T context. Second, RecX negatively modulates RecA filament growth (Cárdenas et al., 2012;Le et al., 2017). In the presence of IPTG, the proportion of unsegregated nucleoids in recX pcrA T was similar to the pcrA T control, but counteracted the formation of filamented cells (Figure 2). Third, for RecU, which has two activities (see above): in the absence of RecU, ∼5% of total cells were anucleated as previously described , suggesting that cell division occurred in regions that had not received a nucleoid, and unsegregated nucleoids accounted up to ∼15% of total cells.
In the presence of IPTG, most of recU pcrA T cells (∼60%) had unsegregated nucleoids, and ∼45% of cells were present as "filaments" (Figure 2), suggesting that PcrA may promote SDSA prior to the formation of a double-HJ that can be resolved to NCO and CO by the RecU HJ resolvase in concert with the RuvAB branch migration translocase Cañas et al., 2014). Altogether, the data presented in Figures 1, 2 revealed certain paradoxes. First, the inviability of PcrA depletion requires RecJ, RecO, and RecA, but under these conditions a chromosomal segregation defect was observed, suggesting that PcrA processes branched DNA structures formed at replication or at replication-transcription stalled forks, but with the help of accessory proteins (e.g., RecJ, RecO) a formed RecA nucleoprotein filament may be dismantled by PcrA. Alternatively, PcrA removes proteins bound to stalled forks to indirectly allow the formation of branched structures. Second, PcrA depletion halts cell proliferation, initiates accumulation of unprocessed branched intermediates, and additively reduces repair-by-recombination in the recQ or rarA context. Finally, PcrA depletion exacerbates the segregation defect of recU cells, with only ∼30% having normal chromosomal segregation, and in the absence of both negative RecA modulators there is a synergistic repair-by-recombination defect.
To study whether the lack of PcrA destabilization of transcription complexes is the primary cause of inviability, the helD pcrA T , recD2 pcrA T , hepA pcrA T , ywqA pcrA T, and mfd pcrA T strains were constructed as described (Materials and methods). PcrA depletion, which decreases cell survival in the recD2 context , was used as control. Except in the mfd strain, the H 2 O 2 or MMS concentrations used were not sufficient to reveal a reduced viability phenotype (Figure 3).
Upon addition of IPTG, the lethality observed following PcrA depletion was not suppressed by recD2 inactivation (Figure 3). In the presence of IPTG, the viability was significantly decreased (by ∼25-fold) in recD2 pcrA T when compared to the pcrA T strain. Addition of both IPTG and H 2 O 2 or MMS significantly reduced cell survival (by ∼8-and ∼5-fold, respectively) in the recD2 pcrA T cells when compared to the pcrA T strain (Figure 3), suggesting that pcrA is not epistatic to recD2 in response to H 2 O 2 -or MMS-induced DNA damage.
Inactivation of the helD helicase partially suppresses the DNA repair defect of recF15, recO, and recR cells (Carrasco et al., 2001). The absence of HelD did not suppress cell inviabilily upon PcrA depletion; in this background, cell viability was not significantly decreased when compared to the pcrA T control upon addition of IPTG (Figure 3, yellow bar [+ IPTG]). Upon addition of IPTG and MMS, cell survival was significantly decreased (by ∼4-fold), when compared to the pcrA T strain (Figure 3, green bar [+ IPTG and MMS]). In contrast, in the absence of E. coli HelD and UvrD cells remain recombination proficient and apparently are as capable of repairing MMSinduced DNA damage as the wt control (Mendonca et al., 1993). Addition of IPTG and H 2 O 2 marginally decreased cell survival (by ∼2-fold) (Figure 3, blue bar [+ IPTG and H 2 O 2 ]).
The absence of HepA or YwqA did not suppress cell inviability upon PcrA depletion (Figure 3). Cell viability was significantly decreased (by ∼8-fold) in the ywqA pcrA T when compared to the pcrA T control, but not in the hepA pcrA T (decreased by ∼3-fold) when compared to the pcrA T control (Figure 3).
Addition of IPTG and H 2 O 2 did not significantly reduce survival in hepA pcrA T or ywqA pcrA T cells when compared to the pcrA T strain (Figure 3). The presence of IPTG and MMS significantly reduced cell survival in the hepA pcrA T , but marginally reduced cell survival in ywqA pcrA T cells when compared to the pcrA T strain (Figure 3). These data altogether suggest that decreasing the probability of backtracking events contributes to maintaining genome stability, but not to suppress the lethality of PcrA depleted cells. Unlike in E. coli cells (Shaw et al., 2008;Jin et al., 2011;Liu et al., 2015), we have little information about how the B. subtilis HepA or YwqA ATPase propels backward translocation of the RNAP along the DNA template or release a sequestered RNAP.
E. coli cells lacking Mfd show a weak sensitivity to UV irradiation (Witkin, 1969). E. coli Mfd recruits UvrA to the site of a roadblock that stalls RNAP. This activity is crucial FIGURE 3 | PcrA lethality is suppressed by mfd inactivation, but not by recD2, helD, hepA, ywqA inactivation. Log phase cultures of wt, single or double mutant strains were diluted and plated on LB agar containing 500 µM IPTG (yellow bars) or lacking it (gray bars). Lethality assays showing cell viability upon PcrA depletion in recD2, helD, hepA, ywqA, or mfd cells. Log phase cultures of indicated strains were diluted and plated on LB agar containing IPTG and 0.2 mM H 2 O 2 (blue bars) or IPTG and 1.3 mM MMS (green bars). Experiments were performed at least four times. The dotted lines mark the survival rate upon PcrA depletion. Data are shown as mean fractional survival ± SEM.
for the recognition and removal of a stalled RNAP, but Mfd is subsequently displaced by UvrB to initiate transcription coupled repair Sancar, 1993, 1994;Ayora et al., 1996;Manelyte et al., 2010;Epshtein, 2015;Ho et al., 2018Ho et al., , 2020Le et al., 2018). In contrast, inactivation of B. subtilis mfd renders cells significantly sensitive to the UV mimetic 4nitroquinoline-1-oxide and also to oxidative non-bulky lesions as those generated by exposure to MMS (Ayora et al., 1996). Our results suggested that Mfd is also required in B. subtilis cells to repair non-bulky lesions (Figure 3). It is likely that in B. subtilis: (i) bulky and non-bulky DNA lesions stall RNAP; and (ii) Mfd interacts with and dislodges RNAP from the damaged DNA template.
Upon addition of IPTG, inactivation of mfd significantly suppressed (P < 0.001) the lethality induced by PcrA depletion (Figure 3), but the colonies were minute and with an area ∼19-fold smaller than in the absence of IPTG, suggesting that Mfd and PcrA play a crucial role in response to a replicative stress. The mfd mutation significantly suppressed the sensitivity of mfd pcrA T cells to H 2 O 2 (addition of both IPTG and H 2 O 2 ), but mfd pcrA T cells showed a nonsignificant decrease when compared to the pcrA T strain in the presence of both IPTG and MMS (Figure 3). It is likely that the pcrA gene is epistatic to mfd in response to MMSinduced DNA lesions. Inactivation of E. coli mfd partially suppresses the sensitivity to UV irradiation in the uvrD context (Epshtein et al., 2014).

CONCLUSIONS
We show that PcrA depletion lethality is suppressed by recJ, recO16, or recA inactivation, but not by addAB, recQ, rarA, recX, or recU inactivation when cells are grown in rich medium (Figures 1A,B). These data suggest that PcrA depleted cells primarily die due to their inability to resuscitate replisomes blocked by a RecA-ssDNA complex. Indeed, RecO loads RecA onto SsbA-coated ssDNA and a RecA nucleoprotein filament downregulates initiation of PriA-dependent DNA replication in vitro (Vlasic et al., 2014), and PcrA depletion inviability requires RecA for replication re-start (Million-Weaver et al., 2015). This is consistent with the observation that PcrA depletion inviability also requires RecO, which loads RecA onto ssDNA (Carrasco et al., 2015), but not AddAB. In a minimal synthetic medium, the pcrA inactivation lethality is also suppressed by inactivation of the recO or recR positive mediators or a leaky mutation in the positive recF17 modulator (Petit and Ehrlich, 2002).
In E. coli cells, the synthetic lethality of uvrD and rep mutations is partially suppressed by recJ or recQ inactivation in minimal medium (Lestini and Michel, 2008). We can envision that the discrepancies observed between E. coli and B. subtilis cells are related to genetic differences between these genetically distant bacteria. First, B. subtilis cells have two RecQ-like helicases, RecQ and RecS, with the latter potentially masking the outcome, whereas E. coli has only RecQ. Second, E. coli cells have two proteins (UvrD and RecX) to actively dismantle a RecA nucleoprotein filament (Petrova et al., 2015;Le et al., 2017), whereas B. subtilis cells have four different proteins (PcrA,RecX,RecU,and RecD2) to do this job (Anand et al., 2007;Le et al., 2017;Torres et al., 2017;Serrano et al., 2018). PcrA was also found to be necessary to survive DNA damage. The pcrA gene is not epistatic to genes involved in end resection (addAB, recJ, recQ), RecA mediators (recO16), or negative RecA modulators (recX, recU) in response to MMS-or H 2 O 2 -induced DNA damage, but it is epistatic to the recA gene, suggesting that PcrA also contributes to repair-by-recombination via poorly understood mechanisms. The role of the positive RecA modulator RarA upon PcrA depletion requires further studies.
As it has been seen previously for Srs2 (Marini and Krejci, 2010), PcrA might play a putative role in promoting SDSA, which does not entail the generation of COs. Depletion of PcrA leads to additive, in recJ, recO and recA, and to synergic accumulation of unsegregated chromosomes in recQ, rarA recU backgrounds. The dual activities of these proteins (see above) mask the interpretation of our results.
Finally, we show that PcrA depletion lethality is suppressed by mfd inactivation, but not by recD2, helD, hepA, or ywqA inactivation (Figure 3). We show that pcrA is not epistatic to recD2, helD, or hepA in response to non-bulky DNA damage, but it is epistatic to mfd. The role of the poorly characterized YwqA ATPase upon PcrA depletion requires further studies.
PcrA and Mfd might act on RTCs, both dependent (Ayora et al., 1996) and independently of the nucleotide excision repair pathway (Figure 3). The PcrA and Mfd translocases physically interact with stalled RNAPs at lesions on the DNA template. PcrA is a pro-backtracking factor by promoting forward RNAP translocation, and Mfd might be an anti-backtracking that dislodges a stalled RNAP, as previously postulated for the isolated protein in vitro (Selby and Sancar, 1993;Ayora et al., 1996;Park et al., 2002;Deaconescu et al., 2006;Epshtein et al., 2014;Sanders et al., 2017;Ho et al., 2018;Le et al., 2018). It is likely that when damaged template bases interfere with RNAP progression, it backtracks and becomes transiently arrested. Under this condition, Mfd and PcrA, which interact with UvrA and UvrB, respectively, are crucial factors involved in mitigating RTCs in the presence of DNA lesions that are or not targeted by transcription coupled repair. Thus, we propose that PcrA is crucial to remove a stalled RNAP that would otherwise hinder DNA replication even in the presence of DNA lesions that are not targeted by transcription coupled repair. In other words, in the absence of PcrA, RNAP may not be evicted from the damage site by Mfd, leading to a harmful genotoxic stress that induces lethality. PcrA allows genome duplication to occur concurrently with other essential DNA transactions (replication, transcription, repair, segregation). Partial PcrA depletion sensitizes cells to severe DNA transactions and its major role is to work in concert with recombination and repair proteins at stalled DNAP/RNAP complexes to facilitate replication progression beyond the conflict point.

DATA AVAILABILITY STATEMENT
The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation.

ACKNOWLEDGMENTS
We are grateful to Marie A. Petit for the yqhH ( hepA) and ywqA strains, to Houra and Christopher Merrikh for the pcrA-ssrA sspB degron strain, and to Marielle C. Gruenig and Michael M. Cox for the early steps of recL16 mapping. We thank C. Marchisone for technical assistance and S. Ayora for comments on the manuscript.