Chromosome Segregation and Cell Division Defects in Escherichia coli Recombination Mutants Exposed to Different DNA-Damaging Treatments

Homologous recombination repairs potentially lethal DNA lesions such as double-strand DNA breaks (DSBs) and single-strand DNA gaps (SSGs). In Escherichia coli, DSB repair is initiated by the RecBCD enzyme that resects double-strand DNA ends and loads RecA recombinase to the emerging single-strand (ss) DNA tails. SSG repair is mediated by the RecFOR protein complex that loads RecA onto the ssDNA segment of gaped duplex. In both repair pathways, RecA catalyses reactions of homologous DNA pairing and strand exchange, while RuvABC complex and RecG helicase process recombination intermediates. In this work, we have characterised cytological changes in various recombination mutants of E. coli after three different DNA-damaging treatments: (i) expression of I-SceI endonuclease, (ii) γ-irradiation, and (iii) UV-irradiation. All three treatments caused severe chromosome segregation defects and DNA-less cell formation in the ruvABC, recG, and ruvABC recG mutants. After I-SceI expression and γ-irradiation, this phenotype was efficiently suppressed by the recB mutation, indicating that cytological defects result mostly from incomplete DSB repair. In UV-irradiated cells, the recB mutation abolished cytological defects of recG mutants and also partially suppressed the cytological defects of ruvABC recG mutants. However, neither recB nor recO mutation alone could suppress the cytological defects of UV-irradiated ruvABC mutants. The suppression was achieved only by simultaneous inactivation of the recB and recO genes. Cell survival and microscopic analysis suggest that chromosome segregation defects in UV-irradiated ruvABC mutants largely result from defective processing of stalled replication forks. The results of this study show that chromosome morphology is a valuable marker in genetic analyses of recombinational repair in E. coli.


Introduction
Homologous recombination (HR) is indispensable for accurate and efficient repair of potentially lethal genomic lesions, such as double-stranded DNA breaks (DSBs) and single-stranded DNA gaps (SSGs) [1][2][3][4]. In Escherichia coli and many other bacterial species, homologous recombination almost entirely depends on the RecA protein. This recombinase plays a key role during the central, synaptic stage of the recombination process that includes reactions of homologous DNA pairing and DNA strand exchange. The synaptic stage is prepared by binding RecA protein to single-stranded DNA (ssDNA), which is mediated by RecBCD and RecFOR protein complexes [3,5]. The mediator proteins play a crucial role by helping RecA to overcome the SSB protein during competition for ssDNA. Depending on the mediator complex involved in this process, two major recombination pathways, RecBCD and RecFOR, are distinguished in E. coli (reviewed in [1,3]). Polymerisation of RecA molecules on ssDNA in 5 -3 direction gives rise to a nucleoprotein filament (termed RecA filament), which subsequently invades available DNA duplexes and searches for Microorganisms 2023, 11, 701 2 of 30 homology [5]. After pairing with homologous DNA, the invading RecA filament displaces one of the resident strands thus forming three-strand recombination intermediate called D-loop. During further reactions, D-loop is transformed into a four-strand intermediate named Holliday junction (HJ). Eventually, HJ is resolved into separate recombinant DNA molecules by the RuvABC protein complex that possesses both branch migration and endonuclease (resolvase) activities [6]. HJ may also be processed by an alternative, but only partially understood mechanism that involves the RecG helicase and possibly, an unknown nuclease activity [7][8][9].
DSBs represent the most severe form of DNA damage. They arise in the chromosome after treatment with various DNA-damaging agents (e.g., ionising radiation, UV light, bleomycin, nalidixic acid, etc.) or after induction of nucleolytic cleavage [4]. However, DSBs may also occur spontaneously during normal, unperturbed growth of cell cultures. Spontaneous DSBs result mainly from DNA replication accidents that occur when replication fork encounters a damaged DNA template or different obstacles on its path [10]. The RecBCD pathway is responsible for the majority of recombinational repair at DSBs in wildtype E. coli [1]. The RecBCD enzyme binds to the blunt or nearly blunt double-stranded DNA (dsDNA) ends, and then unwinds and simultaneously degrades both strands of the DNA duplex [11]. When the enzyme meets a regulatory octanucleotide sequence called Chi, its 3 -5 exonuclease activity is silenced while its 5 -3 exonuclease activity is increased. This Chi-dependent modification allows RecBCD to produce a long 3 -ending ssDNA tail [12]. The modified RecBCD also gains the ability to direct loading of the RecA protein onto newly produced ssDNA, thus promoting formation of the RecA filament [13,14].
In wild-type E. coli, the RecFOR pathway is used primarily for recombinational repair at SSGs [15], which arise in chromosomes when replication forks skip noncoding lesions (such as UV-induced photoproducts), or arise at stalled replication forks. UV light is a natural DNA-damaging agent that is frequently used in DNA repair studies, particularly those focused on SSG repair. The primary DNA lesions induced by UV are pyrimidine dimers that are subject to nucleotide excision repair (NER) [16]. However, the presence of pyrimidine dimers may alter DNA replication and lead to formation of secondary lesions such as SSGs and DSBs which require recombinational repair in order to be mended [17,18]. SSGs contain a region of ssDNA that is immediately available for RecA loading. Prior to RecA loading, the ssDNA region can be further enlarged by the action of RecQ helicase and several ssDNA exonucleases of different polarities [19]. However, the critical role in initiation of SSGs repair is played by RecFOR proteins that catalyse loading of RecA protein onto ssDNA thus enhancing the formation of RecA filaments within the SSGs [15].
The formation of DSBs in UV-irradiated bacterial cells is well documented, however, the underlying molecular mechanisms are quite complex and still not fully elucidated. DSBs in UV-irradiated cells may arise from: breakage of regressed replication forks [20,21], overlapping NER tracts during removal of closely spaced pyrimidine dimers located on opposite DNA strands [22,23], replication forks running into non-repaired SSGs [18], and cleavage of ssDNA within the SSGs [17,24,25]. Irrespective of their origin, the repair of DSBs in UV-irradiated cells proceeds mainly via RecBCD recombination pathway [21,25].
DNA damage not only interferes with the primary DNA functions (such as replication and transcription), but may also affect the global genome structure. Bacterial cells exposed to various DNA-damaging agents display pronounced changes in nucleoid morphology (such as nucleoid compaction and aggregation) and lack of normal nucleoid segregation [26,27]. These perturbations on the DNA level are usually accompanied by cell division defects including cell filamentation and aberrant divisions that produce anucleate cells [26,[28][29][30]. The extent, duration and reversibility of the above cytological changes depend on the dose of the DNA-damaging agent applied as well as on the genetic background of the affected cells.
Studies on E. coli and Bacillus subtilis have revealed that nucleoid compaction provoked by UV light or nalidixic acid largely depends on the functional RecA protein and formation of RecA filaments [26,27,31]. This suggests that ongoing recombinational repair may participate in chromosome reconfiguration after DNA damage. In line with this is finding that inactivation of the RecBCD enzyme (by a recB null mutation) suppresses nucleoid aggregation in E. coli cells after gamma (γ) irradiation [30]. In contrast, the ruv, recG and ruv recG mutants of E. coli which are deficient in resolution of recombination intermediates, show drastic nucleoid segregation and cell division defects after UV-or γ-irradiation [28][29][30]. Moreover, moderate chromosome segregation and cell division defects are present even in unirradiated, normally grown ruv and ruv recG mutants [29], and this phenotype is exacerbated in strains carrying additional recombination-related mutations [32]. This is compatible with finding that recombination intermediates (HJs) occasionally appear in exponentially growing cells as a result of the repair of spontaneous DNA damage [33].
In the present work, we have conducted a parallel analysis of different types of DNA damage (i.e., different DNA substrates), different mechanisms of their repair, and the consequent cytological effects. We have studied chromosome morphology and segregation after three different DNA-damaging treatments in E. coli strains carrying mutations in presynaptic (recB, recO) and postsynaptic (ruvABC, recG) recombination functions. The DNA-damaging treatments used were: (i) chromosome cleavage caused by expression of the yeast I-SceI endonuclease, (ii) γ-irradiation, and (iii) UV-irradiation. The recombination mutants were also checked for their survival after particular DNA-damaging treatment. The results obtained served us to correlate typical cytological patterns of examined mutants with presumed recombinational repair defects associated with particular mutations. In this way, we systematised and supplemented the previous results of our and other research groups showing that chromosome morphology is a valuable marker in genetic analyses of recombinational repair pathways in E. coli. Our analysis enabled us to get novel insight into complex genetic interactions that occur during recombinational repair in E. coli, particularly in UV-irradiated cells that engage both major recombination repair pathways and which also use some of the recombination proteins for rescue of stalled replication forks.

Bacterial Strains, Growth Media and Conditions
All E. coli strains used for DNA repair and cytological experiments (Table S1 (Supplementary Materials)) were derivatives of strain MG1655 [34]. They were grown in LB liquid medium or on LB plates [35]. Cell growth in liquid medium was monitored on spectrophotometer, by measuring optical density at 600 nm (OD 600 ). New strains were constructed by P1 transduction [35]. Mutations in the recombination genes (deletions or insertions) transferred by P1 transduction were associated with antibiotic resistance markers. Therefore, most transductants could be directly selected on LB plates supplemented with appropriate antibiotics: apramycin (Apra), 60 µg/mL; chloramphenicol (Cm), 15 µg/mL; kanamycin (Km), 25 µg/mL; tetracycline (Tc), 10 µg/mL; and trimethoprim (Tm), 100 µg/mL. In addition, transductants were checked for their UV sensitivity (UV s ) phenotype (Table S1). The ∆ruvABC::cam mutation could not be directly introduced into the TRM387 background, which already contained the chloramphenicol resistance marker. In this case, the ∆ruvABC::cam mutation was co-tranduced with the close marker eda-51::Tn10. Co-transductants were first selected on plates with tetracycline and then screened for the ∆ruvABC::cam mutation using a UV sensitivity test.

Measurement of Bacterial Survival after Introduction of DSBs by I-SceI Endonuclease
To study specifically DSB repair, we used an engineered E. coli strain TRM387 carrying both an arabinose-inducible cassette for expression of the yeast I-SceI endonuclease, and an I-SceI cleavage site cloned in the chromosomal argE locus [8]. During growth in the presence of arabinose, this strain expresses I-SceI nuclease which then cleaves its target site and creates a DSB with two ends that have four nucleotides-long 3 overhangs [8].
For the purpose of this study, TRM387 derivatives carrying mutations in various recombination genes were constructed by P1 transduction (Table S1). The DSB repair capacity of constructed strains was estimated essentially according to the procedure described previously [8]. Bacterial cultures were grown from a single colony in 20 mL of LB medium at 37 • C, and their growth was monitored by measuring OD 600 ( Figure S1). When the cultures reached the exponential phase (at OD 600 of 0.2), each of them was divided into two equal parts. L-arabinose was added to one half of the culture (to induce expression of the I-SceI endonuclease), and glucose to the other (to prevent I-SceI expression). Both sugars were added to final concentrations of 0.2% (w/v). Incubation of the split cultures was continued at 37 • C for another 30 and 60 min. At these intervals, samples were taken for determination of cell viability. Cells from each culture were first washed by centrifugation and resuspension in phosphate buffer. Bacterial suspensions were then diluted in the same buffer to obtain a series of decimal dilutions. Five 10-µL aliquots of each dilution were spotted on LB medium plates. Colonies of viable cells were scored after 24 to 72 h of incubation at 37 • C. The number of colony-forming units (CFU) for each cell sample grown in the presence of arabinose was expressed relative to the CFU value of the corresponding control sample grown with glucose. The effect of arabinose-induced I-SceI expression on cell growth was measured spectrophotometrically ( Figure S1).

Fluorescence Microscopy of Bacterial Cells after Induction of I-SceI Expression
Exponential cultures of TRM387 derivatives were grown as described above. After each culture reached an OD 600 of 0.2, 2 mL samples were taken for microscopic analysis as a control. Arabinose was then added to the cultures (to a final concentration of 0.2% w/v) and incubation was continued for the next 60 min. Thereafter, the cells were centrifuged and transferred to arabinose-free medium in which they were grown for an additional 60 min. Finally, 2 mL cell samples were taken for microscopy. Control and arabinose-grown samples were centrifuged and fixed in 500 µL of 0.1% osmium tetroxide for 20 min. After fixation, the cells were centrifuged again and resuspended in 500 µL of 0.2M cacodylate buffer (pH 7.0). Their nucleoids were stained by adding the fluorescence dye 4 ,6-diamidino-2phenylindole (DAPI) to a final concentration of 1 µg/mL. After 20 min of staining, control cells and cells grown in arabinose were centrifuged and resuspended in 100 and 400 µL of cacodylate buffer, respectively. Glass slides for the microscopy were prepared as follows: 50 µL of 2% low gelling temperature agarose (Sigma-Aldrich, St. Louis, MO, USA) prewarmed at 48 • C was spotted on each glass slide and covered with siliconised coverslip (dimensions 18 × 18 mm). After 2-3 min at room temperature, the coverslip was removed and 3 µL of prepared bacterial suspension was applied to a thin layer of solidified agarose. The agarose with cells was covered with a clean coverslip whose edges were then sealed with transparent nail polish. Cells were observed under Zeiss AxioVert 35 microscope adapted for combined phase contrast and fluorescence microscopy, and equipped with Zeiss Ph3 Plan-NEOFLUAR (100x/1.30 OIL) objective. Images were acquired using Pixera Pro150ES digital camera and Twain Viewfinder v.3.0 capture program.

Measurement of Bacterial Survival after UV-and γ-Irradiation
Bacteria were grown from a single colony in LB medium at 37 • C and their growth was followed by measuring OD 600 ( Figure S2). For UV-irradiation, exponential bacterial cultures (at OD 600 of 0.2) were diluted in phosphate buffer to obtain a series of decimal dilutions. Five 10-µL aliquots of each dilution were spotted on LB medium plates and exposed to different doses of UV light (wavelength, 254 nm) at a dose rate of 0.25 J m −2 s −1 .
The irradiation was carried out under the low pressure mercury germicidal lamp (Phillips), in conditions of subdued light to prevent photoreactivation. The dose rate was determined by VLX-3W radiometer (Bioblock).
In γ-irradiation experiments, cells from exponential cultures were sedimented by centrifugation, resuspended in the same volume of cold phosphate buffer, and irradiated on ice with a 60 Co γ-ray source at a dose rate of 3 Gy/s. The appropriate decimal dilutions of the cells in phosphate buffer were plated on LB agar plates. In both types of irradiation experiments, the colonies of survivors were counted after 24 to 72 h of incubation at 37 • C. The number of CFU obtained for each irradiated sample was expressed as a percentage of the CFU value of the respective non-irradiated control.

Fluorescence Microscopy of UV-and γ-Irradiated Cells
Exponential bacterial cultures were prepared as described above. For UV-irradiation, cells from 6 mL of bacterial culture were harvested by centrifugation and resuspended in the same volume of ice cold phosphate buffer. The cell suspension was transferred into a glass Petri dish (diameter 10 cm) and irradiated with UV dose of 5 J m −2 (at dose rate of 0.25 J m −2 s −1 ) with swirling. Cells from 5 mL of irradiated suspension were precipitated by centrifugation, resuspended in 10 mL of LB medium and grown at 37 • C in water bath with shaking. Samples for microscopy were taken after 2 h of postirradiation incubation.
For γ-irradiation, 5 mL of cell suspension in phosphate buffer was transferred into a plastic tube and irradiated on ice with a 60 Co γ-ray source at a dose of 100 Gy (dose rate 3 Gy/s). Irradiated cells were spun down by centrifugation, resuspended in 10 mL of LB, and grown as described above. Samples for microscopy were taken after 2 h of postirradiation incubation. In both types of irradiation experiments, the procedure of cell sampling and preparation of cells for microscopy was identical to that described above (see Section 2.3). After both γand UV-irradiation cell growth was monitored by measuring OD 600 ( Figure S2).

Statistical Analyses
Statistical significance of differences in survival (after I-SceI expression, gammaradiation and UV radiation at 60 min, 200 Gy and 50 J/m 2 , respectively) was assessed by one-way ANOVA followed by Tukey HSD test. Statistical tests were carried out in R (version 4.2.0), with aov function for ANOVA (from the base package "stats"), and HSD.test (from the package "agricolae", v.1. [3][4][5] for Tukey HSD test with the grouping output. Pairwise p-values reported for the Tukey test are the p-values from the TukeyHSD function (from the base package stats) which reports the pairwise p-values adjusted for multiple testing. Differences in the production of filamentous and anucleate cells between different strains were assessed by the chi-square test of independence (function chisq.test from the base package "stats"). Post-hoc analysis was performed through pairwise chi-square tests between all the strains using function pairwiseNominalIndependence from the package "rcompanion" v.2.4.16. The pairwise p-values have been bonferroni-corrected for multiple testing. Due to the large number of comparisons and to keep the text readable, we abstained from citing all the p-values in the text. Instead, p-values are available in the Supplementary Materials. All the statements comparing different strains in the Results section are in agreement with the results of the statistical analysis.

Effects of recB, recO, ruvABC and recG Mutations on Cell Survival and Morphology after Introduction of DSBs by I-SceI Endonuclease
A set of different recombination mutants was constructed in a genetic background carrying both an arabinose-inducible cassette for expression of the yeast I-SceI endonuclease, and an I-SceI cleavage site cloned in the chromosomal argE locus. The mutants were grown in the presence of arabinose in order to promote I-SceI-mediated DSBs and samples were removed after 30 and 60 min to determine cell viability. As expected from previous studies [8,36], the recombination proficient (wild-type) strain showed only a moderate decrease in survival, whereas a much stronger effect was observed with the recB mutant which is deficient for DSB repair ( Table 1). The recO mutant behaved essentially as the wild-type strain ( Table 1, see also [36]). Additionally, the recO mutation had a negligible effect in the recB background (Table 1). These later findings are in line with the notion that the RecFOR pathway has only a marginal role in DSB repair.
Inactivation of ruvABC and recG genes (which encode major postsynaptic functions) had different effects on cell survival after I-SceI expression. The ruvABC mutant showed similar survival as the wild-type both after 30 min and 60 min of growth in arabinose (Table 1). This contrasts somewhat to the previous work that reported twofold lower survival of ruvABC mutant relative to wild-type after 30 min growth in arabinose [8]. In our experimental conditions, such a difference between ruvABC and wild-type cells was obtained only after 90 min of incubation in arabinose (not shown). In contrast, the recG mutant proved to be significantly more sensitive than the wild-type after 30 min and 60 min of I-SceI expression ( Table 1). The double ruvABC recG mutant showed the strongest drop of survival due to synergistic effects of the two mutations. Except for small discrepancy mentioned above, our results essentially corroborate those described previously [8]. Table 1. Survival of different E. coli mutants after I-SceI endonuclease expression. Statistical analysis is provided in Tables S2 and S3.

Strain
Relevant Genotype Survival a (30 min) Survival a (60 min) ∆recB ∆recO ∆ruvABC recG 0.014 ± 0.0025 0.0047 ± 0.0018 a Cell survival as measured after 30-and 60-min growth periods in 0.2% arabinose compared with identical growth periods in 0.2% glucose. The values are averages ± standard deviations of results of three independent experiments. All strains carry the arabinose-inducible attB::P BAD I-SceI expression cassette.
We have next tested various combinations of mutations in presynaptic (recB and recO) and postsynaptic (ruvABC and recG) genes. In brief, all strains carrying recB mutation in combination with ruvABC, recG, and ruvABC recG mutations showed essentially the phenotype of the single recB mutant (Table 1). Notably, the recB ruvABC recG mutant showed 5-to 6-fold better survival than ruvABC recG mutant suggesting that inactivation of recB precludes toxic (abortive) recombination that occurs in ruvABC recG mutant. Such a conclusion was corroborated by the cytological analysis of the above strains (see further in the text). Taken together, these results are compatible with a view that RecBCD enzyme acts before RuvABC and RecG enzymes during DSB repair. As expected from the above results, inactivation of the recO gene had no additional effects in ruvABC, recG, and ruvABC recG mutants (Table 1). Also, as expected, the recB recO mutant and its ruvABC, recG and ruvABC recG derivatives showed similarly low survival ( Table 1).
The series of strains used in DSB repair essay was also examined for cytological changes associated with introduction of DSBs. For that purpose, the strains were first grown in the presence of arabinose for 60 min, and then transferred to arabinose-free medium and grown for additional 60 min. Upon completion of cultivation, the cells were examined under a fluorescence microscope. The wild-type population contained a significant number of filaments with large unsegregated nucleoids, and with DNA-free zones next to the cell poles ( Figure 1, Table 2). Anucleate cells formed by separation from the ends of the filaments were also observed in the population. The described cytological changes were further enhanced in ruvABC and recG mutants, and especially in double ruvABC recG mutants. All these strains showed pronounced nucleoid aggregation and produced numerous anucleate cells (Figures 1 and S3, Table 2).
In contrast to the strains mentioned above, the recB mutant displayed less pronounced cytological changes after I-SceI expression. This mutant produced fewer filaments than the wild-type strain ( Table 2). The filaments contained DNA that was evenly dispersed throughout the cell volume, leaving no DNA-free zones in the cells ( Figure 2).   Consistent with this, no anucleate cells were observed in recB mutant culture ( Table  2).
The recB ruvABC, recB recG, and recB ruvABC recG mutants were much more similar to the single recB mutants than the corresponding ruvABC, recG or ruvABC recG mutants ( Figure 2, Table 2). In other words, the recB mutation ameliorated cytological defects associated with ruvABC and/or recG mutations.  Consistent with this, no anucleate cells were observed in recB mutant culture ( Table 2). The recB ruvABC, recB recG, and recB ruvABC recG mutants were much more similar to the single recB mutants than the corresponding ruvABC, recG or ruvABC recG mutants ( Figure 2, Table 2). In other words, the recB mutation ameliorated cytological defects associated with ruvABC and/or recG mutations.
In contrast to the effects of the recB mutation, inactivation of the recO gene had no strong cytological effect in either wild-type or in the ruvABC, recG, and ruvABC recG backgrounds ( Figure 3, Table 2). As expected, the recB recO mutant and its ruvABC, recG and ruvABC recG derivatives showed a similar cytological pattern ( Figure 4, Table 2). Additionally, this pattern was similar to that observed in the recB mutant, with or without additional ruvABC, recG and ruvABC recG mutations (compare Figures 2 and 4, Table 2). Taken together, these results strongly suggest that nucleoid aggregation and other cytological defects caused by I-SceI expression in ruvABC, recG, and ruvABC recG mutants result from incomplete DSB repair, i.e., from accumulation of unresolved recombination intermediates formed on the RecBCD recombination pathway.

Effects of recB, recO, ruvABC and recG Mutations on Cell Survival and Morphology after Exposure to γ-Irradiation
Gamma irradiation is an agent that induces a variety of lesions to DNA. However, DSBs are generally considered to be responsible for most of the lethal effects associated with γ-irradiation [16]. To further study the role of the two major recombination pathways in DSB repair, we examined the effects of recB, recO, ruvABC and recG mutations on cell survival after exposure to different doses of γ-irradiation. The mutants used for this experiment were constructed in the genetic background of the strain MG1655 (Table S1). The radiation doses applied only slightly affected the survival of the wild-type strain ( Figure 5A), whereas they caused a strong decrease of survival of the recB mutant ( Figure 5B). Consistent with previous results [8], survival curves for individual ruvABC and recG mutants show similar modestly reduced survival, while the combined ruvABC recG mutant showed a dramatic reduction in cell survival ( Figure 5A). The combination of recB and ruvABC mutations resulted in a phenotype quite similar to that of the single recB mutant, whereas recB and recG mutations produced a moderate synergistic effect ( Figure 5B). The triple recB ruvABC recG mutant was basically as sensitive as the recB recG mutant. Interestingly, the recB ruvABC recG mutant survived γ-irradiation significantly better than the ruvABC recG strain ( Figure 5A,B), reminiscent of the situation observed after induction of I-SceI gene (see Table 1).
In contrast to the effects of the recB mutation, inactivation of the recO gene had no strong cytological effect in either wild-type or in the ruvABC, recG, and ruvABC recG backgrounds ( Figure 3, Table 2). As expected, the recB recO mutant and its ruvABC, recG and ruvABC recG derivatives showed a similar cytological pattern ( Figure 4, Table 2). Additionally, this pattern was similar to that observed in the recB mutant, with or without additional ruvABC, recG and ruvABC recG mutations (compare Figures 2 and 4, Table 2). Taken together, these results strongly suggest that nucleoid aggregation and other cytological defects caused by I-SceI expression in ruvABC, recG, and ruvABC recG mutants result from incomplete DSB repair, i.e., from accumulation of unresolved recombination intermediates formed on the RecBCD recombination pathway.     Inactivation of the recO gene significantly reduced the survival of γ-irradiated cells ( Figure 5C). It also produced a significant synergistic effect in combination with the recB mutation ( Figure 5D). These results are in accord with previous studies showing that RecFOR pathway contributes to DNA repair after γ-irradiation, although its contribution is modest compared to that of the RecBCD pathway [37][38][39]. Combining the recO mutation with ruvABC or recG mutations resulted in a further moderate reduction in survival when compared to the single ruvABC and recG mutations ( Figure 5C). However, the recO mutation did not result in the further reduction in survival of the ruvABC recG mutant (compare A and C in Figure 5). As expected, the recB recO double mutant and its ruvABC, recG or ruvABC recG derivatives had very similar survival curves ( Figure 5D). The collective results described above suggest that RuvABC complex works predominantly on the RecBCD pathway, whereas the RecG protein works on both recombination pathways during DNA repair after γ-irradiation.
Microscopic analysis of the wild-type strain after γ-irradiation showed moderate chromosome aggregation, together with cell filamentation and production of anucleate cells ( Figure 6, Table 3), which is consistent with our previous results [30]. All cytological defects were more severe in ruvABC, recG, and ruvABC recG mutants ( Figure 6). In particular, ruvABC and ruvABC recG mutants showed an extreme production of anucleate cells (Table 3) which was caused by frequent cell divisions in the absence of chromosome segregation. Chromosome segregation defect was most obvious in the ruvABC recG filaments, which had large nucleoids and long DNA-free regions ( Figure 6).
As observed previously [30], the γ-irradiated recB mutant showed seemingly healthier morphology than the wild-type. Although some recB cells were filamentous (Figure 7), their DNA seemed to be on average less aggregated than in the wild-type cells (compare Figures 6 and 7). γ-irradiated recB mutant did not produce DNA-less cells (Table 3). Furthermore, the recB mutation suppressed extreme chromosome aggregation in ruvABC, recG, and ruvABC recG mutants (Figures 6 and 7). Interestingly, the recB recG and recB ruvABC recG mutants were morphologically almost identical to the recB mutant, while recB ruvABC mutant still displayed residual ruvABC phenotype that was particularly evident in a significant number of DNA-less cells (Table 3). Such a phenotype of the recB ruvABC mutant was somewhat surprising given that this mutant had γ-survival curve similar to that of the recB single mutant ( Figure 5B, Table S5). Therefore, it is obvious that chromosome morphology and segregation are very sensitive to the lack of RuvABC function, so that microscopic examination can reveal a minor component of the ruvABC phenotype that remains unnoticed when measuring the survival of irradiated cells.
We drew the following conclusions from the above results. First, cytological defects of the γ-irradiated recG mutant are associated with a defective postsynaptic phase of RecBCDmediated DSB repair. Second, most of the cytological defects of the γ-irradiated ruvABC mutant are associated with the RecBCD pathway, while the remainder of the defects are independent of RecBCD and occur in the presence of RecG function.
We also examined the cytological effects of γ-irradiation in strains carrying the recO mutation, either alone or in combination with other mutations of interest. Compared to the wild-type strain, the recO single mutant showed slightly stronger filamentation, nucleoid aggregation, and DNA-less cell production ( Figure 8, Table 3). Introduction of the recO mutation into the ruvABC strain did not produce significant cytological effects when compared to the single ruvABC mutation (Figure 8, Table 3). The introduction of the recO mutation into the recG and ruvABC recG strains also did not produce strong cytological effects. Finally, we checked the postirradiation morphology of the double recB recO mutant and its derivatives carrying additional ruvABC, recG, or ruvABC recG mutations. All tested mutants showed essentially the same phenotype characterised by the absence of nucleoid aggregation and anucleate cells (Figure 9, Table 3). From these results, we infer that postirradiation DNA reconfiguration in single recO mutant is associated with RecBCD-mediated DNA repair. The absence of anucleate cells in recB recO ruvABC mutant (Table 3) indicates that inactivation of recO was necessary to abolish the residual chromosome segregation defect of the recB ruvABC mutant. The latter finding suggests that a minor part of unresolved recombination intermediates that affect nucleoid morphology in γ-irradiated ruvABC mutant comes from recombination processes occurring via the RecFOR pathway.

Effects of recB, recO, ruvABC and recG Mutations on Cell Survival and Morphology after Exposure to UV-Irradiation
UV light is a DNA-damaging agent that induces SSGs and DSBs, the DNA lesions that both require recombination to be mended [1,3,18]. The survival curves presented in Figure 9 show that single recB and recO mutations moderately reduced cell survival after

Effects of recB, recO, ruvABC and recG Mutations on Cell Survival and Morphology after Exposure to UV-Irradiation
UV light is a DNA-damaging agent that induces SSGs and DSBs, the DNA lesions that both require recombination to be mended [1,3,18]. The survival curves presented in Figure 9 show that single recB and recO mutations moderately reduced cell survival after exposure to UV light, whereas the combined recB recO mutations conferred an extreme UV sensitivity. These results are in accord with previously published data with recB and recF mutations [39]. Reminiscent of the γ-radiation survival curves, in Figure 5, the single ruvABC and recG mutants showed moderate UV sensitivity while double ruvABC recG mutant showed a dramatic drop of survival ( Figure 10A) (see also [40]).
The recB ruvABC mutant showed similar UV sensitivity as the single recB and ruvABC mutants ( Figure 10B, Table S9), suggesting that RecBCD and RuvABC complexes work in the same repair pathway. In contrast, a significant synergistic effect was observed with the double recB recG mutant, which is in accord with previous results [41]. Furthermore, the recB ruvABC recG mutant was only slightly more sensitive to UV than the recB recG mutant ( Figure 10B). Again, the recB ruvABC recG mutant survived significantly better than ruvABC recG mutant (compare Figure 10A,B), thus repeating the survival pattern observed in experiments with I-SceI induction and γ-irradiation (see above). Thus, inactivation of the RecBCD enzyme precluded toxic accumulation of recombination intermediates in UV-irradiated ruvABC recG cells.
Combining the recO mutation with the recG mutation indicated a possible synergy in reducing cell survival ( Figure 10C). The synergistic effect was pronounced with double recO ruvABC mutant suggesting that RuvABC largely acts outside the RecFOR pathway (i.e., on the RecBCD pathway, see above). The triple recO ruvABC recG mutant was equally sensitive as ruvABC recG mutant ( Figure 10A,C). Finally, combining the recB recO pair of mutations with ruvABC, recG, or both, resulted in similar UV sensitivity, i.e., all the multiple mutants behaved essentially as the parental recB recO strain ( Figure 10D).
The applied UV dose of 5 J/m 2 had mild effect on cellular morphology of the wild-type strain ( Figure 11). As in I-SceI-and γ-experiments, more dramatic effect was observed with ruvABC mutant, which showed pronounced chromosome segregation defect accompanied with an extreme production of anucleate cells ( Figure 11, Table 3). The recG mutant dis-played similar phenotype, although with lower production of anucleate cells. Again, the overall morphological defects were most pronounced in the double ruvABC recG mutant, which formed long filaments with highly condensed large nucleoids and numerous anucleate cells (Figure 11, Table 3). The above results are consistent with the results of the previous studies [28,29]. The recB mutation itself did not affect strongly the morphology of UV-irradiated wild-type cells, however, it almost completely abolished chromosome segregation defects in recG mutants (Figure 11). In accord with a previous study [29], the recB mutation had quite moderate effect on the morphology of UV-irradiated ruvABC mutants ( Figure 11). The effect of the recB mutation was manifested mostly in an approximately 40% reduction in the number of anucleate cells in the ruvABC population, with an almost equal increase in the number of normal-sized cells containing DNA (Table 3). This change suggests that the recB mutation leads to a slight improvement in chromosome segregation in UV-irradiated ruvABC mutants. Nevertheless, the overall cytological picture of UV-irradiated ruvABC and ruvABC mutants is basically very similar in that the production of anucleate filamentous cells is very high (Figure 11, Table 3). This result, together with simila Nevertheless, the overall cytological picture of UV-irradiated ruvABC and recB ruvABC mutants is basically very similar in that the production of anucleate and filamentous cells is very high (Figure 11, Table 3). This result, together with similar UV-survival curves for recB, ruvABC, and recB ruvABC strains (see Figure 10A,B) is consistent with a model in which RuvABC resolvase acts prior to RecBCD in processing stalled and regressed replication forks [20,21].
The recB mutation significantly alleviated chromosome segregation defect as well as other cytological defects of ruvABC recG double mutant ( Figure 11, Table 3). This was in accord with beneficial effect of the recB mutation on UV-survival of ruvABC recG cells. Namely, at a UV dose of 5 J/m 2 used for microscopic experiments, the survival of recB ruvABC recG mutant was approximately two orders of magnitude higher than in ruvABC recG mutant ( Figure 10A,B).
Somewhat unexpectedly, the UV-irradiated recB ruvABC recG mutant had less disturbed morphology than either ruvABC or recB ruvABC mutants (Figure 11). This was most obvious in significantly reduced production of anucleate cells in the triple mutant (Table 3). Additionally, the recB ruvABC recG mutant showed less pronounced nucleoid aggregation and cell filamentation ( Figure 11, Table 3). These observations indicate that combination of recB and recG mutations partially suppresses cytological defects of ruvABC mutants.
After UV-irradiation, the recO mutant showed pronounced morphological defects displayed as defective nucleoid partition, anucleate cell production and strong filamentation ( Figure 11, Table 3). The UV-irradiated recO ruvABC mutant showed stronger nucleoid condensation than the single recO mutant. As a result, the recO ruvABC mutant formed filamentous cells with longer DNA-free zones next to cell poles ( Figure 11). In fact, the phenotype of the recO ruvABC mutant was similar to that of the ruvABC mutant, except for a reduced number of anucleate cells in the former strain ( Table 3). The reason for this difference could be in longer cell division delay in the double mutant as inferred from filamentation that was much stronger than that in the ruvABC mutant (Table 3). Indeed, when microscopic analysis was repeated an hour later (i.e., after 3 h of postirradiation incubation), the recO ruvABC mutant showed twice the production of anucleate cells and three-fold reduction in number of filaments (data not shown). In other words, the cytology of the recO ruvABC mutant three hours postirradiation resembles that of the ruvABC mutant two hours postirradiation. Cell morphology of the UV-irradiated recO recG double mutant was quite similar to that of the single recO mutant, except for a slightly increased number of filamentous cells in the double mutant ( Figure 11, Table 3). Furthermore, the morphologies of the recO ruvABC recG and ruvABC recG mutants were also similar, except for the slightly reduced number of anucleate cells after the addition of the recO mutation.
Finally, we investigated the morphology of the UV-irradiated recB recO mutant and its ruvABC, recG and ruvABC recG derivatives. In brief, the recB recO mutant and its derivatives had the same phenotype; none of these strains showed nucleoid aggregation or produced anucleate cells ( Figure 11, Table 3). Hence, the recB mutation suppressed the aforementioned cytological defects of UV-irradiated recO mutants, while the recB recO mutation combination was necessary to achieve the same suppressive effect in ruvABC and ruvABC recG mutants.

Discussion
In this paper, we studied cytological changes in E. coli cells exposed to three different DNA-damaging treatments. We were primarily focused on changes in chromosome morphology and segregation caused by DNA damage, as well as on the accompanying disorders in cell division. Cell morphology was monitored in wild-type strain and in recombination mutants carrying recB, recO, ruvABC, and recG mutations, either alone or in various combinations. The four mutations were selected assuming that two of them (recB and recO) block the initiation stage of the two major recombination pathways, while the remaining two mutations (ruvABC and recG) inactivate major recombination intermediate resolution functions. The results of microscopic analyses were correlated with survival data obtained in the accompanying DNA repair experiments. To "calibrate" our experimental system, we first performed experiments with I-SceI endonuclease expression knowing that it induces exclusively plain DSBs.
Induction of DSBs by I-SceI caused a chromosome partition defect and production of anucleate cells in wild-type, ruvABC, recG, and ruvABC recG strains, although with different intensities. While these cytological disorders were only mildly expressed in the wild-type strain, they were manifested more drastically in mutants defective for resolution of recombination intermediates (Figure 1). Inactivation of recB gene completely suppressed all cytological disorders in strains mentioned above (Figure 2). Additionally, the recB mutation was shown to be epistatic to ruvABC, recG, and ruvABC recG mutations during survival measurements after I-SceI induction ( Table 1). In contrast, introduction of the recO mutation proved to be neutral in DNA repair experiments, and had very little effect in the microscopic experiments ( Figure 3, Table 1). Taken together, these findings clearly indicate that attempted DSB repair via the RecBCD pathway influences DNA distribution within the cell as well as its transmission upon cell division. We assume that the DNA strand exchange between sister chromosomes causes a temporary delay in chromosome segregation and disruption of cell division. In the absence of postsynaptic RuvABC and/or RecG functions, the RecBCD-mediated DSB repair leads to accumulation of recombination intermediates (Figure 12a) that more persistently interfere with chromosome partition and cell division.
The results of our experiments with γ-irradiation largely mimicked those from the I-SceI induction experiments. First, γ-irradiation caused similar cytological defects as I-SceI induction in wild-type, ruvABC, recG, and ruvABC recG strains ( Figure 6, Table 3). Second, the recB mutation has eliminated the most of nucleoid segregation defects and anucleate cell production in the above strains ( Figure 7). Third, the recB mutant and its ruvABC, recG, and ruvABC recG derivatives had generally similar γ-survival curves ( Figure 5). All these findings are in accord with the assumptions that (i) DSBs are the most lethal DNA lesions caused by γ-irradiation [16], (ii) the RecBCD pathway is the main route for DSB repair in E. coli [1,11] (Figure 12a), and (iii) the cytological changes in γ-irradiated E. coli are mostly associated with RecBCD-mediated DSB repair [30] (this paper). However, our results also reveal that the RecFOR pathway affects the morphology and distribution of nucleoids in γ-irradiated cells, although to a much lesser extent than the RecBCD pathway. The RecFOR pathway was responsible for a minor part of nucleoid partition defects in γ-irradiated ruvABC mutant. Interestingly, these defects were also found to be RecG-dependent, suggesting that the RecG helicase acts prior to the RuvABC complex in the RecFOR pathway. These findings also imply that a small proportion of γ-induced cytological changes are associated with the repair of DNA lesions other than DSBs. Since γ-radiation causes various types of oxidative base damage [16], it is possible that some of these lesions lead to the formation of SSGs.
UV-irradiation produced the most complex cytogenetic pattern in our study. This is not surprising given that UV-irradiation causes SSGs and DSBs whose repair requires the activity of both major recombination pathways [1,18]. Recombination, or at least some of its functions, are required to rescue replication forks stalled at UV-induced pyrimidine dimers [20,21,42]. Cytological defects caused by UV-irradiation in wild-type, ruvABC, recG, and ruvABC recG strains were quite similar to those caused by two other DNAdamaging treatments used in this study. However, important differences were observed when recB or recO mutations were introduced in some of the above backgrounds. In particular, neither the recB nor the recO mutation alone suppressed the cytological defects of the UV-irradiated ruvABC mutant ( Figure 11, Table 3). Suppression of these defects was achieved only by simultaneous inactivation of the recB and recO genes. Seemingly, the latter finding is consistent with the assumption that the RecBCD and RecFOR proteins operate on separate repair pathways, both of which require the RuvABC resolvase. However, given the complexity of disorders in DNA metabolism caused by UV radiation and the complexity of cellular response to these disorders, we need to consider other possible models that could explain our results.
Similar cytology of UV-irradiated ruvABC and recB ruvABC mutants (Figure 11), as well as the similarity of their survival curves (Figure 10), is consistent with a model predicting that RuvABC resolvase acts prior to RecBCD in processing stalled and reversed replication forks [20,21] (Figure 12b). Namely, it has been suggested that part of the replication forks stalled at pyrimidine dimers move backward (driven by RecA-mediated re-annealing of template strands) and thus create Holliday junctions (HJs). By cleaving these HJs, the RuvABC resolvase produces DSBs that are repaired by the RecBCD enzyme [21] ( Figure 12b). Based on this model, one could hypothesise that the cytological defects of the UV-irradiated ruvABC mutant result largely from unresolved HJs created by replication fork reversal. Furthermore, our results show that the recO mutation reduces the chromosome segregation defects present in UV-irradiated recB ruvABC mutant ( Figure 11). As part of the above model, this could mean that the RecFOR complex participates in the reversal of the stalled replication fork and its conversion into a HJ. However, the finding that the recO mutation alone has only a modest effect on the cytological phenotype of the ruvABC mutant cannot simply fit into the above scenario. This finding is in better agreement with the results of a previous study suggesting that UV-induced reversal and breakage of replication forks is largely independent of the RecFOR complex [21]. In addition, pronounced cytological defects of recO ruvABC mutants are ameliorated by the recB mutation, suggesting that at least in the recO background, the RecBCD enzyme acts before the RuvABC complex.
Interestingly, the UV-irradiated recO mutant itself shows a significant chromosome partition defect and anucleate cell formation, both of which are efficiently suppressed by the recB mutation ( Figure 11, Table 3). This puzzling observation could be explained by the assumption that inactivation of the RecFOR pathway leads to a redirection of DNA repair to the RecBCD pathway, presumably by conversion of non-repaired SSGs to DSBs [17,18,24] ( Figure 12c). Increased RecBCD-mediated DNA repair activity could be in turn manifested on cytological level through enhanced nucleoid partition arrest and increased production of anucleate cells. Conceivably, these cytological defects would be further exacerbated by inactivation of the RuvABC resolvase.
The idea that inactivation of the RecFOR pathway enhances formation of DSBs after UV-irradiation originates from early studies in which chromosome fragmentation was measured by sucrose gradient sedimentation in NER-deficient uvrB mutants and their recF, recB and recF recB derivatives [24,25]. Due to the extreme UV sensitivity of the strains used, chromosome fragmentation was measured at very low UV doses (up to 1 J/m 2 ) [24,25]. More recently, the measurement of chromosome fragmentation in UV-irradiated recB cells by pulsed-field gel electrophoresis has shown that inactivation of the RecFOR pathway increases the formation of DSBs at UV doses up to 4 J/m 2 [21]. However, at higher UV doses, the recF recB mutant showed a gradual decrease in chromosome fragmentation compared to its recF + counterpart [21]. The direct application of the latter findings to our work is not simple due to differences in genetic backgrounds and growth conditions of the strains used, and possibly also due to differences in dosimetry. In particular, the much stronger UV sensitivity of the conditional recBC (Ts) mutant used in the previous study [21] compared to the recB null mutant used in our work ( Figure 10) raises the possibility that the latter strain accumulated less DNA lesions at nominally the same UV doses. Therefore, it is possible that, at UV dose used in our cytological experiments (i.e., at 5 J/m 2 according to our dosimetry), the recO mutation enables the conversion of unrepaired SSGs to DSBs and thus enhances RecBCD-dependent DNA repair activity.
It has been proposed that the RecFOR complex, together with the RecA protein, participates in the protection and reactivation of stalled replication forks in UV-irradiated E. coli cells [42,43]. This process involves a reversal of stalled forks, which enables NER enzymes access to blocking lesion. According to this model, the reversed forks remain almost intact, undergoing only limited exonucleolytic processing before replisome reloading and replication restart [42,43]. However, as mentioned above, some experimental data strongly suggest that part of the regressed replication forks are broken by the RuvABC resolvase [20,21,40]. It has also been suggested that this latter process is largely RecFORindependent at low UV doses [21]. Currently, it is not known which of the two mechanisms is dominant in UV-irradiated wild-type cells. It is possible that the choice of the mechanism used for rescuing stalled replication forks depends on the dose of UV radiation, i.e., the density of UV lesions on the DNA template [21].
The RecG protein is involved in the processing of recombination intermediates, as well as in the resetting and reactivation of stalled replication forks [44]. UV-survival curves obtained in our work suggest that RecG acts mostly in the RecFOR pathway ( Figure 10). This may indicate an important role of RecG in the repair of persistent SSGs that occur when replication forks skip pyrimidine dimers on the lagging strand template. Conceivably, RecG could be required to process Holliday junctions created during RecFOR-mediated SSG repair. However, chromosome segregation defects as well as the production of anucleate cells in UV-irradiated recG single mutants appear to be entirely related to recombination in the RecBCD pathway ( Figure 11). This suggests that part of the DNA repair defect of UVirradiated recG mutants cannot be directly detected at the cytological level. Furthermore, our results show that the recG mutation aggravates chromosome segregation defects of UVirradiated ruvABC mutants, but modestly alleviates these defects in recB ruvABC mutants ( Figure 11, Table 3). Given that ruvABC and recB ruvABC mutants have similar UV-survival and cytology, the above findings suggest that RecG may act both prior and after the RecBCD enzyme in UV-irradiated cells. It was shown previously that inactivation of RecG slightly increases chromosome fragmentation in UV-irradiated recB mutants, suggesting that RecG activity to some extent counteracts replication fork reversal and/or breakage [21]. Our results suggest that in the absence of RecG, a part of UV repair (and to lesser extent also γ-repair) is re-directed toward RecBCD pathway without the mediating role of the RuvABC resolvase. We can speculate that the RecG protein is involved in the RecFOR-dependent stabilisation of stalled replication forks, and that in the absence of RecG, some of the stalled forks disintegrate independently of the RuvABC resolvase. However, our results do not rule out the involvement of RecG protein in the repair of persistent SSGs. In addition to its presumed role in processing of recombination intermediates, the RecG protein could also be involved in stabilisation of initial homologous joints during SSG repair. If so, part of the strand exchange events during SSG repair in recG mutants would be interrupted at an early stage, leading to SSG persistence and a higher risk of its conversion to DSBs.
The next question is how the lack of RuvABC complex affects the repair of SSGs and consequently, chromosome morphology. UV-survival curves do not show any synergistic effect of ruvABC and recB mutations ( Figure 10A,B), suggesting that the RuvABC complex does not play an important role in the repair of persistent SSGs. Perhaps, the function of RuvABC at SSGs may be efficiently replaced by RecG. Nevertheless, given our finding that the ruvABC mutation can strongly affect cell morphology without having an equally strong effect on cell survival, it is possible that part of the morphological disorders in UV-irradiated ruvABC mutants is associated with defective repair of persistent SSGs. Thus, cytological disorders in UV-irradiated ruvABC mutants are likely the cumulative result of defects in the processing of reversed replication forks as well as defects in the repair of DSBs and persistent SSGs. In any case, the results of this work once again indicate the complexity of the effects of UV radiation on the structure and function of DNA, as well as the intertwining of recombination pathways involved in UV repair.
All treatments that damage the chromosome and/or interfere with its replication or segregation simultaneously cause a halt in cell division (reviewed in [45]). This phenomenon is partially related to the action of the cell division inhibitor protein SulA, which is synthesised as part of the SOS response. Another inhibitory mechanism is based on the nucleoid occlusion system. This mechanism generally prevents septum formation in the cell zone occupied by chromosomal DNA, thus avoiding septation over the nucleoid (the so-called guillotine effect). In the case of defective chromosome segregation, the large nucleoid associated with the cell division inhibitor SlmA remains at midcell and interferes with septation at its normal position, leading to cell filamentation [45]. It is generally accepted that division inhibition mechanisms serve to coordinate chromosome replication and segregation with cell division when DNA metabolism is disturbed. However, none of the inhibitory mechanisms can completely prevent aberrant divisions resulting in anucleate cells or cells with "guillotined" nucleoids [45,46]. In particular, the previous and present data clearly show that the initiation of cell division often precedes chromosome segregation in cells recovering from DNA damage, thus creating a high proportion of anucleate cells in the population ( [28,29,46] and this work).
Induction of the SOS system, including the synthesis of the SulA protein, depends on the loading of the RecA protein to single-stranded DNA that is generated after DNA damage (reviewed in [47]). The loading of RecA protein to ssDNA is mediated by the RecBCD enzyme in the case of DSBs or the RecFOR complex in the case of SSGs. Accordingly, SOS-dependent inhibition of cell division after DNA damage might be expected to depend on the availability of the appropriate RecA loading complex. Indeed, expression of the sulA gene in cells in which DNA has been damaged by I-SceI nuclease is completely dependent on the RecBCD enzyme [36], while in cells exposed to UV radiation, maximal expression of the sulA gene depends dominantly on the RecFOR complex [48]. A mixed situation is present in γ-irradiated cells where both recombination pathways contribute significantly to the induction of the SOS response [49].
Our experiments with I-SceI expression show that filamentation is on average stronger in strains with a functional RecBCD enzyme (Table 2), which could be a combined effect of SOS (SulA)-dependent inhibition of cell division and nucleoid occlusion. Filamentation is further enhanced in strains that show a severe defect in chromosome segregation (i.e., in ruvABC and/or recG mutants), possibly due to persistent SOS-dependent division inhibition and stronger nucleoid occlusion. After expression of I-SceI, the recB mutants are unable to induce the SOS response [36] and thus show only slight filamentation (Table 2), which is probably the result of disturbed DNA arrangement in some cells. Namely, although the recB cells generally do not show a strong disorder in the segregation of chromosomes, they sometimes have DNA dispersed throughout the cell volume (Figure 2), which possibly generates the nucleoid occlusion effect. The same explanation can be applied to recB recO mutants that show a similar phenotype as recB mutants ( Figure 4).
As mentioned above, in γ-irradiated cells, the induction of the SOS response depends on both RecA loading systems [49]. Therefore, a certain degree of SOS (SulA)-dependent filamentation can be expected in recB or recO single mutants after γ-irradiation, which is in agreement with our results (Table 3 and Figures 7-9 and 11). In UV-irradiated cells the SOS response depends on the RecFOR complex [48]. However, at moderate UV doses (up to 20 J/m 2 ), recFOR mutations cause only a short delay in the induction of the SOS response, but not its inhibition [50,51]. This explains the UV-induced filamentation we observed in both recB and recO mutants. It should be noted that in recO mutants, filamentation may be additionally affected by nucleoid occlusion (Figure 11).
However, it is interesting that postirradiation filamentation is also pronounced in double recB recO mutants (see Table 3 and Figures 9 and 11), although the double mutants should be severely deficient for SOS induction [48,49]. Additionally, both UV-irradiated and γ-irradiated recB recO filaments often show partial nucleoid segregation (Figures 9 and 11), which should reduce the effect of nucleoid occlusion. However, it is possible that despite some nucleoids being separated, their position in the cell is not correct, thereby interfering with septation. Furthermore, since recB recO mutants are poorly viable [39,52] and extremely sensitive to both types of radiation ( [39], this paper), it is also possible that cell filamentation occurs as a result of a general deregulation of cellular processes associated with cell dying. Finally, the possibility that there is an additional but still unknown mechanism in E. coli that inhibits cell division after DNA damage cannot be ruled out. An SSG occurs when the replication fork stops at a pyrimidine dimer (black triangle) on the leading strand template. The gap is then coated with RecA protein (white circles) in a process facilitated by the RecFOR complex [42] or an unknown factor (probably a replisome component) [21]. Homologous pairing catalysed by RecA leads to fork reversal, with concomitant replisome disassembly. Reversal of the replication fork allows NER enzymes to access the pyrimidine dimer and remove it. HJ formed by replication fork reversal is cleaved by the RuvABC resolvase thus creating a DSB. The RecBCD enzyme initiates the repair of DSB resulting in the restoration of the replication fork. (c) Minor sources of DSBs after UV and γ-irradiation. The replication fork runs into a single-stranded DNA interruption on the DNA template and collapses. In UV-irradiated cells, transient ssDNA breaks may arise due to ongoing NER. Alternatively, replication fork collapse can Figure 12. Models depicting the possible sources of DSBs and the corresponding repair pathways following the three different DNA-damaging treatments used in this work. (a) Classical DSB induced either by I-SceI endonuclease or γ-irradiation has two dsDNA ends that are repaired by the RecBCD pathway of homologous recombination (HR). (b) In UV-irradiated cells, most DSBs are caused by the breakage of stalled replication forks. The drawing shows the progression of the replication fork on a DNA template damaged by UV radiation (grey ovals represent a replisome). An SSG occurs when the replication fork stops at a pyrimidine dimer (black triangle) on the leading strand template. The gap is then coated with RecA protein (white circles) in a process facilitated by the RecFOR complex [42] or an unknown factor (probably a replisome component) [21]. Homologous pairing catalysed by RecA leads to fork reversal, with concomitant replisome disassembly. Reversal of the replication fork allows NER enzymes to access the pyrimidine dimer and remove it. HJ formed by replication fork reversal is cleaved by the RuvABC resolvase thus creating a DSB. The RecBCD enzyme initiates the repair of DSB resulting in the restoration of the replication fork. (c) Minor sources of DSBs after UV and γ-irradiation. The replication fork runs into a single-stranded DNA interruption on the DNA template and collapses. In UV-irradiated cells, transient ssDNA breaks may arise due to ongoing NER. Alternatively, replication fork collapse can occur at unrepaired SSGs. In γ-irradiated cells, some DSBs may arise when replication fork encounters a site of ongoing base excision repair or an unsealed ssDNA break. Some ssDNA breaks could be widened by ssExos into SSGs that could also cause collapse of the replication forks [49].

Supplementary Materials:
The following supporting information can be downloaded at: https: //www.mdpi.com/article/10.3390/microorganisms11030701/s1, Table S1: E. coli strains used in this work. Figure S1. Growth of different TRM387 derivatives. Figure S2. Effects of γand UV-irradiation on growth rates of different recombination mutants of E. coli. Figure S3. Formation of anucleate cells in ruvABC mutants after introduction of double-strand DNA breaks by I-SceI endonuclease. Table S2. E. coli strains divided into groups (Tukey HSD test, alpha = 0.05) based on the data on survival at 60 min after I-SceI expression (Table 1). Table S3. Statistical analysis of differences in survival at 60 min after I-SceI expression (Table 1) presented as pairwise p-values (p adj) from the Tukey HSD test, which controls for multiple testing. Table S4. Statistical analysis of differences in the occurrence of filamentous, anucleate and regular cells in different strains after 60 min of I-SceI expression ( Table 2). Table S5. E. coli strains divided into groups (Tukey HSD test, alpha = 0.05) based on the data analysed on γ-survival at 200 Gy (see also Figure 4). Table S6. Statistical analysis of differences in survival after 200 Gy of γ-irradiation (see also Figure 4) presented as pairwise p-values (p adj) from the Tukey HSD test, which controls for multiple testing. Table S7. Statistical analysis of differences in the occurrence of filamentous, anucleate and regular cells in different strains after UV dose of 5 J/m 2 ( Table 3). Table S8. Statistical analysis of differences in the occurrence of filamentous, anucleate and regular cells in different strains after γ-dose of 100 Gy (Table 3). Table S9. E. coli strains divided into groups (Tukey HSD test, alpha = 0.05) based on the data on UV-survival at 50 J/m 2 (see also Figure 9). Table S10. Statistical analysis of differences in survival after 50 J/m 2 of UV (see also Figure 9) presented as pairwise p-values (p adj) from the Tukey HSD test, which controls for multiple testing.