An Epistasis Analysis of recA and recN in Escherichia coli K-12

Abstract

RecA is essential for double-strand-break repair (DSBR) and the SOS response in Escherichia coli K-12. RecN is an SOS protein and a member of the Structural Maintenance of Chromosomes family of proteins thought to play a role in sister chromatid cohesion/interactions during DSBR. Previous studies have shown that a plasmid-encoded recA4190 (Q300R) mutant had a phenotype similar to ∆recN (mitomycin C sensitive and UV resistant). It was hypothesized that RecN and RecA physically interact, and that recA4190 specifically eliminated this interaction. To test this model, an epistasis analysis between recA4190 and ∆recN was performed in wild-type and recBC sbcBC cells. To do this, recA4190 was first transferred to the chromosome. As single mutants, recA4190 and ∆recN were Rec⁺ as measured by transductional recombination, but were 3-fold and 10-fold decreased in their ability to do I-SceI-induced DSBR, respectively. In both cases, the double mutant had an additive phenotype relative to either single mutant. In the recBC sbcBC background, recA4190 and ∆recN cells were very UV^S (sensitive), Rec⁻, had high basal levels of SOS expression and an altered distribution of RecA-GFP structures. In all cases, the double mutant had additive phenotypes. These data suggest that recA4190 (Q300R) and ∆recN remove functions in genetically distinct pathways important for DNA repair, and that RecA Q300 was not important for an interaction between RecN and RecA in vivo. recA4190 (Q300R) revealed modest phenotypes in a wild-type background and dramatic phenotypes in a recBC sbcBC strain, reflecting greater stringency of RecA’s role in that background.

BIOCHEMICAL processes that maintain genomic integrity are essential for all organisms. This is necessary because DNA damage from internal or external sources halting DNA replication can arise during every round of genome duplication. In many cases, this damage, either directly or indirectly, results in a double-strand break (DSB). Organisms have multiple sets of biochemical pathways to deal with these situations. Homologous recombination (HR) refers to a broad set of pathways that allow duplexes of DNA with similar or homologous sequences to interact with a resultant change in linkage. Double-strand-break repair (DSBR) refers to a subset of these pathways that are used to fix DSBs, using an intact homologous duplex as a guide. Typically, a recombinase (Rad51, RecA, and RadA in Eukarya, Bacteria, and Archaea, respectively; Brendel et al. 1997; Lin et al. 2006; Chintapalli et al. 2013) polymerizes on single-strand DNA (ssDNA) liberated from the DNA ends created by the break, forming a protein-DNA helical filament. This filament then searches a second duplex (often a sister chromosome) for a homologous region. When found, the polymerized proteins exchange the homologous strands, healing the break at the molecular level. This process often requires some DNA synthesis to replace DNA destroyed in the initial processing of the DSB. When using HR, the cell usually avoids introducing mutations during the repair process. Many organisms have a second pathway, nonhomologous end joining (NHEJ), that repairs DSBs by joining the two broken ends together [reviewed for Bacteria (Pitcher et al. 2007; Bertrand et al. 2019) and Eukaryotes (Chang et al. 2017)]. This process can introduce mutations at the site of the break (Chang et al. 2017). A key activity in NHEJ is provided by a Structural Maintenance of Chromosomes (SMC) type of protein that holds the two DNA ends in place, while other enzymes process and ligate the DNA ends together (Sjögren and Nasmyth 2001; Hirano 2002; Nasmyth and Haering 2005; Haering et al. 2008; Peters et al. 2008). In organisms where these two pathways coexist, they typically operate independently of one another, using different sets of enzymes.

The relationship between HR and NHEJ is less clear in the model organism Escherichia coli. While this organism has both RecA recombinase and SMC-like RecN proteins, they appear to work in the same HR pathway instead of separate HR and NHEJ pathways. Strong evidence for this is seen from an experiment by Meddows et al. when they performed an epistasis analysis of recA269::Tn10 and ∆recN266 mutations with ionizing radiation (Meddows et al. 2005). They found that while both single mutants were sensitive to ionizing radiation to different degrees, the double mutant was no more sensitive than the more sensitive of the two. It has also been reported that after mitomycin C treatment, RecN-Flag can be precipitated with an RecA antibody, suggesting a physical interaction between the two proteins (Vickridge et al. 2017).

In E. coli, the regulation of expression of recA and recN are also intertwined. Both are members of the SOS response (Courcelle et al. 2001). Upon DNA damage, the SOS response upregulates many genes at the level of transcription that are involved in DNA repair, mutagenesis, and inhibition of cell division. RecA acts as a sensor for the SOS response. It polymerizes on ssDNA, creating a RecA-ssDNA filament. This structure is both an allosteric effector of the self-proteolysis activity of the transcriptional repressor of the SOS response, LexA (Little 1991; Luo et al. 2001), and a key structure in HR as mentioned above. During induction of the SOS response, recA and recN transcription increases 10- and 20-fold, respectively. Cells in exponential phase growth have substantial basal levels of RecA (2900–10,300 molecules per cell) and RecN (130–410 molecules per cell), depending on media, as revealed by ribosome profiling (Li et al. 2014).

recA mutants are recombination deficient and hypersensitive to UV light, ionizing radiation, and mitomycin C treatment (Clark and Margulies 1965; Lusetti et al. 2003). RecA is a 38-kDa protein with a central core domain bracketed by smaller N- and C-terminal domains (Story et al. 1992). The loading of RecA onto ssDNA is catalyzed by either the RecBCD complex (encoded by the recB, recC, and recD genes) or the RecF, RecO, and RecR (RecFOR) proteins, depending on the type of DNA substrate [reviewed in Cox (2007), Michel and Leach (2012), Bell and Kowalczykowski (2016)]. RecBCD processes the end of the DNA to liberate a region of ssDNA that RecA can then use for loading (Kowalczykowski 2000; Singleton et al. 2004; Cockram et al. 2015). Mutants of recB or recC are recombination deficient and extremely sensitive to many types of DNA-damaging agents, particularly those creating DSBs (Low 1968; Willetts and Mount 1969; Sinden and Cole 1978). RecFOR facilitates RecA loading on ssDNA gaps coated with Single-Stranded DNA Binding Protein (Tseng et al. 1994; Umezu and Kolodner 1994; Morimatsu and Kowalczykowski 2003; Morimatsu et al. 2012). These gaps are formed when DNA replication forks encounter certain types of DNA damage (Howard-Flanders et al. 1968; Sassanfar and Roberts 1990). Mutations in recF, recO, or recR have very little effect on conjugational recombination or DSBR in an otherwise wild-type background. These mutants are, however, sensitive to UV irradiation (Horii and Clark 1973; Kolodner et al. 1985; Mahdi and Lloyd 1989a,b).

It was found that recBC mutant cells carrying suppressor mutations in sbcB and sbcC genes (sbcBC, suppressors of recBC) are capable of DSBR, UV resistant, and able to promote recombination of linear DNA substrates during conjugation and P1 transduction (Kushner et al. 1971, 1972; Lloyd and Buckman 1985). It was later found that HR and DNA repair in these recBC sbcBC cells were dependent on a number of other genes, including recF, recJ, recN, recO, recQ, recR, ruvA, ruvB, and ruvC [reviewed in Michel and Leach (2012); Bell and Kowalczykowski 2016]. Mutations in any one of these genes make the recBC sbcBC cell much more sensitive to DNA-damaging agents, remove the ability to do recombination and repair DSBs (Horii and Clark 1973; Lloyd et al. 1983, 1984; Nakayama et al. 1984; Kolodner et al. 1985; Lloyd and Buckman 1985, 1991; Wang and Smith 1988; Mahdi and Lloyd 1989a). A long-standing interpretation of these results is that the sbcBC suppressor mutations allow the proper processing of the double-strand DNA ends to produce a fairly stable DNA substrate with a 3′ ssDNA extension, such that RecFOR can now load RecA onto the ends of this DNA to begin the process of HR and DSBR.

RecN is a 61-kDa protein that is a member of the SMC family of proteins (Cobbe and Heck 2004; Sanchez et al. 2008; Grove et al. 2009). Found in all kingdoms of life, the SMC proteins play roles in chromosome condensation, segregation, sister chromatid cohesion, and recombinational DNA repair (Graumann 2001; Hirano 2005, 2006). SMC proteins have globular N- and C-terminal domains that interact. The middle of the protein is linear, with two coiled-coil domains separated by a hinge region. The active structure is a dimer where the two hinge regions interact to allow the two proteins to form a pincer or V-like structure. Many SMC proteins also interact with a second companion protein that helps them close off or gather DNA in the V-like structure (Hirano 2016; Uhlmann 2016). RecN deviates from the typical SMC protein in two ways. First, it has shorter and more rigid coiled-coil domains (Pellegrino et al. 2012), and second, no companion protein has yet been identified for RecN. For these reasons, Pellegrino et al. have suggested that RecN spirals around two DNA duplexes to hold them together instead of stabilizing them inside a pincer or V-like structure.

E. coli recN is strongly upregulated during the SOS response (Courcelle et al. 2001) and was showed to be involved in a SOS-induced sister chromatid cohesion process after genotoxic stress (Vickridge et al. 2017). RecN is recognized for degradation by the ClpXP protease through residues at its C-terminal end (Nagashima et al. 2006; Neher et al. 2006). Proteolysis is thought to be the way the cell returns RecN levels to homeostasis after being induced to high levels during the SOS response. Overproduction of RecN in E. coli leads to defects in cell division and nucleoid partitioning (Warr et al. 2019). recN mutants are sensitive to ionizing radiation, I-SceI cleavage, and mitomycin C treatment (Picksley et al. 1984; Meddows et al. 2005). They are also defective in conjugational recombination and sensitive to UV radiation in recBC sbcBC strains (Lloyd et al. 1983).

As mentioned above, Meddows et al. have provided evidence that RecN is involved in the RecA-mediated recombinational repair of DNA (i.e., DSBR) (Meddows et al. 2005). Several other groups have provided independent experiments using different approaches that also support this hypothesis (Nagashima et al. 2006; Keyamura et al. 2013; Uranga et al. 2017; Keyamura and Hishida 2019). However, the molecular mechanism of how this occurs is not yet clear. Do RecA and RecN physically interact or do they just belong to the same pathway? Keyamura et al. isolated recA Q300R (in a recA null mutant) that, when expressed from a plasmid, mimicked two ΔrecN mutant phenotypes: sensitivity to mitomycin C (a cross-linking DNA-damaging agent) and resistance to UV irradiation (Keyamura et al. 2013). In contrast, recA null mutants are extremely sensitive to both types of DNA damage (Clark and Margulies 1965; Lusetti et al. 2003). Keyamura et al. further showed that both recA Q300R and ΔrecN cells were filamentous after mitomycin C treatment, and had similar abnormal nucleoid morphology (Keyamura et al. 2013). Additionally, after mitomycin C treatment, the number of nucleoid-associated, plasmid-expressed GFP-RecN foci were reduced in recA Q300R and ΔrecA cells (Keyamura et al. 2013). Based on these data, Keyamura et al. suggested that recA Q300R specifically prevents RecA from physically interacting with RecN, but allows RecA Q300R to participate in other cellular DNA repair processes (repair at ssDNA gaps and the SOS response). The lack of this interaction prevents RecN from executing its function in DSBR.

In this report, we tested whether recA Q300R [hereafter called recA4190 (Q300R)] specifically removes RecA’s ability to interact with RecN in vivo through an epistasis analysis. We did this by first placing recA4190 (Q300R) on the chromosome at its native locus, under control of its native promoter, and then by measuring the phenotypes of the recA4190 (Q300R) and ΔrecN single mutants and the recA4190 (Q300R) ΔrecN double mutant. If the two mutations are epistatic (they act in the same pathway), then the magnitude of the phenotypes should not be additive when comparing the single mutants with the double mutant (it should show the greater of the two). If they are not epistatic (they act in different pathways), then the magnitude of the phenotypes should be additive. We did these tests in an otherwise wild-type background and in a recBC sbcBC background where ΔrecN has a greater and more important role, as judged by larger phenotypic effects (Lloyd et al. 1983; Picksley et al. 1984). Our results show that as single mutants, recA4190 and ∆recN were Rec⁺ as measured by transductional recombination, but were 3-fold and 10-fold decreased in their ability to do I-SceI-induced DSBR, respectively. In both cases, the double mutant had an additive phenotype relative to either single mutant. In the recBC sbcBC background, recA4190 and ∆recN cells were very UV^S, Rec⁻, had high basal levels of SOS expression, and an altered distribution of RecA-GFP structures. In all cases, the double mutant had an additive phenotype. These results do not support the hypothesis that the recA4190 (Q300R) specifically affects an interaction between RecA and RecN. They also suggest that recA4190 (Q300R) reveals a greater stringency for RecA’s role in DNA repair in the recBC sbcBC background, relative to wild type.

Materials and Methods

Bacterial strains and growth conditions

All bacterial strains are derivatives of E. coli K-12 and are characterized in Supplemental Material, Table S1. The strains were generated using either linear transformation or P1 transduction, according to previously described protocols (Willetts et al. 1969; Datsenko and Wanner 2000). Transformants and transductants were selected on 2% agar plates containing either Luria broth (1% tryptone, 0.5% yeast extract, and 1% sodium chloride) or 56/2 minimal medium (Willetts et al. 1969) supplemented with 0.2% glucose, 0.001% thiamine, 0.02% arginine, 0.005% histidine, 0.02% proline, 0.01% leucine, 0.01% threonine, and appropriate antibiotics (Willetts et al. 1969). Ampicillin was used at 50 µg/ml, chloramphenicol at 25 µg/ml, kanamycin at 50 µg/ml, and tetracycline at 10 µg/ml. The cells were purified on the same type of media on which they were selected and grown at 30° or 37°. L-arabinose was used for induction of the λ Red expression plasmid pKD46 in a final concentration of 0.5% (w/v).

Plasmid construction and transfer of recA alleles onto the chromosome

DNA oligonucleotide primers and plasmids are described in Tables S2 and S3, respectively. A schematic of recA gene mutagenesis is given in Figure S1. recA4190 (Q300R) and recA4190-gfp alleles were initially constructed on plasmids using a 326 bp fragment of the recA gene between PmeI and BlpI sites that was synthesized by Integrated DNA Technologies. The fragment also contained a novel BsiEI site at the amino acid residue Q300 of recA (a change from CGGTCAG to CGGTCgt) to facilitate further screening by restriction fragment length polymorphism. The recA4190 gene fragment was amplified with prSJS1100 and prSJS468 and digested with PmeI and BlpI enzymes. Plasmids pNR117 and pSJS1598 were digested with PmeI and BlpI enzymes and appropriate bands were extracted from the gel before they were ligated with the recA4190 gene fragment using T4 DNA ligase. The ligation mixtures were transformed into SS10168 chemically competent cells to create plasmids pAK11 (8.8 kb) and pAK18 (9.6 kb), respectively. Kanamycin- and ampicillin-resistant colonies were purified and screened for UV resistance and for green patches (in case of pAK18 transformants). Plasmid DNAs from successful transformants were purified and screened for restriction polymorphism using prSJS1090 and prSJS1093. The resulting strains were called SS12954 and SS13491, respectively. Both plasmids were subjected to DNA sequence analysis with prSJS507, prSJS1090, prSJS1093, and prSJS563 to confirm presence of the recA4190 mutation.

pAK11 and pAK18 were used to transfer recA4190 and recA4190-gfp alleles, respectively, onto the chromosome using λ Red recombination method (Datsenko and Wanner 2000). The plasmid pAK11 was linearized by digestion with BamHI-HF and BsaI enzymes. The plasmid pAK18 was linearized by digestion with BamHI-HF, BsaI, and NheI enzymes, followed by PCR with prSJS1655A and prSJS1656A (3.9 kb product). Column purified linear DNA fragments were transformed into SS10191 electrocompetent cells expressing λ Red recombinase from the pKD46 plasmid (Datsenko and Wanner 2000). Successful transformants were selected on LB-Kan and screened for ampicillin sensitivity by PCR with prSJS1655A and prSJS1656A, and for restriction polymorphism using prSJS1090 and prSJS1093. Presence of the recA4190 mutation was verified by sequencing with prSJS468 and prSJS1093 for pAK11 linear transformants; and with prSJS507, prSJS1524, prSJS1093, and prSJS563 for pAK18 linear transformants. The resulting strains were named SS12986 and SS13493, respectively.

Preparation cells for microscopy

Cells were grown in LB broth overnight at 37° with aeration. Then, 100 µl of the culture was added to 3 ml of 56/2 minimal media and grown for 3 hr at 37° with aeration into early log phase. A total of 3–5 µl of the log phase culture was loaded onto a 2% agarose slap prepared from 56/2 minimal medium and low-melting agarose. A coverslip was mounted on top and the slides were incubated for 3–4 hr at 37° before imaging. Cells were visualized using a Nikon E600 microscope equipped with automated filter wheels, shutters, CoolLED light source, and an ORCA-ER camera, as previously described (McCool et al. 2004; Renzette et al. 2005). Phase contrast and fluorescence images were taken for at least nine different fields of view (three fields on three different days) under total magnification of ×600.

Analysis of microscopic images

The images were analyzed with the following software: I-Vision (BioVision Technologies), OpenLabs 5.5.1 (Improvision), SuperSegger (Stylianidou et al. 2016), and MATLAB R2016a and MATLAB R2019a (MathWorks). Individual cells were outlined using SuperSegger. Strains were analyzed for number of cells, cell area, number and shape of fluorescent structures, as well as intensity of the fluorescence signal, using specially written MATLAB programs. The cell area is given as a total number of pixels in a cell, where one pixel is equal to 0.107296 µm. Statistical analysis of average cell area and average relative fluorescence intensity (RFI) was performed with a two-sample Student’s t-test, assuming unequal variances. Statistical analysis of the cell distribution by number of fluorescent structures and distribution of fluorescent structures by shape (see Table 3) was performed with a chi-square test of homogeneity for R × C contingency tables (Ott and Longnecker 2001). P-values are reported in the footnotes for each table.

Table 3. Effect of the ∆recN and recA4190 mutations on the characteristics of RecA-GFP fluorescent structures.

					% Cells witha				% Structuresa,b
Strain	recA	recN	Average cell areaa,c	Structures/aread	0 Structures	1 Structures	2 Structures	≥3 Structures	Circular	Undecided	Linear	Cell count
SS13721	+	+	178 ± 2	2.4	66	28	5	1	77	13	10	1307
SS13723	+	Δ	317 ± 17	3.1	52	30	10	8	41	25	34	1273
SS13499	4190	+	250 ± 12	2.6	71	19	5	5	36	21	43	1601
SS13724	4190	Δ	547 ± 39	4.2	45	27	12	16	19	27	54	1008

^aThe average cell area of ΔrecN (SS13723) or the recA4190 (SS13499) single mutant is significantly different from the wild type (SS13721) (P < 0.001) by the Student’s t-test. The average cell area of recA4190 ΔrecN (SS13724) double mutant is significantly different from either of the single mutants (P < 0.001). The cell distribution by number of structures of ΔrecN or the recA4190 single mutant is significantly different from the wild type (P < 0.001) by the chi-square test of homogeneity for an R × C contingency table. The cell distribution by number of structures of recA4190 ΔrecN double mutant is significantly different from either of the single mutants (SS13723 and SS13499) (P < 0.025 and P < 0.001, respectively). The distribution of structures by shape of ΔrecN or the recA4190 single mutant is significantly different from the wild type (P < 0.001) by the chi-square test of homogeneity for an R × C contingency table. The distribution of structures by shape of recA4190 ΔrecN double mutant is significantly different from either of the single mutants (SS13723 and SS13499) (P < 0.001 and P < 0.005, respectively).

^bA structure was defined as circular if its length is between one- and twofold of its width. A structure was defined as linear if its length is more than threefold longer than its width. Otherwise, a structure was defined as undecided. Errors shown are the SEM of the entire cell population.

^cAll strains have ygaD1::kan recAo1403 recA4155, 4136::gfp-918 (A206E) (Renzette et al. 2005) in the JC7623 (recBC sbcBC) (Kushner et al. 1971) background. Cells were grown as mentioned in the Materials and Methods. Average cell area is given in number of pixels. Errors shown are the SEM of the entire cell population.

^dThe total number of structures divided by the total area of cells multiplied by 1000.

UV survival assay

Assay for UV sensitivity has been described elsewhere (Sandler et al. 1996). Briefly, strains were grown overnight at 37°. Then, 0.2 ml of the cultures was diluted in 13 ml 56/2 minimal media and grown for 4 hr into the log phase at 37° with aeration. The cultures were transferred to the sterile 10 mm petri dishes and, with gentle agitation, were irradiated by UV light from two General Electric 15W germicidal lamps at a rate of 1 J/m²/sec. Samples were collected as 1 ml aliquots at time intervals and serially diluted up to 10⁻⁷. Then, 5 µl of each dilution were spotted on the LB agar and incubated for 24 hr at 37°. Survival at a certain dose was measured as a ratio between number of colony-forming units in the radiated sample and nonirradiated control.

Transduction analysis

Qualitative analysis of the abilities of various mutants to perform intrachromosomal recombination was performed using P1 phage transduction assay. The recipient strains were grown into exponential phase (∼10⁸ cells/ml). A P1 transducing lysate from CAG5052 strain carrying metB1 argE⁺ btuB3191::Tn10 was used to infect into the recipient strains carrying metB⁺ argE3 btuB⁺. The titer of P1 CAG5052 lysate was (1.45 ± 0.26) × 10¹⁰ PFU/ml in the SS6321 strain and (1.77 ± 0.33) × 10¹⁰ PFU/ml in the JC7623 (recBC sbcBC) strain. Tetracycline-resistant (Tet^R) transductants were selected and scored from three independent experiments. Recombination proficiency for each recipient strain was reported as the percentage of Tet^R recombinants per 10⁸ donors. Errors were calculated as indirect measurement errors using SD of the means obtained from the Tet^R recombinants score and the titer of the P1 lysate. For each cross, 96 recombinants (32 per experiment) were selected and screened for the ability to grow on minimal media in the presence or absence of methionine and arginine. Statistical analysis of distribution of transductants by phenotype in Figure 4 was performed with a chi-square test of homogeneity for R × C contingency tables (Ott and Longnecker 2001). P-values are reported in the footnotes for each table.

Figure 4.

Cotransduction analysis to determine recombination proficiency of the recA4190 ΔrecN mutant. Top: Crossover events that occur between a donor DNA (CAG5052) from the transducing P1 lysate and a recipient DNA. Bottom: Recipient strains are indicated in the table and a donor strain is CAG5052. The ratio of the recombination proficiency of each strain to that of the recA⁺ recN⁺ strain is shown. Recombinants are scored for the phenotypes as the percentage from to the total number of screened transductants indicated in the last column. Recombination proficiency indicated as <0.002 refers to values below detection in this assay. Statistical analysis was done using the chi-square test of homogeneity for an R × C contingency table. For the JC13509 background, the distribution of transductants by phenotype of ΔrecN (SS10360) or the recA4190 (SS12990) single mutant was not found to be significantly different from the wild type (SS6321) (P = 0.12 and P = 0.7, respectively). The distribution of transductants by phenotype of recA4190 ΔrecN (SS13459) double mutant is significantly different from either of the single mutants (SS10360 and SS12990) (P < 0.025). For the JC7623 (recBC sbcBC) background, the distribution of transductants by phenotype of ΔrecN (SS12613) is significantly different from the wild type (JC7623) (P < 0.025). The distribution of transductants by phenotype of recA4190 (SS12991) was not found to be significantly different from the wild type (JC7623) (P = 0.25).

DSBR assay

Assay for cell survival after transient chromosomal cleavage at a single I-SceI site has been described elsewhere (Meddows et al. 2004, 2005). Briefly, 12.5 µl of the overnight cultures were diluted in 2.5 ml of LB and grown for 2 hr at 37° to early log phase. The cells were split in two 1 ml aliquots and treated with either arabinose or glucose to a final concentration of 0.2% (w/v). After 30 min of incubation in the shaking water bath, the cells were diluted in 56/2 buffer and 5 µl of each dilution were plated on LB agar in duplicates. Plates were incubated for 8 hr at 37°. Survival was measured as a ratio between number of colony-forming units in arabinose culture and glucose culture.

Data availability

Strains and plasmids are available upon request. The authors affirm that all data necessary for confirming the conclusions of the article are present within the article, figures, and tables. Supplemental material available at figshare: https://doi.org/10.25386/genetics.12786266.

Results

Additive effects of DSBs in recA4190 (Q300R) and ΔrecN mutants

As mentioned above, null mutants of recA and recN are sensitive to DSBs caused by ionizing radiation and I-SceI cuts. In the case of ionizing radiation, these mutations have been shown to be epistatic. If recA4190 (Q300R) identified a residue in RecA required for RecN function in the repair of DSBs, then the two mutations recA4190 (Q300R) and ΔrecN should also be epistatic. That is, the double mutant should be as sensitive to DSBs as the more sensitive of the two single mutants. To test this, we first cloned the recA4190 allele on a plasmid and then transferred it onto the chromosome using λ Red recombineering (Figure S1 and Materials and Methods). We then assessed the ability of the recA4190 (Q300R), ΔrecN, and recA4190 (Q300R) ΔrecN mutants to survive DSBs caused by transient induction of the I-SceI endonuclease as previously described (Meddows et al. 2005). Briefly, a single I-SceI site has been placed at the cynX locus and an arabinose-inducible I-SceI endonuclease gene has been inserted at the λ att site, such that expression and then cutting at the cynX site occurs after the addition of arabinose. Table 1 shows that the recA4190 (Q300R) and ΔrecN single mutants have a 3-fold and 16-fold decrease in viability, respectively, after treatment with I-SceI relative to wild type. The viability of the recA4190 (Q300R) ΔrecN double mutant was decreased ∼60-fold. This is much greater than either of the single mutants. Table 1 also shows that recA938::cat and recB268::Tn10 null mutant strains had >500-fold decrease in viability as expected. We concluded that recA4190 (Q300R) and ΔrecN mutations remove functions from different pathways (an additive effect) that are needed for cell survival after a single DSB caused by I-SceI. We also note that recA4190 (Q300R) has a very mild defect relative to ΔrecN on the cell’s ability to survive a DSB.

Table 1. Survival of the cells carrying mutations in recN and recA after transient chromosomal cleavage at a single site with I-SceI endonuclease.

Strain	recA	recN	recB	Survivala
APS121	+	+	+	0.33 ± 0.10
SS12685	+	Δ	+	0.02 ± 0.0004
SS13453	4190	+	+	0.10 ± 0.02
SS13463	4190	Δ	+	0.0054 ± 0.0026
SS2989	+	+	268::Tn10	0.0016 ± 0.0007
SS12857	938::cat	+	+	0.00022 ± 0.00006

^aThe strains carry an arabinose-inducible ΔattB::P_BAD I-SceI expression cassette and a single I-SceI cleavage site inserted at the cynX gene in the MG1655 background (Meddows et al. 2004). Values given are means ± SD from three to six independent experiments.

The UV^S phenotypes of recA4190 and ΔrecN mutations in a recBC sbcBC background are additive

To gain further insight into the relationship between recA4190 (Q300R) and ∆recN, we measured the survival of the mutants after UV irradiation in the recBC sbcBC background. As mentioned above, mutations in recB and recC inactivate RecBCD, leading to a deficiency in conjugational and transductional recombination as well as sensitivity to UV light, ionizing radiation, and mitomycin C (Low 1968; Willetts and Mount 1969; Kushner et al. 1972). Indirect suppressors sbcB15 and sbcC201 restore recombination proficiency, UV resistance, and mitomycin C resistance to the level of the wild type (Kushner et al. 1971, 1972; Lloyd and Buckman 1985). This test was done in this background because it has been shown that recN mutations decrease the survival of recBC sbcBC dramatically after UV treatment, whereas there is little decrease in otherwise wild-type strains (Lloyd et al. 1983; Picksley et al. 1984). Figure 1 shows that recA4190 (Q300R) and ∆recN were more sensitive to UV light compared to the recA⁺ recN⁺ control strain. Combination of recA4190 (Q300R) and ∆recN in the recBC sbcBC background caused a significant increase in UV sensitivity compared to either single mutant. We concluded that the recA4190 (Q300R) and ∆recN belong to different epistatic groups for UV sensitivity in a recBC sbcBC background.

Figure 1.

Survival of the recA4190 ΔrecN mutant after UV irradiation. The strains in the figure are JC7623, SS12613 (ΔrecN), SS12991 (recA4190), and SS13448 (recA4190 ΔrecN). These strains have the additional genotype of recB21 recC22 sbcB15 sbcC201 in the AB1157 background (Kushner et al. 1971). WT, wild type.

In the recBC sbcBC background, the recA4190 ∆recN double mutant has more filamentous cells that are highly induced for SOS than either single mutant

Many mutant strains that are missing a DNA repair function have higher basal levels of SOS expression (McCool et al. 2004). We want to test whether the recA4190 (Q300R) and ΔrecN single mutants would show higher basal levels of SOS expression than wild type, and if the recA4190 ΔrecN double mutant would show additive amounts in the recBC sbcBC background. We suspected that the chromosomal recA4190 (Q300R) allele would be competent to induce SOS expression because the plasmid-borne recA4190 (Q300R) allele was capable of stimulating the LexA cleavage reaction in vivo (Keyamura et al. 2013). To test if these alleles increased the basal levels of SOS expression, we added a sulAp-gfp reporter at the λ attachment site (attB) in the recBC sbcBC strain and measured SOS expression (and cell area) in individual cells. Since the recBC sbcBC background is sulA⁺ sulB⁺, any SOS expression would also delay cell division and this would be detected by an increase in cell area. The cell area and basal levels of SOS expression of the three strains and controls were determined using phase contrast and fluorescence microscopy and image analysis. Figure 2 presents samples of the phase contrast and fluorescence images of cells, and the data are summarized in Table 2 in two ways. First, Table 2 shows the average cell area and average RFI as proxies for SOS expression. Second, the information was binned according to cell area into two groups, ≤400 pixels and >400 pixels (slightly more than two average cells) and then the average RFI (SOS expression) of these two groups of cells were determined.

Figure 2.

The recA4190 ΔrecN mutant produces an additive effect on both cellular filamentation and SOS induction in the JC7623 (recBC sbcBC) background. As explained in the Materials and Methods, the cells were grown on 2% agarose slabs prepared from minimal media for 3–4 hr at 37°C, before imaging the cells with standard phase contrast and fluorescence microscopy. The strains in the figure are SS13709 (WT), SS13710 (ΔrecN), SS13711 (recA4190), SS13720 (recA4190 ΔrecN), and SS13712 (ΔrecA). The Bar, 10 µm. WT, wild type.

Table 2. Effect of the ΔrecN and recA4190 mutations on cell size and basal levels of SOS expression in JC7623 (recBC sbcBC) background.

Strain	recA	recN	Average cell areaa,b	RFI (SOS)b,c	Cell area ≤ 400		Cell area > 400		Cell count
					RFI (SOS)	% Population	RFI (SOS)	% Population
SS13709	+	+	149 ± 1	6.5 ± 0.2	6.4	99	18.1	1	1657
SS13710	+	Δ	212 ± 8	10.0 ± 0.3	8.6	93	28.1	7	1627
SS13711	4190	+	238 ± 10	12.0 ± 0.4	10.4	92	29.6	8	1344
SS13720	4190	Δ	521 ± 30	19.3 ± 0.5	14.2	82	42.3	18	1565
SS13712	Δ	+	164 ± 2	2.04 ± 0.02	2.0	98	2.1	2	1414

^aCells were grown as mentioned in the Materials and Methods. Average cell area is given in number of pixels. Errors shown are the SEM of the entire cell population.

^bStatistical analysis was done using the Student’s t-test. The average cell area of ΔrecN (SS13710) or the recA4190 (SS13711) single mutant is significantly different from the wild type (SS13709) (P < 0.001). The difference in average cell area of recA4190 ΔrecN (SS13720) double mutant and either of the single mutants (SS13710 and SS13711) is significant (P < 0.001). The level of SOS expression of either single mutant (SS13710 or SS13711) is significantly different from the wild type (P < 0.001). The difference in the level of SOS expression between recA4190 ΔrecN double mutant and either of the single mutants (SS13710 and SS13711) is significant (P < 0.001).

^cLevels of the SOS expression were monitored using a sulAp-gfp transcriptional reporter inserted into the λ attachment site (McCool et al. 2004). SOS expression was quantified as an average relative fluorescence intensity (RFI) of pixels for the entire population of cells normalized to the average fluorescence intensity of the background of the images. Errors shown are the SEM of the entire cell population.

Table 2 shows that both ∆recN and recA4190 (Q300R) had greater average cell areas than the wild type (212, 238, and 149 pixels, respectively). The basal level of SOS was elevated approximately twofold in both single mutants relative to wild type [an average RFI of 10.0 and 12.0 for ∆recN and recA4190 (Q300R), respectively, compared to an average RFI of 6.5 for the wild type]. A total of 99% of wild-type cells had areas ≤400 pixels and the average RFI was 6.4, while the 1% of cells with an area >400 pixels had an average RFI of 18.1. The ∆recN mutant revealed that 93% of their population had ≤400 pixels and exhibited an average RFI of 8.6, while the 7% of cells >400 pixels showed an average RFI of 28.1. The recA4190 (Q300R) mutant showed that 92% of the cells were ≤400 pixels and had an average RFI of 10.4, while the remaining 8% of cells had an average RFI of 29.6.

Once again to test epistasis between recA4190 (Q300R) and ∆recN, the recA4190 (Q300R) ∆recN double mutant was measured and displayed a significant increase in average cell area and the basal levels of SOS expression (521 pixels and RFI of 19.3, respectively) compared with either of the single mutants (see above). The binned data showed that 82% of cells had an area of ≤400 pixels and had an average RFI of 14.2, while the other 18% of cells had an average RFI of 42.3. We concluded that both recA4190 (Q300R) and the ΔrecN increased the cell area and basal levels of SOS expression relative to wild type, and that the phenotypes of these two mutations were additive in the recA4190 (Q300R) ΔrecN double mutant.

In recBC sbcBC cells, the recA4190 (Q300R) and ∆recN mutations both affect, and are additive in the double mutant, for the number and distribution of shapes of RecA-GFP fluorescent structures

About 15% of a population of exponential phase cells undergo a RecA-dependent recombination event yielding a crossover of flanking markers (Steiner and Kuempel 1998). Similarly, ∼10–15% of exponential phase cells have a RecA-GFP structure (Renzette et al. 2005). These presumably identify places where replication fork collapse has resulted in DSBs. The number of these structures increases after DNA damage (Renzette et al. 2005, 2007). The structures formed initially are circular and elongate over time, to become linear before they dissociate (Renzette et al. 2007). The relevance of the shape of the structure to the mechanistic functions of the RecA filaments during recombination is less clear. However, studies have shown that RecA-GFP forms long structures called “bundles” in response to a DSB and that these bundles likely play a role in the homology search during recombinational DNA repair (Lesterlin et al. 2014; Ghodke et al. 2019).

We tested whether recA4190 (Q300R) and ΔrecN would lead to changes in the numbers and/or distribution of shapes of RecA-GFP structures that form in exponential phase cells. To measure RecA-GFP structures, we used a strain containing ygaD1::kan recAo1403 recA4155, 4136::gfp-918 (A206E) (hereafter called recA-gfp) (Renzette et al. 2005). This has been characterized and shown to yield a five- to eightfold decrease in recombination/repair ability relative to recA⁺ depending on the assay used. However, it is proficient enough at recombination to allow the growth of dam mutants that require recA for growth (Marinus 2000). To measure the effect of recA4190 (Q300R) on RecA-GFP structures, we introduced this mutation into our standard recA-gfp construct described above, and placed it on the chromosome at the recA locus. We call this construct recA4190-gfp.

To measure the effect of ΔrecN and recA4190 (Q300R) on RecA-GFP structures, the appropriate mutations were combined in a recBC sbcBC background. Populations of these cells were grown to exponential phase and imaged using phase contrast and fluorescence microscopy. Samples of the fluorescence images are shown in Figure 3. The images containing the cells and fluorescent structures were then analyzed using methods described in the Materials and Methods. Table 3 summarizes the data on average cell area, number of structures per area of cell, distribution of number of structures per cell, and distribution of structures by shape.

Figure 3.

The recA4190 ΔrecN mutant displays an additive effect on RecA-GFP fluorescent structures in the JC7623 (recBC sbcBC) background. As explained in the Materials and Methods, the cells were grown on 2% agarose slabs prepared from minimal media for 3–4 hr at 37°C, before imaging the cells with standard phase contrast and fluorescence microscopy. The strains in the figure are SS13721 (WT), SS13723 (ΔrecN), SS13499 (recA4190), and SS13724 (recA4190 ΔrecN). The Bar, 10 µm. WT, wild type.

Table 3 shows that the cell area of the recA-gfp strains varies with respects to the recA and recN mutations, in a pattern similar to the set of strains in Table 2. Hence, the presence GFP fused to RecA had little effect on the phenotype of cell area. It is seen in Table 3 that strains with recA4190-gfp and ΔrecN single mutants had an ∼8% and 29% increase in the number of structures per cell area respectively, relative to wild type. The distributions of the structures in cells for both recA4190-gfp and ΔrecN single mutants, according to the chi-square test for homogeneity for an R × C contingency table (Ott and Longnecker 2001), were significant (P < 0.001). The recA4190-gfp ΔrecN double mutant had 62% and 35% more structures per area relative to the recA4190-gfp and ΔrecN single mutants, respectively. The statistical analysis of the distributions showed that the change of the recA4190-gfp and the ΔrecN single mutants to the double mutant had a P value <0.001 and 0.025, respectively. We concluded that for both the ΔrecN and recA4190 (Q300R) mutants, the number of fluorescent structures per area and their distribution in the recBC sbcBC cells increased relative to wild type, and their values are additive with respects to the recA4190 (Q300R) ΔrecN double mutant.

Table 3 also shows the analysis of the fluorescent structures with respects to their shape: either circular, linear, or undecided (see footnote b in Table 3 for the definition of each shape). While the wild-type cells had mostly circular structures (77% circular and 10% linear), both ΔrecN and recA4190 (Q300R) mutants showed significant [P < 0.001 relative to wild type) changes in the distribution of circular and linear structures (41% circular and 34% linear for the ΔrecN mutant, and 36% circular and 43% linear for the recA4190 (Q300R) mutant]. The recA4190 (Q300R) ΔrecN double mutant also revealed a change in distribution (and in a similar direction: less circular and more linear) relative to either of the single mutants (19% circular and 54% linear structures). These changes were also significantly different (P < 0.005). We concluded that both the ΔrecN and recA4190 (Q300R) mutants showed an increase in the number of linear structures relative to wild type in the recBC sbcBC background, and their values were additive with respects to the recA4190 ΔrecN double mutant.

The effect of recA4190 (Q300R) and ΔrecN mutations on recombination in the wild-type and recBC sbcBC backgrounds

To test if recA4190 (Q300R) and ΔrecN mutations affect the process of HR, we measured the abilities of these mutants individually and as the double mutant, to affect P1 transductional recombination in both wild-type and recBC sbcBC backgrounds. In addition to measuring the frequency of transductions of the selectable marker (btuB3191::Tn10) per 10⁸ PFU, we monitored an inheritance of the nearby unselected markers metB and argE (see top part of Figure 4). This was done to analyze the locations of the crossover events. The data are shown in Figure 4. In the wild-type background using the recA4190 (Q300R) and ΔrecN single mutants as recipients, these crosses showed slightly decreased frequencies of recombination as compared to the wild type. These differences were not significant. The double mutant, however, showed a significant two- to threefold decrease relative to either single mutant. The distribution of crossovers was not significantly different for either of the single mutant relative to wild type (P > 0.1). But, as with the transductional frequency, the double mutant showed a change in the distribution of crossovers relative to either of the single mutants (P < 0.025). We conclude that the recA4190 and ΔrecN single mutations had little, if any, effect on the transductional recombination frequencies or the distribution of crossovers by themselves. However, the recA4190 (Q300R) ΔrecN double mutant had a significant difference relative to either single mutant for both of these phenotypes.

The transductional frequencies of either single mutant in the recBC sbcBC background were dramatically lower: 300- or 30-fold relative to wild type for the recA4190 and ΔrecN single mutants, respectively. Strikingly, no Tet^R transductants were detected in the cross where the recA4190 (Q300R) ΔrecN double mutant was used as the recipient. This was similar to the ∆recA control. While the distributions of crossovers are reported for these transductions, one should note that the frequencies were so low that we did not considered them statistically useful. We concluded that in the recBC sbcBC background, both the recA4190 (Q300R) and ΔrecN single mutants lowered the ability to do recombination, and this decrease was additive in the recA4190 (Q300R) ΔrecN double mutant.

A surprising result from this part of the study was that no transductants were detected that had crossover events in regions I and IV, using the recBC sbcBC background as the recipient (no recA or recN mutations). This is most striking given that this was the most frequent class in the recBC⁺ sbcBC⁺ crosses.

Discussion

An established body of evidence in the literature suggests that in E. coli, RecN is involved in RecA-mediated pathways of recombinational repair (see above). Yet, it still remains unclear if RecA and RecN physically interact or are just members of the same pathway. recA4190 (Q300R) was identified as a phenocopy of ∆recN in terms of sensitivity to mitomycin C and resistance to UV irradiation (Keyamura et al. 2013). This and other data suggested that recA Q300 is a candidate for a residue in RecA where RecN may interact. Alternately, it could a site in RecA that is required indirectly for RecN function. To critically test these ideas, we first placed the recA4190 (Q300R) allele on the chromosome and then performed an epistasis analysis of recA4190 (Q300R) and ∆recN mutants in two backgrounds: recBC⁺ sbcBC⁺ and recBC sbcBC. Phenotypes measured were survival after a single DSB, UV resistance, average cell area, basal levels of SOS expression, ability to form RecA-GFP fluorescent structures, and recombination proficiency. For all of these phenotypes, we detected an additive effect of the recA4190 ∆recN double mutant compared to the either of the single mutants. These data do not support the hypothesis that the RecN and RecA proteins interact either directly or somehow indirectly through the RecA Q300 residue. The data suggest that while the two mutations have the same phenotype in terms of resistance to mitomycin C and UV irradiation, they seem to show these identical phenotypes for different, independent mechanisms. It should be emphasized that these data do not say anything about how or if RecA and RecN actually do interact.

Aside from testing the specific hypothesis of whether RecN and RecA interact through the RecA Q300 residue, this paper further characterizes the recA4190 (Q300R) and ∆recN mutations using a number of assays where they have not yet been characterized. First, it was shown that both mutations make the cell very UV^S in the recBC sbcBC background. While it was known that recN mutations increase the UV^S of recBC sbcBC cells (Lloyd et al. 1983; Picksley et al. 1984), this was not known for recA4190 (Q300R). This was unexpected given that the recA4190 (Q300R) mutation was originally characterized as UV resistant when expressed from a plasmid in an otherwise wild-type cell. This suggests that recA4190 (Q300R) is specifically affecting the way in which RecA interacts with the DNA repair machinery in a recBC sbcBC background (vs. a recBC⁺ sbcBC⁺ strain). This is true to a greater degree for HR, as measured by P1 transduction (Figure 4). The decrease in recombination ability of either ∆recN or recA4190 (Q300R) in a recBC⁺ sbcBC⁺ strain is less than twofold, while in the recBC sbcBC background, there is a 30- or 300-fold decrease. This is particularly relevant in this study as transductional recombination is similar to repair of a DSB that is produced as a result of replication fork collapse [“Ends Out” Recombination (Smith 1991), reviewed in Michel and Leach (2012)]. It has been previously shown that mutations in this region within the C-terminal domain of RecA affect the ability of the second duplex of DNA to interact with the RecA-ssDNA helical filament in vitro (Kurumizaka et al. 1996). Others have shown that another mutation at the 300 position, recA Q300A, when expressed from a plasmid, is nearly completely deficient in recombination and DNA repair in vivo. It is, however, still able to stimulate LexA cleavage in vivo (Adikesavan et al. 2011). Lastly, both recA4190 (Q300R) and ∆recN single mutants appear to have an increased basal levels of SOS expression. This is not uncommon among mutants that have defects in their DNA repair pathways and usually indicates that they have an increased amount of ssDNA available to RecA to bind, and this is likely to be a result of the lesser ability to properly repair DNA damage arising after replication fork collapse.

It is interesting that the ∆recN or recA4190 (Q300R) mutations have a similar effect on the shape of recA-gfp structures. They increase the number of linear (vs. circular) structures in the population. These linear structures or bundles are also seen after UV damage (Renzette et al. 2007). It is important to realize in this case, that the increase in the linear structures is associated with a decrease in the ability to do recombination and DNA repair. If the linear structures or bundles are indicative of the repair of DSBs, we may be seeing more of these structures because there are more DSBs, the cells are failing to successfully repair the DNA or they are engaged in the process longer than normal.

Another interesting observation to come from this study is that the placement of crossovers in a recBC⁺ sbcBC⁺ vs. recBC sbcBC strain is very different. One sees that the predominant class in the recBC⁺ sbcBC⁺ strain has most crossovers at the ends of the interval studied (I→IV). This is what might be expected for the RecBCD pathway operating on the ends of the transducing fragment and initiating “Ends Out” recombination (Smith 1991). The second most prevalent class for the wild-type recipient is the double crossover that requires four crossover events instead of two (I→II and III→IV). This was unexpected as this class is typically the least frequent class seen. In the recBC sbcBC recipient, the pattern is completely different. Here, the I→III crossover is the most prevalent class. While these results are intriguing, there are many explanations that could lead to these results. For instance, they could be due to unknown idiosyncrasies of this region of the chromosome (i.e., Chi sites in the region). Thus, more experiments are needed to be done before any general statements can be made about the distribution of crossovers during P1-mediated recombination in these backgrounds.

Mitomycin C treatment causes interstrand DNA cross-links (Tomasz et al. 1987). These, in turn, can cause DSBs if replication forks run into a cross-link itself or the repair of the cross-link. The repair of cross-links is complicated involving the coordination between nucleotide excision repair and HR [reviewed in Clark and Sandler (1994), Weng et al. (2010)]. The published data show that while a strain with ΔrecN or multiple copies of recA4190 (Q300R) is very sensitive to mitomycin C (Keyamura et al. 2013), a strain with a single copy of recA4190 (Q300R) is only mildly sensitive (where ΔrecN is fivefold more sensitive) to a single DSB caused by I-SceI (Table 1). It is also likely that the I-SceI-induced break is made away from a replication fork. These differences in the types of DNA damage, how the DSBs are created and their relationship to the DNA replication fork are important considerations, and may provide clues for understanding the differences in defects of recA4190 (Q300R) and ΔrecN mutations.

recA4190 (Q300R) is an interesting allele because it encodes a protein that has wild-type activity for some DNA substrates under some conditions, but not others. These altered specificities of RecA’s abilities may reveal clues about RecA’s function under these different conditions. This allele allows RecA to repair UV damage, induce the SOS response, and perform HR, but has trouble repairing interstrand cross-links caused by mitomycin C treatment in an otherwise wild-type cell. In a recBC sbcBC strain, recA4190 (Q300R) affects UV survival and the ability to recombine DNA, but has no apparent effect on the SOS response. The recBC sbcBC background is unique in the way it has coopted the RecFOR pathway to process double-strand ends in the absence of RecBCD for RecA binding. It is unlikely that recA4190 (Q300R) affects some interaction with RecFOR, as recBC⁺ sbcBC⁺ strains are fully UV resistant (Keyamura et al. 2013).