RecN: A tunable switch for DNA repair choice and stress tolerance in Zymomonas mobilis

Abstract

Engineering polyploid industrial microorganisms is hindered by their intrinsic capacity to repair induced mutations, limiting the efficiency of genome editing and directed evolution. Using the ethanologenic bacterium Zymomonas mobilis— a polyploid alphaproteobacterium that exhibits exceptionally efficient microhomology-mediated end joining (MMEJ)— we demonstrate that RecN is essential for MMEJ and homologous recombination (HR) in vivo. Strikingly, a specialized mutant RecN-K35A, with strongly impaired ATP hydrolysis, specifically blocks MMEJ while leaving HR fully intact. The physiological importance of RecN-mediated MMEJ is highlighted by the cell elongation phenotype and increased stress sensitivity observed in the RecN-K35A mutant. Based on this connection, we developed a high-phosphorus cultivation strategy that increases cellular DNA content and significantly enhanced ethanol fermentation efficiency under industrial stress conditions. In summary, this work defines RecN as a key ATP-dependent effector of MMEJ and positions it as a potential engineering target for modulating DNA repair pathway choice and stress tolerance in Z. mobilis. Moreover, the essential role of RecN in both HR and MMEJ suggests that RecN-deficient polyploid strains could facilitate directed evolution by preventing repair of newly introduced mutations, offering a new strategy for strain improvement.

1. Introduction

Engineering polyploid industrial microorganisms for enhanced bioproduction remains challenging due to their intrinsic capacity to repair induced mutations, which limits the efficiency of both directed evolution and genome editing [1]. During industrial fermentation, microbial cell factories encounter multiple stresses—including high osmolarity [2], inhibitors [3,4], and product toxicity [5]—that can cause DNA damage, particularly DNA double-strand breaks (DSBs). DSBs represent one of the most lethal forms of DNA damage, posing a severe threat to genomic integrity and cell survival. Efficient and accurate DSB repair is therefore critical for maintaining robust production performance, yet strategies for rationally modulating repair pathways to improve strain robustness remain underexplored.

Bacteria have evolved sophisticated repair mechanisms, primarily homologous recombination (HR) and the error-prone non-homologous end joining (NHEJ) [6]. HR is a high-fidelity repair process that requires an intact sister chromatid as a template and is orchestrated by a suite of proteins including the central recombinase RecA, which catalyzes strand invasion and exchange [7]. In contrast, NHEJ directly ligates broken ends with minimal or no homology, often preserving genome structure at the cost of potential mutations [8]. Between these two extremes lies microhomology-mediated end joining (MMEJ)—an alternative error-prone pathway that utilizes short homologous sequences exposed by limited end resection to align ends prior to ligation, resulting in characteristic deletion scars at repair junctions [9].

The structural maintenance of chromosomes (SMC) protein family member, RecN, has been studied in various bacterial species with potentially distinct roles [10]. In Bacillus subtilis, RecN acts as an early responder that forms dynamic repair centers on the nucleoid upon DSB induction, where it restructures chromosome architecture and recruits repair factors [11] In Escherichia coli, RecN has been shown to topologically bind circular single-stranded DNA (CssDNA), circular double-stranded DNA (CdsDNA), or linear dsDNA with RecA-bound overhangs, leveraging ATP hydrolysis to stimulates DNA strand invasion and subsequent RecA-mediated synapsis and strand exchange. [12,13]. This functional interplay is interdependent, as RecN stabilizes RecA filaments while RecA, in turn, enhances RecN's ATPase activity and directs its focal assembly [14,15]. A dramatic manifestation of this cooperation is the recently described “supercompaction” of DNA, a mid-nucleoid condensation event driven by RecN and RecA that is crucial for survival following severe DSB damage [16,17]. Beyond HR, RecN modulates end compatibility for error-prone pathways. It has been shown to stimulate polynucleotide phosphorylase (PNPase) to resect or blunt 3′-ssDNA overhangs, thereby preparing DNA ends for NHEJ [18]. However, the role of RecN in MMEJ has remained unexplored, representing a gap in this ability to rationally design repair pathways for synthetic biology applications.

Zymomonas mobilis is an alphaproteobacterium of significant biotechnological interest due to its high ethanol production capacity [19]. Importantly, this organism is polyploid, harboring approximately 10-20 genome copies per cell under standard growth conditions [20]. Previous work from this laboratory demonstrated that Z. mobilis exhibits remarkably efficient MMEJ-mediated repair of CRISPR-induced genomic DSBs [21], making it an attractive model for studying this pathway. The organism's DNA end resection machinery is predicted to be AddAB based [22], consistent with its alphaproteobacterial lineage.

Here, we demonstrate that RecN serves as a molecular switch governing DNA repair pathway choice in Z. mobilis, with its ATPase activity being critical for MMEJ. RecN function correlates with stress tolerance, enabling a nutritional strategy to enhance ethanol fermentation (Fig. 1). Moreover, the essential role of RecN in both HR and MMEJ suggests that RecN-deficient polyploid strains could facilitate directed evolution by impairing repair of induced mutations. Together, this work provides a framework for engineering repair pathways to improve microbial cell factory robustness and engineerability.

Fig. 1.

Experimental and Technical Design

This study was designed to elucidate the molecular role of RecN in DNA repair pathway choice and to translate this knowledge into strategies for improving industrial fermentation robustness. The experimental workflow consisted of three integrated phases: (i) genetic dissection of RecN function using CRISPR-based DSB repair assays in Z. mobilis, (ii) biochemical characterization of RecN ATPase and DNA annealing activities, (iii) cell morphology and DNA content analysis and (iv) application of findings to develop a phosphorus-supplementation strategy for enhanced ethanol production under stress.

2. Materials and methods

2.1. Bacterial strains and growth conditions

Z. mobilis ZM4mrr and its derivative mutants were cultured at 30 °C in Rich Media (RM) medium (20 g/L glucose, 10 g/L yeast extract, and 2 g/L KH₂PO₄) [21]. E. coli strains DH5α and BL21, used for molecular cloning and manipulation and for protein expression respectively, were cultured at 37 °C in Luria–Bertani (LB) medium at 37 °C with shaking. Where appropriate, spectinomycin was added to a final concentration of 100 μg/mL for both Z. mobilis and E. coli.

2.2. Plasmid construction

All plasmids were constructed using standard molecular cloning techniques. Self-targeting plasmids (pCut series) for endogenous CRISPR-Cas mediated DSB induction were assembled by cloning a 32-bp “Repeat-Spacer-Repeat” sequence, targeting specific genes (e.g., ZMO0346), into the pEZ15Asp vector under the control of the cas1 promoter, as previously described [21]. Genome editing plasmids (pGE series) were derived from their corresponding pCut plasmids by inserting a donor DNA fragment, which was generated by splicing by overlap extension PCR (SOE-PCR) and contained homology arms flanking the target site [21]. For protein expression, the wild-type (WT) recN and the recN-K35A mutant coding sequences were cloned into pET30a (+) vector for E. coli expression or cloned into pEZ15Asp plasmid under control of the constitutive ZMO0367 (glucose-6-phosphate dehydrogenase) gene promoter for Z. mobilis expression. All primers used are listed in Supplementary Table 1.

2.3. Generation of Z. mobilis mutants

The recN deletion and point mutant strains were generated via endogenous CRISPR-Cas genome editing. Briefly, the respective editing plasmid was electroporated into Z. mobilis ZM4mrr competent cells using a Bio-Rad Gene Pulser (1.6 kV, 200 Ω, 25 μF, 0.1 cm cuvette). Post-electroporation, cells were recovered in RM medium for 3 h at 30 °C before plating on RM agar [21]. Individual colonies were screened by colony PCR, and the mutations were verified by DNA sequencing. Due to the polyploid nature of Z. mobilis, we performed three rounds of colony purification and re-streaking to ensure segregation of all chromosome copies to the mutant allele. For the ΔrecN strain, complete loss of the recN gene was confirmed by PCR using primers flanking the deletion locus. For the RecN-K35A point mutant, the presence of the mutation and absence of WT sequence were confirmed by sequencing PCR products amplified from genomic DNA. To rule out secondary/compensatory mutations, we independently isolated and characterized two additional clones for each mutant strain, which exhibited phenotypes consistent with the primary isolates.

2.4. Protein expression and purification

The pET30a-based RecN expression plasmids were transformed into the E. coli BL21 (DE3) cells. Protein expression was auto-induced in ZYM-5052 medium for 4 days at 20 °C [23]. Cells were harvested by centrifugation, resuspended in Lysis Buffer (20 mM HEPES pH 7.5, 500 mM NaCl, 20 mM imidazole), and lysed using a French press. The clarified lysate was applied to a 1 mL HisTrap HP column (Cytiva), and the His-tagged protein was eluted with Elution Buffer (20 mM HEPES pH 7.5, 500 mM NaCl, 120 mM imidazole). Eluted fractions containing the target protein were pooled and concentrated for subsequent assays.

2.5. In vivo plasmid end-joining assay

The pEZ15Asp plasmid was linearized with FastDigest SalI, SmaI, or KpnI restriction enzymes (Thermo), according to the manufacturer's introduction. Complete digestion was confirmed by agarose gel electrophoresis, and the linear DNA fragments were gel-purified. A total of 100 ng of purified linear or circular (control) plasmid was electroporated into Z. mobilis ZM4mrr strains. Transformation efficiency was calculated as the number of colony-forming units (CFU) per μg of DNA on spectinomycin plates, with values representing the mean of three biological replicates. To control for differences in strain competence, we measured transformation efficiency with supercoiled pEZ15Asp plasmid for each strain and used these values to normalize linear plasmid transformation data where indicated.

2.6. In vitro DNA ligation assay

The substrate plasmid p4NcoI (a pBBr-derived plasmid containing four NcoI sites) 500 ng was digested with FastDigest NcoI (1 μl; Thermo) in 10 μl at 37 °C for 20 min. Complete digestion was confirmed by agarose gel electrophoresis showing four fragments of expected sizes. The digestion reaction was then supplemented with T4 DNA Ligase (5 U) in the presence or absence of purified RecN protein (15 μM) to 20 μl and incubated at 22 °C for 120 min. The reaction mixtures were transformed into E. coli DH5α, and the transformation efficiency was determined. Colonies were analyzed by PCR and sequencing to determine the pattern of re-ligated fragments. Control reactions lacking T4 DNA ligase were included to assess the contribution of residual undigested plasmid.

2.7. DNA annealing assay

5′-end FAM- or Cy5-labeled oligonucleotides (50 nM - 8 μM) in different lengths were used to incubate with their completely or partially complementary oligonucleotides (Supplementary Table 1) in the DNA annealing buffer (20 mM HEPES, 500 mM NaCl, pH = 7.5) on ice for 15 min. Then, RecN (2.5 μM) with or without ATP (1 mM) was added, and the reaction was incubated at 30 °C for 1 min. For competition assays, a 100-fold molar excess of unlabeled competitor oligonucleotide was added after the initial annealing step. Reactions were analyzed on 10% native polyacrylamide gels in 1x TBE buffer, and fluorescent signals were visualized using a Typhoon FLA 7000 imager (GE Healthcare).

2.8. ATPase assay

ATP (2 μM) and RecN or RecN-K35A (0.08 μM) with or without linear dsDNA (0.025 μM) was added as described previously [24], and the reaction (20 mM HEPES pH 7.5, 500 mM NaCl, 20 mM imidazole) was incubated at 30 °C for 1 min. Reactions were analyzed by ATP assay kit (Beyotime, S0026), and Luminescence signals (Relative Light Units) were visualized using a microplate reader (PerkinElmer, Victor Nivo 3S). Using the blank group as a baseline, the luminescence values of the experimental groups were subtracted to calculate ATP consumption and subsequently determine the ATPase activity rate. All assays were performed in triplicate.

2.9. Cell morphology and DNA content analysis

For cell size analysis, strains were grown to mid-log phase, stained with Carbol Fuchsin Staining Kit (OriLeaf R22015), and imaged using a Nikon Eclipse 80i microscope. Cell dimensions were measured from at least 100 cells per strain using ImageJ software. For DNA content quantification, cells were fixed, stained with DAPI (1 μg/mL) and the membrane dye FM4-64 (4 μM), and visualized by fluorescence microscopy. DAPI fluorescence intensity of individual cells was quantified from over 1000 cells per condition using a high-content screening system (Revvity, Opera Phenix).

2.10. Fermentation analysis

Fermentation was conducted in Rich Media (RM) with varying initial glucose (75 or 120 g/L) and KH₂PO₄ (2, 8, or 16 g/L) concentrations. Where indicated, p-methylbenzaldehyde was added to 0.5 g/L to simulate lignocellulosic hydrolysate inhibitors. Culture growth was monitored by measuring optical density at 600 nm (OD₆₀₀). Glucose and ethanol concentrations in 0.2 μm-filtered culture supernatants were quantified via High-Performance Liquid Chromatography (HPLC) using an Aminex HPX-87H column.

2.11. Statistical analysis

Data are presented as mean ± standard deviation from at least three independent experiments. Statistical significance was determined using two-tailed Student's t-test for pairwise comparisons or one-way ANOVA with Tukey's post-hoc test for multiple comparisons. P-values <0.05 were considered significant.The median value was calculated as the middle observation in the ranked dataset. Data are presented as median $\pm$ interquartile range (IQR). The median was determined by sorting all values in ascending order and selecting the value at the midpoint. For datasets with an even number of observations, the median was calculated as the average of the two middle values.

3. Results

3.1. RecN is essential for both MMEJ and HR pathways in Z. mobilis

To investigate the potential role of RecN in MMEJ, we leveraged this established endogenous type I-F CRISPR-Cas self-targeting system in Z. mobilis ZM4mrr [21,25]. In this system (Fig. 2A), electroporation of a plasmid (pCut) carrying a mini-CRISPR array targeting a specific genomic locus directs the endogenous CRISPR-Cas machinery to create a site-specific DSB. Repair of this DSB occurs primarily through MMEJ in Z. mobilis [21,25]. When a donor DNA fragment with homology to the target region is provided on the plasmid (pGE), repair can also occur through HR, enabling precise gene deletion [26].

Fig. 2.

RecN is essential for both HR and MMEJ pathways in Z. mobilis. A. Illustration of HDR (homology directed repair) and MMEJ-mediated repair of DSBs generated by the endogenous I-F CRISPR-Cas system. The pCut plasmid, carrying only an artificial mini-CRISPR targeting genomic DNA, and the pGE plasmid, containing an additional donor fragment for HDR mediated gene deletion. These plasmids were electroporated into Z. mobilis ZM4mrr cells and generated DSBs at specific genome loci, facilitating repair via MMEJ or HDR. B. Relative transformation efficiencies of pCut0346 and pGE0346 compared to the empty plasmid pEZ15Asp in WT and ΔrecN strains. C. Relative transformation efficiencies of pCut plasmids and pGE plasmids targeting ZMO0346, ZMO0675, ZMO1062, ZMO1314, and ZMO1934 genes in WT (red) and ΔrecN (blue) strains. Heatmaps show the efficiency ratio of WT to ΔrecN at the corresponding target sites on Z. mobilis genome. Error bars indicate standard deviations from three experiments. ∗, p < 0.05; ∗∗, p < 0.01; ∗∗∗, p < 0.001. D. The proportion of MMEJ (red) and HDR (blue) in WT and ΔrecN transformants was assessed across 23 clones (Fig. S1 A-D).

We first constructed a recN deletion mutant (ΔrecN) and confirmed complete loss of the recN gene by PCR and sequencing after three rounds of colony purification to ensure segregation of all chromosome copies. We assessed its capacity to repair DSBs generated by the self-targeting plasmid pCut0346, which directs cleavage to the ZMO0346 gene locus (Fig. 2A). Strikingly, the transformation efficiency of pCut0346 was nearly abolished in the ΔrecN strain compared to the WT (Fig. 2B). This defect was not locus-specific, as similar dramatic reductions were observed with self-targeting plasmids against five additional genomic sites (Fig. 2C). These results indicate that RecN is important for efficient repair of endogenous I-F CRISPR-induced DSBs, which we have previously shown to be repaired primarily by MMEJ [21].

We next tested the requirement for RecN in HR using a genome-editing plasmid (pGE0346), which provides a donor DNA template for precise deletion of ZMO0346 via HR. The transformation efficiency of pGE0346 was also severely compromised in the ΔrecN mutant (Fig. 2B). Quantification of repair efficiencies confirmed that both MMEJ and HR were abolished in the absence of recN gene, dropping from ∼10% in the WT to undetectable levels in the mutant (Fig. 2D). Thus, RecN functions as a central hub controlling both major DSB repair pathways in Z. mobilis, making it a potential target for modulating repair outcomes in Z. mobilis or other industrial strains.

3.2. RecN promotes DNA end-joining in vivo and in vitro

We further probed RecN's function using a plasmid re-circularization assay. Linearized pEZ15Asp plasmid with either cohesive (SalI, KpnI) or blunt (SmaI) ends was electroporated into WT and ΔrecN cells (Fig. 3A). The ΔrecN strain exhibited a severe defect in re-ligating all linearized plasmids, with transformation efficiencies reduced by 5 to 10-fold compared to the WT (Fig. 3B). All recovered plasmids from colonies had precise ligations at the restriction sites, confirming repair via simple end-joining.

Fig. 3.

RecN facilitates DNA end-joining in vivo and in vitro. A. Illustration of in vivo repair of restriction endonucleases-digested pEZ15Asp plasmid in WT or ΔrecN strain. The pEZ15Asp plasmid was digested by FastDigest SalI, SmaI or KpnI and transformed into Z. mobilis ZM4mrr strain afterwards and selected on antibiotic selection plates. Completely digestion was confirmed by agarose gel analysis. B. Relative transformation efficiencies of linearized plasmids with various distinct ends in WT and ΔrecN strains. Error bars indicate standard deviations from three independent experiments. ∗, p < 0.05. C. Illustration of in vivo repair of NocI-digested pBBr plasmid at multiple NcoI sites and in vitro ligated with T4 DNA ligase in presence or absence of RecN. Afterwards, the reaction mix was transformed into E. coli DH5α and selected on antibiotic selection plates. F and R: primers used amplify the ligated plasmids in single colonies. D. Transformation efficiency of the reaction mix containing NcoI, RecN and T4 DNA ligase treated pBBr plasmid. Solid and empty circles: presence or absence of NcoI, RecN, T4 DNA ligase or the pET30a plasmid. Error bars indicate standard deviations from three experiments. Letters on the columns indicate significant differences (p < 0.05). E. Agarose gel analysis of the PCR products amplifying the single colonies isolated on the plates from (D). The bands corresponded to the different ligated plasmids shown in (C).

To determine if RecN acts directly on DNA ends, we established an in vitro ligation assay using a modified plasmid, p4NcoI, which contains four NcoI sites (Fig. 3C). Digestion produces four fragments, and their re-ligation by T4 DNA ligase can generate multiple plasmid species. The presence of RecN protein significantly increased the transformation efficiency of the ligation reaction mixture by several orders of magnitude compared to reactions lacking RecN (Fig. 3D). Moreover, sequence analysis of the resulting plasmids revealed a stark contrast in ligation complexity: RecN strongly biased the reaction towards the incorporation of all three insert fragments (83.3% of plasmids with RecN vs. 54.2% without) (Fig. 3E). Crucially, DNA sequencing confirmed that these RecN-assisted ligations accurately restored the original plasmid sequence at the junctions. This demonstrates that RecN not only enhances the overall efficiency of end-joining but also directly facilitates the precise pairing and co-ligation of multiple DNA ends in vitro, promoting the accurate reconstruction of complex molecular substrates.

3.3. RecN stimulates DNA annealing in an ATP-dependent manner

Given its role in end-joining, we hypothesized that RecN might promote the annealing of microhomologous sequences. We incubated complementary 59-nt oligonucleotides with increasing concentrations of RecN (0-7.5 μM) and observed a clear dose-dependent increase in dsDNA formation by native PAGE analysis (Fig. 4A and B). This annealing activity exhibited distinct length dependence - while RecN effectively facilitated the annealing of shorter 29-nt oligonucleotides, it showed no detectable activity with 20-nt fragments under identical conditions in vitro (Fig. 4C and D). Further characterization revealed that RecN could promote annealing even with oligonucleotides containing terminal mismatches or overhangs (Fig. 4E), demonstrating its flexibility in handling various substrate configurations.

Fig. 4.

RecN facilitates DNA annealing in vitro. A. Illustration of FAM-labeled single-stranded DNA (ssDNA) annealing with complementary ssDNA to form FAM-labeled double-stranded DNA (dsDNA). B and C. Non-denaturing PAGE analysis of a 59 nt (B) or 29 nt (C) FAM-labeled ssDNA with its complementary non-labeled ssDNA in presence or absence of RecN. D. Illustration of the effect of ATP on RecN-mediated ssDNA annealing. Firstly, RecN-mediated annealing of FAM-labeled ssDNA (F) and Cy5-labeled complementary ssDNA (R) to form double-stranded DNA (dsDNA) was carried out in presence or absence of ATP. Then, 10-folds more unlabeled ssDNA was added to compete the labeled complementary ssDNA. E. Non-denaturing PAGE analysis of the competition of unlabeled ssDNA (R) in the RecN-mediated annealing reactions as shown in (D). F,G. Non-denaturing PAGE analysis of RecN-mediated annealing of short ssDNAs with blunt or 3’/5′ overhangs (F) or mistch(G). The length of the ssDNA fragments was indicated; + or – indicated presence or absence of RecN protein in the reaction.

Most importantly, we found that ATP profoundly enhanced RecN's annealing capability. In a competitive assay where annealing reactions were challenged with a 100-fold molar excess of unlabeled competitor DNA, the addition of 1 mM ATP significantly increased the amount of stable labeled dsDNA formed by RecN (Fig. 4F and G). This ATP-dependent enhancement was consistently observed across a wide range of substrate concentrations (50 nM to 8 μM), indicating that ATP enables RecN to actively and processively drive the annealing of complementary strands.

3.4. ATPase activity of RecN is essential for MMEJ in Z. mobilis

RecN belongs to the SMC protein family and requires functional nucleotide-binding domains for its activity. Structural modeling, using the E. coli RecN structure as template, identified Lys35 (K35) in the Walker A motif as essential for ATP catalyzing [27](Fig. 5A and B). Consistent with this, the purified RecN-K35A mutant protein was completely defective in ATP-stimulated DNA annealing in vitro (Fig. 5D). Recombinant Z. mobilis RecN showed ATPase activity with a specific activity regardless of presence of absence of dsDNA (Fig. 5E). The RecN-K35A mutant showed significantly reduced ATPase activity (Fig. 5E), especially in the presence of linear dsDNA activating RecN to form multimers that encapsulate ATP, as described previously [24]. When introduced into the Z. mobilis chromosome confirmed by sequencing after segregation, the RecN-K35A mutation recapitulated the severe end-joining deficiency of the ΔrecN null mutant across all tested linearized plasmids with sticky or blunt end (Fig. 5F).

Fig. 5.

ATPase activity of RecN is essential for MMEJ. A. Schematic representation of the RecN protein. Walker A and Walker B motif were indicated. Orange and cyan indicated the head domain, gray indicated the coiled coil domain. Multiple Sequence Alignment of RecN Walker A motif from different bacterial species (Fig. S2). B. Quasi-Atomic Model of RecN. The red indicated the mutation in the enlarged box. C. Bar chart showing ATPase activity of WT (gray dot) versus K35A mutant RecN (light gray dot), with and without dsDNA. Error bars indicate standard deviations from three independent experiments. ∗∗∗∗, p < 0.001,∗∗∗, p < 0.001,∗, p < 0.05. D. Non-denaturing PAGE analysis of a FAM-labeled ssDNA and a Cy5-labeled complementary ssDNA in presence RecN WT or mutant (K35A). E. Relative transformation efficiencies of linearized plasmids with various distinct ends in WT and RecN K35A strains. Error bars indicate standard deviations from three independent experiments. ∗, p < 0.05. F. Relative transformation efficiencies of pCut0346 and pGE0346 compared to the empty plasmid pEZ15Asp in WT and RecN K35A strains. G. The proportion of MMEJ and HR in WT and RecN K35A transformants carried pCut0346 or pGE0346 plasmids. Totally 23 clones for each transformation were used to measure MMEJ and HR efficiency. Error bars indicate standard deviations from three experiments (Fig. S1 A-F).

Strikingly, pathway-specific analysis revealed a clear functional separation. While transformation efficiency of the MMEJ-dependent self-targeting plasmid pCut0346 was drastically reduced in the RecN-K35A mutant, the HR-dependent genome-editing plasmid pGE0346 showed transformation efficiency comparable to WT levels (Fig. 5F). Consequently, MMEJ efficiency was specifically abolished in the mutant, whereas HR remained unaffected (Fig. 5G). This separation-of-function demonstrates that MMEJ requires a higher threshold of RecN ATPase activity than HR, establishing RecN as a tunable molecular switch governing repair pathway choice. The RecN-K35A allele thus provides a valuable tool for constructing strains with enhanced HR fidelity by eliminating MMEJ-mediated mutations while maintaining overall repair capacity.

3.5. MMEJ contributes to the maintenance of normal cell morphology under stress conditions

To assess the physiological importance of MMEJ, we examined the consequences of its specific disruption. Quantitative analysis revealed that the RecN-K35A mutant, defective specifically in MMEJ but proficient in HR, exhibited a remarkable cell elongation phenotype, with over 15.79% of cells showing at least a 2-fold increase in length compared to WT under normal growth conditions (Fig. 6A and B). This morphological defect, characteristic of cell division impairment potentially associated with unresolved DNA damage, suggests that the HR pathway alone may be insufficient to maintain normal cell morphology under standard laboratory conditions. Treatment with the DNA-crosslinking agent mitomycin C (MMC) further exacerbated this phenotype, with both the RecN-K35A and ΔrecN mutants displaying significantly greater cell elongation compared to WT cells at the MMC concentration tested (160 ng/mL) (Fig. 6B). Similarly, compared with untreated WT (Median = 1.462 μm, IQR = 1.236 - 1.779 μm), 10% ethanol stress intensified the elongation phenotype in both mutant strains (cell length: RecN-K35A: Median = 3.596 μm, IQR = 2.648 - 5.536 μm; ΔrecN: Median = 3.995 μm, IQR = 2.641 - 5.689 μm; WT: Median = 3.030 μm, IQR = 2.173 - 4.211 μm; OE recN: Median = 2.847 μm, IQR = 1.867 - 4.325 μm) (Fig. 6B). These results indicate that RecN contributes to the maintenance of normal cell morphology (probably through maintaining genomic integrity) under both standard and stress conditions in Z. mobilis [28].

Fig. 6.

Elevated DNA concentration enhances cellular stress tolerance. A. Fluorescence microscopy analysis of exponentially growing WT (gray), recN-K35A (green), ΔrecN (yellow), and OErecN (red) cells after 24 H exposure to 160 ng/ml MMC, 10% ethanol, or 10% ethanol + 16 g/L KH₂PO₄. OErecN strain carried a pEZ15Asp vector cloned with recN constitutive expression cassette, and other strains carried an empty pEZ15Asp vector. Cells were stained with DAPI (nucleoid) and FM4-64 (membrane), and imaged to assess nucleoid morphology and cell length.White scale bar = 1 μm. B. Cell length measurements were obtained from micrographs of fixed, fuchsin-stained cells, with each group containing data from >120 cells. C. DNA density represented by mean gray value of DAPI spots by high-content imaging. n > 500 spots from >20 Fields of View per group. D, E. Fermentation profiles of the strains of Zymomonas mobilis in medium containing 2 g/L (black), 8 g/L (green), 16 g/L (red) KH₂PO₄ controlled at 120 g/L glucose or 75 g/L glucose with 0.5 g/L p-Methyl benzaldehyde, and 30 °C under fermentative conditions: square symbols for glucose conditions, dot symbols for ethanol conditions. Symbols are mean values from three independent experiments.

3.6. RecN-mediated DNA repair capacity correlates with stress tolerance

Building on the observation that MMEJ deficiency causes genomic instability, we asked whether cellular DNA content correlates with stress resilience. RecN overexpression strain (OErecN) was constructed based on a multicopy plasmid pEZ15Asp with the constitutive glucose-6-phosphate dehydrogenase gene (ZMO0367) promoter, and other strains (WT, ΔrecN and RecN-K35A mutant) were transformed with an empty pEZ15Asp vector to eliminate the impact caused by plasmid DNA when study the intracellular DNA density. Strikingly, RecN-overexpressing cells exhibited a 1.42-fold increase in intracellular DNA density compared to WT under normal growth conditions, as quantified by DAPI staining (Fig. S3). This elevated DNA content was associated with markedly improved ethanol tolerance, and more normal cell morphology under 10% ethanol stress (Fig. 6A and B). We reasoned that increased genomic DNA content might provide redundant templates for damage tolerance, thereby facilitating more efficient DNA repair.

Guided by this correlation, we developed a simple nutritional engineering strategy to enhance genome maintenance capacity by increasing intracellular DNA content through phosphorus supplementation. Elevating KH₂PO₄ concentration to 8-16 g/L (versus standard 2 g/L) effectively increased cellular DNA density across all strains except recN overexpression strain (Fig. 6C) and substantially suppressed ethanol stress-induced cell elongation (Fig. 6A and B), supporting the premise that enhanced DNA biosynthesis compensates for repair deficiencies by providing multiple genomic copies.

This insight translated directly into improved fermentation performance. In fermentation with high glucose concentration (120 g/L), increasing KH₂PO₄ concentration to 8 g/L significantly improved glucose consumption and elevated final ethanol titer by approximately 13.33% (Fig. 6D). Most notably, under simulated inhibitory conditions resembling lignocellulosic hydrolysates, the high-phosphorus strategy (16 g/L KH₂PO₄) enabled nearly complete glucose utilization and achieved a 2.95-fold increase in ethanol production from 11.83 g/L to 34.88 g/L (Fig. 6E). These results demonstrate that nutritional enhancement of genome maintenance capacity, informed by RecN biology, provides a simple and scalable strategy for improving industrial bioethanol fermentation.

4. Discussion

While the eukaryotic MMEJ pathway is well characterized, its bacterial counterpart remains poorly understood. Here we demonstrate that the SMC-like protein RecN is an essential, ATP-dependent regulator of MMEJ in Z. mobilis. The RecN-K35A mutation, which severely impairs ATP hydrolysis, specifically disrupts MMEJ without affecting HR, establishing that MMEJ requires a higher threshold of RecN ATPase activity than HR. The severe genomic instability in recN-K35A mutants demonstrates that MMEJ functions not simply as a backup for HR but as a fundamental pathway for genome maintenance.

The core protein machinery driving bacterial MMEJ represents a unique evolutionary adaptation, distinct from the eukaryotic paradigm. While eukaryotic MMEJ employs the MRN complex with CtIP for resection initiation [29] and typically utilizes longer microhomologies (5-25 bp) [30], the bacterial pathway employs a different strategy. In bacteria such as E. coli, LigA serves as the primary ligase [31,32], while RecN provides critical ATP-dependent regulation. The RecBCD complex may perform initial end processing, and bacteria appear to operate with shorter microhomology tracts (1-10 bp) compared to eukaryotes [21]. The absence of Polθ and recruitment of an SMC-family protein highlight potential differences in the mechanistic strategies employed by bacterial MMEJ compared to its eukaryotic counterpart. Whether other bacterial factors, such as DNA polymerases or replisome components, participate in MMEJ remains an open question for future investigation.

The balance between MMEJ and HR in bacteria is likely influenced by multiple factors. Template availability is one established determinant-the presence of homologous donor DNA suppresses MMEJ in favor of HR [21]. Cellular energy status could potentially represent an additional layer of regulation, given that RecN's ATPase activity strongly contributes to MMEJ-associated repair in this assays. However, this remains speculative, as both HR and MMEJ involve ATP-dependent enzymes (RecA, AddAB, LigA), and severe ATP depletion would likely compromise all repair pathways. This finding that RecN-K35A retains some ATPase activity (∼25%) in vitro suggests partial inactivation, yet this mutant shows clear separation-of-function in plasmid assays. This could indicate that a threshold of ATPase activity is required for MMEJ but not HR, or that the mutation differentially affects interactions with pathway-specific partners. The observation that RecN-K35A protein is functional in HR pathway in vivo supports the interpretation that its defects are due to impaired function rather than protein loss.

Our work provides a framework for integrating fundamental repair biology with bioprocess design. The RecN-deficient strains we characterized offer an unexpected application: in polyploid industrial microorganisms, where multiple genome copies can mask or repair induced mutations [33,34], disabling RecN could facilitate directed evolution by preventing repair and allowing mutations to accumulate. This strategy could accelerate strain improvement programs that rely on adaptive evolution. Moreover, we translated DNA repair insights into practical application. Leveraging the RecN-stress tolerance correlation, we developed a high-phosphorus strategy that enhanced ethanol fermentation under inhibition, establishing genome maintenance support as a novel bioprocessing strategy.

In conclusion, this work identifies RecN as the core ATP-dependent effector of bacterial MMEJ and establishes a direct link between repair pathway choice, genome stability, and fermentation performance. Future studies should define the signals governing RecN's functional switching between repair pathways and characterize the complete MMEJ complex.

Author contributions

S.P. and N.P. conceived the study. J.H., and X.W. conducted the experiments. S.P., J.H., X.W. and N.P. participated in the discussion and analysis of the results. N.P. acquired the funding, supervised the work. S.P. and N.P. wrote the manuscript. All authors contributed to review and editing.

Funding

This work was supported by grants from the Hubei Natural Science Foundation (2024AFD125), the Hubei Key Technology R&D Program (2025BCA006), the Hubei Special Project for Science Development (No. 2024CSA060), the Major Special Project for the Development of Agricultural Microbial Industry (NYWSWZX2025-2027-10). The open access charge was funded by the Hubei Special Project for Science Development (No. 2024CSA060).

Declaration of competing interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Appendix A. Supplementary data

The following is the Supplementary data to this article:

Data availability

Plasmids, strains, and other unique reagents generated in this study are available upon request.