Bacteria often use overlapping or redundant systems to sustain essential pathways such as PG synthesis. To identify genes with significant roles in elongasome activity in B. subtilis, we constructed a strain (designated Δ4) lacking all four class A PBPs (aPBPs), and which therefore relies solely on the elongasome for PG synthesis during cell elongation (McPherson and Popham, 2003). A Tn-Seq approach was employed to identify genes essential in the Δ4 strain but not in the wild-type (WT) background. We identified the ecsAB operon as having numerous mariner transposon insertions in WT, but very few in the Δ4 strain (Figure 1—figure supplement 1). We verified conditional essentiality of ecsA by determining the plating efficiency of a clean, unmarked deletion mutant (ΔecsA) in a ponA depletion background in the presence and absence of the genes encoding the other 3 aPBPs (pbpD, pbpF, pbpG). Interestingly, ecsA was not only essential in the Δ4 background but also with depletion of ponA alone (Figure 1A). Mutations that impair PG synthesis can often be rescued by growth on plates amended with 20 mM MgSO₄, which leads to decreased activity of autolysins and thereby helps restore balance between PG synthesis and degradation pathways (Formstone and Errington, 2005). Indeed, an ΔecsAΔponA mutant was viable when streaked on high Mg plates, and growth was Mg-dependent (Figure 1B).

EcsA has been designated as part of an ABC-type transporter involved in the expression and secretion of proteins (Leskelä et al., 1999). Deletion of ecsA has a profound effect on the intramembrane protease RasP, with similar phenotypes noted for the ecsA and rasP deletion mutants (Heinrich et al., 2008). Consequently, we tested whether the essential role of EcsA in the ΔponA strain was due to RasP. Indeed, viability of ΔrasPΔponA, like ΔecsAΔponA, depended on high Mg concentrations (Figure 1B). The above data highlight the importance of the EcsAB-RasP pathway in maintaining viability in the absence of aPBPs.

Upregulation of elongasome activity is known to alleviate aPBP defects (Meeske et al., 2016). Based on the observed conditional essentiality, we hypothesized that the EcsAB-RasP pathway might functionally compensate for the absence of aPBPs. As a first test of this hypothesis, we measured sensitivity to moenomycin (MOE), a specific inhibitor of aPBP-associated TG activity (Van Heijenoort et al., 1978; Chen et al., 2019). Indeed, ecsA and rasP mutants were MOE sensitive with a four-fold decrease in minimum inhibitory concentration (MIC) relative to WT (Table 1). This was not due to a general growth defect: ecsA and rasP single mutants grew as well as WT in the absence of MOE, albeit with some lysis in stationary phase (Figure 2A), consistent with previous observations (Heinrich et al., 2008). This antibiotic sensitivity could be complemented by ectopic expression of ecsAB or rasP, respectively (Figure 2—figure supplement 1). Moreover, ΔecsAΔrasP had a similar MOE sensitivity as ΔrasP (Figure 2A), suggesting that the synthetic lethality of ecsA with ponA is mediated through its known downstream effect on the activity of RasP (Heinrich et al., 2008). In contrast to MOE, the ΔrasP and ΔponA mutants had a similar sensitivity as WT when tested for sensitivity to antibiotics that act on substrates common to both the elongasome and aPBP-dependent pathways of PG synthesis. For example, both nisin (Wiedemann et al., 2001) and vancomycin (Watanakunakorn, 1984) bind the common lipid II intermediate (Figure 2—figure supplement 1). Together, these results suggest that the EcsAB-RasP pathway is critical when aPBPs are compromised, but not as a general response to inhibition of PG synthesis.

We next sought to test antibiotics that, unlike MOE, inhibit aPBPs at their TP active site. We reasoned that a stress response important for elongasome activity should also provide resistance to antibiotics that inhibit aPBPs, assuming they do not also interfere with the bPBPs essential for the elongasome. We tested 4 β-lactams (cefuroxime, oxacillin, ampicillin and penicillin G) for their inhibition profiles against ΔrasP and ΔponA strains. Oxacillin and cefuroxime (CEF) were previously suggested to preferentially inhibit aPBPs (Sassine et al., 2017; Sharifzadeh et al., 2020), whereas penicillin G preferentially inhibits bPBPs (Sassine et al., 2017). Consistently, oxacillin and CEF had highest activity against ∆rasP, whereas penicillin G and ampicillin had the highest activity against ∆ponA, which encodes the major aPBP, PBP1 (Figure 2B). These results support the idea that the EcsAB-RasP pathway functionally compensates either for the absence of aPBPs or for their chemical inhibition at either the TG (MOE) or TP (CEF) active sites.

Interestingly, the ∆ponA mutant was actually more CEF resistant than WT. Thus, PBP1 inactivated by CEF may be deleterious to the cell. This is suggestive of futile cycling, a process in which inactivation of the TP active site leads to an ongoing generation and degradation of uncrosslinked PG strands driven by the aPBP-associated TG (Cho et al., 2014; Waxman et al., 1980). To explore this idea further, we treated WT cells with sub-inhibitory concentrations of two drugs simultaneously, MOE and CEF, that inhibit the two different active sites of the aPBP proteins. If CEF results in futile cycling, we reasoned that MOE might antagonize this effect. In contrast, MOE and CEF together resulted in synergistic inhibition (Figure 2—figure supplement 2). This is consistent with the same target drug synergy model, as previously described for E. coli protein synthesis inhibitors (Yilancioglu, 2019) and drugs used to treat human diseases (Jia et al., 2009), but does not support the hypothesis of CEF-dependent futile cycling.

RasP functions as an intramembrane protease for the activation of multiple stress response pathways, and our results suggest it may be important for PG synthesis when aPBPs are missing or inhibited. RasP proteolytically inactivates the anti-sigma factors RsiW (regulator of σ^W) (Schöbel et al., 2004), RsiV (regulator of σ^V) (Hastie et al., 2013) and RsgI (regulator of σ^I) (Liu et al., 2017). In the absence of RasP, these σ factors can not be activated. RasP also cleaves FtsL, a cell division protein (Bramkamp et al., 2006). To determine which of these RasP targets may contribute to elongasome activity, we took advantage of the fact that MOE and CEF selectively inactivate aPBPs. Therefore, MOE and CEF resistance provides a readout of elongasome function. We tested mutants lacking each of the three RasP-dependent sigma factors or containing Δ25FtsL, coding for a functional, but truncated FtsL (deleted in amino acids 2–26) variant that is not subject to cleavage by RasP (Bramkamp et al., 2006). The ΔecsA and ΔrasP mutants were 4-fold more sensitive to MOE than WT (0.4 vs. 1.6 µg/mL), whereas for ΔponA the (MIC) was >1.6 µg/mL (Table 1; Figure 3—figure supplement 1). The MIC was unaffected by deletion of sigW or sigV or by the non-cleavable FtsL (1.6 µg/mL). However, the ΔsigI mutant was significantly more sensitive to MOE with the MIC being 0.4 µg/mL, similar to ΔrasP. This suggests that σ^I is required for optimal function of the MOE-insensitive elongasome.

Similar results were observed when CEF sensitivity was monitored (Figure 3A). Of the known RasP targets, σ^I contributes the most to CEF resistance. Moreover, the ΔsigWΔsigI mutant phenocopies the ΔrasP mutant, suggesting that activation of σ^I and σ^W largely accounts for the role of RasP in CEF resistance. In addition, the sensitivity of the ΔecsA and ΔrasP mutants was not further increased by mutation of sigW or sigI (Figure 3—figure supplement 2), indicative of them being in the same pathway. Finally, deletion of rsgI, encoding the σ^I anti-sigma factor, led to a significant decrease in CEF sensitivity of the ΔecsA and ΔrasP mutants. ΔrsgI was more sensitive to CEF compared to WT, which may be due to increased activity of σ^I and its associated autolysins. In contrast, deletion of rsiW, encoding the σ^W anti-sigma factor, led to a much less pronounced effect (Figure 3—figure supplement 2). Thus, σ^I plays a dominant role in intrinsic CEF resistance, and as expected this activity relies on the RasP-dependent degradation of the RsgI anti-sigma factor.

The importance of σ^I in the absence of aPBPs was confirmed by determining the plating efficiency of ΔsigIΔponA double mutant (Figure 3B). The double mutant could survive with high Mg²⁺, but was unable to grow on LB. This synthetic lethality of the ΔsigIΔponA and ΔrasPΔponA strains was suppressed by ectopically expressing the sigI gene from the leaky promoter P_spac(hy). Thus, decreased σ^I activity can fully explain the ∆rasP antibiotic sensitivity phenotypes, and we therefore conclude that one or more members of the σ^I regulon must facilitate growth under conditions of impaired aPBP activity.

Next, we sought to identify the σ^I-dependent genes important for survival in the absence of aPBPs. Of the genes directly regulated by σ^I (Ramaniuk et al., 2018), five (mreBH, lytE, gsiB, fabI and bcrC) have known or likely roles related to cell envelope functions. GsiB is a general stress response protein (Michna et al., 2016) and FabI is involved in fatty acid synthesis (Heath et al., 2000). BcrC functions in undecaprenylpyrophosphate recycling (Bernard et al., 2005; Zhao et al., 2016; Radeck et al., 2017b), and MreBH and LytE are both elongasome-associated proteins. MreBH, one of three MreB-family proteins that associate with the elongasome, sequesters and directs the LytE endopeptidase to the sites of insertion of new peptidoglycan (Carballido-López et al., 2006). To further define the role of σ^I in sustaining viability during aPBP inhibition, we conducted CEF/MOE sensitivity assays using single mutants of σ^I-controlled genes. The mreBH, lytE and bcrC single mutants exhibited slightly higher sensitivity for both CEF and MOE (Figure 4—figure supplement 1), however, they did not entirely phenocopy the sigI phenotype. The ΔmreBHΔlytE double mutant exhibited the same level of CEF and MOE sensitivity as both the rasP and sigI mutants (Figure 4A–B). Thus, these results suggest that the EcsAB-RasP-σ^I pathway primarily acts through MreBH and LytE to control elongasome function.

To further validate the importance of MreBH and LytE, we created deletion mutants in the ΔponA background (Figure 4C). A ΔmreBHΔponA double mutant could be constructed only when the cells were initially plated on LB supplemented with high Mg²⁺. Once constructed, however, this mutant and the ΔlytEΔponA double mutant did not exhibit a plating defect on LB. In contrast, the triple mutant of ΔmreBHΔlytEΔponA was synthetic lethal and could not be plated on LB agar without Mg²⁺ supplementation. These data suggest an additive role for MreBH and LytE in the effective functioning of the elongasome, likely due to the ability of LytE to retain some function in the absence of MreBH, and MreBH having functional roles beyond localization of LytE.

B. subtilis has two partially redundant D,L-endopeptidases, LytE and CwlO, which are collectively essential for cell viability (Hashimoto et al., 2012). The involvement of σ^I in the expression of lytE has already been established since both ΔlytEΔcwlO and ΔsigIΔcwlO are synthetic lethal (Salzberg et al., 2013). Consistently, ΔrasPΔcwlO was also synthetic lethal (Figure 4—figure supplement 1). To confirm that LytE activity was reduced in the rasP and sigI mutants we evaluated the autolytic potential of the cells. Cells were treated with sodium azide, which disrupts membrane potential and activates autolysins (Jolliffe et al., 1981; Wang et al., 2014). By monitoring the time taken for a 50% reduction in optical density, we found that the ΔlytE mutant had a lower rate of autolysis (Figure 4D). Similar to ΔlytE, we observed that ΔrasP, ΔsigI and ΔmreBH also had lower autolytic potential, consistent with a role in affecting LytE expression or activity.

Next, we evaluated the expression levels of mreBH and lytE in ΔrasP, ΔsigI and ΔponA mutants (Figure 4E). In the ΔponA mutant, mreBH was significantly upregulated, whereas lytE was unchanged. In ΔsigI, both mreBH and lytE expression was significantly lower. This suggests that ΔponA cells require higher levels of MreBH to direct the autolytic activity of LytE to support optimal elongasome function, and that activation of σ^I mediates increased mreBH expression. As a result, the reduced expression of mreBH in ΔrasP and ΔsigI strains likely contributes to the synthetic lethality with ΔponA.

We complemented the conditional essentiality of mreBH and lytE by ectopically expressing each of these genes individually as well as in combination in different mutant backgrounds. These strains were used to evaluate the relative importance of each gene upon inhibition of PBP1 by monitoring their CEF resistance. Although ectopic expression of mreBH complements the CEF sensitivity of ΔmreBH, it is unable to restore CEF resistance to the ΔmreBHΔlytE double mutant (Figure 5A). However, when both mreBH and lytE were ectopically expressed, the strain was significantly more CEF resistant than WT (Figure 5A). Similarly, induction of mreBH modestly increased CEF resistance of ΔrasP (Figure 5B), but not a ΔrasPΔlytE double mutant. Similar results were obtained in cells where pbpD, pbpF and pbpG were deleted (data not shown) indicating no indirect effect of MreBH on these aPBPs. In ΔsigI, however, mreBH expression alone had no significant impact on CEF resistance, perhaps due to reduced availability of LytE. Thus, increasing MreBH levels likely functions to increase elongasome activity by facilitating the localized action of LytE. Conversely, the P_spac(hy)lytE overexpression construct could not be introduced into the ΔrasP and ΔsigI mutants. We speculate that high LytE, in cells that have reduced expression of mreBH, leads to delocalized and unregulated autolysin activity. Collectively, these results further support a model in which a major role of MreBH is in directing LytE to sites of ongoing, elongasome-dependent PG synthesis.

The elongasome is critical for the maintenance of rod-shape, as judged by the spherical morphology of conditional mutants that are depleted for either the RodA transglycosylase or the two class B PBPs that provide transpeptidase activity (Boylan and Mendelson, 1969; Wei et al., 2003). The maintenance of rod shape is also affected by the balance between the directional motion of the elongasome and the random diffusive motion of PBP1 (Dion et al., 2019). Any imbalance in the activities of the two systems can lead to change in cell morphology. Overexpression of MreB or other elongasome proteins leads to cells that are longer and thinner, whereas overexpression of PBP1 leads to shorter and wider cells (Dion et al., 2019). Thus, we hypothesized that the effects of the σ^I regulatory system (acting through mreBH and lytE) on elongasome function would be revealed by monitoring cell morphology. We imaged WT, ΔrasP, ΔsigI, ΔmreBH, ΔlytE, ΔmreBHΔlytE and ΔponA cells and quantified the cell length and width using MicrobeJ (Ducret et al., 2016). Indeed, ΔrasP, ΔsigI and ΔmreBHΔlytE mutants were significantly shorter (Figure 5C) and wider (Figure 5D) compared to the WT, which indicates that these cells were primarily utilizing PBP1 for PG synthesis. ΔmreBH and ΔlytE mutants individually also had lower elongasome activity. In contrast, the ΔponA mutant formed significantly thinner cells due to PG synthesis being contributed mainly by the elongasome. These data support the conclusion that the rasP, sigI and mreBH-lytE genes all support elongasome function.

Next, we took advantage of the ΔrasPΔponA synthetic lethality to isolate suppressors that grow on LB agar plates. Using whole-genome resequencing, we identified three strains with point mutations in walK (Ala241Asp, Ser385Leu, Asp274Ala). WalK is the sensor kinase of the essential two-component system WalKR, which regulates cell wall metabolism (Takada and Yoshikawa, 2018). WalR has binding sites upstream of sigI, mreBH and lytE and activates expression of these genes under heat stress (Huang et al., 2013). In addition to their regulation by σ^I, sigI and lytE also have σ^A-dependent promoters. WalR may function in conjunction with the σ^A holoenzyme, which together with σ^I controls lytE expression (Tseng et al., 2011). Taking into account the importance of WalKR in the expression of sigI, mreBH and lytE, we characterized one of the suppressor mutants of WalK, wherein aspartate 274 is changed to alanine (D274A).

Residue 274 lies in the cytoplasmic Per-Arnt-Sim (PAS) domain of WalK (Figure 6A). PAS domains have been linked to signal sensing (Taylor and Zhulin, 1999) and may be involved in protein dimerization (Huang et al., 1993). Recently, the cytoplasmic PAS domain of S. aureus WalK was found to bind zinc at a site including D274. Moreover, mutation in this binding site, which is highly conserved in WalK orthologs (Monk et al., 2019), led to increased kinase activity. We therefore hypothesized that the WalK^D274A suppressor (denoted as WalK*) led to increased activity of the WalKR two-component system. We used CRISPR mutagenesis to introduce the walK* allele into WT cells and then confirmed that this allele suppressed the synthetic lethality of ΔrasPΔponA (Figure 6B).

We next aimed to test the effect of WalK^* on gene expression and cell wall homeostasis. The sigI and lytE genes can be expressed through their σ^A promoter after activation by WalR (Salzberg et al., 2013; Tseng et al., 2011). However, mreBH lacks an annotated σ^A promoter, implying that the expression of mreBH may rely on WalR activation of the σ^I holoenzyme. To test this hypothesis, we measured CEF sensitivity of walK*ΔrasP and walK*ΔsigI strains (Figure 6C). Although walK* increased CEF resistance of the ΔrasP strain, it could not rescue the ΔsigI strain. This supports the idea that WalR may act in conjunction with σ^I to activate transcription of mreBH, and thereby augment elongasome activity. Increased activation of WalK* can lead to increased expression of not only lytE, but also cwlO (Takada and Yoshikawa, 2018). This could lead to elevated autolysin levels that might account for the higher CEF sensitivity of walK* alone compared to WT.

We further quantified the mRNA levels of mreBH and lytE in the walK* strain and in the walK*ΔrasP and walK*ΔsigI strains (Figure 6D). The walK* allele led to increased expression of both mreBH and lytE. Moreover, these levels were similar to that observed in the ΔponA background, suggesting that deletion of ponA leads to a compensatory increase in mreBH and lytE mediated by the WalKR. However, they were lower for the walK*ΔsigI strain. These data suggest that walK* leads to increased activation of WalR, which then leads to increased transcription of sigI and thereby of mreBH and lytE. This ultimately leads to the survival of the ΔrasPΔponA strain. These data also validate the importance of RasP and σ^I in the regulation of MreBH and LytE and their significant impact on elongasome activity, especially in the ΔponA background.

While our results suggest a critical role for σ^I in aPBP-elongasome homeostasis through its regulation of MreBH and LytE, previous studies have indicated that the extracytoplasmic (ECF) sigma factor σ^M also plays a significant role in B. subtilis cell wall homeostasis. σ^M regulates the expression of rodA, mreB, mreC and mreD (core components of the elongasome), as well as ponA and other genes involved in PG synthesis (Eiamphungporn and Helmann, 2008; Luo and Helmann, 2012). To determine the relative contribution of σ^M to cell survival during aPBP inhibition, we used P_M* mutations that selectively inactivate σ^M-dependent promoters of genes encoding elongasome components. We constructed the P_M*rodA and P_M*ponA strains that are unable to upregulate rodA and ponA, respectively, and a P_M*maf strain that cannot upregulate the mreBCD genes located downstream of the intragenic P_M inside maf (Eiamphungporn and Helmann, 2008). We also constructed the double mutant P_M*rodA P_M*maf strain. The CEF sensitivity of P_M*rodA and P_M*rodA-P_M*maf was similar to that of the sigM mutant (Figure 7A). Neither P_M*maf nor P_M*ponA were CEF sensitive. Thus, under conditions where CEF has inhibited PBP1, σ^M helps restore peptidoglycan synthesis primarily by increasing the expression of rodA to increase elongasome activity. In contrast, the double mutants of ΔecsAΔsigM, ΔrasPΔsigM and ΔsigIΔsigM revealed an additive effect with respect to CEF sensitivity (Figure 7B). Thus, the role of the elongasome in PG synthesis can be regulated through two-independent pathways: the EcsAB-RasP-σ^I pathway acts by regulating MreBH and LytE, and the σ^M pathway acts through RodA.

All stains were grown in lysogeny broth (LB) medium at 37°C. Liquid cultures were aerated on an orbital shaker at 300 rpm. Glycerol stocks were streaked on LB agar plates and incubated overnight at 37°C. Conditionally synthetic lethal strains were grown in LB medium with 20 mM MgSO₄.

Bacterial strains used in this study have been listed in the Key Resources Table. For all deletion mutants, primary strains were ordered from the BKK/BKE collection available at the Bacillus Genetic Stock Centre (BGSC) (Koo et al., 2017). These gene deletions with the antibiotic cassette (kanamycin or erythromycin) were then transformed into our WT 168 strain using natural competence induced in modified competence (MC) medium. rasP, ecsA and ponA deletion strains had very low natural competence. Thus, other mutations were introduced using SPP1 phage transduction as described (Kearns et al., 2005). The null mutants were constructed using pDR244, which removes the resistance cassette leading to clean in-frame deletions (Koo et al., 2017). The resulting gene deletions (designated Δ) were confirmed with check primers listed in Supplementary file 1.

Genes were ectopically expressed at amyE under promoter P_spac(hy) using pPL82 plasmid (Quisel et al., 2001). MreBH was also expressed at the lacA locus under xylose inducible promoter P_xyl using plasmid pECE741 (Popp et al., 2017). The respective genes were amplified from genomic DNA using primers listed in Supplementary file 1. The required restriction enzyme sites (and if required a ribosome binding site (RBS)) were incorporated in the primers used for gene amplification. CRISPR-Cas9 mutagenesis was carried out using pJOE8999 plasmid as described before (Altenbuchner, 2016). The primers used to construct the repair fragment and guide RNAs are in Supplementary file 1. The whole sequence of the genes was confirmed by Sanger sequencing (Biotechnology Resources core facility at Cornell University).

The transposon-sequencing (Tn-Seq) was performed using modified pMarA (Le Breton et al., 2006). pMarA is a plasmid harboring the mariner-Himar1 transposase gene and a temperature-sensitive replicon to select for transposition events. Two MmeI sites were introduced flanking the BstXI and PstI sites to generate plasmid pMarA1 (HE8334). The plasmid was transformed into WT Bacillus subtilis and ΔpbpDFG ponA::erm mutant at 28°C selecting for Kan^R on LB plates supplemented with 10 mM MgSO₄ (final concentration) to generate strain HB20725 and HB20738, respectively. Liquid cultures of HB20725 and HB20738 harboring plasmid-borne transposons were grown at 28°C in liquid LB medium with 10 mM MgSO₄ to mid-exponential phase (OD₆₀₀ ~0.4), diluted and spread on LB plates containing kanamycin and 10 mM MgSO₄. Plates were incubated overnight at 48°C to select for transposition events, and the ones with distinct single colonies (not too crowded, and about 500 colonies per plate) were pooled together. Two hundred and forty plates with a total of >100,000 independent colonies were pooled together for each strain, and their genomic DNA was isolated. For each strain, 10 µg of genomic DNA was digested using MmeI, purified and ligated with sequencing adaptors. Illumina sequencing was performed and DNA adjacent to the transposon insertion sites were matched to Bacillus subtilis reference genome NC_000964.3 using CLC workbench version 8.5.1. Matching results were visualized using CLC workbench, and quantified using Tn-seq Explorer software (Solaimanpour et al., 2015). For visualization of transposon insertions, IGV genome browser was used (Robinson et al., 2011).

For plating efficiency (spot dilution) assays, the cultures were grown in LB medium with 20 mM MgSO₄ to ~0.4 OD₆₀₀. 1 mL of culture was centrifuged at 5000 rpm for 5 min and resuspended in LB medium (without MgSO₄). 10-fold serial dilutions were done in LB medium and 10 µL was plated/spotted on LB agar plates, allowed to air-dry for 10–15 min, and incubated overnight at 37°C.

Cultures were grown in LB medium to ~0.4 OD₆₀₀. 1 µL of this culture was inoculated in each well containing 200 µL of LB media with the required drug concentration. Honeycomb 100-well plates were used for the assay. The increase in the OD₆₀₀ of the culture was monitored real-time using Bioscreen C growth curve analyzer (Growth curves USA). Readings were taken at every 15 min interval up to 24 hr under constant shaking conditions at 37°C. For MIC determination, two-fold increase in the drug concentration was screened ranging from (0.2 to 1.6 µg/mL). The minimum concentration which inhibited the growth (less than 0.2 OD₆₀₀) up to at least 10 hr of incubation was considered as the MIC for the strain.

Antibiotic sensitivity was screened by determining the zone of inhibition using a disc diffusion assay. Cultures were allowed the grow up to ~0.4 OD₆₀₀. 100 µL of this culture was added to 4 mL of top agar (0.75% agar) kept at 50°C to prevent it from solidifying. This was poured on to 15 mL LB agar plates (1.5% agar). The top agar was allowed to air-dry for 30 min. A Whatmann paper filter disc of 6 mm was then put on the top agar. The required amount of drug was added on the disc immediately. The plates were incubated overnight at 37°C and the diameter of the clear zone of inhibition was measured. For all histograms, the zone of inhibition (Y-axis) starts from 6 mm which is the disc diameter. For strains having the inducible promoter P_xyl, both the top agar and LB agar plates were made with 0.1% xylose.

200 µL of cells (~0.4 OD₆₀₀) were added in each well of a 100-well honeycomb plate. To this, 0.05 M of sodium azide (from 5 M stock) was added. Immediately, the real-time monitoring of the decrease in OD₆₀₀ was started with Bioscreen C. Readings were taken every 15 min for up to 24 hr. The time at which 50% of the cells had lysed was noted for each mutant. The time taken (in hours) was plotted as lysis time for each strain. Sodium azide stock was prepared fresh before every experiment.

Gene expression for mreBH and lytE was determined by real-time PCR using primers in Table S2. RNA was purified from 1.5 mL of ~0.4 OD₆₀₀ cells using the RNeasy Kit from Qiagen as per the manufacturer’s instructions. 2 µg of RNA was used to prepare 20 µL of cDNA to achieve a final concentration of 100 ng/µL using High capacity cDNA reverse transcription kit from Applied Biosystems. The gene expression levels were measured using 100 ng of cDNA using 0.5 µM of gene specific primers and 1X SYBR green (Bio-Rad) in CFX connect real-time system from Bio-Rad. gyrA was used a house-keeping gene. Gene expression values (2^-Δct) were plotted after normalization with gyrA.

A very thin agar pad was prepared on slides from 0.8% agarose. 10 µL of cells (~0.4 OD₆₀₀) were spotted and allowed to air dry for 10 min before putting on a cover slip. Cells were imaged using Olympus BX61. Images were captured using Cooke Sensicam camera system under 100X magnification with immersion oil. The images were then analyzed for their length and width using MicrobeJ (Ducret et al., 2016), a plugin for imageJ (Schneider et al., 2012).

Spontaneous suppressors were picked from LB agar plates for ΔecsAΔponA and ΔrasPΔponA. 12 suppressors were selected from each background and their chromosomal DNA extracted using Qiagen DNA extraction kit. DNA was sequenced using the Illumina platform at the Biotechnology Resources core facility at Cornell University. The results were trimmed, mapped and aligned with the ΔecsAΔponA and ΔrasPΔponA backgrounds using CLC genomics workbench.

All the experiments were performed with a minimum of 3 biological replicates. For microscopy images, at least 100 cells per strain were quantified for their cell length and width. One-way ANOVA was used to calculate the statistical significance. Tukey’s comparison test was used to determine significance between all the strains. P-value cut-offs have been mentioned in the figure legends. Different letters represent data which are significantly different. Same letter represents mean values which are not statistically different. Significance between two strains was determined using student’s t-test.