Strains and plasmids used in this study are listed in Supplementary Tables S1 and S2. All E.coli strains used in the reported experiments are derivatives of MG1655 (Guyer et al., 1981). Bacterial cells were grown in lysogeny broth (LB; 1% tryptone, 0.5% yeast extract, 0.5% NaCl) or minimal M9 medium supplemented with a carbon source (0.2% glucose or arabinose). Antibiotics were used at 25 μg/ml (chloramphenicol; Cm), 37.5 μg/ml (kanamycin; Kan), or 50 μg/ml (ampicillin; Amp).

Strains with multiple deletion mutations were made by sequential introduction of each deletion via P1 transduction followed by removal of the aph cassette using FLP recombinase expressed from pCP20, leaving a frt scar sequence at each deletion locus (Datsenko and Wanner, 2000). The correct orientation of the DNA flanking frt sequences in multiple deletion mutants was confirmed for all deletions in each mutant.

Growth conditions prior to microscopy are described in the figure legends. Prior to imaging, cells were immobilized on 2% agarose pads containing 1X M9 salts and covered with #1.5 coverslips. Micrographs were obtained using a Leica DM2500 LED microscope equipped with a Leica DFC7000 GT camera, Fluo Illuminator LRF 4/22, HC PL APO 100x/1.40 Oil Ph3 objective lens, and Leica Las X acquisition software.

HC611 was mutagenized with the Kan^R mariner transposon delivered by conjugation from the donor strain MFDpir/pSC189, which is auxotrophic for diaminopimelic acid (DAP; Ferrières et al., 2010). The donor strain was grown in LB supplemented with 300 μM DAP and 50 μg/ml ampicillin. The recipient strain HC611 was grown in M9-arabinose medium supplemented with 0.2% casamino acids. The cells were washed and resuspended in LB supplemented with 0.2% arabinose and 300 μM DAP to 1/10th of the original culture volume. Equal volumes of donor and recipient were mixed together and 50 μl aliquots were spotted on LB agar supplemented with 0.2% arabinose and 300 μM DAP. Donor-only and recipient-only aliquots were also spotted in parallel to verify proper selection of transconjugants. Mating was allowed to proceed at 37°C for 3 h. Cells were resuspended in LB and transconjugants were then selected on LB agar supplemented with 0.2% arabinose and 37.5 μg/ml kanamycin overnight at 37°C. About 1.4 × 10⁵ colony-forming units were scraped from agar plates by suspension in LB, glycerol was added to a final concentration of 15%, and aliquots were frozen at −80°C.

An aliquot of the transposon mutant library of HC611 was thawed and serially diluted in LB. The serial dilutions were spread on LB agar and incubated at 37°C to select for suppressors of the mepM mepS synthetic lethality. The serial dilutions were also spread on LB agar supplemented with 0.2% arabinose to determine the suppressor frequency. The suppressor frequency of the transposon mutant library was 1.6 × 10⁻³, which was about 20-fold higher than the mock library of HC611. The transposon insertion sites of the suppressors were determined by arbitrarily primed PCR (ArbPCR). The first round of ArbPCR was performed with the primer pair 5′- GGCCACGCGTCGACTAGTACNNNNNNNNNNGATGC - 3′ and 5′- ATCGGCAAAATCCCTTATAAATC -3′. The second nested PCR was performed with 5′- GGCCACGCGTCGACTAGTAC -3′ and 5′- AATAGGAACTTCGGAATAGGAAC -3′ to increase specificity and sensitivity. The resulting PCR products were sequenced with the primer 5′- AATAGGAACTTCGGAATAGGAAC -3′ that anneals to the transposon sequence to identify the transposon–chromosome junctions.

Cells in 2 ml cultures with an OD₆₀₀ of about 0.5 were harvested by centrifugation. Cell pellets were resuspended in 0.1 ml 1X Laemmli sample solution, sonicated, and heated at 95°C for 5 min. Then, proteins were separated by SDS/PAGE and transferred onto a PVDF membrane. To reduce non-specific binding, the membrane was blocked by incubation in 5% skimmed milk in 1X PBS containing 0.05% Tween-20 (PBS-T) for 2 h. The blocked membrane was then incubated overnight with the appropriate primary antibodies, monoclonal anti-FLAG M2 antibody (Sigma, 1:10,000) or anti-RpoA 4RA2 Mouse IgG1 (BioLegend, 1:4,000), at 4°C. The membrane was washed four times with PBS-T and then probed with appropriate secondary antibodies conjugated with horseradish peroxidase, anti-mouse IgG HRP-linked antibody (Cell Signaling Technology, 1:20,000), and incubated for 3 h at room temperature. After washing the membrane with 1× PBS-T five times, the membrane was overlaid with ECL-SuperKine^™ West Femto Maximum Sensitivity Substrate (Abbkine) for anti-FLAG immunoblotting or Miracle-Star^™ Western blot detection system (iNtRON) for anti-RpoA immunoblotting, and chemiluminescence was imaged using the ImageQuant LAS 4000 mini imaging system (GE healthcare).

Prc, NlpI, and MepH were overexpressed and purified with a 6xHis-SUMO (H-SUMO) tag fused to the N-terminus (Mossessova and Lima, 2000; Marblestone et al., 2006). Before the protein purification steps, the plasmid clones, pWJ24, pWJ25, or pWJ26, were transformed into the E.coli BL21(DE3)/pLysS strain and the fresh transformants were inoculated into 10 ml of LB supplemented with ampicillin, chloramphenicol, and 0.2% glucose and grown overnight at 37°C. These cultures were diluted 1:100 into 0.5 liter of LB containing ampicillin and 0.04% glucose, and grown at 37°C. When the OD₆₀₀ reached 0.5, protein expression was induced by the addition of 1 mM IPTG and grown at 30°C for an additional 4 h. Then, cells were centrifuged at 6,000 × g for 10 min at 4°C. The cell pellets were resuspended in 25 ml of 1X PBS (Bio-Rad, 10 mM sodium phosphate, 150 mM sodium chloride, pH 7.8) supplemented with 20 mM imidazole, and stored at −80°C. The resuspended cell pellets were lysed by sonication and centrifuged at 50,000 × g for 90 min at 4°C to remove cell debris. The supernatant was loaded onto 1 ml of Ni²⁺-NTA agarose equilibrated with 1X PBS containing 20 mM imidazole. The column was then washed with 20 ml of 1X PBS containing 20 mM imidazole. Bound protein was eluted sequentially with 1 ml of 1X PBS containing increasing amounts of imidazole (50, 100, 150, 200, 250, 300, or 500 mM). The peak fractions containing the eluted proteins were pooled, mixed with 50 μg His6-tagged UlpI (SUMO protease) to cleave the His6-SUMO-tag, and dialyzed against 1X PBS overnight at 4°C to remove imidazole. The dialyzed protein was passed onto 1 ml of Ni²⁺-NTA agarose equilibrated with 1X PBS to remove the released His6-SUMO tag, His6-tagged UlpI, and other contaminating proteins. Untagged Prc and NlpI were collected in the flow through, aliquoted, snap-frozen in liquid nitrogen, and stored at −80°C.

MepH was soluble in the His-SUMO-tagged form but precipitated after cleavage of the His-SUMO tag. Thus, MepH was prepared in the form of His-SUMO-MepH by dialyzing the peak fractions of His-SUMO-tagged MepH in 1X PBS without treatment of UlpI.

Among the genes that encode DD-endopeptidases in E. coli, mepS, mepM, and mepH have been shown to be collectively required for cell wall assembly, but the cellular functions of the other DD-endopeptidases remain unclear. To examine whether other DD-endopeptidases can also function in PG assembly in vivo, we tested if overexpression of individual DD-endopeptidase genes suppresses the growth defect of the mepS mepM double mutant in LB medium. To facilitate characterization of the MepS^— MepM^— growth defect, we constructed the strain HC611 (P_ara::mepS, ΔmepM) whose mepS promoter was replaced with the araBAD promoter in combination with the mepM deletion. Then, plasmids that express each DD-endopeptidase gene from the IPTG-inducible tac promoter were introduced into HC611. Suppression of the MepS^— MepM^— growth defect was tested on LB agar lacking arabinose but instead supplemented with IPTG to deplete mepS expression and induce the expression of the DD-endopeptidases. In addition to mepS and mepM, overexpression of mepH, pbpG, mepA, and ampH also suppressed the growth defect (Figure 1A). Although strains overexpressing DD-endopeptidases other than mepS and mepM formed much wider cells than the wild-type strain, overexpression of mepH, pbpG, and mepA prevented the lysis of HC611 in LB medium, showing that these DD-endopeptidases can at least function in maintaining cell wall integrity (Figure 1B).

As most DD-endopeptidase genes seem to be expressed at a level comparable to that of mepM according to ribosome profiling data in E. coli (Li et al., 2014), we suspected that there could be regulatory mechanisms that inhibit these gene products from functioning in PG expansion under normal laboratory growth conditions. We reasoned that inactivation of such regulatory mechanisms might suppress the growth defect of the mepS mepM double mutant. To identify suppressors of the MepS^— MepM^— growth defect, a random transposon insertion library of HC611 was generated using growth medium supplemented with arabinose, and the library was selected on LB agar not supplemented with arabinose. Transposon insertions of the isolated suppressor mutants were mapped to either prc or nlpI (Figure 2A). The suppressive effect of the prc or nlpI mutation was then confirmed by constructing the Δprc and ΔnlpI derivative strains of HC611 and examining their growth phenotype (Figure 2B).

As prc and nlpI encode the proteolytic system that degrades MepS, it seemed possible that mutation of prc or nlpI might suppress the growth defect by stabilizing MepS produced by basal expression from the arabinose-inducible promoter in HC611. Thus, we also examined the effect of prc or nlpI mutation in the strain with complete deletions of mepS and mepM. As a ΔmepS ΔmepM double mutant strain survives in an M9-minimal medium lacking casamino acids (Singh et al., 2012; Kim et al., 2021), the ΔmepS ΔmepM Δprc and ΔmepS ΔmepM ΔnlpI triple mutant strains were generated using minimal medium. The growth defect of the ΔmepS ΔmepM double mutant was also suppressed by mutation of prc or nlpI, indicating that prc and nlpI are involved in the negative regulation of DD-endopeptidases other than MepS and MepM (Figure 2C).

To identify DD-endopeptidase(s) whose activity is required for the survival of the Δprc and ΔnlpI suppressors of the MepS^— MepM^— growth defect, we mutated individual DD-endopeptidase genes in the Δprc and ΔnlpI derivative strains of HC611 and examined the growth phenotypes in rich medium. An obvious growth defect was observed only upon introduction of an mepH deletion. Inactivation of MepH caused a severe growth defect in the ΔnlpI suppressor on LB agar (Figure 3A). Although the mepH mutation did not significantly impair growth of the Δprc suppressor on normal LB agar, it caused a severe growth defect when LB agar lacking NaCl was used as a growth medium (Supplementary Figure 1; Figure 3B). These results indicated that MepH is the DD-endopeptidase critical for suppression of the MepS^— MepM^— growth defect by inactivation of Prc or NlpI.

We also assessed the effect of prc, nlpI, and mepH mutations on PG homeostasis by comparing the lysis phenotype and morphology of HC611 and its mutant derivatives. nlpI mutation suppressed HC611 lysis, but it did not fully restore the regular rod shape and resulted in the formation of much wider cells than wild type (Figures 3C,E). On the other hand, prc mutation suppressed the shape defect as well as the lysis of HC611 (Figures 3D,F). In addition, consistent with the spot dilution assay result, inactivation of MepH abrogated the suppressive effect of the prc and nlpI mutations and led to cell lysis (Figures 3E,F).

Although it was shown that mepS, mepM, and mepH are collectively essential for the survival of E.coli in minimal medium (Singh et al., 2012), our result showed that prc mutation suppresses this essentiality even in rich medium (Supplementary Figure 1). Thus, we further examined the Δprc- or ΔnlpI-mediated suppression of the synthetic lethality of mepS, mepM, and mepH mutations using various growth media (Supplementary Figure 2). In M9-minimal medium, nlpI mutation also suppressed the synthetic lethality. However, the ΔnlpI suppressor made mucoid colonies unlike the Δprc suppressor, indicating that nlpI mutation only weakly suppresses the synthetic lethality. Prc mutation suppressed the synthetic lethality between mepS, mepM, and mepH much more robustly than the nlpI mutation, but even the Δprc suppressor was not able to survive if LB agar lacking NaCl was used as the growth medium. Thus, these results indicated that mepS, mepM, and mepH are indeed the DD-endopeptidases that play a major role in PG assembly in E.coli, although it is unclear how prc mutation suppresses the synthetic lethality much more strongly than nlpI mutation.

Requirement of MepH for Δprc- or ΔnlpI-mediated suppression of the MepS^— MepM^— growth defect suggested that MepH might also be negatively controlled by Prc and NlpI. Thus, we assessed the effects of nlpI or prc overexpression on the survival of the MepS^— MepM^— strain in minimal medium. If MepH is under the proteolytic control by Prc and NlpI, overexpression of prc or nlpI might accelerate MepH degradation and cause synthetic lethality with the mepS and mepM mutations in minimal medium. Consistent with this idea, overexpression of prc or nlpI in HC611 caused a severe growth defect on M9 glucose agar similar to that caused by mepH mutation (Figures 4A,B). Importantly, overexpression of prc or nlpI did not cause a similar growth defect in the wild-type strain, indicating that the growth defect arises due to synthetic lethality with inactivation of MepS and MepM rather than a general toxicity of prc or nlpI overexpression.

To examine the effects of various DD-endopeptidase-related mutations on cellular MepH levels, we fused a 3X FLAG tag to the C-terminus of MepH at the native chromosomal locus. The resulting MepH-FLAG appeared functional, since strains with the mepH-FLAG fusion showed growth phenotypes similar to strains with untagged mepH (Figures 5A,B). As the immunoblot signal for MepH-FLAG was relatively weak, we looked for a strain where the MepH-FLAG signal could be readily detected. We suspected that MepH levels might increase in the mepS mepM strain because MepH activity is essential for the survival of the mepS mepM strain. Indeed, MepH-FLAG levels slightly increased in the mepS mutant and increased more than 10-fold in the ΔmepS ΔmepM strain compared with the MepH level in the wild-type strain, suggesting that the low DD-endopeptidase activity somehow induces mepH expression or delays MepH degradation in the ΔmepS ΔmepM strain (Figures 5C,D).

As MepH levels are higher in the ΔmepS ΔmepM strain, and it is also relevant to assess the effects of prc and nlpI mutations in the strain where we can observe suppression of the growth defect, we compared MepH levels in the ΔmepS ΔmepM strain background (Figure 6A). The prc and nlpI mutations caused a significant increase in MepH levels. The MepH-FLAG signal increased more than 5-fold by the prc mutation and about 3.5-fold by the nlpI mutation, indicating that MepH is negatively controlled by the Prc/NlpI proteolytic system. We also compared the levels of other DD-endopeptidases between the ΔmepS ΔmepM strain and its Δprc and ΔnlpI derivatives by fusing the FLAG tag to the C-terminus of each gene at their native chromosomal loci (Figure 6A). The levels of PBP7 (pbpG), MepA, and AmpH also increased slightly in the Δprc and ΔnlpI derivatives, but the increase was less than 2-fold, indicating that Prc and NlpI are not significantly involved in the negative regulation of these DD-endopeptidases.

Next, we monitored MepH levels in the ΔmepS ΔmepM strain and its prc and nlpI mutant derivatives after inhibition of protein synthesis to test if Prc and NlpI are involved in MepH degradation (Figures 6B,C). While MepH levels rapidly decreased in the ΔmepS ΔmepM strain, it did not noticeably decrease in its Δprc derivative strain over the same time interval. MepH levels decreased slightly in the ΔnlpI derivative strain, but the rate of decrease was lower than the parental strain. These results suggested that the higher MepH levels in the prc and nlpI mutants are due to reduced degradation rather than increased production, indicating that Prc and NlpI are indeed involved in the proteolytic control of MepH. The slow decrease of MepH levels in the nlpI mutant suggested that MepH can be degraded by Prc in an NlpI-independent way.

To determine if MepH is a direct substrate of the Prc/NlpI proteolytic system, we purified each protein and tested if MepH is degraded in the presence of Prc and NlpI (Figure 7). We used His6-SUMO-tagged MepH instead of MepH because untagged MepH was not soluble in our experimental conditions. His6-SUMO-tagged MepH will be referred to as MepH for convenience. MepH was stable when incubated alone or with NlpI, but was degraded rapidly in the presence of Prc. The degradation was further accelerated when NlpI was also added to the reaction. Interestingly, the degradation of MepH by Prc was initially fast even without NlpI, but slowed down when MepH concentration was reduced in the reaction lacking NlpI. On the other hand, MepH was rapidly degraded to an undetectable level in the reaction that contained both Prc and NlpI, suggesting NlpI is indeed involved in MepH degradation, especially when MepH concentration is low.