Endopeptidase Regulation as a Novel Function of the Zur-Dependent Zinc Starvation Response

Bacteria encode a variety of adaptations that enable them to survive during zinc starvation, a condition which is encountered both in natural environments and inside the human host. In Vibrio cholerae, the causative agent of the diarrheal disease cholera, we have identified a novel member of this zinc starvation response, a cell wall hydrolase that retains function and is conditionally essential for cell growth in low-zinc environments. Other Gram-negative bacteria contain homologs that appear to be under similar regulatory control. These findings are significant because they represent, to our knowledge, the first evidence that zinc homeostasis influences cell wall turnover. Anti-infective therapies commonly target the bacterial cell wall; therefore, an improved understanding of how the cell wall adapts to host-induced zinc starvation could lead to new antibiotic development. Such therapeutic interventions are required to combat the rising threat of drug-resistant infections.

N-acetylglucosamine (NAG) and N-acetylmuramic acid (NAM) sugar strands (2) (Fig. 1A). Adjacent PG strands are linked to each other via peptide side chains, enabling the PG to assemble into a mesh-like structure called the sacculus (3). In Gram-negative bacteria, the sacculus is a single PG layer that is sandwiched between an inner and an outer membrane (4). This thin wall must be rigid enough to maintain cell shape and to contain high intracellular pressure (3,5); however, the wall must also be flexible enough to accommodate cell elongation, cell division, and the insertion of transenvelope protein complexes (6). This requirement for both rigidity and flexibility necessitates continuous remodeling of the cell wall, which is accomplished by a delicate interplay between PG synthesis and degradation. Inhibition or dysregulation of either process can cause growth cessation or cell lysis, rendering the mechanisms of cell wall turnover an attractive target for new antibiotic development (7,8).
PG synthesis is mediated by penicillin binding proteins (PBPs; the targets for beta-lactam antibiotics) and shape, elongation, division, and sporulation (SEDS) proteins (9). These proteins collectively catalyze cell wall synthesis through two main reactions, namely glycosyltransfer, to add new PG monomers to the glycan strand, and transpeptidation (TP) to crosslink the peptides of adjacent strands (2). Cell wall turnover is mediated by "autolysins," a collective term for diverse and often redundant enzymes (amidases, lytic transglycosylases, and endopeptidases) that are able to cleave PG at almost any chemical bond (6). Endopeptidases (EPs), for example, hydrolyze the peptide crosslinks that covalently connect adjacent PG strands. EPs are crucial for cell elongation in several well-studied Gram-positive and Gram-negative rod-shaped bacteria (10-12), presumably because EPs create gaps in the PG meshwork to allow for the insertion of new cell wall material. Consistent with this proposed role, EP overexpres- , and metallo-endopeptidase domain (M23, green). Protein domains were annotated using UniProt (64). aa, amino acids. (C) Promoter::lacZ transcriptional reporters for each endopeptidase were spotted onto LB (top row) and M9 (bottom row) agar containing X-Gal. A blue colony (ϩ) indicates that the promoter is actively transcribed. Wild-type (WT) and ⌬lacZ mutant strains are included as positive and negative controls, respectively. sion promotes PBP activity in Escherichia coli, likely through the generation of initiation sites for PG synthesis (13).
While EPs are essential for growth, they are also main drivers of PG degradation after inhibition of PBPs (14,15). Thus, EP activity must be tightly controlled under normal growth conditions. EPs in two divergent bacterial species (E. coli and Pseudomonas aeruginosa) are proteolytically degraded to adapt to conditions that require changes in PG cleavage activity (16,17), such as the transition into stationary phase. In Bacillus subtilis, EP expression is regulated by growth-phase dependent sigma factors (18)(19)(20)(21). However, it is not known how EP expression is modified in response to specific environmental stresses.
In this study, we investigate the genetic regulation of specialized EPs in Vibrio cholerae, the causative agent of the diarrheal disease cholera. V. cholerae encodes three nearly identical EPs that are homologous to the well-characterized D,D-endopeptidase MepM in E. coli (10). Each EP contains a LysM domain that likely binds PG (22) and a putatively Zn 2ϩ -dependent M23 catalytic domain that hydrolyzes peptide cross-links (23) (Fig. 1B). We previously showed that two of these homologs (ShyA and ShyC) are housekeeping EPs that are collectively essential for V. cholerae growth (12). The gene encoding the third EP, shyB, is not transcribed under standard laboratory conditions (LB medium), and thus little is known about its biological function. To elucidate the role of ShyB, we conducted a transposon screen to identify mutations that promote shyB expression in LB. We found that shyB is induced by zinc starvation and, unlike the other two M23 EPs, ShyB enzymatic activity is resistant to treatment with the metal chelator EDTA. These data suggest that ShyB acts as an alternative EP to ensure proper PG maintenance during zinc starvation. Importantly, this represents the first characterization of an autolysin that is controlled by Zur-mediated zinc homeostasis and provides insight into how other Gram-negative bacteria might alter EP activity in zinc-limited environments.

RESULTS
shyB is repressed in LB but transcribed in minimal medium. The hydrolytic activity of EPs needs to be carefully controlled to maintain cell wall integrity. We therefore considered it likely that specialized EPs are transcriptionally regulated and only induced when required. To test this hypothesis, we examined the expression patterns of V. cholerae's LysM/M23 endopeptidases using lacZ transcriptional fusions. We first compared promoter activity on LB and M9 agar, as our previous work had shown that a ΔshyB mutation exacerbates a ΔshyA growth defect in M9 minimal medium (12). The P shyA ::lacZ and P shyC ::lacZ reporters generated a blue colony color on both LB and M9 agar (Fig. 1C); thus, these promoters are actively transcribed in either medium. Quantification of ␤-galactosidase (LacZ) activity in liquid culture showed that P shyA transcription does not vary between LB and M9 media, whereas P shyC promoter activity was slightly lower, yet still robust, in M9 minimal medium (Fig. S1). These data are consistent with ShyA and ShyC's role as the predominant growth-promoting EPs (12). In contrast, P shyB ::lacZ yielded blue colonies and produced detectable levels of ␤-galactosidase activity in M9 minimal medium only, indicating that the shyB promoter is induced in M9 medium but tightly repressed in LB ( Fig. 1C and S1).
shyB is induced by zinc starvation. To elucidate the specific growth conditions that favor shyB expression, we sought to identify the genetic factors regulating shyB transcription. To this end, we subjected the shyB transcriptional reporter strain to Himar1 mariner transposon mutagenesis and screened for lacZ induction (blue colonies) on LB agar. After two independent rounds of mutagenesis (50,000 total colonies), the screen yielded 26 blue insertion mutants. These were divided into two distinct classes according to colony color, with 12 dark-blue and 14 light-blue colonies. Strikingly, arbitrary PCR (24) mapped all 26 transposon insertions to only two chromosomal loci; both contained genes whose products play roles in zinc homeostasis, vc2081-2083-znuABC (light-blue colonies) and vc0378-zur (dark-blue colonies) ( Fig. 2A). znuABC encodes V. cholerae's high-affinity zinc uptake system (25), while Zur is a Fur family Endopeptidase Regulation via Zinc Starvation Response ® transcriptional regulator and the central repressor in the zinc starvation response (26). Under zinc-rich conditions, Zur and its Zn 2ϩ corepressor bind to promoters containing a "Zur box" and block transcription (27). Under low-zinc conditions, Zur dissociates from promoters to induce the zinc starvation response (28). This response includes genes encoding zinc uptake systems (i.e., znuABC and zrgABCDE) (25) and zinc-independent paralogs that replace proteins that ordinarily require zinc for function (i.e., ribosomal proteins) (29).
To validate these transposon hits, we constructed clean deletions of zur and znuA in the P shyB ::lacZ reporter strain. Deletion of either gene resulted in activation of the shyB promoter, as indicated by blue colony color on LB agar (Fig. S2A) or ␤-galactosidase activity measured in LB broth (Fig. 2B). P shyB repression was restored by expressing the deleted genes in trans ( Fig. 2B and S2A). Thus, shyB is induced under conditions that are expected to either mimic (zur inactivation) or impose [znuA(BC) inactivation] zinc starvation.
If zinc starvation is the factor inducing shyB expression in M9 medium, we would expect the P shyB ::lacZ reporter to be repressed by external zinc addition. Indeed, supplementing M9 agar plates with 10 M ZnSO 4 was sufficient to turn off the shyB promoter in a wild-type (WT) background (Fig. 2C), whereas repression could not be achieved by the addition of other transition metals (iron and manganese) (Fig. S2B). Zinc supplementation repressed the reporter in a ΔznuABC mutant but not in a Δzur mutant (Fig. 2C). This suggests that the P shyB activation in the ΔznuABC mutant is caused by zinc deficiency, while activation in the Δzur mutant is due to the loss of zinc-sensing repression mechanism. To quantify the effect of zinc on shyB promoter activity, we grew the WT, Δzur mutant, and ΔznuABC mutant reporter strains in M9 medium supplemented with increasing concentrations of zinc and measured ␤-galactosidase activity. As expected, P shyB ::lacZ expression in WT and in ΔznuABC mutant tapered off at higher zinc concentrations, whereas expression in the Δzur mutant was zinc independent (Fig. 2D). These data demonstrate that shyB transcription is repressed by high zinc availability and that the repression mechanism requires Zur.
Zur directly binds the shyB promoter. Given Zur's well-defined role as a zincsensing transcriptional regulator (27) and its requirement for P shyB repression in zincrich media, we hypothesized that Zur directly binds the shyB promoter. To test this, we retrieved a Zur box sequence logo built from 62 known regulatory targets in the Vibrionaceae family (30,31) and aligned it with the shyB promoter region. This alignment identified a highly conserved Zur box, which is characterized by an inverted AT-rich repeat (Fig. 3A). We used 5=-rapid amplification of cDNA ends (RACE) to locate the shyB transcriptional start site (tss) (ϩ1) and found that the putative Zur box overlaps with both the Ϫ10 region and the tss. A bound Zur-Zn 2ϩ complex at this position likely prevents RNA polymerase binding and thereby prevents transcription (32).
To determine if Zur binds the shyB promoter in vitro, we then incubated purified Zur with a labeled DNA probe encoding the P shyB Zur box. Binding was assessed in the presence of ZnCl 2 using an electrophoretic mobility shift assay (EMSA). As evident by a band shift, Zur forms a complex with the P shyB DNA in vitro (Fig. 3B, lanes 1 and 2). To examine DNA binding specificity, a 100-fold molar excess of unlabeled specific (S) or nonspecific (NS) competitor DNA was included in the binding reaction. The S competitor, which carries a sequence identical to that of the labeled probe, effectively sequestered Zur and increased the amount of unbound labeled probe (Fig. 3B, lane 3). Meanwhile, the NS competitor was ineffective at binding Zur (Fig. 3B, lane 4). These data indicate that the shyB promoter contains an authentic Zur box, and we conclude that shyB is a novel member of the Zur regulon.
ShyB supports growth in chelator-treated medium. As shyB is part of the Zurmediated zinc starvation response, we hypothesized that V. cholerae relies on ShyB endopeptidase activity when zinc availability is low. To induce zinc starvation and robustly derepress the Zur regulon, V. cholerae strains were grown in M9 minimal medium supplemented with TPEN [N,N,N=,N=-tetrakis(2-pyridylmethyl)ethylenediamine], a cell-permeable metal chelator with high affinity for zinc (33). As expected from our genetic analysis, TPEN addition resulted in the production of ShyB protein (as measured by Western blotting), which could be reversed by adding zinc (Fig. S3A). Endopeptidase Regulation via Zinc Starvation Response ® We first tested whether native shyB could restore growth to the ΔshyAC mutant under zinc starvation conditions. shyA and shyC deletions were generated in a parent strain expressing an isopropyl thio-␤-D-galactopyranoside (IPTG)-inducible copy of shyA (lacZ::P tac -shyA), as these genes are conditionally essential in rich medium (12). In the absence of IPTG, we found that chelation with either TPEN or EDTA (a more general divalent metal ion chelator) induced growth of the ΔshyAC mutant but not in the mutant that additionally lacked shyB ( Fig. 4A and S4A). As expected, chelationdependent growth of the ΔshyAC mutant could be suppressed by adding zinc (Fig. 4B  and S4A). These data suggest that induction of shyB alone is sufficient to sustain V. cholerae growth, and synthetic lethality of shyA and shyC is due to the lack of shyB expression under laboratory growth conditions. Consistent with this interpretation, exogenous shyB expression restored growth to the ΔshyABC P tac -shyA mutant (Fig. S4B), and we were able to generate a ΔshyAC knockout in a Δzur background that grows robustly in LB medium (Fig. S4C). Growth of each strain was monitored by the optical density at 600 nm in a Bioscreen C 100-well plate. Error bars report standard error of the mean (SEM) for three independent biological replicates. (A and B) Growth curves on a log scale are shown for WT (green circle), ⌬shyAC lacZ::P tac -shyA mutant (blue square), and ⌬shyABC lacZ::P tac -shyA mutant (red triangle). (C and D) Growth rates (per hour) of WT (solid black), single EP mutants (solid gray), and double EP mutants (striped) were calculated from changes in optical density during exponential phase. Error bars report the SEM for three independent biological replicates. Statistical difference relative to the WT was assessed using a one-way ANOVA followed by Dunnett's multiple-comparison test (****, P Ͻ 0.0001; ***, P Ͻ 0.001; *, P Ͻ 0.05). (E) Phase-contrast images of V. cholerae (C and D) sampled from mid-log phase (scale bar ϭ 5 M).
A shyB deletion alone did not result in a significant defect in growth rate or morphology in TPEN-treated M9 medium ( Fig. 4C and E); however, autolysins often need to be deleted in combination to result in a substantial phenotype (11). We therefore generated all possible combinations of LysM/M23 endopeptidase deletions to broadly dissect the relevance of zinc concentrations for EP activity. ⌬shyA mutant cells were somewhat enlarged under either growth condition but did not exhibit a strong growth rate defect (Fig. 4C to E). The ΔshyAB double mutant failed to grow in TPEN-treated M9 medium (Fig. 4C), and cells exposed to this condition were aberrantly thick and long (Fig. 4E). This indicates that ShyC, the only essential LysM/M23 EP in the ΔshyAB mutant, cannot support growth in zinc-starved medium. In zinc-replete medium, the ΔshyAC lacZ::P tac -shyA mutant failed to grow in the absence of IPTG and displayed a similar aberrant cell morphology (Fig. 4D and E), consistent with a lack of shyB expression under this condition. This trade-off in synthetic lethality partners in low-zinc (ΔshyAB mutant) and high-zinc (ΔshyAC) growth media tentatively suggests that ShyB may function as a replacement for ShyC during zinc starvation. ShyC protein levels, as measured by Western blotting, were not reduced in the presence of TPEN (Fig. S3A), ruling out the possibility that ΔshyAB mutant lethality reflects downregulation or degradation of ShyC. Rather, these observations suggest that ShyC activity is more sensitive to zinc chelation than the other EPs. Alternatively, TPEN might induce changes in PG architecture that make it resistant to cleavage by ShyC.
ShyB is an EDTA-resistant D,D-endopeptidase in vitro. ShyB is predicted to be a zinc-dependent D,D-endopeptidase, but biochemical evidence is lacking. Thus, we measured the in vitro hydrolytic activity of each EP against V. cholerae sacculi. Recombinant Shy proteins were purified without the N-terminal signal sequence (ShyA Δ1-35 , ShyB Δ1-34 , and ShyC Δ1-33 ) (Fig. S5); as a negative control, we purified ShyB Δ1-34 with a mutation (H370A) in the active site that is expected to abolish activity. Each EP was incubated with purified V. cholerae sacculi; the soluble PG fragments released by digestion, as well as the remaining insoluble pellet, were treated with muramidase to process long glycan strands into their subunits (Fig. 5A). The resulting muropeptides were then separated using ultrahigh-performance liquid chromatography (UPLC) and quantified by spectrophotometry (see Materials and Methods for details).
The muramidase-digested insoluble fraction contains PG that was not released by EP digestion (Fig. 5B). The no-enzyme control showed a large peak corresponding to D44 dimers, indicating that these peptide crosslinks are abundant in the V. cholerae sacculus substrate (Fig. 5B). This D44 peak was noticeably absent from the ShyA-, ShyB-, and ShyC-digested sacculi. Instead, each of the Shy endopeptidases (but not the H370A mutant) hydrolyzed sacculi and generated a profile of soluble fragments (Fig. 5C). ShyA and ShyC produced similar muropeptide profiles, indicating similar hydrolytic activity in vitro, while the ShyB chromatogram revealed more peaks with shorter retention times (Fig. S6). Tandem mass spectrometry (MS/MS) analysis determined that these ShyBgenerated peaks correspond to un-cross-linked oligo-NAG-NAM-tetrapeptide chains [M4] 2-4, as well as small amounts of M4, M4N, and M0 -M4 monomers. Consistent with a differential cleavage activity, ShyB was able to further process PG predigested with ShyA or ShyC, while these EPs only slightly modified ShyB-digested PG (Fig. S7).
The soluble EP digestion products were further treated with muramidase (Fig. 5D). Each of the EP-treated samples contained a large peak near 4 min, corresponding to a M4 monomer, in addition to smaller amounts of M4N monomer and chains of monomers (M0 -M4 and M4 -M4N). These peaks were absent from the negative controls (no-enzyme and ShyB H370A). There was virtually no D44 peak detected in the Shy-treated samples, indicating that all three LysM/M23 EPs function as a D,Dendopeptidase in vitro (Fig. 5B). We speculate that the apparent unique activity of ShyB on whole sacculi may reflect its ability to process substrate with a more diverse set of conformations than ShyA and ShyC.
M23 domains typically require a coordinated zinc ion to carry out PG hydrolysis (23). We have previously demonstrated that ShyA requires zinc for activity in vitro (12), and Endopeptidase Regulation via Zinc Starvation Response ® others have modeled zinc in the active site of a ShyB crystal structure (34). Based on its regulation by Zur, we hypothesized that ShyB evolved to function in zinc-limited environments. To test this, we repeated the in vitro PG hydrolysis assays under metal-limited conditions by using the divalent cation chelator EDTA. Strikingly, ShyB EP activity was largely unaffected by EDTA at the wide range of concentrations tested (0.1 mM to 20 mM) ( Fig. 6 and Fig. S8). ShyA and ShyC had reduced activity in 0.5 mM EDTA and suffered a total loss of activity at higher concentrations, consistent with results previously obtained for ShyA (12). ShyC activity appeared to be more sensitive to EDTA than was ShyA (at 1 and 5 mM), but this difference was not statistically significant. These in vitro assays suggest that ShyB has a high affinity for, or can function without, divalent cations like zinc.

DISCUSSION
Functionally redundant endopeptidases support cell growth. The importance of EPs has been established in both Gram-negative and Gram-positive bacteria (10)(11)(12), supporting the long-standing hypothesis that autolysins create space in the PG meshwork for the insertion of new cell wall material (8). As with other autolysins, EPs are  Fig. 5B) and normalizing to the no-EDTA treatment (100% activity) and no-enzyme control (0% activity). Error bars represent the SEM of at least two biological replicates, and statistical significance was assessed using a two-way ANOVA followed by Dunnett's multiple-comparison test (***, P Ͻ 0.0001).  (65), show the arrangement of a LysM/M23 endopeptidase gene (mepM/yebA, green) and adjacent zinc importer genes (znuABC, blue) in 12 Gram-negative organisms. shyB, in V. cholerae O1 El Tor N16961, exists as a single gene on a Vibrio seventh pandemic island (VSP-II), separate from the znuABC locus. Arrows indicate the approximate location of the unidirectional or bidirectional promoter and site of Zur binding. Asterisks signify published data sets that support Zur-and/or zinc-dependent regulation of the endopeptidase. Evolutionary history of the species shown was inferred using alignments of 16S rRNA genes (66) and the neighbor-joining method for tree construction (67) in MEGA7 (68). Evolutionary distances were computed using the maximum composite likelihood method (69) and are in units of number of base substitutions per site.
Endopeptidase Regulation via Zinc Starvation Response ® often functionally redundant under laboratory growth conditions but exhibit slight differences in cellular localization (12,20,46), substrate specificity (10,47), and relative abundance during each growth phase (11,46). Our previous work in V. cholerae identified three LysM/M23 zinc metallo-endopeptidases; two (ShyA and ShyC) are housekeeping enzymes that are conditionally essential for growth, while the role of the third (ShyB) has remained unknown (12). In this study, we define shyB as a new member of the Zur regulon and demonstrate that ShyB can replace the other EPs in vivo when derepressed by zinc limitation. This is a novel mechanism for regulating autolysins and establishes a link between two essential processes, cell wall turnover and metal ion homeostasis.
Zinc availability affects the expression and activity of cell wall hydrolases. We initially observed that the shyB promoter is active on M9 medium and repressed on LB agar. LB contains ϳ12.2 M zinc (48), while presumably the concentrations are much lower for the defined M9 medium (which does not contain any added zinc). Consistent with a role for the zur regulon in shyB regulation, we found that supplementing M9 medium with nanomolar concentrations of zinc was sufficient to repress the shyB promoter in liquid culture (Fig. 2D). As a cautionary note, this suggests that V. cholerae is starved for zinc in M9 medium (and possibly other minimal media as well), a complication not usually considered when interpreting results obtained in this medium.
Based on its membership in the Zur regulon, it is likely that ShyB evolved to function in low-zinc environments. Indeed, ShyB endopeptidase activity in vitro is unaffected by (49), even at chelator concentrations (e.g., 20 mM) that induce levels of metal starvation that far exceed those encountered in nature (Fig. 6). We found that a mutation in a metal-coordinating residue (H370A) abolished ShyB activity (Fig. 5), but further biochemical characterization of the active site is required to explain how ShyB retains EP activity in low-zinc environments. It is possible that the ShyB active site (i) binds zinc with very high affinity, (ii) utilizes an alternative metal cofactor, or (iii) functions independently of a bound metal cofactor.
ShyA appears to have an intermediate ability to function in low-zinc environments. The ability to sustain growth in the presence of TPEN or EDTA indicates that ShyA function is less affected by metal starvation than ShyC; however, ShyA's activity can still be inhibited by higher concentrations of EDTA in vitro. Collectively, our data suggest a model where ShyA is the predominant housekeeping endopeptidase. ShyC appears to partially overlap in function with ShyA when zinc availability is high. During zinc starvation, we speculate that shyB is derepressed specifically to compensate for a loss of ShyC activity.
ShyB can function as the sole EP under both high-and low-zinc conditions ( Fig. 4A  and S5), so why is it repressed under normal growth conditions rather than just replacing ShyA and ShyC altogether? One possible explanation is that ShyB activity might be more destructive than the other EPs, requiring more careful control. Indeed, our activity assays demonstrated that ShyB has a slightly altered activity profile, processing the sacculus into smaller fragments with shorter retention times (Fig. 5C) and doing so at a higher rate (Fig. S7D to F). Additionally, native ShyB protein levels (constitutively transcribed in a Δzur mutant) are substantially lower than those of either ShyA or ShyC (Fig. S3B), suggesting that less enzyme is required to support growth.
Zur-regulated endopeptidases are present in divergent Gram-negative bacteria. Zur-regulated EPs are a novel adaptation to zinc limitation; however, ShyB activity is not essential to wild-type V. cholerae growth under the laboratory conditions tested. This might be due to functional overlap with the other Shy EPs, but it should also be noted that pandemic V. cholerae horizontally acquired shyB on a pathogenicity island (VSP-II) recently in its evolutionary history (in the 1960s) (50). It is thus possible that ShyB activity is of advantage under conditions that also positively selected for acquisition of VSP-II, such as pathogenesis.
Importantly, Zur-regulated EPs are not confined to Vibrionaceae, as divergent Gramnegative bacteria encode conserved ShyB/MepM/YebA homologs adjacent to the Zur-controlled znu operon (Fig. 7). Transcriptomic data from both Y. pestis and E. coli support the prediction that this EP is upregulated along with the zinc importer, though it is unclear whether these EPs can also be transcribed from a Zur-independent promoter internal to znuA. Parts of the zinc starvation response (e.g., zinc importers) have been shown to be required for host colonization by divergent human pathogens (51)(52)(53)(54), since vertebrates and other hosts sequester metals as a form of nutritional immunity (55). It has also been shown that ShyB homolog YebA in Y. pestis is important for virulence in the plague pathogen (56). It is therefore tempting to speculate that the regulation of endopeptidase activity in low-zinc environments may play a general role in pathogenesis.

MATERIALS AND METHODS
Bacterial growth conditions. Cells were grown by shaking (200 rpm) at 37°C in 5 ml of LB medium unless otherwise indicated. M9 minimal medium with glucose (0.4%) was prepared with ultrapure Milli-Q water to minimize zinc contamination. When appropriate, antibiotics were used at the following concentrations: streptomycin, 200 g ml Ϫ1 ; ampicillin, 100 g ml Ϫ1 ; kanamycin, 50 g ml Ϫ1 ; and chloramphenicol, 5 g ml Ϫ1 . IPTG (200 M) was added to all liquid and solid media if required to sustain V. cholerae growth. 5-Bromo-4-chloro-3-indolyl-␤-D-galactopyranoside (X-Gal; 40 g ml Ϫ1 ) was added to plates for blue-white screening.
5=-Rapid amplification of cDNA ends. The shyB transcription start site was identified with 5=-rapid amplification of cDNA ends (5=-RACE). To obtain a shyB transcript, the Δzur mutant was grown in LB at 37°C until cells reached mid-log phase (optical density at 600 nm [OD 600 ], 0.5), and RNA was extracted using TRIzol reagent and acid-phenol-chloroform (Ambion). DNA contamination was removed through two RQ1 DNase (Promega) treatments and additional acid-phenol-chloroform extractions. cDNA synthesis was performed with MultiScribe reverse transcriptase (Thermo Fisher) and a shyB-specific primer (SM270). cDNA was column purified and treated with terminal transferase (New England BioLabs) to add a homopolymeric cytosine tail to the 3= end. The cDNA was amplified through two rounds of touchdown PCR with a second gene-specific primer (SM271) and the anchored abridged primer (Thermo Fisher). The PCR product was Sanger sequenced using primer SM271.
Electrophoretic mobility shift assay. The LightShift chemiluminescent EMSA kit (Thermo Fisher) was used to detect Zur promoter binding. Forty-one-base pair complementary oligonucleotides (SM264/ 265) containing the putative shyB Zur box, with and without a 5= biotin label, were annealed according to commercial instructions (Integrated DNA Technologies). Twenty-microliter binding reaction mixtures contained buffer, poly(dI-dC) (50 ng l Ϫ1 ), ZnCl 2 (5 M), labeled probe (1 pmol), and purified Zur (600 nM). Unlabeled specific or nonspecific competitor oligonucleotides were added in 100-fold molar excess. Reaction mixtures were incubated on ice for 1 h, electrophoresed on a 6% DNA retardation gel (100 V, 40 min), and wet transferred to a Biodyne B membrane (100 V, 30 min; Thermo Fisher) in a cold room. The membrane was developed using chemiluminescence according to the manufacturer's instructions and imaged using a Bio-Rad ChemiDoc MP imaging system.
Sacculi digestion assay and separation by UPLC. Peptidoglycan from stationary-phase V. cholerae cells was extracted and purified via SDS boiling and digested with muramidase (63). Ten microliters of sacculi and 10 g of enzyme were mixed in 50 l buffered solution (50 mM Tris-HCl [pH 7.5], 100 mM NaCl) in the absence or presence of EDTA (0 to 20 mM). Digestions were incubated for 16 h at 37°C, and enzymes were inactivated by boiling during 10 min and centrifugation for 15 min at 22,000 ϫ g. The soluble fraction and the insoluble pellet were separated, and each sample was further digested with muramidase. All soluble products were reduced with sodium borohydride, their pH was adjusted, and they were injected into a Waters UPLC system (MA, USA) equipped with an Acquity UPLC BEH C 18 , 130-Å, 1.7-m, 2.1 mm by 150 mm column (Waters) and a dual-wavelength absorbance detector. Eluted fragments were separated at 45°C using a linear gradient from buffer A (0.1% [vol/vol] formic acid) to buffer B (0.1% [vol/vol] formic acid, 40% [vol/vol] acetonitrile) in a 12-min run with a flow rate of 0.175 ml min Ϫ1 and detected at 204 nm. Muropeptide identity was confirmed by MS/MS analysis, using a Xevo G2-XS quadrupole time of flight (Q-TOF) system (Waters Corporation, USA) and the same separation conditions. Growth curve analysis. Strains were grown overnight in LB-streptomycin with IPTG. Cells were washed in 1ϫ phosphate-buffered saline (PBS) and subcultured 1:10 into M9-glucose plus IPTG. After 2 h shaking at 37°C, cells were washed and subcultured 1:100 into M9-glucose containing combinations of TPEN (250 nM), ZnSO 4 (1 M), and IPTG (200 M). The growth of each 200-l culture in a 100-well plate was monitored by optical density at 600 nm (OD 600 ) on a Bioscreen C plate reader (Growth Curves America).
Microscopy. Cells were imaged under phase contrast on an agarose patch (0.8% agarose in M9 minimal medium) using a Leica DMi8 inverted microscope.