EstG is a novel esterase required for cell envelope integrity in Caulobacter

Summary

Proper regulation of the bacterial cell envelope is critical for cell survival. Identification and characterization of enzymes that maintain cell envelope homeostasis is crucial, as they can be targets for effective antibiotics. In this study, we have identified a novel enzyme, called EstG, whose activity protects cells from a variety of lethal assaults in the α-proteobacterium Caulobacter crescentus. Despite homology to transpeptidase family cell wall enzymes and an ability to protect against cell wall-targeting antibiotics, EstG does not demonstrate biochemical activity towards cell wall substrates. Instead, EstG is genetically connected to the periplasmic enzymes OpgH and BglX, responsible for synthesis and hydrolysis of osmoregulated periplasmic glucans (OPGs), respectively. The crystal structure of EstG revealed similarities to esterases and transesterases, and we demonstrated esterase activity of EstG in vitro. Using biochemical fractionation, we identified a cyclic hexamer of glucose as a likely substrate of EstG. This molecule is the first OPG described in Caulobacter and establishes a novel class of OPGs, the regulation and modification of which is important for stress survival and adaptation to fluctuating environments. Our data indicate that EstG, BglX, and OpgH comprise a previously unknown OPG pathway in Caulobacter. Ultimately, we propose that EstG is a novel enzyme that, instead of acting on the cell wall, acts on cyclic OPGs to provide resistance to a variety of cellular stresses.

eTOC:

Daitch et al. establish a novel osmoregulated periplasmic glucan (OPG) pathway in Caulobacter. EstG is a periplasmic esterase that acts on a cyclic OPG to promote cell envelope homeostasis and protect against cell stress. The enzymes OpgH and BglX also contribute to envelope integrity through synthesis and hydrolysis of Caulobacter OPGs.

Introduction

The bacterial cell envelope is a multi-component structure that protects bacteria from the external environment. The Gram-negative cell envelope consists of the inner and outer membranes, with the periplasm between them ¹. The peptidoglycan (PG) bacterial cell wall forms a protective meshwork in the periplasm that prevents cell lysis ². During growth, PG metabolic enzymes synthesize, modify, and hydrolyze the PG. Two major classes of PG synthetic enzymes include glycosyltransferases and transpeptidases (TPases), which catalyze polymerization and crosslinking of glycan strands, respectively ³. For most bacteria, PG is an essential structural component, and the primary biosynthetic PG enzymes are essential during growth and division ³. PG enzymes are therefore targets of bactericidal antibiotics, such as β-lactams, which inhibit the TPase activity of penicillin-binding proteins (PBPs) ⁴. Though PG enzymes are effective antibiotic targets, disruption of other components of the envelope can also sensitize cells to stress ^5,6. Understanding all elements of the cell envelope is crucial for identifying new drug targets.

In addition to PG, the periplasm of proteobacteria may contain glycopolymers important for maintaining cell envelope integrity called osmoregulated periplasmic glucans (OPGs). OPGs are glucose polymers made in the periplasm and are thought to function as osmoprotectants ⁷. Across Gram-negative species, OPGs vary in size and geometry, exhibiting linear, branched, and/or cyclic structures depending on which OPG enzymes are produced ^7,8. OPGs may also be modified with, for example, functional groups derived from phospholipids or products of intermediary metabolism, which can influence polymer charge ^7,8. In some bacteria, OPGs are implicated in stress tolerance, as disruption of OPG genes results in increased sensitivity to antibiotics and cell envelope stresses ^7,9. Despite decades of research on OPGs, we have limited knowledge about the diversity of OPG structures, modifications, and metabolic enzymes across bacteria.

In this study, we sought to identify factors required to survive cell wall stress in the α-proteobacterium Caulobacter crescentus, a well-studied model for morphogenesis ¹⁰. A genetic screen revealed an uncharacterized protein required to survive cell wall stress that we named EstG. Although EstG is annotated as a member of the TPase superfamily, it has no detectable activity towards PG. Genetic, structural, and biochemical data indicate that EstG instead acts in the OPG metabolic pathway in Caulobacter. An unbiased mass spectrometry approach identified a native substrate of EstG as a periplasmic, cyclic hexamer of glucose. This is the first OPG identified in Caulobacter and establishes a new class of OPGs in α-proteobacteria. We propose that EstG is a novel enzyme that acts on cyclic OPGs to fortify the cell envelope and provide resistance to a variety of cellular stresses.

Results

EstG is essential for suppression of toxic cell wall misregulation

This study initiated with our interest in understanding Caulobacter cell division, which is orchestrated by FtsZ. We previously demonstrated that expression of a variant of ftsZ lacking the C-terminal linker (called ΔCTL) results in misregulation of PG enzymes and cell death, similar to the effects of β-lactam treatment (Figure 1A) ¹¹. We leveraged ΔCTL toxicity to understand mechanisms of stress survival. To this end, we conducted a screen to identify spontaneous suppressors of ΔCTL-induced lethality (Figure 1A) ¹². Whole genome sequencing of suppressors revealed mutations in genes largely involved in nutrient stress responses (spoT ¹³, cdnL ^12,14, and phoB ¹⁵) (Figure 1A, Data S1). Although each of the suppressors reduced growth rate on their own, slow growth by itself was insufficient to suppress ΔCTL-induced lethality (Figure S1A–F). We were especially intrigued by the identification of suppressing mutations in spoT, since SpoT mediates the stringent response in Caulobacter and has been implicated in antibiotic resistance ¹³.

Figure 1: EstG is required to suppress ΔCTL-mediated lethality.

A. Phase contrast images of ΔCTL and suppressors +/− ΔCTL production. Indicated strains are grown with 0.3% glucose (−ΔCTL) or 0.3% xylose (+ΔCTL) for 7 hours before imaging. Scale bar, 2 μm. Amino acid X represents a premature stop codon. B. Line plot of transposon insertion frequency along the gene locus for CCNA_01638 (named estG) and neighboring genes as determined by transposon sequencing (Tn-Seq) analysis in wild type (WT; EG865), high (p)ppGpp production (RelA’, EG1799), and ΔCTL with high (p)ppGpp production (ΔCTL+RelA’, EG1616). C. Growth curve of strains EG1229 (WT) and EG3116 (ΔestG) with and without ΔCTL production (+/− 0.3% xylose) as monitored by OD₆₀₀. D. Phase contrast images of WT and ΔestG from the 24-hour timepoint of the growth curve in panel C. Scale bar, 2 μm. See also Figure S1 and Data S1 and S2.

SpoT is the only Rel-Spo homolog encoded in Caulobacter and contains synthetase and hydrolase domains that produce and degrade the signaling alarmone (p)ppGpp, respectively. A deletion of spoT results in cells devoid of (p)ppGpp ¹⁶. However, the suppressing mutations were in a regulatory region that releases inhibition of synthesis, suggesting these mutations result in higher (p)ppGpp levels ^17,18. We asked if elevated (p)ppGpp was sufficient to suppress ΔCTL-induced lethality using an inducible, constitutively activated form of the Escherichia coli (p)ppGpp synthase, RelA (hereafter RelA’) (Figure S1G) ¹⁹. Indeed, induction of relA’ suppressed ΔCTL-mediated lethality, whereas production of the enzymatically-dead control, RelA’-dead, did not (Fig S1G).

To better understand how high levels of (p)ppGpp suppress ΔCTL-induced lethality, we conducted comparative transposon sequencing (Tn-Seq) to identify genes that were synthetically lethal with ΔCTL production, using the following strains: wild type (WT), RelA’-producing, and RelA’-producing with ΔCTL. Notably, we identified a gene, CCNA_01638 (hereafter named estG for Esterase for Stress Tolerance acting on Glucans), that appeared to be essential only in the presence of ΔCTL (Figure 1B, Data S2). estG acquired abundant transposon insertions in WT and RelA’ backgrounds, suggesting that it is dispensable in those strains. However, there were almost no transposon insertions in estG in RelA’-producing cells that also produced ΔCTL, indicating estG is essential in the presence of ΔCTL (Figure 1B). EstG is an uncharacterized protein that is annotated as a β-lactamase family protein in the transpeptidase superfamily, which includes numerous PG enzymes.

To validate our Tn-Seq findings, we deleted estG in a strain with xylose-inducible production of ΔCTL and compared its growth to that of WT Caulobacter carrying xylose-inducible ΔCTL (Figure 1C–D). The two strains grew comparably in the absence of xylose (Figure 1C, solid lines). Production of ΔCTL in a WT background resulted in filamentation and lysis over time (Figure 1C–D, black). Notably, producing ΔCTL in a ΔestG background resulted in faster lysis compared to ΔCTL in a WT background (Figure 1C–D, grey).

Though we initially identified estG in a screen for factors required for (p)ppGpp-mediated suppression of ΔCTL-induced stress, we ultimately determined that estG is unrelated to (p)ppGpp. Instead, it was additional antibiotic stress (used to select for the P_xyl-relA’ construct) in the presence of ΔCTL stress that rendered estG essential. To show this, we first deleted estG in a ΔCTL background suppressed by a hyperactive spoT mutant allele (SpoTΔACT_668–712) which has increased (p)ppGpp levels ¹⁷. Production of ΔCTL in this background lacking estG was not lethal, suggesting that estG protection is separate from (p)ppGpp-mediated suppression of ΔCTL (Figure S1I). Moreover, production of ΔCTL in ΔestG cells carrying a gentamycin-resistance marker and xylose-inducible relA’-dead was lethal, with cells showing less growth than when ΔCTL was produced in the ΔestG mutant background alone (Figure S1H). Despite the lack of a relationship between estG and (p)ppGpp, we were intrigued by EstG and sought to explore its relationship to surviving PG and antibiotic stress.

estG is non-essential in unstressed conditions, but required to survive cell wall stress

Our Tn-Seq results indicated that estG was non-essential in a WT background (Figure 1B) ²⁰. We therefore generated a deletion of estG (ΔestG, confirmed by western blotting (Figure S2A)) and compared its growth and morphology to WT. ΔestG cells grew comparably to WT by optical density (Figure 2A) and spot dilution (Figure 2B), though colony size of the ΔestG strain was slightly smaller than WT. ΔestG cells were also morphologically identical to WT (Figure 2C). Therefore, estG is non-essential under normal growth conditions, but becomes essential during ΔCTL induction.

Figure 2: ΔestG does not impact cell viability or growth in unstressed conditions.

A. Growth curve, B. spot dilutions, and C. phase contrast images of wild type (WT, EG865) and ΔestG (EG2658). Culture dilutions are as indicated. Scale bar, 2 μm. D. Minimum inhibitory concentrations (MIC) of WT (EG865) and ΔestG (EG2658) against peptidoglycan (PG)− and ribosome-targeting antibiotics. Measurements in μg/mL. Mec=mecillinam; Vanc=vancomycin; Amp=ampicillin; Fos=fosfomycin; Cef=cephalexin; Spec=spectinomycin; Tet=tetracycline. Asterisk (*) represents value with a secondary zone of light inhibition. See also Figures S2 and S3 and Table S1

Because estG is essential during ΔCTL production, we hypothesized that EstG may be required to survive other cell wall stresses, such as cell wall-targeting antibiotics. To test this, we measured the minimum inhibitory concentrations (MIC) of a variety of antibiotics against WT and ΔestG cells (Figure 2D). ΔestG cells were more sensitive to every cell wall-targeting antibiotic tested, as well as to some ribosome-targeting antibiotics, compared to WT. To confirm that this sensitivity was attributable to loss of EstG, we complemented ΔestG with a vanillate-inducible copy of estG and showed that resistance to ampicillin was restored (Figure S2B). This indicates a broadly important role of EstG during cell wall stress.

While exploring the role of EstG, we noticed the gene immediately downstream from estG, CCNA_01639, is also annotated as a β-lactamase family protein and wondered if the two might be functionally related. CCNA_01639 has high sequence identity to EstG (52%), and both are predicted to reside in the periplasm ²¹. Despite similarity to EstG, CCNA_01639 showed no change in transposon insertion frequency in the presence of ΔCTL (Figure 1B). Similarly, deletion of CCNA_01639 did not result in hypersensitivity to ampicillin or cephalexin (Figure S2C). Moreover, the double deletion, ΔestGΔCCNA_01639, phenocopied the single ΔestG mutant (Figure S2C). Since ΔCCNA_01639 cells had no detectable phenotype, we focused the remainder of our study on characterizing EstG.

EstG is periplasmic with no detectable activity against the cell wall

EstG is 462 amino acids long and has an N-terminal signal sequence, with cleavage predicted between residues 30 and 31 ²¹. To study the localization of EstG, we expressed an inducible EstG-β-lactamase (EstG-BlaM) fusion protein in a β-lactamase deficient strain (ΔblaA; BlaA is the primary β-lactamase in Caulobacter ²²). These cells will only be resistant to ampicillin if EstG contains a functional signal sequence to transport the fused β-lactamase to the periplasm ²³. The EstG-BlaM strain, when plated in the presence of inducer, displayed resistance to ampicillin, thus validating the predicted periplasmic localization of EstG (Figure S2D).

The classification of EstG as a β-lactamase family protein as well as the hypersensitivity of ΔestG cells to ΔCTL and PG-targeting antibiotics suggested that EstG might act as a β-lactamase. However, purified EstG displayed negligible activity against nitrocefin, a substrate used to detect β-lactamase activity in vitro (Figure S2E), compared to a Caulobacter enzyme with β-lactamase activity, EstA ²⁴. This, however, does not rule out an activity against the cell wall, so we next tested for ability to bind PG. In vitro, purified EstG pelleted with Caulobacter PG, whereas a control protein (glutathione S-transferase) remained soluble (Figure S2F). Despite binding PG, EstG did not have detectable activity against any of the most abundant muropeptide species or purified PG sacculi in vitro (Figure S3A–G). Finally, there were no significant differences between the chemical composition of PG isolated from ΔestG cells and PG from WT, as analyzed by muropeptide analysis (Figure S3H–I, Table S1). Considering EstG’s lack of activity against cell wall substrates in vitro, we hypothesized that EstG’s substrate is novel and not directly related to PG metabolism.

estG interacts genetically with opgH, which encodes a putative OPG synthase

To search for the molecular function of EstG in an unbiased fashion, we isolated spontaneous suppressors of the ampicillin sensitivity of ΔestG cells (Data S1). We were most intrigued by a mutation in opgH, a periplasmic glucan glucosyltransferase (OpgH_L480P) (Figure 3A). OpgH has been characterized in other organisms as the synthase of OPGs ⁷. OpgH is the only homolog of known OPG-biosynthetic enzymes encoded in the Caulobacter genome, but the presence of OpgH indicates the existence of an undiscovered OPG pathway. Isolation of a suppressing mutation in OpgH led us to hypothesize that EstG could function in OPG metabolism.

Figure 3: opgH_L480P and opgH_L434P suppress ΔestG sensitivities.

A. Spot dilutions of WT (EG865), opgH_L480P (EG3369), ΔestG (EG2658), ΔestG ampicillin suppressor (EG3105), ΔestG; opgH_L480P (EG3371) grown on PYE agar alone or with 50 μg/mL ampicillin. Culture dilutions are as indicated. B. Schematic diagraming predicted topology of OpgH with grey boxes representing transmembrane domains and corresponding amino acids labeled. Asterisks represent approximate location of suppressing point mutations from the ΔCTL (EG1569) and ΔestG suppressors (EG3105). C. Spot dilutions of indicated strains on PYE agar alone or with added 0.5 mM vanillate and/or 50 μg/mL ampicillin. Strains are WT (EG865), ΔestG (EG2658), P_van-opgH (EG3375), ΔestG + P_van-opgH (EG3377), P_van-opgH_L434P (EG3577), and ΔestG + P_van-opgH_L434P (EG3579). See also Figure S4 and Data S1.

To characterize the suppressing mutation in opgH, we generated the mutation (opgH_L480P) in a clean genetic background, in the presence or absence of estG. In the absence of stress, opgH_L480P did not impact growth, but did restore ΔestG cells to WT colony size (Figure 3A). In the presence of ampicillin, opgH_L480P completely restored growth in a ΔestG background (Figure 3A). The opgH_L480P mutation in a WT background also caused moderate growth defects in the presence of ampicillin (Figure 3A).

We hypothesized that the opgH_L480P mutation might result in a loss of function, as the proline substitution is located within a predicted transmembrane domain (Figure 3B) ^25,26. To ensure that OpgH_L480P was stably produced, we assessed the levels of a 3x-Flag tagged version of the L480P mutant protein produced from the opgH locus and saw no difference compared to WT (Figure S4A). We then tested if OpgH_L480P could suppress ΔestG mutant sensitivity to stress in the presence of WT OpgH by expressing vanillate-inducible opgH_L480P. Indeed, expression of opgH_L480P suppressed ΔestG mutant sensitivity to ampicillin in a dominant fashion (Figure S4B). opgH is reported to be essential in Caulobacter according to published Tn-Seq data and our own dataset (²⁷, Data S2), and we were unable to generate an opgH deletion strain. We conclude that the OpgH_L480P mutant protein suppresses the sensitivities of the ΔestG mutant by altering OpgH activity or function.

Interestingly, in revisiting our ΔCTL suppressors, we discovered an independent suppressing mutation in opgH that restored growth in the presence of ΔCTL (Data S1). The mutant protein, OpgH_L434P (Figure 1A), also carries a leucine to proline mutation within a predicted transmembrane domain (Figure 3B). We tested if, like L480P, the L434P mutant protein could suppress the sensitivity of a ΔestG mutant in a dominant fashion. Indeed, the OpgH_L434P mutant completely restored growth of a ΔestG mutant in the presence of ampicillin (Figure 3C). Collectively, our suppressor analyses solidify a genetic link between estG and opgH.

estG and bglX have a synthetic sick phenotype

To gain further insight into EstG function, we examined estG on the Fitness Browser database ²⁸. This database includes sensitivities of a library of transposon mutants in Caulobacter to numerous environmental conditions and reports on each gene’s fitness profile. This resource reflected the ΔestG mutant sensitivities to cell wall antibiotics and also revealed genes with a similar sensitivity profile to estG when disrupted. The top hit for co-fitness with estG was an uncharacterized gene, bglX (CCNA_01162), predicted to encode a β-D-glucoside glucohydrolase. Co-fitness between estG and bglX was demonstrated by the shared sensitivities of strains with transposon insertions in these genes, the highest being sensitivity to the cell wall-targeting antibiotic carbenicillin. The BglX homolog in Pseudomonas aeruginosa cleaves glucose polymers (including OPGs) in vitro, but BglX homologs are otherwise uncharacterized ²⁹.

We tested for activity of purified Caulobacter BglX as a glucohydrolase in vitro against the reporter substrate 4-nitrophenyl-β-D-glucopyranoside (pNPG), where hydrolysis of pNPG results in a color change that can be measured over time ²⁹. BglX hydrolyzed pNPG in a concentration-dependent manner, confirming its activity as a glucohydrolase (Figure 4A), while EstG displayed no activity against pNPG (Figure S5A). In vivo, we determined that bglX is non-essential and that its loss does not affect growth or morphology (Figure 4B–C). As predicted, we found that a ΔbglX mutant shares the antibiotic sensitivities we observed in the ΔestG mutant (Figure 2D) and that BglX localizes to the periplasm (Figure S5B). Their similar sensitivity profiles indicated a possible genetic interaction between estG and bglX. When both estG and bglX were deleted (ΔestGΔbglX), cells had a growth defect when compared to WT or either single deletion mutant (Figure 4B). ΔestGΔbglX cells had a notable length defect when compared to WT or either single deletion mutant, as quantified by principle component analysis ³⁰ (Figure 4C, Figure S5C). The double deletion also yielded a lower MIC for all tested antibiotics compared to either of the single deletions, confirming a synthetic sick interaction between estG and bglX (Figure 2D).

Figure 4: BglX is a glucosidase that interacts genetically with estG.

A. 4-Nitrophenyl-β-D-glucopyranoside (pNPG) hydrolysis assay with purified BglX or GST at indicated amounts measured at OD₄₀₅. B. Growth curve and C. phase contrast images of WT (EG865), ΔestG (EG2658), ΔbglX (EG3279), and ΔestGΔbglX (EG3282). Scale bar, 2 μm. D. Spot dilutions on PYE agar with 50 μg/mL ampicillin +/− 50 mM vanillate or 0.3% xylose of WT (EG865), ΔestG (EG2658), P_van-bglX (BglX O/E, EG3384), ΔestG+ P_van-bglX (EG3385), ΔbglX (EG3279), P_xyl-estG (EG2759), and ΔbglX+ P_xyl-estG (EG3425). E. Minimum inhibitory concentrations (MIC) of WT (EG865), ΔestG (EG2658), ΔbglX (EG3279), and ΔestGΔbglX (EG3282) against peptidoglycan (PG)− and ribosome-targeting antibiotics. Measurements in μg/mL. Mec=mecillinam; Vanc=vancomycin; Amp=ampicillin; Fos=fosfomycin; Cef=cephalexin; Spec=spectinomycin; Tet=tetracycline. Asterisk (*) represents value with a secondary zone of light inhibition. F. Spot dilutions of WT (EG865), ΔestG (EG2658), ΔbglX (EG3279), and ΔestGΔbglX (EG3282) on PYE agar alone or with added 50 μg/mL ampicillin and/or 20 mM NaCl. Culture dilutions are as indicated. See also Figure S5.

From the genetic interaction between estG and bglX, we hypothesized that EstG and BglX fulfill a similar function. If so, overexpression of one of the enzymes may compensate for loss of the other. We found that, indeed, overproduction of BglX in ΔestG cells completely rescued the β-lactam sensitivity of a ΔestG mutant (Figure 4D). Surprisingly, the reverse was not true—overproduction of EstG did not compensate for loss of bglX (Figure 4D). Therefore, though there is a genetic interaction between estG and bglX, these results, and differences in their biochemical activities in vitro (Figure 4A, S5A, and below), suggest that EstG and BglX are not functionally redundant.

ΔestG and ΔbglX cells have similar sensitivities to OPG-deficient mutants in other bacteria

Inspired by the genetic links that implicated estG in the OPG pathway, we wondered whether other aspects of the ΔestG phenotype align with OPG mutants in other bacteria. In P. aeruginosa, OPG production is important for resistance to ribosome-targeting aminoglycoside antibiotics ⁹. Similarly, we found that ΔestG, ΔbglX, and ΔestGΔbglX cells all have decreased MIC values when treated with spectinomycin or tetracycline (Figure 4E). In E. coli, OPG synthesis mutants demonstrate increased sensitivity to outer membrane detergents ³¹. We therefore assessed estG and bglX mutants for sensitivity to the detergent sodium deoxycholate (NaDOC). Indeed, NaDOC impaired growth of ΔestG and ΔbglX mutants, and almost entirely inhibited growth of the double mutant (Figure S5D). These sensitivities of ΔestG and ΔbglX strains are consistent with a role for EstG and BglX in maintaining cell envelope integrity via the OPG pathway.

In some organisms, OPG production is thought to compensate for a decrease in environmental osmolarity. In low osmolarity media, OPGs in E. coli comprise up to 5% of the dry weight, while in high osmolarity media, OPGs account for as low as 0.5% of the dry weight ⁷. With our hypothesis that EstG and BglX are involved in the OPG pathway, we altered media osmolarity to assess reliance on OPGs in our mutants. When grown in complex media (peptone yeast extract (PYE)), ΔestG, ΔbglX, and ΔestGΔbglX mutants are all hypersensitive to 50 μg/mL ampicillin (Figure 4F). However, these sensitivities are almost completely alleviated when the osmolarity is increased (PYE+sodium chloride (NaCl)) (Figure 4E). We observed a similar result when Tris-HCl was provided as an osmolyte instead of NaCl (Figure S5E). This osmolarity-dependent rescue further supports a link between EstG, BglX, and OPGs, and led us hypothesize that EstG acts on OPGs.

EstG structurally resembles and functions as an esterase in vitro

To obtain more insight into a putative substrate for EstG, we determined its structure to 2.1 Å resolution using X-ray crystallography (Figure 5A, PDB ID 7UIC). The EstG final map shows well-defined density for amino acids 30 to 352 and 367 to 444 with excellent geometry (Figure 5A, Table S2). EstG is annotated as a member of the transpeptidase superfamily, and within this family are well-studied PG enzymes with an α/β hydrolase fold. EstG displays a seven stranded, antiparallel β-sheet sandwiched by the N- and the C-terminal helices in the front and other helices in the back (Figure 5A). The hydrolase domain in EstG is formed by amino acids 30 to 121 and 218 to 444 and displays two highly conserved motifs ²⁴. Motif I comprises a Ser-X-X-Lys sequence, residues 101–104 in EstG (Figure 5B) located at the beginning of helix α2, similar to the structure of EstB, a cytoplasmic esterase from Burkholderia gladioli (PDB ID 1CI8, Figure S6A). Motif II contains a highly conserved Tyr, Tyr218 in EstG, which acts as a base to activate the serine nucleophile (Figure 5B, Figure S6). Motif I and II are both located in the active site at about 2.7 Å from each other (Figure 5A and B).

Figure 5: EstG is structurally similar to esterases in the β-lactamase family.

A. The structure of EstG displays an α/β hydrolase fold. Ribbon diagram of residues 30–444 with the N-terminal residues 30 to 121 colored in orange, 122–217 colored grey, 218 to 444 in yellow. B. Zoom in of putative active site identified by homology to esterases. Ser101 (S101) of motif I is 2.7 Å away from Tyr218 (Y218) of motif II. The active site has a sulfate (SO₄) and a Tris (TRS) molecule bound. C. The structural alignment of EstG+ TRS + SO₄ (PDB ID 7UIC) with EstB bound to diisopropyl fluorophosphate (DFP) (PDB ID 1CI8 ³³, colored in light grey) displays the partial overlap of the sulfate to the phosphonate of DFP. D. p-nitrophenyl butyrate (pNB) hydrolysis of purified EstG, EstG_S101A, and GST at 10 μM measured at OD₄₀₅. Data represents technical triplicate and error bars are standard deviation. E. Spot dilutions of WT (EG865), ΔestG (EG2658), and EstG_S101A (EG2990) on PYE agar plates with 6 μg/mL cephalexin. Culture dilutions are as indicated. The S101A mutation results from T301G mutation in the chromosomal copy of estG. See also Figure S6 and Table S2.

In total, we determined three structures of EstG: EstG bound to tris (EstG+TRS), EstG bound to tris and sulfate (EstG+TRS+SO₄), and EstG bound to tris, sulfate, and tantalum bromide (EstG+TRS+SO₄+(Ta₆Br₁₂)²). The structures are very similar with a pairwise root-mean-square deviation ranging from 0.24 to 0.26 Å for amino acids 398–401 as calculated with SSM Coot ³². The binding of a SO₄ molecule close to Motif I and Motif II correlates with the presence of clear electron density for the loop 346–357 (PDB ID 7UIC, 7UIB, Figure 5B, Figure S6B–E). Structural alignment of EstG+TRS+SO₄ with EstB bound to diisopropyl fluorophosphate (DFP, PDB ID 1CI9) highlights the partial overlap between the SO₄ in EstG and the DFP bound to catalytic serine residue in EstB (Figure 5C).

In EstG, residues 122 to 217 are on top of the hydrolase fold (Figure 5A). Within it, residues 138 through 151 define an insertion of a hairpin formed by strand β4-β5 (Figure S6A,S6F–G) which is also present in the transesterase enzyme, simvastatin synthase (Aspergillus terreus LovD, PDB ID 4LCM). Notably, this hairpin is absent in EstB. Structural alignment over those deposited in the PDB highlights structural conservation among enzymes in this family. The most similar proteins to EstG by structure are esterases (EstB, PDB ID 1CI8), transesterases (LovD, PDB ID 3HLB), carboxylesterases (PDB ID 4IVK), PBP homologs (PDB ID 2QMI), and D-amino acid amidases (PDB ID 2DNS). Interestingly, D-amino acid amidases and aminohydrolases also lack the hairpin insertion described for EstG and the transesterase LovD (Figure S6F–G).

Based on the structural similarity of EstG to EstB, a cytoplasmic esterase with an unknown substrate, we sought to compare the two enzymatically. Like EstG (Figure S2E, S3), EstB has no β-lactamase or peptidase activity ³³. EstB does, however, demonstrate esterase activity which can be detected in vitro using p-nitrophenyl butyrate (pNB) as a substrate ³³. Hydrolysis of pNB causes a color change that can be measured over time. We found that EstG significantly hydrolyzed pNB as compared to the negative control (Figure 5D). We sought to create a catalytically dead variant of EstG by mutating the predicted active site serine, Ser101, to alanine (S101A; or T301G in the DNA sequence). Consistent with our prediction, the S101A mutant protein did not hydrolyze pNB in vitro (Figure 5D). We also made the S101A mutation in the chromosomal copy of estG and found that the mutant phenocopied the β-lactam sensitivity of ΔestG cells (Figure 5E). These data establish the essentiality of EstG’s enzymatic activity in protecting the cell against stress and confirms activity of EstG as an esterase.

EstG enzymatically modifies a cyclic hexasaccharide periplasmic glucan

EstG can act as an esterase in vitro and our genetic and osmolarity data implicate OPGs as a substrate. However, Caulobacter OPGs have never been characterized, and the absence of homologs of most characterized OPG-metabolizing enzymes precludes predicting which OPGs may be present. To identify the native substrate of EstG, we took an unbiased biochemical approach. We first isolated the periplast fraction of WT cells, where EstG and its substrate should both reside (Figure 6A). We hypothesized that Caulobacter OPGs might be of similar size to E. coli OPGs and therefore further fractionated to isolate components between 1 and 10 kDa. The sample was boiled to inactivate proteins, leaving sugars or other heat-resistant metabolites intact. In vitro, we combined this 1–10 kDa periplast isolate with purified EstG or the catalytically dead mutant protein, EstG_S101A. We then separated molecules in the treated periplast by high-performance liquid chromatography (HPLC) and selected for peaks that decreased in abundance when mixed with EstG, but not when mixed with EstG_S101A. Peaks of interest were then identified by mass spectrometry. Using this approach, we identified a molecule that decreased in abundance ~40% when incubated with EstG (Figure 6B, Figure S7A–B), indicating that EstG modified this substrate in some way. The mass of the parental ion indicated that the molecule resembled α-cyclodextrin (α-CD), a cyclic, hexameric glucose polymer. Notably, the MS/MS spectra for this molecule in the periplast + EstG_S101A (top half of Figure 6C), most closely matched the library spectra for α-CD (bottom half of Figure 6C). Greater than 80% of the fragmentation signal generated from our experiments matched the ion profile for α-CD. We next attempted to detect chemical modification of α-CD by EstG using the same workflow. However, due to the complexity of the periplast fraction and the small expected amount of modified α-CD, we were not successful. Though this small, cyclic sugar is a novel structure for an OPG, it is consistent with the existence of cyclic OPGs in other bacteria ⁷.

Figure 6: A cyclic hexameric glucose is a native substrate of EstG.

A. Schematic outlining the method for isolating periplasmic contents (periplast, blue) and sequential fractionation. Periplast 1–10 kDa was then combined with EstG or EstG_S101A, and contents were separated and identified with LCMS. B. A box-plot displaying the relative abundances of the cyclic hexaglycan with error bars across four technical replicates in the samples of periplast alone, periplast + EstG, or periplast + EstG_S101A. Mass of the parental ion is 973.3223 m/z. C. MS/MS spectra with the experimental spectra observed in one of the injections of the periplast + EstG_S101A of the parental ion from panel B as the top of the mirror plot. Bottom half of the mirror plot is the mzCloud reference spectra for α-cyclodextrin (α-CD). D. p-nitrophenyl butyrate (pNB) hydrolysis of purified EstG or EstG_S101A with increasing amounts of α-CD showing concentration dependent inhibition. E. Michaelis–Menten saturation curve of the rates of pNB hydrolysis with EstG or EstG + 10 mM α-CD to show competitive inhibition of the active site. Rate was determined by the slope of the pNB hydrolysis curve at the indicated pNB concentration. Rate is presented as molecules of pNB hydrolyzed per minute. Parenthesis next to values for V_max and K_m represent 95% confidence interval. See also Figure S7.

We next sought to validate α-CD as an EstG substrate in vitro. If α-CD is a substrate for EstG, addition of α-CD to the pNB hydrolysis assay should inhibit pNB hydrolysis through competition for the active site. Indeed, increasing amounts of α-CD reduced EstG’s hydrolysis of pNB in a concentration-dependent manner (Figure 6D). To confirm that α-CD competitively inhibits EstG, we measured the rate of pNB hydrolysis with increasing concentrations of pNB and a consistent amount of α-CD. For a competitive inhibitor, we expect to see a constant V_max and an increased K_m value with added α-CD. By plotting the rate of hydrolysis +/− α-CD, we found that the V_max values of the two curves were similar (Figure 6E, Figure S7C). However, the K_m values differed, at 11.2 mM without α-CD and 53.7 mM with α-CD (Figure 6E, Figure S7C). These data indicate that α-CD interacts directly with the active site of EstG and is thus structurally similar to the native substrate. Collectively, our data suggest that EstG modifies a previously uncharacterized cyclic, hexameric OPG in a novel manner, thereby contributing to cell envelope homeostasis during stress (Figure 7).

Figure 7: EstG protects the cell envelope against stress through its activity on cyclic OPG polymers.

Cell envelope homeostasis during stress is maintained through the actions of EstG and the putative OPG pathway in Caulobacter. We propose that OpgH takes cytoplasmic UDP-glucose to synthesize small, cyclic OPG molecules into the periplasm (light blue). We hypothesize that BglX hydrolyzes these OPGs and EstG adds a modification (green circles) to produce an OPG species that is protective against stress (lightning bolt). When OPG metabolism is dysregulated, cell envelope integrity is lost, resulting in hypersensitivity to a variety of environmental changes and antibiotic stresses (represented by yellow lightning bolt). OM=outer membrane, PG=peptidoglycan, IM=inner membrane.

Discussion

It is clear from our work that there are undiscovered proteins and pathways that play critical roles in maintaining cell envelope homeostasis. Our identification of EstG and its role in the Caulobacter OPG pathway suggests there might be unexplored substrates of other TPase family enzymes. We identified EstG through a Tn-Seq screen as an essential factor for surviving ΔCTL-induced cell wall stress (Figure 1). Though estG is non-essential in unstressed conditions (Figure 2), ΔestG cells are more sensitive to cell envelope stresses (Figure 2D, Figure S5D). Despite its homology to TPase family proteins, EstG does not detectably modify the PG (Figure S3, Table S1). Instead, genetic interactions with opgH (Figure 3) and bglX (Figure 4) implicate EstG in the OPG pathway. In vitro biochemistry revealed a periplasmic substrate of EstG as a cyclic hexamer of glucose, which is the first reported OPG in Caulobacter (Figure 6). Our work establishes a framework to understand the OPG metabolic pathway and its consequences in Caulobacter. We propose that OpgH synthesizes a cyclic hexamer of glucose, EstG adds a modification to create an OPG with properties that protect the cell envelope, and BglX regulates OPG abundance through its glucohydrolase activity (Figure 7).

EstG is classified as a β-lactamase family protein within the TPase superfamily, which is why it stood out as an attractive candidate in our screen. Of characterized proteins, EstG shares overall structural similarity to EstB from B. gladioli, another enzyme in the β-lactamase family. They both contain an active site serine, but EstG lacks the common esterase motif, G-X-S-X-G, present in EstB. This esterase motif is not required for EstB’s hydrolase activity, however ³³. EstG also demonstrates structural similarity to the transesterase LovD from A. terreus, which converts Monacolin J to Lovastatin by addition of a 2-methylbutyryl side chain ³⁴. Specifically, both EstG and LovD contain an insertion of a hairpin that is missing in EstB and other members of this family of enzymes (Figure S6). Based on this similarity and our genetic, biochemical, and physiological data, we hypothesize that EstG acts as a transesterase to add a functional group to an α-CD-like OPG that contributes to protecting the cell against stress (Figure 7).

Though not required under normal growth, our data demonstrate the importance of EstG acting on OPGs and implies an essential role for OPGs in stress survival. OPGs had not been previously identified in Caulobacter, though the presence of an opgH homolog in the genome was reported ⁸. OPGs in several α-proteobacteria of the orders Rhizobiales and Rhodobacterales are characterized and have a variety of structures, consisting of family II, III, and IV OPGs ⁸. These OPGs can range from 10–25 glucose monomers, but all three classes are cyclic polymers, as opposed to the linear family I OPGs found in γ-proteobacteria. We were surprised to find that the only opg gene in Caulobacter is opgH. As we report a cyclic OPG, we would expect other OPG genes responsible for cyclizing and modifying OPGs to be present in Caulobacter. Interestingly, other α-proteobacteria encode OPG metabolic enzymes that are not homologs of the opg genes in E. coli including chvA and chvB in Agrobacterium tumefaciens, ndvA and ndvB in Sinorhizobium meliloti, and cgs and cgt in Brucella abortus ⁷. These genes imply the existence of a wide variety of OPG enzymes and structures across bacteria. We propose that EstG and BglX are additional examples of enzymes with unique roles in OPG synthesis, modification, and/or hydrolysis.

Both our own characterization of the ΔestG strain and information in the Fitness Browser database ^28,35 indicated a wide range of antibiotic sensitivities. Those we tested (Figure 2D) include many classes of PG- and ribosome-targeting antibiotics such as β-lactam, glycopeptide, phosphonic, aminoglycoside, and tetracycline antibiotics (Figure 2D). Fitness Browser additionally indicated sensitivities to a DNA-gyrase-targeting antibiotic and an inhibitor of lipid A biosynthesis. Collectively, this establishes the hypersensitivity of ΔestG cells to antibiotics that target at least four different cellular processes. We looked for similarities among these drug classes but found no obvious chemical similarities. For instance, nalidixic acid and ampicillin are relatively small, while vancomycin is a large glycopeptide. Though most molecules tested were polar and uncharged, others, such as chloramphenicol and sodium deoxycholate are charged. Ultimately, these broad antibiotic sensitivities support the idea of a global cell envelope defect resulting from loss of EstG activity, and not a sensitivity specific to a particular molecular feature.

Mutants of OPG enzymes in diverse bacteria typically have pleiotropic phenotypes, including those discussed for estG and bglX mutants as well as defects in motility, biofilm formation, and/or virulence ⁷. Despite the impact of OPGs on important cellular behaviors and properties, we do not know the mechanism(s) behind OPG-mediated effects. One model suggests OPGs function as osmoprotectants by establishing a Donnan equilibrium across the outer membrane. The idea is that production of negatively charged OPGs in the periplasm creates a high concentration of fixed, charged molecules that cannot cross the outer membrane. The accumulation of charged OPGs attracts counterions to the periplasm, and maintains a Donnan membrane potential across the outer membrane, allowing for iso-osmolarity of the periplasm and cytoplasm ^36,37. The Donnan potential has also been suggested to play a role in permeability of the envelope to antibiotics ³⁸. These mechanisms, however, presume that OPGs are highly charged, which is not the case in all bacteria ⁷. Though we were not able to determine the exact EstG-mediated modification on Caulobacter OPGs, it is possible that EstG adds a charged moiety in order to mediate the Donnan potential and protect the cell envelope.

Our data suggest that the activities of EstG and BglX on Caulobacter OPGs contribute to osmoprotective properties (Figure 4E). It is possible that more mechanistic insight can be revealed by comparison to OPG pathways in other organisms. For instance, the E. coli OpgH enzyme links nutrient availability with cell size by inhibiting FtsZ when UDP-glucose levels are high ³⁹. This is likely not a conserved function of Caulobacter OpgH, as it lacks most of the N-terminal FtsZ-interacting region. Suppressor mutations within opgH have also been identified in E. coli that implicate OpgH with envelope homeostasis. A nonsense mutation in opgH was isolated in a lipopolysaccharide (LPS) mutant that together conferred resistance to envelope stressors ⁴⁰. Due to the integral role of LPS in outer membrane integrity, it was proposed that either the lack of OPGs or loss of OpgH reduces membrane permeability to antibiotics, thus conferring resistance. However, unlike the opgH suppressing mutations identified in this study, the E. coli opgH nonsense mutation was recessive to WT. Though this suggests a different mechanism of suppression, it does not rule out a role for Caulobacter OpgH in regulating membrane permeability. Two spontaneous opgH mutants were also isolated in Vibrio cholerae that suppressed the hyperosmotic lethality of a lytic transglycosylase (LTG) mutant ⁴¹. This model suggested that LTG mutants inadequately recycle PG products, resulting in excessive periplasmic crowding ⁴¹. Additional production of OPGs exacerbated this periplasmic crowding⁴¹. Unlike for LTGs, our identification of an OPG substrate for EstG indicates a direct link to OPG metabolism, rather than a consequence of molecular crowding. An important avenue for future work includes functional studies of OpgH, BglX, and EstG, as well as comprehensive determination of the structures and modifications of Caulobacter OPGs. These insights can ultimately bridge our gap in understanding of the mechanistic role of OPGs in the Caulobacter envelope.

STAR Methods

RESOURCE AVAILABILITY

Lead contact

Further information and requests for resources and reagents should be directed to and will be fulfilled by the lead contact, Erin Goley (egoley1@jhmi.edu).

Materials availability

Plasmids, strains, and antibodies generated in this study will be provided upon requests sent to the lead contact.

Data and code availability

• The accession number for the Tn-seq data reported in this paper is BioProject: PRJNA820779 and accession numbers SRX14637780, SRX14637779, and SRX14637778 and can be accessed at https://www.ncbi.nlm.nih.gov/bioproject/?term=PRJNA820779. The final coordinates of EstG bound to TRS, EstG bound to SO₄ and TRS, EstG bound to (Ta₆Br₁₂)² have been deposited in the PDB with accession codes 7UDA, 7UIC and 7UIB respectively. All LCMS data has been uploaded to the MASSIVE public repository ⁶⁶ under accession MSV000089142.

• This paper does not report original code.

• Any additional information required to reanalyze the data reported in this paper is available from the lead contact upon request.

EXPERIMENTAL MODEL AND SUBJECT DETAILS

Strains

All strains, plasmids, and primers used in this study can be found in Methods S1.

C. crescentus NA1000 cells were grown at 30°C in peptone-yeast extract (PYE) medium. E. coli Rosetta(DE3)/pLysS cells were grown at 30°C in Luria-Bertani (LB) medium. Xylose or glucose were used at concentrations of 0.3% (w/v) for induction experiments. Antibiotics were used in liquid (solid) medium at the following concentrations for Caulobacter growth: gentamycin, 1 (5) μg/mL; kanamycin, 5 (25) μg/mL; spectinomycin, 25 (100) μg/mL. Streptomycin was used at 5 μg/mL in solid medium. E. coli antibiotics were used in liquid (solid) medium as follows: ampicillin, 50 (100) μg/mL; gentamicin, 15 (20) μg/mL; kanamycin, 30 (50) μg/mL; and spectinomycin, 50 (50) μg/mL. For growth curves, a Tecan Infinite M200 Pro plate reader measured absorbance every 30 minutes at OD₆₀₀ of a 100 μL culture volume in a 96 well plate in biological triplicate with intermittent shaking. For spot dilution assays, mid-log cells were diluted to an OD₆₀₀ of 0.05 and serially diluted up to 10⁻⁶ before spotting 5 μL of each dilution onto a PYE plate with indicated inducer and/or antibiotic. Plates were incubated at 30°C for 48 hours, or until the appearance of colonies at the lowest dilution in the control strain. To determine the minimum inhibitory concentration (MIC), mid-log phase cells were diluted to OD₆₀₀ of 0.5 and 200 μL were spread out onto a PYE plate. Antibiotic strips with increasing concentration of antibiotic were placed on the dried plate, inverted, and grown at 30°C for 48 hours. Some MIC values were estimated by loss of growth on plates with a range of antibiotic added to the media.

METHOD DETAILS

Plasmid construction

Plasmids (Methods S1) can be found in the supplemental information. Plasmid construction involved standard molecular cloning procedures including PCR, restriction digestion, and ligation or NEBuilder HiFi DNA Assembly Cloning Kit (NEB). Mutagenesis to generate blaB_S101A was performed using a QuikChange Lightning Multi Site-Directed Mutagenesis Kit (Agilent Genomics).

Atypical strain construction

We were unable to generate the following strains in low osmolarity PYE media, so they were constructed in M2G minimal media: EG3116 (ΔCTL+ΔestG), EG3369 (opgH_L480P), EG3371 (ΔestG+OpgH_L480P), and EG3377 (P_van-opgH). For a 500 mL batch of M2G plates, 465 mL of water and 7.5 g agar (1.5%) were autoclaved. Once cooled, 25 mL of 5x M2 salts, 500 μL of 500 mM MgSO₄, 500 μL of 10 mM FeSO₄ 10 mM EDTA (Sigma F-0518), and 0.3% glucose were added. Additional antibiotics or media supplements needed for selection were also added at this time.

Phase-contrast microscopy

Exponential phase cells were spotted on 1% agarose pads and imaged using a Nikon Eclipse Ti inverted microscope equipped with a Nikon Plan Fluor 100X (NA1.30) oil Ph3 objective and Photometrics CoolSNAP HQ² cooled CCD camera. Images were processed using Adobe Photoshop.

Cell shape analysis

Binary masks of phase contrast images of log phase cells were inputted into Celltool ³⁰. The shape mode of interest was plotted as single data points. Prism was used for graphing calculated terms.

Suppressor screening and whole genome sequencing

For the ΔCTL suppressor screen, Caulobacter strains EG937 or EG1214 strains were inoculated from individual colonies and grown overnight in PYE media (with no inducer) until stationary phase. Cells were plated on PYE agar plates containing 0.3% (w/v) xylose to induce ΔCTL expression and incubated at 30°C until the appearance of colonies (suppressors). Suppressors were tested for growth in PYE media with 0.3% xylose overnight. Immunoblotting with FtsZ-antiserum was used to confirm xylose-induced ΔCTL expression. Genomic DNA was extracted from suppressors using Qiagen DNeasy Blood and Tissue Kit. Mutations were identified from MiSeq analysis of genomic DNA from suppressor strains. Spontaneous suppressors of ΔestG were isolated by plating ΔestG (EG2658) on PYE+100 μg/mL ampicillin and isolating resistant colonies. Resistance was confirmed by spot dilution on plates containing 50 μg/mL ampicillin. Genomic DNA was extracted from suppressors using Qiagen Dneasy Blood and Tissue Kit and sent to Microbial Genome Sequencing Center (MiGS) for whole genome sequencing and BreSeq analysis.

Transposon library preparation, sequencing, and analysis

Transposon libraries were prepared, sequenced, and analyzed using the same methods as previously described in Woldemeskel et al. and Lariviere et al. (Woldemeskel et al., 2020; Lariviere et al., 2019). Tn-Seq libraries were generated for WT (EG865), RelA’ (EG1799) and ΔCTL+RelA’ (EG1616). 1L PYE cultures were harvested at OD₆₀₀ of 0.4–0.6, washed 5 times with 10% glycerol, and electroporated with the Ez-Tn5 <Kan-2> transposome (Epicentre, Charlotte, North Carolina). Cells recovered at 30°C shaking for 90 minutes, and plated on PYE-Kan plates. The RelA’ library was plated on PYE-Kan with gentamycin and 0.003% xylose to induce RelA’ expression. ΔCTL+RelA’ library was plated on PYE-Kan plates with spectinomycin, streptomycin, gentamycin, and 0.003% xylose to induce RelA’ and ΔCTL. Colonies were scraped off plates, combined, resuspended to form a homogeneous solution in PYE, and flash frozen in 20% glycerol. The Dneasy Blood and Tissue Kit (Qiagen, Hilden, Germany) was used to extract genomic DNA from each pooled library. Libraries were prepared for Illumina Next-Generation sequencing through sequential PCR reactions. The initial PCR round used arbitrary hexamer primers with a Tn5 specific primer going outward. The second round used indexing primers with unique identifiers to filter artifacts arising from PCR duplicates. Indexed libraries were pooled and sequenced at the University of Massachusetts Amherst Genomics Core Facility on the NextSeq 550 (Illumina, San Diego, California).

Sequencing reads were first demultiplexed by index, each library was concatenated and clipped of the molecular modifier added in the second PCR using Je ⁴³:

java -jar /je_1.2/je_1.2_bundle.jar clip F1 = compiled.gz LEN = 6

Reads were then mapped back to the Caulobacter crescentus NA1000 genome (NCBI Reference Sequence: NC_011916.1) using BWA ⁴⁴ and sorted using Samtools ⁴⁵:

bwa mem -t2 clipped.gz | samtools sort -@2 - > sorted.bam

Duplicates were removed using Je ⁴³ and indexed with Samtools ⁴⁵ using the following command:

java -jar /je_1.2/je_1.2_bundle.jar markdupes I = sorted.bam O = marked.bam M = METRICS.txt MM = 0 REMOVE_DUPLICATES = TRUE
samtools index marked.bam

The 5’ insertion site of each transposon were converted into .wig files comprising counts per position and visualized using Integrative Genomics Viewer (IGV) ^46,47. Specific hits for each library were determined with coverage and insertion frequency using a bedfile containing all open reading frames from NC_011916.1 and the outer 20% of each removed to yield a clean and thorough insertion profile. This was determined using BEDTools ^48,49 and the following commands:

bedtools genomecov -5 -bg marked.bam > marked.bed
bedtools map -a NA1000.txt -b marked.bed -c 4 > output.txt

Tn-Seq data have been deposited in the Sequence Read Archive (SRA) under accession numbers listed in the Data and Code Availability section.

Protein purification

All purified proteins were overproduced in E. coli Rosetta(DE3)pLysS from the following plasmids: His₆-EstG-His₆, pEG1622; His₆-EstG_S101A-His₆, pEG1706; His₆-EstA, pEG1950; His₆-BglX-His₆, pEG1779. Cells were induced with 1mM IPTG for 4 hours at 30°C. Cell pellets were resuspended in Column Buffer A (50 mM Tris-HCl pH 8.0, 300 mM NaCl, 10% glycerol, 20 mM imidazole, 1 mM β-mercaptoethanol) flash frozen in liquid nitrogen and stored at −80°C. To purify the His-tagged proteins, pellets were thawed at 37°C, and 10 U/mL DNase I, 1 μg/mL lysozyme, and 2.5 mM MgCl₂ were added. Cell slurries were left on ice and occasionally inverted for 45 minutes, then sonicated and centrifuged for 30 minutes at 15,000 × g at 4°C. The protein supernatant was then filtered and loaded onto a pre-equilibrated HisTrap FF 1mL column (Cytiva, Marlborough, Massachusetts). The His-tagged proteins were eluted in 30% Column Buffer B (same as Column Buffer A but with 1M imidazole, adjusted to pH 8). Peak fractions were concentrated and applied to a Superdex 200 10/300 GL (Cytiva) column equilibrated with EstG storage buffer (50 mM HEPES-NaOH pH 7.2, 150 mM NaCl, 10% glycerol, 1 mM β-mercaptoethanol (BME)). Peak fractions were combined, concentrated, and snap-frozen in liquid nitrogen and stored at −80°C.

Immunoblotting

Purified His₆-EstG-His₆ was dialyzed into PBS and used to immunize a rabbit for antibody production (Pocono Rabbit Farm & Laboratory, Canadensis, Pennsylvania). To affinity purify the EstG antisera, His₆-EstG-His₆ in EstG storage buffer was coupled to Affigel 10 resin (Bio-Rad, Hercules, California). After washing the resin 3 times with cold water, add approximately 10 mg of protein to 1 mL of Affigel 10 resin to rotate at 4°C for 4 hours. 75 mM Tris pH 8.0 was added to terminate the reaction and left to rotate at 4°C for 30 minutes. EstG-resin was washed in a column with the following cold reagents: 10 mL EstG storage buffer, 15 mL Tris-buffered saline (TBS), 15 mL 0.2 M glycine-HCl pH 2.5 with 150 mM NaCl, 15 mL TBS, 15 mL guanidine-HCl in TBS, and 20 mL TBS. EstG antisera was combined with EstG-resin, and incubated, rotating, overnight at 4°C. Unbound sera flowed through the column and was washed with 25 mL TBS, 25 mL TBS with 500 mM NaCl and 0.2% Triton X-100, and a final wash of 25 mL TBS. Bound Anti-EstG was eluted with 0.2 M glycine pH 2.5 and 150 mM NaCl, dialyzed into TBS, and diluted 1:1 with glycerol. Anti-EstG antibody specificity was validated by western blot to recognize a band in wild type lysate that is absent in a ΔestG mutant.

Western blotting was performed using standard lab procedures. Cells in log phase were isolated and lysed in SDS-PAGE loading buffer and boiled for 10 minutes. For a given experiment, equivalent OD units of cell lysate were loaded. SDS-PAGE and transfer of protein to nitrocellulose membrane were performed using standard procedures. Antibodies were used at the following concentrations: Antibodies used were EstG-1:1000; SpmX-1:10,000 ⁵⁰; Flag-1:1,000 (Sigma, St. Louis, Missouri); CdnL-1:2,500 ¹² and 1:10,000 of HRP-labeled α-rabbit (for EstG, SpmX, and CdnL) or α-mouse (for Flag) secondary antibody (PerkinElmer) on nitrocellulose membranes. Chemiluminescent substrate (PerkinElmer) was added to facilitate protein visualization via an Amersham Imager 600 RGB gel and membrane imager (GE).

In vitro pNB hydrolysis or pNPG assay

To test for serine hydrolase activity using p-nitrophenyl butyrate (pNB, Sigma), indicated proteins were used at 10 μM in a 50 μL reaction containing 50 mM Tris-HCl pH 8. pNB was added last to the samples at a concentration of 4 μM. Absorbance was measured every minute at 405 nm for 10 minutes. To test for glucosidase activity using 4-Nitrophenyl-β-D-glucopyranoside (pNPG, Sigma), indicated proteins were used at listed concentrations in a 50 μL reaction containing 50 mM Tris-HCl pH 8. pNPG was added last at a final concentration of 4 μM. Absorbance was measured every minute at 405 nm for 10 minutes.

Nitrocefin hydrolysis assay

To assess β-lactamase activity through hydrolysis of nitrocefin, 10 μM of indicated proteins were mixed with 100 μM nitrocefin (Calbiochem, Sigma) in a reaction buffer containing EstG storage buffer (50 mM HEPES-NaOH pH 7.2, 150 mM NaCl, 10% glycerol, 1 mM BME) to a final volume of 100 μL. Absorbance was measured at 492 nm every 10 minutes for 4 hours.

Sacculi purification and PG binding assay

Sacculi for PG binding assay were prepared as previously described in Meier et al ⁵¹. Wild type (EG865) Caulobacter cells were grown in 1L of PYE at 30°C to an OD₆₀₀ of 0.5. Cells were pelleted by centrifugation at 6,000 × g for 10 minutes and resuspended in 10 mL of 1X PBS. The cells were added dropwise to a boiling solution of 4% SDS where they were continuously mixed and boiled for 30 minutes, then incubated overnight at room temperature. Sacculi were pelleted by ultracentrifugation at 42,000 × g in an MLA-80 rotor for 1 hour at 25°C and remaining pellet was washed four times with ultra-pure water with a final resuspension in 1 mL PBS with 20 μL of 10 mg/mL amylase, left at room temperature overnight. Then the sacculi were pelleted at 90,000 × g in an MLA-130 rotor for 15 minutes at 25°C and washed three times with ultra-pure water, with a final resuspension in 1 mL of PG binding buffer (20 mM Tris-HCl pH 6.8, 1 mM MgCl₂, 30 mM NaCl, 0.05% Triton X-100). To each reaction, 6 μg of each protein was added to either PG or buffer. Reactions were left on ice for 30 minutes and then centrifuged for 30 minutes at 90,000 × g in the MLA-130 rotor at 4°C. Supernatant was saved and the pellet was resuspended in PG binding buffer and saved as the PG bound isolate. SDS-PAGE loading dye was added to a final concentration of 1X to each sample and run on an SDS-PAGE gel, Coomassie stained, and imaged.

PG purification and analysis

PG samples were analyzed as described previously ^52,53. In brief, samples were boiled in SDS 5% for 2 h and sacculi were repeatedly washed with MilliQ water by ultracentrifugation (110,000 × g, 10 min, 20°C). The samples were treated with muramidase (100 μg/mL) for 16 hours at 37°C. Muramidase digestion was stopped by boiling and coagulated proteins were removed by centrifugation (10 min, 22,000 × g). The supernatants were first adjusted to pH 8.5–9.0 with sodium borate buffer and then sodium borohydride was added to a final concentration of 10 mg/mL. After reduction during 30 min at room temperature, the samples pH was adjusted to pH 3.5 with orthophosphoric acid. UPLC analyses of muropeptides were performed on a Waters UPLC system (Waters Corporation, USA) equipped with an ACQUITY UPLC BEH C18 Column, 130 Å, 1.7 μm, 2.1 mm × 150 mm (Waters, USA) and a dual wavelength absorbance detector. Elution of muropeptides was detected at 204 nm. Muropeptides were separated at 45°C using a linear gradient from buffer A (formic acid 0.1% in water) to buffer B (formic acid 0.1% in acetonitrile) in an 18-minute run, with a 0.25 mL/min flow.

To test the activity of EstG against cell wall substrates, sacculus or purified muropeptides were used as substrate. Reactions were performed in triplicates and contained 10 μg of purified enzyme, 50 mM Tris-HCl pH 7.5, 100 mM NaCl, and 10 μg of purified Caulobacter sacculus or 5 μg of purified M4, M5, D44 or D45, in a final 50 μL reaction volume. Reactions were incubated at 37°C for 24 h, then heat inactivated (100°C, 10 min) and centrifuged (22,000 × g, 15 min), for separation of soluble and pellet fractions. Soluble fractions were adjusted to pH 3.5. Pellet fractions were resuspended in water and further digested with muramidase for 16 h at 37°C. Muramidase reactions were reduced and adjusted to pH 3.5 as explained before. Both soluble and muramidase digested samples were run in the UPLC using the same PG analysis method described above.

Relative total PG amounts were calculated by comparison of the total intensities of the chromatograms (total area) from three biological replicas normalized to the same OD600 and extracted with the same volumes. Muropeptide identity was confirmed by MS/MS analysis, using a Xevo G2-XS QTof system (Waters Corporation, USA). Quantification of muropeptides was based on their relative abundances (relative area of the corresponding peak) normalized to their molar ratio. The program GraphPad PRISM^® Software (Inc., San Diego, California, www.graphpad.com) was used for all statistical analyses. To determine the significance of the data, the t-test (unpaired) was performed.

Crystallography, Data Collection, Structure Determination and Refinement

EstG protein purified for crystallography was prepared the same way as described above, with the exception of the storage buffer changed to 50 mM HEPES-NaOH pH 7.2, 150 mM NaCl, 1 mM DTT. Crystals of wild type EstG were grown by vapor diffusion in hanging drops set up with a Mosquito LCP robot (SPT Labtech, Melbourn, United Kingdom). Crystal growth was monitored using a crystallization imager ROCKIMAGER (Formulatrix, Bedford, Massachusetts). High quality crystals grew with a reservoir solution containing 20% PEG500 MME, 10% PEG20000, 0.1 M Tris/Bicine pH 8.5 and 90 mM mixture of sodium nitrate, sodium phosphate dibasic and ammonium sulfate (called EstG+SO₄+TRS) or 20% PEG500 MME, 10% PEG20000, 0.1 M Tris/Bicine pH 8.5 and 100 mM mixture of DL-Alanine, Glycine, DL-Lysine and DL-Serine, (called EstG+TRS). Crystals grown in the first condition were soaked in 500 mM Tantalium bromide heavy metal solution for 1 hour (crystals called EstG + TaBr). Crystals were flash-cooled in mother liquor. Data of crystals of EstG +TRS (PDB ID 7UDA) were collected at National Synchrotron Light Source-II at beamline 17-ID-2 (FMX) on a Dectris EIGER X 16M while crystals of EstG in complex with SO₄ and TRS (PDB ID 7UIC, EstG+SO₄+TRS) and of EstG bound to tantalum bromide (PDB ID 7UIB, EstG+TaBr) were collected at 17-ID-1 (AMX) on a Dectris EIGER X 9M detector. Diffraction data were collected on a vector defined along the longest axis of the crystal ⁵⁴. The datasets were indexed, integrated, and scaled using fastdp, XDS, and aimless ⁵⁵. All EstG crystals belong to tetragonal space group and diffracted from 2.09 to 2.62 Å.

Since the N-terminal and C-terminal sequence of EstG differed from available homologous proteins, a model of EstG to use in molecular replacement was generated with the RoseTTAFold package ⁵⁶. RoseTTAFold model weights as of July 16, 2021, UniRef30 clusters as of June 2020, PDB templates as of March 3, 2021, and the BFD ⁵⁷ were used during model prediction. A C-terminal segment (Pro443-Arg462) that was predicted to extend as a random coil away from the molecular envelope was truncated from the model with the lowest predicted coordinate error to generate the final molecular replacement search model. The structure of EstG was determined by molecular replacement using PHASER ⁵⁸ with the RoseTTAFold model of EstG as a search model ⁵⁶. The data were refined to a final resolution of 2.47, 2.09 and 2.62 Å using iterative rounds of refinement with REFMAC5 ⁵⁹ and manual rebuilding in Coot ³². Structures were validated using Coot ³² and the PDB deposition tools. Each of the three models have more than 95 % of the residues in the preferred regions according to Ramachandran statistics (Table S2). Figures were render in PyMOL (v2.2.3, Schrödinger, LLC).

Comparison with other beta lactamase binding proteins

A search using PDBeFOLD ⁶⁰ was conducted using EstG as a search model. Among them carboxyesterases, penicillin binding protein EstY29, and simvastatin synthase (PDBs 4IVK ⁶¹, 4P87 ⁶², 3HLB ⁶³) aligned with root-mean-square deviations of 1.39, 1.62, 1.82 Å over 404, 387 and 400 amino acids, respectively. The structure of EstG was used to analyze and display the primary, secondary and quaternary structure of homologous proteins with ENDscript ⁶⁴.

Cell fractionation

Cells were fractionated into periplasm and spheroplast using the previously described methods in Judd et al, except that 2 μg/mL lysozyme was used ⁶⁵. Briefly, 1 liter of cells were grown at 30° to an OD₆₀₀ of 0.5 in PYE. Cells were pelleted at 6,000 × g for 10 minutes and the supernatant removed. The pellet was resuspended in 10 mL of periplasting buffer (50 mM Tris-HCl pH 8.0, 18% sucrose, and 1 mM CaCl₂) and then 2 μg/mL of lysozyme and 1 mM EDTA was added. Contents were left on ice for 30 minutes and then spun at 3,140 × g for 5 minutes. The supernatant (periplast fraction) was carefully removed to a fresh tube, and filtered through a 0.22 μm filter. The filtered periplast fraction was fractionated using a 10 kDa concentrator (Millipore Amicon) and the flow through (periplasm contents less than 10 kDa) was saved. This was then concentrated using a 1 kDa cutoff concentrator until the final volume is approximately 1 mL. The 1 mL fraction was then boiled at 100°C for 10 minutes. This final periplast 1–10 kDa fraction was stored at 4°C until ready to use. This 1–10 kDa periplast fraction was incubated with 10 μM of indicated proteins (EstG or EstG_S101A), and left on ice for 30 minutes prior to LCMS analysis.

LCMS Analysis

All analysis was performed on a Dionex UHPLC and Q Exactive quadrupole Orbitrap system (Thermo Fisher, Waltham, Massachusetts). Two microliters of each reaction and unreacted input was injected directly onto a HyperSil Gold C-18 2.1mm × 150mm reversed phase chromatography column. Analytes were separated using an increasing gradient that consisted of 0.1% formic acid in LCMS grade water as buffer A and 0.1% formic acid in LCMS grade acetonitrile as buffer B. Due to the hydrophilic nature of glucans, the gradient began with a 2-minute acquisition at 100% buffer A with a rapid ramp to 100% buffer B by 15 minutes before returning to baseline conditions for the remainder of the 20 minute experiment. The Q Exactive was operated in positive ionization mode using a data dependent acquisition method. An MS1 scan was acquired at 140,000 resolution with a scan range of 150 to 1500 m/z. The three most abundant ions from each MS1 scan were isolated for fragmentation using a three-step collision energy of 10, 30 and 100 and the fragment scans were obtained using 15,000 resolution. Ions with unassigned charge states or more than 3 charges were excluded from fragmentation. To prevent repeat fragmentation any ion within 5 ppm mass deviation of the selected ion was excluded from additional fragmentation for 30 seconds. The complete LCMS method in vendor .meth format and a text adaptation have been uploaded to LCMSMethods.org under the following DOI (dx.doi.org/10.17504/protocols.io.36wgq7djkvk5/v1). All Thermo .RAW instrument files have been uploaded to the MASSIVE public repository ⁶⁶ under accession MSV000089142. The vendor .RAW files and processed results can be accessed during the review process using the following link: ftp://MSV000089142@massive.ucsd.edu and reviewer password EstG725.

LCMS Data Analysis

All downstream data analysis was performed with Compound Discoverer 3.1 and Xcalibur QualBrowser 2.2 (Thermo Fisher). Briefly, all MS1 ions with a signal to noise of greater than 10:1 from the vendor .RAW files were considered for downstream analysis. The LCMS files were chromatographically aligned using an adaptive curve on all ions within a maximum mass shift of 2 minutes and with less than a 5 ppm mass discrepancy. The files were also normalized to compensate for concentration and loading differences between samples using a constant mean normalization. Ion identities were assigned using the mzCloud and ChemSpider databases using a maximum mass tolerance of 5ppm against library entries. In addition, a similarity search algorithm and custom compound class scoring module were used to flag ions that exhibited common glucose ions following fragmentation. Compounds of interest were flagged in the resulting output report by use of custom filter that eliminated ions that were of decreased abundance in the EstG reacted periplasm relative to both the unreacted periplast fraction and the periplast fraction treated with the EstG_S101A.

LCMS results

A total of 1,166 LCMS features were identified in the study. After removal of background signal and ions with an m/z of less than 600, 13 prospective ions were identified that appeared to be downregulated following incubation with the EstG protein. Of these molecules only one possessed a fragmentation pattern consistent with a glucan polymeric structure. This ion demonstrated an exact match by mass and an 83.7% fragment similarity to the cyclic hexasaccharide α-cyclodextrin. Figure 6C is a mirror plot that demonstrates the level of fragment sequence match between the fragmentation of this ion and α-cyclodextrin.

QUANTIFICATION AND STATISTICAL ANALYSIS

Cell shape analysis was performed using Celltool, as indicated in the Method Details section. Tn-Seq analysis using Je, BWA, Samtools, BEDTools, Integrative Genomics Viewer, and the edgeR package in the Bioconductor suite, as described in the Method Details section. Crystallography data is rendered from PyMOL (v2.2.3, Schrödinger, LLC). Crystallography data used PHASER, REFMAC5, Coot and the PDB deposition tools. LCMS analysis completed using Compound Discoverer 3.1 and Xcalibur QualBrowser 2.2 (Thermo Fisher). Prism GraphPad was used perform statistical analyses. Crystallography data was outlined based on Ramachandran statistics (Table S2). Information regarding individual statistical test parameters can be found in the figure legends.

Supplementary Material

1

Data S1. Suppressors of ΔCTL lethality and ΔestG ampicillin resistance. Related to Figures 1 and 3. A) Extragenic suppressors of ΔCTL-induced lethality. B) Extragenic suppressors of ΔestG ampicillin sensitivity.

2

Data S2. Tn-Seq results for ΔCTL suppression by ppGpp. Related to Figure 1. A) Transposon insertion frequency and ratio of insertion frequency comparing WT to ΔCTL+RelA’. B) Transposon insertion frequency and ratio of insertion frequency comparing RelA’ to ΔCTL+RelA’.

3

Methods S1. Strains and plasmids used in this study. Related to STAR Methods. A) Caulobacter strains used in this study. B) Plasmids used in this study.

KEY RESOURCES TABLE.

REAGENT or RESOURCE	SOURCE	IDENTIFIER
Antibodies
α-EstG primary antibody	This study	N/A
α-SpmX primary antibody	50	N/A
α-Flag primary antibody	Sigma	Cat# F3165; RRID:AB_259529
α-CdnL primary antibody	12	N/A
α-rabbit secondary antibody	PerkinElmer	Cat# NEF812001EA; RRID:AB_2571640
α-mouse secondary antibody	PerkinElmer	Cat# NEF822001EA; AB_2650498
Bacterial and virus strains
NEB Turbo Competent E. coli	NEB	Cat# C2984I
XL10-Gold Ultracompetent Cells	Agilent	Cat# 210513
Chemicals, peptides, and recombinant proteins
D-(+)-Xylose	Sigma-Aldrich	Cat# X1500-500G
DEXTROSE ANHYDROUS (Glucose)	Thermo Fisher Scientific	Cat# BP350500
IPTG	Thermo Fisher Scientific	Cat# 15529019
Western Lightning Plus-ECL, Enhanced Chemiluminescence Substrate	PerkinElmer	Cat# NEL103E001EA
MECILLINAM MEC 0.016-256 Test Strips	Liofilchem	Cat# 92017
VANCOMYCIN VA 0.016-256 Test Strips	Liofilchem	Cat# 92057
FOSFOMYCIN FOS 0.016-256 Test Strips	Liofilchem	Cat# 92078
SPECTINOMYCIN SPEC 0.064-1024 Test Strips	Liofilchem	Cat# 920141
TETRACYCLINE TET 0.016-256 Test Strips	Liofilchem	Cat# 921141
Ampicillin	Fisher Scientific	Cat#BP1760-25
Cephalexin	Millipore-Sigma	Cat#C4895
Lysozyme	Millipore-Sigma	Cat#L6876
DNAse I	New England Biolabs	Cat#M0303L
4-Nitrophenyl butyrate pNB	Millipore-Sigma	Cat# N9876
4-nitrophenyl-β-D-glucopyranoside pNPG	Millipore-Sigma	Cat# N7006
α-cyclodextrin α-CD	Millipore-Sigma	Cat# C4642
Nitrocefin	Millipore-Sigma	Cat# 484400
Vanillate	Millipore-Sigma	Cat#94770
Deposited data
Tn-Seq Sequencing Data	BioProject	Bioproject: PRJNA820779
Crystal Structure	PDB	PDB: 7UDA, 7UIC and 7UIB
Experimental models: Organisms/strains
Strains used in this study	See table S5	N/A
Oligonucleotides
Oligonucleotides used in this study	See Methods S1	N/A
Recombinant DNA
Plasmids used in this study	See Methods S1	N/A
Software and algorithms
GraphPadPrism	GraphPad Software	RRID:SCR_002798
Adobe Photoshop	Adobe	RRID:SCR_014199
Celltool	30	http://zplab.wustl.edu/celltool/
Je	43	https://gbcs.embl.de/portal/tiki-index.php?page=Je
BWA	44	http://bio-bwa.sourceforge.net
Samtools	45	http://www.htslib.org/
Integrative Genomics Viewer	47	https://software.broadinstitute.org/software/igv/
BEDTools	48,49	https://bedtools.readthedocs.io/en/latest/
Bioconductor	48	https://www.bioconductor.org/
Adobe Illustrator	Adobe	RRID:SCR_010279
Nikon Elements Imaging Software	Nikon	https://www.nikonmetrology.com/en-us/product/nis-elements-microscope-imaging-software
RoseTTAFold	56	https://www.ipd.uw.edu/2021/07/rosettafold-accurate-protein-structure-prediction-accessible-to-all/
BFD	57	https://www.nature.com/articles/s41467-018-04964-5
PHASER	58	https://www.phaser.cimr.cam.ac.uk/index.php/Phaser_Crystallographic_Software
REFMAC5	59	https://www.ccp4.ac.uk/html/refmac5.html
PDBeFOLD	60	ebi.ac.uk/msd-srv/ssm/
ENDscript	64	https://endscript.ibcp.fr/ESPript/ENDscript/
Coot	32	https://www2.mrc-lmb.cam.ac.uk/personal/pemsley/coot/

Highlights:

• Cells lacking EstG are sensitive to an array of antibiotic and other stresses

• EstG has esterase activity and acts on a cyclic osmoregulated periplasmic glucan

• Mutations in OpgH rescue antibiotic sensitivity of the ΔestG mutant

• BglX is involved in the OPG pathway and impacts envelope homeostasis

Inclusion and Diversity

One or more authors of this paper self-identifies as an underrepresented ethnic minority in their field of research or geographical location. One or more of the authors of this paper self-identifies as a gender minority in their field of research. One or more of the authors of this paper self-identifies as a member of the LGBTQIA+ community. One or more authors of this paper received support from a program designed to increase minority representation in their field of research.