Peptidoglycan precursor synthesis along the sidewall of pole-growing mycobacteria

Rod-shaped mycobacteria expand from their poles, yet d-amino acid probes label cell wall peptidoglycan in this genus at both the poles and sidewall. We sought to clarify the metabolic fates of these probes. Monopeptide incorporation was decreased by antibiotics that block peptidoglycan synthesis or l,d-transpeptidation and in an l,d-transpeptidase mutant. Dipeptides complemented defects in d-alanine synthesis or ligation and were present in lipid-linked peptidoglycan precursors. Characterizing probe uptake pathways allowed us to localize peptidoglycan metabolism with precision: monopeptide-marked l,d-transpeptidase remodeling and dipeptide-marked synthesis were coincident with mycomembrane metabolism at the poles, septum and sidewall. Fluorescent pencillin-marked d,d-transpeptidation around the cell perimeter further suggested that the mycobacterial sidewall is a site of cell wall assembly. While polar peptidoglycan synthesis was associated with cell elongation, sidewall synthesis responded to cell wall damage. Peptidoglycan editing along the sidewall may support cell wall robustness in pole-growing mycobacteria.

Model, rod-shaped organisms such as Escherichia coli and Bacillus subtilis elongate across a 3 broad swath of the cell (1, 2). Mycobacterial cells, by contrast, extend from narrower polar 4 regions (3-9). Circumscription of growth to discrete zones poses spatial challenges to the 5 bacterial cell. For example, if polar growth and division are the only sites of cell wall synthesis in 6 mycobacteria, the entire lateral surface of the cell must be inert (3,(10)(11)(12). Such an expanse of 7 non-renewable surface could leave the cell vulnerable to environmental or immune insults. Since cell wall peptidoglycan synthesis is critical for bacterial replication, it is often used to 1 0 localize the sites of growth and division. Intriguingly, D-amino acid probes, which in other 1 1 species have been shown to incorporate into peptidoglycan (1,11,13), label both the poles and 1 2 sidewall of mycobacteria (5,(13)(14)(15)(16)(17). The localization of these molecules is supported by the 1 3 detection of peptidoglycan synthetic enzymes at the mycobacterial cell tips and periphery (5, 9, 1 4 18-20). However, both intracellular and extracellular incorporation pathways have been labeling patterns (21). Intracellular uptake implies that the probe enters the biosynthetic pathway 1 7 at an early stage, and therefore marks nascent cell wall. Extracellular incorporation, on the other 1 8 hand, suggests that the probe enters the pathway at a later stage and/or is part of enzymatic 1 9 remodeling of the macromolecule in question. The extent to which peptidoglycan synthesis and 2 0 remodeling are linked is not clear (10,22,23) and may vary with species and external milieu. In 2 1 Mycobacterium tuberculosis, for example, there is indirect but abundant data that suggest that 2 2 there is substantial cell envelope remodeling during infection when growth and peptidoglycan 2 3 synthesis are presumed to be slow or nonexistent (7). An intracellular metabolic tagging method for the cell wall would be an ideal tool for determining 2 6 whether tip-extending mycobacteria can synthesize peptidoglycan along their lateral surfaces. 1 At least two pieces of evidence suggest that D-alanine-D-alanine dipeptide probes are 2 incorporated into peptidoglycan via the cytoplasmic MurF ligase (24,25). First, derivatives of D-3 alanine-D-alanine rescue the growth of Chlamydia trachomatis treated with D-cycloserine, an 4 antibiotic that inhibits peptidoglycan synthesis by inhibiting the production and self-ligation of D-5 alanine in the cytoplasm (24). Second, B. subtilis cells stripped of mature peptidoglycan by 6 lysozyme treatment retain a small amount of dipeptide-derived fluorescence (25). While these supplement 1). These data suggest that lipid-linked peptidoglycan precursors are synthesized 1 at lateral sites in addition to their expected localization at the poles. However, our standard 2 experimental protocol for detecting envelope labeling is to perform CuAAC on fixed cells. 3 Because formaldehyde fixation can permeabilize the plasma membrane to small molecules, 4 labeled material may be intracellular, extracellular or both. Dipeptide labeling could therefore 5 read out lipid I/II on the cytoplasmic face of the plasma membrane, uncrosslinked lipid II on the 6 periplasmic side, or polymerized peptidoglycan.    We next sought to address whether these molecules could be used to build the peptidoglycan 1 9 polymer. Transglycosylases from both the PBP (penicillin-binding proteins) and SEDS (shape, 2 0 elongation, division, and sporulation) families stitch peptidoglycan precursors into the existing 2 1 meshwork (Figure 3-figure supplement 1, (14,32,51,(55)(56)(57)(58)). If peptidoglycan precursors 2 2 are polymerized along the lateral surface of the mycobacterial cell, at least a subset of these 2 3 periplasmic enzymes must be present at the sidewall to assemble the biopolymer. Two conserved PBPs in mycobacteria are likely responsible for most of the peptidoglycan 2 5 polymerization required for cell viability, PonA1 and PonA2 (7,19,59). Published images of 1 4 PonA1-mRFP and PonA1-mCherry localization suggested that the fusion proteins might 1 decorate the mycobacterial sidewall in addition to the cell tips (9,18,19), but the resolution of 2 the micrographs did not allow for definitive assignment. Therefore, we first verified the 3 localization of PonA1-mRFP. We found that a subset of this fusion protein indeed homes to the 4 lateral cell surface (Figure 5-figure supplement 3A).

6
We were concerned that overexpression of PonA1-mRFP causes aberrant polar morphology 7 and is toxic to M. smegmatis (18,19) and about the propensity of mCherry to cluster (60). Since 8 our attempts to produce PonA1 fusions with different fluorescent proteins were unsuccessful, 9 we opted to take a complementary, activity-based approach. Fluorescent derivatives of β -lactam 1 0 antibiotics bind specifically and covalently to PBPs, and therefore have been used to image 1 1 active enzyme in both protein gels and intact cells (61). Our images of whole cells labeled with 1 2 Bocillin, a BODIPY conjugate of penicillin, were in agreement with those from a previous 1 3 publication (20), and seemed to indicate that Bocillin binds both the poles and sidewall of M. BODIPY dye, we considered the possibility that Bocillin might nonspecifically associate with the 1 6 greasy mycomembrane. Fluorescence across the cell surface was diminished by pre-treating 1 7 cells with the β -lactam ampicillin, which prevents peptidoglycan assembly by binding to PBPs, 1 8 but not D-cycloserine, which inhibits peptidoglycan synthesis in a PBP-independent manner 1 9 ( Figure 5-figure supplements 3B, 3C). These experiments suggest that at least some of the 2 0 sidewall labeling of Bocillin is specific, and therefore, that PBPs are present and active in these 2 1 locations. Expansion of the mycobacterial envelope is concentrated at the poles. Our data indicate 2 4 that peptidoglycan precursors are made and likely polymerized both at the poles and sidewall.

5
Peptidoglycan synthesis is often presumed to mark sites of bacterial cell growth. However, 1 5 dispersed elongation has not been reported in mycobacteria. Accordingly we performed a pulse 1 chase experiment to test whether cell expansion correlates with sites of metabolic labeling. After 2 marking peptidoglycan with RADA, we tracked labeled and unlabeled cell surface during 15 min 3 (~10% generation time) of outgrowth. While we cannot rule out sidewall expansion below our 4 limit of detection, the fluorescence dilution in this experiment was consistent with previous 5 reports ( (3-5, 7, 14, 62) and restricted to the mycobacterial poles (  Muramidase treatment increases peptidoglycan synthesis along the sidewall. What is the 9 function of peptidoglycan assembly that does not directly contribute to physical expansion of the 1 0 cell? We hypothesized that one role of growth-independent cell wall synthesis might be repair.

1
More specifically, we reasoned that insertion of peptidoglycan building blocks directly along the 1 2 cell periphery would enable a real-time, comprehensive response to damage ( Figure 6A). Cell 1 3 wall repair that is restricted to sites of mycobacterial growth, by contrast, would be confined to decreases colony-forming units (63). We have also shown that spheroplasts generated by 1 7 combined glycine and lysozyme treatment lack peptidoglycan (64). Together these data indicate 1 8 that the enzyme is able to access and damage peptidoglycan in intact cells. We challenged M. In this work we aimed to address the seemingly discrepant observations that, on the one hand, 4 mycobacteria expand from their tips (Figure 5-figure supplement 4, (3-5, 7, 14)), and on the 5 other, metabolically-labeled cell wall and synthetic enzymes are detectable at both the poles 6 and along the sidewall (Figures 2, 5-figure supplement 3, (5,9,(18)(19)(20)). The first step to 7 resolving this conundrum was to unambiguously identify sites of peptidoglycan synthesis.

8
Although the D-amino acid probes that we and others have developed for peptidoglycan labeling 9 have been extensively used for marking the cell wall (21), in most cases it has not been clear  We show that dipeptide probes rescue the growth of a DdlA mutant ( Figure 4B) and incorporate 2 0 into lipid-linked peptidoglycan precursors (Figures 5B, 5C). To our knowledge, this is the first 2 1 direct demonstration that peptidoglycan precursors can be metabolically labeled in vivo without alkDADA-derived fluorescence was stable to pre-treatment with imipenem, an antibiotic that 2 5 targets this class of enzymes (Figure 3, (42)). These data suggest that the dipeptide is unlikely 1 7 to be a direct substrate for L,D-transpeptidases. It is possible that D,D-carboxypeptidases cleave 1 a small proportion of alkDADA prior to incorporation and release D-alanine and alkDA. In this 2 scenario, apparent alkDADA labeling in M. smegmatis may be a combination of intracellular 3 alkDADA incorporation via MurF (Figures 4, 5), extracellular alkDA incorporation by L,D-   the heterogeneity in polar dominance observed previously for M. tuberculosis (15). Although we 1 cannot rule out a contribution from cyan autofluorescence, these experiments also suggest that 2 sidewall envelope metabolism may be even more prominent in M. tuberculosis than in M. 3 smegmatis, comprising half of the total cell output. What is the physiological role for cell wall synthesis that does not directly contribute to growth or 6 division? It is possible that peptidoglycan assembly along the lateral surface of the 7 mycobacterial cell is simply a byproduct of synthetic enzymes that are en route to the polar 8 elongasome or the divisome. Having active enzymes at the ready could enable efficient 9 coordination between cell growth and septation. We think that this model is less likely, however, we find that cell wall synthesis along the sidewall is enhanced upon exposure to peptidoglycan-1 7 degrading enzymes ( Figure 6B). More broadly, the ability to tailor the entire cell surface, not 1 8 just the ends, should enable rapid adaptation to external stimuli. Such activity may be 1 9 particularly important for M. tuberculosis, a slow-growing organism that must survive a hostile, 2 0 nutrient-poor environment.   with an episomal plasmid carrying the Cre recombinase and a sucrose negative selection 1 7 marker. Strains were cured of this plasmid by repeated passaging in the presence of sucrose.  were identified as having only two peaks. peptidoglycan precursors, we adopted the assay developed in (48) with some modifications. M.

9
smegmatis was inoculated in 100 mL of 7H9 medium and grown to mid-log phase at 37 ºC.

0
Where applicable, MurJ was depleted by 8 hours of anhydrotetracycline-induced protein 2 1 degradation as described (50). The bacteria were then divided into 25 mL cultures that were 2 2 subjected or not to freshly-prepared 80 µg/mL vancomycin and/or 10 µg/mL D-cycloserine. After 2 3 one hour of incubation at 37 ºC, bacteria were collected by centrifugation and cell pellets were 2 4 normalized by wet weight. 200-300 mg wet pellet was resuspended in 500 µL 1% glacial acetic 2 5 acid in water. 500 µL of the resuspended pellet mixture was transferred into a vial containing 500 µL of chloroform and 1 mL of methanol and kept at room temperature (RT) for 1-2 hours 1 with occasional vortexing. The mixture was then centrifuged at 21,000x g for 10 min at RT and 2 the supernatant was transferred into a vial containing 500 µL of 1% glacial acetic acid in water 3 and 500 µL chloroform, and vortexed for 1 min. After centrifugation at 900x g for 2 min at RT, 4 three phases were distinguishable: aqueous, an interface, and organic. We collected the lipids To detect lipid-linked precursors that had been metabolically labeled with alkDADA or azDADA, Selectively Blocks Apical Growth in CMN Group Bacteria. mBio 8. acetylglucosamine-modified proteins via the N-acetylgalactosamine salvage pathway. Communications 5:4981.      Agents and Chemotherapy 57:5940-5945. mycobacterial cell envelope assembly for initiation and stabilization of polar growth.

9
Cytoskeleton.   and black chemical structures denote probes used in one and two step labeling, respectively. without (top; 34<n<42) and with (bottom; 9<n<31) visible septa. We defined the dim pole (dp; 1 5 dark purple) as the sum of the fluorescence intensity over the first 25% of the cell; the sidewall 1 6 (sw; medium purple) as the sum from 25% to 75%; and the bright pole (bp; light purple) as the 1 7 sum over the final 25% of the cell. Fluor distrib, fluorescence distribution. AU, arbitrary units. Error bars, +/-standard deviation. smegmatis was pretreated or not with the drugs at the fold-MIC indicated and labeled as in B.

2
Experiment was performed twice in triplicate with similar results. One data set is shown. The Combinations that produce Bliss scores greater than, equal to, or less than 1 are respectively 2 interpreted as antagonistic, additive, or synergistic interactions.  via SPAAC or to alkyne biotin via CuAAC then detected as in A. results. dp, dim pole (dark purple); sw, sidewall (medium purple); bp, bright pole (light purple).

1
Fluor distrib, fluorescence distribution. AU, arbitrary units.  TPase, transpeptidase. and 10-fold serial dilutions were spotted onto LB agar.  Signal was normalized to cell length and to total fluorescence intensity. Cells were oriented such that the brighter pole is on the right hand side of the graph. D, M. smegmatis (Msm) and M. tuberculosis (Mtb) were labeled with HADA for 15 min and 2 hours, respectively, then washed and fixed. Fluorescence was quantitated as in C for cells without (top; 34<n<42) and with (bottom; 9<n<31) visible septa. We defined the dim pole (dp; dark purple) as the sum of the fluorescence intensity over the first 25% of the cell; the sidewall (sw; medium purple) as the sum from 25% to 75%; and the bright pole (bp; light purple) as the sum over the final 25% of the cell. Fluor distrib, fluorescence distribution. AU, arbitrary units. 0.1 0.3 0.5 1 2 0.5 0.9 1.0 1.0 0.9 0.9 1 0.9 0.9 0.9 0.9 0.9 2 0.9 0.9 0.9 0.9 0.9 4 0.9 0.9 0.9 0.9 0.8 8 0.8 0.9 0.9 0.9 0.8 fold MIC dcs Bliss score % inhibition D fold MIC imi A, Schematic of the theoretical routes of D-amino acid (DA) and D-alanine D-alanine (DADA) probe incorporation. LDT, L,Dtranseptidase, DDT, D,D-transpeptidase (DDTs). For more details on the peptidoglycan synthesis pathway, see Fig. S3.
B, Sensitivity of HADA (blue), NADA (green), RADA (red), alkDA (light grey), alkDADA (dark grey) and OalkTMM (black) to antibiotics. Imi, imipenem + clavulanate; amp, ampicillin + clavulanate; dcs, D-cycloserine; vanc, vancomycin. M. smegmatis was pretreated or not with the indicated antibiotics at 2X MIC for 30 min then incubated an additional 15 min in the presence of probe. The bacteria were then washed and fixed. The alkyne-bearing probes were detected by CuAAC with azido-CR110 and quantitated by flow cytometry. Experiment was performed 3-4 times in triplicate. For each biological replicate, the averaged median fluorescence intensities (MFI) of the drug-treated samples were divided by the MFI of untreated bacteria. Data are expressed as the average percentage of untreated labeling across the biological replicates. Error bars, +/-standard deviation.
C, Effect of antibiotic dose on alkDA-derived fluorescence. M. smegmatis was pretreated or not with drugs at the fold-MIC indicated and labeled as in B. Experiment was performed 3 times in triplicate. For each biological replicate, the averaged MFI of the control (no drug, no alkDA but subjected to CuAAC) was subtracted from the averaged MFI of the drug-treated sample. This was then divided by the averaged MFI of untreated control (no drug but incubated in alkDA and subjected to CuAAC) from which the control MFI had also been subtracted. Data are expressed as the average percentage of untreated labeling across the biological replicates. Error bars, +/-standard deviation. D, Left, combined effects of imipenem and D-cycloserine on alkDA-derived fluorescence. M. smegmatis was pretreated or not with the drugs at the fold-MIC indicated and labeled as in B. Experiment was performed twice in triplicate with similar results. One data set is shown. The percent of untreated labeling was calculated as in C and subtracted from 100 to obtain the percent inhibition. Right, Bliss interaction scores for each pair of doses in left-hand graph were calculated as (E I +E D -E I E D )/E I,D where E I is the effect of imipenem at dose i, E D is the effect of D-cycloserine at dose d and E I,D is the observed effect of the drugs at dose i and dose d. Combinations that produce Bliss scores greater than, equal to, or less than 1 are respectively interpreted as antagonistic, additive, or synergistic interactions. E, Dose-dependent effect of imipenem on alkDADA (grey), HADA (blue) and RADA (red) labeling. M. smegmatis was pretreated or not with imipenem at the fold-MIC indicated and labeled as in B. Experiment was performed 3 times in triplicate and one representative data set is shown. For each technical replicate, the averaged median fluorescence intensities (MFI) of the drugtreated samples were divided by the averaged MFI of untreated bacteria. Data are expressed as the average percentage of untreated labeling across the technical replicates. Error bars, +/-standard deviation.
F, Wildtype M. smegmatis pre-treated or not with 2X MIC imipenem and untreated ΔldtABE and complement (c ldtABE) were labeled with alkDA (white), HADA (blue), NADA (green) or RADA (red) and processed as in B. Experiment was performed 2-10 times in triplicate. Representative data from one of the biological replicates is shown here. Error bars, +/-standard deviation. A, Detection of lipid-linked peptidoglycan precursors from organic extracts of M. smegmatis. Endogenous D-alanines were exchanged for biotin-D-lysine (BDL) via purified S. aureus PBP4. Biotinylated species detected by blotting with streptavidin-HRP. MurJ (MviN) depletion strain was incubated in anhydrotetracycline (atc) to induce protein degradation. Other strains were treated with vancomycin (vanc), D-cycloserine (dcs) or a combination prior to harvesting. Wt, wildtype; blank, no sample run; no extract, BDL and PBP4 alone a , organic extract from MurJ-atc; b , organic extract from MurJ+atc; hk, heat-killed.
B, Detection of lipid-linked peptidoglycan precursors labeled by alkDADA in vivo. Wildtype and Δalr strains were incubated in alkDADA and treated or not with the indicated antibiotics prior to harvest. Alkyne-tagged species from organic extracts were ligated to picolyl azide biotin via CuAAC then detected as in A. BDL/PBP4, endogenous precursors from MurJ+atc were subjected to in vitro exchange reaction as in A.
C, Detection of lipid-linked peptidoglycan precursors labeled by azDADA in vivo. Δalr was incubated in azDADA. Azide-tagged species from organic extracts were ligated to DIFO-biotin via SPAAC or to alkyne biotin via CuAAC then detected as in A. -lys/mut + lys/mut fluor distrib dp sw bp A replication PG synthesis PG repair 5 µm -lys/mut + lys/mut B Figure 6. Peptidoglycan synthesis is redistributed to the sidewall upon cell wall damage. A, Model for the spatial organization of peptidoglycan (PG) synthesis and repair with respect to mycobacterial growth and division. B, M. smegmatis was pretreated or not with lysozyme and mutanolysin for 30 min then incubated an additional 15 min in the presence of alkDADA. The bacteria were then washed and fixed and subjected to CuAAC with azido-CR110. Images obtained by conventional fluorescence microscopy were quantified for 25<n<40 cells as in Fig. 2 except that i. signal was not normalized to total fluorescence intensity, only to cell length, and ii. cells with and without visible septa were analyzed together. Experiment was performed twice with similar results. dp, dim pole (dark purple); sw, sidewall (medium purple); bp, bright pole (light purple). Fluor distrib, fluorescence distribution. AU, arbitrary units.