The C. crescentus genome encodes a complete PG recycling pathway.

De novo PG biosynthesis starts with the conversion of fructose-6-phosphate, supplied by central carbon metabolism, to glucosam-ine-6-phosphate. This key metabolite is then transformed by multiple enzymatic steps into the membrane-attached PG precursor lipid II, which is flipped to the periplasmic face and incorporated into the existing PG sacculus by glycosyltransferases and transpeptidases. To enable PG remodeling, growth and cell separation, the PG meshwork is cleaved by a diverse set of enzymes, such as lytic transglycosylases (LTs), endopeptidases (EPases), amidases and carboxypeptidases (CPases). PG fragments released by these activities are imported into the cytoplasm by the permease AmpG or two homologs of the phospho-transferase system (PTS) transporter NagE. In the cytoplasm, the peptide is released by AmiR and the sugars separated by NagZ. Afterwards, two independent pathways feed the two sugars back into the de novo PG biosynthetic pathway. The enzymes are named as established for E. coli and P. aeruginosa, with the exception of the C. crescentus homolog of AmpD, which was renamed to AmiR due to its different domain organization. The symbols are defined in the legend on the right.

List of PG recycling enzymes found in E. coli and P. aeruginosa and the corresponding homologs of C. crescentus identified by reciprocal BLAST analysis.

The AmiR and NagZ homologs are functional and have a broad substrate specificity.

(A) Overview of the PG digestion assay used to assess the hydrolytic activities of AmiR and NagZ. (B) HPLC chromatograms showing the products generated upon incubation of anhydro-muropeptides from LT-treated PG sacculi without added proteins (Buffer), with AmiR or with a catalytically inactive AmiR variant (AmiR*). The identities of anhMurNAc-peptides were determined based on their known retention times (Glauner, 1988). The AmiR reaction product GlcNAc-1,6-anhMurNAc was identified by mass spectrometry (Figure 2–figure supplement 4A). (C) HPLC chromatograms showing the products generated upon incubation of anhydro-muropeptides from LT-treated PG sacculi without added proteins (Buffer), with NagZ or with a catalytically inactive NagZ variant (NagZ*). Disaccharide-containing anhydro-muropeptides were identified based on their known retention times (Glauner, 1988). The identities of anhMurNAc-peptides were validated by mass spectrometry analysis of the reaction products obtained after NagZ treatment of anhydro-muropeptide mixtures (Figure 2–figure supplement 4B).

Metabolomic analysis reveals aberrant levels of PG recycling intermediates in recycling-deficient strains.

(A) Overview of the analysis pipeline used to identify cytoplasmic PG recycling products. (B) Relative levels of the indicated anhMurNAc-peptide species in the cytoplasm of C. crescentus wild-type, ΔamiR (PR033) and ΔamiR ΔampG (PR221) cells, measured by metabolomics analysis. For each anhydromuropeptide, the mass spectrometric peak areas were normalized against the mean obtained for the ΔamiR mutant. The statistical significance of differences between the wild type and the mutant strains was determined using an unpaired two-sided Welch’s t-test. ns indicates p-values > 0.1. (C) Relative levels of the GlcNAc–anhMurNAc disaccharide in C. crescentus wild-type, ΔampG (PR207), ΔamiR (PR033) and ΔnagZ (PR188) cells, assessed as for panel B. The mass spectrometric peak areas were normalized against the mean obtained for the ΔnagZ mutant. Statistical significance was determined as described for panel B.

AmpG, AmiR and NagZ are required for proper cell shape and β-lactam resistance in C. crescentus.

(A) Phase contrast images of C. crescentus wild-type, ΔampG (PR207), ΔamiR (PR033) and ΔnagZ (PR188) cells harvested in the exponential and stationary growth phase. Scale bar: 5 µm. (B) Superplots showing the distribution of cell lengths in populations of C. crescentus wild-type, ΔampG (PR207), ΔamiR (PR033) and ΔnagZ (PR188) cells in the exponential (exp) and stationary (stat) growth phase. Small dots represent the data of three independent replicates (shown in light blue, teal and dark grey; n=100 cells per replicate). Large dots represent the mean values of the three datasets, with their average indicated by a red horizontal line. The statistical significance (p-value) of differences between conditions was assessed using an unpaired two-sided Welch’s t-test. ns indicates p-values > 0.1. (C) Serial dilution spot assay investigating the growth of wild-type, ΔampG (PR207), ΔamiR (PR033), ΔnagZ (PR188) and ΔblaA (CS606) cells on agar plates with different concentrations of ampicillin.

The GlcNAc recycling pathway is dispensable for proper cell division and β-lactam resistance.

(A) Phase contrast images of C. crescentus ΔnagK (PR255), ΔnagA1 (PR256) and ΔnagA2 (PR257) cells harvested in the exponential and stationary growth phase. Scale bar: 5 µm. (B) Superplots showing the distribution of cell lengths in populations of C. crescentus wild-type, ΔnagK (PR255), ΔnagA1 (PR256) and ΔnagA2 (PR257) cells in the exponential (exp) and stationary (stat) growth phase. Data (n=100 cells per replicate) are presented as described for Figure 4B. The statistical significance (p-value) of differences between conditions was assessed using an unpaired two-sided Welch’s t-test. ns indicates p-values > 0.1. (C) Serial dilution spot assays investigating the growth of C. crescentus ΔnagK (PR255), ΔnagA1 (PR256) and ΔnagA2 (PR257) cells on agar plates containing different concentrations of ampicillin. The cells were spotted on the same plates as those depicted in Figure 3. (D) Volcano plot illustrating differential protein abundance in C. crescentus wild-type cells grown in M2G minimal medium with GlcNAc compared to plain M2G medium. Grey dots represent proteins identified by mass spectrometry. The x-axis indicates the log2 of the average difference in the peptide counts for the two different conditions (n=3 independent replicates). The y-axis shows the -log10 of the corresponding p-value. Proteins encoded in the two nag gene clusters are highlighted in color and annotated. The colors correspond to those used in panel E. (E) Schematic representation of the two nag gene clusters. The predicted functions of the gene products are specified in the legend on the right.

C. crescentus recycles anhMurNAc through a functional MurU pathway.

(A) Phase contrast images of C. crescentus ΔanmK (PR252) and ΔamgK (PR262) cells, harvested in the exponential and stationary growth phase. Scale bar: 5 µm. (B) Superplots showing the distribution of cell lengths in populations of C. crescentus wild-type, ΔanmK (PR252), and ΔamgK (PR262) cells in the exponential (exp) and stationary (stat) growth phase. Data (n=100 cells per replicate) are presented as described for Figure 4B. The statistical significance (p-value) of differences between conditions was assessed using an unpaired two-sided Welch’s t-test. ns indicates p-values > 0.1. (C) Serial dilution spot assay investigating the growth of C. crescentus ΔanmK (PR252) and ΔamgK (PR262) on plates containing different concentrations of ampicillin. The cells were spotted on the same plates as those depicted in Figure 3. (D) Levels of MurNAc in the cytoplasm of C. crescentus wild-type and ΔamgK (PR262) cells, measured by metabolomics analysis through quantification of the corresponding mass spectrometric peak areas. The mass spectrometric peak areas were normalized against the mean obtained for the ΔamgK mutant. The statistical significance of differences between the two strains was determined using an unpaired two-sided Welch’s t-test. (E) Analysis of the growth of C. crescentus wild-type, ΔnagK (PR255) and ΔamgK (PR262) cells on agar containing a fosfomycin gradient. The fosfomycin concentrations on the MIC test strip are indicate in the legend on the right.

Recycling of the disaccharide moiety plays only a minor role in cell division and β-lactam resistance.

(A) Phase contrast images of C. crescentus ΔamgK ΔnagK (PR261) cells, harvested in the exponential and stationary growth phase. Scale bar: 5 µm. (B) Superplots showing the distribution of cell lengths in populations of C. crescentus wild-type and ΔamgK ΔnagK (PR261) cells in the exponential (exp) and stationary (stat) growth phase. Data (n=100 cells per replicate) are presented as described for Figure 4B. The statistical significance (p-value) of differences between conditions was assessed using an unpaired two-sided Welch’s t-test. ns indicates p-values > 0.1. (C) Serial dilution spot assays investigating the growth of C. crescentus ΔamgK ΔnagK (PR261) cells on agar plates containing different concentrations of ampicillin. The cells were spotted on the same plates as those depicted in Figure 3.

BlaA function is not regulated at the levels of transcription, protein accumulation or enzymatic activity.

(A) Organization of the putative five-gene operon containing the blaA gene. The annotation or ORF number of the open reading frames and the predicted functions of their gene products are indicated. (B) Activity of the promoter likely driving the expression of the blaA-containing operon in different strain backgrounds. The C. crescentus wild-type, ΔampG (PR207), ΔamiR (PR033) and ΔnagZ (PR188) strains were transformed with a plasmid carrying a 400 bp fragment of the CCNA_02225 upstream region fused to a β-galactosidase reporter gene and assayed for reporter activity. Strains transformed with the empty plasmid served as negative controls (NC). The underlying data are shown as grey dots. The statistical significance (p-value) of differences between strains was calculated using an unpaired two-sided Welch’s t-test. ns indicates p-values > 0.1. (C) β-lactamase activity of C. crescentus wild-type, ΔampG (PR207), ΔamiR (PR033) and ΔnagZ (PR188) cells, as determined by monitoring the hydrolysis of the chromogenic β-lactam nitrocefin in permeabilized cells over time. (D) Volcano plots showing differential protein abundance in C. crescentus ΔampG (PR207), ΔamiR (PR033) and ΔnagZ (PR188) cells compared to wild-type cells. Grey dots represent proteins identified by mass spectrometry. The x-axis indicates the log2 of the average difference in the peptide counts for the two different conditions (n=4 independent replicates). The y-axis shows the -log10 of the corresponding p-value. The BlaA protein is highlighted in magenta. (E) Serial dilution spot assays investigating the growth of C. crescentus wild-type, ΔampG (PR207), ΔamiR (PR033), ΔnagZ (PR188) and ΔblaA cells (CS606) on agar plates containing different concentrations of aztreonam.

PG recycling defects increase the sensitivity of the septal PG synthetic machinery to ampicillin and reduce PG precursor biosynthesis.

(A) Phase contrast images of C. crescentus wild-type and ΔamiR (PR033) cells, incubated for 3 h in the presence of different concentrations of ampicillin. Scale bar: 5 µm. (B) Quantification of the fraction of lysed cells in cultures of C. crescentus wild-type and ΔamiR (PR033) cells, incubated for 3 h in the presence of different concentrations of ampicillin. The bars represent the mean (±SD) of three independent replicates (n=100 cells per condition). The underlying data are shown as grey dots. The statistical significance (p-value) of differences between strains was calculated using an unpaired two-sided Welch’s t-test. ns indicates p-values > 0.1. (C) Time-lapse microscopy analysis showing C. crescentus wild-type and ΔamiR (PR033) cells in a microfluidic flow cell before and after ampicillin treatment. After 3 h of growth in antibiotic-free medium, the cells were shifted to medium containing 20 µg/mL ampicillin. Images were taken at 5 min intervals for a total duration of 8 h. Shown are representative frames of the time-lapse series. The full sequences are shown in Figure 9-Video 1 and Figure 9-Video 2. (D) Image of an SDS-gel showing the labeling of PBPs with the fluorescent β-lactam bocillin-FL in the absence or presence of other β-lactams. Crude membrane fractions of C. crescentus ΔblaA (CS606) cells were preincubated with the indicated concentrations of ampicillin, 5 µg/mL cefalexin (Cfx) or 100 µg/mL mecillinam (Mec), treated with bocillin-FL and then subjected to SDS gel electrophoresis to separate the labeled PBPs prior to fluorescence imaging. (E) Levels of UDP-GlcNAc and UDP-MurNAc in the cytoplasm of C. crescentus wild-type, ΔampG (PR207), ΔamiR (PR033) and ΔnagZ (PR188) cells, measured by metabolomics analysis through quantification of the corresponding mass spectrometric peak areas. For each PG precursor, the mass spectrometric peak areas were normalized against the mean obtained for the wild-type strain. The statistical significance of differences between the wild type and the mutant strains was determined using an unpaired two-sided Welch’s t-test. ns indicates p-values > 0.1; nd: not detectable.

The cell filamentation phenotype of ΔamiR cells is rescued by hyper-activation of the septal FtsW-PBP3 PG biosynthetic complex.

(A) Phase contrast images of C. crescentus ftsW::ftsW* (ML2103) and ΔamiR ftsW::ftsW* (PR246) cells, harvested in the exponential and stationary growth phase. (B) Superplots showing the distribution of cell lengths in populations of C. crescentus ftsW::ftsW* (ML2103) and ΔamiR ftsW::ftsW* (PR246) cells in the exponential (exp) and stationary (stat) growth phase. Data (n=100 cells per replicate) are presented as described for Figure 4B. The statistical significance (p-value) of differences between conditions was assessed using an unpaired two-sided Welch’s t-test. ns indicates p-values > 0.1. (C) Serial dilution spot assay investigating the growth of C. crescentus ftsW::ftsW* (ML2103), ΔamiR (PR033) and ΔamiR ftsW::ftsW* (PR246) cells on agar plates containing no or 10 µg/mL ampicillin.

Model for the critical role of PG recycling in C. crescentus growth and β-lactam.

(A) PBPs mediates the incorporation of new cell wall material into the growing PG sacculus. The PG precursors required for this process are synthesized in the cytoplasm, using building blocks that are provided by both de novo synthesis and PG recycling. (B) If β-lactams enter the periplasm, they compete with PG precursors for binding to their PBP targets. Most antibiotic molecules are rapidly hydrolyzed by the β-lactamase BlaA. However, a minor fraction manages to interact with PBPs before they are captured and degraded. The resulting inactivation of a small proportion of PBPs can be tolerated by cells, as long as they still contain sufficient active PBPs to maintain cell growth and division. (C) Upon disruption of PG recycling, the level of PG monomers strongly decreases, leaving the substrate binding sites of PBPs unoccupied for longer periods of time and thus increasing their accessibility to β-lactam molecules. As a result, a larger fraction of PBPs is inactivated, leading to decreased cell wall integrity and ultimately cell lysis. This effect may be aggravated by the reduced activity of the remaining, active PBPs due to an insufficient supply of PG precursors and by the accumulation of anhydro-muropentapeptides in the periplasm, which could interfere with the activities of PG biosynthetic enzymes, thereby further decreasing the incorporation of new material into the existing sacculus.

Substrate specificity of AmiR.

(A) Structural models of AmiR and the catalytically inactive AmiR* variant, generated with AlphaFold 3 (Abramson et al., 2024). The magnified region shows the predicted catalytic site of AmiR and the amino acid substitutions introduced to generate AmiR*. (B) HPLC chromatograms showing the products generated upon incubation of anhydro-muropeptides from LT-treated PG sacculi for 16 h without added proteins (Buffer) or with the indicated concentrations of AmiR. (C) HPLC chromatograms showing the products generated upon incubation of muropeptides from muramidase-treated PG sacculi for 16 h without added proteins (Buffer) or with the indicated concentrations of AmiR. Reaction products were identified based on their known retention times (Glauner, 1988).

Lack of AmiR activity towards intact PG sacculi.

(A) Overview of the PG digestion assay used to assess the hydrolytic activity of AmiR on intact PG sacculi. (B) HPLC chromatograms showing the products generated upon incubation of intact PG sacculi without added proteins (Buffer), with AmiR and the catalytically inactive AmiR variant (AmiR*) and subsequent cellosyl-mediated cleavage of the pre-treated sacculi into muropeptides. Muropeptides were identified based on their known retention times (Glauner, 1988).

Activity of NagZ with muropeptides.

(A) Structural models of NagZ and the catalytically inactive NagZ* variant, generated with AlphaFold 3 (Abramson et al., 2024). The magnified region shows the predicted catalytic site of NagZ and the amino acid substitution introduced to generate NagZ*. (B) HPLC chromatograms showing the products generated upon incubation of muropeptides from muramidase-treated PG sacculi without added proteins (Buffer), with NagZ or with a catalytically inactive NagZ variant (NagZ*). Disaccharide-containing muropeptides were identified based on their known retention times (Glauner, 1988). The identities of MurNAc-peptides were validated by mass spectrometry analysis of the reaction products obtained after NagZ treatment of anhydro-muropeptide mixtures (Figure 2–figure supplement 4B).

Mass spectrometric identification of AmiR and NagZ reaction products.

(A) Identification of the GlcNAc–anhMurNAc disaccharide generated by treatment of anhydro-muropeptides with AmiR. Sacculi isolated from E. coli D456 (Edwards and Donachie, 1993) were treated with the LT Slt70 E. coli. The graph shows the chromatogram obtained after separation of the sugar-containing reaction products by HPLC. The observed m/z ratios of the main product in peak 1 and the calculated molecular mass of the GlcNAc-1,6-anhMurNAc disaccharide are given in the table on the right. (B) Identification of the anhMurNAc-peptide species generated by treatment of anhydro-muropeptides with NagZ. Sacculi isolated from E. coli CS703-1 (Meberg et al., 2001) were treated with the LT MltA from E. coli. The graph shows the chromatogram obtained after separation of the reaction products by HPLC. The table on the right gives the observed m/z ratios of the main product detected contained in the indicated peaks as well as the calculated molecular masses of the anhMurNAc-peptides assigned to these peaks.

Morphological characterization of catalytically inactive AmiR and NagZ variants.

(A) Phase contrast images of C. crescentus amiR::amiR* (PR173) and nagZ:nagZ* (PR196) cells harvested in the exponential and stationary growth phase. Scale bar: 5 µm. (B) Superplots showing the distribution of cell lengths in populations of wild-type, ΔamiR (PR033), amiR::amiR* (PR173), ΔnagZ (PR188) and nagZ::nagZ* (PR196) cells in the exponential (exp) and stationary (stat) growth phase. Data (n=100 cells per replicate) are presented as described for Figure 4B. The statistical significance (p-value) of differences between conditions was assessed using an unpaired two-sided Welch’s t-test. ns indicates p-values > 0.1. The data for wild-type, ΔamiR and ΔnagZ cells are taken from Figure 3 and shown as a reference.

Growth of PG recycling-deficient mutants in minimal medium.

(A) Phase contrast images of C. crescentus wild-type, ΔampG (PR207), ΔamiR (PR033) and ΔnagZ (PR188) cells grown in minimal (M2G) medium and harvested in the stationary growth phase. (B) Superplots showing the distribution of cell lengths in populations of wild-type, ΔampG (PR207), ΔamiR (PR033) and ΔnagZ (PR188) cells grown in rich medium (PYE) and minimal medium (M2G). Data (n=100 cells per replicate) are presented as described for Figure 4B. The statistical significance (p-value) of differences between conditions was assessed using an unpaired two-sided Welch’s t-test. ns indicates p-values > 0.1. The distributions of cell lengths in rich medium are taken from Figure 3 and shown as a reference.

Redundant role of the LT SdpA in the production of anhydro-muropeptides.

(A) Serial dilution spot assay investigating the growth of C. crescentus ΔsdpA (AM399) cells on agar plates containing different concentrations of ampicillin. The cells were spotted on the same plates as those depicted in Figure 3. (B) Phase contrast images of C. crescentus ΔsdpA (AM399) harvested in the exponential or stationary growth phase. (C) Superplots showing the distribution of cell lengths in populations of wild-type and ΔsdpA (AM399) cells harvested in the exponential (exp) and stationary (stat) growth phase. Data (n=100 cells per replicate) are presented as described for Figure 4B. The statistical significance (p-value) of differences between conditions was assessed using an unpaired two-sided Welch’s t-test. ns indicates p-values > 0.1. (D) Levels of the indicated anhMurNAc-peptide species in the cytoplasm of C. crescentus wild-type, ΔamiR (PR033) and ΔamiR ΔsdpA (PR260) cells, measured by metabolomics analysis through quantification of the corresponding mass spectrometric peak areas. For each anhydro-muropeptide, the mass spectrometric peak areas were normalized against the mean obtained for the ΔamiR mutant. The statistical significance of differences between the wild type and the mutant strains was determined using an unpaired two-sided Welch’s t-test. ns indicates p-values > 0.1.

The two genes forming an operon with amiR are dispensible for PG recycling or β-lactam resistance.

(A) Genomic context of amiR. The gene names and the predicted functions of the respective gene products are indicated. (B) Levels of the indicated anhMurNAc-peptide species in the cytoplasm of C. crescentus wild-type, ΔamiR (PR033) and ΔamiR ΔtraX (PR258) cells, measured by metabolomics analysis through quantification of the corresponding mass spectrometric peak areas. (C) Volcano plot showing differences in protein abundance in C. crescentus ΔregX (PR154) cells in comparison to wild-type cells. Grey dots represent proteins identified by mass spectrometry. The x-axis indicates the log2 of the average difference in the peptide counts for the two different conditions (n=3 independent replicates). The y-axis shows the -log10 of the corresponding p-value. Proteins with a fold change smaller than 0.5 or larger than 2 are highlighted in color and numbered. Details on these proteins are given in Figure 8–table supplement 1. (D) Phase contrast images of ΔtraX (PR153) and ΔregX (PR154) cells, harvested in the exponential or stationary growth phase. Scale bar: 5 µm. (E) Superplots showing the distribution of cell lengths in populations of C. crescentus wild-type, ΔtraX (PR153) and ΔregX (PR154) cells in the exponential (exp) and stationary (stat) growth phase. Data (n=100 cells per replicate) are presented as described for Figure 4B. The statistical significance (p-value) of differences between conditions was assessed using an unpaired two-sided Welch’s t-test. ns indicates p-values > 0.1. (F) Serial dilution spot assays investigating the growth of C. crescentus wild-type, ΔtraX (PR153), ΔregX (PR154) and ΔblaA (CS606) cells on agar plates in the absence or presence of ampicillin.

Proteins found to be differentially accumulated in ΔregX cells compared to wild-type cells.

The table lists the ORF numbers and predicted functions of the numbered proteins in Figure 8–Figure supplement 1C.

Changes in protein accumulation upon deletion of CCNA_02225.

(A) Volcano plot showing differential protein abundance in C. crescentus ΔCCNA_02225 (PR218) vs. wild-type cells. Grey dots represent proteins identified by mass spectrometry. The x-axis indicates the log2 of the average difference in the peptide counts for the two different conditions (n=3 independent replicates). The y-axis shows the -log10 of the corresponding p-value. The 20 proteins with the highest fold change according to Manhattan distance are highlighted in color and numbered. The functional categorization of these genes is indicated in the lower left corner of the plot. Proteins encoded in the five-gene operon are labeled in orange. (B) List of the most highly deregulated proteins in the ΔCCNA_02225 (PR218) mutant. The table gives the gene name or ORF number, the fold change, the p-value and the predicted function for the proteins numbered in panel A.

Muropeptide composition of PG sacculi from wild-type and ΔamiR cells.

Shown are HPLC chromatograms resolving the muropeptides obtained by muramidase treatment of PG sacculi isolated from stationary wild-type or ΔamiR (PR033) cells. The muropeptides contained in the different peaks are listed in Figure 9–table supplement 1. A summary of the data obtained is given in Figure 9–table supplement 2.

Muropeptide composition of peptidoglycan isolated from stationary C. crescentus wild-type and ΔamiR cells.

The table gives the relative abundance of the indicated muropeptide species, calculated from the areas of the corresponding peaks in the HPLC chromatograms from Figure 9–Figure supplement 1. Two independent replicates were analyzed per strain.

Overview of the muropeptide species identified in peptidoglycan from stationary C. crescentus wild-type and ΔamiR cells.

The table summarizes the relative abundance of different muropeptide species, calculated from the values listed in Figure 9–table supplement 1. Two independent replicates were analyzed per strain.

Ampicillin sensitivity of PG recycling-defective mutants in the presence of GlcNAc.

Shown is a serial dilution spot assay investigating the growth of C. crescentus wild-type, ΔampG (PR207), ΔamiR (PR033), ΔnagZ (PR188) and ΔblaA (CS606) cells on agar plates containing different concentrations of ampicillin in the presence of 0.3% GlcNAc.