Introduction

In most bacteria, the cytoplasmic membrane is surrounded by a layer of peptidoglycan (PG), which protects the cell from osmotic lysis, maintains cell shape and serves as a physical barrier against environmental assaults (Egan et al., 2020). PG is composed of a dense meshwork of glycan strands made of alternating N-acetylglucosamine (GlcNAc) and N-acetylmuramic acid (MurNAc) moieties that are connected by short peptides containing unusual D-amino acids (Vollmer et al., 2008a). The integrity of the PG layer is critical for bacterial fitness and survival, which makes it a major target of medically relevant antibiotics (Bugg et al., 2011).

PG biosynthesis initiates in the cytoplasm (Barreteau et al., 2008; Höltje, 1998), starting with the transformation of fructose-6-phosphate to UDP-MurNAc. Next, this key intermediate is modified with a tripep-tide, synthesized stepwise by dedicated amino acid ligases, and then completed by the addition of a preformed D-Ala–D-Ala unit, leading to UDP-MurNAc-pentapeptide (Figure 1). The MurNAc-pentapeptide moiety is then transferred to the lipid carrier undecaprenol phosphate and linked to GlcNAc. The product, lipid II, is then flipped across the cytoplasmic membrane (Sham et al., 2014), where is it used by glycosyl-transferase (GTases) to polymerize the glycan chains. Transpeptidases (TPases) then crosslink the peptides of neighboring glycan strands, thereby establishing the mesh-like architecture of PG (Egan et al., 2020; Höltje, 1998). The most abundant group of TPases are the DD-TPases, which cleave the C-terminal D-Ala– D-Ala unit of a pentapeptide and then transfer the resulting tetrapeptide to an adjacent acceptor peptide (Sauvage et al., 2008). This group of enzymes is also known as penicillin-binding proteins (PBPs), because they recognize penicillin and other β-lactams as non-cognate substrates, based on the structural similarity of their β-lactam ring to the D-Ala–D-Ala moiety of uncrosslinked peptides. These compounds form a stable adduct with the active-site serine of PBPs and thus irreversibly inhibit their crosslinking activity, thereby weakening the cell wall and inducing cell lysis (Cho et al., 2014; Mora-Ochomogo and Lohans, 2021). The sensitivity of DD-TPases to β-lactams represents a vulnerability that is widely exploited to treat bacterial infections, but resistance to this class of antibiotics is rapidly spreading among pathogens (Lima et al., 2020).

Figure 1.

Cells usually contain both monofunctional PBPs, which contain only a transpeptidase domain, and bifunctional PBPs, which contain an additional glycosyltransferase domain and thus have polymerization as well as crosslinking activity (Sauvage et al., 2008; Vollmer and Bertsche, 2008). To enable the tight spatio-temporal control of PG biosynthesis, these enzymes are often part of dynamic multi-protein complexes that include regulatory factors and cytoskeletal elements (Egan et al., 2020; Typas et al., 2012). In most rod-shaped bacteria, lateral cell wall elongation is carried out by the elongasome, a complex organized by the actin-like protein MreB (Jones et al., 2001). Multiple elongasomes are distributed along the longitudinal axis of the cell and move around its circumference in helical tracks (Garner et al., 2011; van Teeffelen et al., 2011), incorporating new cell wall material in a dispersed manner. Their PG biosynthetic activity is mediated by the GTase RodA and the monofunctional DD-TPase PBP2 (Cho et al., 2016; Meeske et al., 2016), whose functions are closely coupled (Rohs et al., 2018). Cell division, by contrast, is mediated by the divisome, which is orchestrated by polymers of the tubulin homolog FtsZ (Bi and Lutkenhaus, 1991; Bisson-Filho et al., 2017; Yang et al., 2017), with the glycosyltransferase FtsW and the monofunctional DD-TPase PBP3 acting as the main PG biosynthetic enzymes (Taguchi et al., 2019). The division process also involves bifunctional PBPs, such as PBP1B in E. coli (Bertsche et al., 2006; Navarro et al., 2025), although this class of enzymes also appears to act outside larger complexes in the repair of cell wall (More et al., 2019; Pazos and Vollmer, 2021).

To enable the incorporation of new cell wall material into the existing sacculus, the PG meshwork needs to be opened at specific sites – a task that is achieved by a diverse set of autolysins (Figure 1) (van Heijenoort, 2011; Vollmer et al., 2008b). DD- and LD-endopeptidases cleave the peptides and the crosslinks between them, while amidases remove the entire peptide stem from the glycan chain. The glycan chains are cleaved by glucosaminidases and muramidases. The latter include lytic transglycosylases (LTs), which break the β-1,4-glycosidic bond between MurNAc and GlcNAc, leading to the concomitant formation of a 1,6-anhydro bond in the MurNAc moiety (Dik et al., 2017; Vollmer et al., 2008b). The activity of autolytic enzymes generates small PG fragments, called muropeptides. Only a small fraction of these is released into the environment (Goodell and Schwarz, 1985), whereas the majority is transported into the cytoplasm, where they are metabolized and reused in PG synthesis. This process is known as PG recycling and has so far been studied mainly in gammaproteobacteria such as Escherichia coli or Pseudomonas aeruginosa (Dik et al., 2018; Gilmore and Cava, 2025; Park and Uehara, 2008).

During one division cycle, more than 50% of the PG sacculus are broken down in E. coli. Most of the turnover products in this and other Gram-negative species are 1,6-anhydro-muropeptides (Vollmer et al., 2008b). They are transported into the cytoplasm by the proton-driven Major Facilitator Superfamily (MFS) permease AmpG (Cheng and Park, 2002; Jacobs et al., 1994), where the sugars and the peptide are separated from each by the combined action of the LD-carboxypeptidase (LD-CPase) LdcA (Templin et al., 1999), the amidase AmpD (Höltje et al., 1994) and the β-hexosaminidase NagZ (Cheng et al., 2000). The GlcNAc moiety is then recycled by conversion to GlcNAc-6-P and then GlcN-6-P, which can either be transformed into the PG precursor UDP-MurNAc or directed into central carbon metabolism by deamination to fructose-6-P (Gilmore and Cava, 2025; Vollmer et al., 2008b). The fate of the anhMurNAc moiety, on the other hand, varies between species. In E. coli, it is converted to GlcNAc-6-P (Uehara et al., 2005), whereas P. aeruginosa employs the MurU pathway to convert anhMurNAc into UDP-MurNAc, thereby channeling it directly back into PG biosynthesis (Borisova et al., 2017; Fumeaux and Bernhardt, 2017; Gisin et al., 2013) (Figure 1). As the last component, the peptide is typically broken down to single amino acids that are subsequently reused to synthesize new peptide stems (Gilmore and Cava, 2025). An alternative route has been reported for E. coli, which trims the released peptide by removal of the terminal D-ala residues and then attaches the resulting tripeptide to UDP-MurNAc using the peptide ligase Mpl (Mengin-Lecreulx et al., 1996).

The recycling of PG degradation products is an efficient way to save energy and resources, which may be particularly relevant during growth in nutrient-limiting conditions (Gilmore and Cava, 2025). For human commensals or pathogens, it also represents an important strategy for evading detection by the innate immune system, which senses released muropeptides through the NOD1 and NOD2 receptors (Trindade and Chen, 2020). Apart from that, some gammaproteobacteria such as Citrobacter freundii and P. aeruginosa use PG recycling as a sensory mechanism to detect the presence of β-lactam antibiotics and initiate a resistance response (Dik et al., 2018). These species possess the inducible β-lactamase AmpC, whose gene is controlled by the the LysR-type transcriptional regulator AmpR. The activity of AmpR is modulated through its interaction with muropeptides (Jacobs et al., 1994; Lindquist et al., 1989). In the absence of β-lactams, it is associated with UDP-MurNAc-pentapeptide, acting as a repressor that prevents ampC expression (Jacobs et al., 1997). When cells are exposed to β-lactam antibiotics, the disruption of PG metabolism leads to an increased accumulation of anhydro-muropeptides, thus raising the levels of PG recycling intermediates in the cytoplasm. Some of these intermediates, especially anhMurNAc-pentapep-tide (Dietz et al., 1997; Vadlamani et al., 2015), can interact with AmpR and competitively displace its co-repressor UDP-MurNAc-pentapeptide. As a result, AmpR transforms into a transcriptional activator that stimulates ampC expression, thereby establishing β-lactam resistance (Dietz et al., 1997; Dik et al., 2018; Jacobs et al., 1997). While PG recycling and its impact on cell physiology as well as β-lactam resistance have been extensively investigated in gammaproteobacteria, these processes are still poorly investigated in other bacterial lineages.

Here, we studied PG recycling in the alphaproteobacterium Caulobacter crescentus, a species intrinsically resistant to β-lactam antibiotics due to production of the β-lactamase BlaA, extending recent work on this topic by Modi et al. (Modi et al., 2025). We found that C. crescentus encodes a functional PG recycling pathway and characterized the activities of key enzymes, using a combination of biochemical, cell biological, metabolomics and proteomics analyses. Our results demonstrate that PG recycling is required to sustain proper cell morphology and division in this species, especially upon transition to stationary phase. They further show that C. crescentus cells regulate their PG recycling pathway to control the flux of internal and external GlcNAc into PG biosynthesis and central carbon metabolism, while employing the MurU pathway to directly convert anhMurNAc back into the PG precursors. Importantly, PG recycling-deficient cells show severe defects in β-lactam resistance, despite normal BlaA production and activity. For the model β-lactam ampicillin, this phenotype can be explained by increased sensitivity of the septal TPase PBP3 to ampicillin-mediated inhibition, likely resulting from a reduced supply of PG building blocks in the absence of PG recycling. Together, our findings indicate that β-lactam efficacy is modulated by the availability of PBP substrates, uncovering a new aspect of the complex interplay between this major class of antibiotics and their cellular targets.

Results

The genome of C. crescentus encodes a full PG recycling pathway

To initiate the analysis of PG recycling in C. crescentus, we first mined the genome of this species for homologs of proteins that are central to the well-studied PG recycling pathway of E. coli. Reciprocal BLAST analysis revealed clear counterparts for the muropeptide permease AmpG as well as for most enzymes involved in the processing of the imported anhydro-muropeptides and the transformation of their sugar and peptide moieties into new PG building blocks (Table 1 and Figure 1). Notably, no homologs were found for the LD-CPase LdcA and the peptide ligase Mpl, which trim tetrapeptides to tripeptides and attach these to UDP-MurNAc, respectively. Another key enzyme missing in C. crescentus is the etherase MurQ, a critical player in the recycling of anhMurNAc that converts MurNAc-6-P into GlcNAc-6-P and thus enables its reutilization through the GlcNAc recycling/degradation pathway (Jaeger et al., 2005; Uehara et al., 2006). Instead, we identified a three-step pathway initially found in P. aeruginosa (Borisova et al., 2017; Fumeaux and Bernhardt, 2017; Gisin et al., 2013) that leaves the MurNAc moiety intact and uses it directly to regenerate UDP-MurNAc. These findings suggest that the two sugar moieties of anhydro-muropeptides are recycled independently in C. crescentus. The protein homologs identified are generally well-conserved at the sequence level (Table 1) and show the same domain architectures as their E. coli and P. aeruginosa counterparts. The only exception is the homolog of AmpD, in which the amidase_2 domain is fused to a C-terminal PG_binding 1 domain that is not present in the E. coli protein. Due to this difference, this protein will be referred to as ‘AmiR’ (Amidase for PG Recycling) throughout the present study.

Table 1.

Predicted key players in the PG recycling pathway are functionally active

To test the prediction that C. crescentus can recycle PG degradation products, we aimed to verify the activity of the two first enzymes in the recycling pathway, AmiR and NagZ, which mediate the release of the sugar and peptide moieties, thereby providing the basis for their refunneling into PG biosynthesis (see Figure 1). To this end, pentapeptide-rich PG sacculi isolated from a DD-carboxypeptidase-deficient E. coli strain were treated with the LT Slt from E. coli (Betzner and Keck, 1989) to release anhydro-muropeptides (Figure 2A). These were then first incubated with AmiR or a catalytically inactive variant of the protein (AmiR*) in which three conserved amino acid residues coordinating the essential Zn²⁺ cofactor of AmpD-like amidases (Généreux et al., 2004; Liepinsh et al., 2003; Uehara and Park, 2007) were replaced by alanine (Figure 2–Figure supplement 1A). HPLC analysis of the reaction products revealed that AmiR was able to remove the peptide stems from all detectable anhydro-muropeptides, regardless of the length of the peptide or the number of crosslinked peptides (Figure 2B). By contrast, no amidase activity was detected in the control reactions containing the mutant AmiR* protein, confirming the critical role of the metal cofactor in the amidase reaction. To determine whether, despite its broad substrate specificity, AmiR had a preference for certain anhydro-muropeptides, we repeated the enzyme assay with limiting enzyme concentrations. The results showed that AmiR efficiently hydrolyzed tetrapeptide-containing fragments, whereas anhydro-muropeptides with tripeptide or pentapeptide stems were cleaved with lower efficiency (Figure 2–Figure supplement 1B). Notably, unlike E. coli AmpD (Höltje et al., 1994), it also showed activity towards muropeptides that were produced with the muramidase cellosyl and thus contained reducing MurNAc instead of anhMurNAc moieties, although the efficiencies were slightly lower under these conditions (Figure 2–Figure supplement 1C). By contrast, no activity was detectable with undigested peptidogylcan sacculi, consistent with a role of the protein in the recycling of PG fragments (Figure 2–Figure supplement 2). Together, these results confirmed that AmiR was a genuine amidase with a role in PG recycling. Next, we used the same approach to study the biochemical activity of C. crescentus NagZ. For this purpose, anhydro-muropeptides or reducing muropeptides were incubated with wild-type NagZ or a catalytically inactive variant (NagZ*) in which an essential aspartic acid residue at the catalytic site (Bacik et al., 2012) was replaced with an alanine (Figure 2–Figure supplement 3A). NagZ removed the GlcNAc moiety from both types of muropeptides and, in each case, hydrolyzed all detectable muropeptide variants, irrespective of the length of the peptide or the degree of crosslinkage (Figure 2C and Figure 2–Figure supplement 3B). By contrast, NagZ* did not show any activity. Thus, similar to its E. coli homolog (Cheng et al., 2000), NagZ acts as an N-acetyl-β-D-glucosaminidase with broad substrate specificity. Together, these findings demonstrate that C. crescentus possesses the enzymes mediating the two initial key steps of PG recycling.

Figure 2.

In E. coli, PG turnover products are imported by various transporters with different binding specificities (Gilmore and Cava, 2025), but only AmpG recognizes muropeptides containing a GlcNAc–anhMurNAc unit, as typically produced by LTs (Chang et al., 2025; Cheng and Park, 2002; Jacobs et al., 1994). To clarify whether the AmpG homolog of C. crescentus has a similar role, we used a metabolomics approach to detect imported anhydro-muropeptides, using the NagZ cleavage products anhMurNAc-pentapeptide, -tetrapeptide and -tripeptide as a proxy (Figure 3A). In the wild-type strain, the levels of these PG fragments were close to the detection limit, suggesting that they are rapidly processed into smaller units. Consistent with this notion, and in line with previous results (Jacobs et al., 1994; Simpson et al., 2023), a strong accumulation of these species was observed in a ΔamiR mutant, which lacks the amidase to remove the peptide stem (Figure 3B). Upon the deletion of ampG in the ΔamiR background, the three anhydro-muropeptides were again barely detectable, confirming that AmpG functions as the primary importer of LT-generated PG degradation products in C. crescentus.

Figure 3.

The analysis of cytoplasmic PG degradation intermediates also provided a means to obtain further insight into the nature and relevance of the reactions catalyzed by AmiR and NagZ. Consistent with the results of the in vitro activity assays (compare Figure 2C), the accumulation of anhMurNAc-peptide species in the absence of AmiR (Figure 3B) shows that NagZ can hydrolyze the glycosidic bond in the GlcNAc–anhMurNAc unit while the peptide stem is still attached. Moreover, it indicates that the reaction catalyzed by AmiR is a bottleneck in the PG recycling pathway, because the removal of the peptide appears to be critical for downstream reactions to occur. To clarify whether cleavage of the GlcNAc–anhMurNAc bond was required to further process the disaccharide unit after peptide removal, we assessed the levels of the free disaccharide in different mutant background. Metabolomics analysis revealed that GlcNAc–anhMurNAc was barely detectable in the wild-type strain, suggesting that AmiR catalyzes the rate-limiting step in the initial decomposition of anhydro-muropeptides. The disaccharide was also not enriched in ΔampG and ΔamiR cells, but highly accumulated in the ΔnagZ mutant (Figure 3C). Thus, the two sugars need to be separated before further processing steps can initiate.

Together, our analyses verify that C. crescentus possesses the machinery required to import anhydromuropeptides and break them down into their sugar and peptide components. Moreover, they indicate that AmiR and NagZ have a broad binding specificity, so that the initial steps of the recycling process do not occur in a strictly linear fashion.

PG recycling is necessary for proper cell division and β-lactam resistance in C. crescentus

Although a considerable fraction of the PG sacculus is turned over and recycled during each cell division cycle, the model species E. coli and P. aeruginosa typically do not display any obvious morphological phenotype upon disruption of the PG recycling pathway (Gilmore and Cava, 2025). The only exception are E. coli cells with a defect in the DD-carboxypeptidase LdcA, which promotes the erroneous ligation of UDP-MurNAc with unprocessed tetrapeptides, blocking the addition of the terminal D-Ala–D-Ala unit and inducing cell lysis in the stationary growth phase (Templin et al., 1999). To determine the relevance of PG recycling in C. crescentus, we examined the morphology of ΔampG, ΔamiR and ΔnagZ cells growing in rich (PYE) medium by light microscopy. In each case, the mutation caused significant cell filamentation and widening, with the defects becoming more pronounced upon transition of the cells from the exponential to the stationary growth phase (Figure 4A,B). Notably, strains in which amiR and nagZ genes were replaced with alleles encoding the catalytically inactive AmiR* or NagZ* showed phenotypes similar to those of the respective deletion mutants, verifying that the defects observed were due to the absence of the enzymatic activities (Figure 4–Figure supplement 1). These results indicate that PG recycling is critical for proper cell division and, to some extent, also lateral growth in C. crescentus, particularly under nutrient limitation. Notably, the lack of AmpG or AmiR, which prevents the recycling of the peptide stem (Figure 3B), caused a more severe phenotype than the lack of NagZ, which blocks the processing of the free GlcNAc– anhMurNAc disaccharide (Figure 3C). Therefore, the supply of amino acids rather than sugars appears to be the major limiting factor in PG precursor biosynthesis. To further investigate the effect of nutrients on the phenotypes of the different mutant strains, we also analyzed cells grown to stationary phase in minimal (M2G) medium. Interestingly, none of the mutant strains showed obvious morphological defects under these conditions (Figure 4–Figure supplement 2), suggesting that PG recycling is less important at lower growth rates or that the high glucose content of M2G medium compensates for the lack of PG recycling by promoting de novo PG synthesis.

Figure 4.

Apart from promoting growth, PG recycling was found to critically contribute to the regulation of β-lactam resistance in C. freundii and P. aeruginosa, because these species use the level of anhMurNAc-pentapeptide as a readout to sense the presence of β-lactams and control the production of the β-lactamase AmpC (Dietz et al., 1997; Dik et al., 2018; Jacobs et al., 1997). Previous work has also suggested a link between PG recycling and β-lactam resistance in C. crescentus (Modi et al., 2025), which produces a metallo-β-lactamase (BlaA) conferring intrinsic high-level resistance to ampicillin and related β-lactam antibiotics (Docquier et al., 2002; West et al., 2002). To further clarify the underlying mechanism, we analyzed the ability of the C. crescentus ΔampG, ΔamiR and ΔnagZ mutants to grow on nutrient-agar (PYE) plates at different ampicillin concentrations (Figure 4C). Wild-type cells tolerated high levels (50 µg/mL) of the antibiotic without any apparent fitness defects. In comparison, all three mutant strains showed a significant reduction in ampicillin resistance. The strongest effect was observed for ΔamiR cells, whose sensitivity approached that of ΔblaA cells (Figure 4C). Surprisingly, the ΔampG mutant showed the least pronounced resistance phenotype, although it exhibited the largest fraction of filamentous cells (compare Figure 4A,B), indicating that the defects in β-lactam resistance and cell morphology are not strictly connected. Together, these findings show that PG recycling is important for proper growth and β-lactam resistance in C. crescentus.

Notably, previous work has shown that inactivation of the divisome-associated soluble LT SdpA also leads to a severe decrease in ampicillin resistance in otherwise wild-type cells (Zielinska et al., 2017). This observation suggested that SdpA could represent the main LT in C. crescentus, producing the largest share of the anhydro-muropeptides that are fed into the PG recycling pathway. Consistent with this notion, cells lacking SdpA were highly sensitive to even low concentrations of ampicillin (Figure 4–Figure supplement 3A). Surprisingly, however, their resistance defect was more pronounced that of any of the PG recycling-deficient mutants investigated (compare Figure 4C). Moreover, in accordance with previous work, ΔsdpA cells did not show any signs of cell filamentation, indicating that their reduced ampicillin resistance may not be primarily due to a decrease in the supply of anhydro-muropeptides available for PG recycling (Figure 3–Figure supplement 3B,C). Metabolomics analysis indeed showed that the absence of SdpA barely affected the levels of cytoplasmic anhMurNAc-peptide species accumulating in the ΔamiR background (Figure 3–Figure supplement 3D). This finding supports the idea that SdpA is not the main producer of anhydro-muropeptides and likely affects β-lactam resistance through a different pathway, e.g., by contributing to proper cell wall biosynthesis and/or maintenance.

The recycling of GlcNAc involves two partially redundant enzymatic systems

After removal of the peptide and cleavage of the GlcNAc–anhMurNAc moiety, the two sugars are recycled in different ways. In E. coli, GlcNAc is transformed to GlcNAc-6-P by NagK, which is then reintroduced into de novo PG biosynthesis by conversion into GlcN-6-P, catalyzed by the deacetylase NagA (Figure 1). Bioinformatic analysis revealed that this branch of the PG recycling pathway is conserved in C. crescentus. Interestingly, however, this species contains two NagA homologs (75% identity), which we have named NagA1 and NagA2 (Table 1). To determine the physiological relevance of GlcNAc recycling, we generated C. crescentus ΔnagK, ΔnagA1 and ΔnagA2 mutants and analyzed their cell morphologies and resistance to ampicillin. None of the strains showed a significant morphological (Figure 5A,B) or resistance (Figure 5C) phenotype, again suggesting that the supply of sugars is not a major constraining factor in PG biosynthesis. Since it was conceivable that NagA1 and NagA2 had redundant functions, we also aimed to generate a double mutant lacking both nagA homologs to fully interrupt the second reaction of the GlcNAc recycling branch. Surprisingly, this was not possible, even though blocking the first step by deletion of nagK did not cause any obvious fitness defect, suggesting that the accumulation of GlcNAc-6-P in the complete absence of NagA activity (Plumbridge, 2009) may be toxic to the cells, possibly by allosterically activating the GlcN-6-P deaminase NagB (Álvarez-Añorve et al., 2016) and thus preventing de novo PG biosynthesis.

Figure 5.

In E. coli, the expression of nagA is controlled by GlcNAc-6-P, which interacts with the transcription factor NagC, thereby converting it from a repressor to an activator (Plumbridge, 1991). The system can also be activated by periplasmic GlcNAc, because E. coli possesses a phosphotransferase system (PTS) permease (NagE) that takes up and concomitantly phosphorylates this sugar, releasing it in the cytoplasm as GlcNAc6-P (Mukhija and Erni, 1996; Plumbridge, 2009). Since C. crescentus possesses two nagE homologs (Table 1), we decided to analyze its response to externally added GlcNAc to better understand the regulation of NagA1 and NagA2 in this species. Comparing the total proteomes of wild-type cells grown in minimal (M2G) medium with or without GlcNAc, we found that the presence of sugar strongly stimulated the expression of two distinct but similar gene clusters, containing nagA1 and nagA2 respectively (Figure 5D,E). One of these clusters (CCNA_00452-CCNA_00458) has previously been implicated in the uptake and metabolism of GlcNAc in C. crescentus, based on the fact that it includes a gene for an outer-membrane TonB-dependent receptor that transports GlcNAc and is essential for its utilization (Eisenbeis et al., 2008). The other genes in the cluster encode a NagE-like PTS permease, its cognate PTS phosphotransferase, NagA2 and a NagB homolog (Figure 5E), thus providing the complete set of proteins required to import GlcNAc from the periplasm and then funnel it either into PG biosynthesis or central caron metabolism (see Figure 1). Finally, the cluster contains genes for a β-N-acetylhexoaminidase, catalyzing the release of GlcNAc from chitin oligomers, and a transcriptional regulator of unknown function, which does not have an obvious homolog in E. coli. The second cluster (CCNA_00566-CCNA_00574) includes homologs of all the above-mentioned genes and additionally features genes for a predicted mannosidase and a predicted mannonate dehydratase, suggesting that it could be additionally involved in the metabolism of mannose derivatives. Collectively, the results of the mutational and proteomics analyses suggest that NagA1 and NagA2 are functionally redundant enzymes whose synthesis is activated by their substrate GlcNAc-6-P.

anhMurNAc is recycled through the MurU pathway

Bioinformatic analysis suggested that C. crescentus contains all enzymes of the P. aeruginosa MurU pathway, which recycles anhMurNAc independently of the GlcNAc recycling branch, converting it directly back to UDP-MurNAc (Borisova et al., 2017; Fumeaux and Bernhardt, 2017; Gisin et al., 2013) (Figure 1 and Table 1). To verify the functionality of this predicted pathway and clarify its impact on cell physiology, we introduced deletions in the anmK or amgK gene, thereby blocking essential steps in the reaction sequence. Microscopic analysis revealed that resulting single mutants did not have any significant morphological defects (Figure 6A,B). Importantly, however, both strains showed a moderate increase in their susceptibility to ampicillin (Figure 6C). This finding indicates that the MurU pathway is active and relevant for β-lactam resistance in C. crescentus, thus playing a more important role than the GlcNAc recycling branch (compare Figure 5A-C). To determine whether this pathway was the only means for C. crescentus to metabolize anhMurNAc, we tested whether its disruption would lead to an accumulation of recycling intermediates. Metabolomics analysis indeed revealed a considerable increase in the cytoplasmic level of MurNAc in cells lacking the MurNAc kinase AmgK, suggesting the absence of alternative recycling pathways for the anhMurNAc moiety (Figure 6D). Notably, previous work has shown that the formation of UDP-MurNAc from recycled anhMurNAc by the MurU pathway reduces the sensitivity of P. aeruginosa to the antibiotic fosfomycin, which inhibits MurA and thus hinders the de novo synthesis of this central PG precursor (Borisova et al., 2014). We reasoned that this effect should also be observable in C. crescentus if the MurU pathway was active in this species. Antibiotic resistance assays revealed that the ΔamgK mutant indeed showed higher sensitivity to fosfomycin than the wild-type strain or the ΔnagK mutant, which is defective in GlcNAc recycling (Figure 6E). Taken together, these findings strongly suggest that anhMurNAc is recycled via the MurU pathway in C. crescentus, thereby enhancing the supply of precursors for peptidoglycan biosynthesis. This process may be more significant than GlcNAc recycling (compare Figure 5A-C) because a reduced production of GlcN-6-P may be compensated by an increased production of this intermediate from fructose-6-phosphate, supplied by central carbon metabolism.

Figure 6.

As the recycling of GlcNAc and anhMurNAc is mediated by two distinct metabolic routes, cells blocked in one of these routes can still recycle half of the sugars generated by anhydro-muropeptide processing, potentially explaining the relatively mild phenotypes observed for the different single mutants. To clarify the effect caused by inhibiting both sugar recycling pathways, we generated a ΔnagK ΔamgK double mutant and analyzed its cell morphology and ampicillin resistance. Cells of this strain showed a slightly higher tendency to form filaments than the wild-type strain (Figure 7A,B). Interestingly, however, its susceptibility to ampicillin was not reduced beyond the level seen for the ΔamgK single mutant (Figure C). This observation may be attributed to the presence of GlcNAc in the rich medium used to grow the cells, which can be taken up and concomitantly phosphorylated by the NagE1 and NagE2 permease systems, thereby compensating for the loss of the GlcNAc kinase activity normally provided by NagK (see Figure 1).

Figure 7.

Defects in PG recycling do not change the levels or activity of the β-lactamase BlaA

In gammaproteobacteria such as P. aeruginosa and C. freundii, β-lactam resistance is established by increased synthesis of the β-lactamase AmpC in the presence of excess anhydro-muropentapeptide, mediated by the transcriptional regulator AmpR (Dietz et al., 1997; Dik et al., 2018; Jacobs et al., 1997). Previous work has suggested that the C. crescentus β-lactamase BlaA is constitutively active (Docquier et al., 2002; Modi et al., 2025), raising the question of how its protective effect could be dependent on PG recycling. To address this conundrum, we aimed to reinvestigate the regulaton of BlaA activity at the transcriptional, protein accumulation and activity level.

Inspecting the genomic contexts of PG recycling genes (Table 1), we found that amiR is located in a three-gene operon that also includes genes for a predicted drug/metabolite transporter (named TraX) and a PaiB-family transcriptional regulator (named RegX), suggesting that these proteins could be involved in PG recycling and/or the regulation of β-lactam resistance (Figure 8–Figure supplement 1A). However, in line with the dominant role of the permease AmpG in anhydro-muropeptide import (see Figure 3), the absence of TraX did not markedly affect the accumulation of anhMurNAc-peptides observed in the ΔamiR background. Similarly, whole-cell proteome analysis showed that the regulon of RegX did not include blaA or any genes related to PG recycling. Apart from this, neither ΔtraX nor ΔregX showed any obvious defects in cell division or ampicillin resistance, indicating that these two genes are not involved in PG-related processes (Figure 8–Figure supplement 1B-F).

Figure 8.

Interestingly, the blaA gene is located in a five-gene operon, which additionally encodes a predicted transcriptional regulator and three proteins involved in methionine biosynthesis (Figure 8A). To test whether the expression of this operon changes in response to β-lactam stress or disruption of the PG recycling pathway, we fused its upstream region (400 bp) to a β-galactosidase reporter gene (lacZ) and analyzed it for promoter activity in the ΔampG, ΔamiR and ΔnagZ backgrounds both in the absence or in the presence of ampicillin. The results showed that lacZ expression was constant in all conditions tested (Figure 8B), indicating that the operon as a whole was not subject to transcriptional regulation. However, it was possible that the transcriptional regulator controlled blaA expression by acting on an internal promoter. To test this hypothesis, we compared the whole-cell proteomes of the wild-type strain and a mutant lacking the regulatory gene (CCNA_02225). Surprisingly, in the mutant strain, the methionine biosynthetic proteins encoded by the genes flanking blaA were significantly upregulated, whereas the levels of BlaA remained essentially unchanged (Figure 8–Figure supplement 2). This finding may suggest that blaA is expressed from a strong and constitutive internal promoter, so that moderate changes in the transcription of the whole operon do not markedly affect the overall levels of the blaA transcripts (Plumbridge, 1996). Finally, we tested whether the activity of BlaA could be regulated post-transcriptionally at the level of enzyme activity. To this end, we incubated cells with the chromogenic β-lactam nitrocefin (O’Callaghan et al., 1972) and used its property of developing red color upon cleavage to measure β-lactamase activity in vivo. We found the ΔampG, ΔamiR and ΔnagZ mutants showed behaviors similar to that of the wild-type strain, while the ΔblaA mutant lacked activity, validating the assay (Figure 8C). Consistent with this finding, whole-cell proteomic analyses showed that cells impaired in PG recycling accumulated BlaA at levels comparable to those in the wild-type strain (Figure 8D). Collectively, these results demonstrate that the severe reduction in β-lactam resistance observed in PG recycling-deficient strains is not due to changes in BlaA accumulation or activity. This conclusion is supported by the results of growth assays showing that defects in the PG recycling pathway also reduce the resistance of cells to aztreonam, a β-lactam antibiotic that is not cleaved by metallo-β-lactamases such as BlaA (Bush et al., 1982) (Figure 8E). Thus, the ampicillin sensitivity phenotypes observed are likely due to changes in the susceptibility and/or activity of target proteins rather than impaired degradation of this antibiotic in the absence of PG recycling.

PG recycling defects sensitize the PG biosynthetic machinery to the inhibitory effect of ampicillin

To shed more light on the connection between PG recycling and β-lactam resistance, we aimed to investigate the growth defects induced by the exposure of cells to ampicillin in more detail. To this end, we cultivated wild-type and ΔamiR cells in liquid medium containing different concentrations of the antibiotic and analyzed their morphology by light microscopy. Wild-type cells tolerated substantial concentrations of ampicillin without noticeable changes in cell shape (Figure 9A,B). However, at 500 µg/mL, they developed a pronounced filamentous phenotype, while exposure to 5,000 µg/mL led to growth arrest and cell lysis. For the ΔamiR mutant, similar phenotypes were observed, but cell filamentation and lysis already occurred at the lowest ampicillin concentration tested and were consistently more pronounced than for the wild type across all conditions (Figure 9A,B). To better characterize the development of the observed morphological defects, we monitored the growth behavior of the two strains following a shift from plain rich medium to medium supplemented with a low concentration of ampicillin (20 µg/ml). Whereas the wild-type strain grew normally in both conditions (Figure 9C and Figure 9–video 1), ΔamiR cells started to become filamentous and swollen shortly after exposure to the antibiotic and were unable to divide (Figure 9C and Figure 9–video 2). Together, these results indicate that, in C. crescentus, ampicillin preferentially targets cell division, although it also affects lateral growth when applied at elevated concentrations. Moreover, they suggest that PG recycling defects enhance the efficacy of ampicillin by rendering the PG biosynthetic machinery more sensitive to its inhibitory effect.

Figure 9.

To shed more light on the inhibitory activity of ampicillin, we aimed to identify its target proteins. For this purpose, we adapted an assay enabling the visualization of PBPs in SDS-gels after labeling with the fluorescent β-lactam bocillin-FL (Zhao et al., 1999). Using ΔblaA cells to prevent β-lactam degradation, this approach detected all known functional transpeptidases of C. crescentus except for PBP2 (Strobel et al., 2014; Yakhnina and Gitai, 2013) as well as the predicted carboxypeptidase CrbA (Figure 9D). However, when these proteins were incubated with ampicillin prior to the addition of bocillin-FL, all but one of them (PbpX) remained unlabeled, indicating that ampicillin had blocked their active-site serine, thereby preventing the subsequent reaction with bocillin-FL. Among the ampicillin targets identified were the essential, cell division-specific monofunctional transpeptidase PBP3 as well as the bifunctional PBPs Pbp1A, PbpY and PbpC. The labeling of PBP3 and some bifunctional PBPs was also abolished by preincubation with cephalexin, a β-lactam that specifically targets PBP3 in E. coli (Kocaoglu and Carlson, 2015), whereas the PBP2-specific β-lactam mecillinam had only minor effects on the labeling pattern (Figure 9D). Given that Pbp1A, PbpY and PbpC have been shown to be dispensible for cell division in C. crescentus (Strobel et al., 2014; Yakhnina and Gitai, 2013), these results indicate that the cell filamentation phenotype induced by ampicillin is primarily due to inhibition of PBP3 activity.

PG recycling defects may increase ampicillin sensitivity by limiting the supply of PG precursors

The results obtained suggest that impaired PG recycling does not affect BlaA activity and may rather enhance the inhibitory effect of ampicillin on PG biosynthesis. It was conceivable that the accumulation of unrecycled PG degradation products in the periplasm could impair the activity of certain PG biosynthetic enzymes, thereby disrupting PG homeostasis. However, sacculi isolated from wild-type and ΔamiR cells did not show any significant differences in their muropeptide composition (Figure 9–Figure supplement 1), suggesting that a defect in PG recycling does not perturb the global balance of PG biosynthetic and autolytic processes. Notably, mutants blocked in the PG recycling pathway also exhibit a conditional cell division defect, even when not exposed to ampicillin. It was therefore possible that their higher β-lactam sensitivity was linked to a decrease in the overall efficiency of PG biosynthesis, caused by impaired replenishment of PG precursors from recycling products. Metabolomics analysis revealed that the ΔampG, ΔamiR and ΔnagZ mutants indeed showed significantly lower levels of UDP-sugars than the wild-type strain (Figure 9E). The absence of AmpG and NagZ led to a more than 50% decrease in the UDP-GlcNAc pool, confirming that de novo synthesis alone is not sufficient to satisfy the cellular demand for this central metabolite. By contrast, UDP-GlcNAc levels remained unchanged in ΔamiR cells, consistent with the observation that GlcNAc can still be released from imported anhydro-muropeptides in the absence of amidase activity (see Figures 2C and 3C). Even stronger effects were observed on the UDP-MurNAc pool, which was drastically reduced in all three mutant strains (Figure 9E). Its depletion in the absence of PG recycling suggests that the formation of UDP-MurNAc constitutes a bottleneck in the de novo PG biosynthetic pathway, with cells relying heavily on anhMurNAc recycling via the MurU pathway to maintain an adequate supply. Collectively, these results demonstrate that PG recycling is critical to provide sufficient activated sugars for PG precursor biosynthesis. However, although the ΔampG, ΔamiR and ΔnagZ mutants show a very similar decrease in cytoplasmic UDP-MurNAc levels, their morphological and resistance defects differ significantly, indicating that the reduced availability of UDP-sugars may not be the only reason for the phenotypes observed. Consistent with this notion, an increase in cytoplasmic GlcNAc-P levels through supplementation of the growth medium with GlcNAc (see Figure 5) was not sufficient to improve ampicillin tolerance in the three mutant strains (Figure 9–Figure supplement 2).

It was conceivable that the limited supply of PG precursors in the absence of proper PG recycling causes cell filamentation and lysis by reducing the level of association between PBPs and their substrates, thereby slowing their catalytic activity. To test this idea, we investigated whether the cell division defect of the ΔamiR mutant could be ameliorated by artificially increasing the activity of the septal PG biosynthetic machinery. To this end, employed a hyperactive (A246T) variant of the glycosyltransferase FtsW (FtsW*) that leads to increased activity of the septal FtsW-PBP3 PG biosynthetic complex (Lariviere et al., 2019; Modell et al., 2014), resulting in elevated cell constriction rates and shorter cell lengths, especially in stationary phase (Figure 10A,B). Interestingly, introduction of the ftsW* allele almost completely rescued the cell filamentation phenotype of the ΔamiR mutant in stationary phase, although it had only little effect in exponentially growing cells, possibly because PG precursors are more limiting under conditions of rapid growth. Thus, hyperactivation of the FtsW-PBP3 complex can, at least in part, reverse the cell division defect induced by defective PG recycling, suggesting that the cell filamentation phenotype observed in the mutant backgrounds could indeed be due to a reduced activity of this complex. Notably, however, the presence of the ftsW*allele did not relieve the ampicillin sensitivity of ΔamiR cells (Figure 10C), indicating that an increased activity of the septal PG biosynthetic complex does not reduce its ampicillin-mediated inactivation.

Figure 10.

Discussion

Bacterial PG biosynthesis has been studied intensively over the past decades. Importantly, apart from the proper spatiotemporal control of peptidoglycan incorporation and cleavage during cell growth and cell division, the recycling of PG degradation products has emerged to play a major role in cellular fitness and survival. This process not only conserves valuable resources (Gilmore and Cava, 2025) but also plays an important role in infection by enabling bacteria to evade the human innate immune system (Trindade and Chen, 2020). Moreover, it was found to critically contribute to AmpR-mediated regulation of β-lactam resistance in P. aeruginosa and various enterobacterial pathogens (Dik et al., 2018). Although substantial advances have been made in elucidating the mechanisms and physiological roles of PG recycling within Gammaproteobacteria, these aspects remain incompletely characterized across other bacterial lineages (Gilmore and Cava, 2025). In this study, we demonstrate that the alphaproteobacterial model organism C. crescentus possesses a functional PG recycling pathway that includes components from both the E. coli and P. aeruginosa systems, extending the results of a recent study on this topic (Modi et al., 2025). We find that its function is required for maintaining an adequate supply of peptidoglycan precursors, complementing de novo synthesis. C. crescentus cells impaired in PG recycling show marked defects in cell division and cell wall integrity, in particular upon transition to the stationary growth phase. Moreover, they exhibit a pronounced decrease in intrinsic β-lactam resistance, although the levels and activity of their major β-lactamase BlaA remain unchanged. Blocking critical steps in the recycling pathway therefore emerges as a promising new strategy to combat β-lactam resistant pathogens and thus mitigate the global antibiotic crisis.

Our analyses revealed that C. crescentus possesses almost all core PG recycling enzymes identified in gammaproteobacteria. Notable exceptions are the LD-CPase LdcA (Templin et al., 1999), which trims sugar-linked and free tetrapeptides to tripeptides, as well as the murein peptide ligase Mpl, which re-attaches free peptide stems to UDP-MurNAc, thereby obviating the need for their de novo synthesis (Mengin-Lecreulx et al., 1996). In E. coli, the activity of LdcA is critical for cell envelope integrity, because its inactivation leads to the accumulation of free tetrapeptides, which are then erroneously ligated with UDP-MurNAc, giving rise to non-crosslinkable tetrapeptide-containing PG precursors, whose incorporation into peptidoglycan severely destabilizes the cell wall structure (Templin et al., 1999). In some gammaproteobacteria, the function of LdcA is taken over by other, unrelated peptidases (Dai et al., 2021; Hernández et al., 2020; Simpson et al., 2021), opening the possibility that C. crescentus possesses a thus-far unknown peptide-processing LD-CPase. However, metabolomic analysis showed that ΔamiR cells accumulated high concentrations of anhMurNAc-tetrapeptide in their cytoplasm (Figure 3B), making the existence of such an enzyme unlikely. In line with the absence of Mpl, C. crescentus may thus not recycle tripeptides but rather degrade peptide stems released by AmiR into individual amino acids, which are subsequently reused for de novo PG biosynthesis. A similar situation may be seen for many other Gram-negative and Gram-positive species that lack both an LdcA and an Mpl homolog (Gilmore and Cava, 2025).

Apart from the removal of the peptide stem by AmiR, the cleavage of the GlcNAc–anhMurNAc moiety by NagZ is a critical step in the initial processing of imported anhydro-muropeptide species. The results of our enzyme assays (Figures 2) and metabolomic analyses (Figure 3) show that these two reactions do not follow a defined order but occur independently of each other. Once released, the two amino sugars follow distinct metabolic routes. GlcNAc is phosphorylated to GlcNAc-6-P and then deacetylated by NagA, yielding GlcN-6-P, which is then either used as the starting point of de novo PG biosynthesis or, alternatively, fed into central carbon metabolism through deamination to frucose-6-P by NagB (Figure 1). This branch point is critical since GlcNAc is not only produced during peptidoglycan degradation but also acquired from the environment as a readily available carbon source, primarily derived from the breakdown of chitin (Beier and Bertilsson, 2013). In the presence of external GlcNAc, two related gene clusters are highly upregulated, each of them including genes for an outer-membrane TonB-dependent receptor, a PTS permease together with its cognate phosphorylation system as well as a homolog of the deacetylase NagA and the deaminase NagB (Figure 5D,E). Their regulation is likely mediated by GntR-type transcriptional regulators encoded in the same clusters, which may sense the accumulation of GlcNAc-6-P as the first common cytoplasmic intermediate of the PG recycling and GlcNAc uptake pathways, thereby adapting the flux through the GlcNAc recycling/utilization pathway to substrate availability. While one of the TonB-dependent receptors (CCNA_00455) was shown to be specific for GlcNAc (Eisenbeis et al., 2008), the substrate range of the other transport systems still remains to be determined. Overall, the GlcNAc recycling/utilization system of C. crescentus shows parallels to that of E. coli, which also involves two different PTS permeases, NagE and ManXYZ, for GlcNAc transport (Plumbridge and Vimr, 1999). However, E. coli lacks a GlcNAc-specific TonB-dependent receptor, shows a different organization and regulation of the genes involved and only possesses a single set of NagA and NagB proteins. Even more pronounced differences are observed for the recycling of anhMurNAc, which in both species initiates with the formation of anhMurNAc-6-P. In E. coli, this intermediate is converted to GlcNAc-6-P by the etherase MurQ and thus funneled into the GlcNAc recycling pathway (Jaeger et al., 2005; Uehara et al., 2006). C. cresentus, by contrast, lacks this enzyme and instead uses the alternative MurU pathway, a short-cut that transforms MurNAc-6-P to UDP-MurNAc and thus feeds it back directly into PG biosynthesis (Figures 6 and 7). This pathway has so-far only been observed in a few other species such as the gammaproteobacteria P. aeruginosa (Borisova et al., 2017; Fumeaux and Bernhardt, 2017; Gisin et al., 2013) and Legionella pneumophila (Ratna et al., 2025) as well as Tannerella forsythia (Hottmann et al., 2021), a member of the Bacteroidia. Our results, combined with those of a previous study (Modi et al., 2025), show that it is more widespread than previously anticipated and also conserved in the alphaproteobacterial lineage.

In previously studied model species, PG recycling was found to be dispensable under laboratory conditions and defects in this process did not lead to any obvious morphological defects (Gilmore and Cava, 2025), except for the lack of LdcA, which causes cell rounding and lysis in stationary phase in E. coli (Templin et al., 1999). In C. crescentus, it was also possible to introduce single deletions in all PG recycling-related genes, although it was not possible to create a strain lacking both nagA homologs simultaneously, suggesting that the accumulation of GlcNAc-6-P may be toxic to the cells. Notably, however, and contrary to a previous report (Modi et al., 2025), most mutants displayed cell division defects and cell filamentation, especially when transitioning to stationary phase. These phenotypes were particularly pronounced for ΔampG and ΔamiR cells, which are impaired in anhydro-muropeptide uptake or peptide stem cleavage. By contrast, single disruptions in the sugar recycling pathways generally did not result in cell elongation, suggesting that the observed filamentation phenotype is linked to impaired peptide recycling. Interestingly, while completely abolishing sugar recycling through inactivation of NagZ did not cause any obvious morphological defects, blocking both the GlcNAc and anhMurNAc recycling branches led to cell filamentation. This conundrum may be explained by the accumulation of metabolic intermediates in the double mutant that may cause misregulation of PG biosynthesis by affecting gene expression or enzyme activities. The precise link between PG recycling and cell division remains to be clarified. However, we observed that blocking the initial steps of PG recycling leads to a severe reduction in the levels of PG precursors such as UDP-GlcNAc and UDP-MurNAc (Figure 9E). This effect may be even more pronounced for the peptide-containing downstream products of the PG biosynthetic pathway, which were not detectable with our methods, in particular in mutant backgrounds preventing the recycling of the peptide stem. Importantly, the cell division defects caused by impaired PG recycling were alleviated by hyperactivation of the FtsW-PBP3 complex (Figure 10). Together, these findings suggest that reduced PG precursor levels impair the association of divisome-associated GTases and PBPs with their substrates, leading to lower catalytic rates and a slow-down of cell constriction. This effect may be further aggravated in the presence of ampicillin, because β-lactam-mediated inhibition of transpeptidase activity can hyperactivate the GTase domain of bifunctional PBPs. As a consequence, they produce an abundance of non-crosslinked glycan strands that are rapidly degraded after synthesis, leading to further depletion of the lipid II pool (Banzhaf et al., 2012; Cho et al., 2014; Kohlrausch and Höltje, 1991). Moreover, apart from decreasing PG synthase activity, defects in PG recycling may also cause an accumulation of anhydro-muropeptides in the periplasm, which could potentially interfere with normal PG biosynthesis by occupying binding sites or promoting non-productive side-reactions. Interestingly, elongation appears to continue normally under these conditions, likely because it requires significantly less PG incorporation than cell constriction and the formation of the new cell poles. Notably, the limitation of PG precursors can be compensated by enhancing the overall catalytic activity of the FtsW-PBP3 complex and thereby increasing its reaction rate under conditions of reduced binding site occupancy.

Strikingly, defects in PG recycling not only caused cell filamentation but also strongly reduced the intrinsic resistance of C. crescentus to ampicillin. This phenomenon was most pronounced for mutants with defects in the first PG recycling steps that completely abolished peptide and/or sugar recycling (Figure 4). A milder effect was observed for mutants blocked in the MurU pathway, whereas disrupting GlcNAc recycling did not produce any obvious ampicillin sensitivity phenotype. Our results suggest that the formation of UDP-MurNAc as the first committed step in PG biosynthesis represents a metabolic bottleneck (Figure 9E), making anhMurNAc recycling through the MurU shunt essential for maintaining adequate PG precursor levels. UDP-GlcNAc, by contrast, which is used in the biosynthesis in various cell envelope polysaccharides, can be more readily replenished because its biosynthesis is supported not only by de novo synthesis and peptidoglycan (PG) recycling but also by the uptake of external GlcNAc from the growth medium (Eisenbeis et al., 2008) (Figure 5D,E). Notably, even high concentrations of external GlcNAc were not sufficient to restore ampicillin resistance in any of the PG recycling-deficient mutants tested (Figure 9–Figure supplement 2). Contrary to a previous report (Modi et al., 2025), this finding supports the notion that their β-lactam sensitivity is not attributable to an inadequate supply of GlcNAc for PG precursor biosynthesis.

A role of PG recycling in β-lactam resistance has been previously reported for gammaproteobacteria such as P. aeruginosa and C. freundii. In these species, β-lactam exposure leads to an increase in the cytoplasmic level of anhMurNAc-pentapeptide, thereby stimulating its interaction with the transcriptional regulator AmpR and inducing expression of the ampC β-lactamase gene (Dik et al., 2018). However, recent work has indicated that different mechanisms may be at work in Alphaproteobacteria (Gilmore and Cava, 2022; Modi et al., 2025). Our data show that, in C. crescentus, the accumulation and activity of the main β-lactamase BlaA remain unchanged in the presence of ampicillin and upon inhibition of PG recycling. The lower ampicillin resistance level of PG recycling-deficient mutants may therefore not be due to less efficient inactivation of ampicillin but rather an increased susceptibility of PBPs to this antibiotic. This effect could potentially be explained by competition of ampicillin and PG precursors for the substrate-binding sites of target PBPs (Figure 11). In wild-type cells, the abundant supply of precursor ensures the saturation of PBPs with their cognate substrate. As a consequence, most ampicillin molecules will be cleaved by BlaA before they manage to covalently modify their target proteins, leaving the vast majority of PBPs in a functional state. However, if defects in PG recycling limit the availability of PG precursors, the substrate binding sites of PBPs remain unoccupied for extended periods of time. As a consequence, they become more susceptible to interaction with residual ampicillin molecules, leading to an increase in the fraction of inactive enzymes. The resulting defect in cell division and cell wall integrity may be further aggravated by the reduced activity of the remaining PBPs due to inadequate substrate levels and the interference of periplasmic anhydro-muropeptides with proper PG biosynthesis. In different PG recycling-deficient mutants, these factors may be manifested to varying degrees, which could account for differences in the severity between the cell filamentation and ampicillin sensitivity phenotypes in many of the mutant strains.

Figure 11.

Collectively, our results demonstrate that PG recycling is critical for proper growth, cell division and β-lactam resistance in C. crescentus, thus playing a more significant role in bacterial fitness and survival than previously appreciated. Moreover, they support the idea that bacterial β-lactam resistance is a complex trait, determined not only by specific resistance proteins but also by the supply of PG building blocks and the dynamics of PG biosynthesis. Given the widespread occurrence of PG recycling among bacteria, it will be important to determine the impact of this process on cell viability and antibiotic resistance in other species and to evaluate its inhibition as a potential new strategy to potentiate the efficacy of β-lactam antibiotics in antibacterial therapies.

Methods

Experimental models

The bacterial strains used in this study are derivatives of Caulobacter crescentus NA1000 (CB15N) (Poindexter, 1964), Escherichia coli TOP10 (Thermo Fisher Scientific, Germany) or Escherichia coli Rosetta(DE3)pLysS (Merck, Germany). The sequences of all genes and proteins investigated are based on the genome sequence of C. crescentus NA1000 (Nierman et al., 2001).

Cultivation of bacterial strains

C. crescentus NA1000 (CB15N) and its derivatives were grown at 28 °C in peptone-yeast-extract (PYE) medium (Poindexter, 1964), double-concentrated PYE (2xPYE) medium or M2 salts-glucose (M2G) minimal medium (Hottes et al., 2004) under aerobic conditions, shaking at 210 rpm. To prepare PYE plates, liquid medium was solidified by the addition of 1.5% agar. To ensure the maintenance of integrative plasmids, media were supplemented with kanamycin (5 µg/mL for liquid cultures, 25 µg/mL for plates). To counter-select against cells carrying the sacB marker, solid PYE medium was supplemented with 3% D-sucrose. E. coli TOP10 (Invitrogen, Germany) was used for cloning purposes, whereas E. coli Rosetta(DE3)pLysS (Novagen, Germany) was used for protein overproduction. Cells were grown aerobically at 37 °C in LB Broth or on LB Agar (Carl Roth, Germany), supplemented when appropriate with antibiotics at the following concentrations (µg/mL; liquid/solid): kanamycin (30/50), ampicillin (200/200), chloramphenicol (20/30).

Plasmid and strain construction

The bacterial strains, plasmids and oligonucleotides used in this study are listed in Tables S1-S3. E. coli TOP10 (Invitrogen) was used as host for cloning purposes. C. crescentus was transformed by electroporation (Ely, 1991). Gene replacement was achieved by double homologous recombination using the counter-selectable sacB marker (Thanbichler and Shapiro, 2006). Proper chromosomal integration or gene replacement was verified by colony PCR.

Antibiotic susceptibility assay

Overnight cultures were serially diluted (10^-2 to 10^-7) in PYE medium and spotted on PYE plates containing ampicillin (10-50 µg/mL) or aztreonam (95-105 µg/mL) at the indicated concentrations. When appropriate, the media were additionally supplemented with 0.3% GlcNAc. The plates were incubated for three days at room temperature (25 °C) and imaged using a ChemiDocTM XRS+ Imager (Biorad, Germany) and Image Lab™ software (version 5.0) (Biorad, Germany). To account for the high instability of ampicillin in aqueous solution, ampicillin stocks were prepared freshly before every experiment.

The resistance against fosfomycin was assessed using Fosfomycin MIC Test Strips (Liofilchem, Italy). Overnight cultures were diluted with PYE medium to an OD₆₀₀ of 0.1. Subsequently, 100 µL of the suspensions were added to 20 mL of sterile liquid PYE agar that had been cooled to slightly above the gelation temperature (∼45 °C). After mixing, the suspension was immediately poured into cultivation dishes. After solidification of the medium, the plates were briefly dried and the test strips were applied on their top. Subsequently, they were incubated for three days at 28 °C and imaged using a ChemiDocTM XRS+ Imager and Image Lab™ 5.0 software (Bio-Rad Laboratories, Germany).

Widefield fluorescence microscopy

Unless indicated otherwise, cells were grown in PYE medium, transferred to 1% agarose pads and imaged with a Zeiss Axio Imager.Z1 microscope equipped with a Plan Apochromat 100×/1.45 Oil DIC and a 100×/1.40 Oil Ph3 M27 phase contrast objective. Images were recorded with a pco.edge 4.2 sCMOS camera (PCO). An X-Cite 120PC metal halide light source (EXFO, Canada) and an ET-DAPI filter cube (Chroma, USA) were used for the detection of fluorescence signals. Images were recorded with VisiView 4.0.0.14 (Visitron Systems, Germany) and processed with Fiji 1.53 (Schindelin et al., 2012). Cell length measurements were performed with Oufti (Paintdakhi et al., 2016), using phase-contrast images as an input.

Microfluidics-based ampicillin sensitivity assay

Microfluidic-based imaging was performed using a CellASIC ONIX B04A plate for bacterial cells (Merck, Germany) that was connected to a CellASIC® ONIX2 microfluidic system (Merck, Germany) and imaged with an Zeiss Axio Observer.Z1 (see above) in a heatable enclosure adjusted to 28 °C. Exponentially growing C. crescentus wild-type and ΔamiR cells were diluted to an OD₆₀₀ of 0.2 and flushed into different chambers of the microfluidic plate. They were then first incubated for 3 h with a constant flow of PYE medium and then for 5 h with a constant flow of PYE medium containing 20 µg/mL ampicillin at a pressure of 5 kPa. DIC images were taken at 5-min intervals to follow cell growth and division.

β-lactamase activity assay

ß-lactamase activity was assessed using the chromogenic cephalosporin derivative nitrocefin (Merck, Germany) as a substrate (O’Callaghan et al., 1972). Cells were grown to exponential phase in 3 mL PYE medium, harvested by centrifugation for 5 min at 7,000 ×g and room temperature, and resuspended in 1 mL PBS. After transfer of the suspensions into a 24-well polystyrene microplate (Carl Roth, Germany), the reactions were started by the addition of 100 µL nitrocefin (1 mg/mL). The hydrolysis of nitrocefin was then monitored by measuring the increase in absorbance at 482 nm every 2 min over a period of 180 minutes at 30 °C using an Epoch™ 2 microplate reader (BioTek, California, United States).

Identification of ampicillin target proteins

C. crescentus CS606 (ΔblaA) cells were grown to exponential phase in 100 mL PYE medium, harvested, washed with PBS and lysed by three passages through a French press at 10,000 psi. Insoluble material, including the membrane fraction, was collected by centrifugation for 15 min at 21,000 ×g (4 °C). After resuspension of the pellet in 2 mL of PBS, 100-µL aliquots were incubated with 0, 1, 10 or 100 µg/mL ampicillin, 5 µg/mL cefalexin or 100 µg/mL mecillinam for 2 h at room temperature. The membranes were collected again by centrifugation (1 min, 13,000 ×g, room temperature), washed twice with PBS and resuspended in a final volume of 100 µL PBS containing 5 µg/mL Bocillin-FL (Thermo Fisher Scientific, Germany). After 10 min of incubation, the membranes were collected by centrifugation (1 min, 13,000 ×g, room temperature), washed with PBS, dissolved in 30 µL of 2x SDS sample buffer (0.125 M Tris Base, 20 % [v/v] glycerol, 4 % [w/v] SDS, 200 mM DTT, 0.01 % [w/v] bromophenol blue, pH 6.8), heated for 10 min at 90 °C and loaded onto an 8% SDS-polyacrylamide gel. After separation by electrophoresis, proteins labelled with Bocillin-FL were detected with a ChemiDoc^TM MP imager (Bio-Rad Laboratories, Germany) using a 526 nm short-pass filter.

β-galactosidase assay

Cultures were grown to exponential phase in PYE medium. After determining the cell densities (OD₆₀₀), the cells were harvested by centrifugation for 7 min at 12,000 ×g and 4 °C and resuspended in 1 mL of Z-buffer (60 mM Na₂PO₄, 40 mM NaH₂PO₄, 10 mM KCl, 1 mM MgSO₄, 50 mM β-mercaptoethanol, pH 7.0). Following the addition of 100 µL chloroform and 0.1 % SDS, the suspensions were briefly agitated and then incubated for 7 min at room temperature to permeabilize the cells. 500-µL samples of the suspensions were mixed with 500 µL of Z-buffer. The reactions were then started by the addition of 200 µL o-nitrophenyl-β-D-galactopyranoside (4 mg/mL), incubated at room temperature until the mixtures turned yellow, and stopped by the addition of 500 µL 1 M Na₂CO₃. Cell remnants were pelleted by centrifugation and the extinction of the supernatant was measured photometrically at 550 nm and 420 nm. β-galactosidase activities were then calculated as follows: Activity (in Miller units)=1000*(OD₄₂₀–1.75*OD₅₅₀)/(t*V*OD₆₀₀), where t is the time of incubation (min) and V the culture volume used for the reaction (mL).

Protein purification

To purify AmiR and AmiR* (H26A/H133A/D143A), E. coli Rosetta(DE3)pLysS was transformed with plasmids encoding His₆-SUMO-tagged derivatives (Marblestone et al., 2006) of the proteins (pPR029 or pPR041) and grown in LB medium to an OD₆₀₀ o 0.4. After reduction of the growth temperature to 18 °C, the cultures were supplemented with isopropyl-β-D-thiogalactopyranoside (IPTG) to a final concentration of 1 mM, and incubated further overnight. The cells were harvested by centrifugation (12 min, 5000 ×g, 4 °C), washed in buffer A (50 mM Tris-HCl, 500 mM NaCl, 10 mM imidazole, 10% glycerol, pH 7.0) and stored at -80 °C until needed. After thawing, the cells were resuspended in buffer A containing 100 μg/mL phenylmethylsulfonyl fluoride (PMSF) and 10 μg/mL DNaseI and lysed by three passages through a French press at a pressure of 10,000 psi. The lysate was cleared by centrifugation (30 min, 25,000 ×g, 4 °C) and loaded onto a His-Trap HP 5 mL Ni-NTA affinity column (Cytiva, Germany) equilibrated with buffer A. After washing of the column with buffer A, protein was eluted at a flow rate of 2 mL/min with a linear gradient of imidazole generated by mixing buffer A and buffer B (50 mM Tris-HCl, 500 mM NaCl, 250 mM imidazole, 10% glycerol, pH 7.0). The peak fractions were pooled, supplemented with SUMO protease (His₆-Ulp) (Marblestone et al., 2006) at a molar ratio of 1:1,000 relative to the purified protein, and dialyzed overnight at 4 °C against buffer C (50 mM Tris-HCl, 500 mM NaCl, 10% glycerol, 1 mM dithiothreitol, pH 7.0). The solution was then loaded again onto a His-Trap HP 5 mL column to remove His₆-SUMO and His₆-Ulp. Flow-through fractions containing the protein of interest were collected, concentrated using Amicon® Ultra centrifugation filters (10kDa MWCO; Merck, Germany), and loaded onto a Superdex 75 GL10/300 size-exclusion column (Cytiva, Germany) equilibrated with buffer D (50 mM HEPES-NaOH, 500 mM NaCl, pH 7.0) and eluted at a flow rate of 10 mL/min. Fractions containing the protein of interest at high concentration and purity were combined, concentrated, snap-frozen in liquid nitrogen and stored at -80 °C until further use.

To purify NagZ and NagZ* (D259A), His₆-SUMO-tagged derivatives of the proteins were overproduced in E. coli Rosetta(DE3)pLysS carrying suitable plasmids (pPR102 or pPR107) as described above. The fusion proteins were then purified on a His-Trap HP 5 mL column, cleaved with SUMO protease, and separated from His₆-SUMO and His₆-Ulp by a second affinity purification step essentially as described for AmiR, with the exceptions that buffers A and B contained 300 mM instead of 500 mM NaCl and buffer C contained 150 mM instead of 500 mM NaCl. Flow-through fractions from the His-Trap HP column that contained the protein of interest at high concentration and purity were concentrated, snap-frozen in liquid nitrogen and stored at -80 °C.

Muropeptide analysis

The C. crescentus wild type and the ΔamiR mutant were grown to stationary phase (OD₆₀₀ ≈ 1.2) in 400 mL PYE medium and harvested by centrifugation at 16,000 ×g and 4 °C for 30 min. The pelleted cells were resuspended in 6 mL of ice-cold double-ionized water (_ddH₂O) and added dropwise to 6 mL of a boiling solution of 8% sodium dodecylsulfate (SDS), which was vigorously stirred. The suspension was boiled for an additional 30 min, and double-deionized water was added regularly to maintain a constant volume. Subsequently, the lysate was cooled to room temperature and stored at 4 °C until further use. Peptidoglycan was isolated following previously described procedures (Glauner, 1988; Takacs et al., 2010) and digested with the muramidase cellosyl (kindly provided by Hoechst, Frankfurt, Germany). The resulting muropeptides were reduced with sodium borohydride and separated by HPLC as described previously (Takacs et al., 2010). The identity of eluted fragments was assigned based on the retention times of known muropeptides (Takacs et al., 2010).

AmiR and NagZ activity assays

To test the activity of AmiR towards anhydro-muropeptides, peptidoglycan from E. coli D456 (Edwards and Donachie, 1993), which lacks three major DD-carboxypeptidases and thus has elevated levels of pentapeptides, was treated with the soluble lytic transglycosylase Slt70 from E. coli (Banzhaf et al., 2020; Betzner and Keck, 1989). The products were incubated with wild-type AmiR or a catalytically inactive AmiR derivative (AmiR*) in a total volume of 50 µl in buffer H (20 mM HEPES-NaOH, 50 mM NaCl, 1 mM MgCl₂, pH 7.5) at 37 °C for 16 h in a thermal shaker set to 900 rpm. The reactions were stopped by incubation at 100 °C for 10 min in a dry-bed heater. After clearance of the mixtures by centrifugation at 17,000 ×g for 10 min, the supernatants were acidified to pH 4.0 – pH 4.5 with 20% phosphoric acid. Subsequently, the reaction products were separated by HPLC as described previously (Izquierdo-Martinez et al., 2023). The activity of AmiR towards muropeptides with reducing ends was tested in a similar manner. However, peptidoglycan from E. coli D456 was treated with the muramidase cellosyl from Streptomyces coelicolor (Bräu et al., 1991). After incubation with the enzyme, heat inactivation and centrifugation, the supernatants were mixed with an equal volume of 0.5 M sodium borate (pH 9.0), followed by reduction of the muropeptides with ∼1 mg of sodium borohydride. The samples were then acidified to pH 4.0 – pH 4.5 with 20% phosphoric acid and analyzed by HPLC as described previously (Glauner, 1988).

The enzymatic activity of NagZ towards anhydro-muropeptides or reducing muropeptides was analyzed as described for AmiR, but with the following modifications. Peptidoglycan from E. coli CS703-1 (Meberg et al., 2001), which lacks all major DD-carboxypeptidases, was treated with the lytic transglycosylase MltA from E. coli (Lommatzsch et al., 1997) to produce anhydro-muropeptides or with cellosyl to produce muropeptides with reducing ends. The lytic products were incubated with NagZ or a catalytically inactive NagZ variant (NagZ*) (10 µM) and processed as described above.

To test the activity of AmiR towards undigested peptidoglycan, sacculi isolated from E. coli D456 (Glauner, 1988; Takacs et al., 2010) were incubated with AmiR or AmiR* (10 µM) in buffer H at 37 °C for 16 h in a thermal shaker set to 900 rpm. The reactions were stopped by heating at 100 °C for 10 min in a dry-bed heater. Subsequently, the sacculi were digested with cellosyl overnight. After heat inactivation of cellosyl at 100 °C for 10 min and clearance of the mixtures by centrifugation, the reaction products were reduced with borohydride. The samples were then acidified to pH 4.0 – pH 4.5 with 20% phosphoric acid and analyzed by HPLC as described previously (Glauner, 1988).

Disaccharide-containing muropeptides and anhydro-muropeptides were identified based on their known retention times and peak patterns (Glauner, 1988). The identities of the GlcNAc–anhMurNAc disaccharide generated by AmiR and the anhMurNAc-peptides generated by NagZ were determined by HPLC separation of the reaction products and mass spectrometric analysis of the different peak fractions (Figure 2–Figure supplement 4).

Metabolomics analysis

Anhydro-muropeptides were quantified essentially as described previously (Jacobs et al., 1994; Simpson et al., 2023). In brief, cultures were grown overnight in 10 mL PYE medium (OD₆₀₀ = 1.2) and harvested by centrifugation for 10 min at 21,000 ×g and 4 °C. The pelleted cells were washed three times with 0.9% NaCl, resuspended to a final volume of 100 µL in 0.9% NaCl and heated for 7 min at 98 °C. The suspension was cleared by centrifugtation for 15 min at 21,000 ×g. The supernatant was isolated, centrifuged again to remove residual cell debris, and then subjected to liquid chromatography-mass spectrometric (LC-MS) analysis for semi-quantitative determination of PG recycling or biosynthesis intermediates. The chromatographic separation was performed on an Agilent Infinity II 1290 HPLC system using a ZicHILIC SeQuant column (150 × 2.1 mm, 3.5 μm particle size, 100 Å pore size) connected to a ZicHILIC guard column (20 × 2.1 mm, 5 μm particle size) (Merck KgAA) at 25° C and at a constant flow rate of 0.3 mL/min, with mobile phase A being 0.1% formic acid in 99:1 water:acetonitrile (Honeywell, Morristown, NJ, USA) and phase B being 0.1% formic acid in 99:1 acetonitrile:water (Honeywell, Morristown, NJ, USA). The injection volume was 5 µl. The mobile phase profile comprised the following steps and linear gradients: 0 – 1 min constant at 80% B; 1 – 10 min from 80 to 10% B; 10 – 12 min constant at 10% B; 12 – 12.1 min from 10 to 80% B; 12 to 14 min constant at 80% B. An Agilent 6546A QTOF mass spectrometer was used in high-resolution, positive mode with a mass range of 70 – 1700 m/z and a scan rate of 1.5 spectra/second. An electrospray ionization source was used at the following conditions: ESI spray voltage 4000 V, nozzle voltage 500 V, sheath gas 250° C at 12 l/min, nebulizer pressure 60 psig and drying gas 100 °C at 11 l/min. Compounds were identified based on their accurate mass and retention time, applying an extraction tolerance of 5 ppm. Chromatograms were integrated using MassHunter software (Agilent, CA, USA). Relative abundance was determined based on the peak area.

To determine the cytoplasmic amounts of the PG precursors UDP-GlcNAc and UDP-MurNAc, the strains to be analyzed were cultivated overnight in PYE medium, followed by determination of their optical densities (OD₆₀₀). Subsequently, 1-mL samples of the cultures were mixed with 1 mL of 70% LC-MS-Grade MeOH (Merck, Germany) and subjected to centrifugation at 13,000 ×g and -9 °C for 10 min. To extract the endometabolome, the pelleted cells were mixed with 100 µL per OD₆₀₀ unit of extraction fluid (50% LC-MS-Grade MeOH, 5 mM Tris, 0.5 mM EDTA) and the same amount of chloroform, followed by incubation at 4 °C for 2 h. After separation of the phases by centrifugation (max. speed, -9 °C, 10 min), the upper phase was isolated, filtered using PTFE membrane filters (0.2 µm pore size) (Sartorius, Germany) and subjected to LC-MS analysis. The chromatographic separation was performed on an Agilent Infinity II 1290 HPLC system using a SeQuant ZIC-pHILIC column (150 × 2.1 mm, 5 μm particle size, peek coated, Merck) connected to a guard column of similar specificity (20 × 2.1 mm, 5 μm particle size, Phenomoenex) at 40° C and at a constant flow rate of 0.1 mL/min, with mobile phase A composed of 10 mM ammonium acetate in water (pH 9) supplemented with medronic acid to a final concentration of 5 μM and mobile phase B composed of 10 mM ammonium acetate in 90:10 acetonitrile:water (pH 9). The injection volume was 5 µl. The mobile phase profile comprised the following steps and linear gradients: 0 – 1 min constant at 75 % B; 1 – 6 min from 75 to 40% B; 6 to 9 min constant at 40% B; 9 – 9.1 min from 40 to 75% B; 9.1 to 20 min constant at 75% B. An Agilent 6546A QTOF mass spectrometer was used in high-resolution, negative mode with a mass range of 50 – 1700 m/z and a scan rate of 1.5 spectra/second. An electrospray ionization source was used at the following conditions: ESI spray voltage 4000 V, nozzle voltage 500 V, sheath gas 300° C at 12 L/min, nebulizer pressure 20 psig and drying gas 150° C at 11 L/min. Compounds were identified based on their accurate mass and retention time applying an extraction tolerance of 5 ppm. Chromatograms were integrated using MassHunter software (Agilent, Santa Clara, CA, USA). Relative abundance was determined based on the peak area.

Whole-cell proteomics

Cells were grown to exponential phase in 10 mL PYE medium, harvested by centrifugation (15,000 ×g, 10 min, 4 °C), washed three times with ice-cold PBS and resuspended in 300 μL of lysis buffer (2% sodium lauroyl sarcosinate [SLS], 100 mM ammonium bicarbonate). The samples were then heated for 10 min at 90 °C, followed by sonication with a VialTweeter Sonicator (Hielscher Ultrasonics, Germany). Proteins were reduced with 5 mM Tris(2-carboxyethyl)phosphine (Thermo Fischer Scientific, Germany) at 90 °C for 15 min and alkylated using 10 mM iodoacetamid (Sigma Aldrich, Germany) at 20 °C for 30 min in the dark. Subsequently, they were precipitated by the addition of a sixfold excess of ice-cold acetone and incubation for 2 h at -20 °C, followed by two washing steps with methanol. After drying, the proteins were reconstituted in 0.2% SLS and their amount was determined using a bicinchoninic acid (BCA) protein assay (Thermo Scientific, Germany). Subsequently, 50 µg protein were subjected to tryptic digestion by incubation with 1 µg of trypsin (Serva) in 0.5% SLS at 30 °C overnight. After digestion, SLS was precipitated by adding a final concentration of 1.5% trifluoroacetic acid (TFA, Thermo Fisher Scientific, Germany). The peptides were desalted using CHROMAFIX C18 solid phase extraction cartridges (Macherey-Nagel, Germany), which were prepared by the addition of acetonitrile and subsequent equilibration with 0.1% TFA. They were then loaded on the equilibrated cartridges, washed with a mixture of 5% acetonitrile and 0.1% TFA, and finally eluted with a mixture of 50% ACN and 0.1% TFA. After drying, the peptides were reconstituted in 0.1% TFA and then analyzed using liquid-chromatography-mass spectrometry on an Exploris 480 instrument connected to an UltiMate 3000 RSLCnano HPLC system and a Nanospray Flex ion source (all Thermo Scientific, Germany). The following gradient was used for separation: 94% solvent A (0.15% formic acid) and 6% solvent B (99.85% acetonitrile, 0.15% formic acid) to 25% solvent B over 40 min, followed by an increase to 35% solvent B over 20 min at a flow rate of 300 nl/min. Peptides were ionized at a spray voltage of 2.3 kV, with the ion transfer tube temperature set at 275 °C. 445.12003 m/z was used as internal calibrant.

MS raw data were acquired on an Orbitrap Exploris 480 instrument (Thermo Scientific, Germany) in data-independent acquisition (DIA) mode. The funnel RF level was set to 40. For DIA experiments, full MS resolution was set to 120.000 at m/z 200. The AGC target value for fragment spectra was set at 3000%. 45 windows of 14 Da were used with an overlap of 1 Da between m/z 320-950. Resolution was set to 15,000 and IT to 22 ms. A stepped HCD collision energy of 25, 27.5, 30 % was used. MS1 data were acquired in profile, MS2 DIA data in centroid mode.

The analysis of DIA data was performed with DIA-NN version 1.8 (Demichev et al., 2020), based on a protein database for Caulobacter crescentus NA1000 retrieved from UniProt (UniProt Consortium, 2025) to generate a dataset-specific spectral library. The DIA-NN suite was used to perform noise interference correction (mass correction, RT prediction and precursor/fragment co-elution correlation) and peptide precursor signal extraction on the DIA-NN raw data. The following parameters were used: Full tryptic digest was allowed with two missed cleavage sites, oxidized methionines and carbamidomethylated cysteins. “Match between runs” and “Remove likely interferences” were enabled. The precursor FDR was set to 1%. The neural network classifier was set to the single-pass mode, and protein inference was based on genes. The quantification strategy was set to any LC (high accuracy). Cross-run normalization was set to RT-dependent. Library generation was set to smart profiling. DIA-NN outputs were further evaluated using the SafeQuant script (Ahrné et al., 2013; Glatter et al., 2012), modified to process DIA-NN outputs. A detailed summary of the proteomics data obtained is provided in Supplementary File 1.

Bioinformatic analysis

Protein similarity searches were performed with BLAST (Altschul et al., 1990), using the BLAST server of the National Institutes of Health (https://blast.ncbi.nlm.nih.gov/Blast.cgi). Protein structures were predicted with Alphafold 3 (Abramson et al., 2024) and visualized using PyMOL v2.6.0a0 (Schrödinger LLC). Proteomics data were visualized with VolcaNoseR (Goedhart and Luijsterburg, 2020).

Quantification and statistical analysis

Details on the number of replicates and the sample sizes are given in the figure legends. Datasets were displayed using the SuperPlotsOfData web application (Goedhart, 2021), and an unpaired two-sided Welch’s t-test implemented in the application was used to assess the statistical significance of differences between strains or conditions. Standard deviations were calculated in Microsoft Excel 2019.

Availability of biological material

The plasmids and strains used in this study are available from the corresponding author upon request.

Data availability

The proteomics data generated in this study have been deposited to the ProteomeXchange Consortium via the PRIDE partner repository (Perez-Riverol et al., 2025) under the dataset identifier PXD069004. All other data supporting the findings of this study are included in the main text or the supplementary material.

Acknowledgements

We thank Julia Rausch for excellent technical assistance and Daniela Vollmer for the purification of peptidoglycan. This work was supported by the University of Marburg (core funding to M.T.), the Max Planck Society (Max Planck Fellowships to M.T.) and the United Kingdom Biotechnology and Biological Sciences Research Council (grant BB/W013630/1 to W.V.).

Additional information

Author contributions

P.R. and M.T. conceived the study. P.R. generated plasmids and strains, performed the in vivo experiments, purified the proteins and prepared the samples for metabolomics analysis. A.M. generated plasmids and strains. J.B. conducted the enzyme assays and muropeptide analysis. N.P. performed the metabolomics analysis. P.R., J.B., N.P., W.V. and M.T. analyzed the data. V.M. and M.T. secured funding and supervised the study. P.R. and M.T. wrote the manuscript, with input from all other authors.

Funding

Max Planck Society (Max Planck Fellowship)

• Martin Thanbichler

United Kingdom's Biotechnology and Biological Sciences Research Council (BB/W013630/1)

• Waldemar Vollmer

Additional files

Figure 9–video 1. Effect of ampicillin on the growth of C. crescentus wild-type cells. C. crescentus wild-type cells were cultivated in a microfluidic flow cell for 3 h in antibiotic-free medium and then shifted to medium containing 20 µg/mL ampicillin. Phase contrast images were taken at 5-min intervals. Scale bar: 5 µm.

Figure 9–video 2. Effect of ampicillin on the growth of C. crescentus ΔamiR cells. C. crescentus ΔamiR (PR033) cells were cultivated in a microfluidic flow cell for 3 h in antibiotic-free medium and then shifted to medium containing 20 µg/mL ampicillin. Phase contrast images were taken at 5-min intervals. Scale bar: 5 µm.

Supplementary file 1

Supplementary table 1

Supplementary table 2

Supplementary table 3