The peptidoglycan is an essential component of the bacterial cell wall, which provides a mechanical barrier against the turgor pressure of the cytoplasm thereby preventing cells from bursting and lysing (Vollmer and Bertsche, 2008). The peptidoglycan also determines the shape of bacterial cells and is intimately integrated into the cell division process since the barrier to the osmotic pressure needs to be maintained during the entire cell cycle. These functions depend upon the net-like structure of the peptidoglycan macromolecule that is made of glycan strands cross-linked by short peptides (Figure 1A). It is assembled from disaccharide-peptide subunits consisting of β,1→4 linked N-acetylglucosamine (GlcNAc) and N-acetyl muramic acid (MurNAc) and a stem peptide linked to the D-lactoyl group of MurNAc (Figure 1B; Mengin-Lecreulx et al., 1982). Polymerization of the subunit is mediated by glycosyltransferases for the elongation of glycan chains and by transpeptidases for cross-linking stem peptides carried by adjacent glycan chains.

Figure 1 with 3 supplements see all

Download asset Open asset

The last cross-linking step of peptidoglycan polymerization is catalyzed by two types of structurally unrelated transpeptidases (Biarrotte-Sorin et al., 2006), the L,D-transpeptidases (LDTs) (Mainardi et al., 2008) and the D,D-transpeptidases (Zapun et al., 2008), the latter belonging to the penicillin-binding protein (PBP) family. PBPs (Figure 1C) and LDTs (Figure 1D) use different acyl donor substrates (pentapeptide versus tetrapeptide) resulting in the formation of 4→3 versus 3→3 cross-links, respectively (Hugonnet et al., 2016; Mainardi et al., 2005). PBPs are inactivated by all classes of β-lactams, whereas LDTs are inactivated by carbapenems only (Mainardi et al., 2007). In wild-type Escherichia coli, LDTs are fully dispensable for growth, at least in laboratory conditions (Magnet et al., 2008). Accordingly, these enzymes have a minor contribution to peptidoglycan cross-linking in the exponential growth phase (Vollmer and Bertsche, 2008; Glauner, 1988). However, selection of mutants resistant to ceftriaxone, a β-lactam of the cephalosporin class, results in a full bypass of PBPs by the YcbB LDT (Hugonnet et al., 2016). The bypass requires overproduction of both the (p)ppGpp alarmone and YcbB. The peptidoglycan of the mutant grown in the presence of ceftriaxone only contains 3→3 cross-links following PBP inhibition.

D,D-carboxypeptidase (DDC) and L,D-carboxypeptidase activities sequentially remove D-Ala residues at the 5th and 4th positions of stem peptides containing a free carboxyl end, which are present in the acceptor position of dimers and in monomers. DDCs, which belong to the PBP family, hydrolyze the D-Ala⁴-D-Ala⁵ amide bond of stem pentapeptides to generate stem tetrapeptides in mature peptidoglycan (Sauvage et al., 2008). Since the reaction is nearly total in E. coli, stem pentapeptides are found in very low abundance unless DDCs are inactivated by β-lactams. The L,D-carboxypeptidase activity of LDTs hydrolyzes the DAP³-D-Ala⁴ amide bond generating tripeptide stems from tetrapeptide stems (Magnet et al., 2008). Since this reaction is incomplete the peptidoglycan contains combinations of tripeptide and tetrapeptide stems in monomers and in the acyl acceptor position of dimers. Removal of D-Ala residues by DDCs and LDTs occurs before or after cross-linking, leading to the polymorphism in the acceptor position of dimers depicted in Figure 1C and D.

By removing D-Ala⁵ from pentapeptide stems, DDCs eliminate the essential pentapeptide donor of PBPs and generate the essential tetrapeptide donor of LDTs (Mainardi et al., 2002; Hugonnet et al., 2016). By removing D-Ala⁴ from tetrapeptide stems, LDCs eliminate the essential tetrapeptide donor of LDTs thereby potentially limiting the formation of 3→3 cross-links. Thus, DDC and LDC activities control both the extent of peptidoglycan cross-linking and the relative contributions of D,D-transpeptidases and LDTs to peptidoglycan cross-linking.

The interconversion between monomers and cross-linked dimers involves not only PBPs and LDTs in the biosynthetic direction but also hydrolytic enzymes, referred to as endopeptidases, which cleave the cross-links. Endopeptidases belonging to the PBP family are specific of 4→3 cross-links whereas other endopeptidases cleave both 4→3 and 3→3 cross-links (Voedts et al., 2021). Among eight endopeptidases with partially redundant functions, at least one enzyme is essential in the context of the formation of 4→3 cross-links whereas two endopeptidases, MepM and MepK, are required in the context of 3→3 cross-links (Voedts et al., 2021; Singh et al., 2012).

Morphogenesis of E. coli cells depends upon controlled expansion of the peptidoglycan and requires scaffolding proteins, actin (MreB) and tubulin (FtsZ) homologues (den Blaauwen et al., 2008). These proteins are involved in the recruitment of peptidoglycan polymerases and in the coordination of their activities with that of hydrolases (Vollmer and Höltje, 2001). The cell cycle involves two phases, namely the MreB-dependent elongation of the side wall at a constant diameter and the FtsZ-dependent formation of the septum at mid-cell. These processes are thought to involve two distinct multienzyme complexes, the elongasome and the divisome, and specific enzymes, in particular class B PBP2 and PBP3, which mediate peptidoglycan cross-linking in the side wall and in the septum, respectively (Höltje, 1996). Accordingly, specific inactivation of PBP2 by mecillinam or of PBP3 by aztreonam results in the growth of E. coli as spheres or as filaments, respectively.

The average chemical composition of peptidoglycan is well known. Most enzymes involved in the formation and cleavage of glycosidic and amide bonds have been identified (at least one enzyme and generally several enzymes for each bond). However, the spatial arrangement of glycan strands is still a matter of debate. Most models propose a regular arrangement of glycan strands parallel to the cell surface and perpendicular to the long axis of the cell as depicted in Figure 1A; Typas et al., 2011. Alternatively, it has been proposed that the glycan strands protrude perpendicularly to the surface of the cytoplasmic membrane (Meroueh et al., 2006; Dmitriev et al., 2003). The lack of direct experimental evidence originates from the heterogeneity of the peptidoglycan macromolecule that prevents the analysis of the polymer by radiocrystallography. NMR spectroscopy also failed to determine the precise and complete structure of the polymer due to its heterogeneity and its important flexibility, as measured by solid-state NMR relaxation (Kern et al., 2008; Sharif et al., 2009). The mode of insertion of new glycan strands in the growing peptidoglycan is also not fully characterized. According to the ‘three-for-one’ model, peptidoglycan expansion involves the insertion of three neo-synthesized glycan strands to the detriment of the hydrolysis of one existing strand, which acts as a docking strand (Vollmer and Höltje, 2001). Alternatively, glycan strands may be inserted according either to a ‘one-at-a-time’ model or a ‘two-at-a-time’ model, the latter being supported by the dimeric nature of certain D,D-transpeptidases (Charpentier et al., 2002). All these models postulate that peptidoglycan expansion requires cleavage of cross-links by endopeptidases and obey to the ‘make-before-break’ principle, which proposes that cross-links in the existing peptidoglycan are hydrolyzed by endopeptidases only if the newly synthesized glycan strands are cross-linked and can sustain the osmotic pressure of the cytoplasm (Vollmer and Höltje, 2001).

At each generation, about half of the disaccharide-peptide units is released from the peptidoglycan by the combined action of endopeptidases and lytic transglycosylases (Goodell, 1985a; Johnson et al., 2013). The latter enzymes catalyze a non-hydrolytic cleavage of the β,1→4 MurNAc-GlcNAc glycosydic bond and generate GlcNAc-anhydro-MurNAc-peptides fragments via MurNAc cyclisation at positions 1 and 6. These fragments are transported into the cytoplasm by the AmpG permease and recycled according to the pathway depicted in Figure 1—figure supplement 1. The key features of the recycling pathway are that the L-Ala-γ-D-Glu-DAP tripeptide is directly recycled whereas the glucosamine moiety of GlcNAc and MurNAc reenters into the peptidoglycan biosynthesis pathway at the level of glucosamine-6-phosphate. The relationships between peptidoglycan synthesis and recycling are poorly understood. The ‘three-for-one’ growth model predicts that the docking strand is recycled; conversely, the ‘one-at-a-time’ and ‘two-at-a-time’ models do not imply that recycling is a consequence of peptidoglycan synthesis.

The peptidoglycan expansion models described above were previously investigated by pulse labeling of peptidoglycan with radioactive DAP. Indeed, the models imply specific acceptor-to-donor radioactivity ratios (ADRR) in cross-linked stem peptides (Figure 1—figure supplement 2; de Jonge et al., 1989). The ADRR may potentially vary from zero to infinite for (i) exclusive cross-linking of neo-synthesized (radioactive) donor stems to existing (unlabeled) acceptor stems (new→old; ADRR=0), (ii) the formation of cross-links between neo-synthesized (radioactive) donor and acceptor stems (new→new; ADRR=1), and (iii) cross-linking of existing (unlabeled) donor stems to neo-synthesized (labeled) acceptor stems (old→new; infinite ADRR). This method is intrinsically inaccurate since newly synthesized subunits issued from the recycling of stem peptides are mistakenly identified as material from the existing wall. Moreover, the method provides access to the average composition of dimers in labeled donor and acceptor stems rather than the labeling status of stem peptides that have directly participated in the same cross-linking reaction. Thus, an ADRR of 1 can originate either from dimers containing neo-synthesized stem peptides in the donor and acceptor positions (new→new) or, alternatively, from a combination of dimers containing a radioactive DAP either in the donor (new→old) or acceptor (old→new) position in equimolar amounts. This limitation was disregarded in previous studies since dimers containing a single labeled DAP residue in the donor position were considered to be rare, at least in wild-type E. coli grown in the absence of β-lactam. This is arguably the case for dimers generated by PBPs since pentapeptide stems are rapidly converted to tetrapeptide stems by DDCs thereby precluding existing stems from participating as donors in the formation of 4→3 cross-links by D,D-transpeptidases. Obviously, this premise does not apply to LDTs since these enzymes use tetrapeptide stems as donors (Figure 1). Here, we describe a new method based on the full labeling of the entire peptidoglycan macromolecule with stable isotopes of carbon and nitrogen combined with high-resolution mass spectrometry (MS). This method does not suffer from the limitations listed above and our results show that it enables a thorough investigation of peptidoglycan synthesis and recycling in a single experiment. We applied our method to the comparison of the modes of expansion of the peptidoglycan in bacterial strains relying on PBPs, and, for the first time, on LDT YcbB, or on both types of enzymes for peptidoglycan cross-linking. We show how previously described β-lactam-induced malfunctioning of the bacterial cell wall synthesis machinery (Cho et al., 2014) affects peptidoglycan synthesis and recycling. We also report how LDT YcbB rescues inhibition of PBPs by participating in an unprecedented mode of peptidoglycan polymerization. The results described in this report were obtained, thanks to the richness of the structural information brought by MS and MS/MS analyses of all the labeled/unlabeled sugar and amino acid residues of peptidoglycan subunits, as opposed to the radioactive signal located in the DAP moiety only. The mass data analysis pipeline automating the identification of labeled muropeptides is described in this study and may be generally applied to the screening of the mode of action of antibacterial agents acting on peptidoglycan synthesis.

Previous studies of the expansion of the peptidoglycan macromolecule relied on labeling with [³H] or [¹⁴C]DAP (de Jonge et al., 1989). Quantitatively, the number of glycan strands inserted at a time was estimated by determining the acceptor-to-donor radioactivity ratio (ADRR), that is, the radioactivity ratio in the acceptor and donor positions of dimers (Figure 1—figure supplement 2; Burman and Park, 1984). By analogy, we define here the acceptor-to-donor ratio in neo-synthesized stems (ADRNS). ADRR and ADRNS determinations rely on opposite labeling schemes involving pulse labeling with radioactive DAP versus incorporation of newly synthesized (unlabeled) subunits into the fully labeled peptidoglycan, respectively. The latter strategy was chosen because of the ease of obtaining fully labeled cells with non-radioactive heavy isotopes. Switching the medium from labeled to unlabeled was chosen, rather than the unlabeled-to-labeled switch, since this facilitates the detection of neo-synthesized isotopologues present at a low abundance at early times after the medium switch. This is due to the fact that muropeptides from the existing peptidoglycan, which are preponderant at the beginning of the kinetics, are fully labeled and hence have higher m/z values than those of muropeptides containing ¹²C and ¹⁴N light isotopes following the medium switch. Consequently, sodium adducts to the fully labeled muropeptides do not appear in the relevant region of the chromatogram that contains muropeptides of lower m/z value due to the incorporation of light isotopes.

Pulse labeling with radioactive DAP suffers from four major limitations that are not encountered in our labeling strategy. First, optimization of radioactive DAP incorporation requires mutations that prevent endogenous DAP synthesis (lysA) and conversion of DAP to L-Lys (dapF) (Burman and Park, 1984; Glauner and Höltje, 1990; Goodell, 1985a; Goodell and Schwarz, 1985b; Figure 1—figure supplement 1). Starvation prior to the addition of radiolabeled DAP was used to minimize the lag between the addition of radioactive DAP into the culture medium and its incorporation into peptidoglycan (Glauner and Höltje, 1990; Burman and Park, 1984). However, depleting the intracellular DAP pool compromised the integrity of the peptidoglycan layer (Burman et al., 1983b), affected cell morphology (Verwer and Nanninga, 1980), and reduced turnover (Burman et al., 1983b). Our labeling scheme is not affected by these confounding factors since it does not require any metabolic engineering. A second limitation of the DAP labeling procedures is their inability to discriminate between muropeptides originating from the existing peptidoglycan, from the recycled L-Ala-D-iGlu-DAP tripeptide, or from the existing UDP-MurNAc-pentapeptide pool, because all of them contain non-radioactive DAP. In contrast, our procedure easily discriminates between these three types of muropeptides since they correspond to different isotopologues, namely uniformly labeled subunits and hybrids h2 and h3, respectively (Figure 4). The latter hybrids are likely to contribute to the lag phase observed for the incorporation of radioactive DAP into peptidoglycan in previous reports (Glauner and Höltje, 1990; Burman and Park, 1984). Accordingly, a lag phase was observed in our study when h2 and h3 hybrids containing [¹³C]- and [¹⁵N]-labeled DAP were excluded from the dataset (Figure 10). A third limitation of the radioactive DAP labeling procedures for ADRR determination originates from the fact that they provide access only to the average radioactivity content of donor and acceptor stem peptides. Indeed, these procedures are based on purification of individual dimers, derivatization of their free amino group located in the acceptor stem, acid hydrolysis, and chromatography to separate DAP and derivatized DAP originating from the donor and acceptor stems, respectively (Burman et al., 1983a). Our procedure is richer in information because it provides direct access to the structure of dimers enabling identification of new→new, new→old, old→new, and old→old dimers based on differences in mass and fragmentation patterns. The fourth limitation stems from the fact that exploring peptidoglycan recycling by the radioactive DAP labeling procedures requires use of specific mutants and study designs. For example, Jacobs et al., 1994, used mutants deficient in the production of the AmpG and Opp permeases in which peptidoglycan fragments are not recycled in the cytoplasm but released in the culture medium (Jacobs et al., 1994). In contrast, our approach enables investigating peptidoglycan recycling and the mode of peptidoglycan expansion in a single experiment in the absence of any metabolic perturbation since it does not require altering the recycling pathway and is solely based on the determination of the structure of peptidoglycan fragments.

Figure 10

Download asset Open asset

Strain BW25113Δ6ldt offered the possibility to investigate the mode of insertion of new subunits by PBPs in the absence of any LDT. Aztreonam and mecillinam were used to specifically inhibit synthesis of septal and lateral peptidoglycan, respectively. In the presence of aztreonam, the ADRNS for side wall synthesis was 0.14, 5 min after the medium switch and increased to 0.20, 0.34, 0.50, and 0.64 at 20, 40, 60, 80 min, respectively (Figure 11). At this time scale, existing disaccharide-peptide units are gradually replaced by newly synthesized disaccharide-peptides accounting for the increase in new→new dimers and the increase in the ADRNS value. The extrapolation of the ADRNS at t=0 (time of the medium switch) is therefore relevant to the evaluation of the mode of insertion of new subunits into the peptidoglycan. The extrapolated ADRNS value of 0.08 (Figure 8—figure supplement 3) was similar to the ADRR value of 0.10 reported by de Jonge and collaborators based on extrapolating data to a synchronized culture exclusively containing elongating cells (de Jonge et al., 1989). In the presence of mecillinam, the ADRNS values marginally increased over time, providing an extrapolated value of 0.31 for t=0. This value is lower than the ADRR of 0.64 obtained by extrapolating data to a synchronized culture exclusively containing constricting cells (de Jonge et al., 1989). The origin of this difference is unknown but could involve the fact that blocking cell elongation by mecillinam may not fully prevent incorporation of newly synthesized subunits into the peptidoglycan since this drug does not prevent the polymerization of glycan chains by the elongasome (Uehara and Park, 2008). Together, our results are in agreement with previous analyzes (de Jonge et al., 1989) that have shown single-strand insertion of newly synthesized peptidoglycan into the side wall versus insertion of multiple strands for septal peptidoglycan synthesis. In the absence of mecillinam and aztreonam, the ADRNS value (0.14) was intermediate between the values observed in the presence of aztreonam (0.08) and mecillinam (0.31), as qualitatively expected for the combined contributions of cross-links located in the side wall and in the septum and the lower contribution of the latter to the cell surface.

Figure 11

Download asset Open asset

In the absence of β-lactam, E. coli M1.5 relied both on the D,D-transpeptidase activity of PBPs and the LDT activity of YcbB for peptidoglycan cross-linking, leading to a high peptidoglycan content in both 4→3 (56%) and 3→3 (44%) cross-linked dimers (Supplementary file 1a). The proportions of dimers containing two newly synthesized stems (new→new) or one newly synthesized and one existing stem (new→old) were similar for the two types of cross-links (Figure 7). Thus, the D,D-transpeptidase activity of PBPs and the LDT activity of YcbB were involved in similar modes of insertion of new strands into the existing peptidoglycan of strain M1.5. The ADRNS values were in the order of 0.17 and 0.22 for 4 → 3 and 3 → 3 cross-links, respectively. These values indicate that the neo-synthesized glycan strands were predominantly but not exclusively cross-linked to the existing peptidoglycan for both types of transpeptidation reactions. Peptidoglycan polymerization might involve single-strand insertion into the side wall and insertion of multiple strands into the septum, as discussed above for strain Δ6ldt. The striking similarities between dimers generated by PBPs and YcbB raise the possibility that the mode of insertion of new subunits is not determined by the transpeptidases but by other components of the peptidoglycan polymerization complexes, such as the scaffolding proteins.

The peptidoglycan of strain M1.5 grown in the presence of ampicillin was mainly (99%) cross-linked by YcbB following the inhibition of the D,D-transpeptidase activity of PBPs (Figure 9). Strikingly, the Tri→Tetra and Tri→Penta dimers were almost exclusively of the new→new type. This pattern was not found in strains M1.5 and Δ6ldt grown in the absence of β-lactam (Figures 7 and 8). The prevalence of new→new dimers may be a consequence of the inhibition of the transpeptidase activity of PBPs by ampicillin that results in the uncoupling of the cross-linking and transglycosylation reactions (Uehara and Park, 2008; Cho et al., 2014). This in turn results in the accumulation of linear glycan strands that are eventually cross-linked to each other by YcbB because they are not effectively degraded due to a mutation present in M1.5 that prevents production of lytic transglycosylase Slt70 (Voedts et al., 2021; Cho et al., 2014). The resulting new→new cross-linked material was almost exclusively connected to the existing peptidoglycan by the formation of Tri→Tri dimers because new→old isotopologues were only detected in this type of dimers. This implies that the acceptors of the cross-linking reactions in the existing peptidoglycan were tripeptides, whereas existing tetrapeptides also acted as acceptors in the absence of ampicillin (Figure 7). The nearly exclusive use of existing tripeptides as acceptors in the presence of ampicillin could be accounted for by the fact that these tripeptides originate from cleavage of 3→3 cross-links liberating the tripeptide stems present in the donor position of the existing dimers. This model implies a cooperation between YcbB and an endopeptidase that provides the tripeptide acceptor substrate of the reaction by cleaving existing 3→3 cross-links (Figure 12). This enzyme could be MepK since this endopeptidase preferentially cleaves 3→3 cross-links in vitro, is essential for bacterial growth only in strains overproducing YcbB, and is not inhibited by ampicillin (Voedts et al., 2021). According to this model, the reactions catalyzed by MepK and YcbB are concerted rather than sequential. This mode of expansion of the peptidoglycan network differs from the ‘make-before-break’ models that postulate that neo-synthesized glycan strands are cross-linked to free stem peptides of the existing peptidoglycan prior to the cleavage of existing cross-links. In the absence of ampicillin, the hydrolysis of 4→3 cross-links by endopeptidases belonging to the PBP family could generate tetrapeptide acceptors in addition to the tripeptide acceptors exclusively used in the presence of the drug.

Figure 12

Download asset Open asset

Cho et al., 2014 proposed that the antibacterial activity of β-lactams does not merely rely on a loss of function, that is, the inactivation of the D,D-transpeptidase activity of PBPs by acylation of their catalytic Ser residue, but by inducing a toxic malfunctioning of the peptidoglycan biosynthesis machinery. This involves uncoupling of transglycosylation and transpeptidation, accumulation of polymerized glycan chains, and their degradation by lytic transglycosylase Slt70 (Banzhaf et al., 2012; Cho et al., 2014; Uehara and Park, 2008). The resulting futile cycle depletes resources and prevents bacterial growth. Exposure of ΔsltY cells to mecillinam was reported to result in the polymerization of an ‘aberrant peptidoglycan’ enriched in 3→3 cross-links. In comparison to this previous report (Cho et al., 2014), our analysis revealed that overproduction of YcbB results in substantially different modifications of peptidoglycan metabolism. First, M1.5 grown in the absence of β-lactam released a high proportion of its peptidoglycan per generation (56%) (Figure 13). This indicates that formation of 3→3 cross-links by YcbB is sufficient to stimulate peptidoglycan degradation in the absence of β-lactams. Under these conditions, Slt70 was not the main lytic enzyme since this activity is impaired in M1.5 (Voedts et al., 2021). Exposure to β-lactam had an additional stimulating effect on peptidoglycan degradation that reached 78% per generation for M1.5 grown in the presence of ampicillin (Figure 13). Second, overproduction of YcbB enabled growth of M1.5 in the presence of β-lactams indicating that the ‘aberrant peptidoglycan’ described by Cho et al., 2014, can in fact sustain bacterial growth. This involved an unprecedented mode of peptidoglycan synthesis implicating polymerization of linear glycan chains, their cross-linking by YcbB, and the incorporation of the resulting neo-synthesized material into the existing peptidoglycan by the concerted action of LDT YcbB and of β-lactam-insensitive endopeptidase MepK.

Figure 13

Download asset Open asset

MS/MS spectra have been provided in Supplementary data file. The software developments required to predict and analyze the labeled/unlabeled muropeptide ions isotopic clusters either in MS or MS/MS experiments are hosted at https://gitlab.com/kantundpeterpan/masseltof, (copy archived at swh:1:rev:c45a7e66a8a89c436d41b8431938b35acfa4e53b) and published under a Free Software license.

1. 1. Adusumilli R
   2. Mallick P

   (2017) Data Conversion with ProteoWizard msConvert

   Methods in Molecular Biology (Clifton, N.J.) 1550:339–368.

   https://doi.org/10.1007/978-1-4939-6747-6_23

   • PubMed
   • Google Scholar
2. 1. Arbeloa A
   2. Hugonnet J-E
   3. Sentilhes A-C
   4. Josseaume N
   5. Dubost L
   6. Monsempes C
   7. Blanot D
   8. Brouard J-P
   9. Arthur M

   (2004) Synthesis of mosaic peptidoglycan cross-bridges by hybrid peptidoglycan assembly pathways in gram-positive bacteria

   The Journal of Biological Chemistry 279:41546–41556.

   https://doi.org/10.1074/jbc.M407149200

   • PubMed
   • Google Scholar
3. Software

   1. Atze H

   (2021) kantundpeterpan/masseltof, version swh:1:rev:c45a7e66a8a89c436d41b8431938b35acfa4e53b

   GitHub.

   https://gitlab.com/kantundpeterpan/masseltof
4. 1. Banzhaf M
   2. van den Berg van Saparoea B
   3. Terrak M
   4. Fraipont C
   5. Egan A
   6. Philippe J
   7. Zapun A
   8. Breukink E
   9. Nguyen-Distèche M
   10. den Blaauwen T
   11. Vollmer W

   (2012) Cooperativity of peptidoglycan synthases active in bacterial cell elongation

   Molecular Microbiology 85:179–194.

   https://doi.org/10.1111/j.1365-2958.2012.08103.x

   • PubMed
   • Google Scholar
5. 1. Biarrotte-Sorin S
   2. Hugonnet J-E
   3. Delfosse V
   4. Mainardi J-L
   5. Gutmann L
   6. Arthur M
   7. Mayer C

   (2006) Crystal Structure of a Novel β-Lactam-insensitive Peptidoglycan Transpeptidase

   Journal of Molecular Biology 359:533–538.

   https://doi.org/10.1016/j.jmb.2006.03.014

   • PubMed
   • Google Scholar
6. 1. Burman LG
   2. Raichler J
   3. Park JT

   (1983a) Evidence for diffuse growth of the cylindrical portion of the Escherichia coli murein sacculus

   Journal of Bacteriology 155:983–988.

   https://doi.org/10.1128/jb.155.3.983-988.1983

   • PubMed
   • Google Scholar
7. 1. Burman LG
   2. Reichler J
   3. Park JT

   (1983b) Evidence for multisite growth of Escherichia coli murein involving concomitant endopeptidase and transpeptidase activities

   Journal of Bacteriology 156:386–392.

   https://doi.org/10.1128/jb.156.1.386-392.1983

   • PubMed
   • Google Scholar
8. 1. Burman LG
   2. Park JT

   (1984) Molecular model for elongation of the murein sacculus of Escherichia coli

   PNAS 81:1844–1848.

   https://doi.org/10.1073/pnas.81.6.1844

   • PubMed
   • Google Scholar
9. 1. Charpentier X
   2. Chalut C
   3. Rémy M-H
   4. Masson J-M

   (2002) Penicillin-binding proteins 1a and 1b form independent dimers in Escherichia coli

   Journal of Bacteriology 184:3749–3752.

   https://doi.org/10.1128/JB.184.13.3749-3752.2002

   • PubMed
   • Google Scholar
10. 1. Cho H
    2. Uehara T
    3. Bernhardt TG

    (2014) Beta-lactam antibiotics induce a lethal malfunctioning of the bacterial cell wall synthesis machinery

    Cell 159:1300–1311.

    https://doi.org/10.1016/j.cell.2014.11.017

    • PubMed
    • Google Scholar
11. 1. Datsenko KA
    2. Wanner BL

    (2000) One-step inactivation of chromosomal genes in Escherichia coli K-12 using PCR products

    PNAS 97:6640–6645.

    https://doi.org/10.1073/pnas.120163297

    • PubMed
    • Google Scholar
12. 1. de Jonge BL
    2. Wientjes FB
    3. Jurida I
    4. Driehuis F
    5. Wouters JT
    6. Nanninga N

    (1989) Peptidoglycan synthesis during the cell cycle of Escherichia coli: composition and mode of insertion

    Journal of Bacteriology 171:5783–5794.

    https://doi.org/10.1128/jb.171.11.5783-5794.1989

    • PubMed
    • Google Scholar
13. 1. den Blaauwen T
    2. de Pedro MA
    3. Nguyen-Distèche M
    4. Ayala JA

    (2008) Morphogenesis of rod-shaped sacculi

    FEMS Microbiology Reviews 32:321–344.

    https://doi.org/10.1111/j.1574-6976.2007.00090.x

    • PubMed
    • Google Scholar
14. 1. Dmitriev BA
    2. Toukach FV
    3. Schaper K-J
    4. Holst O
    5. Rietschel ET
    6. Ehlers S

    (2003) Tertiary structure of bacterial murein: the scaffold model

    Journal of Bacteriology 185:3458–3468.

    https://doi.org/10.1128/JB.185.11.3458-3468.2003

    • PubMed
    • Google Scholar
15. 1. Glauner B

    (1988) Separation and quantification of muropeptides with high-performance liquid chromatography

    Analytical Biochemistry 172:451–464.

    https://doi.org/10.1016/0003-2697(88)90468-x

    • PubMed
    • Google Scholar
16. 1. Glauner B
    2. Höltje JV

    (1990) Growth pattern of the murein sacculus of Escherichia coli

    The Journal of Biological Chemistry 265:18988–18996.

    https://doi.org/10.1016/S0021-9258(17)30613-0

    • PubMed
    • Google Scholar
17. 1. Goodell EW

    (1985a) Recycling of murein by Escherichia coli

    Journal of Bacteriology 163:305–310.

    https://doi.org/10.1128/jb.163.1.305-310.1985

    • PubMed
    • Google Scholar
18. 1. Goodell EW
    2. Schwarz U

    (1985b) Release of cell wall peptides into culture medium by exponentially growing Escherichia coli

    Journal of Bacteriology 162:391–397.

    https://doi.org/10.1128/jb.162.1.391-397.1985

    • PubMed
    • Google Scholar
19. 1. Höltje JV

    (1996) A hypothetical holoenzyme involved in the replication of the murein sacculus of Escherichia coli

    Microbiology (Reading, England) 142 (Pt 8):1911–1918.

    https://doi.org/10.1099/13500872-142-8-1911

    • PubMed
    • Google Scholar
20. 1. Hugonnet J-E
    2. Mengin-Lecreulx D
    3. Monton A
    4. den Blaauwen T
    5. Carbonnelle E
    6. Veckerlé C
    7. Brun YV
    8. van Nieuwenhze M
    9. Bouchier C
    10. Tu K
    11. Rice LB
    12. Arthur M

    (2016) Factors essential for L,D-transpeptidase-mediated peptidoglycan cross-linking and β-lactam resistance in Escherichia coli

    eLife 5:e19469.

    https://doi.org/10.7554/eLife.19469

    • PubMed
    • Google Scholar
21. 1. Jacobs C
    2. Huang LJ
    3. Bartowsky E
    4. Normark S
    5. Park JT

    (1994) Bacterial cell wall recycling provides cytosolic muropeptides as effectors for beta-lactamase induction

    The EMBO Journal 13:4684–4694.

    https://doi.org/10.1002/j.1460-2075.1994.tb06792.x

    • PubMed
    • Google Scholar
22. 1. Johnson JW
    2. Fisher JF
    3. Mobashery S

    (2013) Bacterial cell-wall recycling

    Annals of the New York Academy of Sciences 1277:54–75.

    https://doi.org/10.1111/j.1749-6632.2012.06813.x

    • PubMed
    • Google Scholar
23. 1. Kern T
    2. Hediger S
    3. Müller P
    4. Giustini C
    5. Joris B
    6. Bougault C
    7. Vollmer W
    8. Simorre J-P

    (2008) Toward the Characterization of Peptidoglycan Structure and Protein−Peptidoglycan Interactions by Solid-State NMR Spectroscopy

    Journal of the American Chemical Society 130:5618–5619.

    https://doi.org/10.1021/ja7108135

    • PubMed
    • Google Scholar
24. 1. Langella O
    2. Rusconi F

    (2021) mineXpert2: Full-Depth Visualization and Exploration of MS(n) Mass Spectrometry Data

    Journal of the American Society for Mass Spectrometry 32:1138–1141.

    https://doi.org/10.1021/jasms.0c00402

    • PubMed
    • Google Scholar
25. 1. Magnet S
    2. Bellais S
    3. Dubost L
    4. Fourgeaud M
    5. Mainardi J-L
    6. Petit-Frère S
    7. Marie A
    8. Mengin-Lecreulx D
    9. Arthur M
    10. Gutmann L

    (2007) Identification of the L,D-transpeptidases responsible for attachment of the Braun lipoprotein to Escherichia coli peptidoglycan

    Journal of Bacteriology 189:3927–3931.

    https://doi.org/10.1128/JB.00084-07

    • PubMed
    • Google Scholar
26. 1. Magnet S
    2. Dubost L
    3. Marie A
    4. Arthur M
    5. Gutmann L

    (2008) Identification of the L,D-transpeptidases for peptidoglycan cross-linking in Escherichia coli

    Journal of Bacteriology 190:4782–4785.

    https://doi.org/10.1128/JB.00025-08

    • PubMed
    • Google Scholar
27. 1. Mainardi J-L
    2. Morel V
    3. Fourgeaud M
    4. Cremniter J
    5. Blanot D
    6. Legrand R
    7. Frehel C
    8. Arthur M
    9. Van Heijenoort J
    10. Gutmann L

    (2002) Balance between two transpeptidation mechanisms determines the expression of beta-lactam resistance in Enterococcus faecium

    The Journal of Biological Chemistry 277:35801–35807.

    https://doi.org/10.1074/jbc.M204319200

    • PubMed
    • Google Scholar
28. 1. Mainardi J-L
    2. Fourgeaud M
    3. Hugonnet J-E
    4. Dubost L
    5. Brouard J-P
    6. Ouazzani J
    7. Rice LB
    8. Gutmann L
    9. Arthur M

    (2005) A novel peptidoglycan cross-linking enzyme for A beta-lactam-resistant transpeptidation pathway

    The Journal of Biological Chemistry 280:38146–38152.

    https://doi.org/10.1074/jbc.M507384200

    • PubMed
    • Google Scholar
29. 1. Mainardi J-L
    2. Hugonnet J-E
    3. Rusconi F
    4. Fourgeaud M
    5. Dubost L
    6. Moumi AN
    7. Delfosse V
    8. Mayer C
    9. Gutmann L
    10. Rice LB
    11. Arthur M

    (2007) Unexpected inhibition of peptidoglycan LD-transpeptidase from Enterococcus faecium by the beta-lactam imipenem

    The Journal of Biological Chemistry 282:30414–30422.

    https://doi.org/10.1074/jbc.M704286200

    • PubMed
    • Google Scholar
30. 1. Mainardi J-L
    2. Villet R
    3. Bugg TD
    4. Mayer C
    5. Arthur M

    (2008) Evolution of peptidoglycan biosynthesis under the selective pressure of antibiotics in Gram-positive bacteria

    FEMS Microbiology Reviews 32:386–408.

    https://doi.org/10.1111/j.1574-6976.2007.00097.x

    • PubMed
    • Google Scholar
31. 1. Mengin-Lecreulx D
    2. Flouret B
    3. van Heijenoort J

    (1982) Cytoplasmic steps of peptidoglycan synthesis in Escherichia coli

    Journal of Bacteriology 151:1109–1117.

    https://doi.org/10.1128/jb.151.3.1109-1117.1982

    • PubMed
    • Google Scholar
32. 1. Mengin-Lecreulx D
    2. van Heijenoort J
    3. Park JT

    (1996) Identification of the mpl gene encoding UDP-N-acetylmuramate: L-alanyl-gamma-D-glutamyl-meso-diaminopimelate ligase in Escherichia coli and its role in recycling of cell wall peptidoglycan

    Journal of Bacteriology 178:5347–5352.

    https://doi.org/10.1128/jb.178.18.5347-5352.1996

    • PubMed
    • Google Scholar
33. 1. Meroueh SO
    2. Bencze KZ
    3. Hesek D
    4. Lee M
    5. Fisher JF
    6. Stemmler TL
    7. Mobashery S

    (2006) Three-dimensional structure of the bacterial cell wall peptidoglycan

    PNAS 103:4404–4409.

    https://doi.org/10.1073/pnas.0510182103

    • PubMed
    • Google Scholar
34. 1. Montón Silva A
    2. Otten C
    3. Biboy J
    4. Breukink E
    5. VanNieuwenhze M
    6. Vollmer W
    7. den Blaauwen T

    (2018) The Fluorescent D-Amino Acid NADA as a Tool to Study the Conditional Activity of Transpeptidases in Escherichia coli

    Frontiers in Microbiology 9:2101.

    https://doi.org/10.3389/fmicb.2018.02101

    • PubMed
    • Google Scholar
35. 1. Morè N
    2. Martorana AM
    3. Biboy J
    4. Otten C
    5. Winkle M
    6. Serrano CKG
    7. Montón Silva A
    8. Atkinson L
    9. Yau H
    10. Breukink E
    11. den Blaauwen T
    12. Vollmer W
    13. Polissi A

    (2019) Peptidoglycan Remodeling Enables Escherichia coli To Survive Severe Outer Membrane Assembly Defect

    MBio 10:e02729-18.

    https://doi.org/10.1128/mBio.02729-18

    • PubMed
    • Google Scholar
36. 1. Sauvage E
    2. Kerff F
    3. Terrak M
    4. Ayala JA
    5. Charlier P

    (2008) The penicillin-binding proteins: structure and role in peptidoglycan biosynthesis

    FEMS Microbiology Reviews 32:234–258.

    https://doi.org/10.1111/j.1574-6976.2008.00105.x

    • PubMed
    • Google Scholar
37. 1. Sharif S
    2. Singh M
    3. Kim SJ
    4. Schaefer J

    (2009) Staphylococcus aureus peptidoglycan tertiary structure from carbon-13 spin diffusion

    Journal of the American Chemical Society 131:7023–7030.

    https://doi.org/10.1021/ja808971c

    • PubMed
    • Google Scholar
38. 1. Singh SK
    2. SaiSree L
    3. Amrutha RN
    4. Reddy M

    (2012) Three redundant murein endopeptidases catalyse an essential cleavage step in peptidoglycan synthesis of Escherichia coli K12

    Molecular Microbiology 86:1036–1051.

    https://doi.org/10.1111/mmi.12058

    • PubMed
    • Google Scholar
39. 1. Templin MF
    2. Ursinus A
    3. Höltje JV

    (1999) A defect in cell wall recycling triggers autolysis during the stationary growth phase of Escherichia coli

    The EMBO Journal 18:4108–4117.

    https://doi.org/10.1093/emboj/18.15.4108

    • PubMed
    • Google Scholar
40. 1. Typas A
    2. Banzhaf M
    3. Gross CA
    4. Vollmer W

    (2011) From the regulation of peptidoglycan synthesis to bacterial growth and morphology

    Nature Reviews. Microbiology 10:123–136.

    https://doi.org/10.1038/nrmicro2677

    • PubMed
    • Google Scholar
41. 1. Uehara T
    2. Park JT

    (2008) Growth of Escherichia coli: significance of peptidoglycan degradation during elongation and septation

    Journal of Bacteriology 190:3914–3922.

    https://doi.org/10.1128/JB.00207-08

    • PubMed
    • Google Scholar
42. 1. Verwer RW
    2. Nanninga N

    (1980) Pattern of meso-DL-2,6-diaminopimelic acid incorporation during the division cycle of Escherichia coli

    Journal of Bacteriology 144:327–336.

    https://doi.org/10.1128/jb.144.1.327-336.1980

    • PubMed
    • Google Scholar
43. 1. Virtanen P
    2. Gommers R
    3. Oliphant TE
    4. Haberland M
    5. Reddy T
    6. Cournapeau D
    7. Burovski E
    8. Peterson P
    9. Weckesser W
    10. Bright J
    11. van der Walt SJ
    12. Brett M
    13. Wilson J
    14. Millman KJ
    15. Mayorov N
    16. Nelson ARJ
    17. Jones E
    18. Kern R
    19. Larson E
    20. Carey CJ
    21. Polat İ
    22. Feng Y
    23. Moore EW
    24. VanderPlas J
    25. Laxalde D
    26. Perktold J
    27. Cimrman R
    28. Henriksen I
    29. Quintero EA
    30. Harris CR
    31. Archibald AM
    32. Ribeiro AH
    33. Pedregosa F
    34. van Mulbregt P
    35. SciPy 1.0 Contributors

    (2020) SciPy 1.0: fundamental algorithms for scientific computing in Python

    Nature Methods 17:261–272.

    https://doi.org/10.1038/s41592-019-0686-2

    • PubMed
    • Google Scholar
44. 1. Voedts H
    2. Dorchêne D
    3. Lodge A
    4. Vollmer W
    5. Arthur M
    6. Hugonnet J-E

    (2021) Role of endopeptidases in peptidoglycan synthesis mediated by alternative cross-linking enzymes in Escherichia coli

    The EMBO Journal 40:e108126.

    https://doi.org/10.15252/embj.2021108126

    • PubMed
    • Google Scholar
45. 1. Vollmer W
    2. Höltje JV

    (2001) Morphogenesis of Escherichia coli

    Current Opinion in Microbiology 4:625–633.

    https://doi.org/10.1016/s1369-5274(01)00261-2

    • PubMed
    • Google Scholar
46. 1. Vollmer W
    2. Bertsche U

    (2008) Murein (peptidoglycan) structure, architecture and biosynthesis in Escherichia coli

    Biochimica et Biophysica Acta 1778:1714–1734.

    https://doi.org/10.1016/j.bbamem.2007.06.007

    • PubMed
    • Google Scholar
47. 1. Zapun A
    2. Contreras-Martel C
    3. Vernet T

    (2008) Penicillin-binding proteins and beta-lactam resistance

    FEMS Microbiology Reviews 32:361–385.

    https://doi.org/10.1111/j.1574-6976.2007.00095.x

    • PubMed
    • Google Scholar

Decision letter after peer review:

Thank you for submitting your article "Heavy isotope labeling and mass spectrometry reveal unexpected remodeling of bacterial cell wall expansion in response to drugs" for consideration by eLife. Your article has been reviewed by 3 peer reviewers, and the evaluation has been overseen by Bavesh Kana as the Senior and Reviewing Editor. The following individual involved in review of your submission has agreed to reveal their identity: Tanneke den Blaauwen (Reviewer #3).

The reviewers have discussed their reviews with one another, and the Reviewing Editor has drafted this to help you prepare a revised submission.

Essential revisions:

1. While the data on the one-by-one strand insertion in the lateral wall is supported by all the presented data, there seems to be a conflict in the data presented for the insertion of new peptidoglycan strands during cell division. The calculation of the ratio of new vs old peptidoglycan presented in Supplementary figure 12 for bacteria treated with mecillinam (that grow as cocci and perform only cell division) indeed suggests that multiple peptidoglycan strands are inserted from the beginning of the chase. However, this data is based on abundance of labeled, hybrid and unlabeled peptidoglycan fragments presumably presented in Figure 8. The data on panel D, left graph does not show any immediate presence in the peptidoglycan of unlabeled peptidoglycan fragments. In fact, these are as low in abundance as in none treated or aztreonam treated cells (that elongate but do not divide). These two figures are hard to reconcile.

2. As indicated in several figure legends, the data presented is from two independent experiments. Does this mean that the authors only performed twice the chase experiments? The experiment itself is rather simple to perform and it's the downstream analysis that is extremely time consuming. How confident are the authors of the reproducibility of the data?

3. The cells are grown in mecillinam and aztreonam, which inhibit length growth and division, respectively. The cells are expected to have a round and filamentous morphology, respectively, but this is not shown in the manuscript, where a very "non wild-type" strain is used for the experiments. The concentration of the antibiotics is perhaps based on the literature, but maybe this strain has difference susceptibility to the antibiotics? Using the wild type strain in these experiments would be important to demonstrate relevance of the findings.

4. In the absence of LD-TPases the insertion of new PG is predominantly new-old for the first 20 minutes, whereas in the M1.5 strain, this is 50% new-old and 50% new-new. Their mass doubling times are 67 and 90 min, respectively. Both strains should contain dividing cells directly after the medium change. If septum formation is multistrand insertion, one would expect new-new in both strains. Then aztreonam is added to the δ 6 LD-TPases strain and the mode of insertion remains the same (new-old), suggesting that length growth is single strand insertion. However, this cannot be concluded because with division it was also single stranded. Then in the presence of mecillinam the authors claim that they see new-old and new-new insertion and therefore, claim that division requires multiple strand insertion. However, in the graph of the mecillinam cells, (1) it looks as if the PG synthesis is considerably slowed down in comparison to the other two situations. This is unexpected given the that the mass doubling time did not change according to Table S2. (2) If one extrapolates the lines in the graph, they are not that different from the strains without antibiotics or with aztreonam. (3) The new-old dominates, which one does not expect from cells that enlarge predominantly through division. The explanation that the futile cycle of PG TGase activity may result in insertion of old-new strand by PBP1b and 1a is plausible, but it does not prove that septation is multistrand. Would it be possible, to calculate the amount of surface generated through septation and elongation per min and predict the expected increase in old-new and new-new in the M1.5 strain to see if the observed data would fit the proposed model?

5. In general, it is a pity that no wild-type strain is used for the new method. The conclusion of the paper is very important, i. e. new PG is inserted strand by strand likely by the combined action of an TPase and an endopeptidase. The absence of data on a WT strain diminishes the soundness of the conclusion. Since it can be expected that this paper will be cited for a long time by many people, it is important that it is complete. Experiments with the wild type should be reported.

6. The discrimination of labelled versus unlabelled assumes that if one isotope is present, the molecule is considered to be labelled and if no isotopes are present the molecule is unlabelled. A situation must exist (as is outlined in Figure S14 and how the mass spectra data are poled in S15) in which for instance glucosamine is made from a mixture of labelled and unlabelled 13C and existing labelled DAP (which was also reported as having a considerable cellular pool) is coupled to an newly synthesized (unlabeled) D-Ala-D-Ala. This would result in a high number of different MS peaks which all present the same structure. Assuming that the relative abundance of these variable labelled molecules will be the same for all muropeptides, but H1 and H2, is 5 min enough to cause every newly synthesized muropeptide to be fully unlabelled during the chase or is there a mixture for some time and how longs do these mixtures last? Maybe this gives some information on the presence of the pools of the various parts of the muropeptides?

7. The GlcNAc-anhydro-MurNAc- peptides must have a discrete mass from the standard muropeptides, but are not identified as such. Why not? It gives information in the length of the glycan strands and as such is interesting.

8. Page 20: "Thus, the absence of old-new 3-3 cross-linked Tri-Tetra dimer indicates that YcbB discriminates neo-synthesized tetrapeptide stems (used as donors) from existing tetrapeptide stems (used as acceptors)" Could it be that these crosslinks were below the detection limit? It is important because If YcbB is indeed only able to make 3-3 crosslinks from newly synthesized PG, this implies that repair is only possible if a TGase polymerizes PG. It has been shown by Morè et al., (Morè, N., Martorana, A.M., Biboy, J., Otten, C., Winkle, M., Serrano, C.K.G., et al. (2019) Peptidoglycan Remodeling Enables Escherichia coli To Survive Severe Outer Membrane Assembly Defect. mBio 10.) that YcbB works together with PBP1B and PBP6a and the absence of old-new may explain and support this further.

Other corrections:

1. In the introduction, the authors mention "lytic glycosyltransferases". The term "lytic transglycosylase" seems more appropriate.

2. Ln 62: a comma is missing after "in E. coli".

3. Ln 104: Change "NMR" to "Solution-state NMR" because solid-state NMR, although limited in its applications, has been successful in characterizing the tertiary structure of peptidoglycan in bacteria. For example, "Staphylococcus aureus peptidoglycan tertiary structure from carbon-13 spin diffusion" (JACS, 2009).

4. The Figure S2. ADRR calculation assumes the PG model which consists primarily of PG dimers with 50% crosslinking efficiency. The literature is in good agreement with an average PG crosslinking efficiency in E. coli observed at 50% (ranging from 30~60% depending on the strains),1 but in addition to a significant number of monomers (30 to 55%) and dimers (41 to 51%), there are trimers (4 to 10%) and tetramers (less than 1%) that are found in E. coli.2 These oligomeric muropeptide species are low in their abundance, however, a statement on how these oligomers are consistent with the insertion models would be important.

5. Ln 211: insert ", Peak A" after "The second peak of the isotopic cluster".

6. Ln 224: Change "cluster D" to "Peak D".

7. Ln 227: Change "clusters" to "Peaks".

8. Ln 231: "(99% for each isotope)" mentioned in the text is for the isotopic enrichments for the heavy-labeled glucose and ammonium chloride. However, the isotope incorporation efficiency in E. coli seems to be less than 99% because of the observed spectrum of Figure 2D which shows the larger peak intensities for Peaks A, B, and C compared to the corresponding ratio shown in the simulated spectrum (Figure 2E). Since isotopic distribution is highly sensitive to dilution, perhaps 25 ml of preculture used to inoculate 1L of labeled media was grown in unlabeled M9. If so, then the dilution is approximately 25ml/(100ml + 25ml) = 2.4% and thus the expected isotopic incorporation efficiency is 96.5% (97.5% * 99%) which may account for the differences in the observed and calculated Peaks for A, B, and C? This is a minor point since it does not change the overall analysis.

9. Ln 251, Ln 264: "Table 2" is missing from the manuscript.

10. Ln 308: insert "heavy-isotope labeled" before "minimal medium".

11. Ln 467 "Tri-Tri and Tetra-Tri dimers formed by PBPs and YcbB, respectively" should be changed to "Tri-Tri and Tetra-Tri dimers formed by YcbB and PBPs, respectively".

12. Ln 468: "were almost exclusively of the new-old type (Figure 7)" But Figure 7 A (Tri-Tri) and B (Tetra-Tri) show an even mixture of both new-old and new-new. Hence, the statement "exclusively of new-old" is not entirely correct.

13. Ln 468-469: "In contrast, Tri-Tetra and Tetra-Tetra dimers formed by these enzymes were a mixture of new-new and new-old" but the figure shows only new-new.

14. Ln 475: The statement "Thus, the mode of insertion of PG subunits depended upon the structure of the acceptor stem … rather than upon the type of transpeptidase" is based on the observation from Figure 7 which shows that panels A and B (Tri as acceptor), and panels C and D (Tetra as acceptor) showed similar kinetic behaviors. The addition of a few sentences for an explanation would add clarity to the statement.

15. Figure 7 provides insight into the flux of how much new (unlabeled) PG is incorporated. For example, in Figure 7 C (Tri-Tetra 3-3 CL) at 85 mins the observed N-N is at ~80%, O-O at ~10%, and N-O at ~10%, then the relative total amount of Old PG subunit (2*10% + 10% = 30%) to the New PG subunit (2*80% + 10% = 170%) for Tri-Tetra of 3-3 CL is 3:17. The approximate ratio of old to new for Tetra-Tetra of 4-3 is approximately 1:19. This indicates that the newly formed PG subunits (nascent PG) are incorporated into both 4-3 and 3-3 CL with a tetrapeptide stem, although more actively to 4-3 CL. In contrast, the approximate ratio of old to new for Tri-Tri (3-3 CL) and Tetra-Tri (4-3) is 9:17 and 1:2, respectively. This indicates that the newly formed PG subunit in the form of a tripeptide acceptor is not readily incorporated into the cell wall. Hence, the insertion into PG shows dependence on the PG stem structure of the acceptor. Alternatively, the concentrations of Tri and Tetra are not independent but coupled, most likely high concentration of Tetrapeptide PG stems (both monomers and dimers) as shown Figure 6 (b-f) is likely to be the determinant for the observed kinetics.

16. Ln 519: Please explain the amounts of antibiotics that were added, Mec (2.5 ug/ml) and AZT (12 ug/ml), to E. coli before the medium switch.

17. Ln 557 Figure 9A: "… grown in the presence of ampicillin", at what time point was the sample collected?

18. Ln 546-549: "New-new rather than new-old isotopologues were predominantly detected among dimers……. In contrast, the Tri-Tetra dimers in the PG of M1.5 grown in the absence of ampicillin mainly contained the new-old isotopologues (Figure 7)". Figures below show kinetics for Tri-Tetra dimers from Figure 9 (left) and Figure 7C (right). Contrary to the statement, PG of M1.5 grown in the absence of ampicillin (Figure 7C, right) contained a significant amount of new-new isotopologues.

19. For Figures 5, 7, 8, and 9, please provide which program was used to fit the data. Also, what was the kinetic model used for the fit? Does the model assume sequential event of A to B to C (conversion from old-old to new-old and then to new-new)? An explanation for a buildup of new-old in Tri-Tri but not in Tri-Tetra in E. coli M1.5 grown in the presence of ampicillin (as shown in Figure 9C below) would be helpful.

20. Ln 577-579: The rationale for switching the medium from labeled to unlabeled was that it "facilitates the detection of neo-synthesized isotopologues present at a low abundance at early times after the medium switch." It was not clear how this facilitates the detection of neo-synthesized isotopologues compared to switching from unlabeled to label medium. Additional sentences explaining this would be of interest to the readers.

21. Ln 610: "old-new" should be changed to "old-old".

22. Ln 671-673: Confusing statement. Ln 672: "new-old" should be "new-new"?

23. Figure S12 is of great significance and thus recommend moving this as one of the main figures in the text.

24. Page 20 "Kinetic analyses revealed that the Tri-Tri and Tetra-Tri dimers formed by PBPs and YcbB, respectively, were almost exclusively of the new-old type (Figure 7). Should these be YcbB and PBPs, respectively.

25. In Figure 6, S7, S8 and S9 strain M1.4 is used, but this strain is not described in the method section.

26. The E. coli BW25113 rph mutation is a frame shift that leads to pyrimidine starvation. To have an optimal DNA replication cycle in M9 medium it needs some supplements, which were not added according to the Materials and methods. This may not affect the PG mode of insertion but perhaps the authors can keep this in mind for future experiments? All strains used in this study are derivatives of this strain. E. coli BW25113 [Δ(araD-araB)567 Δ(rhaD-rhaB)568 ΔlacZ4787 (::rrnB-3) hsdR514 rph-1, lacI+] BW25113 Needs uracil 20 ug/ml, when grown in minimal medium. KF Jensen J Bacteriol. 1993 Jun;175(11):3401-7. Grenier, F., Matteau, D., Baby, V., & Rodrigue, S. (2014). Complete Genome Sequence of Escherichia coli BW25113. Genome Announcements, 2(5). doi:10.1128/genomeA.01038-14. Needs also thymidine, 2 ug/ml according to this paper. J Bacteriol. 1998 Jun;180(11):2992-4. Determining the optimal thymidine concentration for growing Thy- Escherichia coli strains. Molina F1, Jiménez-Sánchez A, Guzmán EC.

Essential revisions:

1. While the data on the one-by-one strand insertion in the lateral wall is supported by all the presented data, there seems to be a conflict in the data presented for the insertion of new peptidoglycan strands during cell division. The calculation of the ratio of new vs old peptidoglycan presented in Supplementary figure 12 for bacteria treated with mecillinam (that grow as cocci and perform only cell division) indeed suggests that multiple peptidoglycan strands are inserted from the beginning of the chase. However, this data is based on abundance of labeled, hybrid and unlabeled peptidoglycan fragments presumably presented in Figure 8. The data on panel D, left graph does not show any immediate presence in the peptidoglycan of unlabeled peptidoglycan fragments. In fact, these are as low in abundance as in none treated or aztreonam treated cells (that elongate but do not divide). These two figures are hard to reconcile.

The data presented in Figure 8 and Supplementary Figure S12 (Figure 11 in the revised manuscript) are extracted from the same experiments but cannot be compared as suggested by the reviewer since they concern the kinetics of variations in the percentage of the various isotopologues (new→new, new→old, and old→old) versus the ADRNS, respectively. The latter, ADRNS, provides an estimate of the acceptor-to-donor ratio in neo-synthesized stems in dimers. It therefore does not depend upon the absolute rate of replacement of old (heavy) by new (light) stems. Instead, it provides an estimate of the proportion of new and old stems in the acceptor and donor position of dimers. The data appearing in the two figures are correct: Even though the absolute rate of replacement of old stems by new ones is low in the presence of mecillinam (Figure 8D; right panel), it is clear that the old→new and new→new dimers both increase at the beginning of the kinetics resulting in an ADRNS of 0.31 at 4 min (Figure 11). In contrast, the appearance of new→new dimers is delayed in the presence of aztreonam resulting in a lower ADRNS (0.14 at 5 min; Figure 11; extrapolation of 0.10 at t = 0). In the presence of aztreonam, the ADRNS values increase with time as expected from multiple and independent rounds of insertion of new material at the same location. We added three sentences in the legend to Figure 11 to indicate that data presented in this figure originate from the treatment of data appearing in Figure 8. These sentences also explain the origin of the apparent discrepancy underscored by the reviewer.

Additional sentences appearing in the legend to Figure 11 are as follows:

“Curves are obtained by linear regression. ADRNS were deduced from the relative abundance of new→old and new→new dimers appearing in Figure 8. Please note that the ADRNS was the highest for growth in the presence of mecillinam (this figure) in spite of the fact that the overall rate of replacement of old stems by new stems was the lowest for this growth condition, as evidenced by the slow rise of both the new→old and new→new dimers (Figure 8). This apparent discrepancy is accounted for by the fact that the ADRNS is a ratio solely based on neo-synthesized dimers.”

2. As indicated in several figure legends, the data presented is from two independent experiments. Does this mean that the authors only performed twice the chase experiments? The experiment itself is rather simple to perform and it's the downstream analysis that is extremely time consuming. How confident are the authors of the reproducibility of the data?

The reviewer is correct in stating that the experiments were performed twice. However, each one of the two experiments includes an independent analysis of at least five biological samples recovered at various time points (peptidoglycan extractions, digestion with muramidases, separation of the muropeptides by rpHPLC, and isotopologue identification by mass spectrometry). This approach provides evidence that the determination of muropeptide isotopologues between data points from the same experiments and between two fully independent experiments are consistent and reproducible. Our strategy was to focus on extensive characterization of a limited number of samples rather than generating a larger set of less extensively characterized data. Nonetheless, in the additional experiment addressing the next comment (comment 3), we generated a set of four biological repeats and the reviewers will find an analysis of the isotopologue measurement variability added to our answers as a spreadsheet file entitled “Variability analysis, answer to comment 2”.

3. The cells are grown in mecillinam and aztreonam, which inhibit length growth and division, respectively. The cells are expected to have a round and filamentous morphology, respectively, but this is not shown in the manuscript, where a very "non wild-type" strain is used for the experiments. The concentration of the antibiotics is perhaps based on the literature, but maybe this strain has difference susceptibility to the antibiotics? Using the wild type strain in these experiments would be important to demonstrate relevance of the findings.

We determined the minimal inhibitory concentrations (MICs) of aztreonam and mecillinam against strains BW25113 and BW25113Δ6ldt (additional Supplementary Table S3). Aztreonam and mecillinam were used at concentrations (2.5 µg/mL and 12 µg/mL, respectively) that were thus inhibitory for both strains (as we initially expected).

Strain BW25113Δ6ldt does not produce any L,D-transpeptidase. In strain BW25113, these enzymes marginally contribute to peptidoglycan cross-linking since YcbB and (p)ppGpp are produced at “wild-type” levels (Hugonnet et al. 2016). Thus, the MICs were expected to be similar for both strains and were experimentally shown to be so in the additional analysis presented in Table S3 (introduced Lines 481 to 483 of the revised manuscript):

“Aztreonam and mecillinam were used at inhibitory concentrations corresponding to 200- and 20-fold the MIC of the drugs, respectively (Supplementary Table S3).”

As requested by the reviewers, we performed a morphology analysis (additional Figure 8 – figure supplement 1). Mecillinam at 2.5 µg/mL and aztreonam at 12 µg/mL resulted in the expected changes in bacterial cell morphology, the formation of filaments or of spheres, respectively. The figure was introduced Lines 483-486 of the revised manuscript:

“Exposure to aztreonam or mecillinam resulted in the morphological changes expected for the inhibition of septum formation (filamentation) or of side wall synthesis (rounding), respectively (Figure 8 – figure supplement 1).”

The procedure to prepare the sample is described Lines 734 to 739 of the revised manuscript:

“For microscopy analyses, 1 mL of the culture used for peptidoglycan analysis was withdrawn and fixed by addition of paraformaldehyde (final concentration 4%; v/v). The sample was incubated at 4 °C for 15 min, centrifuged, the pellet was washed twice with 1 mL of phosphate buffer saline (PBS), and resuspended in 1 mL of PBS. Bacteria were imaged using agarose-pad slices (1% agarose in PBS). Phase contrast images of bacteria were taken using a 100x magnification objective on a Nikon TI inverted microscope.”

As recommended by the reviewers, we have repeated the full experiment with the parental “wild-type” strain BW25113 in the absence of β-lactam and in the presence of aztreonam or mecillinam. The results appear in Figure 8 – figure supplement 2. The initial choice of the BW25113Δ6ldt strain was motivated by the highly simplified muropeptide profile of this strain (absence of tripeptide and of 3→3 cross-linked dimers) that was essential to extensively characterize all the cross-linked muropeptides. We maintain that this choice was justified and additionally provides the requested analysis of a “wild-type” strain, i.e. the parental strain used for construction of BW25113Δ6ldt, which does not harbor any known mutation in genes involved in peptidoglycan metabolism. As expected from the similarity between the muropeptide compositions of strains BW25113 and BW25113Δ6ldt, the ADRNS values were similar for BW25113Δ6ldt and parental “wild-type” strain BW25113 (Figure 8 – figure supplement 3). This additional experiment establishes that the results obtained with BW25113Δ6ldt (in the first version of the manuscript) were not biased by the absence of the six ldt genes.

The analyses performed with the parent “wild-type” strain grown in the absence of β-lactams or in the presence of aztreonam or mecillinam appear in an additional section at the end of the results (Lines 506 to 522):

“Mode of insertion of peptidoglycan subunits in a bacterial strain harboring the wild-type complement of PBP and L,D-transpeptidase genes. Our last objective was to characterize the mode of peptidoglycan synthesis in a wild-type background with respect to genes involved in peptidoglycan synthesis. Toward this aim, we analyzed the peptidoglycan from the parental “wild-type” strain BW25113. Figure 8- figure supplement 2 provides a comparison of the kinetics obtained for both the BW25113∆6ldt and parental BW25113 strains grown in the absence of β-lactams and in the presence of aztreonam or mecillinam. This analysis showed that the conclusions drawn from the analyses performed with strain BW25113Δ6ldt (Figure 8) are not biased by the deletion of the six ldt genes. The ADRNS values were similar for both strains (Figure 8- figure supplement 3) indicating that the mode of insertion of new stems in the Tetra→Tetra dimers is not affected by the deletion of the six ldt genes. For the sake of completeness, we also compared the kinetics of appearance of muropeptides resulting from the activity of L,D-transpeptidases that are present in low amount in the parental strain BW25113 and are more abundant in mutant M1.5 due to overproduction of YcbB (Supplementary Table S1). Qualitatively, the kinetics were similar for muropeptides extracted from these two strains (Figure 8- figure supplement 4). Overall, the replacement of old by new muropeptides was more rapid for the “wild type” strain than for mutant M1.5.”

4. In the absence of LD-TPases the insertion of new PG is predominantly new-old for the first 20 minutes, whereas in the M1.5 strain, this is 50% new-old and 50% new-new. Their mass doubling times are 67 and 90 min, respectively. Both strains should contain dividing cells directly after the medium change. If septum formation is multistrand insertion, one would expect new-new in both strains. Then aztreonam is added to the δ 6 LD-TPases strain and the mode of insertion remains the same (new-old), suggesting that length growth is single strand insertion. However, this cannot be concluded because with division it was also single stranded. Then in the presence of mecillinam the authors claim that they see new-old and new-new insertion and therefore, claim that division requires multiple strand insertion. However, in the graph of the mecillinam cells, (1) it looks as if the PG synthesis is considerably slowed down in comparison to the other two situations. This is unexpected given the that the mass doubling time did not change according to Table S2. (2) If one extrapolates the lines in the graph, they are not that different from the strains without antibiotics or with aztreonam. (3) The new-old dominates, which one does not expect from cells that enlarge predominantly through division. The explanation that the futile cycle of PG TGase activity may result in insertion of old-new strand by PBP1b and 1a is plausible, but it does not prove that septation is multistrand. Would it be possible, to calculate the amount of surface generated through septation and elongation per min and predict the expected increase in old-new and new-new in the M1.5 strain to see if the observed data would fit the proposed model?

As indicated in our answers to comment #2, our data actually do show, based on ADRNS determination, that the mode of insertion is different in the presence of aztreonam (mostly single strand) and in the presence of mecillinam (a combination of single and multi-strand insertion). Our answers to the additional points raised in comment #4 are as follows:

(i) The author states that “it looks as if the PG synthesis is considerably slowed down [in the presence of mecillinam] in comparison to the other two situations. This is unexpected given the that the mass doubling time did not change according to Table S2.”

The analysis proposed by the reviewer should take into account the relative contributions of cell poles and side walls to the cell surface in addition to overall growth. If the length of the side wall in the long axis of bacterial cells is twice the cell diameter, then the contribution of the poles and side walls to the cell surface is 33% and 67%, respectively. It is thus expected that the overall synthesis of new material is slowed down to a greater extent by blocking elongation of the side wall than by blocking formation of the poles. The kinetics observed in the absence of the drugs can be accounted for by an additive combination of kinetics obtained in the presence of mecillinam and aztreonam with a greater contribution of the latter condition (extension of side walls). Regarding the variation in OD, it is not obvious that the increase in the cell surface is directly proportional to the optical density. In the legend to Table S2 (first version of the manuscript, footnote b), we have clearly indicated that generation times for the aztreonam and mecillinam conditions were deduced from the variation in the OD₆₀₀ during 90 min. We think that these data should appear in the manuscript as an estimate of potential increases in the biomass. Obviously, these estimates do not reflect steady state growth. Thus, they should not be over-interpreted to relate changes in peptidoglycan composition to putative growth rates that cannot be precisely defined for all growth conditions used in this study.

2) “If one extrapolates the lines in the graph, they are not that different from the strains without antibiotics or with aztreonam.”

As indicated in the first version of the manuscript, extrapolation should be made for t=0 min (Figure 11; revised version, formerly Figure S12). When the proportion of new material increases in the wall, the probability that new stems are cross-linked to unlabeled material inserted after the medium switch increases. It is therefore expected that extrapolations for long incubation times can be convergent for different modes of insertion (single versus multiple strands) as observed at 40 min in Figure 11.

3. “The new-old dominates, which one does not expect from cells that enlarge predominantly through division. The explanation that the futile cycle of PG TGase activity may result in insertion of old-new strand by PBP1b and 1a is plausible, but it does not prove that septation is multistrand. Would it be possible, to calculate the amount of surface generated through septation and elongation per min and predict the expected increase in old-new and new-new in the M1.5 strain to see if the observed data would fit the proposed model?”

The proposed evaluation of the amount of surface generated through septation and elongation per min is feasible. However, we think that this analysis would be extremely speculative due to numerous confounding factors. (i) The rate of expansion of the cell surface may vary during the entire experiment and between individual bacterial cells. (ii) Steady state growth is unlikely to occur in the presence of β-lactam. (iii) Peptidoglycan recycling may be affected by exposure to drugs and is likely to differently affect the rate of synthesis of the septum and of side walls. For these reasons, we would like to stay with the initial conclusions that side wall synthesis mostly involves single strand insertion of new strands, whereas synthesis of the septum involves insertion of multiple strands. We think that we cannot provide a more refined estimate of the proportion of new material inserted as single versus multiple strand in bacteria treated with β-lactams.

5. In general, it is a pity that no wild-type strain is used for the new method. The conclusion of the paper is very important, i. e. new PG is inserted strand by strand likely by the combined action of an TPase and an endopeptidase. The absence of data on a WT strain diminishes the soundness of the conclusion. Since it can be expected that this paper will be cited for a long time by many people, it is important that it is complete. Experiments with the wild type should be reported.

As stated in our answer to comment #2, we have added to our study an extensive analysis of a “wild-type” strain, i.e. the parental strain BW25113, that does no harbor any known mutation directly affecting peptidoglycan synthesis. The results appear at the end of the Results section under the subheading “Mode of insertion of peptidoglycan subunits in a bacterial strain harboring the wild-type complement of PBP and L,D-transpeptidase genes” (Lines 506 to 522) and in three additional figures (Figure 8 – figure supplement 2, 3 and 4).

6. The discrimination of labelled versus unlabelled assumes that if one isotope is present, the molecule is considered to be labelled and if no isotopes are present the molecule is unlabelled. A situation must exist (as is outlined in Figure S14 and how the mass spectra data are poled in S15) in which for instance glucosamine is made from a mixture of labelled and unlabelled 13C and existing labelled DAP (which was also reported as having a considerable cellular pool) is coupled to an newly synthesized (unlabeled) D-Ala-D-Ala. This would result in a high number of different MS peaks which all present the same structure. Assuming that the relative abundance of these variable labelled molecules will be the same for all muropeptides, but H1 and H2, is 5 min enough to cause every newly synthesized muropeptide to be fully unlabelled during the chase or is there a mixture for some time and how longs do these mixtures last? Maybe this gives some information on the presence of the pools of the various parts of the muropeptides?

Our analyses are based on the characterization of hybrid muropeptides by tandem mass spectrometry. This analysis showed that the major peaks can be assigned to mosaic structures with a defined isotopic composition (e.g. Figure 4). A first conclusion is that the main isotopologues are made of combinations of labeled and unlabeled moieties (amino acids, glucosamine, acetate, and lactate) as opposed to a hybrid composition of labeled and unlabeled nuclei within these building blocks. Specifically, the fact that we only detected muropeptides with fully labeled or fully unlabeled glucosamine moieties indicates that glucosamine is not merely made of a mixture of heavy and light isotopes in agreement with the pathways for de novo synthesis of glucosamine and for the recycling of GlcNAc and MurNAc residues (Figure 1 – figure supplement 1). Hybrid h3 clearly originated from a pool of UDP-MurNAc-pentapeptide as expected from the abundance of this precursor in cytoplasm (Mengin-Lecreulx et al. 1982, cited in the first version of the manuscript). The h3 pool levels off over time, in contrast to h1 and h2, as expected from the exhaustion of a preexisting pool. The latter hybrids, h1 and h2, originate from recycling of existing (labeled) peptidoglycan. It is therefore expected that their accumulation will only level off when the proportion of old (labeled) peptidoglycan decreases due to insertion of new material. Regarding DAP, it is true that this precursor is also abundant. However, DAP is a precursor for L-lysine synthesis. Thus, the bulk of the DAP pool is used for protein synthesis. For D-Ala-D-Ala, we indeed detected hybrid tetrapeptides that were uniformly unlabeled except for the terminal D-Ala. We did not perform any pulse-chase experiment and could therefore not provide any insight into the size of the precursor pools and the fluxes in metabolic pathways.

7. The GlcNAc-anhydro-MurNAc- peptides must have a discrete mass from the standard muropeptides, but are not identified as such. Why not? It gives information in the length of the glycan strands and as such is interesting.

We agree that the determination of content of anhydro-MurNAc versus MurNAc-peptide could provide an indication of the length of glycan chains. However, in the context of our analysis, they are two caveats. First, the current determination of the isotopic composition of muropeptides relies on the detection of isotopologues with the same chemical composition as purified in a single peak of rpHPLC chromatograms. The determination of the proportion of the isotopologues is therefore extremely precise since it is not affected by the purification yields and the efficacy of ionization in the electrospray ion source. In contrast, estimates of the length of glycan chain relies on the determination of material present in different rpHPLC fractions. Our chromatographic gradient is optimized for MS analysis since is enables recovering muropeptides free from salt following lyophilization. However, the resolution of peaks is not sufficient for reliable determination of minor muropeptide species. Second, such analyses would only provide estimates of the average length of the glycan chains in a potentially heterogeneous material depending on the extents of peptidoglycan polymerization and recycling. Unfortunately, our study design does not allow us to adequately answer to the reviewer’s request. In addition, we think that discussing the length of glycan chains would introduce an entire new set of concepts in our study, which is already, we believe, highly complex for the readers. As shown in the figure below, the kinetics for muropeptides and anhydro-muropeptide were similar indicating that the behavior of the material at the end of the glycan chains does not substantially differ from that of the inner material. We would therefore like not to discuss anhydro-muropeptides in this study.

8. Page 20: "Thus, the absence of old-new 3-3 cross-linked Tri-Tetra dimer indicates that YcbB discriminates neo-synthesized tetrapeptide stems (used as donors) from existing tetrapeptide stems (used as acceptors)" Could it be that these crosslinks were below the detection limit? It is important because If YcbB is indeed only able to make 3-3 crosslinks from newly synthesized PG, this implies that repair is only possible if a TGase polymerizes PG. It has been shown by Morè et al., (Morè, N., Martorana, A.M., Biboy, J., Otten, C., Winkle, M., Serrano, C.K.G., et al. (2019) Peptidoglycan Remodeling Enables Escherichia coli To Survive Severe Outer Membrane Assembly Defect. mBio 10.) that YcbB works together with PBP1B and PBP6a and the absence of old-new may explain and support this further.

Thank you for drawing our attention to this interesting implication of our results. As proposed by the reviewer, we included a short paragraph on the required participation of the glycosyltransferase activity of PBP1b to peptidoglycan remodeling mediated by YcbB (Lines 452 to 461 of the revised manuscript):

“In this study, we analyzed the participation of YcbB to global peptidoglycan synthesis following overproduction of this L,D-transpeptidase and of the alarmone (p)ppGpp in combination with inhibition of PBPs in certain experiments. Previous work concluded that YcbB also participates in the repair of peptidoglycan in a wild-type background (Morè et al. 2019). In addition to YcbB, this process required glycosyltransferase and D,D-carboxypeptidase activities provided by PBP1b and PBP6a, respectively. This implies that peptidoglycan repair proceeds by polymerization of new glycan chains that are cross-linked by PBPs. It is worth noting that the mode of insertion of new peptidoglycan subunits delineated in the current study is fully compatible with this model since we detected only one type of 3→3 cross-linked dimers (new→old and not old→new).”

Other corrections:

1. In the introduction, the authors mention "lytic glycosyltransferases". The term "lytic transglycosylase" seems more appropriate.

The text has been corrected as required (Line 124 of the revised manuscript).

2. Ln 62: a comma is missing after "in E. coli".

A comma has been added as required (Line 67 of the revised manuscript).

3. Ln 104: Change "NMR" to "Solution-state NMR" because solid-state NMR, although limited in its applications, has been successful in characterizing the tertiary structure of peptidoglycan in bacteria. For example, "Staphylococcus aureus peptidoglycan tertiary structure from carbon-13 spin diffusion" (JACS, 2009).

The reference cited in the first version of the manuscript (Kern et al., 2008) was also based on solid state NMR. We agree that the sentence in the first version of the manuscript was not sufficiently qualified. We propose to replace the sentence:

“NMR also failed to determine the structure of the polymer due to its heterogeneity and its important flexibility as measured by NMR relaxation (Kern et al., 2008).”

By the following sentence that incorporates the suggested citation:

“NMR spectroscopy also failed to determine the precise and complete structure of the polymer due to its heterogeneity and its important flexibility, as measured by solid state NMR relaxation (Kern et al., 2008, Sharif et al., 2009).”

Changes appear Lines 109 to 111 of the revised manuscript.

4. The Figure S2. ADRR calculation assumes the PG model which consists primarily of PG dimers with 50% crosslinking efficiency. The literature is in good agreement with an average PG crosslinking efficiency in E. coli observed at 50% (ranging from 30~60% depending on the strains),1 but in addition to a significant number of monomers (30 to 55%) and dimers (41 to 51%), there are trimers (4 to 10%) and tetramers (less than 1%) that are found in E. coli. 2 These oligomeric muropeptide species are low in their abundance, however, a statement on how these oligomers are consistent with the insertion models would be important.

We agree that trimers and tetramers contribute to the ADRR as calculated from data based on the previous radioactive labeling techniques. The situation is different for our study since the analysis of dimers is fully independent from that of trimers and tetramers. We advocate that including the analysis of trimers and tetramers is beyond the scope of the current study. In addition, structure assignment is hampered by the following issues: (i) The abundance of muropeptide trimers (7%) and tetramers (<1%) is too low to permit the required MS/MS analyses. (ii) For trimers and tetramers, all isomers are not resolved in individual rpHPLC peaks. This introduces ambiguity in the MS/MS analyses. (iii) The number of isotopologues for each isomer is very high, resulting in overlapping isotopic clusters. Thus, unambiguously assigning the structure of isotopologues is out of reach for molecules as complex as trimers and tetramers. We added a sentence Lines 285 to 288 of the revised manuscript to indicate that small amounts of trimers and tetramers were present in our preparations but that we were unable to characterize them: “The peptidoglycan preparations also contained small amounts of trimers and tetramers that were not analyzed in the current study because their low abundance, poor resolution by rpHPLC, and complex isotopologue composition precluded their full characterization by MS and MS/MS.”

5. Ln 211: insert ", Peak A" after "The second peak of the isotopic cluster".

The text has been modified as requested (Line 202 of the revised manuscript).

6. Ln 224: Change "cluster D" to "Peak D".

The text has been modified as requested (Line 214 of the revised manuscript).

7. Ln 227: Change "clusters" to "Peaks".

The text has been modified as requested (Line 217 of the revised manuscript).

8. Ln 231: "(99% for each isotope)" mentioned in the text is for the isotopic enrichments for the heavy-labeled glucose and ammonium chloride. However, the isotope incorporation efficiency in E. coli seems to be less than 99% because of the observed spectrum of Figure 2D which shows the larger peak intensities for Peaks A, B, and C compared to the corresponding ratio shown in the simulated spectrum (Figure 2E). Since isotopic distribution is highly sensitive to dilution, perhaps 25 ml of preculture used to inoculate 1L of labeled media was grown in unlabeled M9. If so, then the dilution is approximately 25ml/(100ml + 25ml) = 2.4% and thus the expected isotopic incorporation efficiency is 96.5% (97.5% * 99%) which may account for the differences in the observed and calculated Peaks for A, B, and C? This is a minor point since it does not change the overall analysis.

We were aware of the impact of the inoculum, if performed in the unlabeled medium, on the labeling efficacy. As described in the first version of the manuscript, both the preculture and culture were performed in the labeled medium to avoid any isotopic dilution. We added a short sentence in parenthesis to emphasis this fact in the “Results” section (Lines 222 and 223 of the revised manuscript): “please, note that both the preculture and the culture were performed in labeled medium to avoid isotopic dilution.” We also added a sentence in the “Materials and methods” section (Lines 726 to 727 of the revised manuscript): “Thus, both the preculture and the culture were performed in labeled liquid medium to avoid isotopic dilution.”

We agree that the chromatogram displayed in Figure 2E shows a slight enrichment in isotopologues containing one or two light isotopes in comparison to the simulated spectrum. Correcting the extent (99%) of labeling reported for ¹³C- and ¹⁵N-labeling of glucose and ammonium chloride by the manufacturer of these compounds could be used to improve the fit between simulated and observed spectra. However, introducing this modification, which is in the order of 0.2% (98,8% instead of 99%), will require complex description and justification, with minimal improvement the manuscript.

9. Ln 251, Ln 264: "Table 2" is missing from the manuscript.

We apologize for this error. The text should have referred to Table 1 (corrected Line 234 and 247 of the revised manuscript).

10. Ln 308: insert "heavy-isotope labeled" before "minimal medium".

The suggestion has been incorporated into the revised manuscript. The requested modification has been inserted Line 291 of the revised manuscript

11. Ln 467 "Tri-Tri and Tetra-Tri dimers formed by PBPs and YcbB, respectively" should be changed to "Tri-Tri and Tetra-Tri dimers formed by YcbB and PBPs, respectively".

We apologize for this error and have corrected the text accordingly (Line 425 of the revised manuscript).

12. Ln 468: "were almost exclusively of the new-old type (Figure 7)" But Figure 7 A (Tri-Tri) and B (Tetra-Tri) show an even mixture of both new-old and new-new. Hence, the statement "exclusively of new-old" is not entirely correct.

We respectfully disagree with the comment of the reviewer. The proportion of the new→new isotopologue remained close to 0% in the first 20 min for the Tri→Tri (panel A) and Tetra→Tri (panel B) dimers whereas that of the new→old reached 18% during the same time period. The first version of the text may have been misleading since we did not mention that we were considering early times after the medium switch. The corresponding modifications appear in lines 426 to 427 of the revised manuscript: “…. at early times after the medium switch (i.e. ≤ 20 min) (Figure 7A and 7B).”

13. Ln 468-469: "In contrast, Tri-Tetra and Tetra-Tetra dimers formed by these enzymes were a mixture of new-new and new-old" but the figure shows only new-new.

Again, the text was correct in referring to a mixture of new→new and new→old isotopologues for the Tri→Tetra (panel C) and Tetra→Tetra (panel D) dimers at early times.

To improve clarity, we have modified the text to make clear that we are considering short incubation times after the medium switch. In addition, references to specific panels of Figure 7 were added. We also indicated that incubation for longer time periods resulted in an increase in the proportion of new→new isotopologues for all types of dimers and that this increase was expected since serial insertions of glycan strands at the same location results in the formation of new→new dimers.

These modifications appear in lines 427 to 438 of the revised manuscript: “These dimers contained a tripeptide stem at the acceptor position. In contrast, Tri→Tetra and Tetra→Tetra (formed by YcbB and PBPs; Figure 7C and 7D respectively), which contained a tetrapeptide stem at the acceptor position, were mixtures of new→new and new→old types of dimers at early times after the medium switch. Incubation for longer time periods resulted in an increase in the proportion of new→new isotopologues for all types dimers. This increase was expected since serial and independent insertions of glycan strands at the same location results in the formation of new→new dimers. Note that this increase was more rapid for dimers containing a tetrapeptide stem at the acceptor position as observed for tetrapeptide-containing monomers. As described above, this difference reflects the delayed formation of tripeptide stems from tetrapeptide stems by cleavage of D-Ala4 by the L,D-carboxypeptidase activity of YcbB. At early times, tripeptide-containing new stems may therefore be present in small amounts.”

14. Ln 475: The statement "Thus, the mode of insertion of PG subunits depended upon the structure of the acceptor stem … rather than upon the type of transpeptidase" is based on the observation from Figure 7 which shows that panels A and B (Tri as acceptor), and panels C and D (Tetra as acceptor) showed similar kinetic behaviors. The addition of a few sentences for an explanation would add clarity to the statement.

We have rewritten the conclusion of this section and extended it with an explanation (lines 439 to 443 of the revised version):

“Together, the results shown in Figure 7 revealed that the kinetics of formation of isotopologues containing a tripeptide stem at the acceptor position were similar for the Tri→Tri and Tetra→Tri dimers generated by YcbB and PBPs, respectively. Likewise, the kinetics of formation of dimers containing a tetrapeptide stem at the acceptor position were similar for the Tri→Tetra and Tetra→Tetra generated by YcbB and PBPs, respectively.”

15. Figure 7 provides insight into the flux of how much new (unlabeled) PG is incorporated. For example, in Figure 7 C (Tri-Tetra 3-3 CL) at 85 mins the observed N-N is at ~80%, O-O at ~10%, and N-O at ~10%, then the relative total amount of Old PG subunit (2*10% + 10% = 30%) to the New PG subunit (2*80% + 10% = 170%) for Tri-Tetra of 3-3 CL is 3:17. The approximate ratio of old to new for Tetra-Tetra of 4-3 is approximately 1:19. This indicates that the newly formed PG subunits (nascent PG) are incorporated into both 4-3 and 3-3 CL with a tetrapeptide stem, although more actively to 4-3 CL. In contrast, the approximate ratio of old to new for Tri-Tri (3-3 CL) and Tetra-Tri (4-3) is 9:17 and 1:2, respectively. This indicates that the newly formed PG subunit in the form of a tripeptide acceptor is not readily incorporated into the cell wall. Hence, the insertion into PG shows dependence on the PG stem structure of the acceptor. Alternatively, the concentrations of Tri and Tetra are not independent but coupled, most likely high concentration of Tetrapeptide PG stems (both monomers and dimers) as shown Figure 6 (b-f) is likely to be the determinant for the observed kinetics.

We thank the reviewer for these suggestions and have provided a more detailed analysis of the data that takes them into account (lines 443 to 451 of the revised manuscript):

“In contrast, distinct kinetics were observed between dimers containing a stem tripeptide or tetrapeptide in the acceptor position irrespective of the transpeptidase responsible for their formation (mainly new→old versus a mixture of new→new and new→old). Thus, the mode of insertion of peptidoglycan subunits depended upon the structure of the acceptor stem (tripeptide versus tetrapeptide) rather than upon the type of transpeptidase (PBPs versus YcbB). In addition, it is worth noting that the newly formed peptidoglycan subunits are more rapidly incorporated into dimers containing a tetrapeptide stem in the acceptor position than into dimers containing a tripeptide stem at this position.”

16. Ln 519: Please explain the amounts of antibiotics that were added, Mec (2.5 ug/ml) and AZT (12 ug/ml), to E. coli before the medium switch.

We used inhibitory concentrations of the drugs as mentioned Lines 481 to 483 of the revised manuscript:

“Aztreonam and mecillinam and were used at inhibitory concentrations corresponding to 200- and 20-fold the minimal inhibitory concentration (MIC) of the drugs.

17. Ln 557 Figure 9A: "… grown in the presence of ampicillin", at what time point was the sample collected?

The following text was appended to the sentence (Lines 1084 and 1085 of the revised manuscript):

“….. which was added 5 min prior to the medium switch and to the unlabeled culture medium used for the medium switch corresponding to t=0.”

18. Ln 546-549: "New-new rather than new-old isotopologues were predominantly detected among dimers……. In contrast, the Tri-Tetra dimers in the PG of M1.5 grown in the absence of ampicillin mainly contained the new-old isotopologues (Figure 7)". Figures below show kinetics for Tri-Tetra dimers from Figure 9 (left) and Figure 7C (right). Contrary to the statement, PG of M1.5 grown in the absence of ampicillin (Figure 7C, right) contained a significant amount of new-new isotopologues.

In Figure 7C, the curve for the new→old isotopologues (grey curve) was above that for the new→new isotopologue (purple curve) during the first 20 min of the reaction (for the Tri→Tetra dimer). This indicates that the new-old isotopologues were more abundant than the new-new isotopologues for the Tri→Tetra dimers from the peptidoglycan of M1.5 grown in the absence of ampicillin. We rephrased the sentence to indicate this fact and avoid the use of the word “predominantly” and to indicate that the first 20 min of the reaction were taken into account (Lines 501 to 503):

“In contrast, the new-old isotopologues were more abundant than the new-new isotopologues in the Tri→Tetra dimers from the peptidoglycan of M1.5 grown in the absence of ampicillin at early times (Figure 7).”

19. For Figures 5, 7, 8, and 9, please provide which program was used to fit the data. Also, what was the kinetic model used for the fit? Does the model assume sequential event of A to B to C (conversion from old-old to new-old and then to new-new)? An explanation for a buildup of new-old in Tri-Tri but not in Tri-Tetra in E. coli M1.5 grown in the presence of ampicillin (as shown in Figure 9C below) would be helpful.

The curves are interpolations. This is indicated in the revised legends to the figures. We do not want to propose particular models since there are too many variables (See Figure 1 – figure supplement 3).

20. Ln 577-579: The rationale for switching the medium from labeled to unlabeled was that it "facilitates the detection of neo-synthesized isotopologues present at a low abundance at early times after the medium switch." It was not clear how this facilitates the detection of neo-synthesized isotopologues compared to switching from unlabeled to label medium. Additional sentences explaining this would be of interest to the readers.

The additional sentences appear lines 536 to 542 of the revised manuscript.

21. Ln 610: "old-new" should be changed to "old-old".

We added the old→old isotopologues in the sentence. Please note that the procedure is expected to enable the detection of old→new isotopologues although they were not present in the peptidoglycan samples examined in the current study. Accordingly, the reference to old→new isotopologues was maintained in the sentence (Lines 570 and 571 of the revised manuscript).

22. Ln 671-673: Confusing statement. Ln 672: "new-old" should be "new-new"?

We apologize for this confusing statement. The sentence was ambiguous and has been rephrased, Lines 632 to 634 of the revised manuscript:

The resulting new→new cross-linked material was almost exclusively connected to the existing peptidoglycan by the formation of Tri→Tri dimers because new→old isotopologues were only detected in this type of dimers.

23. Figure S12 is of great significance and thus recommend moving this as one of the main figures in the text.

According to the reviewer’s recommendation, Figure S12 appears in the revised version of the main text as Figure 11.

24. Page 20 "Kinetic analyses revealed that the Tri-Tri and Tetra-Tri dimers formed by PBPs and YcbB, respectively, were almost exclusively of the new-old type (Figure 7). Should these be YcbB and PBPs, respectively.

The text has been corrected as suggested (Line 425 of the revised manuscript).

25. In Figure 6, S7, S8 and S9 strain M1.4 is used, but this strain is not described in the method section.

We apologize for these errors. The mutant is M1.5 (not M1.4).

26. The E. coli BW25113 rph mutation is a frame shift that leads to pyrimidine starvation. To have an optimal DNA replication cycle in M9 medium it needs some supplements, which were not added according to the Materials and methods. This may not affect the PG mode of insertion but perhaps the authors can keep this in mind for future experiments? All strains used in this study are derivatives of this strain. E. coli BW25113 [Δ(araD-araB)567 Δ(rhaD-rhaB)568 ΔlacZ4787 (::rrnB-3) hsdR514 rph-1, lacI+] BW25113 Needs uracil 20 ug/ml, when grown in minimal medium. KF Jensen J Bacteriol. 1993 Jun;175(11):3401-7. Grenier, F., Matteau, D., Baby, V., & Rodrigue, S. (2014). Complete Genome Sequence of Escherichia coli BW25113. Genome Announcements, 2(5). doi:10.1128/genomeA.01038-14. Needs also thymidine, 2 ug/ml according to this paper. J Bacteriol. 1998 Jun;180(11):2992-4. Determining the optimal thymidine concentration for growing Thy- Escherichia coli strains. Molina F1, Jiménez-Sánchez A, Guzmán EC.

We thank the reviewer for this information and will further explore the requirements of strain BW25113 for growth. In our hands, BW25513 does not require the addition of uracil or thymidine to the minimal medium. Genome sequencing did not reveal any mutation in our laboratory subculture of BW25113 in comparison to the reference genome. We also found a publication that questions the growth requirement of MG1655, which also harbors the rph frameshift mutation “The growth medium did not contain uracil, which has been shown to stimulate growth of E. coli MG1655, which has an rph frameshift mutation (Jensen, 1993). However, inclusion of uracil had no effect on logarithmic growth, growth arrest caused by isoleucine starvation, or rescue of growth by addition of isoleucine (data not shown).” (Traxler et al. 2008. 68:1128-1148; DOI: 10.1111/j.1365-2958.2008.06229.x)

Author details

1. Heiner Atze

   Centre de Recherche des Cordeliers, INSERM, Sorbonne Université, Université Paris Cité., Paris, France

   Contribution

   Data curation, Formal analysis, Investigation, Methodology, Software, Writing – original draft

   Competing interests

   No competing interests declared

   "This ORCID iD identifies the author of this article:" 0000-0003-1497-6373

2. Yucheng Liang

   Centre de Recherche des Cordeliers, INSERM, Sorbonne Université, Université Paris Cité., Paris, France

   Contribution

   Conceptualization, Data curation, Formal analysis, Methodology, Software, Writing – review and editing

   Competing interests

   No competing interests declared

3. Jean-Emmanuel Hugonnet

   Centre de Recherche des Cordeliers, INSERM, Sorbonne Université, Université Paris Cité., Paris, France

   Contribution

   Conceptualization, Formal analysis, Funding acquisition, Investigation, Methodology, Project administration, Supervision, Validation, Writing – original draft, Writing – review and editing

   Competing interests

   No competing interests declared

   "This ORCID iD identifies the author of this article:" 0000-0003-4150-0944

4. Arnaud Gutierrez

   Robustesse et Evolvabilité de la vie, Institut Cochin, Université de Paris, INSERM U1016, CNRS UMR 8104, Paris, France

   Contribution

   Data curation, Writing – review and editing

   Competing interests

   No competing interests declared

5. Filippo Rusconi

   1. Centre de Recherche des Cordeliers, INSERM, Sorbonne Université, Université Paris Cité., Paris, France
   2. PAPPSO, Université Paris-Saclay, INRAE, CNRS, Paris, France

   Contribution

   Conceptualization, Data curation, Formal analysis, Funding acquisition, Investigation, Methodology, Project administration, Software, Supervision, Validation, Writing – original draft, Writing – review and editing

   Contributed equally with

   Michel Arthur

   For correspondence

   filippo.rusconi@universite-paris-saclay.fr

   Competing interests

   No competing interests declared

   "This ORCID iD identifies the author of this article:" 0000-0003-1822-0397

6. Michel Arthur

   Centre de Recherche des Cordeliers, INSERM, Sorbonne Université, Université Paris Cité., Paris, France

   Contribution

   Conceptualization, Data curation, Formal analysis, Funding acquisition, Investigation, Methodology, Project administration, Supervision, Validation, Writing – original draft, Writing – review and editing

   Contributed equally with

   Filippo Rusconi

   For correspondence

   michel.arthur@crc.jussieu.fr

   Competing interests

   No competing interests declared

   "This ORCID iD identifies the author of this article:" 0000-0003-1007-636X

• 815

  Page views

• 303

  Downloads

• 2

  Citations

Article citation count generated by polling the highest count across the following sources: Crossref, PubMed Central, Scopus.