A synthetic 5,3-cross-link in the cell wall of rod-shaped Gram-positive bacteria

Significance

The cell wall of bacteria is a biopolymer formed of glycan chains cross-linked by interconnecting peptide stems. Elucidating the structure, biosynthesis, and recycling mechanisms of the cell wall is important to understanding the mechanism of action of β-lactam antibiotics and the development of new antibiotics. Here we use a synthetic biology approach to probe the structural and biosynthetic constraints of cell-wall architecture by introducing noncanonical building blocks into the cell walls of living bacteria. We show that a Gram-positive bacterium remains viable under circumstances where approximately 20% of the cell wall is interconnected by synthetic 5,3-cross-links, a cross-linking arrangement absent in Nature. Characterization of these synthetic cell-wall cross-links highlights key differences in cell-wall recycling between Gram-negative and -positive bacteria.

Abstract

Gram-positive bacteria assemble a multilayered cell wall that provides tensile strength to the cell. The cell wall is composed of glycan strands cross-linked by nonribosomally synthesized peptide stems. Herein, we modify the peptide stems of the Gram-positive bacterium Bacillus subtilis with noncanonical electrophilic d-amino acids, which when in proximity to adjacent stem peptides form novel covalent 5,3-cross-links. Approximately 20% of canonical cell-wall cross-links can be replaced with synthetic cross-links. While a low level of synthetic cross-link formation does not affect B. subtilis growth and phenotype, at higher levels cell growth is perturbed and bacteria elongate. A comparison of the accumulation of synthetic cross-links over time in Gram-negative and Gram-positive bacteria highlights key differences between them. The ability to perturb cell-wall architecture with synthetic building blocks provides a novel approach to studying the adaptability, elasticity, and porosity of bacterial cell walls.

Gram-positive bacteria are monoderms and their single membrane is encased by a multilayered (15–30 nm) cell wall. In contrast, Gram-negative bacteria are diderms and the cell wall is mono- or bilayered (3–6 nm) and resides between the inner and outer membranes (1). Notwithstanding this disparity in general architecture, the chemical structures of the peptidoglycan of each organism are remarkably similar (2). The cell-wall peptidoglycan, or murein, is composed of glycan strands assembled from the saccharide constituents N-acetyl muramic acid (NAM) and N-acetyl glucosamine (NAG), and in each class of bacteria a pentapeptide extends from the lactyl group of the NAM saccharide (Fig. 1A) (3). While variability in the sequence of the peptide stem exists across species, the archetypal structure of the monoderm stem is l-Ala-γ-d-Glu-l-Lys-d-Ala-d-Ala, and in the diderm l-Lys is typically substituted with the carboxy derivative meso-2,6-diaminopimelate (m-DAP) (Fig. 1B) (3, 4). The peptide stems of the cell wall are cross-linked by penicillin-binding protein (PBP) d,d-transpeptidases, and l,d-transpeptidases. In Gram-negative bacteria, the epsilon amino group of m-DAP serves as a nucleophile displacing a terminal d-Ala of an adjacent peptide stem in the formation of d,d-transpeptidase-catalyzed 4,3- and l,d-transpeptidase-catalyzed 3,3-peptide cross-links (5). Both cross-linking arrangements stabilize the cell wall and the l,d-transpeptidase cross-link additionally anchors the cell wall to the outer membrane via Braun’s lipoprotein (6). In Gram-positive bacteria, the epsilon amino group of lysine acts as the corresponding nucleophile in cross-link formation. Due to the absence of an outer membrane in Gram-positive bacteria, l,d-transpeptidase cross-linking is a rare event that may aid in providing mechanical stability. Functional redundancy is evident in cell-wall synthesis as Escherichia coli encodes at least five d,d-transpeptidases and six l,d-transpeptidases, while Bacillus subtilis encodes at least 10 d,d-transpeptidases and two l,d-transpeptidases (5, 7, 8). This redundancy implies that bacteria have evolved a careful balance in cell-wall assembly that forms an ideal biopolymer for bacterial life.

Fig. 1.

(A) The cell wall of Gram-positive bacteria is anchored to the outer leaflet of the membrane by lipoteichoic acids (LTAs; shown in yellow). Walled teichoic acids (WTAs; shown in orange) provide additional support between branches (36). (A, B) The peptidoglycan is composed of glycan that assembles from the disaccharide pair NAG (shown and labeled in dark green)-NAM (light green). Peptide stems (blue) connect the glycan via enzyme-catalyzed cross-linking reactions. The glycan of the cell wall is degraded by muramyl hydrolases, and to a lesser extent by lytic transglycosylases, the latter producing anhydroNAM (shown in red) as the reaction product (37).

The arrangement of the peptide cross-links dictates the dimensionality, proportions, and porosity of the cell wall as well as the shape of the bacterium. To explore the structural constraints of the cell-wall architecture that allow it to meticulously orchestrate essential biological processes and maintain cellular morphology, we have begun to replace the naturally existing canonical cell-wall cross-links (Fig. 1B) with unnatural synthetic cell-wall cross-links. Previously, we used electrophilic noncanonical d-amino acids (d-AAs) to form unnatural synthetic cell-wall 4,3-cross-links in the Gram-negative bacterium E. coli (9). Herein, we extend this approach to the Gram-positive bacterium B. subtilis, revealing differences in synthetic cross-link arrangements, synthetic cross-link–induced effects on cell growth and phenotype, as well as differences in the mode of accumulation of synthetic cell wall within the bacteria.

Results and Discussion

Experimental Approach.

Our previous studies on the formation of unnatural cell-wall cross-links in Gram-negative bacteria depended on the ability of bacterial transpeptidases to incorporate environmental d-AAs into the cell wall, a strategy used by bacteria to reduce the extent of peptide cross-linking (Fig. 2A) (9–12). In seminal studies, others have used this approach to incorporate fluorophores and biorthogonal photo–cross-linkers into the cell wall of diverse bacteria for live-cell cell-wall imaging (13, 14). By using noncanonical electrophilic d-AAs as substrates for the transpeptidases, we were able to replace ∼30% of the natural cell-wall cross-links in E. coli with unnatural synthetic cell-wall cross-links with no accompanying observable effect on bacterial phenotype (9). To determine if a similar strategy can be used to form unnatural cell-wall cross-links in Gram-positive bacteria, we selected B. subtilis. B. subtilis is a model bacterium for the study of the structure of the rod-shaped, spore-forming capable Gram-positive bacterium (15). A key distinction between the cell wall of B. subtilis and other Gram-positive bacteria is the presence of the archetypal Gram-negative m-DAP at the third position of the peptide stem, which in B. subtilis commonly exists as an enzyme-catalyzed amidated amino acid (16).

Fig. 2.

(A) Comparative mechanisms of d,d-transpeptidase-mediated canonical cell-wall cross-linking and noncanonical cell-wall cross-linking by exogenous d-AAs in B. subtilis. (B) Structures of electrophilic noncanonical d-AAs and cognate l-AA controls. (C) Bacterial growth curves of B. subtilis untreated or treated with d-AA 1c or l-AA 1d. (D) Structure of the noncanonical 5,3-cross-linked NAG-NAM-(pentapeptide)-NAG-anhydroNAM-(tetrapeptide) formed by 1c. The gray boxes show a comparison of the synthetic cross-link and the native cross-link. For each noncanonical d-AA the primary synthetic cross-linked muropeptide formed comprises the canonical NAG-NAM-(tetrapeptide) (R′) and NAG-anhydroNAM-(tetrapeptide) (R′′), where the noncanonical d-AA is installed adjacent to the fourth-position d-Ala of the R′ stem, replacing the fifth-position d-Ala. The mass spectra corresponding to the synthetic non–cross-linked and cross-linked cell-wall species formed by (E, F) 1a, (G, H) 1b, and (I, J) 1c are shown. Compounds 2a and 2b form synthetic non–cross-linked muropeptides, but not synthetic cross-linked muropeptides. The structures and masses of each unnatural synthetic cell-wall muropeptide is provided in SI Appendix (SI Appendix, Figs. S3–S8).

In Gram-negative bacteria, noncanonical d-AAs are incorporated into the peptide stem at the fourth position by l,d-transpeptidases and at the fifth position by d,d-transpeptidases (17). In our previous efforts to generate unnatural synthetic cell-wall cross-links, we primarily observed amino acid replacement at the fourth position, likely a result of d,d-carboxypeptidases removing any noncanonical d-AAs incorporated at the fifth position of the stem peptide (9, 17). However, in Gram-positive bacteria, l,d-transpeptidases seldom form cross-links, which limits the ability to incorporate noncanonical d-AAs at the fourth position (17, 18).

As an alternative approach, it was shown previously that knocking out the dacA gene, which encodes the primary d,d-carboxypeptidase PBP5, permits accumulation of fluorescent d-AAs into the fifth position of the peptide stems of B. subtilis (13). We reasoned that this same strategy might be used to incorporate noncanonical electrophilic d-AAs into the fifth position of the peptide stems of B. subtilis ΔdacA. These d-AAs might be appropriately positioned to covalently cross-link with the amidated m-DAP of an adjacent peptide stem resulting in unnatural synthetic 5,3-cross-links (Fig. 2A). To this end, we synthesized five d-AAs as potential structures for synthetic cross-link formation (9). Amino acids 1a, 1b, and 1c are fluorosulfates capable of reacting with amino and hydroxy groups of amino acids in close proximity, and 2a and 2b are somewhat more reactive vinyl sulfonamides (Fig. 2B) (19). Variants with different ring substitution patterns and sidechain lengths were assessed for efficiency of incorporation into cell walls. Previously we had shown that the sidechain length and the nature of the electrophile significantly affected the incorporation of the d-AA and the efficiency of synthetic cross-link formation. Additionally, the l-enantiomers of 1c (1d) and 1a (1e) were prepared as controls.

Synthetic Cross-Link Formation by Noncanonical d-AAs.

We first determined whether the noncanonical d-AAs are toxic to B. subtilis ΔdacA at elevated concentrations. The l-enantiomers of fluorosulfate and vinyl sulfonamide substituted phenylalanine have previously been incorporated into proteins in live bacteria with minimal effect on bacterial growth at 1 mM concentrations (19, 20). In E. coli, addition of the d-AAs and the cognate l-enantiomer to growth media resulted in identical growth curves with significant effects on growth rates beginning at concentrations above 8 mM (9). However, in B. subtilis ΔdacA, we observed increased detrimental effects on cell growth for the noncanonical d-AAs compared to the l-AAs starting at 4 mM (Fig. 2C and SI Appendix, Fig. S1 A–H). Compounds 1a, 1b, 1c, 1d, 1e, 2a, and 2b completely inhibit growth at 8 mM, 8 mM, 8 mM, >8 mM, >8 mM, >8 mM, >8 mM concentrations, respectively.

To determine if the noncanonical d-AAs are incorporated into the cell wall, we initially treated cultures of B. subtilis ΔdacA with 1 mM of each d-AA. Briefly, an overnight culture of B. subtilis ΔdacA was inoculated into Luria Bertani (LB) media supplemented with chloramphenicol and grown to optical density_600nm (OD₆₀₀) = 0.05. Subsequently, cultures were independently treated with 1 mM of each d-AA and grown to OD₆₀₀ = 1.10. Cells were then placed on ice, pelleted, and the cell wall of each culture was isolated following a modified version of the established methodology (9, 21, 22). We analyzed each sample by liquid-chromatography mass spectrometry (LC-MS). The concentration of each sample was normalized to the most abundant native non–cross-linked muropeptide, NAG-NAM-pentapeptide at a mass spectra signal of ∼10⁶ (SI Appendix, Fig. S2 A and B). Masses corresponding to the incorporation of noncanonical d-AAs into the fifth position of the peptide stems were detected for each compound and the corresponding mass spectra signal is provided: 1a (10⁶), 1b (10⁶), 1c (10⁶), 2a (10⁵), and 2b (10⁶) (Fig. 2E, 2G, and 2I, and SI Appendix, Fig. S2 C and D). No incorporation was detected for samples treated with the l-AA controls 1d and 1e.

Next, we analyzed the MS data for a mass corresponding to the synthetic cross-link. The expected structure for a synthetic 5,3-cross-link formed by 1c is shown in Fig. 2D. Mass spectral analysis confirmed the expected synthetic cell-wall 5,3-cross-links for all three fluorosulfates 1a, 1b, and 1c (a complete list of structures is provided in SI Appendix, Figs. S3–S8). The structure of the synthetic 5,3-cross-link formed by 1c was confirmed by LC-MS/MS (SI Appendix, Fig. S9). The synthetic cross-linked muropeptides formed by 1a, 1b, and 1c gave comparable mass spectra signals of ∼10⁶ and were ∼4-fold lower than the signal observed for the synthetic non–cross-linked muropeptides (Fig. 2 E–J). A synthetic cross-linked muropeptide was not detected for vinyl sulfonamides 2a or 2b, which previously formed synthetic cross-links in E. coli (9). Furthermore, we do not detect the reaction product of m-DAP and the free d-AA.

Notably, in our previous studies with E. coli, 1c was poorly incorporated and much less efficient at forming cross-links compared to 1a or 1b (9). This disparity in cross-linking efficiency is likely due to the spatial arrangement of the peptide stems, such that the extended 1c sidechain places the fluorosulfate in close proximity to the epsilon amino group of m-DAP in B. subtilis. Because noncanonical d-AAs are incorporated at the fourth position in E. coli, this extension likely worsens interactions with m-DAP of the adjacent peptide stem. An unexpected feature in our analysis is that the m-DAP donor strand of the primary d-AA cross-linked muropeptide is a tetrapeptide (Fig. 2D), while in the primary native cross-linked muropeptide this same strand is a pentapeptide (SI Appendix, Fig. S3). While the cell encodes other low-activity d,d-carboxypeptidases (23), this structural disparity was unexpected in the ΔdacA strain, and may point to key differences in cell-wall processing for the two muropeptides. We next focused on further characterizing the cross-links formed by d-AA 1c.

Quantification of Synthetic Cell-Wall Composition.

We evaluated the effect of varying concentrations of 1c on the density of synthetic cell-wall cross-links. Cultures were prepared as previously described and independently treated with 1, 2, and 4 mM of 1c. Additionally we prepared an untreated culture and a culture treated with compound 1d as controls. After cell-wall extraction and digestion, the samples were analyzed by high-performance liquid chromatography (HPLC). At the highest concentration tested (4 mM of 1c), the percentage of the most abundant noncanonical d-AA containing synthetic non–cross-linked species (SnC), an amidated NAG-NAM-tetrapeptide-1c, accounted for ∼19% of the total non–cross-linked peptidoglycan; the most abundant native non–cross-linked species (nC) was an amidated NAG-NAM-pentapeptide (Fig. 3 A and B). The primary d-AA containing synthetic cross-linked species (SC), the amidated NAG-NAM-tetrapeptide-1c-tetrapeptide-NAG-anhydroNAM (structure in Fig. 2D), accounted for 19% of the total cross-linked peptidoglycan; the most abundant native cross-linked species (C) was an amidated NAG-NAM-tetrapeptide-pentapeptide-NAG-NAM (Fig. 3 A and B). HPLC analysis showed a concentration-dependent increase in d-AA-containing SnC and SC cell-wall species (Fig. 3B and SI Appendix, Figs. S10–S12). These results indicate that B. subtilis survives in conditions (4 mM of 1c) where 19% of the cell wall has been replaced with noncanonical building blocks. At 2 mM of compound 1c, B. subtilis grows unperturbed and the cell wall contains 12% synthetic cross-links.

Fig. 3.

(A) HPLC trace (205 nm) of isolated peptidoglycan from B. subtilis treated with 1c (4 mM). The peaks corresponding to the most abundant muropeptide species for the native non–cross-linked (nC), native cross-linked (C), synthetic non–cross-linked (SnC), and synthetic cross-linked (SC) are labeled. SnC and SC correspond to the d-AA-modified structures. The black circles denote the amidated species, the gray circles denote the nonamidated non–cross-linked species, and gold circles denote the partially amidated and nonamidated cross-linked species. (B) The percentage of synthetic non–cross-linked and synthetic cross-linked muropeptides for cell wall isolated from B. subtilis after treatment with 1c. Note, percentages are based solely on the most abundant amidated native and synthetic species. A detailed quantification is provided in SI Appendix.

Bacterial Phenotype Imparted by d-AA Cross-Link.

We analyzed B. subtilis by scanning electron microscopy (SEM) for an observable alteration of cellular morphology as a result of the unnatural synthetic crosslinks. B. subtilis ΔdacA was grown to OD₆₀₀= 0.05 and treated with 1 mM, 2 mM, or 4 mM of 1c or 1d for 2 h; untreated wild-type and ΔdacA cells were used as controls. Cells were prepared for SEM analysis as described previously (9). B. subtilis wild-type cells show a characteristic linear rod shape with division at the midcell (Fig. 4A). In comparison, B. subtilis ΔdacA is slightly shorter, but otherwise shares a similar morphology (Fig. 4B). The observation that B. subtilis ΔdacA shows a shortened morphology was previously reported and literature suggests that the shortening of the cells becomes more profound as cells enter into stationary phase (24). In contrast, B. subtilis ΔdacA treated with 4 mM 1c undergo significant cell lysis, as shown in part, by the white arrows (Fig. 4 C and E). Cells that survive treatment appear elongated and spiraled (Fig. 4 C–E). For comparison, cells treated with the l-enantiomer 1d divide successfully and exhibit a cell length comparable to untreated cells (Fig. 4 F–H), however, compared to untreated cells they appear curled which may be due to an off-target interaction of the fluorosulfate with cellular components. This curl effect, when compounded by the elongation of cells caused by treatment with 1c, may contribute to the observed spiraling (Fig. 4 D and E). B. subtilis ΔdacA treated with 1 mM 1c or 1d show no significant change in morphology, while cells treated with 2 mM 1c or 1d show a slight curl, that is more pronounced in the 1c treatment. Additional SEM images at each compound treatment concentration are provided in SI Appendix, Fig. S13 A–F. Collectively, these data show that cells containing synthetic cell-wall cross-links elongate, and likely have an impaired ability to divide.

Fig. 4.

B. subtilis was cultured to OD₆₀₀= 0.05 and either not treated or treated with 4 mM 1c or 1d for 2 h. The bacteria were imaged by SEM and the results are shown. SEM of untreated bacteria (A) B. subtilis wild type and (B) B. subtilis ΔdacA display a linear rod shape. The ΔdacA mutation gives slightly shortened cells. Images of B. subtilis ΔdacA treated with 1c were captured at (C) 1,500× and (D, E) 15,000× magnification. Significant cell lysis is observed as indicated by white arrows. Bacteria that survive treatment by 1c are able to successfully elongate, but cell division appears impaired. Elongated cells display a spiraled phenotype. Images of B. subtilis ΔdacA treated with 1d were captured at (F) 1,500× and (G, H) 15,000× magnification. No cell lysis was observed for bacteria treated with 1d, although cells display a subtle curl. A 1-µM white scale bar is shown in A, B, D, E, G, and H (Top Right). A 5-µM white scale bar is shown in C and F (Top Right).

It has been documented that peptidoglycan at the septum during cell division is often denuded of peptide stems, which are clipped by periplasmic amidases at the lactyl group (25). Saccharides at the septum are directed and remodeled by the cell-division proteins of the divisome as well as by certain cell-wall remodeling enzymes that contain sporulation-related repeat (SPOR) domains, which recognize and bind to denuded glycan (25–27). Additionally, gene deletion of periplasmic cell-wall amidases produces an elongated cell phenotype (28). In our LC-MS analysis, we observe the reaction product of the primary native cross-link with cell-wall amidases, but not the reaction product of the primary synthetic cross-link with cell-wall amidases at either end (i.e., NAG-NAM or NAG-anhydroNAM). We hypothesize that the presence of synthetically modified cell wall interferes with the activity of periplasmic amidases and therefore cell-division–related SPOR-domain–containing enzymes, leading to impaired cell division and the observed phenotype at elevated concentrations of 1c. We do not know at this time why this observation is unique to our model Gram-positive bacterium B. subtilis and not found in our model Gram-negative bacterium E. coli, but hypothesize that it may be due to the substrate specificity and scope of the requisite enzymes involved in division.

A phenotypic comparison of the spiraling in our B. subtilis cells with that of Campylobacter is intriguing. Campylobacter helical shape is controlled by cell-shape–determining (Csd) proteins which often contain lysostaphinlike metalloprotease domains that act on peptidoglycan (29, 30). The exact function of these proteins is a current topic of study (31). Structural and in vitro analysis suggests that at least some of these enzymes are d,d-carboxypeptidases (cleave d-Ala from the fifth position of the peptide stem) and d,d-endopeptidases (cleave cross-links). These enzymatic activities would likely be impaired/modified in a bacterium with a synthetic cell wall. However, due to the difference in enzyme profile and general cell-wall architecture (Campylobacter is Gram-negative) between the organisms, we cannot at this time draw conclusions from the comparison.

Activity of Cell-Wall–Degrading Enzymes Affects Synthetic Cell-Wall Content.

Previously, we observed the formation of an anhydroNAM saccharide on the primary d-AA cross-linked muropeptide in E. coli (9). AnhydroNAM is a reaction product of lytic transglycosylases (LTs), enzymes that degrade the cell wall by nonhydrolytically cleaving the glycosidic bond between NAM and NAG to form an anhydroNAM reaction product (shown in Fig. 2D). LT activity is dominant in Gram-negative bacteria and in isolated cell-wall samples 3.7% of muropeptides contain anhydroNAM, which caps the termini of glycan strands after scission (32). Using LC-MS/MS, we mapped this modification to the saccharide affixed to the peptide stem that is not modified by the d-AA. We did not observe the formation of anhydroNAM on the primary d-AA–containing non–cross-linked muropeptide, nor the peptide stem of the cross-linked muropeptide that is modified by the d-AA. This observation suggests that LTs cannot form anhydroNAM caps on NAM saccharides adjoined to our d-AA-modified peptide stems. Interestingly, in the current study we observed an anhydroNAM modification at this same position (i.e., the saccharide affixed to the peptide stem that is not modified by the d-AA of the cross-linked muropeptide) in B. subtilis (Fig. 2D). Gram-positive bacteria predominantly rely on hydrolytic muramidases to cleave the glycosidic bond between NAM and NAG (33), and the corresponding reaction products of hydrolysis serve as the “caps” on the termini of their glycan strands (34). LT activity in Gram-positive bacteria is less abundant and in isolated native cell-wall samples only 0.4% of muropeptides contain an anhydroNAM (18).

In E. coli we had hypothesized that the noncleavable nature of the synthetic cross-link interferes with the exolytic processivity (disaccharide cleavage from a glycan terminus) of the LTs such that the glycosidic bond positioned at C4 of GlcNAc cannot be cleaved (Fig. 5 A, I-III). As LTs fail to degrade the cell wall of synthetic cross-linked muropeptides, an anhydroNAM imprint is imparted on the synthetic cross-links and the synthetic cell-wall content begins to accumulate as the rate of synthetic cell-wall formation outcompetes the rate of degradation (Fig. 5 A, IV). To assess whether a disruption in LT activity by synthetic cell wall leads to accumulation of synthetic muropeptides, we treated E. coli with 1 mM of 1a for 8 h. After each 2-h interval, we collected 50 mL of culture and performed LC-MS on the isolated muropeptide to determine the synthetic cell-wall content (Fig. 5 B and C). Synthetic cell-wall content increased significantly from 0 to 4-h posttreatment; after 6 h, the accumulation slows, which may be due to a homeostasis that is reached at late-stationary phase (Fig. 5 C and D). To confirm experimentally that the LT activity of E. coli is impaired by both d-AA–containing non–cross-linked and cross-linked muropeptides, we treated E. coli cells with 1 mM of noncanonical control compound O-methyl-d-tyrosine (structure in SI Appendix, Fig. S14A), which structurally resembles 1a, but is nonreactive. We again observed an accumulation pattern that resembles that of culture treated with 1a (SI Appendix, Fig. S14 B–G), demonstrating that LT activity is impaired by the d-AA–modified peptide stems, even if a synthetic cross-link is not formed.

Fig. 5.

(A) Simplified depiction of the proposed glycan degradation route of synthetically modified cell wall in the Gram-negative bacteria E. coli by glycan-cleaving enzymes. The noncanonical d-AAs are shown as yellow circles. The red lines indicate potential cut sites of the enzymes in each respective panel. (B) Bacteria growth curve of E. coli treated with compound 1a (1 mM). Gray circles indicate growth collection points, at which time growth was halted by pelleting and freezing of the bacterial culture. (C) HPLC trace (Abs. 205 nm) of E. coli bacterial cell wall at each collection point. (D) Quantification of SnC and SC muropeptides relative to the total muropeptide concentration of nC and SnC, and C and SC, respectively. (E) Simplified depiction of the proposed glycan degradation route of synthetically modified cell wall in the Gram-positive bacteria B. subtilis by glycan-cleaving enzymes. TAs not shown. (F) Bacteria growth curve of B. subtilis treated with compound 1a (1 mM) (G) HPLC trace (Abs. 205 nm) of B. subtilis bacterial cell wall. (H) Quantification of SnC and SC muropeptides of B. subtilis. A figure key for cell-wall structures is provided in Fig. 1.

In B. subtilis the LTs likely produce similar reaction products to those formed in E. coli (Fig. 5 E, I, II). However, hydrolytic muramidases in B. subtilis cleave and excise both native and synthetic cell wall, which should result in lower accumulation of synthetic cell wall than in E. coli where degradation is more impaired (Fig. 5 E, III, IV). HPLC analysis of cell-wall samples isolated from B. subtilis as a function of time shows a similar cell-wall content to E. coli at 2-h posttreatment. However, in B. subtilis, significant accumulation of synthetic cell wall past 2 h was not observed (Fig. 5 G and H). Hence, a homeostasis is quickly reached between incorporation and subsequent synthetic cross-link formation of noncanonical d-AAs, and excision of synthetic cell wall by hydrolytic muramidases.

Conclusion.

Herein, we showed that the cell wall of a Gram-positive rod-shaped bacterium can be modified with electrophilic noncanonical d-AAs that form nonenzyme catalyzed synthetic cell-wall cross-links to the extent of 19% of the total cell-wall cross-links in the live bacterium. In B. subtilis, a high level of synthetic cell-wall cross-links appears to interfere with septal cell-wall synthesis, but not side-wall cell-wall synthesis, affording the bacterium an elongated morphology. We have also begun to elucidate the enzymatic activities that are impaired by synthetic cell-wall cross-links and the resulting effects on cell-wall synthesis and degradation. The ability to alter the arrangement of synthetic cross-links in live bacteria by gene ablation of cell-wall biosynthetic enzymes allows for the study of highly diverse cell-wall structures. We have shown that nonnative synthetic 4,3-cross-links and 5,3-cross-links, the latter of which is an arrangement entirely absent in Nature, are accommodated within the cell wall of live bacteria at high levels. Additionally, it was recently shown that synthetic 5,5-cross-links can be formed in the cell walls of live bacteria by an azide-alkyne cycloaddition, although this reaction requires exogenous copper and a corresponding copper ligand (35). It is likely that more extensive modifications can be made to the cell-wall architecture using similar approaches.

Materials and Methods

Detailed experimental methods for the biological assays, bacterial growth curves, mass spectrometry analyses, synthetic cell-wall structures, HPLC analyses, SEM, and compound synthesis procedures and characterization data appear in SI Appendix. See SI Appendix, Figs. S1–S14.

Data Availability

All study data are included in the article and/or SI Appendix.