Chemical Biology Tools for examining the bacterial cell wall

Abstract

Bacteria surround themselves with cell walls to maintain cell rigidity and protect against environmental insults. Here we review chemical and biochemical techniques employed to study bacterial cell wall biogenesis. Recent advances including the ability to isolate critical intermediates, metabolic approaches for probe incorporation and isotopic labeling techniques have provided critical insight into the biochemistry of cell walls. Fundamental manuscripts that have used these techniques to discover cell wall interacting proteins, flippases and cell wall stoichiometry are discussed in detail. The review highlights that these powerful methods and techniques have exciting potential to identify and characterize new targets for antibiotic development.

Introduction:

The bacterial cell wall is arguably just as important to human health as it is to the survival of the bacterium. The complex polymers that comprise the cell wall provide bacteria with strength and a barrier to the outside world, allowing them to thrive in a multitude of environments, including the human body. Humans have taken advantage of natural product antibiotics, often produced by bacteria themselves, to target bacterial polymers, yielding some of the most widely used antibiotics to date (Chopra and Roberts, 2001). In this review, we focus on understanding the bacterial cell wall in the context of the threat that antibiotic resistance poses to society. With the announcement that many major pharmaceutical companies will no longer fund research programs in this critical area (Hu, 2018), the onus falls on curious, determined academics to identify new targets for antibiotics and new opportunities to combat resistance. Here we highlight recently reported chemical and biochemical approaches to study bacterial cell wall biosynthesis and maintenance both in the laboratory and in the clinic, focusing on the use of antibiotics. In some cases, antibiotics are part of the toolkit that enables biological insight. In other cases, the biological insights gleaned from applying the tools enable (or have the potential to enable) target discovery. Studies of the bacterial cell wall and the antibiotics that corrupt it are iterative and promote both mechanistic insights and translational applications.

The Target and its Basics.

Scientists have used small molecules to study bacteria since the 19^th century when Christian Gram treated cells with crystal violet and realized that bacteria could be canonically divided into two general classes: Gram-positive and Gram-negative (Gram, 1884). In the present day, sophisticated tools exist to visualize the cell wall and to dissect its composition and biosynthesis at the molecular level (Hsu et al., 2019; Kocaoglu and Carlson, 2016; Radkov et al., 2018; Siegrist et al., 2015; Taguchi et al., 2019a). Here we will review a subset of the new methods, but first offer a brief introduction to cell walls, noting the many recent, detailed reviews on these structures, e.g. (Radkov et al., 2018) and their biosynthesis and maintenance (Taguchi et al., 2019a).

The bacterial cell envelope is a complex structure that provides protection from the external environment, maintains cell shape, and provides resistance to chemical, physical, and biological damage (Figure 1A, (Vollmer et al., 2008)). Nearly all bacterial envelopes have a peptidoglycan (PG) cell wall layer lying just outside the plasma membrane. PG biosynthesis is a highly conserved process in bacteria, starting with UDP-N-acetyl-glucosamine (UDP-GlcNAc) conversion into UDP-N-acetyl-muramic acid (UDP-MurNAc), as the first committed step in PG synthesis (Figure 1B). The subsequent steps involve the addition of amino acids (commonly l-Ala, D-γ-Glu, l-Lys (Gram-positive) or meso-diaminopimelic acid (mDAP; Gram-negative and mycobacteria), and d-Ala- d-Ala) to the UDP-MurNAc lactate moiety. Variation exists within the stem peptide depending on species. For example, Mycobacterium leprae can utilize Gly in place of l-Ala at the one position (Draper et al., 1987); many Gram-positive bacteria and mycobacteria amidate the second position to post-biosynthetically generate D-γ-Gln; and spirochetes include l-Orn at the third position (Schleifer and Kandler, 1972; Vollmer et al., 2008). These and additional variations are highlighted in Vollmer et al., 2008. Stem peptide ligation is followed by the transfer of this PG building block to phosphate polyprenyl to make Lipid I. GlcNAc is added to Lipid I to make Lipid II as the final PG precursor. Lipid II is subsequently flipped across the membrane by the MurJ flippase (Ruiz, 2008; Sham et al., 2014) as the complete PG subunit. Class A Penicillin binding proteins (aPBPs), l,d-transpeptidases, and shape, elongation, division and sporulation (SEDS) proteins in complex with class B PBPs (bPBPs) assemble the PG through two different enzymatic reactions, transglycosylation and transpeptidation (Cho et al., 2016; Emami et al., 2017; Meeske et al., 2016; Taguchi et al., 2019b). PG transglycosylases (TG) link the sugar backbone of the PG subunit to the next unit through a ß-1,4 glycosidic linkage (Vollmer et al., 2008). PG transpeptidases (TP) most commonly connect the fourth amino acid of one peptide chain to the third amino acid (such as mDAP or l-Lys) of an adjacent strand yielding 3-4 cross-linked PG (Vollmer and Seligman, 2010). However, there are more variations that feature diverse connectivity—3-3, 2-4, and 1-3 linkages—and bridge lengths across species. While the biosynthesis of Lipid II is generally conserved across bacteria, the polymerization of the monomer marks the beginning of the diversification process (Egan et al., 2020; Typas et al., 2012). Depending on the activity and the protein-protein interactors of the TGs/TPs, a variety of shapes are formed (Do et al., 2020; Salama, 2020). For example, in H. Pylori, the TG and TP activity is spatially and temporarily regulated along the cell wall, which greatly influences the shape. This diversity of Gram-positive and Gram-negative cells is further enhanced by the inclusion of lipids in the cell wall.

Figure 1.

A. Structural features of the cell wall of Gram-positives, Gram-negatives, and mycobacteria. Abbreviations: PG – peptidoglycan, PM – plasma membrane, LPS – lipopolysaccharide, MM – mycomembrane, OM – outer membrane, AG – arabinogalactan. B. Biosynthetic steps of peptidoglycan (PG): PG biosynthesis occurs in 14 unique biochemical steps, starting with UDP-GlcNAc conversion into UDP-MurNAc; UDP-MurNAc is transferred to the lipid carrier and, subsequently, glycosylated by MurG. This process culminates in Lipid II being flipped across the membrane by MurJ. Transglycosylases (TG) and transpeptidases (TP) incorporate the PG subunits into the growing PG polymer. Ultimately, peptidoglycan is altered through the addition of small and large molecules post synthetically that are not represented here, such as wall teichoic acids (WTA).

In Gram-positive bacteria the PG layer is significantly thicker than that of Gram-negative (Vollmer and Seligman, 2010) and features lipoteichoic and wall teichoic acids (WTA) anchored to the cytoplasmic membrane and the MurNAc 6-OH, respectively. The WTA is assembled largely on the cytoplasmic face. Upon delivery to the extracellular surface, it is anchored to PG by the family of LytR-CpsA-Psr (LCP) enzymes (Kawai et al., 2011; Li et al., 2020; Schaefer et al., 2017). Teichoic acid (TA) elements are essential to the virulence of pathogenic bacteria. They permit bacterial cell adhesion to host cells, in addition to controlling cell wall remodeling by autolysins (Brown et al., 2013).

Gram-negative bacteria have a thinner PG but have additional structural support and protection due to the asymmetric outer membrane consisting of phospholipids and lipopolysaccharides (LPS) (Rojas et al., 2018). LPS is made of three components: lipid A, an oligosaccharide core, and the O-antigen. Towards the end of the lipid A construction, the core oligosaccharide is added, making the lipid oligosaccharide intermediate. The O-antigen is synthesized separately and bound to lipooligosaccharide (LOS), to make LPS, before transport to the outer membrane (Simpson and Trent, 2019). LPS is not anchored to the PG but rather inserted into the outer membrane by complex cellular machinery and hydrophobically adhered via Lipid A.

There is a third class of bacteria that has elements of both Gram-positives and Gram-negatives; it is sometimes referred to as Gram-indeterminate. M. tuberculosis and other members of the Corynebacterineae suborder are phylogenetically related to Gram-positive bacteria, but possess a waxy, relatively impermeable outer membrane-like structure (Figure 1A, (Alderwick et al., 2015)). The core cell wall has a PG layer of intermediate thickness that is covalently bound to the branched arabinogalactan (AG). The AG layer acts as a scaffold for a covalently bound, inner leaflet of mycolic acids. The outer leaflet possesses more mycolic acids, along with intercalated glycolipids such as trehalose monomycolates (TMMs) and trehalose dimycolate (TDM) (Dulberger et al., 2020). The three classes of bacteria discussed here contain some similarities in the composition of their cell wall (i.e. Lipid II, Figure 1B) and differences (i.e. lipid modifications). It has been proposed that some of the lipid modifications that help to differentiate Gram-positive, Gram-negative, and Gram-indeterminate bacteria could provide mechanisms to uniquely target specific bacteria with antibiotic therapy (Jackson et al., 2013; Kuhn, 2019).

Bacterial cell walls have historically made an excellent target for antibiotics. There are multiple ways to kill a bacterial cell, inhibiting DNA replication and halting protein synthesis (Walsh, 2003). However, the bacterial cell wall is an especially attractive target because it is unique to bacteria, meaning that human cells do not contain the biochemical machinery that is required to build and maintain it. In addition, the biosynthesis is largely conserved (Figure 1B), allowing the development of broad-spectrum antibiotics. Unfortunately, bacteria have developed resistance to nearly every cell wall targeting antibiotic, including the well-known ß-lactams, such as penicillin/methicillin, and the antibiotics of “last resort”, the glycopeptides (Kahne et al., 2005). Modes of resistance can be conferred by, but not limited to, the expression of insensitive PBPs, production of ß-lactamases, activity of efflux pumps, and alterations to the cell wall (Walsh, 2000). For a more extensive review of these and other modes of resistance, the authors refer the reader to Walsh et al., 2000. In recent years, there has been an emergence of multi-drug resistant bacteria, particularly in the ESKAPE pathogens (Enterococcus faecium, Staphylococcus aureus, Klebsiella pneumoniae, Acinetobacter baumanii, Pseudomonas aeruginosa, and Enterobacter species) (Boucher et al., 2009; Santajit and Indrawattana, 2016). These bacteria, together with tuberculosis (Bloom and Murray, 1992; Porter and McAdam, 1994) and gonorrhea (Unemo and Shafer, 2014), pose a viable threat with a lack of new antimicrobial compounds (Prevention). Within one year after the first clinical use of a natural or synthetic antibiotic, resistance can develop (Walsh, 2003). Therefore, there is a continuous demand for the development of novel antibiotics via rational design, high throughput screening efforts, and medicinal chemistry campaigns.

In order to identify new bactericidal, cell wall-acting compounds, peptidoglycan biosynthetic pathways (Figure 1) that are less understood and not currently targeted by mainstream antibiotics must be interrogated with appropriate tools. Fluorescent probes have permitted the close study of bacterial structures, and more recently probes such as fluorescent d-amino acids (FDAAs) have enabled methods to screen for novel antibiotics and to identify their mode of action (Culp et al., 2020). There are many excellent reviews that highlight direct imaging of bacterial cell walls to study the structure, biosynthesis, and dynamics of this polymer (Kocaoglu and Carlson, 2016; Radkov et al., 2018; Siegrist et al., 2015; Taguchi et al., 2019a). Here we focus on recent work that merges chemical and biochemical methods including immunoblotting, photo-crosslinking, spectroscopy, and radiolabeling to probe PG biosynthetic pathways and composition that can help inform the development of novel antimicrobial therapeutics.

Using Antibiotics to Isolate Cell Wall Intermediates.

In order to study bacterial cell wall biosynthetic enzymes, one must have access to PG precursor substrates. A common strategy is to use bacterial cell wall inhibitors such as vancomycin and moenomycin to cause a buildup of intermediates. Such a tactic was used by Strominger and Park in the 1950s to identify the UDP-MurNAc intermediate, known as Park’s Nucleotide, using penicillin (Figure 1B) (Park, 1952; Strominger et al., 1959). The lipid-linked precursor, Lipid II, was long-sought in PG biosynthesis (Figure 1B) as it is difficult to synthesize, and its scarcity made in vitro studies of PG polymerization nearly impossible (Lazar and Walker, 2002; Ye et al., 2001). Lipid II is the result of the ß-1,4 glycosylation of Lipid I by MurG with the monosaccharide unit, GlcNAc (Figure 1B). This event produces a disaccharide pentapeptide containing an undecaprenyl pyrophosphate group. A major feat from the Walker Laboratory in 1999 showed that Lipid I derivatives could be transformed with the glycosyltransferase, MurG, to form native Lipid II as well as a variety of derivatives (Ha et al., 1999; Men et al., 1998). This in vitro enzymatic conversion provided the first reliable method to produce workable quantities of Lipid II. Access to Lipid II, in turn, enabled the mechanism of various glycopeptide antibiotics to be studied. For example, vancomycin, as well as other glycopeptides such as dalbavancin (Leimkuhler et al., 2005), has been shown to inhibit TGs directly (Chen et al., 2003).

In 2017, a method to accumulate and isolate Lipid II directly from cells was developed (Figure 2A) (Qiao et al., 2017). Qiao, Kahne, and Walker developed a two-step extraction protocol of Lipid II. Using a similar strategy as Strominger in 1958 and armed with the knowledge that PG TGs are responsible for the last steps of PG biosynthesis (Figure 1B), antibiotics that target these later steps were used to isolate native Lipid II (Figure 2A). Cells treated with moenomycin or vancomycin, which inhibit late-stage biosynthetic enzymes in the outer leaflet of the plasma membrane (Chen et al., 2003; Kahne et al., 2005; Taylor et al., 2006), accumulate a significant amount of Lipid II (Figure 2A). However, when cells were treated with a sublethal dose of fosfomycin, an antibiotic that inhibits the first committed step in PG biosynthesis (Falagas et al., 2016), Lipid II was undetectable (Qiao et al., 2017). The ability to alter the amounts of PG biosynthetic intermediates is a powerful tool for studying both upstream and downstream effects of peptidoglycan synthesis, as evidenced by experiments that have used the Lipid II isolation method to screen for novel antibiotics, assess the function of critical PG biosynthetic enzymes and identify new enzymes involved in PG biosynthesis (as discussed below).

Figure 2.

A. Two-step extraction method for isolating Lipid II from bacterial cultures. B. PBP4 transpeptidase mediated terminal d-Ala exchange with unnatural amino acids. Depicted is the incorporation of Biotin-d-Lys (BDL) to the stem peptide of Lipid II. Lipid II consists of a diphosphate (P) disaccharide backbone, GlcNAc (G) and MurNAc (M), with a pentapeptide chain: Alanine (A), Glutamate (E), and Lysine (K). C. Substituted cysteine accessibility method (SCAM) utilizes single cysteine mutations (orange circle) in a protein of interest in conjunction with two cysteine reactive reagents: MTSES and NEM. MTSES cannot penetrate the membrane and will only react with periplasmic cysteines. NEM can penetrate the membrane and react with both periplasmic and cytoplasmic cysteine residues. Using this method with a cysteine mutant library, it can provide topological information of a protein, in this case a transmembrane protein. D. MurJ-pBPA photocrosslinking assay for detection of Lipid II/MurJ adducts. MurJ encoded with single pBPA mutations undergoes crosslinking with Lipid II upon UV activation. After SDS-PAGE and electroblotting, crosslinked Lipid II is biotinylated in-gel via a BDL exchange reaction and then detected via blotting with streptavidin-HRP. E. Subsequent lysis and a click reaction to attach a fluorophore allow for analysis of mycolate-protein interactions via metabolic incorporation of a bifunctional TMM analogue. N-x-AlkTMM-C15 is metabolically incorporated into the mycobacterial mycomembrane (MM). Covalent crosslinks with MM-associated proteins are induced by UV activation. Subsequent lysis and a click reaction to attach a fluorophore allows for analysis by a variety of techniques.

To interrogate the production of Lipid II in a cellular context, the Walker and Kahne labs have developed an easily accessible technique to label Lipid II from bacterial cultures (Fig 2B) (Qiao et al., 2014). The method involves detection of Lipid II from growing bacterial cells by treating the intermediate with Penicillin Binding Protein 4 (PBP4) to allow for the installation of a biotin tag. This low molecular weight PBP isolated from S. aureus is a promiscuous transpeptidase that, in addition to catalyzing bonds between adjacent muropeptides, can exchange the terminal d-Ala of the muropeptide with various d-amino acids. In vitro, this technique can switch the terminal d-Ala for biotinylated d-Lys (BDL); biotinylated Lipid II is then detectable via Western blot (Figure 2B). This technique provides researchers with a streamlined process to obtain functionalized Lipid II analogues that can be selectively detected, thus, generating a valuable assay for studying PG biosynthetic machinery and screening antibiotics (Cochrane and Lohans, 2020).

In sum, there are at least three methods to access Lipid II and its derivatives have been reported: in vitro biochemical methods using MurG, isolation of Lipid II from bacterial cultures, and Lipid II biotinylation. Collectively, these methods have been applied to diverse bacterial species; S. aureus, Bacillus subtilis, Escherichia coli (Qiao et al., 2017), and M. smegmatis (Garcia-Heredia et al., 2018) have all been successfully used for the Lipid II isolation and/or visualization, despite variation in PG pentapeptide chains (Qiao et al., 2017). Access to these substrates will be important to fully understand the structural diversity present in bacterial cell walls and to characterize pools of PG intermediates during growth. These methods have been useful in studying the mechanism of action of antibiotics, as mentioned above with glycopeptides and more recently with lysobactin (Lee et al., 2016). Research groups have also used them to study the mechanisms of enzymes involved in PG biosynthesis. For example, multiple labs have used the tools to biochemically characterize the TG activity of the SEDS proteins, which have been proposed as new potential targets for antibiotics (Cho et al., 2016; Emami et al., 2017; Meeske et al., 2016; Rohs et al., 2018; Sjodt et al., 2018; Taguchi et al., 2019b). The methods of isolation and detection (Figure 2A/B) have been used in combination in vitro to study the order of addition of WTA precursors to PG intermediates (Figure 3B. This work will be discussed in detail below (see Radiolabeled Substrates). Finally, these methods have been used to identify and study the mechanism of Lipid II transport (Rubino et al., 2020) as discussed in detail in the next section.

Figure 3.

A. Cell wall depictions of S. aureus strains utilized to determine characteristic peaks of PG and TA. B. Assessing the ability of un-cross-linked and cross-linked PG as a substrate for WTA transfer by LcpB. Un-cross-linked PG oligomers featuring a pentaglycine chain are bioenzymatically prepared with a SgtB mutant. The oligomers are then either modified with WTA via LcpB (top) or cross-linked with PBP4 (bottom). Subsequently, these molecules are either subjected to transpeptidation with PBP4 or a ligation reaction with LcpB.

Biochemical Tools to Investigate the PG Enzymes.

For a long time, it was not understood how Lipid II translocates across the lipid bilayer into the periplasmic space for incorporation into the PG meshwork. Multiple proteins were evoked to perform this function, including the SEDS protein FtsW, which was later shown to be a transglycosylase (Egan et al., 2020; Ruiz, 2015; Young, 2014). MurJ was discovered by Ruiz and colleagues to be the elusive flippase (Ruiz, 2008; Sham et al., 2014). In a series of elegant experiments utilizing substituted cysteine accessibility method (SCAM) (Figure 2C) and the protein toxin ColM (Touzé et al., 2012), Ruiz and collaborators were able to create an assay that provided context-dependent monitoring of Lipid II movement. Using an extensive single Cys mutant MurJ library, the authors mapped the topology of MurJ, which was then used in parallel with protein modelling to predict possible dynamic conformations (Butler et al., 2013). Building on these findings, Sham et al. used an in vivo assay with the addition of sulfonating reagents, ColM, or both in either wild-type MurJ or the reversible MurJ mutant to formalize MurJ as the Lipid II flippase (Sham et al., 2014).

With the knowledge that MurJ was the flippase, the biochemical details of its activity were investigated using the Lipid II assays discussed above. MurJ proved challenging to study because, as a flippase, it does not structurally alter the Lipid II precursor when it transports. However, by using the Lipid II detection method (Figure 2B), it was possible to monitor changes in Lipid II’s movement. In 2018, Rubino et al. showed that treatment with a protonophore to disperse the proton-motive force caused Lipid II accumulation in E. coli, suggesting that the transportation of Lipid II is coupled to an electrochemical gradient. The effect of the protonophore mimicked controls known to disrupt MurJ activity. Furthermore, they characterized the conformation of MurJ when the membrane potential is dissipated by probing individual cysteine residues in MurJ (Rubino et al., 2018).

To determine what residues of MurJ are involved in transport, photocrosslinking experiments were used to tether interacting partners. Photocrosslinking has become an invaluable method to determine protein-substrate interactions (Lancia et al., 2014; Parker and Pratt, 2020; Wu and Kohler, 2019). Unnatural amino acid incorporation can be used to install a photoactivatable p-benzoyl-l-phenylalanine (pBPA) in the protein of interest, in this case, MurJ. Excitation of pBPA using 350-365 nm light leads to a reactive diradical that forms a C-H bond to any vicinal functional group within a 3.1 Å reactivity radius (Lancia et al., 2014). In 2012, Okuda et al. utilized this method to determine specific sites where LPS interacts with the LPS transport (Lpt) machinery (Okuda et al., 2012). Briefly, they incorporated pBPA residues in several Lpt transport proteins. After UV activation, photo-crosslinked adducts were identified using immunoblotting with LPS specific antibodies. This method allowed them to take chemical snapshots of LPS transport and determine that shuttling of LPS across the periplasm is accomplished through cytoplasmic ATP hydrolysis. The Ruiz group, in collaboration with the Kahne lab, applied this methodology to probe MurJ flippase activity (Figure 2D) (Rubino et al., 2020). They hypothesized that pBPA incorporated into MurJ would prompt crosslinking to its natural substrate Lipid II. However, the lack of antibodies for Lipid II made it necessary to implement a method that allowed for the detection of the crosslinked adduct. PBP4 was used to incorporate a BDL after photocrosslinking Lipid II to MurJ-pBPA mutants to allow for detection (Figure 2D). They applied this protocol to investigate the role of three essential arginine residues, located in the central cavity of MurJ, that were previously proposed to be key in recognizing the pyrophosphate of Lipid II (Kuk et al., 2019). Using single and multiple Arg→Ala mutants, they observed similar levels of crosslinking compared to wild-type MurJ. However, these mutants displayed impaired ability to flip Lipid II. This methodology permitted the observation of the intermediate transport steps in living cells and provided direct, biochemical evidence that the conserved arginine residues control Lipid II movement through MurJ. Thus through the combination of Lipid II chemical probes, genetic tools, and biochemical conversions, the function of MurJ has been identified and the biochemical mechanisms of Lipid II transport are rapidly being unveiled. This also highlights MurJ as an exciting target for antibiotic development. The ability to track Lipid II in cellular biochemistry assays was critical because it yielded detailed biochemical information (i.e. protein residues, membrane potential) that would not have been possible with other methods.

Metabolic incorporation of photocrosslinking sugars.

The proteins that bind the cell wall and its associated glycoconjugates are also potential antibiotic targets. In contrast to protein-mediated interactions, glycan recognition events are often weak and short-lived. Additionally, glycans are not directly genetically encoded and their biosynthesis is complex, so it is challenging to use standard genetic engineering methods to tag them. The incorporation of functionalized metabolites has allowed a way to bypass these challenges (Campbell et al., 2007).

Several groups have introduced unnatural sugars containing photoactivatable crosslinkers to the cell by hijacking carbohydrate metabolic pathways and capitalizing on enzyme promiscuity (Tanaka and Kohler, 2008; Yu et al., 2012). The mycomembrane (Figure 1A – “MM”) is attached to the mycobacterial cell wall via AG and is a barrier to environmental, immune, and antibiotic insults. However, its protein composition has eluded classic biochemical techniques for a long time, in part because of the difficulty of cleanly separating the covalently bound mycomembrane from other layers of the complex mycobacterial envelope. Kavunja et al. recently developed the first photocrosslinking probes for the mycomembrane to analyze mycolate-protein interactions in vivo (Figure 2E) (Kavunja et al., 2020). They synthesized a TMM analogue that specifically incorporates into the TDM portion of the mycomembrane via previously reported conserved, substrate-promiscuous Ag85 mycoloyltransferases (Fiolek et al., 2019). This analogue contains a bifunctional linker bearing a photoactivatable diazirine group and a clickable alkyne handle. After metabolic incorporation into the cell surface, mycobacteria were irradiated with UV light. The diazirine cross-linked with neighboring proteins that were then enriched following click ligation to an azide-fluorophore-biotin. This method allowed for the identification of both known and previously-undetectable mycomembrane-resident proteins, as well as tracking them by in-gel fluorescence. Similar techniques are likely to be useful in future studies to interrogate different layers of the cell envelope, including PG; especially as DeMeester and coworkers have shown that PG precursors containing the diazirine cross-linkers at the C2 position of MurNAc are accepted by the PG biosynthetic enzymes (DeMeester et al., 2018). Previously, Sakar et al. exploited the MurF ligation process to insert a d-Ala-that was biofunctionalized with an alkyne and photocrosslinking handles into Lipid II to mine the protein interacting partners (Sarkar et al., 2016). Targeted acquisition of cell envelope interactomes may reveal new potential targets for antibiotic therapies (Kavunja et al., 2020).

Spectroscopic Methods to Study the Cell Wall.

The stable isotopes ¹³C and ¹⁵N are effective as probes to elucidate structural features of the bacterial cell wall at the molecular level using spectroscopic methods (Kim et al., 2015; Kim et al., 2014; Yang et al., 2017). In this strategy, the macromolecules of biological interest are unperturbed and uniform enrichment enhances signal output of the NMR spectrum in a non-destructive manner (Nygaard et al., 2015). Romaniuk and Cegelski utilized ¹³C and ¹⁵N cross polarization magic angle spinning (CP/MAS) solid-state NMR (ssNMR) to characterize the composition of uniformly-labeled PG and WTA in S. aureus (Romaniuk and Cegelski, 2018). This technique affords an alternative to solution-based analytical methods that do not permit full characterization of the highly insoluble material. Using this approach, spectra of purified PG and WTA isolates from wild-type and ΔtarO (a mutant that is unable to synthesize WTA), respectively, were first used to identify and quantify characteristic carbon peaks from each component (Figure 3A). Since the sum of the two spectra reproduced the peak intensities of the intact cell wall sample, they were able to quickly determine the composition ratio of TA to PG. The masses for TA and PG calculated by this method were consistent with those determined by phosphate analysis, a more traditional but labor intensive method. Subsequently, the relative composition ratios of the two components in both the stationary and the exponential phases were determined. In the stationary phase PG thickness increased while WTA decreased. This finding was confirmed with selective labeling using either d-[¹⁵N]Ala (WTA) and [¹⁵N]Gly (PG) in ¹⁵N CP/MAS. With these baselines in hand they were able to validate that this analytical method is amenable to determining the composition levels of TA and PG of cell walls with the antibiotic tunicamycin, which inhibits WTA growth. Cegelski et al. have used similar strategies with ssNMR to establish that vancomycin primarily targets transglycosylation over transpeptidation using uniformly ¹³C and ¹⁵N labeled amino acids in S. aureus (Cegelski et al., 2002). Changes to the d-alanine-pentaglycyl bridge-links were unperturbed in the presence of vancomycin, suggesting that transpeptidation is unaffected. Instead, this supports the idea that vancomycin blocks transglycosylation and impedes translocation of Lipid II into the periplasm. Rotational echo double resonance (REDOR) and CP/MAS ssNMR have also been used to discern perturbations to PG and WTA of S. aureus treated with the cyclic decapeptide amphomycin, a drug that is effective against superbugs such as multi-drug-resistant S. aureus and vancomycin-resistant enterococci by targeting bactoprenol-phosphate. Using REDOR, Singh et. al observed ¹⁵N shifts in bridge-links between Gly and l-Lys and the free side chain amine of lysine indicating that the compound induced PG thinning, the accumulation of Park’s nucleotide, and a decrease in alanylation of WTA (Singh et al., 2016). These data suggested that amphomycin acts on the cell wall prior to transglycosylation. Overall, CP/MAS and REDOR analysis of the bacterial cell wall permits rapid assessment of whole cell composition and can be applied to monitor and determine the compositional perturbations caused by antibiotics as showcased in the studies above.

In another example, Calabretta et al. employed ¹³C radiolabeled lipid-linked arabinofuranose donors to study Gram-indeterminate bacteria. These probes were biosynthetically incorporated into the arabinan layer of Corynebacterium glutamicum and M. smegmatis strains unable to produce this polymer due to genetic mutation or treatment with benzothiazinone antibiotics (Calabretta et al., 2019). This procedure circumvents the requirement of metabolic processing of cell wall probes prior to incorporation into the cell wall. Analysis of the soluble arabinan isolated from these models retained characteristic peaks with 2D NMR experiments (¹H-¹³C HSQC, ¹H-¹³C HMBC, and ¹H-¹³C HSQC-TOCSY). Though this study did not use intact cells to gain structural details, it yielded a powerful tool that can be applied to examine structural changes in the lesser-understood mycobacterial cell envelope. Antibiotics are critically needed against mycobacteria to combat public health threats such as tuberculosis, which the Center of Disease Control (CDC) has recently identified as a top AMR threat (Bloom and Murray, 1992; Prevention).

Radiolabeling.

Although NMR methods are useful in determining composition of the cell wall, they do not inform on its biosynthesis. This can be achieved using radioisotopes, as bacterial incorporation of extremely sensitive ionizing radionuclides permits submicromolar (high fM to pM) detection of cell wall products and intermediates of biosynthetic pathways in vivo. This sensitivity is also useful in translational applications for disease diagnosis. Here we discuss the use of ¹⁴C and ¹¹C radionuclides in bacterial cell surface structures to discern mechanistic details of PG synthesis and structure (long-lived) in addition to usage as a diagnostic imaging tool of bacterial infection in vivo (short-lived).

Long-Lived Radioisotopes.

Elucidating some of the lesser-understood biosynthetic processes has the potential to provide new targets for antimicrobials. TAs are important for the structure and virulence of Gram-positive bacteria (Brown et al., 2013), but far less is known about their synthesis compared to that of PG. Recently, Schaefer et. al. investigated the order and process by which WTA are appended to PG. This is useful, as methicillin-resistant S. aureus (MRSA) is resensitized to ß-lactams when WTA biosynthesis is inhibited (Campbell et al., 2011; Farha et al., 2013). They found that WTA is ligated to un-crosslinked PG oligomers and that ligation preference is due to steric restrictions of the LcpB WTA ligase binding site (Schaefer et al., 2018). With isolated Lipid II from S. aureus, featuring a pentaglycine chain for downstream cross-linking, the group bioenzymatically prepared un-crosslinked PG fragments (2-10 carbohydrates in length) with a transglycosylase mutant (SgtB^Y181D) that releases premature PG oligomers (Figure 3B). These fragments were then subjected to transpeptidation with PBP4 followed by treatment with LcpB (WTA ligase) and ¹⁴C radiolabeled LII_A^WTA, a truncated radiolabeled WTA precursor, or just the latter to assess substrate preference. Detection by polyacrylamide gel electrophoresis (PAGE) autoradiography showed that the cross-linked substrate is unable to be ligated to WTA, whereas the un-cross-linked PG fragments serve as an acceptable substrate for the transfer of WTA by LcpB. Additional experiments evaluating the structural requirements for recognition co-crystalized the homologue TagT from B. subtilis with either LI_A^WTA and LII_A^WTA, which indicated a putative binding groove narrow in size. The steric restrictions of this site prompted Schaefer and coworkers to explore if the stem peptide was a necessary feature of recognition by employing chitin, deacetylated chitin, and cellulose-based oligosaccharides as probes of LcpB and TagT. Results indicated that the stem peptide is not a necessary structural feature for recognition of the transfer substrate; however, acetylation of the C2 position of MurNAc is required. These radioisotopic probes enabled Schaefer and coworkers to visually determine if PG intermediates were modified by WTA ligase in this enzymatic assay and suggest the ligation order of WTA precursors to peptidoglycan intermediates. With this fundamental information, new antibiotic targets can be established as a means to stymie the ligation process and ultimately disrupt the integrity of the cell wall of Gram-positive bacteria.

Short-Lived Radioisotopes.

Positron emission tomography (PET) radioisotopes such as ¹¹C, ¹³N, ¹⁵O, and ¹⁸F are extremely sensitive tools that permit whole-body imaging of physiological processes in deep-lying tissues and organs with low nanogram quantities of probe. Unlike stable isotopes, the radiation produced upon decay is detectable outside of the human body. [¹⁸F]-Fluorine deoxyglucose (FDG) is a clinical PET probe frequently used to image cells with higher energy requirements, such as oncogenic cells and sites of inflammation (Zhuang and Alavi, 2002). [¹⁸F]FDG accumulates because it cannot be further metabolized after sequestration into the cell. Since this probe is not bacteria-specific, improved imaging technologies are required to detect active bacterial infections in vivo (Ordonez and Jain, 2018). To address this issue, Rosenberg, Ohliger, and Wilson have developed clinically-relevant ¹¹C radiotracers that can distinguish between sterile inflammation and active infection by metabolically incorporating into both clinically-relevant pathogenic Gram-positive and Gram-negative bacteria (Neumann et al., 2017; Parker et al., 2020; Stewart et al., 2020), utilizing d-amino acid probe incorporation (de Pedro et al., 1997). High incorporation of d-[¹⁴C]Met in both S. aureus and E. coli in the stem peptide of PG led to the development of a ¹¹C-synthetic probe from a d-homocysteinethiolactone precursor (Figure 4). Earlier experimentation had shown that exogenous d-Met is amply integrated into the PG polymer of stationary bacteria (Caparrós et al., 1992; Lam et al., 2009). Moreover, a robust synthetic strategy and efficient labeling probe for d-Ala and d-Glu were not readily available. Using the l-[¹¹C]-Met tracer for imaging in mice modeling myositis, they were able to detect and differentiate between bacterial infection and sterile inflammation, modeled by heat-killed bacteria, in the deltoid muscle with great sensitivity. These probes are likely to be especially useful for regions of the body that are normally sterile such as the musculoskeletal, biliary, and nervous systems, as there will not be significant background from the commensals. The synthetic efficiency of the probe was later improved for use in the clinical setting by using automated synthesis (Neumann et al., 2017; Parker et al., 2020; Stewart et al., 2020). Most recently, Parker et al. expanded the tracers to include amino acids canonically incorporated into the stem peptide, d-[3-¹¹C]-Ala and d-[3-¹¹C]-Ala-d-Ala probes (Figure 4). In vivo studies utilized the d-Ala probe, since d-Ala showed anywhere from 2-3 times greater uptake in E. coli and S. aureus over the dipeptide. Moreover, it was determined that the d-Ala probe generated high levels of incorporation with the previous panel of bacteria tested in the 2019 paper and that these probes were highly selective for bacterial cells over mammalian cells. Infectious imaging challenge experiments also confirmed the ability of the d-Ala probe to distinguish between live bacterial infection and sterile inflammation in mouse models. The tracer was able to monitor cell wall-acting antibiotic (ampicillin) treatment of an E. coli bacterial infection, image infection in the intervertebral space, and proficiently screen for pneumonia in mice. Thus, this class of radiolabeled probes yielded an indispensable bacteria-specific imaging strategy for pathogenic bacteria in difficult-to-access physiological niches in clinical settings comapred to currently approved methods.

Figure 4.

Incorporation of radiolabeled probes, d-[methyl-¹¹C]-Met or d-[3-¹¹C]-Ala, post-biosynthetically via transpeptidases in clinically-relevant bacteria to assess use for in vivo labeling of active bacterial infections.

Conclusion.

Biochemists have been using small molecules to study bacterial cell walls since the 1800s. As improved probes have become available, the ability to study the biochemical phenomena has exploded. Here we have highlighted examples of chemical probes that were used to reveal fundamental cell wall biochemistry or, in some cases, to identify bacterial pathology in a living host. There is a rich intersection between chemical biology and cell walls. It is this intersection that will be powerful in tackling the problem of antibiotic resistance, as a means to identify new targets and to characterize the mechanism of action for promising compounds. We highlighted the recent advances in the field such as monitoring PG precursors to elucidate MurJ, TG and TP activity, establishing new protein targets for antibiotic development through photocrosslinking techniques, applying NMR spectroscopy to capture native cell wall composition, and demonstrating a highly sensitive method of detection of cellular processes using radiolabels in vitro and in vivo. In conjunction with current visualization and imaging techniques, chemical, biochemical, and hybrid methods have the potential to provide both mechanistic and structural information for valuable targets.