Constructing and deconstructing the bacterial cell wall

Abstract

The history of modern medicine cannot be written apart from the history of the antibiotics. Antibiotics are cytotoxic secondary metabolites that are isolated from Nature. The antibacterial antibiotics disproportionately target bacterial protein structure that is distinct from eukaryotic protein structure, notably within the ribosome and within the pathways for bacterial cell‐wall biosynthesis (for which there is not a eukaryotic counterpart). This review focuses on a pre‐eminent class of antibiotics—the β‐lactams, exemplified by the penicillins and cephalosporins—from the perspective of the evolving mechanisms for bacterial resistance. The mechanism of action of the β‐lactams is bacterial cell‐wall destruction. In the monoderm (single membrane, Gram‐positive staining) pathogen Staphylococcus aureus the dominant resistance mechanism is expression of a β‐lactam‐unreactive transpeptidase enzyme that functions in cell‐wall construction. In the diderm (dual membrane, Gram‐negative staining) pathogen Pseudomonas aeruginosa a dominant resistance mechanism (among several) is expression of a hydrolytic enzyme that destroys the critical β‐lactam ring of the antibiotic. The key sensing mechanism used by P. aeruginosa is monitoring the molecular difference between cell‐wall construction and cell‐wall deconstruction. In both bacteria, the resistance pathways are manifested only when the bacteria detect the presence of β‐lactams. This review summarizes how the β‐lactams are sensed and how the resistance mechanisms are manifested, with the expectation that preventing these processes will be critical to future chemotherapeutic control of multidrug resistant bacteria.

1. INTRODUCTION

Bacteria are resilient. Within their single cell their DNA encodes a mere thousands of genes—presumptively, each with an important purpose—that enable the bacterium to survive (if not thrive) within its ecological niche. The partner to this resilience is fastidiousness. Bacterium grow and divide with consummate orchestration.1, 2, 3 As its genome is replicated within its cytoplasm4, 5 the structural entities of its cell envelope—lipids, proteins, glycans, and a peptidoglycan cell‐wall polymer—are synthesized and assembled.6 Minutes later the daughter cells separate by the fracturing of their shared septal cell envelope.7 The daughter bacterium replicates, in size and shape, its progenitor.8, 9, 10, 11 We benefit when we are complemented by bacteria, and when these bacteria rest in their proper ecological niches.12, 13 We suffer when we encounter pathological bacteria outside of these niches.

Among the key drivers of scientific discovery are admiration of elegance, and the necessity to solve problems. Bacteria combine both incentives. The study of death records prior to the introduction of antibiotics in the mid‐20th century shows that bacterial infection is indifferent to gender, age, education, and social status. Within the developed world today, antibiotics and vaccines have withdrawn the fear of words such as cholera, consumption, diphtheria, meningitis, pertussis, pneumonia, scarlet fever, and whooping cough. Human longevity was transformed as a result of the progressive development over the past century of chemotherapy for the control of bacterial infection.14, 15, 16, 17, 18 Accompanying this transformation was the entry into the medial lexicon of a chemical vocabulary—exemplified by bacitracin, penicillin, cephalosporin, and vancomycin—demonstrating the power of antibacterial chemotherapy. Each antibiotic listed in the preceding sentence targets the bacterial cell‐wall polymer as an essential structural edifice. Each of these antibiotics is a natural product, devised by one microbe for advantage against another in a ceaseless internecine effort toward competitive advantage.19, 20, 21 The antibiotic‐producing microbe ensures its immunity by the simultaneous creation of a means to subvert the mechanism of its antibiotic. Over decades of time these existential immunity solutions have escaped by horizontal gene transfer,22, 23, 24, 25 and the stress responses activated by antibiotics have effected mutations in the target that subvert the antibiotic mechanism.26, 27, 28, 29 The result is antibiotic‐resistant bacteria. And indeed, less than a century after the widespread medical use of antibiotics, extensively resistant—and thus powerfully pathogenic bacteria—have arrived.30, 31, 32, 33, 34, 35, 36, 37

Their emergence was not a surprise.38 Notwithstanding the widespread scientific appreciation of the monumental difficulty to secure new generations of antibiotic structure to vanquish these multidrug and extensively‐drug resistant bacteria,39, 40, 41 the paths toward surmounting this response are recognized. These paths include a renewed focus in antibacterial natural‐product discovery,14, 42, 43, 44, 45, 46, 47, 48 the identification of new targets,49, 50, 51, 52 the exploitation of the nutritional needs of the bacteria for the facilitated delivery of antibiotic prodrugs and conjugates,53, 54, 55 the creation of hybrid antibiotic structures,56 understanding the intersection of primary metabolism and antibiotic mechanism,57, 58 and the discernment of effective combinations of antibiotics with other antibiotic, enhancer, or adjuvant structures.59, 60, 61, 62, 63, 64, 65, 66 This essay is a personal perspective, exemplified by the recent efforts of our laboratory, with respect to understanding the intimate relationship between the biochemistry of antibiotic resistance and antibiotic discovery. These efforts combine the paths of new structure, new mechanisms, renewed target evaluation, and the evaluation of antibiotic‐adjuvant combinations that target the bacterial cell wall.

2. THE CELL ENVELOPE OF BACTERIA

A key premise is to turn against bacteria the elegant fastidiousness with which they synthesize their cell envelopes. An understanding of how this turn may be done requires an appreciation of the organization of the bacterial cell envelopes. The cell envelope of Gram‐positive bacteria—bacteria that retain dye‐coloration after gentle washing—consists of two substructures. The first is a lipid bilayer membrane that surrounds the cytoplasm of the bacterium. Above this membrane is the second substructure, in the form of a polymeric cell wall that forms the outer surface of the bacterium. The central polymer of this cell wall is a glycopeptide synthesized by the covalent crosslinking of glycan strands through the peptide stems found on alternate saccharides. This polymer surrounds the entire bacterium. It is called by either of three synonyms: the murein, the sacculus, or (most commonly) the peptidoglycan. The peptide stems of the peptidoglycan have function beyond their primary role of crosslinking: they are used to covalently attach yet different glycan strands (called wall teichoic acids) and proteins.67, 68, 69, 70, 71 The genera of notable pathogenic Gram‐positive bacteria include the Enterococci, the Staphylococci, and the Streptococci. The cell envelope of Gram‐negative bacteria is more complex. Two lipid bilayer membranes enclose a thin but no less structurally important space, the periplasm. The periplasmic space is highly populated by proteins, and is the location in Gram‐negative bacteria for the peptidoglycan polymer. The peptidoglycan of Gram‐negative bacteria is thinner than the peptidoglycan of Gram‐positive bacteria, yet still surrounds the entire bacterium. The inner membrane contains the cytoplasm, while the outer membrane contains the periplasm, which in turn encases the inner membrane. The exterior surface of Gram‐negative bacteria is outer leaflet of this second membrane. While the molecular components of the three other leaflets of the Gram‐negative membranes are structurally similar phospholipids, this outer leaflet is populated with a specialized lipopolysaccharide. The genera of notable Gram‐negative pathogens include the Acinetobacter, the Escherichia, the Enterobacter, the Klebsiella, the Neisseria, the Pseudomonas, and the Salmonella.

3. CELL ENVELOPE‐TARGETING ANTIBIOTICS

A preponderance of the antibiotic entities, including the usually sublimely efficacious β‐lactam antibiotics, interfere with the assembly and/or function of the cell envelope. The β‐lactams inhibit peptidoglycan synthesis. As the chemical entities that comprise the bacterial cell envelopes—the peptidoglycan (similar structurally in both Gram‐positive and Gram‐negative bacteria), the Gram‐positive wall teichoic acid, and the Gram‐negative lipopolysaccharide—each lack eukaryotic counterparts, antibiotics that target their synthesis have intrinsic selectivity and hence generally safety.49, 72, 73, 74 The integrity of these cell envelopes rests not on any one of these entities, but on their seamless integration. Wright opined that we have entered a golden age of understanding as to how antibiotics function.75 The prescience of his statement is evident by scores of ensuing publications. Intimate structural synergy is now recognized between the peptidoglycan and the teichoic acids of the Gram‐positive bacteria,76 and between the peptidoglycan and the outer membrane of the Gram‐negative bacteria.77, 78 Indeed, new opportunities for antibiotic discovery have been identified that interfere with teichoic acid integration into the cell wall of Gram‐positive bacteria69, 79, 80, 81, 82, 83, 84 and with respect to lipopolysaccharide transport in Gram‐negative bacteria.85, 86, 87, 88, 89, 90 Within this universe of opportunity, we exemplify our own efforts toward the mechanistic and structural understanding of key enzymes of the bacteria with roles in cell envelope biosynthesis and antibiotic resistance.

4. MIRACULOUS CHEMISTRY

Miracle is a strong word. The dramatic elimination of mortality from infectious diseases in the mid‐20th century was the result of the widespread clinical availability of the antibiotics—a transformation that largely persists to this day—irrevocably paired the word “miracle” with the word “antibiotics.”91, 92, 93, 94 The antibiotic class that arguably best exemplifies the popular understanding of this pairing is the β‐lactams. The β‐lactam class includes the penicillins, cephalosporins, monobactams, and carbapenems.95 Each structure in each of these subclasses has the eponymous four‐membered cyclic amide (β‐lactam) that is central to their mechanism of action (Scheme 1). This mechanism is disruption of the crosslinking of the peptidoglycan cell‐wall polymer. As intimated previously, this polymer is not static. The peptidoglycan polymer grows in step with the bacterium, and neatly fractures at a septum to separate the mother and daughter cells. At each step, the internal turgor pressure of the bacterium is preserved.96 Fulfilling this obligation with respect to a living polymer is a monumental task. Each bacterium possesses families of glycosyltransferases (to grow the glycan strands); transpeptidases (to interconnect the peptide stems of the glycan strands, as the polymerization event); and amidases, carboxypeptidases, endopeptidases, and glycosylases (to control the extent of stem crosslinking and glycan length, and to enable daughter cell departure).7, 97, 98 Indeed, the bacterial peptidoglycan exists in a constant state of remodeling so as to also fulfill its other obligate tasks, such as cell stiffness99, 100 and cell shape.101, 102, 103, 104, 105 Chemical entities that compromise any aspect of cell‐wall growth and remodeling have antibacterial potential. The targets of the β‐lactams are a family of periplasmic enzymes, each with the same active‐site character (notably, an active‐site lysine base and serine nucleophile) notwithstanding separate (corresponding to specific members of the family) transpeptidase, carboxypeptidase, and endopeptidase activities. The members of this family retain their historical appellation, regardless of their specific activities, as penicillin‐binding proteins (PBPs).69, 106, 107, 108, 109 For example the penicillin‐sensitive, and nearly spherical in its shape, Gram‐positive bacterium Staphylococcus aureus, has four PBPs. The rod‐shaped Gram‐negative bacterium Escherichia coli has 12 PBPs.110 The “binding” inference in their name—that these are enzymes that are inactivated by the β‐lactams, and by a covalent mechanism—is correct.111, 112 Some PBPs are critical. Some are not. If the overall structure of a β‐lactam is optimized so as to match to the critical PBP(s) of particular bacterial species, the resulting loss of function is ultimately bactericidal. With timely intervention in a clinical setting, the infection of the β‐lactam‐susceptible bacterium is vanquished.

Scheme 1.

The structure of the acyl‐d‐Ala‐d‐Ala stem terminus is shown in the upper box. The R substituent is the remainder of the stem structure (see Scheme 2). The four major β‐lactam structural subclasses are shown in the lower box. Their four‐membered β‐lactam substructure is shown in red, and is the reactive core in the structural mimicry of the amide bond connecting the two d‐Ala residues. The R′ and R″ substituents are optimized by medicinal chemists for antibacterial selectivity, potency, pharmacokinetics, and safety

The judicious appearance of the words “optimized,” “ultimately,” and “susceptible” in the two preceding sentences emphasize the themes of the introduction of this review. Bacteria have compelling reasons to evade, subvert, and destroy β‐lactams, and resistant bacteria often do all three. They evade (for example) by active efflux of the antibiotic from the periplasm; they subvert (for example) by PBP mutation and by the recruitment of β‐lactam‐impervious PBPs; they subvert by the creation of persister and tolerant bacterial states113; and they destroy through the expression, as a result of the refinement of the very same lysine‐serine active‐site motif of the PBP,114 of potent hydrolytic (and hence deactivating) enzyme hydrolases of the β‐lactams, the β‐lactamases. The transformation of susceptible bacteria to resistant bacteria is continuous, and hence so is the necessity to reoptimize the clinical β‐lactam structure. The word “ultimately” reflects a changing consensus. It was believed previously that the PBP loss of function was directly bactericidal. PBP inactivation by β‐lactams is seen now as the triggering event that initiates more complex, and perhaps even species‐specific, pathways that culminate in bactericidal cell lysis.115, 116, 117, 118, 119, 120 Perhaps as a consequence (but at the molecular level, for reasons not understood) some resistance mechanisms against the β‐lactams exert a cellular cost. The notorious Gram‐positive pathogen S. aureus does not activate its resistance mechanisms until it detects the presence of a β‐lactam. Likewise, among many no less notorious Gram‐negative bacterial pathogens, β‐lactamase production likewise follows an encounter with a β‐lactam. Both circumstances intrigued us. What molecular mechanisms for β‐lactam detection are being used, and how does this detection produce a resistance response? Could the understanding of these events be used to identify new targets for antibacterial discovery, or to disarm the resistance response?

5. SENSING AND EVADING THE β‐LACTAM ANTIBIOTICS BY A GRAM‐POSITIVE BACTERIUM

S. aureus is a bacterial pathogen of humans and livestock. It is encountered as a β‐lactam‐susceptible strain (referred to customarily as methicillin‐susceptible S. aureus, MSSA) and as β‐lactam‐resistant strains (MRSA, encountered as a variety of strains).121, 122, 123, 124, 125, 126, 127, 128 Here, methicillin refers to a specific second‐generation penicillin that was developed decades ago as especially active against MSSA, as a result of its stability to the hydrolytic deactivation of its β‐lactam substructure (in contrast to first‐generation penicillins) as catalyzed by a plasmid‐encoded β‐lactamase (BlaZ). The acquisition of the gene for the BlaZ β‐lactamase was selected by S. aureus in response to the clinical use of first‐generation penicillins, which were in turn efficiently deactivated by the enzyme. Although methicillin is no longer used clinically, the MSSA and MRSA terminologies persist. Both MSSA, and even more so MRSA, infections have significant morbidity and mortality.129 MSSA is treatable today by a variety of penicillins and cephalosporins.130 MRSA was previously a primarily nosocomial infection, but today is seen routinely also as a community‐acquired infection. A variety of strains ebb and flow unpredictably across the nosocomial and community geographies. MRSA are resistant to all but the most recent β‐lactams. Accordingly, their problematic chemotherapy requires an oxazolidinone (linezolid), one of the newest β‐lactams, or older drugs (such as daptomycin or vancomycin) whose clinical development was renewed in response to the need for anti‐MRSA agents.128, 131, 132, 133 Resistance to each of these clinically‐used antibiotics is encountered. As the BlaZ β‐lactamase found in many MRSA strains is identical to that found in MSSA strains, an additional—and entirely different and more sophisticated—resistance mechanism differentiates MRSA from MSSA.

This additional resistance mechanism was the complementation of the existing PBP family of S. aureus with a new PBP enzyme (now named PBP2a),134, 135 recruited from environmental Staphylococci.136, 137 The primary enzyme catalyst of MSSA peptidoglycan assembly is PBP2. PBP2 is a bifunctional enzyme. One of its two active sites, found proximal to its membrane surface, catalyzes elongation of the glycan strands of the peptidoglycan. At the apex of PBP2a is a second active site. At this site transpeptidase catalysis of stem crosslinking occurs. The transpeptidase active site recognizes the terminal amino acids of the peptide stem, acyl‐d‐Ala‐d‐Ala, as its first substrate. In the first half‐reaction of transpeptidation a nucleophilic serine of the PBP adds to the amide bond between the two d‐Ala amino acids, accepting its acyl moiety and releasing the terminal d‐Ala as a leaving group. In the second half‐reaction this acyl moiety is transferred to the primary amine of the amino acid found at the third position—an l‐Lys or an l‐Lys modified in its side chain by Gly residues—on the stem of an adjacent glycan strand, thus crosslinking the two. β‐Lactams are structural mimics of the acyl‐d‐Ala‐d‐Ala terminus of the stem peptide.138, 139, 140 PBPs are inactivated when β‐lactams are bound as acyl‐d‐Ala‐d‐Ala mimetics at the active site, and complete the first half‐reaction to form an acyl‐enzyme, but are incompetent for acyl‐transfer and thus are inactivated (Figure 1). In contrast to PBP2, PBP2a is a monofunctional enzyme. It is a transpeptidase catalyst. The ability of PBP2a to distinguish in favor of the d‐Ala‐d‐Ala stem as its substrate, and against β‐lactams as inactivators, is exceptional and is rooted in the allosteric control of the conformational states of the enzyme, as explained below. In the presence of β‐lactams the transpeptidase activity of PBP2 is lost, and expression of PBP2a is activated (i.e., antibiotic induction of resistance). Peptidoglycan assembly occurs by coordination of the transglycosylase activity of PBP2 with the transpeptidase activity of PBP2a.147, 148 Presumably in order to accommodate the added steric demand of this collaboration, a new, longer amino terminus for acyl‐enzyme acceptance is created by addition of a (Gly)₅ pentapeptide to the ε‐amine of the lysine.149, 150, 151, 152 Moreover, yet additional crosslinking activity is provided by a second transpeptidase (PBP4) either as a result of resistance mutation and/or increased expression.153, 154, 155, 156, 157, 158 Hence, S. aureus has two primary resistance mechanisms against β‐lactam antibiotics: the plasmid‐encoded BlaZ β‐lactamase (useful against early‐generation β‐lactams) and PBP2a (useful against all but the most recent generation of β‐lactams, such as ceftaroline and ceftobiprole). As noted previously, both resistance mechanisms are activated only when S. aureus detects the presence of a β‐lactam. How is this detection done?

Figure 1.

The stereoview of the active site of S. aureus PBP2a acylated by the cephalosporin, ceftaroline (PDB http://bioinformatics.org/firstglance/fgij//fg.htm?mol=3ZG0).141 Ring opening of the β‐lactam requires catalysis by a lysine general base (K406) that activates acyl‐transfer from the β‐lactam to the nucleophilic serine (S403). In this perspective the serine is below the atoms of the ceftaroline. The third amino acid identified in the figure is the gatekeeper tyrosine (Y446). This tyrosine is in the “bridge” that caps the active site. The opening and closing motions of this bridge, in response to ligand occupancy at the distant allosteric site, control substrate access to the active site. See also Figure 2

As with all critical biochemical pathways, activation requires both check and counter‐check. The genetic aspects of this system have been known for three decades. The mecA gene encoding PBP2a is chromosomal, albeit within an integrated mobile genetic element termed the staphylococcal chromosomal cassette (abbreviated as SCCmec).159 In contrast, the blaZ gene is usually plasmid‐encoded (but is also found in some SCCmec types). The profound regulatory similarity between how PBP2a and the BlaZ β‐lactamase are produced is of note. PBP2a expression is regulated by two proteins (MecI and MecR1), as is the case for BlaZ expression (BlaI and BlaR1).160 The protein and mechanistic similarity between these pairs is such that in some MRSA types the two crosstalk, or BlaI/BlaR1 suffice to activate PBP2a expression.161 Their identities were established: MecI and BlaI are gene repressor proteins and are functionally interchangeable; while MecR1 and BlaR1 are both membrane‐spanning β‐lactam sensor/signal transducer proteins. At the outside of its membrane (and thus exposed to its milieu) MecR1 (BlaR1) presents an extracellular β‐lactam‐sensor domain.162 Its identity as a β‐lactam‐interacting domain correlated with the sequence homology of this domain with the class D serine β‐lactamases.163, 164 The four transmembrane segments of the BlaR1/MecR1 structure create (on the inner leaflet of the membrane, through their loop structures) a metalloproteinase active site, characterized by an H‐E‐X‐X‐H motif found on the cytoplasmic L3 loop (that connects the TM3 and TM4 transmembrane segments).165, 166 In this motif the two histidines (H) are ligands to a Zn(II) Lewis acid, facilitating the glutamate (E) as a general base for peptide hydrolysis. The presumption that this cytoplasmic metalloproteinase active site is activated by β‐lactam engagement of the extracellular sensor domain is proven by a host of experimental studies across several laboratories (Scheme 2). Here, we focus on our contributions toward the now mature understanding of the sensor domain‐metalloproteinase active‐site communication, and to the continuing challenge of understanding how the activation of the metalloproteinase activity subsequently enables PBP2a (and BlaZ) expression.

Scheme 2.

Recognition of the presence of a β‐lactam occurs at the cell surface through acylation of the BlaR^S sensor domain of BlaR1. The zinc‐dependent protease activity of the L3 loop is activated on surface acylation by the antibiotic. This activity degrades the BlaI gene regulator, resulting in the derepression of the antibiotic‐resistance genes, including that for BlaR1 and BlaI. On activation, the protease domain catalyzes autocleavage at two sites (one, F284, is depicted) within the L3 loop. An additional cleavage at the membrane surface by another protease sheds the sensor domain from the membrane

A pivotal conception was that this understanding of the overall BlaR1/MecR1 structure strongly implicated that a generic construct matching the sequence of the C‐terminal sensor domain would express as a soluble protein possessing ring‐opening activity with respect to β‐lactam substrates. This conception was affirmed by independent studies by Kerff et al. (sensor domain of the BlaR1 of Bacillus licheniformis)167 and ourselves (sensor domain of the BlaR1 of S. aureus).168, 169 These studies of the BlaR1 sensor domain (BlaR^S) established not simply sequence homology, but mechanistic alignment between BlaR^S and the class D β‐lactamases with respect to β‐lactam engagement and the formation of a serine acyl‐enzyme as a result of catalytic opening of the β‐lactam. Class D β‐lactamases are distinguished from the other two classes of serine β‐lactamases by the use of an N‐carboxylysine—formed by the reaction of an active‐site lysine with CO₂—as the general base catalyst for β‐lactam ring opening.170, 171, 172 The reactivity of BlaR^S differed from the class D β‐lactamases in that the acylation of its serine was largely irreversible (similar to serine acylation of the PBPs) whereas the class D β‐lactamases were competent for acyl‐enzyme hydrolysis, and so capable of catalytic turnover. The contrasting irreversible acylation of BlaR^S by β‐lactams is the decisive initiating event that alerts S. aureus to the lethal presence of an antibiotic. The basis for the irreversibility derives from the dynamic equilibrium of the reaction of lysine with CO₂ to prepare the requisite carbamate anion required for catalysis of serine acylation by the β‐lactam.171 Following this acylation event, the altered active‐site environment of the acyl‐enzyme173, 174 favors N‐decarboxylation (loss of CO₂) of this carbamate anion to give the β‐amine of lysine.175, 176, 177 The sharply different protein dynamics as a consequence of the sequential N‐acylation/ N‐decarboxylation have been characterized by detailed NMR study of BlaR^S.178, 179 Notably, NMR study documented an association between the L2 loop peptide (as an amphiphilic helical segment) and the BlaR^S protein.180 How the protease domain is activated is not understood, but the process must involve amplification of the conformational change that has crossed the membrane span.181, 182 Loss of the BlaI repressor by proteolytic degradation of the now‐activated protease domain coincides with β‐lactamase expression in both B. licheniformis 183, 184 and S. aureus.185, 186, 187 The protease domain activation is simultaneous to a pair of autoproteolytic reactions within the domain. The precise role of the autoproteolysis of the (cytosolic) L3 loop, as either coincident with acquisition of the Zn‐dependent L3 proteolytic activity165 or as a means of reversal of induction,181 remains uncertain. Our studies with the MecR^S domain as the free protein indicates fundamental mechanistic similarity albeit with differences in the kinetics.188

Activation of PBP2a expression through MecR1/BlaR1 signaling is the primary basis for S. aureus β‐lactam resistance. As noted above PBP2a complements a second S. aureus PBP, PBP2. This fully functional pair is abetted by a third PBP (PBP4, by a mechanism not yet fully deciphered)155, 158 and by “auxiliary enhancing factors,” such as the addition of the (Gly)₅ extension to the amine on the peptide stem customarily used for the crosslinking of the peptidoglycan. The subtle basis for the irrefutable ability of PBP2a to discriminate in favor of its peptidoglycan substrate and against β‐lactam inactivators emerged from analysis of the crystal structure of the soluble enzyme (alone, and following acylation of its serine by various β‐lactams).189, 190 In collaboration with our colleague Juan Hermoso (at the Instituto de Química‐Física Rocasolano, Madrid) we too solved this structure, as a companion to kinetic studies directed toward understanding this discrimination.191 These studies confirmed the conclusion from the initial crystallographic analysis189 of critical conformational motion in the enzyme.192, 193 These motions were responsive to the pH at which the assay was done,194 and not unexpectedly to the presence of peptidoglycan structure. However, the structural basis for the peptidoglycan effect was unexpected as the peptidoglycan also bound to a second site of the PBP2a enzyme (Figure 2).143, 195, 196 Notwithstanding the 60 Å distance of this second site—designated an allosteric site—from the active site, the two are linked by a single sweeping conformational motion that occurs in response to occupancy of the allosteric site by peptidoglycan.144 The focus at the active site of this sweep is an opening motion of the loop structure governing access to the active site. During the necessary process of translocation as PBP2a constructs the cell wall, its active site is closed to both substrate and inactivator. Following translocation, the allosteric site of PBP2a engages existing “template” peptidoglycan so as to trigger the conformational change that opens the active site. Only then is new peptidoglycan structure preferentially admitted for catalysis.141

Figure 2.

The stereoview of the structure of the periplasm‐located domains of the S. aureus PBP2a with a quinazolinone (depicted in green‐colored space‐filling structure) occupying the allosteric site (PDB http://bioinformatics.org/firstglance/fgij//fg.htm?mol=4CJN).142 The quinazolinone has intrinsic antibacterial activity. A single quinazolinone allosteric effector binds in a groove between the allosteric site and a so‐called pedestal domain. The arrow points to the active site. The active‐site serine is approximately 60 Å distant from the allosteric site. The structural change that occurs in the allosteric transformation that controls substrate access to these active‐site serine is understood by crystallographic evidence, simulations and kinetics data143, 144, 145, 146

The identification of this allosteric site had serendipitous consequence. Virtual screening using our PBP2a structure identified two classes of small molecule adjuvants that in murine pharmacological assay restored the antibiotic activity of β‐lactams against S. aureus (as well as against other Gram‐positive pathogens). The first class was characterized by a central 1,2,4‐oxadiazole ring197, 198, 199, 200, 201 and the second class by a quinazolinone ring. The quinazolinones occupy the allosteric site of PBP2a and with this occupancy disrupt the allosteric mechanism.142, 202 They have alone antibacterial activity against a breadth of MRSA strains (in almost all cases with MIC values of less than 0.25 mg l⁻¹). The combination of a quinazolinone with the commercial β‐lactam pair piperacillin‐tazobactam is bactericidal to MRSA at sub‐MICs for all three drugs.203 The outstanding promise of the quinazolinone class has been recognized by other groups, with SAR development confirming independent antibacterial activity against both MRSA204, 205 and Mycobacterium tuberculosis.206 At a time when utmost diligence and creativity is required to combine structure207 and target132 to secure antibacterial activity, these studies from our laboratory validate an experimental approach centered on the protein science.

6. ABETTING THE β‐LACTAM ANTIBIOTICS AGAINST GRAM‐NEGATIVE BACTERIA

Eukaryotes are held hostage to both Gram‐positive and Gram‐negative bacterial infection. Indeed, the latter bacteria are now seen as the more daunting. Gram‐negative bacteria use their second membrane as a barrier to antibiotic permeation, and to populate their periplasmic “moat” with both plasmid‐encoded resistance enzymes and multiprotein motility and virulence complexes.37, 208, 209 Given our long‐standing interest in the β‐lactamases as a pre‐eminent Gram‐negative defense against β‐lactams210 our focus turned several years ago to the relationship between β‐lactamase expression and the peculiar mechanism used by Gram‐negative bacteria to expand (grow) their peptidoglycan polymer. Extensive portions of the Gram‐negative peptidoglycan are removed, degraded, and recycled during peptidoglycan growth.211, 212 Hence, growth is the net difference between peptidoglycan accretion (construction) and removal and recycling (destruction). Although this recycling (as studied in vitro with the Gram‐negative bacterium E. coli) is not essential, cell‐wall recycling is now recognized as an important pathway of many Gram‐negative and Gram‐positive bacteria.213, 214, 215 Notably, cytoplasmic monitoring of the integrity of muropeptide recycling in response to the presence of cell wall‐targeting antibiotics (especially β‐lactams) controls β‐lactamase expression as a key resistance response of many Gram‐negative pathogens.216, 217 The clinical importance of this connection cannot be overstated.218, 219, 220 The breadth of study of this β‐lactamase‐resistance response—from the molecular (antibiotic structure) to the macromolecular (protein‐ligand interaction, biochemical pathways) to the microbiological and the clinical—requires the medium of the comprehensive review.221, 222, 223, 224, 225, 226, 227, 228 Here, we focus on the involvement of a family of important enzymes—the lytic transglycosylases—in peptidoglycan recycling and β‐lactamase induction.229, 230

The lytic transglycosylases (LTs) are periplasmic enzymes (many, but not all, lipoproteins) that catalyze the deconstruction (hence “lytic”) of the polymeric peptidoglycan through the nonhydrolytic cleavage of the glycan strand of the peptidoglycan (hence, the awkward “transglycosylase”). Among the ensemble of cell‐wall deconstructing enzymes that are found in the periplasm,98, 231, 232 the lytic transglycosylases have a central position both with respect to key pathways and events. Their substrate and reaction products are shown in Scheme 3. All Gram‐negative bacteria have several LTs. While the activity itself is critical, functional redundancy among the LTs does not result in any one LT as essential.235, 236 The LTs are active in peptidoglycan growth where their catalysis excises existing peptidoglycan so as to create a template for insertion of new peptidoglycan,237 and in the excision of peptidoglycan so as to allow the insertion through the peptidoglycan of critical macromolecular complexes. An outstanding example of the latter process is the recruitment by enteroaggregative E. coli of the MltE LT in order to excavate its peptidoglycan in order to allow the insertion of a macromolecular Type VI secretion system.238 The Gram‐negative bacterium Pseudomonas aeruginosa is a significant cause of morbidity and mortality as a result of infection of wounds and lungs,239 and β‐lactams are critical components of its chemotherapy.208, 240 P. aeruginosa recycles its peptidoglycan during growth using LTs to excise peptidoglycan segments (these are called “muropeptides”). These muropeptides pass from the periplasm to the cytoplasm through the AmpG transmembrane permease.241, 242 Administration of a β‐lactam that inactivates one particular PBP of P. aeruginosa (PBP4) fundamentally alters the structural composition of the muropeptide pool that passes through AmpG, as evidenced by a dramatic increase in its β‐lactamase (the AmpC β‐lactamase) production.243, 244, 245 In vitro determination of the catalytic behavior of PBP4 toward muropeptide substrates established its dual ability to act as a 4,3‐endopeptidase and a d,d‐carboxypeptidase with respect to these muropeptides.246, 247 Nor is PBP4 the only decisive enzymatic activity for muropeptide processing leading to derepression of the ampC gene. The muropeptide controlling the expression of the ampC gene originates from the catalytic activity of an LT.248, 249 Clinical AmpC hyperexpression (resulting in β‐lactam resistance) results from the mutational loss of function of the AmpD amidase.250, 251 Extensive microbiological,252, 253 crystallographic,254 and enzyme kinetic studies255, 256, 257 (using synthetic muropeptides of precise structure)258 identify the function of AmpD in muropeptide recycling, now in the cytoplasm, as the removal of the peptide stem from the muropeptide. A fourth enzyme (NagZ, also located in the cytoplasm) also participates in the muropeptide processing that leads to AmpC production.259, 260 NagZ is glycoside hydrolase that cleaves the glycosidic bond between the NAG‐NAM disaccharide of muropeptides.260, 261, 262 Loss of activity of NagZ suppresses AmpC production.263 Hence, inhibition of NagZ augments the antibiotic efficacy of β‐lactams.264, 265, 266, 267, 268, 269, 270 Structural reconstruction of all of these observations gives the conclusion that AmpC β‐lactamase correlates with the passage of a particular muropeptide structure from the periplasm to the cytoplasm, through the transmembrane AmpG permease. The muropeptide originates by LT‐dependent excision from the cell wall.248 When PBP4 is catalytically active muropeptide recycling occurs without induction of AmpC. When PBP4 catalytic activity is lost due to inactivation by a β‐lactam antibiotic, the composition of the recycling pool of muropeptides is altered by the enrichment of NAG‐NAM disaccharide that bears the full‐length pentapeptide stem. This muropeptide is an activator for the AmpR gene regulator. It is this activation, as a consequence of shifting the equilibrium due to the aforementioned enrichment process, which unleashes the expression of the ampC gene for the AmpC β‐lactamase production.

Scheme 3.

The reaction catalyzed by the lytic transglycosylases. The glycan strand of the peptidoglycan is cleaved by a nonhydrolytic intramolecular transacetalization.229, 233, 234 Depending on the LT, the cleavage may occur at the terminus of the chain, or within the chain (n ≥ 1). The peptide stem shown in these structures is the complete Gram‐negative “pentapeptide” stem, having the d‐Ala‐d‐Ala dipeptide (fourth and fifth amino acids of the stem) as its terminus

Control of repression and derepression of AmpC occurs through differential complexation of different muropeptides (structures are shown in Scheme 4) by the LysR‐type transcription regulator, AmpR.271, 272, 273 The bicyclic (anhydro) saccharide of the activator originates from LT catalysis. It is the product of removal of the NAG from the disaccharide product of the LT reaction, as depicted in Scheme 3. The repressor is an advanced key metabolite in the recycling process en route to lipid I, and ultimately lipid II, the precursor to the peptidoglycan. The latter is known as the Park nucleotide. The presence of the full‐length peptide stem in both is critical, as the reach of the terminal d‐Ala within the requisite binding cavity of AmpR effector‐binding domain is important for recognition.274, 275 The key activity of PBP4 is d,d‐carboxypeptidase hydrolysis of the terminal d‐Ala‐d‐Ala bond of the stem. In the presence of the PBP4 activity, the pool of muropeptides is depleted of the pentapeptide stem, hence, activation/derepression of AmpR would not take place. At this juncture, recycling of the muropeptides sets in motion in a multistep process entailing removal of the shortened stem peptide, followed by a sequence of events that leads to Park nucleotide. Consequently, the increase in concentration of Park nucleotide—as opposed to that of the activator/derepressor metabolite—would favor its complexation with AmpR, which effectively shuts down expression of ampC, as it signals homeostasis and the absence of the antibiotic‐induced mobilization. Conversely, when the d,d‐carboxypeptidase activity of PBP4 is inactivated by the antibiotic the derepressor metabolite is dominant, leading to induction of expression of the ampC gene. Although the fundamental operation of this regulation was long surmised,216 it is nonetheless gratifying that our crystallographic studies275 of the effector‐binding domain of P. aeruginosa AmpR complement studies with the same domain of the Citrobacter freundii AmpR,276 and further match to our analysis of the structural fluctuation within the muropeptide pool in the absence and presence of β‐lactams.277

Scheme 4.

Derepression of the ampC gene by ligand binding to the effector domain of AmpR in P. aeruginosa coincides with LT excision from the peptidoglycan of a muropeptide that retains the d‐Ala‐d‐Ala stem terminus. Loss of PBP4 catalysis by the presence of a β‐lactam antibiotic allows retention of the d‐Ala‐d‐Ala terminus. Processing in the cytoplasm by NagZ, but not by AmpD, gives the muropeptide on the right. When PBP4 is active, the terminal d‐Ala residue is removed, and the resultant muropeptide pool is directed to the recycling events leading to formation of Park nucleotide (depicted to the left). The elevated concentration of Park nucleotide leads to repression of the system by binding to AmpR and signals homeostasis. This scheme is a simplification. Ligand regulation of AmpR is more complex (and indeed, is not general from one bacterial genus to another) as is discussed elsewhere228

Notwithstanding this seeming maturity of understanding, it would be error to regard as complete the understanding of the control of ampC expression by muropeptide recycling. For example, AmpR is responsive additionally to the nucleotide muropeptides created later in the recycling pathway, and may also be responsive to the different saccharide‐free stems resulting from AmpD catalysis.276, 278 Moreover, P. aeruginosa has a family of 11 LTs, each presumably with separate primary function(s) but overlapping secondary function.279 A credible strategy to create meaningful clinical adjuvants to preserve β‐lactam efficacy would be to suppress muropeptide recycling in its entirety. Indeed, preliminary data support fully this hypothesis.280, 281, 282 One focus of such an effort would be the understanding of the enigmatic functions of the LT family and the respective means to their regulation.283, 284 Accordingly, we have undertaken extensive mechanistic and structural studies of this family. These efforts have led us to the discovery that an ABC‐transporter‐like regulatory module is present in the structure of the P. aeruginosa MltF LT285; that the P. aeruginosa SltB1 LT has a catenane dimer structure; that the P. aeruginosa Slt LT286 has the same compelling doughnut‐shape as is seen in other Slt LTs whose catalytic activity is governed by important conformational changes234, 287, 288, 289; and to a computational mechanism for the smallest LT of E. coli, MltE,233 based on our crystallographic analyses.290, 291

The assertion of an intimate relationship between the LT family and β‐lactam antibiotic efficacy has irrefutable support. In the course of efforts to identify natural products that functionally complemented the mechanism of the β‐lactams, Imada et al. in 1982 identified the bulgecins as natural products that when combined with β‐lactams resulted in deformation of bacteria by a bactericidal bulge formation in the Gram‐negative envelope.292, 293 The molecular targets of the bulgecins—their structure (Figure 3) is recognizable as characteristic of iminosaccharide‐type inhibitors of glycoside hydrolases—were the LT enzymes.294, 295 Moreover, the operon for bulgecin biosynthesis in the bacterium Paraburkholderia acidophilia is contiguous296 with that of the monobactam‐class β‐lactam, sulfazecin.297, 298, 299 The presumption that P. acidophilia coordinates the biosynthetic production of the bulgecins with sulfazecin was implicitly validated by the studies of Imada et al.300 and our own recent findings.301 More recently we documented selectivity in inhibition of the members of the LT family.294, 302 The target for bulgecins would appear to be the Slt, MltD, and MltG LTs out of the 11 known P. aeruginosa LTs.303

Figure 3.

The stereo image of the structure of the complex of the E503Q catalytic‐impaired mutant of the P. aeruginosa Slt LT bound to bulgecin A (bottom structure) (PDB http://bioinformatics.org/firstglance/fgij//fg.htm?mol=6FCQ).286 Bulgecin A (depicted in sphere representation; color by atom types) is bound at the active site of this LT. This active site is located on the threshold of the central opening of this doughnut‐shaped protein

7. CONCLUSION

The presumption that the protein chemistry of the β‐lactams is fully solved, after nearly 80 years of their use in the chemotherapeutic control of bacterial infection, is incorrect. Rather, we are only now appreciative of its complexity. The proteins and pathways among the genera (and species) of bacteria are as profoundly different as they are similar. We see that different bacteria select the most appropriate resistance mechanisms with respect to their physiology—for example, one species selects one particular β‐lactamase while another selects a different one—but broadly speaking we are at a loss to understand the reasons behind this selection.304 The two distinct enzymes clearly serve the needs of those specific bacteria. But is it just merely the fortuitous gene acquisition and random selection of the antibiotic resistance trait that is at play? We discern simple insight and design our next experiment with attention to the singular rather than the plurality of Ockham's razor.305 The allosteric control of S. aureus PBP2a (and perhaps of other PBPs)145, 306 may well be to synchronize loop opening of the active site with protein translocation, but this explanation is almost certainly an oversimplification. The narrative presented here conceptualizes a two‐dimensional pathway for the peptidoglycan construction and deconstruction by P. aeruginosa when the realities of both AmpR regulation227, 307 and cell‐wall biosynthesis are multidimensional.308, 309 Yet this reductionism led to the deduction that the allosteric regulation of PBP2a is a solid basis for small molecule antibacterial discovery, and the parsing of the specificity of bulgecin as a selective inhibitor within the lytic transglycosylase family of P. aeruginosa. Both are small but firm forward steps. Such steps are the substance of scientific momentum.

Fisher JF, Mobashery S. Constructing and deconstructing the bacterial cell wall. Protein Science. 2020;29:629–646. 10.1002/pro.3737