Influence of the α-Methoxy Group on the Reaction of Temocillin with Pseudomonas aeruginosa PBP3 and CTX-M-14 β-Lactamase

The prevalence of multidrug-resistant Pseudomonas aeruginosa has led to the reexamination of older “forgotten” drugs, such as temocillin, for their ability to combat resistant microbes. Temocillin is the 6-α-methoxy analogue of ticarcillin, a carboxypenicillin with well-characterized antipseudomonal properties. The α-methoxy modification confers resistance to serine β-lactamases, yet temocillin is ineffective against P. aeruginosa growth.

ABSTRACT

The prevalence of multidrug-resistant Pseudomonas aeruginosa has led to the reexamination of older “forgotten” drugs, such as temocillin, for their ability to combat resistant microbes. Temocillin is the 6-α-methoxy analogue of ticarcillin, a carboxypenicillin with well-characterized antipseudomonal properties. The α-methoxy modification confers resistance to serine β-lactamases, yet temocillin is ineffective against P. aeruginosa growth. The origins of temocillin’s inferior antibacterial properties against P. aeruginosa have remained relatively unexplored. Here, we analyze the reaction kinetics, protein stability, and binding conformations of temocillin and ticarcillin with penicillin-binding protein 3 (PBP3), an essential PBP in P. aeruginosa. We show that the 6-α-methoxy group perturbs the stability of the PBP3 acyl-enzyme, which manifests in an elevated off-rate constant (k_off) in biochemical assays comparing temocillin with ticarcillin. Complex crystal structures with PBP3 reveal similar binding modes of the two drugs but with important differences. Most notably, the 6-α-methoxy group disrupts a high-quality hydrogen bond with a conserved residue important for ligand binding while also being inserted into a crowded active site, possibly destabilizing the active site and enabling water molecule from bulk solvent to access and cleave the acyl-enzyme bond. This hypothesis is supported by the observation that the acyl-enzyme complex of temocillin has reduced thermal stability compared with ticarcillin. Furthermore, we explore temocillin’s mechanism of β-lactamase inhibition with a high-resolution complex structure of CTX-M-14 class A serine β-lactamase. The results suggest that the α-methoxy group prevents hydrolysis by locking the compound into an unexpected conformation that impedes access of the catalytic water to the acyl-enzyme adduct.

INTRODUCTION

Pseudomonas aeruginosa is a Gram-negative bacterium that is intrinsically resistant to many different classes of antibiotics. It is a common nosocomial pathogen, with most serious infections occurring in immunocompromised and cystic fibrosis patients. Extensive antibiotic usage has worsened susceptibility to conventional antimicrobial agents, and multidrug-resistant strains are becoming increasingly prevalent in health care settings (1, 2). Given the dearth of new effective antibiotics, clinicians have been forced to search for older, “forgotten” drugs that are unaffected by existing resistance mechanisms. One of these drugs, temocillin (6-α-methoxy-ticarcillin), is an extended-spectrum carboxypenicillin that was first developed in the 1980s (Fig. 1) (3).

FIG 1.

Chemical structures of ticarcillin, its α-methoxy analogue temocillin, and other α-methoxy β-lactam antibiotics.

Like other β-lactams, temocillin achieves its bactericidal effect by forming a covalent adduct with the transpeptidase domains of penicillin-binding proteins (PBPs), which are periplasmic enzymes that assemble the peptidoglycan cell wall (4–7). By irreversibly inhibiting the enzymatic action of these enzymes, β-lactams prevent cross-linking of the peptide stem of peptidoglycan, disrupting the integrity of the cell wall, causing cell lysis and death. Each bacterial species has multiple PBPs, many of which are functionally redundant; however, certain PBPs are essential, and these vary by organism. For example, in Escherichia coli, the transpeptidase activities of PBP2 and PBP3 (corresponding to PBP2 and PBP3 in P. aeruginosa, respectively) are both essential (8–15). In comparison, PBP3 alone is essential for P. aeruginosa growth, making it an important drug target for this bacterial pathogen (10, 16).

While β-lactam antibiotics have served as a cornerstone for antibiotic therapy since their clinical implementation in the 1940s, their efficacy is seriously challenged by the emergence of β-lactamases, enzymes that hydrolytically deactivate this class of antibiotic (4, 17–19). One of the strategies to improve the stability of β-lactam drugs against these enzymes includes direct chemical modification of the C-6 (penicillin) or C-7 (cephalosporin) position of the ring (20), as seen in temocillin, moxalactam, and cefoxitin. This chemical function confers resistance to serine β-lactamases (SBLs) by either displacing or obstructing the hydrolytic water, preventing access to the labile ester covalent adduct (20).

Among the known β-lactams that inhibit P. aeruginosa growth is the non-α-methoxy counterpart of temocillin, ticarcillin. Marketed as Timentin (GlaxoSmithKline), ticarcillin is formulated with the β-lactamase inhibitor clavulanate and is a historically useful treatment for P. aeruginosa infections (21–23). However, clavulanate can induce the expression of the chromosomally encoded AmpC class C β-lactamase in P. aeruginosa and does not attenuate the action of AmpC (24, 25). Therefore, temocillin, an antibiotic that is a direct analogue of ticarcillin and is resistant to AmpC and other SBLs should, in theory, function as a monotherapeutic antipseudomonal that can treat SBL-producing strains of P. aeruginosa. However, while temocillin is effective against many other Gram-negative bacteria, such as Enterobacteriaceae, it is a poor inhibitor of P. aeruginosa growth, with MICs ranging from 128 to 256 mg/liter (26, 27).

The disparity in the bactericidal profiles of temocillin and ticarcillin is perplexing, and the reason for the lack of activity of temocillin against P. aeruginosa is unknown. It has been suggested that the efflux pump MexAB-OprM might produce high-level resistance by expelling temocillin from the cell (28). Impaired PBP binding has also been theorized to influence the efficacy of temocillin. Biochemical analysis has shown that reducing the incubation temperature and time appears to improve the binding affinities of temocillin to E. coli PBPs, especially PBP3, suggesting the formation of unstable complexes (29). This echoes other studies demonstrating that a 7-α-methoxy substitution of cephalosporins can have detrimental effects on the binding of β-lactams to some E. coli PBPs, including PBP2 and PBP3 (30). However, no such studies have been performed for P. aeruginosa PBP3 and temocillin, and the molecular and structural basis for the influence of α-methoxy group on β-lactam binding remains largely unknown. Here, we seek to investigate the origin of temocillin’s inferior activity against P. aeruginosa. Considering these previous studies that found that temocillin and other α-methoxy β-lactams have impaired inhibition against E. coli PBP3, coupled with PBP3’s essential role in P. aeruginosa, we were primarily interested in characterizing the binding behavior of temocillin against this protein using biochemical and melting temperature assays. X-ray crystallography experiments were also used to visualize binding poses against PBP3. Furthermore, to compare with PBP3 binding and to illustrate its stability against serine β-lactamase hydrolysis, we elucidate temocillin’s inhibition mechanism of the CTX-M-14 class A β-lactamase using a high-resolution crystal structure.

RESULTS AND DISCUSSION

Compared to ticarcillin, temocillin has a similar on-rate constant but significantly greater off-rate constant for PBP3.

Bocillin FL, a fluorescent analogue of penicillin V, is a useful and validated reporter molecule for measuring inhibition against PBPs (31). Previous studies have found that when bocillin binds to PBPs, a corresponding increase in fluorescence anisotropy occurs; this relationship has been leveraged to quantify reaction kinetics with fluorescence polarization anisotropy (FPA) assays (32, 33). We determined the rate constants for PBP3 with temocillin and ticarcillin using an FPA assay developed previously (33). The results from these experiments are shown in in Table 1 and Fig. 2. Temocillin and ticarcillin exhibit similar initial FPA signal increases, resulting in similar fitted on-rate constant (k_on) values of 77,000 M⁻¹ s⁻¹ and 22,000 M⁻¹ s⁻¹, respectively. As expected for ticarcillin, fixing the off-rate constant (k_off) value to zero produces an accurate fit, suggesting that ticarcillin is covalently bound to PBP3 and not proceeding through the hydrolysis pathway significantly within the time course of the assay (1 h). This observation is consistent with prior results with meropenem, aztreonam, and ceftazidime (33). In contrast, the FPA signal unexpectedly increases when temocillin is tested in the FPA assay (Fig. 2B), indicating that bocillin is replacing the temocillin that initially acylated PBP3. Using these data, we found the fitted k_off value to be 0.001 s⁻¹. This is at least 20-fold higher than the upper limit for the ticarcillin k_off value of <5 × 10⁻⁵ s⁻¹. These results suggest that temocillin initially acylates PBP3 but is subsequently hydrolyzed, similar to the enzymatic turnover mechanism of β-lactamases, albeit at a much lower rate. Indeed, PBPs are known to have modest hydrolytic activity (34, 35). Hydrolyzed β-lactam antibiotics have previously been captured in the active site of P. aeruginosa PBP3, including azlocillin, cefoperazone, and penicillin (36, 37). However, it is paradoxical that a 6-α-methoxy β-lactam would be hydrolyzed at a higher rate than would one lacking this group, since the 6-α-methoxy group normally increases stability against β-lactamases by hindering the deacylation step of the hydrolysis reaction (20).

TABLE 1.

Binding kinetics of temocillin and ticarcillin with P. aeruginosa PBP3 determined by FPA assay

Constant	Ticarcillin without α methoxy	Temocillin with α methoxy	Fold change temocillin/ticarcillin
kon (M−1 s−1)	22,000	77,000	3.5
koff (s−1)	<0.00005	0.001	>20

FIG 2.

Fitting of FPA progress curves for the inhibition of P. aeruginosa PBP3. (A) Ticarcillin. (B) Temocillin.

High-resolution acyl-enzyme complex crystal structures of PBP3 with temocillin and ticarcillin reveal minor, but significant, dissimilarities in their binding modes.

X-ray crystallography was used to visualize the intermolecular interactions and ligand binding poses of temocillin and ticarcillin in the PBP3 active site in order to determine the underlying factors that contribute to the ≥20-fold differences in deacylation rates for these two antibiotics. We solved the complex structures of PBP3 with unambiguous electron density corresponding to a single conformation of temocillin and ticarcillin, determined at resolutions of 2.25 Å and 1.90 Å, respectively (Table 2 and Fig. 3A and B). Continuous electron density is observed between Ser294 and the opened ring of temocillin and ticarcillin, confirming covalent adduct formation. Overall, the conformations of the two acyl-enzymes are extremely similar and superimpose, with a root mean square deviation (RMSD) of 0.477 Å for all protein atoms. Since both ligands are penicillins, the binding modes are unsurprisingly like those of most other β-lactam antibiotics. The β-lactam carbonyl oxygen is placed in the oxyanion hole formed by the backbone amide groups of Ser294 and Thr487. The C-3 carboxylate group of the thiazolidine ring forms classical interactions with the highly conserved KT(S)G motif; in this instance, hydrogen bonds with Ser485 and Thr487 and an electrostatic interaction with Lys484. The amide linker establishes hydrogen bonds with the Thr487 backbone carbonyl group and the side chain of Asn351, part of the highly conserved SXN motif (38, 39). These favorable interactions are important for ligand binding and facilitate acylation by positioning the β-lactam ring near the catalytic serine, Ser294. A hydrogen bond is also observed between Ser349 and the ring nitrogen. In addition, the side chains of temocillin and ticarcillin are positioned in a nearly identical fashion as well. The thiophene ring is placed in an aromatic pocket formed by residues Phe533, Tyr532, and Tyr503, while the side-chain carboxylate forms a salt bridge with Arg489 and a hydrogen bond with Tyr409.

TABLE 2.

Crystallographic data collection and refinement statistics

Parameter	Data by structurea
	PBP3 plus temocillin	PBP3 plus ticarcillin	CTX-M-14 plus temocillin
Data collection statistics
Space group	P21 21 21	P21 21 21	P21
Cell dimensions
a, b, c (Å)	67.13, 81.71, 88.08	68.79, 83.26, 89.22	45.11, 106.46, 47.67
α, β, γ (°)	90.00, 90.00, 90.00	90.00, 90.00, 90.00	90.00, 101.92, 90.00
Resolution (Å)	50.00–2.25 (2.29–2.25)	50.00–1.90 (2.00–1.90)	46.64–1.30 (1.37–1.30)
Rmerge (%)	0.067 (0.683)	0.130 (0.643)	0.081 (0.581)
<I/σ(I)>	14.4 (2.1)	6.5 (2.0)	10.0 (2.0)
Completeness (%)	97.2 (96.3)	86.5 (89.4)	99.9 (99.2)
Redundancy	4.0 (3.8)	5.3 (5.2)	3.7 (3.6)
Refinement statistics
Resolution (Å)	33.57–2.25 (2.34–2.25)	45.59–1.90 (1.97–1.90)	46.64–1.30 (1.35–1.30)
No. reflections/free	22,304/1,059	35,082/1,772	10,860/5,540
Rwork/Rfree	0.182/0.244	0.212/0.276	0.134/0.171
Heavy atoms (no.)	4,065	4,123	4,787
Protein	3,834	3,802	4,020
Ligand/ion	27	32	54
Water	204	289	713
B-factors (Å2)
Protein	38.43	35.58	10.64
Ligand/ion	31.47	31.34	12.74
Solvent	38.60	38.34	25.28
RMS deviations
Bond length (Å)	0.014	0.016	0.012
Bond angle (°)	1.86	1.96	1.59
Ramachandran favored (%)	96.59	96.18	97.68
Ramachandran allowed (%)	3.41	3.61	1.93
Ramachandran outliers (%)	0.00	0.20	0.39

^aValues in parentheses represent highest resolution shells.

FIG 3.

Acyl-enzyme complex crystal structures of P. aeruginosa PBP3. Unbiased F_o-F_c maps, shown in gray, are contoured at 3 σ. Hydrogen bonds between the protein and ligand are shown as dashed lines. The hydrogen bond between Tyr409 and the ligand side-chain carboxylate group is present in both complexes but not visible in the current orientation. (A) Temocillin complex at 2.25-Å resolution. The ligand and protein are colored in orange and green, respectively. (B) Ticarcillin complex at 1.90-Å resolution. The ligand and protein are colored in magenta and cyan, respectively. (C) Comparison of ligand binding poses of temocillin-PBP3 (orange/green) with ticarcillin-PBP3 (magenta/cyan). The hydrogen bonds with temocillin and ticarcillin are shown in orange and magenta, respectively. The α-methoxy group disrupts a high-quality hydrogen bond formed between temocillin and Asn351 of the SXN motif, increasing its distance by 0.47 Å.

Perhaps the most important distinction between the binding modes of ticarcillin and temocillin is the hydrogen bond distance with Asn351. The Asn of SXN in PBPs is almost universally involved in binding to the amide moiety of β-lactam antibiotics and therefore likely to mediate an important interaction in inhibitor binding (39). Furthermore, this residue plays a critical role in substrate recognition, as seen in a complex structure of E. coli PBP6 with a d-amino acid substrate fragment, where Asn of the SXN motif is hydrogen bonding with the amide backbone of this peptide (40). In ticarcillin, the distance between its amide oxygen and the side-chain nitrogen of Asn351 is 2.68 Å, only a 0.04-Å divergence from the optimal 2.72 Å distance described by Pauling et al. (41). Conversely, the α-methoxy group of temocillin projects toward Asn351, causing the amide oxygen to be distanced by 0.47 Å to a total of 3.15 Å (Fig. 3C). While this distance may seem minor, it is the furthest observed between a published structure of PBP3 and a β-lactam antibiotic. Of the 6 structures with β-lactam drugs (noncarbapenems), the average distance between the amide oxygen and Asn351 is 2.75 Å, with none exceeding 2.84 Å (aztreonam) (36, 37, 42, 43).

The disruption of this key hydrogen bond is likely due to the placement of the 6-α-methoxy group in an already-crowded active site, which prevents the amide oxygen from closely engaging Asn351. Although this is partially compensated by the formation of a new and likely weak hydrogen bond between Asn351 and the 6-α-methoxy group, we hypothesize that the bulkiness of this moiety, combined with its rotational degree of freedom, may cause further destabilization of the active site. There are small atomic movements of certain protein residues to avoid steric clashes with the 6-α-methoxy group, most notably, Asn351 (∼0.5 Å) and Tyr407 (∼0.5 Å), in direct contact with the methoxy group, as well as Arg489 (∼1 Å) and Tyr503 (∼0.6 Å) interacting with the ligand distal side chain. These movements result in some relatively close contacts that may indicate minor steric clashes, such as 3.6 Å between the methoxy group and Tyr407Cδ1 and 3.7 Å between Tyr407Cδ1/Cε1 and Tyr409Cδ2. The destabilization effects are more clearly reflected in the disparity between the isotropic B-factors of the active site residues, an indicator of disorder. When taking the B-factors of atoms located within 4.5 Å of the ligand and normalizing them to the overall B-factor of their respective structure, we find that those for PBP3 with temocillin are 3.7% higher (B within 4.5 Å, 39.90 Ų; total B, 38.42 Ų), whereas the B-factors for PBP3 with ticarcillin are 9.2% lower (B within 4.5 Å, 32.27 Ų; total B, 35.55 Ų). Additionally, compared with the 2.87-Ų difference in the overall B-factors of the two complex structures, this disparity is more evident in the side-chain atoms of selective first-shell and second-shell ligand-binding residues, especially those near the protein surface rather than the core, such as Tyr503 (60.01 versus 47.04 Ų comparing the structure of temocillin with that of ticarcillin, respectively), Arg489 (85.63 versus 55.84 Ų, respectively), Tyr328 (72.78 versus 40.99 Ų, respectively), and Asn501 (77.53 versus 64.41 Ų, respectively). Taken together, these values indicate that the active-site residues of PBP3 with temocillin, compared to ticarcillin, may be more flexible.

No water molecules could be identified in the active site of either structure, suggesting that the hydrolytic water that degrades the temocillin-PBP3 complex is not coordinated in the current complex, unlike class A β-lactamases. In fact, the observed conformation of Tyr409 blocks the access of any water molecule to the acyl-enzyme bond and forms a hydrogen bond with the temocillin side-chain carboxylate group. A likely explanation is that by destabilizing the active site, temocillin reduces the energy barrier for the protein to adopt alternative conformations and allows noncoordinated water molecules from the bulk solvent to gain facile access to the acyl-enzyme attachment. In addition, the side-chain carboxylate group can potentially activate the water molecule, similar to substrate-assisted hydrolysis observed in other β-lactams (44–48). It is possible that the Tyr409 side chain swings away from the active site to an “out” conformation in solution, as observed in PBP3 with β-lactams containing an aminothiazole group (42, 43, 49–51), opening a path for a water molecule to approach the acyl-enzyme linkage and possibly allowing the water to replace the Tyr409 hydroxyl group and hydrogen bond with the temocillin carboxylate.

Compared to temocillin, ticarcillin significantly improves the thermal stability of PBP3.

To further investigate the hypothesis that the 6-α-methoxy group may destabilize the active site of P. aeruginosa PBP3, we conducted a thermal shift assay using circular dichroism. By adding temocillin or ticarcillin in molar excess and allowing sufficient time for adduct formation, the differences between the unfolding energy of the apo and covalent complexes can be used to quantify the stabilizing noncovalent interactions (Fig. 4) (52). The melting temperature (T_m) of apo PBP3 was found to be 49.10°C. Incubation with temocillin increased the T_m by 4.42°C (53.52°C), while ticarcillin increased the T_m by 7.93°C (57.03°C). This experiment demonstrates that ticarcillin has more favorable noncovalent interactions than does temocillin. When taken with the biochemical and structural data of this study, it provides evidence that temocillin forms a relatively unstable complex with PBP3 compared with ticarcillin, leading to slow but appreciable hydrolysis and dissociation from the active site.

FIG 4.

Thermal shift experiment for temocillin and ticarcillin with P. aeruginosa PBP3. Secondary structure was monitored by circular dichroism at 222 nm from 40°C to 65°C. PBP3 T_m, 49.10°C; PBP3-temocillin T_m, 53.52°C; PBP3-ticarcillin T_m, 57.03°C.

The acyl-enzyme complex structure of temocillin with CTX-M-14 reveals an unexpected mechanism of β-lactamase stability.

The mechanism by which the α-methoxy group inhibits SBL hydrolysis has been previously analyzed using X-ray crystallography with the cephalosporins cefoxitin and moxalactam (53–55). However, the cephalosporin and penicillin cores are different, and it is unknown if the α-methoxy group interferes with SBL hydrolysis in an identical manner between these two classes of β-lactam compounds. Here, we determined a 1.30-Å resolution structure of temocillin with CTX-M-14, a class A β-lactamase (Table 2 and Fig. 5) (56, 57). In addition to being a plasmid-encoded β-lactamase that is commonly found in β-lactam-resistant bacterial isolates, including P. aeruginosa, CTX-M enzymes are excellent models for understanding the structural basis of class A β-lactamase catalysis and inhibition, since crystals are readily obtained and diffract to very high resolutions (<1.5 Å) (58–61). In addition, as ancestral enzymes, PBPs share a high degree of homology with SBLs, particularly with class A β-lactamases (7, 62, 63). Indeed, many of the active-site residues are strictly conserved, such as the SXN and KT(S)G motifs and the SXXK sequence for ligand acylation (38, 39). The most significant difference between the two families of enzymes, however, is a highly conserved Glu, the catalytic base of class A β-lactamases that activates the adduct-degrading hydrolytic water. Due to the interactions with Glu166 and the nearby Asn170, the catalytic water is well coordinated and present in nearly all crystal structures of class A β-lactamases, unlike PBPs. Other than Glu166 and the catalytic water, the overall active-site architectures are remarkably similar between PBPs and class A β-lactamases. It is therefore unsurprising that the captured conformation of temocillin with CTX-M-14 is like that seen in PBP3. The opened lactam carbonyl forms hydrogen bonds with the backbone amide groups of Ser70 and Ser237 that constitute the oxyanion hole. The C-3 carboxylate of the temocillin thiazolidine ring establishes electrostatic interactions or water-mediated contacts with the positively charged side chains of Lys234 and Arg276 and hydrogen bonds with Thr235 and Ser237. Unlike PBP3, CTX-M-14 lacks an aromatic region in the distal region of its active site, and therefore, hydrogen bonding interactions of temocillin’s side-chain carboxylate group with Asn104 and Asn132 are the principal determinants for the conformation of its side chain, resulting in important differences in the binding modes, as detailed below. This conformation is further stabilized by intramolecular interactions between the thiophene and thiazolidine rings and by intermolecular contacts with Tyr105.

FIG 5.

Acyl-enzyme complex crystal structure of temocillin with CTX-M-14 β-lactamase. An unbiased F_o-F_c map, determined at 1.30-Å resolution, is contoured at 3 σ. The ligand and protein are shown in orange and pink, respectively. Hydrogen bonds between the ligand and protein are shown as dashed lines.

Comparing the complex of CTX-M-14 to PBP3, the placement of temocillin’s carboxylate near Asn132 and Asn104 forces the carbonyl group of the amide to swing into the protein active site and form a hydrogen bond with the catalytic water rather than the hydrogen bond with Asn351 in PBP3 (Asn132 in CTX-M-14) (Fig. 5 and 6A). This in turn causes the α-methoxy group to rotate outward to avoid steric clashes with the carbonyl oxygen while forming favorable nonpolar interactions with Tyr105 and a weak hydrogen bond with Lys73. We posit that temocillin’s amide carbonyl group stabilizes the position of the catalytic water and prevents the water molecule from accessing the acyl-enzyme linkage due to steric exclusion. Interestingly, the conformation of temocillin’s amide linker and α-methoxy group is similar to a previous complex structure of moxalactam with BEL-1, another class A β-lactamase (Fig. 6B), but different from that observed with cefoxitin and CTX-M-9, where the α-methoxy group is oriented toward the catalytic water and Glu166, and the amide linker forms a hydrogen bond with Asn132 and Asn104 (Fig. 6C) (53, 55). CTX-M-9 differs from CTX-M-14 only by a single V231A mutation outside the active site, and its structure and catalytic activity are nearly identical to those of CTX-M-14 (64). In CTX-M-9, the cefoxitin α-methoxy group displaces the hydrolytic water molecule by 1 Å compared with transition state analogue structures while simultaneously serving as a steric barrier between the water and the acyl-enzyme linkage (53).

FIG 6.

Comparison of acyl-enzyme complex crystal structures with α-methoxy β-lactams. (A) Temocillin with CTX-M-14. (B) Moxalactam with BEL-1 (PDB ID 5EUA). (C) Cefoxitin with CTX-M-9 (PDB ID 1YMX). The α-methoxy groups of temocillin and moxalactam are oriented toward Tyr105 (CTX-M-14)/Tyr97 (BEL-1), while in cefoxitin, the α-methoxy group points inward and displaces the hydrolytic water.

The influence of the carboxylate group on the temocillin side-chain conformation mirrors previous observations in the complex structure of moxalactam with BEL-1. Like temocillin, moxalactam also has an arylcarboxylate side chain. This carboxylate group forms hydrogen bonding interactions with Asn125 and water-mediated contacts with the backbone amide groups of Tyr97 and Glu96, as well as the Glu96 side chain (Fig. 6B). It should be noted that the low pH of the crystallization buffer (pH 5.6) might promote a neutral state of the adjacent carboxyl groups of moxalactam and/or Glu96, thus reducing the potential electrostatic repulsion between them. Meanwhile, the phenol ring of the moxalactam side chain establishes π stacking interactions with an alternate conformation of Tyr97 (Tyr105 in CTX-M) (Fig. 6B). Like temocillin, the placement of the moxalactam arylcarboxylate group results in similar conformations of the amide carbonyl group and the α-methoxy moiety. In addition, for both temocillin and moxalactam, even though the carbonyl group is directly responsible for hindering the catalytic water’s attack on the acyl-enzyme bond, the α-methoxy group plays an important role in locking the acyl-enzyme complex in a nonproductive configuration through its interaction with Tyr97 (BEL-1) or Tyr105 (CTX-M-14). In the absence of an α-methoxy group, the side chain can rotate around the C-6/7 carbon and prevent such a clash between the carbonyl group and the catalytic water, permitting the formation of the tetrahedral intermediate and its subsequent collapse. Therefore, when comparing the acyl-enzyme complex structures of temocillin, moxalactam, and cefoxitin, it is noteworthy that although the α-methoxy substituent confers resistance to class A β-lactamases, their manner of involvement is dissimilar and related to how their respective side chains interact with the active site (Fig. 6). However, in contrast to PBP3, where temocillin binding resulted in active-site residues exhibiting higher temperature factors than the average of the protein, all three compounds with the α-methoxy group appear to form stable complexes with class A β-lactamases, with those residues within 4.5 Å of the compound showing temperature factors 16 to 25% lower than the average temperature factors of the whole protein (for CTX-M-14, temocillin, B within 4.5 Å, 7.79 Ų, and total B, 10.39 Ų; for CTX-M-9, cefoxitin, B within 4.5 Å, 10.43 Ų, and total B, 12.41 Ų; for BEL-1, moxalactam, B within 4.5 Å, 21.34 Ų, and total B, 25.48 Ų). These observations likely reflect the favorable interactions between the protein and small molecule due to the fact that these compounds were developed and selected for their ability to inhibit class A β-lactamases.

The addition of an α-methoxy moiety to the β-lactam ring protects this class of antibiotics from SBL hydrolysis. However, our biochemical and biophysical analyses have demonstrated that the α-methoxy group destabilizes the P. aeruginosa PBP3 active site, comparing temocillin with the parent β-lactam ticarcillin. These experiments provide useful insights regarding the diminished effectiveness of temocillin against P. aeruginosa, offering another possible cause alongside previous hypotheses and other potential contributing factors (28, 29). Additionally, a complex structure of temocillin with CTX-M-14, the first such complex with a class A SBL, reveals a conformation that is dissimilar from that previously seen in the structure of cefoxitin with CTX-M-9 but resembles that of moxalactam with BEL-1, revealing that the α-methoxy group has two independent but effective modes of inhibition against class A β-lactamases. The results from these studies may contribute to future inhibitor discovery targeting PBPs and SBLs.

MATERIALS AND METHODS

Protein cloning, expression, and purification.

CTX-M-14 was purified as previously described (53). A soluble construct of P. aeruginosa PBP3 (residues 50 to 579) was cloned into a pET15MHL vector at NdeI/XhoI (His₆ plus TEV protease) restriction sites. Competent Rosetta(DE3)/pLysS cells (Novagen) were then transformed with the cloned plasmid. Cells were grown at 37°C for several hours in 2× yeast extract tryptone (2xYT) medium (Sigma-Aldrich) until the optical density at 600 nm (OD₆₀₀) reached 0.6 to 0.8. Protein expression was induced with 0.5 mM isopropyl-β-d-thiogalactopyranoside (IPTG) overnight at 20°C. Cells were lysed and centrifuged, and the supernatant was loaded onto a HisTrap affinity column (GE Healthcare) and then eluted with a linear gradient of imidazole. The fractions containing the corresponding peak were pooled and buffer exchanged using an Amicon Ultra centrifugal filter (Sigma-Aldrich) to remove NaCl and imidazole. His₆-tagged TEV protease was added overnight at 4°C. The digested protein was loaded again onto the HisTrap affinity column, and the flowthrough was collected, concentrated, and then loaded to a HiLoad 16/60 Superdex 75 size exclusion column (GE Healthcare). Fractions containing PBP3 were pooled and concentrated to 6 mg/ml. The purity of the protein was >95%, as determined by SDS-PAGE.

Fluorescence polarization anisotropy assay.

The P. aeruginosa PBP3 fluorescence polarization anisotropy assay was adapted from previously reported methods (33). Briefly, reactions were performed in 200 μl at ambient temperature (∼22°C) in buffer consisting of 100 mM Tris-HCl (pH 7.0) and 0.01% Triton X-100 using 96-well black polystyrene plates. First, the test compound (ticarcillin or temocillin) at 10× the final concentration was serially diluted in water (2-fold dilutions) down the row of the plate. Then, bocillin FL (Life Technologies/Invitrogen) was added to the test compound to achieve a final concentration of 30 nM and mixed by pipetting. Assays were initiated with a PBP3-buffer-Triton X-100 solution to achieve a final apparent PBP3 concentration of 90 nM. Each assay was performed in triplicate, and 12 compound concentrations were tested. Parallel and perpendicular fluorescence intensities were measured 300 times at 10-s intervals using an Envision plate reader (PerkinElmer). The anisotropy was calculated and analyzed using Global Kinetic Explorer (KinTek), as described previously (33). The PBP reactions follow the scheme below:

$$\text{E+B}\underset{k_{-\text{1}}}{\overset{k_{\text{1}}}{\rightleftharpoons }}\text{E.B}\underset{k_{-\text{2}}}{\overset{k_{\text{2}}}{\rightleftharpoons }}\text{E-B}\underset{k_{-\text{3}}}{\overset{k_{\text{3}}}{\rightleftharpoons }}\text{E+P}\hspace{0.17em}\text{E+I}\underset{k_{-\text{1}}}{\overset{k_{\text{1}}}{\rightleftharpoons }}\text{E.I}\underset{k_{-\text{2}}}{\overset{k_{\text{2}}}{\rightleftharpoons }}\text{E-I}\underset{k_{-\text{3}}}{\overset{k_{\text{3}}}{\rightleftharpoons }}\text{E+H}$$

In this full model, bocillin (B) or the β-lactam (I) binds to PBP3 (E), forming a noncovalent complex (E.B or E.I, respectively), at which point bocillin and the β-lactam acylate PBP3. This acyl-enzyme species (E-B or E-I, respectively) is then hydrolyzed to reform the apo enzyme and the hydrolyzed product (P or H, respectively). Since these reactions are effectively irreversible and the intermediates are indistinguishable, reaction constants can be simplified according to the model presented below:

$$\text{E+B}\stackrel{k_{\text{B}}}{\to }\text{E-B}\hspace{0.17em}\text{E+I}\stackrel{k_{\text{on}}}{\to }\text{E-I}\stackrel{k_{\text{off}}}{\to }\text{E+H}$$

Here, k_on describes the aggregate rate constant of noncovalent binding and bond formation for the β-lactam, while k_off represents the deacylation rate constant. Furthermore, while bocillin does deacylate from PBP3, the inclusion of this rate constant has no significant effect on data fitting and can be excluded, yielding a final model with on-rate constant k_B, or the rate constant of bocillin binding and covalent bond formation. Anisotropy progress curves were fitted according to this simplified model.

Protein crystallization.

Apo crystals of P. aeruginosa PBP3 were grown using a hanging drop apparatus at 20°C. A well solution containing 20% polyethylene glycol 3350 (PEG 3350), 0.2 M calcium acetate, and 6 mg/ml PBP3 was mixed in equal volumes, producing square-like crystals in the P2₁2₁2₁ form. For complex structure determination, fully grown crystals were transferred into a drop containing well solution with 2 mM temocillin or ticarcillin for 2 h before cryoprotection with 27.5% PEG 3350, 0.2 M calcium acetate, and 15% glycerol with subsequent flash freezing.

Large crystals of CTX-M-14 were obtained in a hanging drop apparatus at 20°C using 1.2 M potassium phosphate (pH 8.3) as the crystallization solution in a 1:1 drop ratio. For complex structure determination, CTX-M-14 crystals were transferred to drops with crystallization buffer containing 2 mM temocillin for 2 h and cryoprotected with 30% sucrose and 1.8 M potassium phosphate (pH 8.3) before being flash frozen with liquid nitrogen.

Structure determination.

Data for the PBP3 temocillin complex were collected on the SBC 19-BM beamline at the Advanced Photon Source (APS) in Argonne, IL, and processed with the HKL2000 software suite (65). Data for the PBP3 ticarcillin complex were collected on the GMCA 23-ID beamline at the APS and processed with the iMOSFLM (66). Data for the CTX-M-14-temocillin complex were collected on the SER-CAT 22-BM and processed with iMOSFLM. The CCP4 versions of PHASER and MOLREP were used for molecular replacement using a previously solved apo PBP3 structure (PDB ID 3PBN) and apo CTX-M-14 structure (PDB ID 1YLT) as reference models (67, 68). Rigid and restrained refinements were performed using REFMAC and model building with COOT (69, 70). Protein structure figures were made using PyMOL (Schrödinger, LLC).

Thermal stability assay.

Secondary structure was monitored using circular dichroism (CD) with a Jasco J-815 CD spectropolarimeter coupled to a Peltier cell holder. PBP3 was diluted to 2 μg/ml in 50 mM sodium phosphate (pH 7.0) and incubated with 15 μM temocillin or ticarcillin for 20 min. CD spectra were measured at 222 nm from 40°C to 65°C to monitor melting curves. The data were analyzed using a two-state fitting program with the program SigmaPlot.

Data availability.

The structures have been deposited in the Protein Data Bank with accession numbers 6UN1 for P. aeruginosa PBP3 plus temocillin, 6UN3 for P. aeruginosa PBP3 plus ticarcillin, and 6UNB for CTX-M-14 plus temocillin.