Adding Insult to Injury: Mechanistic Basis for How AmpC Mutations Allow Pseudomonas aeruginosa To Accelerate Cephalosporin Hydrolysis and Evade Avibactam

Pseudomonas aeruginosa is a leading cause of nosocomial infections worldwide and notorious for its broad-spectrum resistance to antibiotics. A key mechanism that provides extensive resistance to β-lactam antibiotics is the inducible expression of AmpC β-lactamase. Recently, a number of clinical isolates expressing mutated forms of AmpC have been found to be clinically resistant to the antipseudomonal β-lactam–β-lactamase inhibitor (BLI) combinations ceftolozane-tazobactam and ceftazidime-avibactam.

ABSTRACT

Pseudomonas aeruginosa is a leading cause of nosocomial infections worldwide and notorious for its broad-spectrum resistance to antibiotics. A key mechanism that provides extensive resistance to β-lactam antibiotics is the inducible expression of AmpC β-lactamase. Recently, a number of clinical isolates expressing mutated forms of AmpC have been found to be clinically resistant to the antipseudomonal β-lactam–β-lactamase inhibitor (BLI) combinations ceftolozane-tazobactam and ceftazidime-avibactam. Here, we compare the enzymatic activity of wild-type (WT) AmpC from PAO1 to those of four of these reported AmpC mutants, bearing mutations E247K (a change of E to K at position 247), G183D, T96I, and ΔG229–E247 (a deletion from position 229 to 247), to gain detailed insights into how these mutations allow the circumvention of these clinically vital antibiotic-inhibitor combinations. We found that these mutations exert a 2-fold effect on the catalytic cycle of AmpC. First, they reduce the stability of the enzyme, thereby increasing its flexibility. This appears to increase the rate of deacylation of the enzyme-bound β-lactam, resulting in greater catalytic efficiencies toward ceftolozane and ceftazidime. Second, these mutations reduce the affinity of avibactam for AmpC by increasing the apparent activation barrier of the enzyme acylation step. This does not influence the catalytic turnover of ceftolozane and ceftazidime significantly, as deacylation is the rate-limiting step for the breakdown of these antibiotic substrates. It is remarkable that these mutations enhance the catalytic efficiency of AmpC toward ceftolozane and ceftazidime while simultaneously reducing susceptibility to inhibition by avibactam. Knowledge gained from the molecular analysis of these and other AmpC resistance mutants will, we believe, aid in the design of β-lactams and BLIs with reduced susceptibility to mutational resistance.

INTRODUCTION

Pseudomonas aeruginosa is a leading opportunistic pathogen notorious for establishing chronic and often fatal nosocomial infections in immunocompromised cancer patients and burn victims (1). It also contributes significantly to the frequent occurrence and spread of chronic respiratory infections in cystic fibrosis patients (2). Not surprisingly, the morbidity and mortality rates associated with this organism are among the highest of all clinically relevant Gram-negative pathogens (3). This is attributed, in part, to its extensive intrinsic multidrug resistance profile, a feature that has helped to establish it as one of the World Health Organization’s top three priority pathogens for research and discovery of new antibiotics (4).

β-Lactam antibiotics are among the suite of drugs against which P. aeruginosa demonstrates resistance (1). This is accomplished primarily through the expression of AmpC (or Pseudomonas-derived cephalosporinase 1 [PDC-1]), a broad-spectrum Ambler class C (5) β-lactamase that hydrolyzes most β-lactams at their diffusion limit (Fig. 1) (6). Chromosomal expression of this cephalosporinase is tightly regulated by the Gram-negative peptidoglycan (PG) recycling pathway and is triggered by the presence of β-lactams (7). Specifically, β-lactams disrupt PG metabolism, altering the relative concentrations of PG metabolites, ultimately leading to derepression of ampC and subsequent production of the AmpC enzyme (8, 9). Furthermore, spontaneous mutations in key PG recycling enzymes can promote constitutive hyperproduction of AmpC, which in turn strengthens resistance to β-lactams (6). Mutations in ampD (AmpD amidase) (10) and dacB (penicillin binding protein 4 [PBP 4]) (11) are the most common examples. Mutational disruption of the PG recycling pathway is just one of several antibiotic resistance mechanisms employed by P. aeruginosa (12, 13).

FIG 1.

Representation of P. aeruginosa AmpC β-lactamase (PDC-1) (PDB code 4HEF) (58). The amino acids defining the ceftolozane-tazobactam- and ceftazidime-avibactam-resistant AmpC mutants investigated in this study are located in the magenta area. Avibactam is depicted as green sticks, and the conserved catalytic residues that compose the active site/binding pocket are depicted as gray sticks. The numbering of all residues represents their position in the immature form of the AmpC protein (i.e., signal peptide included) as described by Mack et al. (31).

The discovery of β-lactamases necessitated the development of β-lactamase inhibitors (BLIs), such as clavulanic acid, sulbactam, and tazobactam (14). These small-molecule compounds are coadministered with specific β-lactam antibiotics and, by inhibiting chromosomal and plasmidic β-lactamases, indirectly enhance the efficacy of these drugs (14). Like β-lactam antibiotics, most BLIs contain a β-lactam ring in their structure (15) and inhibit β-lactamases through the formation of unfavorable acyl-enzyme products that irreversibly abolish their activity (15). This effectively prevents hydrolysis of the antibiotic, allowing it to disrupt peptidoglycan anabolism and facilitate cell lysis via PBP inhibition (16).

Unfortunately, most BLIs are poor inhibitors of AmpC (17). In fact, the clinically relevant BLIs clavulanic acid, tazobactam, and sulbactam have even been shown to induce ampC expression (17). One exception is the recently FDA-approved diazabicyclooctane (DBO) non-β-lactam BLI, avibactam (18). Unlike many other BLIs in clinical use, avibactam does not contain a β-lactam ring; however, the DBO moiety—a five-membered cyclic urea—contains an electrophilic carbonyl carbon analogous to that of the β-lactam ring (18). Therefore, covalent acylation via ring opening is a step common to the inhibition mechanisms of both avibactam and β-lactam-based BLIs (18). The key mechanistic difference between avibactam and β-lactam-based BLIs lies in the deacylation step and, ultimately, the fate of the inhibitor. β-Lactam-based BLIs accomplish β-lactamase inhibition by acylating the enzyme through a one-way kinetically irreversible pathway that is often followed by eventual chemical decomposition of the inhibitor off the acyl-enzyme species (14). Compared to β-lactam-based BLIs, avibactam rapidly acylates its target enzyme and deacylates at a much lower rate. In addition, the ring-opened avibactam adduct to the enzyme deacylates through slow regeneration of its native structure, thereby avoiding inhibitor degradation (18). This ability, in addition to the stable acyl-enzyme complex it forms with β-lactamases, makes avibactam remarkably effective at potentiating the efficacy of coadministered β-lactam antibiotics. Furthermore, its affinity for AmpC β-lactamases broadens its spectrum of activity over that of the classical BLIs. Indeed, avibactam is the first β-lactamase inhibitor with clinically useful inhibitory activity against AmpC (19).

Commercially, avibactam is paired with ceftazidime, while tazobactam accompanies ceftolozane (Fig. 2). Structurally, ceftolozane and ceftazidime differ only in their R2 (3-position) side chains (20). Ceftazidime harbors a pyridine functional group, which provides increased antipseudomonal activity compared to the activities of older cephalosporins (21, 22). Conversely, the R2 side chain of ceftolozane is a bulkier 2-methyl-3-aminopyrazolium moiety containing a 2-aminoethylureido group, which appears to introduce steric hindrance between ceftolozane and the enzyme active site, thereby reducing its susceptibility to hydrolysis (23, 24). Despite their structural differences, ceftazidime and ceftolozane are both extended-spectrum cephalosporins and poor inducers of ampC expression, making them attractive treatment options for complex P. aeruginosa infections (25). Indeed, these β-lactam–BLI combinations are among the most effective forms of combination therapy used to combat infections caused by multidrug-resistant (MDR) P. aeruginosa (25).

FIG 2.

Chemical structures of the β-lactam–β-lactamase inhibitor combinations relevant to this study: (a) ceftazidime-avibactam; (b) ceftolozane-tazobactam. The R1 side chain, β-lactam nucleus, and R2 side chain are highlighted in green, blue, and red, respectively.

Unfortunately, clinical isolates of P. aeruginosa expressing mutant forms of AmpC (i.e., PDC-X) that confer resistance to ceftolozane-tazobactam and cross-resistance to ceftazidime-avibactam have already been reported (26, 27), indicating that the selection of resistance mutations in response to these drug combinations is occurring. The AmpC mutations include substitution mutations in the omega (Ω) loop (28) or in neighboring residues that interact with it (26–29). The Ω loop borders the R1 region (so named because it accommodates the R1 side chains of β-lactams) (6) of the enzyme active site. Intriguingly, several Ω loop deletion mutations, ranging from 5 to 21 amino acids, have also been reported by studies examining AmpC-mediated cephalosporin resistance (28, 30), further revealing that this structural motif is a mutational hot spot for generating cephalosporin resistance (6). Here, we investigate clinically selected AmpC mutations associated with the Ω loop, namely, E247K (PDC-221), G183D (PDC-322), T96I (PDC-222), and ΔG229–E247 (PDC-223) (26, 27, 31). The numbering of these residues is consistent with the structural-alignment-based numbering of class C β-lactamases, or SANC, scheme, proposed by Mack et al. (31). Note that the respective SANC positions of these AmpC mutations are E219, G156, T70, and ΔG202–E219.

Although these AmpC mutants have been shown to be associated with significant reductions in the susceptibility of P. aeruginosa to ceftolozane-tazobactam and ceftazidime-avibactam (26, 27), the molecular mechanism underlying this drug resistance is poorly understood (28, 32). With the exception of E247K (33), these AmpC mutants have only been characterized by microbiological profiling and molecular modeling (26–28, 34, 35). For this reason, we have carried out a detailed study of the catalytic properties of all of these mutant AmpC enzymes in order to provide a quantitative description of the effects of each mutation on antibiotic turnover and inhibitor binding. In this work, we show that all mutants demonstrate enhanced substrate specificity toward both ceftolozane and ceftazidime relative to that of the wild-type (WT) AmpC enzyme PDC-1, and we suggest that this is related to increased enzyme flexibility facilitated by reduced enzyme stability. We also demonstrate that the apparent rate of AmpC acylation by avibactam is reduced significantly for each mutant. Ultimately, we concluded that by increasing both the rate of β-lactam deacylation and the apparent activation barrier of enzyme acylation, these mutations allowed AmpC to hydrolyze ceftolozane and ceftazidime with improved catalytic activity while simultaneously evading inhibition by avibactam. Herein, we provide a detailed kinetic analysis of common AmpC mutations to further elucidate their role in facilitating resistance to β-lactam–BLI combination therapy in P. aeruginosa.

RESULTS AND DISCUSSION

Hydrolysis of ceftolozane and ceftazidime by WT AmpC.

The hydrolysis of ceftolozane and ceftazidime by wild-type (WT) AmpC was analyzed using Michaelis-Menten kinetics (Fig. S1 and S2 in the supplemental material). Table 1 lists the kinetic parameters of the WT and mutant AmpC enzymes with respect to ceftolozane and ceftazidime. The k_cat values of WT AmpC for these antibiotics differ from one another by a factor of 2, with ceftolozane having the larger turnover. Moreover, the Michaelis constant (K_m) of WT AmpC for ceftolozane is 6-fold larger than that for ceftazidime. This notable difference in the enzyme affinities toward ceftolozane and ceftazidime reflects the distinctive properties of their respective R2 side chains (36, 37). Specifically, the 2-methyl-3-aminopyrazolium side chain of ceftolozane contains an additional positive charge in the form of a basic 2-aminoethyluredio moiety (pK_a = 7.95) (Fig. 2) (20). As a consequence of its larger bulk and increased net charge, ceftolozane interacts less favorably with the AmpC active site, thereby lowering the magnitude of binding free energy compared to that of ceftazidime. On balance, WT AmpC demonstrates a modest difference in its specificities toward ceftolozane and ceftazidime, as suggested by their respective k_cat/K_m values (Table 1). It is also clear from the results shown in Fig. 3 that WT AmpC hydrolyzes both ceftolozane and ceftazidime rather poorly.

TABLE 1.

Michaelis-Menten kinetic parameters of WT and mutant AmpC enzymes for ceftolozane, ceftazidime, and nitrocefin^a

Enzyme, parameter	Mean value ± SE for indicated substrateb
	Ceftolozane	Ceftazidime	Nitrocefin
WT
kcat (s−1)	(1.04 ± 0.13) × 10−2	(4.85 ± 0.55) × 10−3	32.48 ± 3.2
Km (μM)	2,107.3 ± 341.2	353.5 ± 81.1	126.12 ± 23.63
kcat/Km (μM−1 s−1)	(4.91 ± 1.01) × 10−6	(1.37 ± 0.35) × 10−5	(2.58 ± 0.54) × 10−1
Fold changec	1	1	1
E247K
kcat (s−1)	1.48 ± 0.25	(6.58 ± 0.40) × 10−1	1.37 ± 0.18
Km (μM)	3,610.2 ± 704.3	1,016.9 ± 90.3	62.86 ± 17.67
kcat/Km (μM−1 s−1)	(4.11 ± 1.05) × 10−4	(6.47 ± 0.70) × 10−4	(2.18 ± 0.67) × 10−2
Fold change	83.7	47.2	0.085
G183D
kcat (s−1)	(1.80 ± 0.17) × 10−1	ND	1.25 ± 0.07
Km (μM)	1,174.3 ± 165.1	ND	73.53 ± 8.34
kcat/Km (μM−1 s−1)	(1.53 ± 0.26) × 10−4	(1.41 ± 0.02) × 10−4	(1.71 ± 0.22) × 10−2
Fold change	31.2	10.3	0.066
T96I
kcat (s−1)	NDd	ND	1.29 ± 0.05
Km (μM)	ND	ND	46.22 ± 4.51
kcat/Km (μM−1 s−1)	(1.99 ± 0.04) × 10−4	(2.27 ± 0.04) × 10−4	(2.78 ± 0.29) × 10−2
Fold change	40.5	16.6	0.107
ΔG229–E247
kcat (s−1)	ND	ND	1.00 ± 0.23
Km (μM)	ND	ND	222.66 ± 80.67
kcat/Km (μM−1 s−1)	(1.22 ± 0.02) × 10−4	(1.38 ± 0.02) × 10−4	(4.50 ± 1.93) × 10−3
Fold change	24.8	10.1	0.017

^aMichaelis-Menten kinetic parameters were obtained from fitting the respective data in Fig. S1, S2, and S3 in the supplemental material to the Michaelis-Menten equation.

^bExperiments were performed in triplicate; plus-minus values represent the standard errors reported from regression analysis.

^cRatio of enzyme specificity constant (k_cat/K_m) of mutant AmpC relative to that of WT AmpC.

^dND, not determined.

FIG 3.

Hydrolysis of 200 μM ceftolozane (a) and ceftazidime (b) by WT and mutant AmpC enzymes measured under conditions described in Materials and Methods. The AmpC enzymes are represented as follows: blue, WT; red, E247K; yellow, G183D; green, T96I; and orange, ΔG229–E247.

These differences notwithstanding, the activity of WT AmpC toward ceftolozane and ceftazidime is still quite low compared to its activity toward older cephalosporins (e.g., the k_cat/K_m of WT AmpC toward cephalothin is 5.1 × 10⁻¹μM⁻¹ s⁻¹) (38). This further highlights the increased protection against β-lactamase-mediated hydrolysis afforded by the continued structural modification of cephalosporins. One such optimization common to both ceftazidime and ceftolozane (but missing from older cephalosporins like cephalothin) is the oxime moiety, which contributes to the hydrophilic R1 group and offers increased protection against β-lactamases (Fig. 2) (24). Indeed, the R1 side chain interacts unfavorably with the conserved Ω loop residues Val239 and Tyr249 (SANC positions 211 and 221, respectively), thereby preventing ceftolozane and ceftazidime from adopting catalytically competent conformations in the enzyme active site (39). Ultimately, as ceftolozane and ceftazidime are designed to resist hydrolysis by β-lactamases, it is not surprising that WT AmpC exhibits poor catalytic activity toward them.

AmpC mutants demonstrate increased catalytic efficiency toward ceftolozane and ceftazidime.

The data in Fig. 3 compare the catalytic activities of WT AmpC toward ceftolozane and ceftazidime with those of each AmpC mutant. It is clear that, while WT AmpC exhibits low catalytic activity toward both ceftolozane and ceftazidime, all four AmpC mutants hydrolyze these antibiotics with highly improved efficiency. The enzymatic properties of each of these mutants in relation to ceftolozane and ceftazidime hydrolysis were analyzed using Michaelis-Menten kinetics, and the results are displayed in Table 1. Figure S1 depicts the Michaelis-Menten plots of the WT and mutant AmpC enzymes toward ceftolozane. The ceftolozane hydrolysis rates of the WT enzyme and the E247K and G183D mutants demonstrate hyperbolic concentration dependence, while those of T96I and ΔG229–E247 are linear. Due to this observed linearity, only the specificity constant (k_cat/K_m) can be calculated from the Michaelis-Menten plots of the T96I and ΔG229–E247 mutants; however, it is clear that the K_m values of these mutants for ceftolozane must be larger than 3,610 μM (40). Therefore, with the exception of the G183D mutant, these AmpC mutations appear to weaken the enzyme’s substrate affinity. Given that the R2 moiety of ceftolozane has a significant net positive charge at neutral pH (20), it is not surprising that the G183D mutation enhances the enzyme’s affinity toward ceftolozane.

The k_cat values in Table 1 indicate that these AmpC mutations also affect the enzyme turnover rate. Although this parameter cannot be directly determined for the T96I and ΔG229–E247 mutants, based on the K_m limiting value of 3,610 μM, it is likely to be larger than 0.4 s⁻¹ for them. In other words, all four mutations increase the rate of ceftolozane turnover by at least an order of magnitude relative to that of WT AmpC. It must be pointed out that the peak concentration of ceftolozane in the plasma during therapy is in the order of 100 mg liter⁻¹ (41) and that this is much lower than the K_m of WT AmpC for this antibiotic. Moreover, the concentration of this antibiotic in the vicinity of the infecting P. aeruginosa cells is likely even lower than it is in the plasma. That is to say, under therapeutic conditions, the AmpC enzyme is operating at the low substrate concentration limit, and its activity is therefore best represented by the specificity constant (k_cat/K_m) (40). As shown by the results in Table 1, the catalytic efficiency of all four AmpC mutants toward ceftolozane was increased by more than 1 order of magnitude relative to that of WT AmpC, with the E247K mutant demonstrating the largest increase. Therefore, the AmpC mutant enzymes have achieved greater catalytic activity toward ceftolozane by enhancing turnover and sacrificing substrate affinity.

The Michaelis-Menten plots of the WT and mutant AmpC enzymes toward ceftazidime are depicted in Fig. S2. The rates of ceftazidime hydrolysis exhibit hyperbolic behavior only for the WT enzyme and the E247K mutant, and their respective K_m values are 353 μM and 1,016 μM (Table 1). The K_m values of the G183D, T96I, and ΔG229–E247 mutants must therefore be larger than 1,016 μM, indicating that all the AmpC mutations reduce enzyme affinity toward ceftazidime considerably. It is clear, however, that these mutations have also led to a significant increase in enzyme turnover with respect to ceftazidime. With a K_m limiting value of 1,016 μM, the k_cat values of these three mutants toward ceftazidime are likely to be greater than 0.14 s⁻¹, which is well over an order of magnitude larger than the k_cat value of WT AmpC. Like ceftolozane, the peak concentration of ceftazidime in the plasma is also in the order of 100 mg liter⁻¹ (42); this is around half the ceftazidime K_m that was determined for WT AmpC. Therefore, the specificity constant (k_cat/K_m) is also the best parameter for comparing the changes observed in ceftazidime hydrolysis activities between the WT and mutant AmpC enzymes. As shown by the results in Table 1, the catalytic efficiencies of all four AmpC mutants toward ceftazidime had also increased by more than 1 order of magnitude relative to that of the WT enzyme, with the E247K mutant again displaying the greatest enhancement. Finally, comparing the kinetic parameters in Table 1 reveals that these mutations affect the hydrolysis of ceftolozane and ceftazidime by AmpC in a very similar manner, despite the fact that they arose only in response to ceftolozane exposure. This is not particularly surprising, however, since the core structure (and R1 side chain) of ceftazidime is analogous to that of ceftolozane (Fig. 2) (20).

The reactivities of WT and mutant AmpC enzymes toward nitrocefin.

The hydrolysis of nitrocefin by the WT and mutant AmpC enzymes was also analyzed using Michaelis-Menten kinetics (Fig. S3), and the results are listed in Table 1. These data show that nitrocefin is a much better substrate for AmpC than either ceftolozane or ceftazidime. Indeed, with the exception of ΔG229–E247, the mutant AmpC enzymes show modest increases in affinity toward nitrocefin relative to that of WT AmpC. Conversely, all of the AmpC mutations reduce enzyme turnover by more than an order of magnitude. The net result of these competing contributions is revealed by the specificity constants of the AmpC mutants, which show at least an order of magnitude reduction in nitrocefin hydrolysis activity compared to that of WT AmpC. This is in clear contrast with the results obtained for ceftolozane and ceftazidime hydrolysis, where the AmpC mutants demonstrate a significant increase in activity compared to that of WT AmpC. The disparate effects of these mutations on enzyme activity can only be expected if the rate-determining step of the nitrocefin hydrolysis mechanism is different from that of ceftolozane or ceftazidime. This will be discussed in more detail below.

AmpC mutants demonstrate reduced susceptibility to inhibition by avibactam.

To investigate the susceptibility of the AmpC mutants toward acylation by avibactam, we compared the effect of avibactam on nitrocefin hydrolysis by each of the AmpC mutants with that of WT AmpC. Specifically, we monitored nitrocefin hydrolysis by the WT and mutant AmpC enzymes in the presence of various concentrations of avibactam and fitted the resulting kinetic profiles to equation 5 (see Materials and Methods) to obtain the pseudo first-order rate constant, k_obs, for each reaction. (Figure 4 depicts typical time courses of the hydrolysis of 100 μM nitrocefin catalyzed by 5 nM WT AmpC and in the presence of 1.25 μM, 2.5 μM, 5 μM, and 10 μM avibactam.) The k_obs values obtained from these fits were then plotted as a function of the avibactam concentration (Fig. 5a and b). The resulting linear dependence (see equation 6 in Materials and Methods) allowed estimation of the apparent rate constants for inhibitor dissociation and binding ($k_{\text{off}}^{\text{app}}$ and $k_{\text{on}}^{\text{app}}$, respectively) associated with the avibactam inhibition mechanism for each AmpC enzyme, as shown in Table 2. The $k_{\text{on}}^{\text{app}}$ and $k_{\text{off}}^{\text{app}}$ values determined for avibactam inhibition of WT AmpC are in agreement with previous measurements reported by Ehmann et al. (43). Intriguingly, the data in Table 2 clearly indicate that all the AmpC mutants experience a reduction in avibactam inhibition relative to the results for WT AmpC. Indeed, each of these mutations has caused at least an order of magnitude drop in the $k_{\text{on}}^{\text{app}}$ rate and around an order of magnitude increase in the $k_{\text{off}}^{\text{app}}$ rate. The overall effect of these rate changes is most clearly demonstrated by the changes in the apparent avibactam inhibition binding constant, $K_{\text{avi-binding}}=k_{\text{off}}^{\text{app}}/k_{\text{on}}^{\text{app}}$, which informs the binding affinity of avibactam for AmpC. That is, these mutations reduce the inhibitory potency of avibactam from the nanomolar to the micromolar concentration scale.

FIG 4.

Typical time courses of the hydrolysis of 100 μM nitrocefin by 5 nM WT AmpC in the presence of 0 μM (blue), 1.25 μM (green), 2.5 μM (yellow), 5 μM (orange), and 10 μM (red) avibactam measured under conditions described in Materials and Methods. The fits of each trace to equation 5 are shown in black. A trace representing 100 μM nitrocefin in the absence of avibactam and WT AmpC is included as a negative control (black).

FIG 5.

Plots of k_obs versus avibactam concentration for WT (a) and mutant (b) AmpC enzymes measured under conditions described in Materials and Methods. Plots of k_obs versus tazobactam concentration for WT (c) and mutant (d) AmpC enzymes measured under conditions described in Materials and Methods. Data points represent average values from three technical replicates, and error bars represent standard deviations. The lines represent the best linear correlations. The AmpC enzymes are represented as follows: blue, WT; red, E247K; yellow, G183D; green, T96I; and orange, ΔG229–E247.

TABLE 2.

Acylation kinetics of WT and mutant AmpC enzymes by avibactam and tazobactam^a

Enzyme, parameter	Mean value ± SE for indicated inhibitorb
	Avibactam	Tazobactam
WT
$k_{\text{on}}^{\text{app}}$ (M−1 s−1)	(6.03 ± 1.21) × 103	(1.96 ± 0.27) × 102
$k_{\text{off}}^{\text{app}}$ (s−1)	(2.7 ± 6.90) × 10−4	(5.0 ± 0.31) × 10−3
$K_{I\text{-binding}}$ (M)c	(4.48 ± 11) × 10−8	(2.55 ± 0.35) × 10−5
R2	0.9907	0.9938
E247K
$k_{\text{on}}^{\text{app}}$ (M−1 s−1)	(7.35 ± 2.27) × 102	(4.56 ± 0.78) × 102
$k_{\text{off}}^{\text{app}}$ (s−1)	(2.5 ± 0.53) × 10−3	(5.3 ± 0.70) × 10−3
$K_{I\text{-binding}}$ (M)	(3.40 ± 1.30) × 10−6	(1.16 ± 0.25) × 10−5
R2	0.9945	0.9909
G183D
$k_{\text{on}}^{\text{app}}$ (M−1 s−1)	(1.36 ± 0.14) × 102	(1.47 ± 0.22) × 102
$k_{\text{off}}^{\text{app}}$ (s−1)	(3.9 ± 0.36) × 10−3	(4.3 ± 0.22) × 10−3
$K_{I\text{-binding}}$ (M)	(2.87 ± 0.40) × 10−5	(2.93 ± 0.46) × 10−5
R2	0.9958	0.9926
T96I
$k_{\text{on}}^{\text{app}}$ (M−1 s−1)	(4.93 ± 0.52) × 102	(4.45 ± 0.30) × 102
$k_{\text{off}}^{\text{app}}$ (s−1)	(6.0 ± 0.69) × 10−3	(5.9 ± 0.41) × 10−3
$K_{I\text{-binding}}$ (M)	(1.22 ± 0.19) × 10−5	(1.33 ± 0.13) × 10−5
R2	0.9889	0.9952
ΔG229–E247
$k_{\text{on}}^{\text{app}}$ (M−1 s−1)	(4.04 ± 1.50) × 101	(6.28 ± 1.9) × 101
$k_{\text{off}}^{\text{app}}$ (s−1)	(1.6 ± 0.19) × 10−3	(2.5 ± 0.05) × 10−3
$K_{I\text{-binding}}$ (M)	(3.96 ± 1.54) × 10−5	(3.98 ± 1.2) × 10−5
R2	0.9962	0.9997

^aAcylation kinetics were obtained by fitting the respective data in Fig. 5 to equation 6.

^bExperiments were performed in triplicate; plus-minus values represent the standard errors reported from regression analysis.

^cApparent inhibition binding constant for inhibitor I (avibactam or tazobactam); plus-minus values represent the standard errors.

For comparative purposes, we also measured the effects of these AmpC mutations on the inhibitory properties of tazobactam, the BLI coadministered with ceftolozane (Fig. 5c and d). As shown by the K_tazo-binding values in Table 2, none of the mutations change the tazobactam inhibitory potency by more than a factor of two relative to that of WT AmpC. This further supports the poor inhibitory activity of tazobactam toward AmpC and suggests that ceftolozane was the selective pressure guiding the mutational resistance of AmpC with respect to this β-lactam–BLI combination. Moreover, these observations are consistent with the results of a previous study in which amino acid substitutions in the class C β-lactamase from Acinetobacter baumannii (ADC-7) were found to significantly reduce the inhibitory potency of avibactam, but not tazobactam, toward this enzyme (44).

The kinetic properties of the AmpC mutant enzymes are consistent with the MICs.

To relate the kinetic properties of the AmpC mutant enzymes to the antibiotic resistance of a bacterial cell, MICs were determined. Table 3 lists the MICs of the P. aeruginosa strain PAO1 ampC knockout mutant (PAΔC) complemented with WT ampC and the ampC mutants for ceftazidime (in the presence and absence of avibactam) and ceftolozane (in the presence and absence of tazobactam). It is evident that each mutant AmpC enzyme increases the MICs of strain PAO1 for these cephalosporins, both alone and in combination with their BLIs, relative to the MIC of WT AmpC. This resistance can be related to the kinetic properties of the mutant AmpC enzymes through the following simple formalism. When a bacterium is placed in growth medium containing antibiotic at concentration [M], the concentration of this antibiotic within the bacterium, [C], can be assumed to follow equation 1, as follows:

$$\frac{d[C]}{\mathit{dt}}=k_{\text{in}}[M]–k_{\text{out}}[C]–k_{\text{rxn}}[C]$$ (1)

where k_in is the average rate constant of transport (passive or active) of antibiotic from the medium to the cell interior, k_out is the average rate constant for the efflux of antibiotic from the cell interior into the surrounding medium, and $k_{\text{rxn}}=\frac{k_{\text{cat}}}{K_m}[E_{\text{total}}]$ is the rate of antibiotic hydrolysis by enzyme E. Eventually, a steady state is established where the concentration of antibiotic within the medium and within the bacterium are related in the following manner:

$$[M]=\frac{k_{\text{out}}+\frac{k_{\text{cat}}}{K_m}[E_{\text{total}}]}{k_{\text{in}}}[C]$$ (2)

Once the concentration of antibiotic within the bacterium surpasses the intracellular concentration of antibiotic required to inhibit growth of the bacterium (C_critical), its growth is inhibited; this corresponds to an [M] value that is equivalent to the MIC. In terms of equation 2, this becomes:

$$\text{MIC}=\frac{k_{\text{out}}+\frac{k_{\text{cat}}}{K_m}[E_{\text{total}}]}{k_{\text{in}}}[C_{\text{critical}}]$$ (3)

Although this model of antibiotic influx is rather crude, it provides a useful qualitative description of the role that β-lactam hydrolysis plays in bacterial susceptibility. Indeed, it is clear from equation 3 that the MIC is strongly influenced by $\frac{k_{\text{cat}}}{K_m}[E_{\text{total}}]$: as this value increases, so does the MIC. If we assume that the parameters k_in, k_out, and [E_total] do not vary greatly between all the complemented PAΔC strains, then variations between their MICs should follow the same trends as our experimentally determined k_cat/K_m values. The ceftolozane MICs for the complemented strains can be ranked as follows: WT < G183D < ΔG229–E247 ≈ T96I < E247K. This roughly follows the trend in ceftolozane hydrolysis efficiency of the AmpC enzymes: WT < ΔG229–E247 ≈ G183D < T96I < E247K (Table 1). Similarly, the ceftazidime MICs for the complemented strains can be ranked as follows: WT < G183D < ΔG229–E247 ≈ T96I < E247K. Again, this trend roughly follows that for ceftazidime hydrolysis efficiency of the AmpC enzymes: WT < ΔG229–E247 ≈ G183D < T96I < E247K. Taken together, these results suggest that the heightened catalytic efficiency of the mutant AmpC enzymes is a key contributor to the reduced susceptibility of the P. aeruginosa strains toward the cephalosporin antibiotics ceftolozane and ceftazidime.

TABLE 3.

Susceptibility profiles of the PAO1 ampC knockout mutant PAΔC complemented with WT ampC and the ampC mutants identified in ceftolozane-tazobactam-resistant P. aeruginosa clinical isolates as described previously^a

Strain	MIC (mg/liter)b
	CAZ	CZA	TOL	C/T
PAO1	2	2/4	0.25	0.25/4
PAΔC	≤1	1/4	0.25	0.25/4
PAΔC/pUCPAmpCWT (PDC-1)	16	2/4	1	0.5/4
PAΔC/pUCPAmpCE247K (PDC-221)	>64	16/4	64	32/4
PAΔC/pUCPAmpCG183D (PDC-322)	32	16/4	16	8/4
PAΔC/pUCPAmpCT96I (PDC-222)	64	8/4	32	16/4
PAΔC/pUCPAmpCΔG229–E247 (PDC-223)	64	32/4	32	16/4

^aFraile-Ribot et al. (26) and MacVane et al. (27).

^bCAZ, ceftazidime; CZA, ceftazidime-avibactam; TOL, ceftolozane; C/T, ceftolozane-tazobactam.

To continue, equation 3 can be expanded to include the effects of the respective coadministered BLIs, tazobactam and avibactam, on the MICs of the complemented strains, as follows:

$$\text{MIC}=\frac{k_{\text{out}}+\frac{k_{\text{cat}}}{K_m}\frac{[E_{\text{total}}]}{\left(1+\frac{[I]}{K_i}\right)}}{k_{\text{in}}}{C}_{\text{critical}}$$ (4)

where [I] is the inhibitor concentration and K_i is the apparent binding constant of the inhibitor.

The ceftazidime-avibactam MICs for the complemented strains are consistent with equation 4. Specifically, the largest avibactam-induced reductions in MICs are observed in the strains expressing the WT, E247K, and T96I AmpC enzymes. This is consistent with the results of the avibactam inhibition assays, which reveal that avibactam has the highest inhibitory potency toward WT AmpC, followed by the E247K and T96I point mutants (Table 2). The strains expressing the G183D and ΔG229–E247 mutant AmpC enzymes experience smaller reductions in MICs; this can also be explained by the notably higher K_avi-binding values associated with these enzymes relative to those of the E247K and T96I point mutants. Conversely, tazobactam affects the ceftolozane MICs of all the strains similarly (i.e., reduces them by 2). This is consistent with the tazobactam inhibition data, which again show that tazobactam binds the WT and mutant AmpC enzymes with similar affinities (Table 2). The observed 2-fold reduction in MICs suggests that tazobactam exhibits some antipseudomonal activity, albeit poor.

AmpC mutants have lower thermal stabilities and are more dynamic than WT AmpC.

To investigate whether the antibiotic resistance mutations in AmpC affect enzyme stability, we determined the melting temperatures (T_m) of the WT and mutant AmpC enzymes (Table 4). The T_m that we have determined for P. aeruginosa PAO1 WT AmpC (PDC-1) (55.33°C) is comparable to those measured for other WT AmpC analogues (33, 45). Interestingly, all the mutations cause a decrease in the melting point of AmpC. This change in protein melting point (ΔT_m) can be related to changes in protein folding free energy ($\Delta \Delta G_{\text{unfolding}}=\Delta G_{\text{unfolding}}^{\text{mutant}}-\Delta G_{\text{unfolding}}^{\text{WT}}$) through the equation $\Delta \Delta G_{\text{unfolding}}=\frac{\Delta H_{\text{unfolding}}}{T_m}\Delta T_m$ (46), where ΔH_unfolding is the unfolding enthalpy, which is positive during thermal denaturation (47), and T_m is the melting temperature of the WT enzyme. That is, these mutations have decreased the stability of the folded state of the AmpC enzyme, thereby increasing its flexibility and dynamic nature at room temperature. Increases in enzyme dynamics and flexibility have been shown to facilitate enzyme function by permitting substrate access to and product release from the active site (48).

TABLE 4.

Melting temperatures of WT and mutant AmpC enzymes

Enzyme	Avg Tm (°C) ± SDa	Avg ΔTm (°C) ± SEb
WT	55.33 ± 0.06	0.00 ± 0.080
E247K	45.80 ± 0.00	−9.53 ± 0.060
G183D	50.43 ± 0.15	−4.90 ± 0.200
T96I	49.23 ± 0.06	−6.10 ± 0.080
ΔG229–E247	51.77 ± 0.06	−3.56 ± 0.080

^aValues represent averages from three technical replicates.

^bChanges in T_m relative to that of WT AmpC.

Furthermore, we investigated the relationship between enzyme flexibility and enzyme activity for WT AmpC and for the E247K, G183D, and T96I point mutants. Single point mutations have very small effects on the ΔH_unfolding of an enzyme (49); therefore, the ΔΔG_unfolding of each of these point mutants is proportional to ΔT_m. The log of specificity constants of the WT and point mutant AmpC enzymes for ceftolozane, ceftazidime, and nitrocefin can therefore be related to the respective changes in protein melting point (Fig. 6). It can be clearly seen from the results in Fig. 6 that, in contrast to nitrocefin hydrolysis, the ability of AmpC to hydrolyze ceftolozane and ceftazidime increases uniformly as the enzyme becomes more dynamic. This observation also confirms that the rate-determining step of the ceftolozane or ceftazidime catalytic cycle is (i) different from that of nitrocefin and (ii) likely coupled to a protein dynamic process.

FIG 6.

Plot of the log of specificity constants (k_cat/K_m) of the WT and point mutant AmpC enzymes against their ΔT_m as defined in the text. The substrates are represented as follows: blue, ceftolozane; red, ceftazidime; and black, nitrocefin. The lines represent the best linear fits to the ceftolozane (R² = 0.909) and ceftazidime (R² = 0.985) data.

The effects of AmpC mutations on the enzyme catalytic cycle.

β-Lactam hydrolysis proceeds by a mechanism involving acylation and deacylation of the β-lactamase (50). Acylation occurs when the catalytic serine launches a nucleophilic attack on the β-lactam carbonyl carbon, forming a high-energy acylation intermediate. This leads to opening of the β-lactam ring and the formation of a low-energy acyl-enzyme complex. Deacylation occurs when a catalytic water attacks this acyl-enzyme complex, facilitating the formation of a high-energy deacylation intermediate that promotes the release of the inactive β-lactam product from the enzyme. This can be represented by scheme 1 (14), as follows, where E is enzyme (AmpC), S is the cephalosporin substrate (ceftolozane, ceftazidime, or nitrocefin), and P is product (i.e., hydrolyzed/inactivated antibiotic substrate):

$$E+S\underset{k_{-1}}{\overset{k_1}{\rightleftarrows }}E:S\underset{k_{-2}}{\overset{k_2}{\rightleftarrows }}E-S\stackrel{k_3}{\to }E+P$$ (scheme 1)

(Note: We propose that the first two steps of this scheme can also represent avibactam binding to AmpC [E]; in that case, S would represent the avibactam inhibitor.) Since avibactam binding to AmpC is reversible and involves ring opening through acylation of the catalytic serine (18), the first two steps of scheme 1 can also represent avibactam inhibition. In scheme 1, two limiting cases can be considered: (i) the rate-limiting step is k₂, where the specificity constant is k₂/K_m, and (ii) the rate-limiting step is k₃, where the specificity constant is k₃/K_m. In case (i), the AmpC mutations should affect the apparent activation energies of the β-lactam hydrolysis reaction and the avibactam binding rates in a similar fashion. To test this, we generated a log-log plot of the specificity constants of the WT and mutant AmpC enzymes for ceftolozane, ceftazidime, and nitrocefin as a function of the avibactam $k_{\text{on}}^{\text{app}}$ (Fig. 7). It is clear that only the effects of the AmpC mutations on the hydrolysis of nitrocefin are correlated with avibactam binding, suggesting that k₂ is the rate-limiting step for nitrocefin hydrolysis. The lack of correlation between avibactam binding and the breakdown of ceftolozane and ceftazidime also indicates that the rate-limiting step of the ceftolozane or ceftazidime catalytic cycle is different from that of nitrocefin. This suggests that k₃ is the rate-limiting step for the breakdown of ceftolozane and ceftazidime. This is consistent with results obtained by Chow et al. (51) for the β-lactamase from Mycobacterium tuberculosis, BlaC, where k₂ was found to be the rate-limiting step for nitrocefin hydrolysis, while k₃ was the rate-limiting step for the breakdown of cefoxitin, a cephalosporin similar to ceftazidime.

FIG 7.

Log-log plot of the specificity constants (k_cat/K_m) of the WT and mutant AmpC enzymes as a function of avibactam $k_{\text{on}}^{\text{app}}$ as defined in the text. The substrates are represented as follows: blue, ceftolozane; red, ceftazidime; and black, nitrocefin. The line represents the best linear fit to the nitrocefin data (R² = 0.927).

Concluding remarks.

The unfortunate emergence of P. aeruginosa clinical isolates overexpressing mutated forms of AmpC is compromising the efficacy of the antipseudomonal β-lactam–BLI combinations ceftolozane-tazobactam and ceftazidime-avibactam. Although these isolates have been characterized by MICs, the AmpC mutations defining them had not been described in detail. We hypothesized that these mutations were enhancing the hydrolytic activity of AmpC toward ceftolozane and ceftazidime. We were surprised to find that not only did these mutations increase the catalytic efficiency of AmpC toward ceftolozane and ceftazidime, they also reduced the enzyme’s susceptibility to inhibition by avibactam. Therefore, the E247K, G183D, T96I, and ΔG229–E247 mutations appear to exert a 2-fold effect on the catalytic cycle of AmpC. First, they increase the apparent activation barrier of the enzyme acylation step, thereby significantly reducing the inhibitory potency of avibactam. This increase in activation energy, however, does not affect the catalytic turnover of ceftolozane and ceftazidime significantly, as the rate-limiting step for the breakdown of these antibiotic substrates is the very slow k₃ deacylation/hydrolysis step. Second, these mutations decrease the stability of the folded state of AmpC, thereby increasing its flexibility. This appears to accelerate deacylation of the β-lactam, resulting in the larger catalytic efficiencies toward ceftolozane and ceftazidime compared to the catalytic efficiency of WT AmpC. It is the contribution of these two effects that allows each AmpC mutant to evade avibactam inhibition while hydrolyzing cephalosporins with enhanced efficiency.

MATERIALS AND METHODS

IPTG (isopropyl-β-d-thiogalactopyranoside) was purchased from Life Technologies; EDTA, phenylmethylsulfonyl fluoride (PMSF), DNase I, SigmaFast EDTA-free protease inhibitor cocktail, bovine serum albumin (BSA), nitrocefin, and ceftazidime were purchased from Sigma-Aldrich; sodium citrate, monosodium phosphate, disodium phosphate, sodium chloride, magnesium chloride, HEPES, and glycerol were purchased from Fisher Scientific; and ceftolozane and tazobactam were provided by Merck & Co., Inc. (Kenilworth, NJ, USA).

Cloning and site-directed mutagenesis.

Wild-type (WT) and mutant ampC open reading frames were amplified by PCR from previously constructed pUCP24 plasmids (26) (WT [PDC-1; accession number NG_049865]; E247K [PDC-221; accession number MF481212]; T96I [PDC-222; accession number MF481213]; and ΔG229-E247 [PDC-223; accession number MF481214]). The forward (5′-GATATcatatgGATCGCCTGAAGGC-3′ [NdeI cut site in lowercase letters]) and reverse (5′-CTATActcgagTCACAGGCCGC-3′ [XhoI cut site in lowercase letters]) oligonucleotide primers used for amplification were designed to exclude DNA encoding the N-terminal signal peptide and the 10 C-terminal residues, respectively. The resulting amplicons therefore encode amino acids D32 to L387 (52). The purified amplicons were directionally cloned into the NdeI/XhoI restriction sites of the pET24b(+) expression vector (Novagen).

The G183D point mutant was generated by site-directed mutagenesis using the Q5 site-directed mutagenesis kit (New England Biolabs [NEB]) and the WT AmpC pET24b(+)-based expression vector as the template DNA. The forward and reverse oligonucleotide primers used to introduce this mutation were 5′-GATCTGTTCGGCTATCTCGCCG-3′ and 5′-GATGCTCGGGTTGGAATAGAGGC-3′, respectively. The fidelity of each construct was confirmed by Sanger sequencing at The Centre for Applied Genomics (Toronto, Ontario).

Susceptibility testing.

The MICs of ceftazidime, ceftazidime plus 4 mg/liter avibactam, ceftolozane, and ceftolozane plus 4 mg/liter tazobactam were determined in triplicate experiments by broth microdilution according to EUCAST recommendations (www.eucast.org) for ampC-deficient P. aeruginosa PAO1 derivatives expressing the different AmpC mutant enzymes from pUCP24-based expression plasmids constructed as previously described (26, 32).

Expression and purification of WT and mutant AmpC enzymes.

WT and mutant AmpC enzymes were expressed and purified using a standardized Escherichia coli-based recombinant protein expression system for use in kinetic studies. Plasmids bearing either WT or mutant ampC were used to transform CaCl₂-competent E. coli BL21-Gold(DE3) for subsequent recombinant protein expression. Transformed E. coli cells were grown aerobically at 37°C in Terrific broth (TB) supplemented with 70 μg/ml kanamycin to an approximate optical density at 600 nm (OD₆₀₀) of 0.8. AmpC expression was then induced with 1 mM IPTG, and the cultures were grown overnight at 18°C with shaking. Cells were harvested via centrifugation (3,440 × g) at 4°C and frozen at −80°C for >24 h before resuspension in lysis buffer (20 mM sodium citrate [pH 5.2], SigmaFast EDTA-free protease inhibitor cocktail, 1 mM EDTA [pH 8.0], 1 mM PMSF, 50 μg/ml DNase I, 5 mM MgCl₂) and subsequent lysis via sonication. Lysate was clarified via high-speed centrifugation (15,400 × g) at 4°C for 30 min and then applied to a 12-ml cation exchange column containing SP Sepharose high-performance resin (GE Healthcare) preequilibrated with 20 mM sodium citrate buffer (pH 5.2). The resin was washed with 20 mM sodium citrate buffer (pH 5.2), and AmpC was subsequently eluted over a linear gradient of 0 to 0.5 M NaCl in the same buffer at 4°C.

Fractions containing AmpC were identified by SDS-PAGE analysis, and protein concentrations (mg/ml) were determined using a NanoDrop One (ThermoFisher) by measuring the absorbance at λ = 280 nm and using the theoretical mass and extinction coefficient (ε) of the protein (ε₂₈₀ = 54,320 M⁻¹cm⁻¹ for WT, E247K, G183D, and T96I; ε₂₈₀ = 52,830 M⁻¹cm⁻¹ for ΔG229–E247). The ProtParam tool of the ExPASy Bioinformatics Resource Portal was used to obtain the theoretical masses and extinction coefficients of these enzymes (53). Fractions containing purified AmpC were pooled and dialyzed overnight against 20 mM HEPES buffer (pH 7.5) at 4°C. The sample was then concentrated down to ∼2 ml using a 10-kDa-molecular-weight cutoff (MWCO) centrifugal filter unit, clarified via centrifugation (16,000 × g) for 15 min, and applied to a 5-ml HiTrap heparin column (GE Healthcare) preequilibrated with 20 mM HEPES buffer (pH 7.5) to remove excess nucleic acid. The resin was washed with 20 mM HEPES buffer (pH 7.5), and AmpC was subsequently eluted over a linear gradient of 0 to 0.4 M NaCl in the same buffer. Fractions containing AmpC were identified by SDS-PAGE, and the concentration and purity were determined as described above. Fractions containing AmpC at a reasonable concentration and purity were pooled and concentrated to ∼5 mg/ml using a 10-kDa MWCO centrifugal filter unit and clarified by centrifugation (16,000 × g) for 15 min at 4°C. Samples were diluted 1:1 in storage buffer (20 mM HEPES buffer [pH 7.5] containing 50% glycerol [final concentration, 25%]), divided into single-use 15-μl aliquots, flash frozen in liquid nitrogen, and stored at −80°C. All AmpC proteins were stable throughout the purification process, with little to no precipitation.

Determining Michaelis-Menten kinetic parameters for WT and mutant AmpC enzymes.

Kinetic assays were performed at 28°C in clear-bottom 96-well microtiter plates (far-UV-transparent Greiner [Monroe, NC, USA] plates for the ceftolozane and ceftazidime assays and Falcon [Tewksbury, MA, USA] plates for the nitrocefin assays) in buffer containing 0.1 M sodium phosphate (pH 7.0) and 0.1 mg/ml BSA. Reaction progress was monitored by absorbance using a SpectraMax iD5 plate reader (Molecular Devices, San Jose, CA, USA), and the data were collected using SoftMax Pro software. Data represent average values from at least three technical replicates that were analyzed using SigmaPlot (Systat Software, Inc., CA, USA). The β-lactamase activity assays were performed under the following conditions, and the initial slopes of the traces were determined by linear regression.

(i) Ceftolozane hydrolysis. The initial rates of ceftolozane hydrolysis were measured by monitoring absorbance changes at λ = 283 nm through challenging 2 μM WT, 0.05 μM E247K, and 0.1 μM G183D, T96I, and ΔG229–E247 AmpC enzymes with ceftolozane concentrations ranging from 0 to 800 μM for 3,600 s at 28°C. Due to the high absorptivity of ceftolozane, activities above 800 μM cannot be accurately measured. Initial rates were determined using the relationship ${\left(\frac{d[P]}{\mathit{dt}}\right)}_{t=0}=\left(\frac{1}{\Delta {\epsilon }_{283}}\right){\left(\frac{\mathit{dA}}{\mathit{dt}}\right)}_{t=0}$, where Δε₂₈₃ = −9,740 M⁻¹cm⁻¹ (54).

(ii) Ceftazidime hydrolysis. The initial rates of ceftazidime hydrolysis were measured by monitoring absorbance changes at λ = 260 nm through challenging 2 μM WT, 0.05 μM E247K, and 0.1 μM G183D, T96I, and ΔG229–E247 AmpC enzymes with ceftazidime concentrations ranging from 0 to 600 μM for 3,600 s at 28°C. Due to the high absorptivity of ceftazidime, activities above 600 μM cannot be accurately measured. Initial rates were determined using the relationship ${\left(\frac{d[P]}{\mathit{dt}}\right)}_{t=0}=\left(\frac{1}{\Delta {\epsilon }_{260}}\right){\left(\frac{\mathit{dA}}{\mathit{dt}}\right)}_{t=0}$, where Δε₂₆₀ = −8,660 M⁻¹cm⁻¹ (55).

(iii) Nitrocefin hydrolysis. The initial rates of nitrocefin hydrolysis were measured by monitoring absorbance changes at λ = 486 nm through challenging 0.250 nM WT, 5 nM E247K, G183D, and T96I, and 10 nM ΔG229–E247 AmpC enzymes with nitrocefin concentrations ranging from 0 to 200 μM for 1,200 s at 28°C. Initial rates were determined using the relationship ${\left(\frac{d[P]}{\mathit{dt}}\right)}_{t=0}=\left(\frac{1}{\Delta {\epsilon }_{486}}\right){\left(\frac{\mathit{dA}}{\mathit{dt}}\right)}_{t=0}$, where Δε₄₈₆ = 20,500 M⁻¹cm⁻¹ (51).

Avibactam and tazobactam inhibition assays.

The kinetics of acylation of the WT and mutant AmpC enzymes by avibactam or tazobactam were measured as previously described (18, 56). Briefly, hydrolysis of 100 μM nitrocefin by AmpC (at 5 nM [WT] or 100 nM [mutants]) was measured by monitoring absorbance changes at λ = 486 nm in the presence of different concentrations of avibactam (1 to 10 μM for the WT, 10 to 90 μM for the E247K mutant, 50 to 400 μM for the G183D and ΔG229–E247 mutants, and 15 to 150 μM for the T96I mutant) or tazobactam (12.5 to 150 μM) for 1,200 s at 28°C. The resulting progress curves were fitted to equation 5 using nonlinear regression to obtain the pseudo first-order rate constant, k_obs.

In the presence of both nitrocefin and avibactam, the kinetic profile of AmpC-catalyzed nitrocefin hydrolysis is expected to follow equation 5 (18, 57), as follows:

$$P=v_{s}t+(v_0–v_s)\frac{(1–e^{–k_{\text{obs}}t})}{k_{\text{obs}}}$$ (5)

where P is the amount of product formed at time t, v₀ and v_s represent uninhibited and fully inhibited enzyme velocity, respectively, and k_obs is an apparent rate constant that exhibits the following linear dependence on inhibitor concentration (18):

$$k_{\text{obs}}=k_{\text{off}}^{\text{app}}+k_{\text{on}}^{\text{app}}\frac{[I]}{\left(1+\frac{[S]}{K_m}\right)}$$ (6)

where $k_{\text{off}}^{\text{app}}$ and $k_{\text{on}}^{\text{app}}$ are the respective apparent rate constants for inhibitor dissociation and binding, [I] is the inhibitor concentration, [S] is the substrate concentration, and K_m is the Michaelis constant. The rate constants $k_{\text{off}}^{\text{app}}$ and $k_{\text{on}}^{\text{app}}$ are therefore key kinetic parameters for comparing the susceptibility of the WT and mutant AmpC enzymes toward avibactam (or tazobactam) acylation.

Melting temperature determination.

Melting temperatures (T_m, °C) of the WT and mutant AmpC enzymes were determined through thermal denaturation experiments performed in triplicate using a NanoTemper Prometheus NT.48 differential scanning fluorimeter (DSF). Briefly, 1.7 mg/ml (50 μM) of enzyme in 20 mM HEPES (pH 7.5), 25% glycerol was loaded into a capillary tube and protein denaturation was followed by monitoring the ratios of fluorescence intensities measured at 330 nm and 350 nm (excitation, 290 nm) as the sample was heated from 20°C to 95°C at a rate of 1°C min⁻¹. The inflection point of the resulting plot defines the melting temperature of the enzyme.