Impaired Inhibition by Avibactam and Resistance to the Ceftazidime-Avibactam Combination Due to the D¹⁷⁹Y Substitution in the KPC-2 β-Lactamase

ABSTRACT

The ceftazidime-avibactam antibiotic combination was recently shown to be at risk for the emergence of resistance under treatment. To gain insight into the underlying mechanism, we have analyzed the catalytic properties of a Klebsiella pneumoniae carbapenemase type 2 (KPC-2) β-lactamase harboring the D¹⁷⁹Y substitution. We show that impaired inhibition by avibactam combined with significant residual activity for ceftazidime hydrolysis accounts for the resistance. In contrast, the D¹⁷⁹Y substitution abolished the hydrolysis of aztreonam and imipenem, indicating that these drugs might provide therapeutic alternatives.

TEXT

Resistance to carbapenems in Enterobacteriaceae is often due to the production of class A β-lactamases belonging to the Klebsiella pneumoniae carbapenemase (KPC) type (1). The most commonly encountered variants worldwide are KPC-2 and KPC-3, which confer high-level resistance to most available β-lactams. The production of KPC β-lactamases is often associated with resistance to other classes of antibiotics, including aminoglycosides, fluoroquinolones, and colistin, leaving few or no therapeutic alternatives (2, 3). Inhibition of KPC enzymes by classical β-lactamase inhibitors (clavulanate, sulbactam, and tazobactam) is not sufficient to restore the activity of β-lactams (4). In this context, a new β-lactam–β-lactamase inhibitor combination, ceftazidime-avibactam (5), has recently obtained regulatory approval in the United States and Europe. The combination has a broad spectrum of activity against multidrug-resistant Enterobacteriaceae and Pseudomonas aeruginosa strains producing class A, C, and some class D β-lactamases (6, 7).

The emergence of resistance to the ceftazidime-avibactam combination has been recently reported in three out of 37 patients infected with carbapenem-resistant Enterobacteriaceae and treated with these drugs (8). The emergence of resistance, which was associated with microbiological failure, was due to amino acid substitutions in the KPC-3 enzymes produced by the K. pneumoniae isolates recovered from these three patients (D¹⁷⁹Y, V²⁴⁰G, and D¹⁷⁹Y associated with T²⁴³M) (9). The impact of these substitutions on the kinetic parameters for hydrolysis of ceftazidime and inhibition by avibactam has not been previously reported. In another study (10), site-directed mutagenesis identified rare substitutions resulting in resistance to the ceftazidime-avibactam combination. A comparison of KPC-2 and KPC-2 D¹⁷⁹N did not reveal any modification of the steady-state rate of ceftazidime hydrolysis or inhibition by avibactam (10). However, an initial burst of ceftazidime hydrolysis was detected for KPC-2 D¹⁷⁹N but not for the parental enzyme (10).

To gain insight into the modifications of the catalytic properties of KPC enzymes that lead to resistance to the ceftazidime-avibactam combination, we have isolated in vitro a mutant derived from a KPC-2-producing strain of Escherichia coli. The resulting KPC-2 enzyme, which was found to harbor the previously described D¹⁷⁹Y substitution, was kinetically characterized with respect to the hydrolysis of various β-lactams and the inhibition by avibactam.

In vitro selection for resistance to the ceftazidime-avibactam combination.

Selection was performed with E. coli TOP10 harboring plasmid pTRC-99kΩbla_KPC-2, which enables expression of the bla_KPC-2 gene encoding KPC-2 under the control of the P_trc promoter (11). Approximately 6 × 10⁹ CFU were plated on Mueller-Hinton (MH) agar (Difco) containing isopropyl-β-d-1-thiogalactopyranoside (IPTG) for the induction of P_trc, a fixed concentration of avibactam (4 μg/ml), and 2-fold-increasing concentrations of ceftazidime (1 to 32 μg/ml). CFU grew after 48 h of incubation at frequencies of 2 × 10⁻⁸, 5 × 10⁻⁹, and 1× 10⁻⁹ on agar plates containing 4, 8, and 16 μg/ml ceftazidime, respectively. Sequencing of the bla_KPC-2 gene from 10 putative mutants obtained at the highest concentrations of ceftazidime (8 or 16 μg/ml) revealed the same point mutation leading to the D¹⁷⁹Y substitution in 5 resistant clones and the wild-type sequence in the remaining clones. The plasmid encoding KPC-2 D¹⁷⁹Y was extracted and introduced by transformation into E. coli TOP10, with selection for resistance to kanamycin conveyed by the vector pTRC-99k. The resulting transformants grew in the presence of ceftazidime (8 μg/ml) and avibactam (4 μg/ml), indicating that the D¹⁷⁹Y substitution was sufficient for resistance to the drug combination, as these concentrations correspond to the susceptible breakpoints by FDA and EUCAST criteria.

Modification of the resistance phenotype resulting from the D¹⁷⁹Y substitution in KPC-2.

The MICs of β-lactams were determined by the microdilution method in MH broth, according to Clinical and Laboratory Standards Institute (CLSI) recommendations (12). Avibactam and clavulanate were used at a fixed concentration of 4 μg/ml in combination with β-lactams. IPTG (500 μM) was added to the microdilution plates to induce production of the β-lactamase. The precultures were grown in MH broth containing IPTG (500 μM) and kanamycin (50 μg/ml) for plasmid maintenance.

E. coli TOP10 producing wild-type KPC-2 was resistant to all tested β-lactams (MIC, ≧128 μg/ml), whereas the ceftazidime-avibactam combination was active (MIC, 1 μg/ml), according to FDA and EUCAST interpretive criteria (Table 1). In contrast, avibactam did not restore susceptibility to ceftazidime in the strain producing the KPC-2 D¹⁷⁹Y variant (MIC, 32 μg/ml), as expected from the selection procedure. Resistance to ceftazidime alone was not abolished by D¹⁷⁹Y (MIC, >128 μg/ml), indicating that the β-lactamase remained active in the hydrolysis of this cephalosporin. The D¹⁷⁹Y substitution decreased the MICs of ampicillin, ceftriaxone, aztreonam, and meropenem, resulting in susceptibility for aztreonam and meropenem (MICs, 0.5 and <0.12 μg/ml, respectively). Interestingly, clavulanate significantly reduced the MIC of ceftazidime against E. coli TOP10 producing KPC-2 D¹⁷⁹Y (from >128 μg/ml to 8 μg/ml). In conclusion, the D¹⁷⁹Y substitution led to the acquisition of resistance to the ceftazidime-avibactam combination and of susceptibility to aztreonam and meropenem. These results are in full agreement with previous analyses of the consequences of the D¹⁷⁹Y substitution in various genetic backgrounds (9).

TABLE 1.

Impact of D¹⁷⁹Y substitution on MICs of β-lactams against E. coli TOP10 producing a KPC-2 β-lactamase in the absence or presence of a β-lactamase inhibitor

β-Lactamase	MIC (μg/ml) of indicated β-lactams with avibactam (+Avi), with clavulanate (+Clav), or in the absence of inhibitor (none)a
	Ampicillin			Ceftazidime			Ceftriaxone	Aztreonam	Meropenem	Imipenem
	None	+Avi	+Clav	None	+Avi	+Clav	None	None	None	None
None	2	2	2	0.25	0.25	0.25	<0.12	0.25	<0.12	0.25
KPC-2	>128	128	>128	>128	1	64	>128	>128	>128	128
KPC-2 D179Y	16	8	8	>128	32	8	32	0.5	<0.12	0.25

^aInhibitors were used at a fixed concentration of 4 μg/ml. Data are the median of the results from five determinations.

Impact of the D¹⁷⁹Y substitution on the hydrolysis of β-lactams by KPC-2.

Wild-type KPC-2 and KPC-2 D¹⁷⁹Y were produced in E. coli TOP10 cells harboring pTRC-99kΩbla_KPC-2 or its derivative encoding KPC-2 D¹⁷⁹Y, as previously described (11). The β-lactamases were purified by anion-exchange and size-exclusion chromatography (11). All kinetic experiments were performed at 20°C in 2-(N-morpholino)ethanesulfonic acid (MES; 100 mM [pH 6.4]) in a Cary 300 spectrophotometer (Agilent), as previously described (11). Steady-state kinetic parameters (k_cat, K_m, and k_cat/K_m) were determined at a fixed concentration of the β-lactamase [E] and various concentrations of the substrate [S]. Initial rate (v_i) values were plotted as a function of [S], according to the equation v_i = k_cat [E][S]/(K_m + [S]). K_m values that could not be evaluated by this method were determined by analyzing competitive hydrolysis of the β-lactam and the chromogenic cephalosporin CENTA (13), as previously described (11, 14). Variations in the molar extinction coefficients (Δε) resulting from hydrolysis of the β-lactam ring were 7,380 M⁻¹ · cm⁻¹ (λ_{415 nm}) for CENTA, −500 M⁻¹ · cm⁻¹ (λ_{235 nm}) for ampicillin, −9,800 M⁻¹ · cm⁻¹ (λ_{256 nm}) for ceftazidime, −400 M⁻¹ · cm⁻¹ (λ_{318 nm}) for aztreonam, −14,500 M⁻¹ · cm⁻¹ (λ_{270 nm}) for ceftriaxone, −2,500 M⁻¹ · cm⁻¹ (λ_{227 nm}) for clavulanate, and −7,280 M⁻¹ · cm⁻¹ (λ_{298 nm}) for meropenem.

The D¹⁷⁹Y substitution severely impaired the hydrolysis of ampicillin, aztreonam, and meropenem, since hydrolysis of these β-lactams by KPC-2 D¹⁷⁹Y was not detectable at the highest enzyme concentration tested (10 μM) (Table 2). Accordingly, KPC-2 D¹⁷⁹Y did not confer resistance to aztreonam and meropenem (Table 1). For ampicillin, the production of KPC-2 D¹⁷⁹Y led to a 4-fold increase in the MIC, suggesting that the enzyme retained residual activity which could not be detected due to the low sensitivity of the in vitro assay for this antibiotic (Δε = −500 M⁻¹ · cm⁻¹ at λ_{235 nm}).

TABLE 2.

Impact of D¹⁷⁹Y substitution on kinetic parameters of KPC-2 for hydrolysis of β-lactams

Kinetic parameter by β-lactam	Resultsb
	KPC-2	KPC-2 D179Y
Ampicillin
kcat (s−1)	200 ± 10	NAa
Km (μM)	200 ± 40	NA
kcat/Km (M−1 · s−1)	(1.0 ± 0.2) × 106	NA
CENTA
kcat (s−1)	100 ± 10	0.11 ± 0.01
Km (μM)	32 ± 8	23 ± 3
kcat/Km (M−1 · s−1)	(3.1 ± 0.8) × 106	(4.8 ± 0.6) × 103
Ceftazidime
kcat (s−1)	>1.4	(1.3 ± 0.2) × 10−3
Km (μM)	>600	19 ± 4
kcat/Km (M−1 · s−1)	(3.7 ± 0.1) × 103	70 ± 20
Ceftriaxone
kcat (s−1)	>25	(7.0 ± 0.1) × 10−4
Km (μM)	>100	0.20 ± 0.03
kcat/Km (M−1 · s−1)	(2.5 ± 0.1) × 105	(3.5 ± 0.5) × 103
Aztreonam
kcat (s−1)	>350	NA
Km (μM)	>5 × 103	NA
kcat/Km (M−1 · s−1)	(6.9 ± 0.3) × 104	NA
Meropenem
kcat (s−1)	1.8 ± 0.2	NA
Km (μM)	27 ± 9	NA
kcat/Km (M−1 · s−1)	(6.7 ± 2.3) × 104	NA
Imipenem
kcat (s−1)	48 ± 5	NA
Km (μM)	66 ± 21	NA
kcat/Km (M−1 · s−1)	(7.3 ± 2.4) × 105	NA
Clavulanate
kcat (s−1)	5.2 ± 0.7	NA
Km (μM)	36 ± 4	NA
kcat/Km (M−1 · s−1)	(1.4 ± 0.3) × 105	NA

^aNA, not applicable, as hydrolysis of β-lactams (100 μM, except for aztreonam at 1,000 µM) was not detected at the highest β-lactamase concentration tested (10 μM). Under these conditions, the lower limits of detection correspond to a turnover of <2 × 10⁻³ · s⁻¹, <2.6 × 10⁻³ · s⁻¹, <1.5 × 10⁻⁴ · s⁻¹, <1.5 × 10⁻⁴ · s⁻¹, and <4 × 10⁻⁴ · s⁻¹ for ampicillin, aztreonam, meropenem, imipenem, and clavulanate, respectively.

^bData are means ± standard errors of the mean (SEM).

The D¹⁷⁹Y substitution led to a >900-fold reduction in the kinetic constant k_cat for hydrolysis of ceftazidime, ceftriaxone, and the chromogenic cephalosporin CENTA. Binding of these drugs to the β-lactamase did not appear to be impaired, since the D¹⁷⁹Y substitution also led to a large reduction in the kinetic constant K_m for ceftazidime (>30-fold) and ceftriaxone (>500-fold), whereas that of CENTA was not significantly altered. Overall, the D¹⁷⁹Y substitution had a moderate impact on the catalytic efficacy (k_cat/K_m) of ceftazidime hydrolysis (a 50-fold decrease) due to a partially compensatory effect of the large decrease in k_cat (>1,000-fold) by a decrease in K_m (>30-fold). Of note, this modification of the kinetic parameters did not prevent the expression of ceftazidime resistance (MIC, >128 μg/ml). The results obtained with ceftriaxone were qualitatively similar, although for this drug, the partial compensation of the decreases in k_cat and K_m was associated with a lower level of resistance (MIC, 32 μg/ml).

Impact of the D¹⁷⁹Y substitution on inhibition of KPC-2 by avibactam.

The rate constant for the carbamylation reaction (k₂/K_i) was evaluated based on the inhibition of CENTA hydrolysis (100 μM) by avibactam at 37°C in MES (100 mM [pH 6.4]) for a two-step reaction (Fig. 1), as previously described (11, 15). The D¹⁷⁹Y substitution led to an approximately 70,000-fold reduction in the carbamylation rate constant k₂/K_i. Thus, resistance to the ceftazidime-avibactam combination conveyed by the D¹⁷⁹Y substitution is accounted for by a drastic reduction in the efficacy of avibactam coupled to substantial residual activity for ceftazidime hydrolysis.

FIG 1.

Kinetics of KPC-2 and KPC-2 D¹⁷⁹Y inhibition by avibactam (Avi). (A) Reaction schemes. (B) Equations for rate constants. v_i and v_s are the initial and steady-state (inhibited) velocities of the reactions, respectively. [I], [P], and [S] are the concentrations of avibactam, hydrolyzed CENTA, and CENTA, respectively. K_m is the kinetic parameter for hydrolysis of CENTA in the absence of avibactam. t, time. (C) Inhibition of CENTA hydrolysis by increasing concentrations of avibactam. Values of k_obs were obtained by fitting equation 1 to progress curves. (D) Parameters k₂/K_i and k₋₂ were determined by fitting equation 2 to the data. Error bars shown are ± standard error of the mean (SEM) from the fit to k_obs.

Selection of mutants coresistant to imipenem and the ceftazidime-avibactam combination.

E. coli TOP10 harboring pTRC-99kΩbla_KPC-2 (ca. 6 × 10⁹ CFU) was plated on MH agar containing avibactam (4 μg/ml), ceftazidime (8 μg/ml), and imipenem (0.12 to 32 μg/ml). Mutants resistant to the triple combination were not obtained (frequency, <2 × 10⁻¹⁰).

The D¹⁷⁹Y substitution leads to decreased hydrolysis of meropenem and imipenem by the enzyme and consequently does not confer resistance to those substrates once produced in E. coli (see above, Table 1). To assess whether secondary mutations could restore resistance to carbapenems, derivatives of E. coli TOP10 producing KPC-2 D¹⁷⁹Y were selected on MH agar containing imipenem (8 μg/ml). Mutants were obtained at a frequency of 2 × 10⁻⁹. Sequencing of the bla_KPC-2 gene from 8 out of 8 mutants revealed a reversion to the wild-type sequence. These results indicate that the alteration of KPC-2 conferring coresistance to the ceftazidime-avibactam combination and to imipenem are not readily selected in vitro.

Conclusions.

Here, we show that a single amino acid substitution in KPC-2 (D¹⁷⁹Y) is sufficient to prevent inhibition of the β-lactamase by avibactam, leading to resistance to the ceftazidime-avibactam combination. In contrast, previous analyses based on site-directed mutagenesis have shown that reduced inhibition by avibactam due to a modification of position 132 of KPC-2 (N¹³²G) does not confer resistance to the ceftazidime-avibactam combination, since the hydrolysis of ceftazidime is drastically reduced (11). Extensive analysis of the Ω-loop of KPC-2 by site-directed mutagenesis (positions 164, 167, 169, and 179) has identified only a few substitutions that result in resistance to the ceftazidime-avibactam combination (R¹⁶⁴ replaced by A or P, and D¹⁷⁹ replaced by A, Q, or N) (10). The underlying mechanism was proposed to involve stabilizing interactions (e.g., hydrogen bonds) of ceftazidime within the KPC-2 active site that prevent avibactam from binding to and inhibiting the β-lactamase (10). The mechanism described in reference 10 is clearly distinct from the impaired carbamylation by avibactam reported for the D¹⁷⁹Y substitution in the current study.

The D¹⁷⁹Y substitution was recently shown to be responsible for the acquisition of ceftazidime-avibactam resistance under treatment in K. pneumoniae isolates producing the closely related enzyme KPC-3 (8, 16). The fact that a single mutation is sufficient for impaired inhibition of KPC β-lactamases by avibactam indicates that the emergence of resistance under treatment may jeopardize the efficacy of the ceftazidime-avibactam combination in the future (17). To date, the emergence of resistance was documented in a single study in ca. 8% of the patients infected with carbapenem-resistant Enterobacteriaceae and treated with the ceftazidime-avibactam combination (8).

The D¹⁷⁹Y substitution abolished the hydrolysis of meropenem and aztreonam by KPC-2 and improved inhibition by clavulanate (Table 2). Thus, these β-lactams and classical β-lactamase inhibitors might provide therapeutic alternatives in case of the emergence of ceftazidime-avibactam resistance. It remains to be determined whether such alternatives could also exist for other modifications of KPC enzymes leading to ceftazidime-avibactam resistance. This might be the case since the mutations selected under treatment (8) also led to large decreases in the MICs of carbapenems (18). Similarly, in vitro selection for resistance to the ceftazidime-avibactam combination in KPC-3-producing K. pneumoniae and Enterobacter cloacae strains led to large decreases in the MICs of carbapenems in the majority of the mutants (19). Our attempts to select mutants resistant to the triple combination of ceftazidime, imipenem, and avibactam were unsuccessful. Our selection for imipenem-resistant derivatives of the mutant producing KPC-2 D¹⁷⁹Y led to reversion to the wild-type sequence at position 179 and to susceptibility to the ceftazidime-avibactam combination. Recently, Shields et al. also reported such reversions, but resistance to carbapenems and ceftazidime-avibactam has been obtained by multistep exposure (≥4 weeks) of ceftazidime-avibactam-resistant K. pneumoniae strains to meropenem, mainly due to modifications of the porin OmpK36 (20).

Carbapenems and classical β-lactamase inhibitors may provide alternatives for use against variants with impaired inhibition by avibactam, as previously observed for substitutions in the β-lactamases from mycobacteria (21). These observations indicate, as previously discussed (21), that the formulation of β-lactamase inhibitors independently from a β-lactam partner would provide clinicians with access to potentially useful therapeutic regimens based on combining β-lactams with approved inhibitors.