ABSTRACT
Impeding, as well as reducing, the burden of antimicrobial resistance in Gram-negative pathogens is an urgent public health endeavor. Our current antibiotic armamentarium is dwindling, while major resistance determinants (e.g., extended-spectrum β-lactamases [ESBLs]) continue to evolve and disseminate around the world. One approach to attack this problem is to develop novel therapies that will protect our current agents. AAI101 is a novel penicillanic acid sulfone β-lactamase inhibitor similar in structure to tazobactam, with one important difference. AAI101 possesses a strategically placed methyl group that gives the inhibitor a net neutral charge, enhancing bacterial cell penetration. AAI101 paired with cefepime, also a zwitterion, is in phase III of clinical development for the treatment of serious Gram-negative infections. Here, AAI101 was found to restore the activity of cefepime against class A ESBLs (e.g., CTX-M-15) and demonstrated increased potency compared to that of piperacillin-tazobactam when tested against an established isogenic panel. The enzymological properties of AAI101 further revealed that AAI101 possessed a unique mechanism of β-lactamase inhibition compared to that of tazobactam. Additionally, upon reaction with AAI101, CTX-M-15 was modified to an inactive state. Notably, the in vivo efficacy of cefepime-AAI101 was demonstrated using a mouse septicemia model, indicating the ability of AAI101 to bolster significantly the therapeutic efficacy of cefepime in vivo. The combination of AAI101 with cefepime represents a potential carbapenem-sparing treatment regimen for infections suspected to be caused by Enterobacteriaceae expressing ESBLs.
INTRODUCTION
Antimicrobial resistance in Gram-negative pathogens is increasing steadily and is projected to substantially impact mortality (i.e., 10 million deaths) by 2050 (1). Novel resistance determinants (e.g., blaNDM-1 [2009] [2] and mcr-1 [2016] [3]) are emerging decennially; in addition, natural selection is expanding our current resistome. In 2015, the World Health Organization (WHO) released a Global Action Plan to address antimicrobial resistance (4). In this plan, the WHO outlined five specific objectives to combat resistance, including the development of novel agents to target multidrug-resistant (MDR) pathogens. β-Lactams are a highly prescribed class of antibiotics, and resistance to β-lactam antibiotics is prominent around the world.
The most common mechanism of resistance to β-lactams among MDR Gram-negative bacteria is the production of β-lactamases. These enzymes hydrolyze β-lactam antibiotics (penicillins, cephalosporins, monobactams, and carbapenems), preventing these agents from reaching their therapeutic target, the penicillin-binding proteins. Four different classes of β-lactamases (classes A, B, C, and D) are expressed by Gram-negative bacteria. Classes A, C, and D are serine-based enzymes, whereas class B β-lactamases are Zn2+ hydrolases (5). Among Enterobacteriaceae, β-lactam resistance is mediated predominantly by class A β-lactamases belonging to the SHV, TEM, and CTX-M families (6–8). Single amino acid substitutions in the SHV and TEM β-lactamases can drastically alter their substrate profiles, allowing them to acquire extended-spectrum β-lactamase (ESBL) activity (9, 10), whereas CTX-M β-lactamases are intrinsically ESBLs. ESBLs hydrolyze expanded-spectrum cephalosporins (e.g., cefotaxime, ceftazidime, and cefepime), and ESBL-producing Enterobacteriaceae constitute a serious public health threat (11). According to the Centers for Disease Control and Prevention (CDC), Enterobacteriaceae are estimated to cause >140,000 infections per year in the United States alone. The CDC also found that >19% of health care-associated infections are caused by ESBL-producing Enterobacteriaceae (11). Moreover, a recent study reported that the incidence of ESBL-producing Enterobacteriaceae has increased significantly in Europe (12).
To circumvent β-lactamases, β-lactamase inhibitors (BLIs) were developed, which typically are partnered with a β-lactam, as BLIs do not normally possess penicillin binding protein (PBP) inhibitory activity. Many class A ESBLs typically are susceptible to inactivation by the clinically approved BLIs clavulanic acid, sulbactam, tazobactam, avibactam, and vaborbactam. However, erosion of the efficacy of BLIs due to the emergence and spread of new β-lactamases has created a need for more potent agents. AAI101, a novel BLI active against ESBLs and other β-lactamases produced by Gram-negative pathogens, has been shown to be efficacious both in vitro and in vivo when combined with a highly potent cephalosporin, cefepime (13, 14).
AAI101 and tazobactam are penicillanic acid sulfone-based BLIs, whose structures differ by the presence of a strategically placed methyl group on the triazole moiety in AAI101 (Fig. 1). The addition of the methyl group to AAI101 results in a compound with a net neutral charge that enhances potency. Thus, like its β-lactam partner cefepime, AAI101 is a zwitterion. Antibiotic zwitterions are able to penetrate the Gram-negative cell wall at a higher rate (15). The goal of this study was to assess the microbiological, biochemical, and in vivo activities of AAI101 using cefepime as the β-lactam partner.
Structures of AAI101, tazobactam, and avibactam.
RESULTS
AAI101 restores the activity of cefepime against isogenic E. coli strains expressing class A β-lactamases.Clinically relevant class A, C, and D β-lactamases were cloned into plasmids and expressed in the Escherichia coli laboratory strain DH10B. MICs against all strains were determined by broth microdilution (Table 1). Of the 25 isogenic E. coli strains expressing a class A β-lactamase, 17, mainly ESBLs and carbapenemases, demonstrated elevated cefepime MICs compared to the MIC for the empty host strain, confirming the activity of the expressed enzymes. AAI101 restored the activity of cefepime against these 17 isolates (≥3 log2 dilution increases compared to the MIC for the host strain devoid of a β-lactamase). This was confirmed for AAI101 at 8 μg/ml combined with piperacillin. In contrast, only 6 of 17 strains possessed piperacillin-tazobactam MICs with a ≥2 log2 dilution increase compared to the MIC of the empty host strain. For 6 of the 8 remaining isolates, which were mainly penicillinases or clavulanic acid-resistant enzymes, piperacillin-tazobactam MICs were elevated. Cefepime alone was already active against those 6 strains, whereas piperacillin-AAI101 was active only against 2 of these 6 isolates. Isogenic E. coli strains producing single class C β-lactamases were susceptible to all agents tested, though AAI101 potentiated the activity of cefepime against Enterobacter cloacae strain NCTC 13406 (harboring a derepressed AmpC), whereas this isolate was resistant to piperacillin in the presence of either AAI101 or tazobactam. Similarly, cefepime alone demonstrated good activity against all five isogenic class D β-lactamase producers examined, whereas only the OXA-51 producer was susceptible to piperacillin when partnered with either tazobactam or AAI101.
Susceptibility testing resultsa
AAI101 is an inhibitor of class A β-lactamases.To assess the inhibitory activity of AAI101 compared to that of tazobactam and avibactam, 50% inhibitory concentrations (IC50s) were determined with periplasmic β-lactamase extracts using selected strains from Table 1. Across all class A β-lactamases tested, AAI101 possessed submicromolar IC50s compared to those of tazobactam and avibactam (Table 2). Some inhibitory activity (8.7 μM) was observed for the class D β-lactamase OXA-1, while the IC50s were ≥10 μM for class C β-lactamases and for OXA-48, OXA-51, and OXA-58. These data support the inhibitory activity of AAI101, particularly against class A β-lactamases, including ESBLs, carbapenemases, and clavulanic acid-resistant β-lactamases.
IC50s for AAI101, tazobactam, and avibactam using periplasmic extracts containing selected β-lactamasesa
Mechanism of inhibition by AAI101.To obtain a more detailed understanding of the inhibitory properties of AAI101, steady-state inactivation kinetic parameters were obtained for AAI101, tazobactam, and avibactam against the following selected class A enzymes: SHV-1, a penicillinase encoded predominantly chromosomally in Klebsiella pneumoniae and largely by a plasmid in E. coli; CTX-M-15, globally the most prevalent ESBL in E. coli and K. pneumoniae; and KPC-2, among the most common class A carbapenemases encountered in Enterobacteriaceae. AAI101 proved to be a potent inactivator of SHV-1, CTX-M-15, and KPC-2. AAI101 acylated SHV-1 faster than avibactam, but at a rate similar to that of tazobactam (Fig. 2, left; Table 3). AAI101, avibactam, and tazobactam displayed similar inactivation kinetics toward CTX-M-15 (Fig. 2, center; Table 3), whereas AAI101 acylated KPC-2 faster than tazobactam, at a rate comparable to that of avibactam (Fig. 2, right; Table 3). Moreover, the observed association rate constant kobs vs. [BLI] plots suggested different modes of inactivation for AAI101 and tazobactam. The rate constants for AAI101 were more similar to those of a diazabicyclooctane (DBO) than to the “classic” BLIs tazobactam, sulbactam, or clavulanic acid, since inactivation by AAI101 was not saturable with increasing concentrations of inhibitor over the range examined.
Inactivation kinetics of SHV-1, CTX-M-15, and KPC-2 by AAI101, tazobactam, and avibactam.
BLI inhibitory efficienciesa
Avibactam demonstrated a slower off-rate (dissociation rate [koff]) from SHV-1, CTX-M-15, and KPC-2 than tazobactam or AAI101 (Table 4). However, the half-lives (t1/2) are all 10 min ± 4 min (mean ± standard deviation), except for KPC-2 and avibactam, whereas the t1/2 of tazobactam with SHV-1 was not measurable. The amount of tazobactam required to achieve an initial velocity close to 0 μM/s was not obtained for SHV-1. Given the high acylation rates, the koff values will not have a physiologic impact so long as the inhibitor is released in intact form (see below). The partition ratios revealed that, by 15 min, more tazobactam was turned over by all three β-lactamases than avibactam or AAI101 (Table 5).
Reactivation rate of the β-lactamase after BLI inhibition
Turnover of BLI molecules during 15 mina
Reaction intermediates revealed by timed electrospray ionization-mass spectrometry (ESI-MS).The reaction course of penicillanic acid sulfones with serine β-lactamases involves various intermediates, as depicted in Fig. 3, with the potential formation of free β-lactamase (16). The major adducts formed include two different +52-Da and +70-Da intermediates, as well as a +88-Da intermediate.
The intermediates of AAI101 formed during reaction with serine β-lactamases (16). Upon acylation, sulfones form an imine, which tautomerizes to the cis-enamine and further to the trans-enamine. The cis-enamine subsequently hydrolyzes to form +88-Da hydrated aldehyde and +70-Da aldehyde adducts that can commence a second round of hydrolysis to regenerate free enzyme. The trans-enamine state is susceptible to attack by the hydroxyl side chain of Ser130, resulting in a cross-linked species of +52 Da. The +52-Da adduct can be hydrated to form a different +70-Da adduct bound to Ser130; this adduct can undergo further hydrolysis to obtain free enzyme. In addition, the cis-enamine, the trans-enamine, and the cross-linked +52-Da adduct can undergo further modification to form a different +52-Da adduct (propynyl enzyme), which can be hydrolyzed to regenerate free enzyme. Regarding the +52-Da propynyl enzyme, we postulate that the acrylate moiety bound to Ser70 is less likely to occur, as it would require a reductive hydrolysis of the enol ester. The Ser130 ester is more stable than the Ser70 ester. However, there is a possibility that the Ser70 ester sits in a more lipophilic part of the active-site pocket or that potentially hydrolytic water molecules are not present. In some cases, the free enzyme becomes inactive, losing 14 Da upon deacylation.
SHV-1, CTX-M-15, and KPC-2 were exposed to AAI101 or tazobactam, reactions were terminated at 1 min, 15 min, and 24 h, and the reaction products examined by ESI-MS. After 1 min, the +88-Da (Fig. 4A, peak labeled 1), +70-Da (Fig. 4A, peak labeled 2), and +52-Da (Fig. 4A, peak labeled 3) adducts were observed for all three β-lactamases with both sulfones.
(A) Mass spectra showing the formation of intermediates during inactivation of SHV-1, CTX-M-15, and KPC-2 by AAI101 or tazobactam at different times. Peaks labeled 1 (+88 Da), 2 (+70 Da), and 3 (+52 Da) correspond to the intermediates shown in Fig. 3. The peak labeled “inactive” represents the −14-Da apoenzyme. (B) Nitrocefin hydrolysis by SHV-1, CTX-M-15, and KPC-2 after 24 h of incubation with AAI101 (red) or tazobactam (gray). The lambda symbols represent wavelength.
A unique apoenzyme peak (−14 Da) appeared during the reactions of SHV-1 with AAI101 and of CTX-M-15 with AAI101 and with tazobactam. These modified β-lactamases (−14 Da) were tested for activity by monitoring the hydrolysis of nitrocefin. None of the −14-Da enzymes was able to turn over nitrocefin, indicating that the enzymes represented by these peaks were enzymatically inactive (Fig. 4B and 5). The −14-Da modification was not observed with KPC-2 upon reaction with AAI101 or tazobactam (Fig. 4A). The timescales for formation of the inactive apoenzyme were faster with AAI101 than with tazobactam for both SHV-1 and CTX-M-15. We hypothesize that the hydroxyl side chain located on the active-site Ser70 has been removed. During KPC-2 incubation with BLIs, apo-KPC-2 was evident within 15 min of incubation with tazobactam, whereas only the acyl-enzyme form was present during incubation with AAI101 (Fig. 4A). After 24 h, KPC-2 was fully active following incubation with AAI101 or tazobactam (Fig. 4B). However, in some replicates (see Fig. 4A, inset, for KPC-2 plus AAI101 at 24 h), acyl–KPC-2 was observed. The chemical and physiological implications of these results are under investigation.
Scheme for interactions of inhibitor I (AAI101 or tazobactam) with CTX-M-15 (E). Formation of the noncovalent complex, E:I, is represented by the dissociation constant, Ki, which is equivalent to k−1/k1. k2 is the first-order rate constant for the acylation step, or formation, of E-I. k3 is the all-inclusive rate formation of the E-I* acyl-enzyme (imine, cis-enamine, trans-enamine, and +88-Da, +70-Da, and +52-Da adducts). The hydrolysis of E-I* to form inactive enzyme (EΔOH) and the various products (P) is represented by the rate constant k4.
Molecular modeling reveals that CTX-M-15 forms more molecular interactions with AAI101 than with tazobactam.Molecular dynamic simulations (MDS) were employed to examine the interactions within the CTX-M-15 active site upon acylation by AAI101 or tazobactam. Based upon timed ESI-MS, the acyl-enzyme structures for tazobactam and for AAI101 were constructed as imine and enamine intermediates. With minimization and MDS for 80 ps, we observed that the imine intermediate was slightly more energetically stable (−11,631 kcal/mol) that the enamine (−11,580 kcal/mol) and achieved equilibration with CTX-M-15 more quickly (25 ps versus >40 ps). Therefore, the imine forms of the BLI complexes in the CTX-M-15 active site were analyzed further.
Examination of the complete MDS trajectories for AAI101 and tazobactam revealed that the methyl group on the triazole ring of AAI101 allows for many classical and nonclassical hydrogen bonding interactions not observed for tazobactam (Fig. 6). AAI101 can form hydrogen bonds with CTX-M-15 Gly238 and Ser272, whereas in tazobactam, the triazole ring is flipped toward the 102-to-108 loop. Upon acylating the active site of CTX-M-15, the additional methyl group on AAI101 enables a different set of interactions to occur than with tazobactam. Given that AAI101 possesses a partition ratio, or turnover number (kcat/kinact [kinact is inactivation rate]), of 1 for CTX-M-15, compared to 2 for tazobactam (see above), the interactions observed may play a role in this process.
Molecular models of CTX-M-15 with AAI101 (pink) and tazobactam (cyan) (top). Hydrogen bonding interactions between active-site residues and the triazole ring of AAI101 (left) or tazobactam (right) during molecular dynamics simulations (MDS) are graphed (bottom). The y axis represents whether a hydrogen bond was observed (red bar) or not (no bar) in the different conformations obtained during the 80-ps MDS. The x axis lists the amino acid position at which that hydrogen bond may occur.
Cefepime and AAI101 possess similar PK profiles.The pharmacokinetics (PK) profile of AAI101, together with that of cefepime, was evaluated in a mouse septicemia model (Table 6; Fig. 7). The PK profile of AAI101 mirrored that of cefepime, thus providing further support for the choice of the pairing of AAI101 with cefepime in this model.
Quantification of cefepime and AAI101 in mouse plasma following subcutaneous coadministration
Pharmacokinetics of cefepime and AAI101 in mouse plasma following a single subcutaneous coadministration of 60 mg/kg of cefepime and 30 mg/kg of AAI101 in an animal infected with K. pneumoniae R-43.
Cefepime-AAI101 demonstrates efficacy in a murine sepsis (peritonitis) model.Three cefepime-resistant clinical isolates of Enterobacteriaceae were used in a murine septicemia model to assess the ability of AAI101 to restore the activity to cefepime in vivo, and one cefepime-susceptible clinical reference isolate (E. coli strain ATCC 25922) served as a control. The isolates were susceptible in vitro to cefepime-AAI101 and demonstrated low frequencies of resistance (10−9) to both cefepime alone and the combination. These results are summarized in Tables 7 and 8, respectively.
Modal MICs of the test strains used in the animal model
Resistance frequencies of the test strains used in the animal model
The 100% lethal doses (LD100s) in the mice for Enterobacteriaceae comprising the septicemia test panel were obtained in the pilot experiments (Table 9). Groups of immunocompetent mice were infected with lethal doses of one or another of the test strains and then dosed over a range of concentrations with either cefepime, cefepime-AAI101 (2:1, wt/wt), or meropenem, and the 50% protective doses (PD50s) for survival during 7 days postinfection were determined (Table 10).
LD100s for test strains used in the animal model
PD50s for cefepime, cefepime-AAI101, and meropenem against tested isolates
Mortality in the control group (treatment with vehicle only) was 100% by 24 h postinfection. Towards the cefepime-susceptible E. coli ATCC 25922, very low doses of cefepime and cefepime-AAI101 (2:1, wt/wt) were equally effective at protecting the infected mice, whereas a slightly higher PD50 was obtained for meropenem (Fig. 8). For animals dosed twice within a few hours of bacterial challenge, cefepime, cefepime-AAI101 (2:1, wt/wt), and meropenem administered at <0.25 mg/kg of body weight sufficed to reduce the pathogen titers below the threshold required to establish a lethal infection for ≥50% of the mice.
Survival curves for cefepime, cefepime-AAI101, and meropenem against E. coli ATCC 25922. Half-log dose range, 0.01 to 30 mg/kg. (A to C) Survival curves for cefepime (A), cefepime-AAI101 (2:1, wt/wt) (B), and meropenem (C). (D) Cumulative survival curves as a function of drug concentration. **, P < 0.01 versus control (vehicle only) group; ***, P < 0.001 versus control (vehicle-only) group.
For the three Enterobacteriaceae isolates resistant to cefepime (MICs of ≥32 μg/ml) due to the expression of ≥2 β-lactamases, the addition of AAI101 had a remarkable ameliorative effect. AAI101 restored the protective efficacy of cefepime more than 10-fold toward the mice infected with the K. pneumoniae strain coproducing an ESBL and a class D carbapenemase and approximately 20-fold toward the K. pneumoniae and E. cloacae strains each coproducing an ESBL and an AmpC (Fig. 9, 10, and 11). For all three cefepime-resistant Enterobacteriaceae, the PD50s for cefepime-AAI101 (2:1, wt/wt) were similar (range, 12.77 to 15.97 mg/kg). Cefepime-AAI101 afforded protection comparable to that of meropenem toward the mice infected with E. cloacae B-143 and better protection than meropenem toward the mice infected with K. pneumoniae B-124, though meropenem proved more protective than cefepime-AAI101 toward the mice infected with K. pneumoniae R-43.
Survival curves for cefepime, cefepime-AAI101, and meropenem against K. pneumoniae R-43. Half-log dose range, 0.1 to 300 mg/kg. (A to C) Survival curves for cefepime, cefepime-AAI101 (2:1, wt/wt) (B), and meropenem (C). (D) Cumulative survival curves as a function of drug concentration. **, P < 0.01 versus control (vehicle only) group; ***, P < 0.001 versus control (vehicle only) group.
Survival curves for cefepime, cefepime-AAI101, and meropenem against K. pneumoniae B-124. Half-log dose range, 0.1 to 300 mg/kg. (A to C) Survival curves for cefepime (A), cefepime-AAI101 (2:1, wt/wt) (B), and meropenem (C). (D) Cumulative survival curves as a function of drug concentration. **, P < 0.01 versus control (vehicle only) group; ***, P < 0.001 versus control (vehicle-only) group. During the study, four animals had to be culled prematurely in order to achieve a humane endpoint (cefepime 10 mg/kg plus AAI101 5 mg/kg cohort, 1 mouse, 80 h postinfection; cefepime 30 mg/kg plus AAI101 15 mg/kg cohort, 3 mice, 80 h postinfection).
Survival curves for cefepime, cefepime-AAI101, and meropenem against E. cloacae B-143. Half-log dose range, 0.1 to 300 mg/kg. (A to C) Survival curves for cefepime (A), cefepime-AAI101 (2:1, wt/wt) (B), and meropenem (C). (D) Cumulative survival curves as a function of drug concentration. **, P < 0.01 versus control (vehicle only) group; ***, P < 0.001 versus control (vehicle-only) group. During the study, two animals had to be culled prematurely in order to achieve a humane endpoint (cefepime 30 mg/kg plus AAI101 15 mg/kg cohort, at 80 h postinfection).
DISCUSSION
Using a collection of isogenic E. coli strains each expressing a single defined β-lactamase, the present study demonstrates that (i) AAI101 restores the activity of cefepime and piperacillin against selected ESBL-producing strains, (ii) AAI101 is more potent than tazobactam against these ESBL producers, (iii) cefepime alone is more active than piperacillin-tazobactam against clavulanic acid-resistant strains, and (iv) cefepime-AAI101 is as effective in vitro as meropenem and imipenem against these β-lactamase producers. These observations suggest that the cefepime-AAI101 combination is likely to be effective against Enterobacteriaceae producing class A, C, and D β-lactamases. This statement is supported by the work of Crandon and Nicolau, who reported previously that cefepime combined with AAI101 was effective in vitro against a panel of cefepime-resistant recent clinical isolates of Enterobacteriaceae, including ESBL and carbapenemase producers (13, 14).
The microbiological and biochemical data indicate that AAI101 differs mechanistically from tazobactam with respect to its inhibitory activity toward class A β-lactamases, including ESBLs, carbapenemases, and clavulanic acid-resistant β-lactamases. For class C and D β-lactamases, AAI101 has less of a direct impact; however, cefepime alone is not efficiently hydrolyzed by most AmpC and OXA β-lactamases, making cefepime an excellent choice as the partner for AAI101.
Kinetic and mass spectrometric measurements support the increased potency of AAI101 over that of tazobactam against SHV-1 and KPC-2 observed during susceptibility testing. MDS suggests that AAI101 forms more hydrogen bonding interactions than tazobactam in the active site of CTX-M-15, which may impede the turnover of AAI101 compared to that of tazobactam (Table 5). The results from our mouse septicemia studies confirm that the addition of AAI101 to cefepime restores the efficacy of cefepime against Enterobacteriaceae producing multiple β-lactamases of diverse classes (ESBLs, AmpCs, and class D carbapenemases) and are concordant with the results of a previous study in which the cefepime-AAI101 combination was highly potent in a neutropenic thigh infection model using cefepime-resistant Enterobacteriaceae (13). Taken together, these findings suggest that AAI101 in combination with cefepime may offer a therapeutic alternative to carbapenems as first-line agents for treatment of Enterobacteriaceae infections by pathogens producing selected class A, C, and D β-lactamases. The novel combination of AAI101 and cefepime is an important addition to our therapeutic armamentarium.
MATERIALS AND METHODS
Strains and plasmids.A panel of 33 isogenic Escherichia coli recombinants carrying diverse single β-lactamase genes was assembled for microbiological and biochemical analysis. The cloning and/or origins of 28 of the bla genes were described previously (17–31). The blaTEM-30, blaCTX-M-14, blaCTX-M-15, blaOXA-51, and blaOXA-58 genes each were synthesized and cloned into the pBC SK(−) phagemid by Celtek Bioscience (Franklin, TN). Codons of the blaOXA-51 and blaOXA-58 genes were optimized for expression in E. coli. All constructs were expressed in E. coli strain DH10B. For protein expression, the leaderless (deletion of nucleotides 1 to 84) blaCTX-M-15 gene was further subcloned into pET24a(+) using NdeI and XhoI restriction enzymes (see below).
For in vivo evaluation in a septicemia model, the following genotyped and well-characterized clinical isolates were chosen: E. coli ATCC 25922; Klebsiella pneumoniae R-43, which carries blaCTX-M-15, blaSHV-12, and blaDHA-1; K. pneumoniae B-124, which possesses blaCTX-M-15 and blaOXA-48; and Enterobacter cloacae B-143, which contains blaCTX-M-15 and a chromosome-encoded derepressed blaAmpC.
Compounds.AAI101 was provided by Allecra Therapeutics SAS (St-Louis, France). For in vitro experiments, avibactam was purchased from Advanced ChemBlocks, Inc. (Burlingame, CA), and tazobactam was obtained from Chem-Impex International, Inc. (Wood Dale, IL). For in vivo work, clinical forms of cefepime (lot number K1-01; Panpharma S.A., Luitré, France) and meropenem (lot number 5JB0439; Actavis, Ltd., Dublin, Ireland) were provided by Atlangram (Nantes, France).
Susceptibility testing.Custom antimicrobial susceptibility testing plates (ThermoFisher Scientific, Cleveland, OH) were used to determine MICs by broth microdilution according to Clinical and Laboratory Standards Institute (CLSI) guidelines (40). Fixed concentrations of AAI101 (4 μg/ml and 8 μg/ml) and tazobactam (4 μg/ml) partnered with a β-lactam antibiotic (piperacillin or cefepime) were used, along with cefepime, imipenem, and meropenem as comparators. The MIC endpoints were defined as the lowest concentration of β-lactam (alone or partnered with a BLI) causing complete inhibition of growth. MIC assays were performed in triplicate on three different days, and modal MICs were determined. Synergy was defined as a ≥3 log2 reduction of the piperacillin or cefepime MIC in the presence of a BLI. Quality-control strains E. coli ATCC 25922, E. coli ATCC 35218 (TEM-1 producer), K. pneumoniae ATCC 700603 (SHV-18 producer), and P. aeruginosa ATCC 27853 (inducible PDC producer) were tested also, as recommended by the CLSI, during each round of MIC determination.
Determination of IC50s for selected β-lactamases.Based on susceptibility testing results, 16 β-lactamases were selected for the determination of IC50s with AAI101, tazobactam, and avibactam. Periplasmic extracts of E. coli constructs producing single β-lactamases were prepared as follows. Cells were grown in lysogeny broth (LB), harvested during exponential growth, and frozen for 18 h at −20°C. Cell pellets were incubated with lysozyme, benzonase nuclease (Merck KGaA, Darmstadt, Germany), and magnesium sulfate in 50 mM Tris-Cl, pH 7.4 for 30 min. Lysates were treated with 2.0 mM EDTA for 5 min, and the cellular debris was removed by centrifugation.
Reaction mixtures were prepared in 10 mM phosphate-buffered saline, pH 7.4 (PBS), unless otherwise noted, with nitrocefin as a chromogenic substrate and various concentrations of BLI. The amount of nitrocefin was slightly different in each assay; a concentration of nitrocefin 3 to 5 times the estimated Km of the β-lactamase for the chromogen was used. IC50s were performed after a 5-min preincubation. Reactions were initiated by the addition to the reaction mixture of periplasmic lysate containing the β-lactamase of interest. Initial velocities at an optical density of 482 nm (OD482) were recorded. Plots of inverse initial velocity versus inhibitor concentration were fitted to a linear equation, and IC50s for the BLIs were determined from the graphs using standard equations. The mean values and standard deviations for triplicate determinations were recorded. When testing class D β-lactamases, 50 mM sodium phosphate buffer, pH 7.2, supplemented with 20 mM sodium bicarbonate was used, as chloride ions may interfere with the activity of class D enzymes.
Purification of β-lactamases.E. coli DH10B pBC SK(−)blaSHV-1, E. coli BL21(DE3) pET24a(+)blaCTX-M-15, and E. coli Origami 2 DE3 pET24a(+)blaKPC-2 (32) were grown in super optimal broth (SOB) to an OD600 of 0.6, isopropyl β-d-1-thiogalactopyranoside was added to the strains with pET24a(+) vectors, and the cells were grown for an additional 2 to 3 h. The cells were pelleted and frozen for 18 h at −20°C. Periplasmic extracts from the cell pellets containing the β-lactamases were used for preparative isoelectric focusing (pIEF). A Sephadex G-100 matrix and commercially prepared ampholines (pH 3.5 to 10.0) were used for pIEF gels. The pIEF gels were run for at least 16 h at 4°C at a constant power of 8 W on a Multiphor II isoelectric focusing apparatus, then the areas of each gel demonstrating β-lactamase activity by nitrocefin overlay were cut from the gel, eluted with PBS on polyethylene glycol columns, and concentrated using an Amicon Ultra-4 concentrator with a molecular weight cutoff of 10,000 Da (Merck KGaA, Darmstadt, Germany). The proteins were purified further using a size exclusion HiLoad 16/600 Superdex column (also an anion exchange HiTrapQ column for KPC-2) on an Äkta fast protein liquid chromatography system (GE Healthcare Life Sciences, Pittsburgh, PA). The purities of the fractions from all preparations were determined by sodium dodecyl sulfate polyacrylamide gel electrophoresis. The gels were stained with Coomassie brilliant blue R250, and the protein concentrations determined by measuring the absorbance at 280 nm and using the specific protein’s extinction coefficient, obtained using the ProtParam tool available through the ExPASy Bioinformatics Resource Portal (33).
Steady-state kinetics.BLI inhibition kinetics were examined for the purified SHV-1, CTX-M-15, and KPC-2 as described previously (34, 35). The kinetics parameters were determined using an Agilent 8453 diode array spectrophotometer (Agilent Technologies, Inc., Santa Clara, CA). All reactions were conducted in PBS at room temperature. The scheme of inhibition and inactivation of β-lactamases by sulfones is depicted in Fig. 5.
(i) Determination of Ki app.A direct competition assay was performed to determine the relative dissociation constant, Ki app, of the inhibitor. Ki app most closely approximates the Ki (k−1/k1) value for BLIs with a slow k2 or acylation rate constant. However, for BLIs that acylate rapidly after Michaelis complex formation, the value more closely approximates the Km. A final concentration of 5× Km of nitrocefin was used as the indicator substrate in the presence of nanomolar concentrations of the β-lactamase. The data were corrected to account for the affinity of nitrocefin (KmNCF) for the β-lactamase according to equation 1, where S is the concentration of substrate used:
(ii) Determination of kinact/KI or k2/K.To determine the inhibitor efficiency (kinact/KI or k2/K), assays were performed using fixed concentrations of enzyme and nitrocefin and increasing molar concentrations of BLI. The progress curves to obtain kobs values for inactivation were fit graphically; kobs is the reciprocal of the ordinate of the intersection of the straight lines obtained from the initial and final steady-state velocities, using equation 2:
(iii) Determination of koff.The off-rates, koff, of the BLIs for the β-lactamases were determined using a previously published method (36). β-Lactamases were incubated with the BLIs at 5× to 10× Ki app for 5 min at room temperature, and then the reaction mixtures were serially diluted and 100 μM NCF added. Progress curves measuring nitrocefin hydrolysis were collected for 1 h, and the data were fitted to equation 2 to obtain koff. Reaction mixtures containing the β-lactamase alone and BLI alone were used as controls.
(iv) Determination of partition ratio.Partitioning of the initial enzyme-inhibitor complex between hydrolysis and enzyme inactivation, kcat/kinact or tn, was obtained by incubating the β-lactamase with increasing concentrations of the BLI at room temperature for 15 min in PBS. The ratio of inhibitor to enzyme necessary to inhibit hydrolysis of nitrocefin by >90% was taken as kcat/kinact.
Timed ESI-MS.Homogenous preparations of the SHV-1, CTX-M-15, and KPC-2 enzymes were subjected to ESI-MS after incubation with the BLIs. A Synapt G2-Si high-resolution quadrupole time-of-flight mass spectrometer (Waters Corp., Milford, MA) equipped with a LockSpray dual electrospray ion source was used to acquire mass spectra. The instrument was calibrated with sodium iodide, using a mass range of 50 to 2,000 m/z. This calibration results in an error of ±5 atomic mass units (amu). All β-lactamase and β-lactamase−BLI reactions were terminated by the addition of 0.1% formic acid and 1% acetonitrile, and the samples were applied to an Acquity H class ultraperformance liquid chromatography (UPLC) instrument on a 1.7-μm, 2.1- by 100-mm Acquity UPLC BEH (ethylene bridged hybrid) C18 column (Waters Corp.) equilibrated with 0.1% formic acid in water. The β-lactamases and β-lactamase−BLI complexes were eluted using gradients with starting concentrations of 90% of 0.1% formic acid in water (mobile phase A) and 10% of 0.1% formic acid in acetonitrile (mobile phase B), reaching final conditions of 15% mobile phase A and 85% mobile phase B by 4 min. A gradient of 19% mobile phase A and 81% mobile phase B was reached by 1 min. The tune settings for each data run were as follows: capillary voltage at 3.5 kV, sampling cone at 35 V, source offset at 35 V, source temperature at 100°C, desolvation temperature at 500°C, cone gas at 100 liter/h, desolvation gas at 800 liter/h, and nebulizer gas at 6.0 bar. The spectra were analyzed using MassLynx version 4.1 (Waters Corp.) and deconvoluted using the MaxEnt1 program.
Molecular modeling.Molecular modeling was used to assess the acyl-enzyme complex formed during interactions between CTX-M-15 and AAI101 and between CTX-M-15 and tazobactam. The crystal structure of CTX-M-15 (PDB code 4HBT) was employed to construct and validate the model using Discovery Studio (DS) 2017 molecular modeling software (Biovia, Dassault Systemes, San Diego, CA). AAI101 and tazobactam were constructed using the DS 2017 Fragment Builder tool and minimized using a DS 2017 Standard Dynamics Cascade. The BLIs were docked automatically into the active site using the DS 2017 CDOCKER module, which utilizes a CHARMm-based molecular dynamics (MD) scheme to dock the ligands into the active sites. The best conformations (e.g., sulfone carbonyl oriented toward the oxyanion hole) were aligned automatically into polar and apolar active-site hotspots of CTX-M-15, and the best scoring poses were reported. The scoring was based on the CDOCKER-energy value (i.e., the internal ligand energy and the receptor-ligand interaction energy), which is reported as a negative value (i.e., −CDOCKER_energy); lower values indicate more favorable binding. This enables the energy to be used as a scoring function. To further optimize the docked poses, a CHARMm minimization step was used. In this step, the Smart Minimization algorithm was utilized, involving 1,000 steps of steepest descent with a root mean square (RMS) gradient tolerance of 0.1 Å, followed by conjugate gradient minimization with an RMS deviation (RMSD) minimization gradient of 0.009 Å. Molecular dynamic simulations (MDS) were conducted over 80 ps, using a standard dynamic protocol.
Resistance frequencies.The frequencies of resistance to cefepime, cefepime in combination with fixed concentrations of 4 μg/ml and 8 μg/ml of AAI101, and meropenem were determined for E. coli ATCC 25922, K. pneumoniae R-43, K. pneumoniae B-124, and E. cloacae B-143. Agar plates containing (i) cefepime at 3× the MIC of cefepime, (ii) cefepime plus 4 μg/ml of AAI101, where the cefepime concentration was 3× the MIC of cefepime obtained in the presence of 4 μg/ml of AAI101, (iii) cefepime plus 8 μg/ml of AAI101, where the cefepime concentration was 3× the MIC of cefepime obtained in the presence of 8 μg/ml of AAI101, and (iv) meropenem at 3× the MIC of meropenem were inoculated with cell suspensions containing 109 to 1010 CFU (the precise inoculum sizes were determined by dilution plating), and the plates incubated at 37°C for 20 h. The resistance frequencies were calculated from the number of colonies showing breakthrough growth.
Animals.Six-week-old pathogen-free RjOrl:Swiss female mice (20 to 24 g) were obtained from Janvier Labs (Le-Genest-Saint-Isle, France); these nonisogenic (outbred) mice, used frequently in bacterial infection models, reflect the heterogeneity of the mouse population better than inbred mice. The mice were housed in cages, given food and water ad libitum, and allowed to adapt to their new environment for 4 days before any procedures were initiated. All the protocols involving animals were approved by the Animal Experiment Committee of Pays de la Loire (France) (authorization number APAFIS#8621-2016092615586292).
Bacterial inoculum and drug preparation.Bacterial cells from overnight cultures of E. coli ATCC 25922, K. pneumoniae R-43, K. pneumoniae B-124, and E. cloacae B-143 were collected by centrifugation (800 × g for 10 min) and washed with cold sterile physiological saline; appropriate dilutions of infecting inocula were prepared with a virulence adjuvant, 5% porcine gastric mucin (Sigma, St. Louis, Missouri). By enhancing the virulence of the inoculum, mucin reduces the infectious dose (37), obviating effects such as endotoxic shock or carryover of toxic metabolites from the spent culture broth.
The drugs were prepared in sterile physiological saline and administered by subcutaneous (s.c.) injection into the neck at 1 h and 4 h after bacterial challenge. A constant volume of drug was administered to each mouse (8 ml/kg/dose); the dosing was adjusted according to the weight of each mouse.
PK study.A single-dose pharmacokinetics (PK) study of cefepime-AAI101 was performed using groups of mice (3 animals per group) infected with K. pneumoniae R-43. At 0.25, 0.5, 1, 2, 3, 4, 6, and 8 h following the subcutaneous administration of cefepime-AAI101 (60 mg/kg cefepime plus 30 mg/kg AAI101), the blood was collected by cardiac puncture (3 mice per time point), placed on ice in Vacuette blood collection tubes (2 ml, containing K2EDTA), and centrifuged; the plasma samples were decanted into Sarstedt tubes and frozen immediately at −80°C, pending shipment on dry ice to Aptuit (Evotec) Center for Drug Discovery & Development (Verona, Italy) for quantification of the plasma cefepime and AAI101 concentrations using a validated high-performance liquid chromatography–tandem MS (HPLC-MS/MS) method.
Septicemia model.Pilot experiments were performed with immunocompetent mice, using 0.6 ml of different titers of each bacterial strain suspended in mucin, to determine the 100% lethal dose (LD100; bacterial load required to kill 100% of mice by 24 h postinfection) for each strain. The precise LD100 values were identified by dilution plating.
Immunocompetent mice were infected intraperitoneally with 0.6 ml of the appropriately diluted cell suspensions corresponding to the LD100 for a given strain. Twenty-five groups of mice (six animals per group) were used for each clinical isolate examined. Group 1 mice were infected but received only vehicle (untreated) and served as a control group. For groups 2 to 9, the mice received half-log dilutions of cefepime monotherapy over a range of 0.1 to 300 mg/kg, except for the mice infected with E. coli ATCC 25922, for which a range of 0.01 to 30 mg/kg was used. For groups 10 to 17, the mice received cefepime in combination with AAI101 (fixed cefepime-AAI101 ratio of 2:1, wt/wt) over a cefepime range of 0.1 to 300 mg/kg, except for the mice infected with E. coli ATCC 25922, for which a range of 0.01 to 30 mg/kg was used. For groups 18 to 25, the mice received meropenem monotherapy over a range of 0.1 to 300 mg/kg, except for the mice infected with E. coli ATCC 25922, for which a range of 0.01 to 30 mg/kg was used.
The mice were observed for mortality and signs of distress for up to 7 days. Survival was recorded at 1, 4, 8, and 24 h postinfection and three times daily on subsequent days until the end of the 7-day observation period. The method of Reed and Muench (cumulative distribution function) was used to calculate the 50% protective doses (PD50s; mg/kg) for cefepime, cefepime-AAI101, and meropenem for each bacterial strain (38).
The percent body weight changes of each animal were recorded daily at 8:00 a.m. and 6:00 p.m. Any mouse judged to be experiencing pain or serious distress (39) received buprenorphine (0.1 mg/kg s.c. twice a day [b.i.d.], sufficient to cover the nocturnal period) over the course of the experiment. Signs of unrelieved suffering triggered the humane endpoint of euthanasia by CO2 inhalation.
ACKNOWLEDGMENTS
We thank Stefano Biondi and Philipp Knechtle for their helpful review of the manuscript.
These studies were supported by a research grant from Allecra. Research reported in this publication also was supported in part by funds and/or facilities provided by the Cleveland Department of Veterans Affairs, the Veterans Affairs Merit Review Program grants number BX002872 (K.M.P.-W.) and BX001974 (R.A.B.) from the United States Department of Veterans Affairs Biomedical Laboratory Research and Development Service, and the Geriatric Research Education and Clinical Center VISN 10 (R.A.B.). Research reported in this publication was supported in part by funds from the National Institute of Allergy and Infectious Diseases of the National Institutes of Health under awards number R01AI100560, R01AI063517, and R01AI072219 (R.A.B.)
The contents do not represent the views of the U.S. Department of Veterans Affairs or the U.S. Government. The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health.
FOOTNOTES
- Received 17 January 2019.
- Returned for modification 19 February 2019.
- Accepted 26 February 2019.
- Accepted manuscript posted online 11 March 2019.
- Copyright © 2019 American Society for Microbiology.