Impact of the double deletion ΔG242-T243 in KPC-2 in the effectiveness of ceftazidime-avibactam and imipenem-relebactam

Abstract

Combinations of β-lactam-diazabicyclooctane inhibitors (DBOs) like ceftazidime-avibactam (CZA) and imipenem-relebactam (IMR) have shown efficacy in treating KPC-2-producing Klebsiella pneumoniae. However, CZA-resistant K. pneumoniae strains have been identified, often linked to substitutions and/or insertions/deletions in three different loops of KPC: (i) the Ω-loop region (amino acids 164-179), (ii) the 237-243 loop; and (iii) the 266-275 loop. This study investigates the impact of the double deletion ΔG242-T243 present in KPC-14. Our results demonstrate that the lower effectiveness of CZA against KPC-14 can be explained by both increased hydrolysis of ceftazidime and a lower affinity and acylation rate by avibactam. In contrast, the IMR combination was efficient in restoring susceptibility to the KPC-14 producing-clone. Although we also observed a lower affinity and acylation rate for relebactam in KPC-14, this reduction in affinity was accompanied by a loss in the carbapenemase activity, finally resulting in an IMR susceptibility phenotype for KPC-14. Expansion of the substrate profile of KPC-14 toward ceftazidime is associated with a trade-off for carbapenems, other penicillins, and cephalosporins, as well as a higher inhibition by clavulanic acid compared to KPC-2. This study provides a better understanding of how deletions in the 237-243 loop affect the effectiveness of novel DBO-combinations and supports the hypothesis that these mutations result in CZA resistance by other different biochemical mechanisms than mutations in the Ω-loop.

Keywords: CZA; DBO; KPC-14; Klebsiella pneumoniae; avibactam; relebactam.

Fig 1

(a) Chemical structure of avibactam (AVI) and relebactam (REL). (b) Kinetic model proposed for the interaction of KPC-2 with AVI and REL (3). In this model, E and I represent the enzyme and the DBO inhibitors, respectively. E:I denotes the noncovalent complex, while E-I represents the enzyme acylated with AVI or REL. Initially, the enzyme is reversibly acylated by the DBO inhibitor. However, for KPC-2, a slow hydrolytic pathway involving the loss of the sulfate and an imine hydrolysis (E-I’) was proposed for both AVI and REL. The final step results in the deacylated enzyme and an oxopiperidine product (P).

Fig 2

(a) Monitoring curves of nitrocefin hydrolysis at 480 nm in the presence of increasing concentrations of clavulanic acid, for KPC-14 and KPC-2. A single representative replicate is shown for each inhibitor concentration tested to simplify the graphic. Compared to KPC-2, the KPC-14 variant is readily inactivated by a concentration of 16 µM of clavulanic acid. (b) Monitoring curves of clavulanic acid hydrolysis at 235 nm over time for KPC-14 and KPC-2. A single representative replicate is shown to simplify the graphic.

Fig 3

(a) Comparative view of the β3-β4 loops of KPC-2 (blue) and KPC-14 (magenta), in which the deletion of G242-T243 (with asterisks) provokes a shortening of the loop, as well as a displacement of residues like V240-Y241 in KPC-14; the β3-β4 sequence is shown in matching colors. (b) Details of the active site of KPC-14, showing the main residues involved in the stabilizing hydrogen bonds network, and the probable position of the acylating (Wat1) and deacylating (Wat2) water molecules (light blue spheres). (c) Surface view of both KPC-2 and KPC-14: W105, Ω loop, and β3-β4 loop are shown as reference. The surface was colored according to the hydropathy scale, using a gradient from the highest (red) to the lowest (white) hydrophobic content.

Fig 4

The structures of KPC-2 (blue) and KPC-14 (orange). (a) Root mean square fluctuation (RMSF) of KPC-2 and KPC-14. The Ω-loop (pink), the 237–243 loop (olive), and the 266–275 (cyan) are highlighted. (b) and (c) structural representations of KPC-2 and KPC-14, respectively, illustrating the Ω-loop (pink), the 237–243 (olive), and the 266–275 loop (cyan). (d) The hydrogen bond interactions between N136 and E166 in KPC-2. No hydrogen bond is observed between N136 and E166 in KPC-14. (e) $\chi$1 dihedral angle density distribution of N136. (f) and (g) Hydrogen bond distances between N136(ND2)-E166(O) and N136(OD1)-E166(N), respectively. The $\chi$2 dihedral angles density distribution of (h) L169 and (i) N170. (j) Structural conformations of L169 and N170 in KPC-2 and KPC-14. The $\chi$1 dihedral angle density distribution of (k) I173 and (l) Y241. (m) The hydrophobic interactions between I173 and Y241 in KPC-2 and KPC-14. The hydrogen bond distances of (n) Y241-K270 and (o) Y241-A267. (p) Y241 forms hydrogen bonds with K270 and A267 in KPC-2, whereas these interactions are absent in KPC-14. (q) Calculated pocket volumes of KPC-2 and KPC-14.

Fig 5

(a) Interaction of KPC-14 with CAZ (magenta) compared to the apo KPC-14 (gray). A possible relocation of N170 may be forced by the ceftazidime’s aminothiazole position upon binding of the substrate. (b) Compared to KPC-2^E166Q (blue ribbon), the aminothiazole ring of CAZ is oriented toward the 237–244 loop compared to KPC-14, avoiding clashes with the Ω-loop and particularly with N170. * Residues are only present in KPC-2. (c) Comparative interaction of KPC-2 (blue) and KPC-14 (magenta) with AVI (left panel), and REL (right panel). The hydrophobic content was shown surrounding residues from the Ω loop, β3-β4 loop, and W105, using a gradient from the highest (red) to the lowest (white) hydrophobic content.