Virtual screening identifies broad-spectrum β-lactamase inhibitors with activity on clinically relevant serine- and metallo-carbapenemases

Abstract

Bacteria are known to evade β-lactam antibiotic action by producing β-lactamases (BLs), including carbapenemases, which are able to hydrolyze nearly all available β-lactams. The production of BLs represents one of the best known and most targeted mechanisms of resistance in bacteria. We have performed the parallel screening of commercially available compounds against a panel of clinically relevant BLs: class A CTX-M-15 and KPC-2, subclass B1 NDM-1 and VIM-2 MBLs, and the class C P. aeruginosa AmpC. The results show that all BLs prefer scaffolds having electron pair donors: KPC-2 is preferentially inhibited by sulfonamide and tetrazole-based derivatives, NDM-1 by compounds bearing a thiol, a thiosemicarbazide or thiosemicarbazone moiety, while VIM-2 by triazole-containing molecules. Few broad-spectrum BLs inhibitors were identified; among these, compound 40 potentiates imipenem activity against an NDM-1-producing E. coli clinical strain. The binary complexes of the two most promising compounds binding NDM-1 and VIM-2 were obtained at high resolution, providing strong insights to improve molecular docking simulations, especially regarding the interaction of MBLs with inhibitors.

Chart 1

Venn diagram reporting the inhibition profile of active compounds towards the BLs panel. Each coloured area refers to a specific BL: CTX-M-15 green, KPC-2 orange, NDM-1 blue, VIM-2 grey, AmpC, pink. Compounds showing inhibition ranging 50–70% or > 70% towards, at least, one target are represented by white-filled and grey-filled circles, respectively, and are located under the curve belonging to that specific enzyme. Those exerting a minimum 30% inhibition towards other BLs are located in overlapping areas. The inhibition percentage exact values are reported in Table 1.

Figure 1

Docking pose of some of the most promising compounds in the corresponding BL binding site. (a) Compound 52 in NDM-1 (80% inhibition; protein PDB code 5zge). (b, c) Compound 24 in NDM-1 (70%; 5zge) and VIM-2 (86%; 2yz3), respectively. (d) Compound 9 in KPC-2 (77%; 3dw0). (e) Compound 7 in CTX-M-15 (57%; 4hbt). (f) Compound 54 in AmpC (51%; 1kdw). Pictures were prepared with Pymol v1.7.6.4.

Figure 2

Binary complexes of compound 24 binding to VIM-2 and of compound 31 binding to NDM-1. (a) Omit map of 24:VIM-2 binary complex shown at 3.5 σ contour level. (b) 31: NDM-1 binary complex. The ligand and the residues lining the pocket are shown as capped sticks, the Zn ions as grey spheres.

Figure 3

(a) Binary complex of compound 24 binding to VIM-2. (b) Superposition of the crystallographic (magenta) and of the originally selected docking pose (yellow) of compound 24 in VIM-2 binding site. (c) Superposition of the crystallographic (magenta) and of the most similar docking pose (yellow) of compound 24 in VIM-2 binding site. The protein of the binary X-ray complex is coloured grey, while the structure used for docking simulation is violet (d). Binary complexes of compound 31 binding to NDM-1. (e). Superposition of the crystallographic (green) and of the best ranked docking pose (orange) of compound 31 in NDM-1 binding site. The protein of the binary X-ray complex is coloured grey, while the structure used for docking simulation is cyan (f). Superposition of the crystallographic (green) and of the originally selected, and most similar docking pose (orange), of compound 31 in NDM-1 binding site. Red spheres correspond to water molecules.