Cyclobutanone Mimics of Intermediates in Metallo-β-lactamase Catalysis

Abstract:

Studies have confirmed that bacterial resistance to β-lactam antibiotics primarily involves hydrolysis by two types of β-lactamases: serine-based β-lactamases (SBLs) and metallo-β-lactamases (MBLs). Cyclobutanoneis a hydrolytically stable β-lactam analogue with potential inhibitory effects against both SBLs and MBLs. This work focuses on solution and crystallographic studies of the interaction between a cyclobutanone carbapenem analogue and the clinically significant MBL SPM-1. ^19F NMR analysis using fluorine-labeled SPM-1 revealed micromolar affinity between Cyclobutanone and SPM-1. The crystal structure of the SPM-1:cyclobutanone complex showed that the hydrated cyclobutanone forms a bond with one of the zinc ions, is stabilized through hydrogen bonding with a zinc-bound water molecule, and exhibits hydrophobic interactions with aromatic residues. These findings support future investigations into using transition-state or intermediate mimics as inhibitors for all β-lactamase classes.

Antibiotic Resistance:

As one of the most widely used antibiotic classes, β-lactams have been in clinical use for over 70 years. However, bacteria can develop resistance to these agents, most commonly through β-lactamases, including nucleophilic serine β-lactamases (SBLs) and zinc-dependent metallo-β-lactamases (MBLs). Although β-lactam/SBL inhibitor combinations like clavulanic acid offer effective treatments, resistant SBL and MBL variants continue to emerge. Carbapenems, once the last line of defense, now face localized resistance due to carbapenemase-producing Gram-negative bacteria such as E. coli and Klebsiella pneumoniae. Key β-lactamase targets include class A and D SBLs and class B MBLs (e.g., IMP-1, VIM-2, SPM-1, NDM-1).

Avibactam, a broad-spectrum SBL inhibitor and non-β-lactam β-lactamase inhibitor, is ineffective against most MBLs and some SBLs. Thus far, no hydrolytically stable inhibitor is known to inhibit both SBLs and MBLs simultaneously.

Mechanism-Based Design:

One strategy for dual inhibition involves designing mimics of common tetrahedral intermediates or transition states (see Figure 2A). Despite numerous proposed MBL-intermediate structures, structural data on intact substrates or close analogs remain limited, hindering inhibitor development. Recent studies, including ours, have explored cyclobutanone analogues of β-lactams as mechanistic probes and broad-spectrum β-lactamase inhibitor templates (see Figure 2B). Early compounds showed weak inhibition of class A SBLs, but we recently found that carbapenem-derived Cyclobutanonei nhibit both SBLs and MBLs. Among tested compounds, Cyclobutanone analogue 1 (Figure 1) showed strong inhibition of class A and C SBLs and moderate activity against IMP-1 MBL. Crystallographic evidence exists for SBL inhibition via hemiketal adduct formation, but how cyclobutanones inhibit MBLs remained unclear.

Interaction Study:

São Paulo MBL (SPM-1), prevalent in Pseudomonas aeruginosa strains from South America, Europe, and North America, is a B1/B2 hybrid with a dinuclear zinc center, making it difficult to inhibit. This study investigated the interaction between compound 1 and SPM-1.

Using ^19F NMR with BTFA-labeled cysteine mutants of SPM-1 (Y152C* and Y58C*), we analyzed binding modes. The α3-loop region of Y152C* showed no significant change upon addition of compound 1, suggesting weak interaction. However, the Y58C* mutant (L3 loop adjacent to the active site) showed peak broadening and shifts upon compound 1 binding, indicating fast exchange near Cys58. Titration experiments gave a K_D of 22 ± 7 μM.

Crystals soaked with excess compound 1 revealed the SPM-1:compound 1 complex. The structure showed a novel space group (P4222), with resolutions of 1.7 Å (apo) and 2.38 Å (complex). Electron density maps indicated compound 1 binds in its hydrated form. The active site contains two zinc ions: Zn1 (tri-histidine coordination) and Zn2 (Asp, Cys, His coordination). Wat1 bridges the two zinc ions. Notably, Zn2 shifts slightly (~0.5 Å) upon ligand binding, but no major structural changes occur in Zn-Wat1, Zn1-Zn2, or metal-ligand distances.

Key interactions include direct binding of compound 1's C4 carboxylate to Zn2 (2.48 Å) and Lys219 (2.91 Å). Weak interactions exist between Zn1 and hydrated C6-oxo atoms (3.5 Å, 4.0 Å). Bridging Wat1 remains present, possibly interacting with both C6 oxygens (2.7 Å).

SPM-1’s active site features a hydrophobic wall formed by residues such as Phe57, Tyr58, Phe79, Phe151, Tyr152, and Tyr228. Compound 1 interacts hydrophobically with several of these, especially Tyr58, Tyr228, and Phe79 (Figure 4B), consistent with ^19F NMR data. Tyr152 is relatively distant (6.5–10.8 Å). α3 helix shows conformational variability across monomers (B-factors 64–75 Ų), supporting α3 flexibility. No conformational changes were observed in the L3 loop.

Experimental Evidence:

Isotopically labeled [^13C]-1 (with ^13C at C6 and C7) was synthesized to assess the solution behavior of the compound. ^13C NMR showed peaks at 102 ppm (C6) and 98.7 ppm (C7), with a J_CC of 38 Hz, consistent with predominant hydrated form. Binding to SPM-1 reduced peak intensities without shifting them, and no new peak near 190 ppm appeared, suggesting no C6 ketone form. Thus, compound 1 binds to SPM-1 primarily as its hydrated form.

Conclusion and Significance:

These findings provide structural insight into one of the closest mimics of intact β-lactam intermediates bound to MBLs. The observed binding mode—carboxylate coordination to Zn2 and Lys219, hydrophobic interactions with surrounding residues—mirrors substrate recognition. Weak interaction between C6 oxygen and Zn1 may reflect reduced affinity due to C6 protonation, supporting interaction with Wat1 instead. The C5–H bond proximity to Zn2 may hinder stronger Zn1-C6 interactions.

While SBLs form stable tetrahedral intermediates via hemiketal bonds with serine, MBLs stabilize hydrolytic open-ring anions. Zn2-anion interactions are thought to be key. Our results demonstrate SPM-1 can bind hydrated Cyclobutanone analogues with good affinity, making these mimics promising leads for broad-spectrum, hydrolytically stable inhibitors. The tetrahedral oxyanion structures of such compounds are valuable templates for dual SBL/MBL inhibition.