Cyclobutanone Mimics of Intermediates in Metallo‐β‐Lactamase Catalysis

Abstract The most important resistance mechanism to β‐lactam antibiotics involves hydrolysis by two β‐lactamase categories: the nucleophilic serine and the metallo‐β‐lactamases (SBLs and MBLs, respectively). Cyclobutanones are hydrolytically stable β‐lactam analogues with potential to inhibit both SBLs and MBLs. We describe solution and crystallographic studies on the interaction of a cyclobutanone penem analogue with the clinically important MBL SPM‐1. NMR experiments using 19F‐labeled SPM‐1 imply the cyclobutanone binds to SPM‐1 with micromolar affinity. A crystal structure of the SPM‐1:cyclobutanone complex reveals binding of the hydrated cyclobutanone through interactions with one of the zinc ions, stabilisation of the hydrate by hydrogen bonding to zinc‐bound water, and hydrophobic contacts with aromatic residues. NMR analyses using a 13C‐labeled cyclobutanone support assignment of the bound species as the hydrated ketone. The results inform on how MBLs bind substrates and stabilize tetrahedral intermediates. They support further investigations on the use of transition‐state and/or intermediate analogues as inhibitors of all β‐lactamase classes.

Carbon tetrachloride was refluxed over P2O5 and distilled for immediate use. (+/-)-Ethyl 2,3dihydrothiophene-3-carboxylate (S3) was prepared as described. [1] Reactions were monitored by thin-layer chromatography on aluminum-backed silica plates, with spots visualized by UV and basic KMnO4 stain prepared according to a standard recipe. 1 H NMR spectra were recorded on a Bruker Avance 300 NMR spectrometer in CDCl3, and acetone-d6. Chemical shifts are reported in parts-per-million (ppm) relative to tetramethylsilane (TMS) and are calibrated to either TMS for spectra in CDCl3, or to residual solvent proton peaks for acetone (2.05 ppm for residual proton; 206.6 ppm for solvent 13 C). Mass spectra were recorded by Dr.
Richard Smith using either a Thermo Scientific Q-Exactive Orbitrap mass spectrometer for experiments requiring electrospray ionization or a JEOL HX110 Double Focusing mass spectrometer for electron impact ionization spectra in the University of Waterloo Mass Spectrometry Facility.

Synthesis of cyclobutanone 1:
The cyclobutanone 1 was prepared as a racemic mixture either via method A (as described previously) [1] or, more conveniently, via method B below.

Method A:
A mixture of the methoxy acids 2α and 2β (51.8 mg, 0.191 mmol, 10:1 dr) was stirred in a solution of 50% MsOH/CH2Cl2 (5 mL) at room temperature (rt) for 1 h. The solution was diluted with EtOAc (50 mL) and washed with H2O until the aqueous washes were no longer acidic (4 × 25 mL). The combined aqueous washes were back-extracted with EtOAc (2 × 25 mL) and CH2Cl2 (2 × 25 mL). The organic extracts were combined, dried over Na2SO4, and concentrated under reduced pressure to give the unsaturated acid 1 as an off-white solid (30.

Method B:
Alternatively, 1 could be prepared in a one-pot three-step procedure from the acid S3 via the acid chloride S4. SOCl2 (4 mL, 55 mmol) was added to a solution of acid S3 (1.475 g, EtOAc (100 mL), then washed with H2O (4 × 75 mL) until the aqueous washes showed a pH of 4. The combined aqueous washes were back-extracted with EtOAc (2 × 100 mL) and the combined organic extracts were dried over Na2SO4 and concentrated in vacuo to give the unsaturated acid 1 as a solid (1.368 g, 5.722 mmol, 94% over 3 steps) that was identical to the material prepared by method A as indicated by 1 H NMR analysis. mixture was transferred to a separating funnel; water (4 mL) and brine (5 mL) were then added. The mixture was washed with CH2Cl2 (5 × 15 mL), and the combined organic portions were dried (Na2SO4) and concentrated. The crude product, a fine beige solid, contained a small amount of apparent phthalate ester impurity (by 1 H NMR). The material was suspended in hexane (3 mL) and briefly exposed to an ultrasonic bath. After the solids had settled, the hexane was carefully withdrawn and the process was repeated. The remaining solid was dried to give 8.0 mg (62%) of [ 13 C]2-S3, which was used directly in the following
The selective incorporation of the label was confirmed by trypsin digestion and MALDI-ToF MS analyses. Circular dichroism and activity assays were followed to confirm that labeling did not affect protein's secondary structure content and activity, respectively. The method validation steps are summarized in Figure S4 as previously reported. [2a] NMR Experiments

F NMR experiments
19 F NMR spectra were recorded using a Bruker AVIII 600 MHz NMR spectrometer equipped with a BB-19 F/ 1 H Prodigy N2 cryoprobe using 5 mm diameter NMR tubes (Norell).
Cyclobutanone analogue was titrated in the assay mixture from a DMSO stock. Spectra were typically obtained using 512 scans. Trifluoroacetic acid (50 μM) was used as an internal NMR standard. Data were processed using TopSpin 3.1 software (Bruker).

C NMR experiments
NMR experiments were carried out using a Bruker AVIII 700 spectrometer equipped with an inverse TCI cryoprobe optimized for 1 H observation and installed with Topspin 3.1 software (Bruker). The assay mixture contained wt SPM-1 (0.84 mM) and ). X-ray data sets were indexed and integrated using XDS [3] or Mosflm, [4] and scaled using Aimless in the CCP4 suite. [5] Structures were solved by molecular replacement using Phaser, [6] using the native SPM-1 structure (PDB ID: 4BP0) [2b] as a search model, and completed by iterative rounds of manual model building in Coot [7] and refinement in Phoenix. [8] Structure validation was assisted by [10] Figure S2. Views from MBL crystal structures highlighting potentially mobile regions. Views of (A) IMP-1 (di-zinc ion B1 MBL, PDB ID: 1JJT) [11] and (B) CphA (monozinc ion B2 MBL, PDB ID: 1X8I)_ENREF_4 [12] highlighting the different mobile regions (L3 loop (red), and α3 region (green)) that characterize the MBL subfamilies. A longer L3 loop (red) is characteristic of the di-Zn B1 MBLs. The B2 MBLs using mono-zinc ion are characterized by an elongated α3 region (green) and a shorter L3 loop (red). Figure S3. View from an SPM-1 crystal structure. The labeled regions, α3 and L3, are highlighted in green and red, respectively. Selected active site residues are shown as sticks in the highlighted red circle. The first Zn(II) is coordinated by 3 His residues: H108, H110 and H197. The second Zn(II) is coordinated by D112, H258 and C216. The figure was created using PyMOL.

Figure S4. Outline scheme of the labeling method and labeling validation steps. [2a]
Recombinant SPM-1 variants (Y58C and Y152C) were generated by site-directed mutagenesis and produced following a three-step based purification method as reported . LC-MS analyses verified the masses of the recombinant proteins which were in agreement with the calculated masses. The observed mass difference between the unlabeled protein and its labeled counterpart corresponded to the attachment of a single CH2COCF3-label per 19 Flabeled SPM-1 positive ion. The specificity of the labeling method was evaluated by MALDI-ToF-ToF analyses following tryptic digestion. CD analyses and activity monitoring followed to confirm that labeling did not affect the protein secondary structure content and activity, respectively. [2a] SPM-1* variants were then used to study the binding of 1.   [13] and of (B) cyclobutanone 1 binding to SPM-1. The C4 carboxylate of cyclobutanone 1 is positioned to interact with both Zn2 (2.48 Å distance) and Lys219 (2.91 Å), a binding mode likely involved in substrate carboxylate binding in B1 MBLs. [13] The interaction of 1 with Zn2 is proposed to mimic the coordination of the cephalosporin-derived dihydrothiazine ring in the anionic intermediate.

Figure S8. Conformations of SPM-1 in cyclobutanone-bound crystal structures. (A)
Superposition of non-complexed SPM-1 (orange) with chain A (grey) and B (blue) of the SPM-1:cyclobutanone complex. While there is no observed conformational change in the L3 loop on inhibitor binding, the α3 region is relatively flexible (B-factors of 75 Å 2 , chain A and 64 Å 2 , chain B), and undergoes a substantial 7 Å shift away from the active site. (B) The positions and movement of Tyr58 and Y152 in unliganded and inhibitor-bound SPM-1. In the crystal structure, Tyr58 is ~4 Å away from bound cyclobutanone, while movement in the α3 region places Y152 5 Å (chain B) or 10 Å (chain A) from the cyclobutanone. [14] and of cyclobutanone 1 binding to SPM-1. Binding of the boronate is proposed to mimic that of bicyclic β-lactams. [14] Structures were superimposed using PyMOL. One oxygen atom of the boronate C-3 carboxylate coordinates to Zn2; the other carboxylate oxygen interacts with Lys224 (NDM-1 and BcII) or Arg228 (via a bridging water molecule in VIM-2) by hydrogenbonding/electrostatic interactions. These interactions are analogous to those with cyclobutanone 1 where the C-4 carboxylate binds Zn2 and Lys219 (similar to hydrolyzed cefuroxime, Figure S7).