19F‐NMR Reveals the Role of Mobile Loops in Product and Inhibitor Binding by the São Paulo Metallo‐β‐Lactamase

Abstract Resistance to β‐lactam antibiotics mediated by metallo‐β‐lactamases (MBLs) is a growing problem. We describe the use of protein‐observe 19F‐NMR (PrOF NMR) to study the dynamics of the São Paulo MBL (SPM‐1) from β‐lactam‐resistant Pseudomonas aeruginosa. Cysteinyl variants on the α3 and L3 regions, which flank the di‐ZnII active site, were selectively 19F‐labeled using 3‐bromo‐1,1,1‐trifluoroacetone. The PrOF NMR results reveal roles for the mobile α3 and L3 regions in the binding of both inhibitors and hydrolyzed β‐lactam products to SPM‐1. These results have implications for the mechanisms and inhibition of MBLs by β‐lactams and non‐β‐lactams and illustrate the utility of PrOF NMR for efficiently analyzing metal chelation, identifying new binding modes, and studying protein binding from a mixture of equilibrating isomers.


Reagents
Tazobactam, meropenem, and sodium piperacillin were from Molekula. Avibactam was a gift from AstraZeneca. Unless otherwise stated, other reagents were from Sigma-Aldrich.

Mutagenesis
The Y58C SPM-1 and F151C SPM-1 variants were generated using the wildtype (wt) pOPINF SPM-1 plasmid. [1] The desired point mutations were achieved employing the QuikChange site directed mutagenesis kit (Stratagene) using the following primers:
The treated proteins were purified using a Superdex S200 column (300 mL) equilibrated with 20 mM Tris pH 7.5, 200 mM NaCl. Fractions containing purified protein were concentrated by centrifugal ultrafiltration (10 kDa cutoffs). The purity of the resulting fractions was ascertained to be > 90% by SDS-PAGE. Concentrations of the purified proteins were determined using a ND-1000 NanoDrop spectrophotometer. Removal of the metal was confirmed by non-denaturing mass spectrometry and activity analyses.

Protein Electrospray Ionization Mass Spectrometry
Low resolution positive ion electrospray mass spectra were recorded using an LCT Premier XE ionization (Micromass ® ) mass spectrometer interfaced with a Acquity ™ Ultra Performance liquid chromatography (UPLC) system using a Acquity ™ UPLCR BEH300 C18 column at 50 ºC (Waters Corporation).

In-Solution Trypsin Digestion
All reagents were prepared in Tris buffer (100 mM, pH 7. MALDI target and allowed to air-dry before analysis.

Circular Dichroism (CD) Experiments
CD measurements were carried out using a Chirascan CD spectrometer (Applied Diffraction data were collected at 100 K on beamline I24 (Diamond Light Source, Didcot, UK). The dataset was indexed and integrated using XDS [5] and scaled using Aimless in CCP4. [6] Data were cut at 1.75 Å as completeness dropped significantly at higher resolutions.
Phases were determined using molecular replacement in Phaser [7] with 4BP0 [2] as the search model (with Y58 mutated to a cysteine). The structure was refined and modelled using Phenix [8] and Coot, [9] respectively. After several rounds of refinement residue 58 was modelled as a cysteine into clearly defined electron density, before undergoing a final round of refinement and model building. Structure validation was assisted by Molprobity [10] and Phenix. Data collection and refinement statistics are given in Table S1. and k cat ) were obtained by determining the initial rate of the reaction at different substrate concentrations. The concentration-dependence of the initial rate was fitted and analyzed using GraphPad Prism 5.01 software to generate Michaelis-Menten curves. [11] Synthesis ML302 and ML302F were synthesized as reported by Brem et al. [12] The isoquinoline derivative (1) was synthesized as reported by van Berkel et al. [11] Production of Hydrolyzed β-Lactams β-Lactamase mediated hydrolysis was used to produce hydrolyzed meropenem and piperacillin. Incubation of 5 equivalents of piperacillin and meropenem with Bacillus cereus m569/H/9 (BcII) produced (5R)-penicilloic acid (PA) [and some (5S)-PA] and hydrolyzed meropenem, respectively. BcII was removed from the reaction mixture using PD-10 columns.

Steady-State Kinetics
Subsequent epimerisation of (5R)-to (5S)-PA was performed under neutral conditions, nonenzymatically, at 4 o C, for an overnight. The hydrolyzed products were obtained via spin filtration using 10 kDa Centricon concentrators. The reaction and the purity of hydrolyzed βlactams were assessed by NMR analyses. BcII was produced and purified as described by van Berkel et al. [11]

NMR Experiments
Spectra were recorded using a Bruker AVIII 600 with BB-19 F/ 1 H Prodigy N 2 cryoprobe operating at 298 K using 5 mm diameter NMR tubes (Norell).

Supplementary Figures
1JJT) [15] and (B) CphA (a mono-Zn(II) binding B2 MBL, PDB ID: 1X8I) [16]           The obtained crystal structure shows that Y58C SPM-1 has the conserved αββα MBL fold. [19]    The results imply that the F151C* label is more exposed to solvent than the Y58C* label (see Fig. S2B).                     Tables   Table S1. Data collection and refinement statistics.  Note, Δδ max = δ max -δ 0 represents the chemical shift difference (in ppm) between δ max and δ 0 , in which δ max = δ observed in the presence of 400 μM of the ligand and δ 0 = δ of the SPM-1* variant in the absence of added ligand: δ 0 = -83.15 ppm for Y58C* SPM-1 and δ 0 = -84.75 ppm for F151C* SPM-1 (Fig. S18)   and slow hydrolysis were monitored (CPMG). [22]  Hydrolyzed products do not appear to bind to SPM-1 (wLOGSY). [22]  Addition of intact avibactam after 24 h led to shifting the equilibrium back to the avibactam.Y58C* SPM-1 complex peak.  K D values were fitted using the following equation: [2] Δδ max = δ max -δ 0 represents the chemical shift difference (in ppm) between δ max and δ 0 , in which δ max = δ observed in the presence of 400 μM of the ligand and δ 0 = δ of SPM-1* variant in the absence of added ligand; Δδ obs = δ obs -δ 0 represents the chemical shift difference (in ppm) between δ obs and δ 0 , in which δ obs = δ observed at an added ligand concentration; [P] t is the total protein concentration; [L] t is the total ligand concentration.

Meropenem
* For slow-exchange systems, the changes in the integration of the resonance corresponding to the protein-ligand complex were fitted instead of the chemical shift changes.
** For very weak binders, an estimate of the K D was given as the observed curves did not reach saturation.
*** K D was not determined for 1,10-o-phenanthroline or (1). The behavior of 1,10-ophenanthroline does not follow standard kinetics (likely due to the competition involving the binding of the metal to the protein and that of the metal to 1,10-o-phenanthroline). Addition of (1) led to very substantial signal broadening such that it merged with the baseline. **** Values are not given for clavulanate and tazobactam since these serine β-lactamase inhibitors undergo hydrolysis to multiple products. [23]