19F NMR Monitoring of Reversible Protein Post‐Translational Modifications: Class D β‐Lactamase Carbamylation and Inhibition

Abstract Bacterial production of β‐lactamases with carbapenemase activity is a global health threat. The active sites of class D carbapenemases such as OXA‐48, which is of major clinical importance, uniquely contain a carbamylated lysine residue which is essential for catalysis. Although there is significant interest in characterizing this post‐translational modification, and it is a promising inhibition target, protein carbamylation is challenging to monitor in solution. We report the use of 19F NMR spectroscopy to monitor the carbamylation state of 19F‐labelled OXA‐48. This method was used to investigate the interactions of OXA‐48 with clinically used serine β‐lactamase inhibitors, including avibactam and vaborbactam. Crystallographic studies on 19F‐labelled OXA‐48 provide a structural rationale for the sensitivity of the 19F label to active site interactions. The overall results demonstrate the use of 19F NMR to monitor reversible covalent post‐translational modifications.


S4
were combined, concentrated (using 10 kDa molecular-weight cut-off Amicon centrifugal filters; EMD Millipore), and loaded onto a 300 mL Superdex 200 column (GE Life Sciences) preequilibrated in gel filtration buffer (50 mM HEPES, pH 7.5, 200 mM NaCl). The Superdex column was washed with one column volume of gel filtration buffer, and 5 mL fractions were collected.
Fractions of the desired purity (>95%), as identified by SDS-PAGE, were combined, buffer exchanged into 50 mM sodium phosphate, pH 7.5, concentrated using an Amicon centrifugal filter, frozen in liquid nitrogen, then stored at -80 °C. The identity and purity of the purified protein was confirmed by MS.
For 1-bromo-3,3,3-trifluoroacetone (BTFA) labelling, [2] protein aliquots were thawed on ice; neat BTFA was added to give a final ratio of 15 BTFA : 1 enzyme. The mixture was incubated on ice for 10 minutes, then buffer exchanged into 50 mM sodium phosphate, pH 7.5 using a preequilibrated PD-10 column (GE Healthcare), then dialyzed overnight against 50 mM sodium phosphate, pH 7.5 using a Slide-A-Lyzer Dialysis cassette (10 kDa molecular-weight cut-off filter). The labelled protein was concentrated using an Amicon centrifugal filter, frozen in liquid nitrogen, and stored at -80 °C. Complete labelling (within detection limits) was confirmed by mass spectrometry ( Figure S2). in CDCl 3 as an external standard. A line broadening of 10 Hz was applied to 13 C-and 19 F-NMR spectra.

Mass Spectrometry
Protein mass spectra were acquired using a Waters Micromass LCT Premier XE spectrometer coupled with an Acquity UPLC system. Samples consisted of 1 µM protein prepared in 50 mM sodium phosphate, pH 7.5. Protein signals were deconvoluted using the MaxEnt1 function of MassLynx V4.1 (Waters).

Circular Dichroism
Circular dichroism spectra were acquired at 25 °C using a Chirascan CD spectrometer (Applied Photophysics model). [3] Samples were made up of 0.2 mg/mL enzyme in 10 mM sodium phosphate, pH 7.5, and a 0.1 cm cuvette was used. The data shown represent the smoothed average of three measurements over the range of 185 nm to 260 nm, using 0.5 nm increments. Each point was averaged for 1 s.

Enzyme Kinetics
Enzymatic hydrolysis of nitrocefin and meropenem was monitored using a PHERAstar FS plate reader (BMG Labtech).

X-ray Crystallography
Purified and 19 F-labelled OXA-48 T213C* was buffer exchanged into 50 mM MES, pH K, were indexed and integrated with XDS, and scaled using SHELX. [4] The structure was solved by molecular replacement using Phaser, [5] using PDB entry 4S2P as a search model. [6] Fitting and refinement were carried out using COOT [7] and PHENIX, [8] until R work and R free no longer converged. The statistics for data collection and refinement are described in Table S2. Figure S1. View from a crystal structure of OXA-48 (PDB 3HBR), [9] showing the positions chosen for 19 F-labelling with 1-bromo-3,3,3-trifluoroacetone. The nucleophilic serine, Ser70, and the carbamylated lysine, KCX73, are represented as white sticks; the residues that were substituted with cysteine residues are shown as yellow sticks.   (Figure 4), and consistent with small molecule hydration studies. [10]   The numbers assigned to these states correspond to the 19 F signals labeled with these numbers in panels A, B, and C. Note that the relative vertical scales of the NMR spectra in panels A, B, and C are independent.   Table S1, which differed in terms of the buffer conditions and enzyme concentrations used), these data show that the ratio of lactone and hydrolysis products are similar for both wild-type OXA-48 and OXA-48 T213C*. Note that the lactone and hydrolysis products are present as a mixture of tautomers and diastereomers. [11] Figure S6. Circular dichroism spectra for the 19 F-labelled OXA-48 variants. The spectra acquired for the 19 F-labelled OXA-48 L158C* and T213C* variants resemble that of wild-type OXA-48. Based on these spectra (and as supported by the kinetic data shown in Table S1 and the crystallographic studies shown in Figure 4), the amino acid substitutions and 19 F-labelling do not appear to significantly impact on the overall folds of the OXA-48 variants, as compared to the wild-type enzyme. In the absence of bicarbonate, the peak corresponding to carbamylated OXA-48 L158C* could not be clearly observed, while a clear peak was present in the corresponding spectrum of OXA-48 T213C*. Based on these NMR spectra, it appeared that carbamylation is relatively disfavoured with the OXA-48 L158C* variant, as compared to the T213C* variant.  Figure 1C, S1); residues of the β5 and β6 strands are shown with a blue background. The high level of sequence variation in the β5-β6 loop between different enzymes suggests this is an appropriate position for the 19 F-label. The nucleophilic serine and carbamylated lysine residues are shown with yellow backgrounds. Sequence alignment was performed using Clustal Omega (https://www.ebi.ac.uk/Tools/msa/clustalo/), and the β-lactamases sequences were retrieved from the Beta-Lactamase DataBase (BLDB). [13]  with and without added sodium bicarbonate. The signals arising from the complex derived from relebactam/zidebactam and OXA-48 T213C* were assigned as the carbamylated (3) and uncarbamylated (4) states. The assignments for OXA-48 T213C* in the absence of inhibitor are based on the spectra shown in Figure 1D.  zidebactam, with 1 mM sodium bicarbonate. As the temperature of the sample was increased, the extent of carbamylation was observed to decrease (i.e., peak 3 decreased in intensity). (C) Proposed structures corresponding to the 19 F signals observed for OXA-48 T213C* with relebactam and zidebactam. These assignments are based in part on those shown in Figure S10. The R groups correspond to the structures of relebactam and zidebactam shown in Figure S10.  Figure 1D.  Figure S8, and as described previously); [12] those in orange are assigned as corresponding to the complex of vaborbactam covalently bound to carbamylated OXA-48 T213C* (resembling what was observed previously for the carbamylated complex of a cyclic boronate with OXA-10). [14] The spectra acquired suggest that OXA-48 T213C* is not fully complexed with vaborbactam (due to the apparent presence of two peaks, and the relatively weak signal intensities), consistent with the relatively poor binding interaction shown in Figure S15. However, we cannot exclude the possibility that the binding of vaborbactam may impact the dynamics of OXA-48 T213C*, thereby impacting signal intensity and broadness.  20.96 †ASU = asymmetric unit. ‡ RMSD = root mean square deviation.