Studies on enmetazobactam clarify mechanisms of widely used β-lactamase inhibitors

Significance Microbial resistance to β-lactam antibiotics mediated by β-lactamase–catalyzed hydrolysis is a major global health concern. Penam sulfones, which are structurally related to penicillins, inhibit clinically important serine β-lactamases (SBLs) by forming transiently stable covalent complexes, thereby protecting β-lactam antibiotics from hydrolysis. The characterization of these complexes and mechanisms of SBL inhibition is important for development of new SBL inhibitors (SBLi). Studies on the mechanism of the new SBLi enmetazobactam employing mass spectrometry and X-ray crystallography inform on its mode of action and also lead to reevaluation of mechanisms of current clinically important SBLi. In addition to insights into the mechanisms of transient SBL inhibition by penam sulfones, the results reveal potential for penam sulfone optimization to enable irreversible SBL inhibition.

The structures of tazobactam and sulbactam are closely related to those of the penicillins; they differ by lack of a C-6 side chain, functionalization of the pro-S methyl group (in case of tazobactam), and by oxidation of the thiazolidine to a sulfone. These differences result in a loss of useful antibacterial activity but a gain of potent SBL inhibition. Although the presence of sulfur in drugs is common [e.g., sulfonamide antibiotics (23)] and there is growing interest in covalently acting drugs (24,25), sulfones are rare in drugs and, as far as we are aware, sulbactam and tazobactam are the only clinically approved sulfone-containing drugs working by covalent reaction with their targets (26)(27)(28).
Since the clinical introduction of the pioneering SBLi, β-lactamases have evolved and SBLi use is increasingly compromised by extended spectrum β-lactamases (ESBLs)

Significance
Microbial resistance to β-lactam antibiotics mediated by β-lactamase-catalyzed hydrolysis is a major global health concern. Penam sulfones, which are structurally related to penicillins, inhibit clinically important serine β-lactamases (SBLs) by forming transiently stable covalent complexes, thereby protecting β-lactam antibiotics from hydrolysis. The characterization of these complexes and mechanisms of SBL inhibition is important for development of new SBL inhibitors (SBLi). Studies on the mechanism of the new SBLi enmetazobactam employing mass spectrometry and X-ray crystallography inform on its mode of action and also lead to reevaluation of mechanisms of current clinically important SBLi. In addition to insights into the mechanisms of transient SBL inhibition by penam sulfones, the results reveal potential for penam sulfone optimization to enable irreversible SBL inhibition. and inhibitor-resistant SBLs (29). Efforts have been made to develop new SBLi, including those with and without a β-lactam. The latter include diazabicyclooctanes (30) and cyclic boronates (31,32). However, β-lactam-containing SBLi remain of most clinical importance. Among SBLi in clinical development, enmetazobactam (formerly AAI-101; Fig. 1) is of particular interest because it is a "simple" N-methylated derivative of the triazole ring of tazobactam (33). In combination with cefepime, enmetazobactam is reported to manifest substantially better antimicrobial properties against class A ESBL-producing strains than the commonly used piperacillin/tazobactam combination (20,33,34).
We report studies on the mechanism of SBL inhibition by enmetazobactam using denaturing and nondenaturing (native) MS methods, NMR spectroscopy, and crystallography. The results led us to reevaluate the mechanisms of SBL inhibition by the clinically important sulfone-containing SBLi, i.e., tazobactam and sulbactam, and reveal limitations on the interpretation of MS studies concerning SBL inhibition.
To investigate the products of the reactions of enmetazobactam and SBLs we employed protein observed positive ion electrospray ionization (ESI) MS. We developed a solid-phase extraction linked to MS (SPE-MS) based method, in anticipation it would be useful for high-throughput assays. Compared to conventional liquid chromatography-MS (LC-MS) techniques, SPE-MS uses cartridges that manifest nonoptimal separation but which require shorter elution times (4 s compared to 3 to 10 min). The sample is applied to a cartridge, washed with 5% (vol/vol) aqueous acetonitrile (4 s), then eluted directly into the spectrometer in 95% (vol/vol) aqueous acetonitrile. As is standard procedure, to protonate proteins for positive ion ESI-MS analysis in both LC-MS and SPE-MS methods an organic acid was added to the solvents (0.1% (vol/vol) formic acid).
By contrast with reported results for enmetazobactam using a conventional LC-MS protocol (20), the SPE-MS analysis of covalent modifications of TEM-116, AmpC EC , and OXA-10 after incubation with enmetazobactam manifested a +314-Da mass increment after 20 min, consistent with formation of species with the full inhibitor mass (e.g., enamine 5 or equivalent mass species; Figs. 2 and 3). A +169-Da mass increment was also observed, likely corresponding to the product of elimination from an imine or enamine intermediate (see below). Only very weak peaks for masses corresponding to the reported (20) further fragmented products, with +52-, +70-, and +88-Da mass shifts (potentially reflecting formation of 6, 7, and 8; Fig.  2), were detected.
One difference between the SPE-and LC-MS techniques is the time that the samples are exposed to formic acid. We observed that addition of 0.1% (vol/vol) aqueous formic acid A B Fig. 1. Sulfone derivatives of penicillins are potent clinically used mechanism-based inhibitors of SBLs. (A) Outline mechanism for penicillin hydrolysis as catalyzed by SBLs; reaction proceeds via an AEC, which is efficiently hydrolyzed. (B) Sulfone derivatives of penicillins are SBLi that react to give one or more hydrolytically stable complex(es), the nature of which was the focus of our work.
to enmetazobactam-reacted samples under partially buffered conditions (50 mM Tris, pH 7.5) prior to SPE-MS analysis promoted AEC fragmentation with OXA-10 and TEM-116 to give +52-, +70-, and +88-Da species ( Fig. 3D and SI Appendix, Fig. S8); for AmpC EC it led to accelerated hydrolysis of the AEC (SI Appendix, Fig. S8). In the acid-treated samples, the À18-Da modifications produced by reaction of enmetazobactam with OXA-10 and TEM-116 were either not observed or were observed to a lesser extent (SI Appendix, Fig. S8).
To further investigate fragmentation of the enmetazobactamderived +314-Da species to give a +169-Da species, we carried out native ESI-MS experiments, which do not require acidpromoted protonation of proteins and allow for analyses using a lower energy input. Samples produced by incubation at room temperature for 10 to 30 min predominantly manifested a +314-Da species (potentially reflecting formation of transenamine 5 or equivalent mass species), with only low amounts of the +169-Da species being observed for all SBLs (Fig. 3B and SI Appendix, Figs. S9-S12). Further analyses using AmpC EC as a model system revealed correlation between levels of the +169-Da species with the gas pressure ( Fig. 3C and SI Appendix, Fig. S12) and collision energy (SI Appendix, Fig.  S13) in the spectrometer. The +169-Da species (at low levels) was also observed on incubation of the AmpC EC -derived AEC at elevated temperature (SI Appendix, Fig. S14).
SPE-MS studies on prolonged reaction over 24 h showed that the +314-Da AmpC EC modifications were relatively stable ( Fig. 5A and SI Appendix, Fig. S16). With a 10-fold excess of enmetazobactam, the unmodified forms of OXA-10 and TEM-116 were fully recovered after 3 h. With TEM-116 and a 100fold excess of enmetazobactam, a +52-Da mass increment species was observed to accumulate over 12 h, but this was not evident after 18 h. After 18 h (with likely near-complete hydrolysis of the 100 equivalents of enmetazobactam), TEM-116 and OXA-10 both showed an additional species corresponding to a mass decrease of À18-Da relative to the unmodified enzymes, with ∼50% conversion to this species for TEM-116 and ∼70% conversion for OXA-10 ( Fig. 5A and SI Appendix, Pathways for reactions of penam sulfones with SBLs. Following initial acyl-enzyme 2 formation the main transient inactivation pathway occurs via thiazolidine ring opening to give species 3-5 which are relatively stable to hydrolysis. Fragmentation of 3-5 can occur in rare cases and is promoted by acid to give 6-8 or heat to give 11. In rare cases fragmentation of 2-5 can result in irreversible inactivation of the SBL to give 9 and 10. Efficient hydrolysis of the β-lactam occurs to give a β-amino acid product 12, which in solution fragments to give 13-16. Our results imply biologically relevant inhibition involves 3-5, or equivalent mass species. In light of the substantial differences in covalently bound species observed with the different MS methods for SBLs on reaction with enmetazobactam, we reinvestigated the mechanisms of tazobactam and sulbactam. Consistent with prior studies (10,(16)(17)(18)(19)(20)(21)(22), with the LC-MS method we observed fragmentation patterns similar to those for enmetazobactam, i.e., full fragmentation to +52-, +70-, and +88-Da species (likely corresponding to 6, 7, and 8; Fig. 2) on incubation with tazobactam (100 equiv.) or sulbactam (500 equiv.) for 20 min ( Fig. 3B and SI Appendix, Fig. S17).
Addition of formic acid [0.1% (vol/vol)] to the otherwise stable TEM-116 or OXA-10 complexes derived by reaction with tazobactam or sulbactam (+300-and +233-Da species), under partially buffered conditions (50 mM Tris, pH 7.50), promoted reaction to give the +52-, +70-, and +88-Da mass increment species in less than 1 h, as observed by SPE-MS ( Fig. 3D and SI Appendix, Fig. S18). As observed with enmetazobactam, generation of the À18-Da species was less favored on acid treatment. In the case of AmpC EC , formic acid addition accelerated regeneration of the unmodified protein (SI Appendix, Fig. S18).
In SPE-MS studies on the prolonged reaction of SBLs with the penam sulfones ( Fig. 5B Table S5 summarizes calculated and observed masses. *Note mass shifts may reflect more than one structure, e.g. 2-5 ( Fig. 2) give the same mass shift. S20), modifications were observed over a longer time span when using more inhibitor (100 equiv.), with regeneration of the unmodified enzyme being observed with lower amounts of inhibitor (10 equiv.), consistent with bifurcating reactivity of AEC 2 leading to hydrolysis or transient inactivation (Fig.  1B). On prolonged incubation (8 h) of TEM-116 and OXA-10 with 100 equiv. of tazobactam, complete conversion to the À18-Da mass shift species was observed by SPE-MS with OXA-10 (suggesting conversion in more than 1 in 100 turnover events), but the À18-Da species was not observed with TEM-116 (Fig. 5B). Small amounts of the À18-Da modification were also observed on incubating OXA-10 with 500 equiv. of sulbactam (SI Appendix, Fig. S20). Over 5 h low levels of a +52-Da species were observed to accumulate for TEM-116, followed by regeneration of the unmodified enzyme ( Fig. 5B and SI Appendix, Fig. S19). We investigated products obtained by SBL-catalyzed hydrolysis of enmetazobactam, tazobactam, and sulbactam in solution by 1 H NMR spectroscopy (750 MHz). In each case, incubation of the inhibitor (100 equiv.) with OXA-10 or TEM-116 manifested efficient turnover in <18 h; AmpC EC turnover was relatively slow (SI Appendix, Figs. S21-S25), consistent with the slower acylation observed in competition studies with FC-5 (SI Appendix, Figs. S5-S7) and SPE-MS (SI Appendix, Figs. S16, S19, and S20). In all cases the penam sulfone reacted efficiently to give fragmentated products, i.e., amine 14 (which appeared stable over the timescale of analysis) and aldehyde 15 (which underwent further decarboxylation and hydration; SI Appendix, Figs. S21-S25).
To further investigate the À18-Da species, native-MS analyses of TEM-116 and OXA-10 samples treated with 100 equiv. of enmetazobactam or tazobactam for 24 h were performed to rule out acid-mediated interference in the generation of the À18-Da modifications; under these conditions the À18-Da modifications were also observed, consistent with the SPE-MS results (SI Appendix, Fig. S26). In accord with a previous report (20), À18-Da-modified SBLs showed no β-lactamase activity when assayed with FC-5 (SI Appendix, Fig. S27).
To test for formation of Dha-containing species 9, the À18-Da protein product was subjected to reaction with a thiol, a reaction used in protein engineering to introduce covalent modifications (38). No reaction was observed on incubation of unmodified OXA-10 with β-mercaptoethanol (BME, 1,000 equiv.), as monitored by SPE-MS (SI Appendix, Fig. S28). By contrast, after complete modification of OXA-10 to give the À18-Da species (8 h with excess tazobactam; validated by SPE-MS) and subsequent addition of BME (1,000 equiv.), ∼80% of the À18-Da protein was converted to a +60-Da species (relative to unmodified OXA-10) within 2 h. This observation is in agreement with the addition of BME to Dha 9 to give 2-hydroxyethyl-cysteine (Dha-BME, 17; Fig. 6). Interestingly, this reaction did not progress further, even after overnight incubation with more BME (SI Appendix, Fig. S28). The extent of conversion to the +60-Da species 17 was increased to ≥98% by simultaneous      Table S5 summarizes calculated/observed masses/mass shifts. *Note mass shifts may reflect more than one structure, e.g. 2-5 and 9-10 ( Fig. 2) give the same mass shift.
incubation of OXA-10 with tazobactam (100 equiv.) and BME (1,000 equiv.) for 12 h and was decreased by prolonged incubation with tazobactam prior to BME addition (SI Appendix, Fig.  S28). These results imply a slow inactivation of the initially formed Dha residue 9 toward reaction with BME, a proposal rationalized by subsequent studies (see below). Reaction of unmodified TEM-116 with BME manifested slow addition of a single BME molecule over 24 h (likely via thiol disulfide exchange with either one of the two disulfidebound Cys residues; SI Appendix, Fig. S29). When TEM-116 was first treated with enmetazobactam and then BME, analogous slow addition of a single BME was observed; no addition of a second BME was observed, even on prolonged incubation, as would be expected for reaction of BME with a Dha residue 9 (SI Appendix, Fig. S29). By contrast with OXA-10, reactivity of the TEM-116 derived À18-Da modified protein with BME could not be increased by simultaneous incubation of TEM-116 with enmetazobactam and BME. These results suggest that if a Dha residue 9 is formed on TEM-116, inactivation of its reaction with BME proceeds more rapidly than with OXA-10.
MS fragmentation (MS/MS) analysis of trypsin-hydrolyzed SBLi-treated samples of OXA-10 and TEM-116 supported the presence of modification of the nucleophilic Ser. With OXA-10 treated simultaneously with tazobactam and BME, a +60-Da modification (in agreement with Dha-BME 17) was observed at Ser67 (SI Appendix, Fig. S30). Notably, whereas MS/MS analysis of the tryptic peptide containing Ser70 residue from unmodified TEM-116 sample showed complete backbone fragmentation (SI Appendix, Fig. S31A), no MS/MS fragmentation of the corresponding enmetazobactam-reacted À18-Da modified peptide was observed beyond residues Lys73 or Ser70, consistent with cross-linking of the side chains of these residues (see below; SI Appendix, Fig. S31B).
Samples of OXA-10, treated with tazobactam or tazobactam in combination with BME as described above (with full conversion to the respective À18and +60-Da species verified by SPE-MS) (SI Appendix, Figs. S32 and S33) were crystallized. Three highresolution structures with two molecules in the asymmetric unit (ASU) were obtained (SI Appendix, Table S7), each using similar crystallization conditions (see Materials and Methods). In a crystal structure of À18-Da modified OXA-10 derived by tazobactam treatment (PDB: 7B3S, 1.85-Å resolution) clear modification of Ser67 to Dha 9 was observed in both chains, as indicated by the planar geometry of the Cα atom compared to Ser67 of unmodified OXA-10 [PDB: 2X02 (39); Fig. 6 and SI Appendix, Fig.  S32]. The structure of OXA-10 incubated simultaneously with tazobactam and BME to give a protein with +60-Da mass increment (PDB: 7B3U, 1.60-Å resolution) showed continuous density consistent with Dha-BME 17 formation, both in chain A and chain B (Fig. 6 and SI Appendix, Fig. S33). Notably, although generation of small amounts of the L(2R)-epimer cannot be ruled out, the addition of BME to Dha 9 appears stereoselective with only the D(2S)-residue being observed. In both chains the hydroxyethyl oxygen is oriented so that the OH is positioned in the same location as the deacylating water in the active site of unmodified OXA-10 (SI Appendix, Fig. S33).
In a second structure of the À18-Da modified OXA-10 (PDB: 7B3R, 1.83-Å resolution), connecting electron density was observed between the C-3 side-chain carbon of the original Ser67 and the side-chain nitrogen of Lys70, indicating partial (∼50%) addition of the Lys N ε -amine to Dha 9, forming a lysinoalanine (Lal) cross-link 10; note this species also manifests a À18-Da mass shift ( Fig. 6 and SI Appendix, Fig. S32). Evidence for cross-linking was observed in both OXA-10 chains. In chain A, clear and continuous electron density for both the Lal crosslink 10 and some unreacted carbamylated lysine was observed. The quality of the electron density map was overall less clear in chain B, though the mF o -DF c polder OMIT map (40) indicates partial cross-linking. However, no clear density for the remaining carbamylated lysine was apparent in chain B, which was therefore not included in the final model (SI Appendix, Fig. S32J). By contrast, with the D(2S)-stereochemistry observed on BME addition to Dha 9, the new chiral center formed by reaction of Lys70 with Dha 9 has, at least predominantly, the L(2S)stereochemistry.

Discussion
Removal of the penicillin C-6 side chain coupled with oxidation of the thiazolidine sulfide to a sulfone converts antibiotics into clinically important SBLi, i.e., tazobactam and sulbactam. Enmetazobactam is a new member of the penam sulfone SBLi class, differing from tazobactam solely by addition of a methyl group on its triazole ring, a modification that confers a permanent positive charge, with consequent potential mechanistic effects (20,33). In accord with prior studies (20,34), we observed only small differences in potency for enmetazobactam compared to the closely related tazobactam (SI Appendix, Tables S1-S3). The reported significantly enhanced potential of enmetazobactam against class A ESBLs in cells (33,34) and in vivo (41) may thus result from improved properties with respect to variables other than potency against isolated SBLs, e. g., outer membrane permeability or efflux pump susceptibility.
By contrast with previous studies employing ESI-MS to investigate SBL inhibition by tazobactam and sulbactam (10,(16)(17)(18)(19)(20)(21)(22), SPE-MS studies on the mechanism by which enmetazobactam inhibits representative SBLs indicate that it reacts to give one or more covalently bound species with the mass of the intact inhibitor. Evidence for the reported fragmentations to +52-, +70-, and +88-Da species (corresponding to 6, 7, and 8 in Fig. 2) was not observed. The SPE-MS studies, however, did provide evidence for partial fragmentation, to give a previously unidentified (i.e., not reported in studies on sulbactam/ tazobactam) +169-Da adduct, potentially resulting from elimination of the triazole and sulfinic acid groups to give alkene 11. However, studies under native-MS conditions imply that formation of this +169-Da species is promoted under high-energy conditions, generated either by collisional activation in the gas phase for MS analysis or by incubation at higher temperature in the aqueous phase, and thus formation of the +169-Da species is likely of limited relevance in physiological conditions. ESI-MS condition influenced fragmentation of the covalent modifications of SBLs by other SBLi has also been reported, e.g., desulfation of avibactam (43,44).
Subsequent studies with enmetazobactam, as well as tazobactam and sulbactam, comparing the results obtained by native-, SPE-, and LC-MS, show that the observed lack of +52-, +70-, and +88-Da species in the SPE-MS studies compared to the LC-MS studies with all three penam sulfones results from a milder sample preparation method and does not reflect a distinctive inhibition mode for enmetazobactam. Formation of species with mass shifts corresponding to the reported fragmentation products 6, 7, and 8 is promoted by the acidic conditions commonly employed to protonate proteins within conventional denaturing ESI LC-MS workflows (e.g. 0.1% formic acid). For all the penam sulfones, SBL inhibition thus appears to result from formation of transitory stable species with the intact inhibitor mass.
By contrast with the SPE-MS studies showing evidence for protein-bound species comprising the full mass of the inhibitors (i.e., trans-enamine 5 or equivalent mass species) during SBL inhibition, we did not observe evidence for unfragmented AEC-derived hydrolysis products (i.e., enamines/imines) by NMR studies in solution (SI Appendix, Figs. S21-S25). Although it is possible that the NMR observed products may (in part) result from fragmentation of enzyme-bound imine/ enamine complexes (3, 4, or 5; Fig. 2) followed by hydrolysis, it is likely that they result from hydrolysis of the nascent acyl-enzyme 2 or imine/enamine intermediates (3, 4, or 5), followed by efficient fragmentation in solution, as reported for nonenzymatic hydrolysis of clavulanic acid (45,46).
Although most studies, including our native-MS and SPE-MS studies, point to an AEC species with the full mass of the inhibitor (i.e., trans-enamine 5 or equivalent mass species) as the major (but not necessarily sole) species for inhibition, the SPE-MS analysis and some crystallographic (22,(47)(48)(49) and UV/Vis-based studies (e.g., ref. 50) provide evidence for rare fragmentation events to give smaller species, e.g. the +169-Da, +52-Da, and À18-Da species. Our results suggest that generation of these species by reaction of SBLs with the current penam sulfones is unlikely to contribute substantially to inhibition in a biological context. However, since some of these species appear resistant to hydrolysis, the future development of sulfone-based SBLi that efficiently generate such species is of interest.
In this regard, it is important to note that by SPE-MS we only observed formation of a +169-Da mass species with enmetazobactam. The additional methyl group of enmetazobactam relative to tazobactam may promote elimination to give the +169-Da species 11, because it ensures a permanent positive charge on the triazole, thereby promoting its loss. A related fragmentation involving loss of SO 2 has been observed for 2βalkenyl penam sulfones with increased activity against class A and C SBLs compared to tazobactam (51). Future studies could thus focus on modification of penam sulfones to promote formation of the +169-Da or equivalent species under biologically relevant conditions.
Most reported evidence suggest a Ser-Ser cross-linked species (i.e., vinyl-ether 8; Fig. 2) as the molecular basis for the +52-Da species, though there are other possibilities (50,52,53). Studies have also shown that the cross-linked +52-Da species may react further to give species with analogous mass increments which confer irreversible inhibition of the SBLs (52). However, in our studies using an excess of tazobactam and enmetazobactam, the small amounts of the +52-Da species that were observed for TEM-116 eventually degraded and did not appear to lead to irreversible inactivation.
A particularly interesting aspect of the prolonged reaction of the penam sulfones with TEM-116 and OXA-10 is formation of catalytically inactive À18-Da species. The combined SPE-MS (including derivatization trials with BME), LC-MS/MS, and crystallographic results imply initial formation of a Dha residue 9, which reacts with N ε -amine of a Lys side chain to give the cross-linked Lal species 10 (Fig. 6). Formation of Lal cross-links can occur on treating proteins with heat or high pH and occurs during biosynthesis of protein-derived lanthipeptide antibiotics (54). To date Lal 10 has rarely been identified in natural protein structures under biological conditions, with the spirochaete flagella hook protein being a notable example (55), and it has not previously been described with a β-lactamase. The different stereoselectivities observed for intermolecular (BME) and intramolecular (Lys) additions to Dha 9 are of interest from the perspective of use of Dha in protein engineering (38).
From the inhibition perspective, the Dha 9 and Lal crosslinked 10 species (which cannot be directly distinguished by intact protein MS) are of particular interest as they are the only modifications we observed leading to apparently irreversible inactivation of both class A and D SBLs. While a range of small molecules are known to react with serine, or, more commonly, cysteine residues, to give Dha 9 which may be utilized for protein engineering purposes (38), these compounds usually lack sufficient selectivity and are thus not useful for enzyme inhibition in a biological context. However, selective Dha formation has also been observed on reaction of the class C β-lactamase P99 with β-sultams (56), suggesting general scope for irreversible SBL inhibition through serine dehydration. The apparent differences in efficiency of Dha 9 formation between tazobactam and enmetazobactam on reaction with TEM-116 and OXA-10 shown here (Fig. 5) highlight opportunities for optimization of penam sulfones to result in irreversible β-lactamase inhibition via Dha/Lal formation.
Overall, our work provides insight into the mechanisms of some of the most widely clinically used β-lactamase inhibitors which will help guide their further optimization. The results also further exemplify the power of protein observed MS for studying mechanism-based inhibition involving covalent reactions. However, they show care should be taken in interpretating the MS results, especially when obtained under conditions far from biologically relevant ones, which should be considered in the light of data from other methods.
Enzyme Production. Recombinant AmpC EC , TEM-116 with an N-terminal His tag, and OXA-10 with a cleavable N-terminal His tag were expressed and purified as previously described (57)(58)(59).
IC 50 s were determined as reported (35). TEM-116 (1 nM), AmpC EC (500 pM), or OXA-10 (250 pM) were incubated with varied inhibitor concentrations at room temperature (r.t.) for the indicated time then assayed using 5 μM FC-5. The apparent inhibitory constant K iapp and the second-order rate constant k inact /K (or k 2 /K) were determined using reported methods (60). SBLs were reacted with the reporter substrate in the presence of varied inhibitor concentrations. Reactions were initiated by SBL addition and immediately monitored for 60 to 120 s (until they plateaued). For TEM-116 (1 nM) assays were carried out in competition with NCF (50 μM); for AmpC EC (100 nM) and OXA-10 (50 nM) FC-5 (5 μM) was used. K iapp 0 values were obtained by linear regression analysis of initial velocities at varied inhibitor concentrations and corrected to account for the substrate concentration and Michaelis constant (K m ) to give K iapp . k obs values were determined by fitting the obtained progress curves to Eq. 1, where P is formed product, P 0 is background signal, V S is velocity of no-inhibitor control, V 0 is velocity of no-enzyme control to estimate fully inhibited enzyme, and t is time: Linear regression of the obtained k obs values against the inhibitor concentration gave k inact /K 0 , which were corrected to account for the substrate concentration and K m to give k inact /K. Dissociation constants (k off ) were determined by the jump dilution method (61). TEM-116 (3 μM) was incubated with the inhibitor (300 μM, 20 min, r.t.). AmpC EC (3 μM) was incubated with the inhibitor (900 μM, 20 min, r.t.). OXA-10 (3 μM) was incubated with tazobactam (900 μM), enmetazobactam (900 μM), or sulbactam (3 mM) (15 min, r.t.). Reactions were serially diluted (to final concentrations of TEM-116: 30 pM; AmpC EC : 10 pM, OXA-10: 10 pM) and assayed using 25 μM FC-5. Reactions were monitored for 30 to 400 min then fitted to Eq. 1, where V S is velocity of no-enzyme control to estimate fully inhibited enzyme and V 0 is velocity of no-inhibitor control. Half-lives for SBL inhibition (t 1/2 ) were obtained using Eq. 2: The partition ratio (k cat /k inact ) between transient inhibition and efficient hydrolysis of the AEC was determined as reported (20). TEM-116 (1 nM), AmpC EC (500 pM), or OXA-10 (250 pM) were incubated with varying concentrations of inhibitors for 60 min at r.t. then assayed with FC-5 (5 μM). The inhibitor-enzyme ratio resulting in ≥90% inhibition was taken as an estimate of the partition ratio. ; Thermo Fisher Scientific) as described by the manufacturer. Samples (3 μL) were loaded into in-house prepared goldcoated capillary needles (Harvard Apparatus) and were injected into a Q-Exactive UHMR Hybrid Quadrupole-Orbitrap spectrometer (Thermo Fisher Scientific) (62). Instrument parameters: capillary voltage 1.2 kV, S-lens RF 200%, mass range from 1,000 to 12,000 m/z, capillary temperature 60°C, resolution of the instrument 17,500 at m/z = 200 (transient time: 64 ms). The noise level was 3, rather than the default of 4.64. In some cases in-source dissociation energy (0 to 100 V) was applied. While analyzing the dependence of elimination from the AmpC EC -enmetazobactam-derived complex, the HCD activation collisional energy and UHV gas pressure were varied. Calibration of the instruments was performed using a 10 mgÁmL À1 solution of CsI in water. Data were analyzed using Xcalibur 4.1 (Thermo Fisher Scientific).
Raw data files were processed using MaxQuant Version 1.6.3.4 with the Andromeda search engine (63,64). Peak lists were searched against individual sequences and potential contaminant proteins. Carbamidomethylation was kept as a fixed modification whereas acetylation (protein N-term), oxidation (methionine), and dehydration (serine) and its corresponding adducts with BME and DTT were variable modifications. The peptide false discovery rate was kept at 1%. Trypsin was set as the protease and up to four missed cleavages were allowed. Spectra identifying cross-linked peptides were manually validated.