Structural Basis of Metallo-β-lactamase Inhibition by N-Sulfamoylpyrrole-2-carboxylates

Metallo-β-lactamases (MBLs) can efficiently catalyze the hydrolysis of all classes of β-lactam antibiotics except monobactams. While serine-β-lactamase (SBL) inhibitors (e.g., clavulanic acid, avibactam) are established for clinical use, no such MBL inhibitors are available. We report on the synthesis and mechanism of inhibition of N-sulfamoylpyrrole-2-carboxylates (NSPCs) which are potent inhibitors of clinically relevant B1 subclass MBLs, including NDM-1. Crystallography reveals that the N-sulfamoyl NH2 group displaces the dizinc bridging hydroxide/water of the B1 MBLs. Comparison of crystal structures of an NSPC and taniborbactam (VRNX-5133), presently in Phase III clinical trials, shows similar binding modes for the NSPC and the cyclic boronate ring systems. The presence of an NSPC restores meropenem efficacy in clinically derived E. coli and K. pneumoniae blaNDM-1. The results support the potential of NSPCs and related compounds as efficient MBL inhibitors, though further optimization is required for their clinical development.

T he β-lactams are one of the most important antibacterial classes; 1 however, their efficacy is increasingly being eroded by resistance, most importantly by β-lactamases. 2 Even carbapenems, often used as "last resort" antibiotics, are often no longer effective due to production of carbapenemases by Enterobacteriaceae strains. 3 Ambler class A, C, and D βlactamases are nucleophilic serine enzymes (serine-β-lactamases, SBLs), whereas class B are metallo-β-lactamases (MBLs). 4 Combinations of a β-lactam antibiotic and a β-lactam containing SBL inhibitor have long been used as a treatment option for bacterial infections; however, no MBL combinations are clinically approved.
Presently, only a few subclasses of SBLs, 5,6 e.g. Class A KPCs and Class D OXAs, are reported to efficiently hydrolyze carbapenems, and these can be countered by clinically available SBL inhibitors, e.g. avibactam. 7,8 Class B MBLs, however, can hydrolyze all carbapenems and β-lactam containing SBL inhibitors. 9−11 MBL inhibition is challenging in part because of the need to obtain activity against a range of relevant enzymes, which vary in their active site details. 12,13 Subclass B1 and B3 MBLs are dizinc ion enzymes, whereas B2 MBLs employ one zinc ion. The B1 MBLs are the most important from a clinical perspective and include the IMP (imipenemase), NDM (New Delhi MBL), and VIM (Verona integron-encoded MBL) MBL subfamilies. 14 Reported MBL inhibitors include bicyclic boronates, thiols, and succinate derivatives ( Figure 1). 15,16 Recently, substituted pyrroles and related compounds have been described as MBL inhibitors in the patent and scientific literature. 17−20 Wachino et al. have reported that substituted pyrroles and furans bearing αcarboxylic acid and N-sulfamoyl functional groups are effective MBL inhibitors, in particular of the B1 subclass. 21 N-1 Sulfamoylpyrrole-2-carboxylates have also been reported as B1 MBL inhibitors, 22 though no structures of them in complex with MBLs are reported. Structurally related sulfonamide-based inhibitors of metalloenzymes are used therapeutically, e.g. as carbonic anhydrase inhibitors with broad clinical utility. 23−25 The N-sulfamoylpyrrole compounds are of mechanistic interest because the (initial) binding mode of some classes of potent βlactamase inhibitors can mimic that of substrates (e.g., clavulanic acid) or tetrahedral intermediates (boronates). 26 We envisaged that the approximately tetrahedral geometry about the sulfamoyl sulfur 27,28 may mimic the tetrahedral intermediate formed during β-lactam hydrolysis, and the Lewis basicity of the oxygen and nitrogen atoms may enable effective coordination to the zinc ion. Here, we report on the mechanism of action and B1 MBL potencies of N-sulfamoyl-substituted pyrrole-2-carboxylic acids (NSPCs).

■ RESULTS AND DISCUSSION
We targeted the synthesis of NSPC 6a, which has a parafluorophenyl substitution at its C3 position, because this substituent has been identified as being preferred in a related series of published pyrrole inhibitors (Scheme 1). 33,34 We aimed to employ mild hydrogenolytic deprotection to prepare the NSPCs because of potential competitive decarboxylation of pyrrole-2-carboxylic acids 35,36 and N1-sulfonyl group cleavage under acidic or basic conditions. 35−37 The efficient synthesis of 6a was readily achieved in seven steps from pyrrole (1) (12% overall yield, Scheme 1). Initial N-sulfonylation of 1 with PhSO 2 Cl was followed by regioselective electrophilic C3bromination using Br 2 . Subsequent directed ortho-metalation and electrophilic trapping with benzylchloroformate (CbzCl) gave C2-substituted benzyl ester 3. The N-Cbz protected sulfamoyl group was installed in good yield by tetrabutylammonium fluoride (TBAF)-mediated N-sulfonyl deprotection, followed by deprotonation of the pyrrole NH with sodium hydride and then electrophilic trapping with zwitterionic sulfamoylating reagent 7. 38 Subsequent Pd-catalyzed Suzuki− Miyaura cross-coupling with 4-fluorophenylboronic acid afforded the C3-aryl derivative 5 as a sodium sulfonylazanide salt which, upon hydrogenation, gave sodium carboxylate 6b (93%). The free acid 6a was obtained from 5 by acidification with aqueous HCl, followed by global deprotection with Pd/C/ H 2 and purification by reverse phase HPLC in 69% yield.
With a robust synthesis of 6 in hand, we synthesized seven other NSPC derivatives varying the C3-pyrrole substituent, with bromopyrrole 4 providing a convenient vector for late-stage diversification with aryl and heteroaryl groups via Suzuki− Miyaura coupling, followed by hydrogenation (8−14, see Supporting Information). Unexpectedly, under the hydrogenation conditions used for the preparation of amino-  19 and related sulfonamide and sulfamoyl inhibitors. 21,22,32 Zinc-chelating functional groups are highlighted in blue. pyrimidine 11, near equimolar quantities of zwitterionic cyclic guanidine 12 were also formed due to over-reduction; 39 both products were separated by preparative HPLC. For the preparation of 14, it was necessary to first install the pinacol boronate ester at the C3 position of pyrrole 4 by Pd-catalyzed Miyaura borylation; the intermediate boronate ester then underwent coupling with the commercial heteroaryl bromide.
Our pIC 50 values for 9 obtained from the fluorescence-based assay are higher than the previously reported pIC 50 values for AMRC272 when using a nitrocefin-based assay. 22 The discrepancy likely reflects different enzymatic assay conditions, protein constructs, and enzyme purification procedures; however, it should be noted that both assays show a clear preference for inhibition of IMP-1 over VIM-2. 22 Crystallography. We investigated the NSPC ligand− enzyme interaction by crystallography and obtained a structure for VIM-1 complexed with 6 (space group: P12 1 1, 1.21 Å resolution, Figure 2). The structure was solved by molecular replacement (PDB: 5N5G), 44 with iterative fitting of 6 at the active site. The two zinc ions (coordinated by H114, H116, D118 (Zn1), H179, C198, and H240(Zn2)) were refined with occupancies of 0.75 for both Zn1 and Zn2. The reduced occupancy for Zn1 is due to partial oxidation of Cys198 to a 3sulfino alanine residue (Csd198), which was modeled and refined in a ratio of Cys (75%) to Csd (25%), as shown in previous work on VIM-1 (PDB 5FQA). 45 Uncomplexed VIM-1:Zn 2 (PDB: 5N5G) 44 has a dizinc bridging hydroxide/water, as do other B1MBls. 26 The NH 2 -of   sulfamoyl group of 6 replaces this "hydrolytic" water, probably in its deprotonated form, though this cannot be discerned from the crystal structure. The C2-carboxylate of 6 ligates to Zn2, as observed in substrate derived complexes and those of inhibitors with analogously placed carboxylates, including CB2/VNRX-5133 ( Figure 1). 41−43 The binding mode of 6 is related to that of bicyclic boronate MBL inhibitors, (e.g., VIM-2:Zn 2 :CB2 (PDB: 5FQC, superimposition Figure 2b)), 43 in which the binding of the two boronbound oxygens mimics the binding modes proposed for the two oxygens in the oxyanion intermediate in MBL catalysis. However, in the NSPC complex, one of these oxygens is "replaced" by the tetrahedral sulfamoyl amino group (NR 2 -S- Figure S1a). Comparison of the Zn(1)−Zn(2) distances in the VIM-1:Zn 2 :6 and the VIM-2:Zn 2 :CB2 complex reveals differences. The Zn(1)−Zn(2) distance is increased in the VIM-2:Zn 2 :CB2 complex to 4.34 Å compared to 3.47 and 3.62 Å 44 in the unligated VIM-2:Zn 2 (PDB: 5N5G) and VIM-1:Zn 2 (PDB: 4NQ2) complexes, respectively 46 (see Supporting Information Figure S1e). Binding of 6, however, does not substantially increase the Zn(1)−Zn(2) distance, i.e. it is 3.60 Å compared to the unligated VIM-1/VIM-2. This distance is similar to those reported for a hydrolyzed VIM-1:Zn 2 :meropenem complex (PDB: 5N5I, Zn(1)−Zn(2): 3.50 Å, Figure S1d) 44 and other unligated VIM family members. 47 Antimicrobial susceptibility testing of the NSPCs in combination with Meropenem, following CLSI guidelines, 49,50 was performed in a minimum inhibition concentration (MIC) antimicrobial assay format with 4 clinically relevant NDM-1 producing strains of Escherichia coli and Klebsiella pneumoniae ( Table 2). At a fixed concentration of 8 μg mL −1 , the Nsulfamoyl pyrroles reduce the meropenem MIC from 64 to ∼0.375 μg mL −1 . In all cases, the MICs for the NSPCs were better than those for VNRX-5133. At a concentration of 0.5 μg mL −1 of the NSPC, the analogue potencies can be compared; 6 and 8−10 exhibit greater potency compared to the higher mass compounds 13 and 14, bearing 4-morpholinophenyl or bicyclic heteroaromatic groups at C3, respectively. These differences may reflect differences in uptake as the pIC 50 s of 13/14 against NDM-1 are consistent with the other analogues (Table 1). Furthermore, the comparative data for aminopyrimidine 11 and its saturated analogue 12 show the latter is significantly less active at 0.5 μg mL −1 , in particular with the B68-1, S117, and IR47 strains, likely reflecting their relative NDM-1 pIC 50 values (Table 1). These data correlate with the submicromolar/ nanomolar enzymatic inhibition for NDM-1 and show the ability of the series to penetrate the cell membrane, at least in the studied Enterobacteriaceae clinical isolates.

■ CONCLUSIONS
Our biochemical and microbiological results combined with recently published data 34,22 reveal the NSPCs as promising MBL inhibitors with particularly potent (low nM) activity against NDM-1 and IMP-1 and submicromolar activity against VIM-1 and VIM-2 enzymes. The structural studies presented here  ACS Infectious Diseases pubs.acs.org/journal/aidcbc Article define an NSPC binding mode very similar to that of the αcarboxylate-and N-sulfamoyl-substituted furans and pyrroles as MBL inhibitors described by Wachino et al. 21 Thus, the likely deprotonated amino group of the tetrahedral sulfamoyl group bridges the two zinc ions replacing the hydrolytic water in a manner reflecting a tetrahedral intermediate in catalysis. The structural analyses suggest that the relatively low activity against VIM-1/VIM-2 may in part reflect differences in inhibitor (and substrate) carboxylate binding mode. Dynamics at the dizinc center of the protein, which are not observable with cryotemperature crystal structures, may also account for the observed differences in inhibition data between the tested MBL enzymes. Indeed, the overall highly conserved active site architecture of the MBL superfamily enzymes supports a wide range of reactions, including nucleic acid hydrolysis and redox reactions, and in some cases, human MBLs are being pursued as drug targets. 51, 52 The compact and polar nature of the NSPCs and related scaffolds suggests that they may have wide utility as inhibitors of MBL superfamily enzymes. However, further derivatization of the NSPCs is warranted to increase potency and spectrum of activity toward the most abundant resistance causing MBLs to restore utility of important β-lactamase antibacterials.

■ METHODS
The experimental procedures describing the synthesis and characterization of the compounds, the evaluation of their biological activity (enzyme assays, in vitro antibacterial susceptibility testing), and X-ray crystallography studies are fully described in the Supporting Information. General Information. Commercially available reagents and solvents were from Merck or Fluorochem and were used as received. All manipulations with air-and moisture-sensitive compounds were carried out under a positive pressure of argon in flame-dried glassware. Reactions under microwave conditions were carried out in Biotage Initiator EXP microwave reactor with Robot Sixty sample processor.
NMR spectra were acquired using a 600 MHz Bruker Avance III HD machine equipped with a 5 mm DCH cryoprobe and a 400 MHz Bruker Avance II equipped with a 5 mm BBFO probe. Chemical shifts were referenced to residual protio-and perdeuterio-solvent resonances (δ H 7.26 and δ C 77.16 for CDCl 3 ; δ H 2.50 and δ C 39.52 for DMSO-d 6 ) as internal standards for 1 H NMR and 13 C NMR spectra, respectively. 19 F NMR spectra were referenced indirectly via the 2 H signal of the lock substance (CDCl 3 or DMSO-d 6 ) and the Ξ( 19 F) value. All NMR spectra were processed with MestReNova software v. 14.1.
Low resolution mass spectrometry (LRMS) data were obtained using a Waters Acquity H-class UPLC with a Sample Manager FTN and a TUV dual wavelength detector coupled to a QDa single quadrupole analyzer using electrospray ionization (ESI). UPLC separation was achieved with a C18 reversedphase column (Acquity UPLC BEH C18, 2.1 × 50 mm, 1.7 μm) operated at 40°C, using a linear gradient of the binary solvent system of buffer A (H 2 O:MeCN:formic acid, 95:5:0.1 v/v/v%) to buffer B (MeCN:formic acid, 100:0.1 v/v%) from 0 to 100% B in 3.5 min, then 1 min at 100% B, maintaining a flow rate of 0.8 mL/min. High resolution mass spectra were recorded using a Bruker μTOF (ESI) spectrometer. The m/z values are reported in Daltons.
Analytical HPLC was carried out using an Ultimate HPLC system (Thermo Scientific) consisting of a LPG-3400A pump (1 mL/min), a WPS-3000SL autosampler, and a DAD-3000D diode array detector (220 and 254 nm) using a Gemini-NX C18 column (4. Data for both analytical and preparative HPLC were acquired and processed using Chromeleon software v. 6.80. 3-Bromo-1-(phenylsulfonyl)-1H-pyrrole (2). N1-Sulfonylation. 53 Sodium hydride (2.20 g, 55.0 mmol, 60 wt %, 1.1 equiv) was added portionwise to the solution of pyrrole (3.47 mL, 50.0 mmol, 1.0 equiv) in dry DMF (150 mL) at 0°C. The obtained mixture was stirred for 1 h at the same temperature (note for this step: a constant flow of nitrogen gas was used to reduce foaming). Benzenesulfonyl chloride (7.66 mL, 66.0 mmol, 1.2 equiv) was added slowly over 5 min at 0°C; the cooling bath was removed, and the reaction was further stirred for 0.5 h at room temperature (starting pyrrole was consumed, TLC). The reaction mixture was carefully quenched with halfsaturated NH 4 Cl (200 mL) at 0°C and diluted with 200 mL of EtOAc. The organic phase was washed with water (4 × 150 mL), brine (150 mL), dried over Na 2 SO 4 , and concentrated under reduced pressure, providing 10.64 g of crude 1-(phenylsulfonyl)-1H-pyrrole as a beige solid which was taken through to the next step without further purification.
2. The more polar fraction (5.87 g) was crystallized from 30 mL of a 20% EtOAc/Hept mixture. The precipitate (1.08 g) was discarded, and the mother liquor was concentrated to give 4.69 g of S1 as a colorless oil, quickly solidifying upon standing.
3-(4-Fluorophenyl)-1-sulfamoyl-1H-pyrrole-2-carboxylic acid (6a). The N-Cbz sodium salt 5 (72.7 mg, 0.137 mmol) was dissolved in 10 mL of EtOAc, transferred to a separatory funnel, and successively washed with 1 M HCl aq (2 × 5 mL), brine (5 mL), and dried over Na 2 SO 4 . The solvent was removed in vacuo to give S8 as a yellow oil in 96% yield (67.2 mg). 1  The oil obtained above (67.2 mg) was hydrogenated according to General Procedure B; then, the crude residue was further purified by prepHPLC to give the desired compound 6a in 72% yield (27. The Supporting Information is available free of charge at https://pubs.acs.org/doi/10.1021/acsinfecdis.1c00104. Synthesis and characterization of compounds and information on enzymatic and microbiological assays and crystallography (PDF)