The expression of beta-lactamases is the most common form of bacterial resistance to beta-lactam antibiotics. To combat these enzymes, agents that inhibit (e.g. clavulanic acid) or evade (e.g. aztreonam) beta-lactamases have been developed. Both the beta-lactamase inhibitors and the beta-lactamase-resistant antibiotics are themselves beta-lactams, and bacteria have responded to these compounds by expressing variant enzymes resistant to inhibition (e.g. IRT-3) or that inactivate the beta-lactamase-resistant antibiotic (e.g. TEM-10). Moreover, these compounds have increased the frequency of bacteria with intrinsically resistant beta-lactamases (e.g. AmpC). In an effort to identify non-beta-lactam-based beta-lactamase inhibitors, we used the crystallographic structure of the m-aminophenylboronic acid-Escherichia coli AmpC beta-lactamase complex to suggest modifications that might enhance the affinity of boronic acid-based inhibitors for class C beta-lactamases. Several types of compounds were modeled into the AmpC binding site, and a total of 37 boronic acids were ultimately tested for beta-lactamase inhibition. The most potent of these compounds, benzo[b]thiophene-2-boronic acid (36), has an affinity for E. coli AmpC of 27 nM. The wide range of functionality represented by these compounds allows for the steric and chemical "mapping" of the AmpC active site in the region of the catalytic Ser64 residue, which may be useful in subsequent inhibitor discovery efforts. Also, the new boronic acid-based inhibitors were found to potentiate the activity of beta-lactam antibiotics, such as amoxicillin and ceftazidime, against bacteria expressing class C beta-lactamases. This suggests that boronic acid-based compounds may serve as leads for the development of therapeutic agents for the treatment of beta-lactam-resistant infections.

The structures of AmpC beta-lactamase from Escherichia coli, alone and in complex with a transition-state analogue, have been determined by X-ray crystallography. The native enzyme was determined to 2.0 A resolution, and the structure with the transition-state analogue m-aminophenylboronic acid was determined to 2.3 A resolution. The structure of AmpC from E. coli resembles those previously determined for the class C enzymes from Enterobacter cloacae and Citrobacter freundii. The transition-state analogue, m-aminophenylboronic acid, makes several interactions with AmpC that were unexpected. Perhaps most surprisingly, the putative "oxyanion" of the boronic acid forms what appears to be a hydrogen bond with the backbone carbonyl oxygen of Ala318, suggesting that this atom is protonated. Although this interaction has not previously been discussed, a carbonyl oxygen contact with the putative oxyanion or ligand carbonyl oxygen appears in most complexes involving a beta-lactam recognizing enzyme. These observations may suggest that the high-energy intermediate for amide hydrolysis by beta-lactamases and related enzymes involves a hydroxyl and not an oxyanion, although the oxyanion form certainly cannot be discounted. The involvement of the main-chain carbonyl in ligand and transition-state recognition is a distinguishing feature between serine beta-lactamases and serine proteases, to which they are often compared. AmpC may use the interaction between the carbonyl of Ala318 and the carbonyl of the acylated enzyme to destabilize the ground-state intermediate, this destabilization energy might be relieved in the transition state by a hydroxyl hydrogen bond. The structure of the m-aminophenylboronic acid adduct also suggests several ways to improve the affinity of this class of inhibitor and points to the existence of several unusual binding-site-like features in the region of the AmpC catalytic site.