Success Stories:
AmpC β-lactamase

Target: Beta-lactamase

AmpC β-lactamase hydrolyzes penicillin and cephalosporin antibiotics, conferring resistance against these drugs to bacteria that express the enzyme. Drugs like clavulanic acid, sulbactam and tazobactam have been long used to reverse β-lactamase-based resistance by inhibiting the enzyme, but these inhibitors are essentially ineffective against class C β-lactamases, such as AmpC. As use of the inhibitors increased, expression of AmpC has spread. Worse still, for many hospital pathogens such as E. cloacae, the inhibitors actually up-regulate the expression of AmpC, leading to overproduction of the enzyme they were meant to inhibit. Thus there is a strong motivation to discover novel inhibitors, unrelated to β-lactams such as penicillins or indeed the inhibitors such as clavulanic acid. In an effort to do so, the Shoichet Lab targeted the structure of AmpC for docking screens.

Site & Ligand Innovations

AmpC is a large and open active site (figure 1) for which only covalent inhibitors like the β-lactams, and pseudo-covalent inhibitors like the boronic acids, were known. On the other hand, many ligand bound structures were available, many of them in fact determined by the student in the lab undertaking the docking screens-Rachel Powers. Rachel used these structures to assign hot spots in the site, which she subsequently exploited in the docking program (what was then NWU_DOCK). The multiple known structures, and their often high quality, allowed her to also determine the positions of several ordered water molecules, which were also used in the docking.
On the ligand side, the major issue was, as always, to calculate parameters for the 250,000 Available Chemical Directory ligands. But this had been largely done for PTP-1B, and we could use what was essentially exactly the same database. As with PTP-1B, about 1000 conformations were calculated for each molecule.


In the docking calculation 250,000 molecules were screened against the AmpC site, resulting in a ranked list of compounds based on electrostatic and non-polar complementarity. Rachel and Federica Morandi picked 56 of the top 500 to test experimentally. Of these, three were competitive, reversible inhibitors, the best of which had a Ki value of 26 µM. Rachel determined the structure of this inhibition by x-ray crystallography, finding that the crystallographic geometry closely matched the docking prediction (figure 2). In subsequent work, Donatella Tondi was able to improve the affinity of this class of inhibitors down to 1 μM, and showed that they were active in cell culture at reversion β-lactamase resistance (Tondi et al & Shoichet, JACS 2005).

Lesson & Caveats

In this AmpC screen, the one decent inhibitor to come out of docking was right for the right reasons: it was a competitive inhibitor whose predicted and subsequent crystallographic structure corresponded closely (figure 2). Also, the chemistry of this compound is novel, being dissimilar to β-lactams or any previously known class of inhibitors. Conversely, the hit rate was low (1/56) and the potency was modest. This highlights both the strengths and the weaknesses of the method. Docking can discover genuinely novel scaffolds for proteins with good structures, which is a strength. On the other hand, because of its many approximations, any prediction for a given compound is unreliable and hit rates will vary. It is important to be able test multiple docking predictions, with a reliable assay, to have some confidence that new inhibitors will be found.


This work was published in Powers RA, Morandi F, Shoichet BK. Structure-based discovery of a novel, non-covalent inhibitor of AmpC β-lactamase. Structure 10 (7), 1013-23 (2002).
[Pubmed | DOI | PDB 1L2S | Download PDF]

© 2000-2015 Blue Dolphin Lead Discovery, LLC. Last updated Feb 3, 2015. Please address comments about this website to bluedolphindiscovery at