Pharmacophore selection and redesign of non-nucleotide inhibitors of anthrax edema factor

Abstract

Antibiotic treatment may fail to protect individuals, if not started early enough, after infection with Bacillus anthracis, due to the continuing activity of toxins that the bacterium produces. Stable and easily stored inhibitors of the edema factor toxin (EF), an adenylyl cyclase, could save lives in the event of an outbreak, due to natural causes or a bioweapon attack. The toxin's basic activity is to convert ATP to cAMP, and it is thus in principle a simple phosphatase, which means that many mammalian enzymes, including intracellular adenylcyclases, may have a similar activity. While nucleotide based inhibitors, similar to its natural substrate, ATP, were identified early, these compounds had low activity and specificity for EF. We used a combined structural and computational approach to choose small organic molecules in large, web-based compound libraries that would, based on docking scores, bind to residues within the substrate binding pocket of EF. A family of fluorenone-based inhibitors was identified that inhibited the release of cAMP from cells treated with EF. The lead inhibitor was also shown to inhibit the diarrhea caused by enterotoxigenic E. coli (ETEC) in a murine model, perhaps by serving as a quorum sensor. These inhibitors are now being tested for their ability to inhibit Anthrax infection in animal models and may have use against other pathogens that produce toxins similar to EF, such as Bordetella pertussis or Vibrio cholera.

Figure 1

(Top) The overall structure of anthrax EF (plus calmodulin [17]) indicating the small area targeted by the inhibitors in this study; (Bottom) detail of the adenylyl cyclase domain of 1K90.pdb, with the Yb ion (green), and the inhibitor included in the co-crystal structure (3'dATP, colored according to atom type) shown as space filling. The magenta lines indicate residues of EF that surround the active (substrate binding) site.

Figure 2

Design of a fragment based pharmacophore using the HINT (Hydropathic INTeractions) program, the lowest energy binding sites of a benzene ring, and two carboxyls and the distances between the three fragments are the basis of a 3D-pharmacophore, suitable for compound library screening with the Unity program. Note that HINT was used again to determine the optimal binding site of larger fragments, as described in Figure 4.

Figure 3

Overview of the fragment based pharmacophore design. (A) Overlay of the initial 3D-pharmacophore designed based on the HINT selected fragments (Figure 2; F1: phenyl ring; F2, F3 carboxyl groups, with distance constraints a, b, c) on a 2D image of the ligand binding site (for 3'dATP) of 1K90 (Poseview [21])); (B) Shows the overlay of the pharmacophore with docking poses (to the 1K90 structure, with the substrate removed) for two of the active compounds identified in the first bioscreening (3-[(9-oxo-9H-fluorene-1-carbonyl)-amino]-benzoic acid, top and 4-[(anthracen-9-ylmethylene)-amino]-2-hydroxy-benzoic acid, bottom). These dockings indicate that the orientation was consistent with the fragment positions in the active site. Compound 1(top), a substituted fluorenone, was our lead for subsequent redesign.

Figure 4

Overall scheme for selecting inhibitors of EF using compound library screening and docking with Flex-X and Autodock. Note the HINT program was used at two levels. In the first, simple fragments were used to design a 3-D pharmacophore, and select a small group of compounds with Flex-X. These compounds were then analyzed to find larger common fragments that were again used to refine the pharmacophore. Several versions of these pharmacophores were then used to screen ZINC [22] to obtain 10,000 compounds that were ranked by docking (to several different crystal structures from the PDB) with Autodock3. Purchased compounds were tested as described in the next section.

Figure 5

Assaying compounds for inhibition of EF induced extracellular cAMP secretion, used RAW 264.7 cells treated with edema toxin (30 nM PA and 70 nM EF). Cells were incubated with the indicated concentrations of our lead inhibitor (Figure 3B, top) and a derivative, LM-1-224 (2-Hydroxy-5-(9-oxo-9H-fluorene-1-carboxamido)benzoic acid). The first bar in each series shows that negligible cAMP is induced when cells are treated with the compounds alone, without EF addition. Cells were incubated for 4 h before assay. cAMP was measured with the Assay designs, Inc direct cAMP kit.