Identification of a Novel Class of PPAR Ligands—The parent 2-BABA scaffold (BVT.142) was identified as a PPARγ ligand and activator as described previously (25). An initial expansion around this scaffold yielded 250 compounds, where the agonists BVT.13, BVT.762, and BVT.763 were substituted in the 5-position with 2-pyrimidinyloxy, 2-thienylmethoxy, and 2-(3-thienyl)ethoxy, respectively (Fig. 1a). These compounds were chosen as good representatives based on both chemical and preclinical properties.
Structural Analysis of Cocrystal Complexes—Initial docking studies of the 2-BABA scaffold to the apoPPARγ-LBD structure (26) in combination with the bioisosteric properties of the TZD headgroup and carboxylate group of the 2-BABA compounds suggested an anchoring mode similar to that of eicosapentaenoic acid to PPARδ (5), with the carboxylic headgroup interacting with Tyr-473 in helix 12. To further elucidate this hypothesis, the crystal structures of BVT.13, BVT.762, and BVT.763 in complex with PPARγ-LBD and a GRIP-1-derived coactivator peptide were solved by molecular replacement. Resolutions of the complexes were 2.8, 3.15, and 2.9 Å, respectively (Table I). The overall fold of the PPARγ-LBD shows only minor differences as compared with the previously determined apoPPARγ structure (26). The major differences are in the region connecting helix 2' and helix 3, a region with high mobility. The 2-BABA complexes superimpose well with the apoPPARγ-LBD structure, resulting in a root mean square deviation of 0.8 Å over 260 C-α atoms (Protein Data Bank accession code 3PRG [PDB] ). Surprisingly, the crystal structures reveal that the three agonists lack a direct interaction with helix 12. The 2-BABA ligands are located in an elongated conformation of the ligand-binding pocket (Fig. 1, b and c) and they occupy a region in proximity with and approximately parallel to helix 3 and in proximity to β-strand 3 and helices 5 and 2'.
The interactions of the molecule with the protein can be divided into four groups. The first group is formed by the interactions to the dichloro-phenyl moiety. This moiety is located in a predominantly hydrophobic pocket of the receptor. Principal interactions are hydrophobic in nature and formed by the side chains of Arg-288, Ala-292, Ile-326, Met-329, Leu-330, and Leu-333 (Fig. 1d). The second group of interactions is formed between the carbonyl oxygen on the peptide linker of the ligand and the side chain sulfur and the backbone carbonyl oxygen of Cys-285. The third group is made up of interactions between the central benzoic acid moiety and the protein. This moiety is located in a narrow groove of the protein formed by the side chains of Cys-285, Arg-288, Ile-341, and Met-364, and the backbone atoms of Gly-284, Cys-285, and Ser-342. The interactions are hydrophobic in nature with the exception of a distinct hydrogen bond formed between the backbone nitrogen of Ser-342 and one of the carboxylate oxygens of the ligand (Fig. 1d). The fourth group of interactions is formed between the heterocyclic ring and the protein. These are also predominantly hydrophobic in nature, where the protein atoms involved belong to the side chains and backbone of Leu-255, Arg-280, and Ile-281.
The pyrimidine ring of BVT.13 has additional interactions with Phe-264 and His-266, residues that are situated in the part of the loop connecting helix 2' and helix 3, which is structured in the BVT.13 complex but not in the other two complexes. The GRIP-1 coactivator peptide is bound in a helical conformation in the predominantly hydrophobic groove formed by helices 3-5 and 12 (AF2), a binding mode identical to that seen for the steroid receptor coactivator 1 helix in the PPARγ-rosiglitazone complex (21). As described previously, Lys-301 and Glu-471 of the LBD stabilize the dipole moment of the coactivator helix by forming a charge clamp (21).
Ligand-induced Transcriptional Activation—To evaluate the ability of BVT.13, BVT.762, and BVT.763 to transcriptionally activate PPARγ and PPAR, a cell-based reporter gene assay was employed. CaCo-2/TC7 cells were transiently transfected with chimeric Gal4-DBD/PPAR-LBD receptors and a 4xGAL4-RE reporter plasmid and subsequently treated with BVT.13, BVT.762, BVT.763 or rosiglitazone in optimized serial dilutions. A strong ligand-dependent activation of PPARγ, with a relative efficacy of 80% compared with rosiglitazone, was seen for all three compounds (Fig. 2a). The calculated EC50 values were 1.3, 0.3, and 0.8 µM, respectively (25). BVT.762 and BVT.763 induced the transcriptional activity of PPARα to a similar efficacy as seen for PPARγ but displayed EC50 values of 5 and 3.8 µM, respectively (25). Interestingly, BVT.13 failed to bind and activate PPARα (25). Furthermore, the 2-BABA compounds were not able to antagonize the activity of 30 nM rosiglitazone in the cell-based reporter gene assay (Fig. 2b). To further determine the specificity of BVT.13, a set of NR LBDs were analyzed in cell-based reporter gene assays. Transcriptional activation by BVT.13 was shown to be selective for the mouse and human PPARγ isoforms (Fig. 3).
Pharmacological Activity of BVT.13—Next, the antidiabetic and antilipidemic effects of BVT.13 in mice were examined. Male ob/ob mice were treated once daily with BVT.13 doses of 30, 100, or 300 mg/kg for 7 days. Plasma concentrations of BVT.13 at the dose of 100 mg/kg were above the lowest limit of quantification for the analytical method at all time points, implicating that the mice were exposed to BVT.13 during the whole 24-h dosing interval. The average steady-state concentration at 100 mg/kg/day was determined in a kinetic study. Using an unbound fraction of 0.6%, as determined for BVT.13 in vitro, the calculations yielded an approximate steady-state unbound concentration of 1 µM. The body weight of both the BVT.13- and rosiglitazone-treated animals increased significantly compared with the vehicle-treated animals, although no difference in food intake was observed (data not shown). After 7 days, levels of plasma glucose, triglycerides, insulin, and free fatty acids were measured. Significant reductions of fasting plasma glucose and triglycerides were seen after treatment with BVT.13 at the dose of 300 mg/kg compared with vehicle treatment. The reductions were comparable with those seen after treatment with rosiglitazone at 5 mg/kg (Fig. 4, a and b). The lower doses of BVT.13 did not significantly reduce the plasma glucose and triglyceride levels (Fig. 4, a and b). The fasting plasma insulin and free fatty acids were significantly reduced after treatment with BVT.13 at the two higher doses (100 and 300 mg/kg) compared with vehicle treatment. The two higher doses of BVT.13 resulted in plasma insulin and free fatty acid levels comparable with those seen after treatment with rosiglitazone (Fig. 4, c and d).