Evolution of pharmacologic specificity in the pregnane X receptor

Abstract

Background: The pregnane X receptor (PXR) shows the highest degree of cross-species sequence diversity of any of the vertebrate nuclear hormone receptors. In this study, we determined the pharmacophores for activation of human, mouse, rat, rabbit, chicken, and zebrafish PXRs, using a common set of sixteen ligands. In addition, we compared in detail the selectivity of human and zebrafish PXRs for steroidal compounds and xenobiotics. The ligand activation properties of the Western clawed frog (Xenopus tropicalis) PXR and that of a putative vitamin D receptor (VDR)/PXR cloned in this study from the chordate invertebrate sea squirt (Ciona intestinalis) were also investigated.

Results: Using a common set of ligands, human, mouse, and rat PXRs share structurally similar pharmacophores consisting of hydrophobic features and widely spaced excluded volumes indicative of large binding pockets. Zebrafish PXR has the most sterically constrained pharmacophore of the PXRs analyzed, suggesting a smaller ligand-binding pocket than the other PXRs. Chicken PXR possesses a symmetrical pharmacophore with four hydrophobes, a hydrogen bond acceptor, as well as excluded volumes. Comparison of human and zebrafish PXRs for a wide range of possible activators revealed that zebrafish PXR is activated by a subset of human PXR agonists. The Ciona VDR/PXR showed low sequence identity to vertebrate VDRs and PXRs in the ligand-binding domain and was preferentially activated by planar xenobiotics including 6-formylindolo-[3,2-b]carbazole. Lastly, the Western clawed frog (Xenopus tropicalis) PXR was insensitive to vitamins and steroidal compounds and was activated only by benzoates.

Conclusion: In contrast to other nuclear hormone receptors, PXRs show significant differences in ligand specificity across species. By pharmacophore analysis, certain PXRs share similar features such as human, mouse, and rat PXRs, suggesting overlap of function and perhaps common evolutionary forces. The Western clawed frog PXR, like that described for African clawed frog PXRs, has diverged considerably in ligand selectivity from fish, bird, and mammalian PXRs.

Figure 1

Chemical structures of PXR activators. Chemical structures of the PXR activators 5β-pregnane-3,20-dione, 5α-androstan-3α-ol, 5β-lithocholic acid, 5α-cyprinol 27-sulfate, 3-aminoethylbenzoate, and 6-formylindolo-[3,2-b]-carbozole. The key bond positions are numbered for the steroids and bile salts, and the lettering of the steroidal rings is indicated for pregnanedione and lithocholic acid. The structure to the right of lithocholic acid illustrates the most stable orientation of the A, B, and C steroid rings for 5β-bile salts (like lithocholic acid) with the A/B cis configuration (referring to the relative orientation of the hydrogen atom substituents on carbon atoms 5 and 10). The structure to the right of 5α-cyprinol sulfate shows the most stable orientation of 5α-bile salts (like 5α-cyprinol sulfate) that prefentially adopt the A/B trans configuration.

Figure 2

PXR activation and steroid pathways. Steroid pathways typical of vertebrates are indicated. (A) Human PXR is activated by a large number of steroid hormones, although typically at micromolar concentrations. The coloring indicates at which concentrations the various steroids activate human PXR (see key in bottom right of panel). (B) Zebrafish PXR is activated by a smaller number of steroid hormones than human PXR, although there is much overlap between the selectivity of the two PXRs. Zebrafish PXR tends to be more sensitive to steroid hormone activation, at least for the functional assay used in this study. The coloring indicates at which concentrations the various steroids activate zebrafish PXR using the same key as in (A). Abbreviations: dehydroepiandrosterone, DHEA; DHEA sulfate; DHEA SO₄; dihydrotesterone, DHT.

Figure 3

Pharmacophore models of PXR activators. Pharmacophore models of PXR activators of (A) human PXR, (B) zebrafish PXR, (C) mouse PXR, (D) rat PXR, (E) rabbit PXR, and (F) chicken PXR. The pharmacophores were generated from the same 16 molecules using Catalyst. The molecules mapped to each pharmacophore are TCDD (green) and 5β-pregnane-3,20-dione (grey). It should be noted that TCDD is inactive in rabbit PXR and only maps to the hydrophobic features. The pharmacophore features are hydrophobic (cyan), hydrogen bond acceptor and vector (green), and excluded volume (grey).

Figure 4

Maximum likelihood phylogeny of VDRs, PXRs, and CARs. Maximum likelihood phylogeny of 49 amino acid sequences of VDRs, PXRs, and CARs (see Methods for details of analysis). Numbered branch labels indicate bootstrap percentages. Node labels 'AncR1', 'AncR2', and 'AncR3' indicate ancestral nodes that were reconstructed (see Additional files 7 and 8).

Figure 5

Conservation of ligand-binding residues. From published X-ray crystallographic structures of human VDR, rat VDR, zebrafish VDR, human PXR, human CAR, and mouse CAR (see Methods for references), amino acid residues that interact with ligands ('ligand-binding residues') were identified. At these amino acid residue positions, the sequences of Ciona intestinalis VDR/PXR, AncR1, AncR2, and AncR3 were compared with the corresponding sequence for human PXR, mouse PXR, rat PXR, rabbit PXR, chicken PXR, Xenopus laevis PXRα, Xenopus laevis PXRβ, zebrafish PXR, human CAR, human VDR, and sea lamprey VDR. The ordinate represents the percent identity of Ciona intestinalis VDR/PXR, AncR1, AncR2, and AncR3 for the corresponding sequences of PXRs, VDRs, or CAR at these ligand-binding residue positions.