BXR, an embryonic orphan nuclear receptor activated by a novel class of endogenous benzoate metabolites

Abstract

Nuclear receptors are ligand-modulated transcription factors that respond to steroids, retinoids, and thyroid hormones to control development and body physiology. Orphan nuclear receptors, which lack identified ligands, provide a unique, and largely untapped, resource to discover new principles of physiologic homeostasis. We describe the isolation and characterization of the vertebrate orphan receptor, BXR, which heterodimerizes with RXR and binds high-affinity DNA sites composed of a variant thyroid hormone response element. A bioactivity-guided screen of embryonic extracts revealed that BXR is activatable by low-molecular-weight molecules with spectral patterns distinct from known nuclear receptor ligands. Mass spectrometry and 1H NMR analysis identified alkyl esters of amino and hydroxy benzoic acids as potent, stereoselective activators. In vitro cofactor association studies, along with competable binding of radiolabeled compounds, establish these molecules as bona fide ligands. Benzoates comprise a new molecular class of nuclear receptor ligand and their activity suggests that BXR may control a previously unsuspected vertebrate signaling pathway.

Figure 1

BXR is a novel member of the steroid receptor superfamily. (a) Sequence of the longest BXR cDNA clone. The DNA and ligand-binding domains are boxed. In-frame termination codons are indicated by asterisks. The sequence probably represents an incomplete cDNA, as the canonical AAUAAA polyadenylation signal is not present upstream of the terminal A residues. (b) Schematic comparison between BXR, other known RXR heterodimeric partners, and the human glucocorticoid receptor. Amino acid sequences were aligned using the program GAP (Devereaux et al. 1984). The similarity between the BXR and other receptors is expressed as percent amino-acid identity. (c) Analysis of BXR mRNA expression. The RNase protection probes used are the following: EF-1a (nucleotides 790–1167); BXR (nucleotides 1314–1560). RNase protection was performed with total RNA from the indicated stages and are abbreviated as follows: (O) total ovary (10 μg); (E) unfertilized egg (40 μg); (2) two-cell (40 μg); (7 and 8) blastula (40 μg); (10 and 11) gastrula (stage 10, 10 μg; stage 11, 8 μg); (13–22) neurula (4 μg); (26 and 33) tailbud (4 μg); (40 and 45) swimming tadpole (4 μg). RNase protection was performed with 20 μg of total RNA from whole embryos or dissected animal caps, marginal zone, and vegetal pole.

Figure 1

BXR is a novel member of the steroid receptor superfamily. (a) Sequence of the longest BXR cDNA clone. The DNA and ligand-binding domains are boxed. In-frame termination codons are indicated by asterisks. The sequence probably represents an incomplete cDNA, as the canonical AAUAAA polyadenylation signal is not present upstream of the terminal A residues. (b) Schematic comparison between BXR, other known RXR heterodimeric partners, and the human glucocorticoid receptor. Amino acid sequences were aligned using the program GAP (Devereaux et al. 1984). The similarity between the BXR and other receptors is expressed as percent amino-acid identity. (c) Analysis of BXR mRNA expression. The RNase protection probes used are the following: EF-1a (nucleotides 790–1167); BXR (nucleotides 1314–1560). RNase protection was performed with total RNA from the indicated stages and are abbreviated as follows: (O) total ovary (10 μg); (E) unfertilized egg (40 μg); (2) two-cell (40 μg); (7 and 8) blastula (40 μg); (10 and 11) gastrula (stage 10, 10 μg; stage 11, 8 μg); (13–22) neurula (4 μg); (26 and 33) tailbud (4 μg); (40 and 45) swimming tadpole (4 μg). RNase protection was performed with 20 μg of total RNA from whole embryos or dissected animal caps, marginal zone, and vegetal pole.

Figure 1

BXR is a novel member of the steroid receptor superfamily. (a) Sequence of the longest BXR cDNA clone. The DNA and ligand-binding domains are boxed. In-frame termination codons are indicated by asterisks. The sequence probably represents an incomplete cDNA, as the canonical AAUAAA polyadenylation signal is not present upstream of the terminal A residues. (b) Schematic comparison between BXR, other known RXR heterodimeric partners, and the human glucocorticoid receptor. Amino acid sequences were aligned using the program GAP (Devereaux et al. 1984). The similarity between the BXR and other receptors is expressed as percent amino-acid identity. (c) Analysis of BXR mRNA expression. The RNase protection probes used are the following: EF-1a (nucleotides 790–1167); BXR (nucleotides 1314–1560). RNase protection was performed with total RNA from the indicated stages and are abbreviated as follows: (O) total ovary (10 μg); (E) unfertilized egg (40 μg); (2) two-cell (40 μg); (7 and 8) blastula (40 μg); (10 and 11) gastrula (stage 10, 10 μg; stage 11, 8 μg); (13–22) neurula (4 μg); (26 and 33) tailbud (4 μg); (40 and 45) swimming tadpole (4 μg). RNase protection was performed with 20 μg of total RNA from whole embryos or dissected animal caps, marginal zone, and vegetal pole.

Figure 2

BXR heterodimerizes with RXR to enable DNA binding. Gel mobility shift analyses of BXR DNA-binding specificity. (a) In vitro-transcribed and -translated proteins were mixed with a cocktail of hormone response elements containing DR0, DR1, PPRE, DR2, MLV–TRE, SPP1, βRARE, GRE, and ERE. (b) BXR and xRXRα proteins were mixed and incubated with the indicated response elements. βRE-1 through βRE-5 are direct repeats of the sequence AGTTCA, separated by 1–5 nucleotides. Reaction conditions and gel electrophoresis were as described (Perlmann et al. 1993). We note also that xRXRα alone binds strongly to the DR-1 motif in this assay and migrates in a position appropriate for xRXRα homodimers. (c) BXR and hRXRα interact in vivo. The indicated plasmids were cotransfected into CV-1 cells along with the reporter tk(gal_p)₃–luc and CMX–βgal. Note the strong suppression of basal transcription when GAL–BXR was added (right). This is characteristic of many known ligand-dependent RXR heterodimeric partners.

Figure 2

BXR heterodimerizes with RXR to enable DNA binding. Gel mobility shift analyses of BXR DNA-binding specificity. (a) In vitro-transcribed and -translated proteins were mixed with a cocktail of hormone response elements containing DR0, DR1, PPRE, DR2, MLV–TRE, SPP1, βRARE, GRE, and ERE. (b) BXR and xRXRα proteins were mixed and incubated with the indicated response elements. βRE-1 through βRE-5 are direct repeats of the sequence AGTTCA, separated by 1–5 nucleotides. Reaction conditions and gel electrophoresis were as described (Perlmann et al. 1993). We note also that xRXRα alone binds strongly to the DR-1 motif in this assay and migrates in a position appropriate for xRXRα homodimers. (c) BXR and hRXRα interact in vivo. The indicated plasmids were cotransfected into CV-1 cells along with the reporter tk(gal_p)₃–luc and CMX–βgal. Note the strong suppression of basal transcription when GAL–BXR was added (right). This is characteristic of many known ligand-dependent RXR heterodimeric partners.

Figure 3

Identification of a BXR agonist. (a) Activation of GAL–BXR (□) and full-length BXR (▪) by fractions from the final purification of BXR agonist. CV-1 cells were transfected with 1 μg of receptor expression plasmid, 5 μg of reporter, and 4 μg of CMX–βgal control plasmid per 5 × 10⁵ cells. 00 254 nm. HPLC fractions were dried, resuspended in 25 μl of methanol, and added in triplicate to cells. (b) Electron impact mass spectrum of the purified BXR agonist. (c) Electron impact mass spectrum of 3-amino-ethyl benzoate. (d) Activation analysis of 10⁻⁴ m 3-AEB (▪) on a variety of GAL4–DNA-binding domain/receptor ligand (□) binding domain chimeras. (e, bottom panel) Chomatogram of polar benzoates from mixed embryonic extracts. Apparent benzoate-containing peaks are indicated as a–d; (top panel) UV spectra of endogenous benzoates in active HPLC fractions from embryonic extracts. (f) Structure of ethyl-3-hydroxyl benzoate glucosamine derived from peak a.

Figure 4

BXR is a benzoate receptor. (a, left panel) Saturable binding of ³H-labeled-4-ABB to purified BXR:hRXRα heterodimers. Purified heterodimeric BXR (2.5 μg) was incubated with the indicated amount of ³H-labeled 4-ABB. Nonspecific binding was determined by competition with excess unlabeled 4-ABB (10⁻⁴ m) and subtracted from total binding. (Right panel) Scatchard transformation of the data in the left panel. K_d was determined by nonlinear regression analysis using GraphPad Prism software. (b) Direct binding of radio-labeled 4-ABB to purified BXR:hRXRα heterodimers. (4-HBB) 4-hydroxyl-butyl benzoate; (2NEB) 2-nitro ethyl benzoate. (c) 4-ABB increases the affinity of BXR:hRXRα heterodimers for the nuclear receptor coactivator SRC-1 in vitro. The total radioactivity in each band was determined by phosphorimager analysis and is reported below the image as fold increase over solvent control. (d) 4-ABB inhibits in vitro interactions between BXR:hRXRα heterodimers and the nuclear receptor corepressor SMRT. Phosphorimager quantitation is shown below the image as fold decrease over solvent control.

Figure 4

BXR is a benzoate receptor. (a, left panel) Saturable binding of ³H-labeled-4-ABB to purified BXR:hRXRα heterodimers. Purified heterodimeric BXR (2.5 μg) was incubated with the indicated amount of ³H-labeled 4-ABB. Nonspecific binding was determined by competition with excess unlabeled 4-ABB (10⁻⁴ m) and subtracted from total binding. (Right panel) Scatchard transformation of the data in the left panel. K_d was determined by nonlinear regression analysis using GraphPad Prism software. (b) Direct binding of radio-labeled 4-ABB to purified BXR:hRXRα heterodimers. (4-HBB) 4-hydroxyl-butyl benzoate; (2NEB) 2-nitro ethyl benzoate. (c) 4-ABB increases the affinity of BXR:hRXRα heterodimers for the nuclear receptor coactivator SRC-1 in vitro. The total radioactivity in each band was determined by phosphorimager analysis and is reported below the image as fold increase over solvent control. (d) 4-ABB inhibits in vitro interactions between BXR:hRXRα heterodimers and the nuclear receptor corepressor SMRT. Phosphorimager quantitation is shown below the image as fold decrease over solvent control.

Figure 4

BXR is a benzoate receptor. (a, left panel) Saturable binding of ³H-labeled-4-ABB to purified BXR:hRXRα heterodimers. Purified heterodimeric BXR (2.5 μg) was incubated with the indicated amount of ³H-labeled 4-ABB. Nonspecific binding was determined by competition with excess unlabeled 4-ABB (10⁻⁴ m) and subtracted from total binding. (Right panel) Scatchard transformation of the data in the left panel. K_d was determined by nonlinear regression analysis using GraphPad Prism software. (b) Direct binding of radio-labeled 4-ABB to purified BXR:hRXRα heterodimers. (4-HBB) 4-hydroxyl-butyl benzoate; (2NEB) 2-nitro ethyl benzoate. (c) 4-ABB increases the affinity of BXR:hRXRα heterodimers for the nuclear receptor coactivator SRC-1 in vitro. The total radioactivity in each band was determined by phosphorimager analysis and is reported below the image as fold increase over solvent control. (d) 4-ABB inhibits in vitro interactions between BXR:hRXRα heterodimers and the nuclear receptor corepressor SMRT. Phosphorimager quantitation is shown below the image as fold decrease over solvent control.

Figure 5

Stereochemical requirements for BXR activation. All compounds were tested as dilution series (10⁻³–10⁻⁹ m) for their ability to activate full-length BXR or GAL–BXR using the cotransfection assay as described above. (a) A number of classes of compounds were tested. For simplicity, one representative example from each group is shown. (1) 3-AEB; (2) benzoate esters ranging from methyl to butyl; (3) 2-,3-, or 4-amino benzoic acids; (4) 3- and 4-amino salicylic acids; (5) o-, m-, and p-phenetedines; (6) 2-,3-, and 4-ethyl anilines; (7) 2-,3-, and 4-amino benzyl alcohols; (8) 2-,3-, and 4-acetamidobenzoic acids; (9) salicylamide. Results with 10⁻⁴ m of each compound are shown. (b) The effects of substituting the amino group with NO₂, OH, or H were tested. (c) The transcriptional effects of increasing the alkyl chain length of the esterified moiety were evaluated using full-length BXR and GAL–BXR in the context of amino benzoates. Results are shown for GAL–BXR.