Convergence of 9-cis retinoic acid and peroxisome proliferator signalling pathways through heterodimer formation of their receptors

Abstract

Peroxisomes are cytoplasmic organelles which are important in mammals in modulation of lipid homeostasis, including the metabolism of long-chain fatty acids and conversion of cholesterol to bile salts (reviewed in refs ¹ and ²). Amphipathic carboxylates such as clofibric acid have been used in man as hypolipidaemic agents and in rodents they stimulate the proliferation of peroxisomes. These agents, termed peroxisome proliferators, and all- trans retinoic acid activate genes involved in peroxisomal-mediated β-oxidation of fatty acids^1–4. Here we show that the receptor activated by peroxisome proliferators⁵ and the retinoid X-receptorα (ref. ⁶) form a heterodimer that activates acyl-CoA oxidase gene expression in response to either clofibric acid or the retinoid × receptor-a ligand, 9-cis retinoic acid, an all-frans retinoic acid metabolite^7,8; simultaneous exposure to both activators results in a synergistic induction of gene expression. These data demonstrate the coupling of the peroxisome proliferator and retinoid signalling pathways and provide evidence for a physiological role for 9-cis retinoic acid in modulating lipid metabolism.

The peroxisome proliferator responsive element (PPRE) located in the rat acyl-CoA oxidase (AOX) promoter is composed of two direct AGG(A/T)CA repeats separated by a single nucleotide (Fig. 1a)^9–11 and thus conforms with previously described retinoid × response elements¹²‘¹³. Indeed, cotransfection of the retinoid × receptor-a (RXRa) expression plasmid in the presence of 9-cis retinoic acid (Fig. 1b, 9-cis RA, RXR) resulted in activation of a reporter construct containing the AOX promoter upstream of the luciferase gene (AOX-LUC). As previously shown⁹, the AOX promoter was also induced by the peroxisome proliferator-activated receptor (PPAR) in the presence of clofibric acid (Fig. 1b, clofibric acid, PPAR). Interestingly, cotransfection of expression plasmids for both PPAR and RXRa in the presence of clofibric acid and 9-cis retinoic acid resulted in a synergistic increase in the activity of the AOX promoter (Fig. 1b, clofibric acid + 9-cis RA, PPAR+RXR).

FIG. 1.

PPAR and RXRα transactivate reporter expression cooperatively through the PPRE. a, The sequence of the PPRE located between nucleotides −558 and −570 of the AOX promoter is compared to that of the idealized DR-1 HRE. The positions of the AGGTCA direct repeats are indicated by arrows. X, nucleotide substitution in this motif, b, CV-1 cells were cotransfected with reporter construct AOX-LUC and either the control (CONT) Rous sarcoma-virus promotor-chloramphenicol acetyl transferase (RS-CAT) or expression plasmids RS-rat(r)PPAR (PPAR) and/or RS-human(h)RXRα (RXR) as indicated. Cells were subsequently treated with carrier dimethylsulphoxide (DMSO) (−) or hormones (+), including clofibric acid, 9-cis retinoic acid (RA), or clofibric acid and 9-cis RA as indicated. Luciferase activity Is presented as per cent normalized response where induced PPAR activity in the presence of clofibric acid is arbitrarily set at 100%. c, CV-1 cells were cotransfected with reporter construct PPRE₃-TK-LUC and either the control RS-CAT or expression plasmids RS-rPPAR and/or RS-hRXRα as Indicated.

METHODS. Expression plasmid RS-rPPAR was constructed by inserting the rat PPAR cDNA (D.J.N., submitted) into the Kpnl/BamHl sites of pRS (ref. ⁶). The RS-human RXRα expression plasmid has been described¹². The reporter plasmid AOX-LUC was constructed by amplification from rat genomic DNA of the 5’ flanking sequences (nucleotides −602 to +20) of the rat AOX promoter by 30 cycles of the polymerase chain reaction (PCR) and directional insertion into BamHI/Xhol-cut. pLUC (ref. ¹²). Oligonucleotides for PCR contained BamHI (5’) and XhoI (3’) restriction sites to facilitate subcloning. The reporter construct PPRE₃-TK-LUC was constructed by inserting 3 copies of an oligonucleotide encoding the PPRE (GTCGACAGGGGA- CCAGGACAAAGGTCACGTTCGGGAGTCGAC)^9–11 in direct orientation into the unique SaII site of the basal reporter construct TK-LUC (ref. ¹²). The orientation of the PPREs was confirmed by sequencing. Transfection assays were performed using CV-1 cells as described^12,18 and modified for automation In 96-well plates⁷. All transfections were performed on a Beckman Biomek Automated Workstation. Transfections contained 5.0 ng of receptor expression plasmid vector (or control RS-CAT), 50 ng of the reporter luciferase plasmid, 50 ng of pRS/3GAL (β-galactosidase) as an internal control, and 90 ng of carrier plasmid pGEM. Cells were transfected for 6 h, washed to remove DNA precipitates, and treated with hormones (1 × 10⁻³ M clofibric acid; 1 × 10⁻⁶ M 9-cis RA) for 36 h. Cell extracts were subsequently prepared and assayed for luciferase and β-galactosidase activities⁷. All experiments were done In triplicate in at least two independent experiments and were normalized for transfection efficiency by using β-galactosidase as the internal control.

To investigate whether the cooperativity between PPAR and RXRα occurred through the PPRE, cotransfection experiments were performed with a reporter construct containing three copies of the PPRE upstream of the thymidine kinase promoter fused to the luciferase gene (PPRE₃-TK-LUC). As expected, cotransfection of either PPAR or RXRα expression plasmids in the presence of clofibric acid or 9-cis retinoic acid, respectively, resulted in induction of reporter expression (Fig. 1c, clofibric acid or 9-cis RA). As in the case of the intact AOX promoter, coexpression of PPAR and RXRα in the presence of both activators resulted in a synergistic increase in PPRE₃-TK- LUC expression (Fig. 1c, clofibric acid + 9-cis RA). Thus, PPAR and RXRα cooperate in transactivating gene expression through the PPRE.

The synergism between PPAR and RXRα on the PPRE suggested the possibility of a heterodimeric interaction between the two receptors. To test this possibility, coimmunoprecipitation experiments were performed using bacterially expressed RXRα and in vitro-synthesized, [³⁵S]methionine-labelled PPAR. As shown in Fig. 2a, when mixed, polyclonal antiserum against RXRα efficiently coprecipitated PPAR, indicating the formation of a solution heterodimer. Using the PPRE as probe, gel mobility shift experiments were performed to examine the binding properties of the PPAR-RXRα complex. Neither PPAR nor RXRα alone bound efficiently to the radiolabelled PPRE (Fig. 2b, lanes 2 and 3). Mixing the PPAR and RXRα proteins, however, resulted in the appearance of a strong, shifted complex (Fig. 2b, lane 4) which was further shifted by inclusion of the polyclonal antiserum (Fig. 2b, lane 5). Thus, PPAR and RXRα are capable of forming a complex in solution that binds DNA in a cooperative fashion.

FIG. 2.

Direct interactions between PPAR and RXRα in the absence or presence of DNA. a, PPAR and RXRα form a complex in solution. Immunoprecipitation was done using in vitro-synthesized [³⁵S]methionine- labelled PPAR in the presence of 150 ng of either bacterially expressed RXRα (lane 2) or control glutathione S-transferase (GST) (lane 1). Polyclonal antiserum prepared against RXRα was used. The position of immunoprecipitated PPAR is indicated by an arrow. Under identical conditions we failed to observe RXRα interactions with radiolabelled glucocorticoid receptor (data not shown, and ref. 14). b, PPAR and RXRα bind cooperatively to the PPRE. Gel mobility shift assays were done using in wtro-synthesized PPAR and/or RXRα as indicated, and ³²P-labelled PPRE oligonucleotide. Pre-immune (PI) or polyclonal antiserum prepared against RXRα (RXRab) were included in the reactions as indicated.

METHODS. PPAR and RXRα RNA was prepared and translated in rabbit reticulocyte lysates as directed by the supplier (Promega). Coimmunoprecipi- tation reactions (20 μl) included 5 of in v/tro-synthesized [³⁵S]methion- ine-labelled PPAR and 150 ng of either bacterially expressed GST-RXR fusion protein¹² or GST alone in 20 mM Tris, pH μl8.0. Proteins were incubated for 20 min on ice before the addition of 5 μl polyclonal antisera prepared against an RXRα peptide (amino acids 214–229). Antigen-antibody complexes were collected by the addition of protein A-Sepharose (Pharmacia) and the immunocomplexes washed three times with 400 μl RIPA buffer (10 mM Tris, pH 8.0, 150mMNaCI, 1% Triton X-100, 1% sodium deoxycholate). Immunoprecipitated complexes were resolved by SDS-polyacrylamide gel electrophoresis on 10% gels which were then fixed in 30% methanol/10% acetic acid, dried, and subjected to autoradiography. Gel mobility shift assays (20 μl) contained 10 mM Tris, pH 8.0,40 mM KCI, 0.05% NP-40,6% glycerol, 1 mM DTT, 0.2 μg poly(dl-dC) and in vitro-synthesized PPAR (2.5 μl) and RXRα (2.5 μl) as indicated in the figure legends. The total amount of reticulocyte lysate was maintained constant in each reaction (5 μl) through the addition of unprogrammed lysate. After a 10-min incubation on ice, 0.5–1 ng of ³²P-labelled oligonucleotide was added and the incubation continued for a further 10 min. DNA-protein complexes were resolved on a 4% polyacrylamide gel in 0.5 xTBE (1 xTBE is 90 mM Tris, 90 mM boric acid, 2mM EDTA, pH 8.0). Gels were dried and autoradiographed at −70°. The PPRE oligonucleotide was 5’-AGCTGGACCAGGACAAAGGTCACGT- TCAGCT-3’.

RXR also forms heterodimers with the receptors for vitamin D, thyroid hormone and retinoic acid (VDR, TR and RAR, respectively), which bind cooperatively to their cognate hormone response elements (HREs)^14–17. In vivo, the selectivity in the hormonal response is achieved through these receptor complexes recognizing direct repeats spaced by 3,4 or 5 nucleotides, respectively (3–4-5 rule)¹⁸. The comparative binding specificity of the PPAR-RXRα complex was assessed by gel mobility shift experiments using as competitors a series of synthetic HREs composed of two AGGTCA direct repeats separated by spacers from zero to five nucleotides in length (DR 0–5; ref. 13). As predicted from in vivo studies¹⁸, heterodimers formed between RXRα and the VDR, TR or RAR interacted most strongly with DR-3, DR-4 and DR-5, respectively (Fig. 3a). In contrast, the PPAR-RXRα complex interacted with the DR-1 oligonucleotide only. The RAR-RXRα complex also bound to DR-1 (Fig. 3a) indicating that there is the potential for competitive interactions between these heterodimers.

FIG. 3.

Binding specificity of the PPAR-RXRα complex on synthetic and natural HREs. a, The binding specificities of the PPAR-RXRα, VDR-RXRα, TR-RXRα and VDR-RXRα complexes were tested by gel mobility shift analysis in the absence (–) or presence of a 4- or 16-fold molar excess of synthetic HREs, DR-0 to DR-5. Radiolabelled oligonucleotides encoding the PPRE, osteopontin VDRE, Moloney leukaemia virus (MLV) LTR TRE, and ß-RARE were used as radiolabelled probes in binding assays involving the PPAR- RXRα, VDR-RXRα, TR-RXRα and RAR-RXRα complexes, respectively. Only the regions of the autoradiographs displaying the bound complexes are shown. Asterisks indicate reactions in which maximal levels of competition were observed, b, The binding specificity of the PPAR-RXRα complex was tested using radiolabelled oligonucleotides encoding the PPRE, 3-ketoacyl- CoA thiolase (3KAT), cellular retinol binding protein type II (CRBPII), chicken ovalbumin (OVAL), DR-1, osteopontin (VDRE), MLV LTR (TRE), and ß-RAR (ß-RARE) HREs. The sequences of these HREs are shown on the right.

METHODS. PPAR, RXRα, TRß, VDR and RXRα RNAs were prepared and translated in rabbit reticulocyte lysates as directed by the supplier (Promega). Gel mobility shift assays were done as for Fig. 2. The 3- keioacyl-CoA thiolase oligonucleotide was 5’-AGC- T CT C AG AGACCTTTGAACCA- CTTC-3’. The CRBPII-RXRE, ß-RARE, MLV TRE, osteo-pontin VDRE, chicken ovalbumin HRE, and synthetic HRE direct repeat series oligonucleotides have been described^12,13,18.

Consistent with the observed DR-1 binding preference, the PPAR-RXRα complex interacted strongly with other natural DR-l-like HREs found in the 3-ketoacyl-CoA thiolase, chicken ovalbumin and cellular retinol-binding protein type II promoters as well as the PPRE (Fig. 3b, lanes 1–4). In contrast, the PPAR-RXRα complex bound only weakly to the retinoic acid response element of the RAR-/3 promoter and the thyroid hormone response element of the Moloney leukaemia virus long terminal repeat, and not at all to the osteopontin vitamin D response element (Fig. 3b, lanes 6–8). The observation that the PPAR-RXRα complex binds a sequence element found in the promoter of the 3-ketoacyl-CoA thiolase gene is particularly interesting, because this gene encodes another of the peroxisomal ß-oxidation enzymes and is induced by peroxisomal proliferators¹⁹. The presence of DR-1-like HREs upstream of the genes for both peroxisomal enzymes suggests that the PPAR- RXRα complex is involved in their coregulation.

The physiological significance of the PPAR-RXRα heterodimer is supported by the observations that both receptors are most highly expressed in the liver and kidney (data not shown, and refs ⁵ and ⁶), and that these tissues are the principal sites of action of peroxisome proliferators²⁰. Thus extracts prepared from these tissues should presumably contain factors that promote PPAR binding to the PPRE. As shown in Fig. 4, liver nuclear extracts have PPRE binding activity (lane 4), but addition of in vitro- synthesized PPAR results in a marked enhancement in DNA binding (Fig. 4; compare lanes 4 and 5), demonstrating a cooperative activity in the extract. Addition of RXRα-specific antiserum to the reaction resulted in a reduction in complex formation to roughly the level in nuclear extracts alone, and the concomitant formation of a new complex with reduced electrophoretic mobility (Fig. 4, lane 6). These results demonstrate that RXRα is the main protein in liver responsible for enhancing PPAR binding to the PPRE in vitro.

FIG. 4.

RXRα present in liver nuclear extracts enhances PPAR binding to the PPRE. Gel mobility shift assays were prepared using ³²P-labelled PPRE oligonucleotide and in vitro- synthesized PPAR and/or RXRα and 0.5 μg of rat liver nuclear extract (NE) as indicated. Pre-immune (PI) or polyclonal antiserum prepared against RXRα (RXRab) was included in the reactions.

METHODS. Gel mobility shift assays were done as described for Fig. 2.

In summary, our results demonstrate that peroxisome proliferators and 9-cis retinoic acid regulate an overlapping set of target genes through the PPAR-RXRα complex. The observation that both retinoids and peroxisome proliferators stimulate transcription of the genes encoding the peroxisomal β-oxidation enzymes^3,4 suggests that 9-cis retinoic acid may function as a hypolipidaemic agent through activation of the PPAR-RXRα heterodimer. Finally, our demonstration that 9-cis retinoic acid is present in the liver and kidney of mouse⁷ is consistent with this molecule serving as an endogenous signal in the modulation of peroxisomal-mediated metabolism of fatty acids and other lipids.