Serine 157, a Retinoic Acid Receptor α Residue Phosphorylated by Protein Kinase C in Vitro, Is Involved in RXR·RARα Heterodimerization and Transcriptional Activity*

Retinoic acid (RA) regulation of cellular proliferation and differentiation is mediated, at least in part, through two related nuclear receptors, RAR and RXR. RA-induced modulation of gene expression leads generally to cellular differentiation, whereas stimulation of the protein kinase C (PKC) signaling pathway is associated with cellular proliferation. Pursuant to our discovery that prolonged activation of PKCs induced a strong decrease in RA responsiveness of a retinoid-inducible reporter gene, we have further investigated the connections between these two signaling pathways. We demonstrate that PKC isoforms α and γ are able to phosphorylate human RARα (hRARα) in vitro on a single serine residue located in the extended DNA binding domain (T box). The introduction of a negative charge at this position (serine 157) strongly decreased hRARα transcriptional activity, whereas a similar mutation at other PKC consensus phosphorylation sites had no effect. The effect on transcriptional activation was correlated with a decrease in the capacity of hRARα to heterodimerize with hRXRα. Thus hRARα is a direct target for PKCα and γ, which may control retinoid receptor transcriptional activities during cellular proliferation and differentiation.

External stimuli (neurotransmitters, hormones, mitogens) activating G-protein-coupled receptors and growth factor receptors induce the activity of phospholipases, which in turn generate membrane lipid metabolites. Among them, diacylglycerol (DAG) 1 and others are able to modulate cellular protein kinase C (PKC) activities needed for sustained cellular responses (1,2). The PKC superfamily contains to date 11 iso-forms encoded by 10 genes. The isoforms can be classified into three groups based on sequence homologies and biochemical properties: conventional PKCs (isoforms ␣, ␤1, ␤2, and ␥) are DAG-and calcium-dependent kinases; novel PKCs (isoforms ␦, ⑀, , ) are dependent only on DAG; and a third group of atypical PKCs (isoforms , /) is unresponsive to DAG and calcium. The isoform (mouse protein kinase D) is a high molecular weight enzyme with specific properties such as a transmembrane domain, but it can be considered an atypical PKC in view of its very low affinity for DAG (for review, see Refs. 3 and 4). All of these kinases need negatively charged phospholipids (phosphatidylserine) to exert their phosphorylating activities, triggering major cellular events such as differentiation and proliferation. Although a clear physiological role for each isozyme has not been established definitively, the existence of these isotypes with unique subcellular and tissue distribution suggests a specialized role for each isoform (for review, see Ref. 5). Most notably, gene transfer experiments have underlined a role of these protein kinases in cellular proliferation and tumor growth progression.
Nuclear receptors are ligand-dependent transcription factors that control a wide range of cellular events, including cellular proliferation and differentiation. Several lines of evidence point to a direct role of the PKC signaling pathway in modulating the transcriptional activity of some members of this superfamily. The thyroid hormone receptor appears to be a target for PKC (6) as well as the human vitamin D receptor (hVDR). hVDR is phosphorylated by PKC␤ in the DNA binding domain (DBD), and phosphorylated serine (Ser-51) is important for hVDR transactivation properties (7,8). Regulation of other nuclear receptors (glucocorticoid receptor, peroxisome proliferator-activated receptor, estrogen receptor) through activation of the MAP kinase cascade has also been documented and suggests that PKC, as an activator of the MAP kinase pathway, might participate indirectly in the control of the transcriptional activity of these nuclear receptors (9 -13).
Our previous experiments pointed to a role of PKC in the regulation of hRAR␣ transcriptional activity. They indicated that chronic treatment of COS-7 cells with TPA led to the specific inhibition of the activity of a retinoid-inducible reporter gene. We determined that hRAR␣ is a substrate for PKC in vitro, and preliminary results suggested that two to four phosphorylation sites occurred out of the ligand binding domain (LBD) of this nuclear receptor (14). In light of these results, we addressed in this study several questions to elucidate further the role of PKC isozymes in the regulation of the transcriptional activity of hRAR␣. We characterized PKC isoforms for their capacity to phosphorylate hRAR␣ in vitro and identified the target amino acid. Mutation of the phosphorylated serine affected hRAR␣ transcriptional activity, and the molecular basis for the observed inhibition was investigated. Our results support the hypothesis that PKC␣ and PKC␥ are direct regu-lators of hRAR␣ transcriptional activity by altering its ability to dimerize with RXRs.
All mutations were checked by restriction analysis and automatic sequencing. Detailed sequence information is available upon request.
Cell Culture and Transfections-HeLa cells were cultured as monolayers in Dulbecco's minimal essential medium supplemented with 10% fetal calf serum. Cells were treated when indicated with TPA and/or retinoids to a final concentration of 10 Ϫ7 M and 10 Ϫ6 M, respectively. Transfections were carried out using the polyethyleneimine coprecipitation method as described (17). The luciferase assay was performed as described (15).
Receptor Purification and Phosphorylation Analysis-A detailled procedure has been published elsewhere (18). Briefly, His 6 -hRAR␣ was overexpressed in Escherichia coli and purified to homogeneity by nickel-immobilized affinity chromatography. The purified polypeptide was phosphorylated and purified further by 8% SDS-PAGE. Gel slices containing the phosphorylated receptor were desiccated and rehydrated with trypsin digestion buffer, and trypsin was added to a 1:50 (w:w) ratio. Peptides were extracted and purified on a C18 column and submitted to Edman degradation reaction for sequencing. Phosphoamino acid determination was as follows. The phosphorylated hRAR␣ was purified by 8% SDS-PAGE and transfer onto a polyvinylidene difluoride membrane. Bands containing the radioactive receptor were cut out and transferred into an Eppendorf tube containing 200 l of 5.7 N HCl. After a 1-h incubation at 110°C, the hydrolysate was dried and resuspended in 20 l of electrophoresis buffer (0.5% acetic acid, 0.5% pyridine at pH 3.5). 10 g of phosphoserine, phosphothreonine, and phosphotyrosine (Sigma) were added to the sample, which was submitted to electrophoresis on a cellulose glass-backed plate for 30 min at 1,000 V (30 mA). Plates were then dried and sprayed with ninhydrin to locate phos-phoamino acid standard. 32 P-Labeled amino acids were identified by autoradiography.
Western Blotting and Antibodies-Antibodies directed against isoforms of PKC were obtained from Transduction Laboratories (Lexington, KY). The anti-MRGS(His) 6 and anti-(His) 5 monoclonal antibodies were purchased from Qiagen. The anti-RAR␣ monoclonal antibody was from Affinity BioReagents (Neshanic Station, NJ). Peroxidase-coupled anti-mouse and anti-rabbit IgGs were from Sigma. Immunodetections were carried out as described previously using the Amersham ECLplus detection system. SDS-PAGE, electrotransfer of proteins, and immunodetection procedures have been described in Ref. 14.
Protein-Protein Interaction Assays-Protocols for glutathione S-transferase pull-down experiments have been described elsewhere (17,19). Matrix-bound receptors were then resolved by 8% SDS-PAGE and detected by autoradiography or quantified using a Storm Phospho-rImager (Molecular Dynamics). The RXR binding assay used in Fig. 6B was carried out as described in Ref. 16. Briefly, native or PKC-phosphorylated purified His 6 -hRAR␣ was diluted to a final concentration of 20 g/ml in 1 ϫ phosphate-buffered saline and adsorbed (ϳ 0.5 g/well) to a Corning 96-well tissue culture-treated plate for 16 h at 4°C. Wells were then washed three times with 1 ϫ phosphate-buffered saline and coated for 3 h with a 5% fetal calf serum solution in 1 ϫ phosphatebuffered saline to saturate nonspecific binding sites. Wells were washed three times with 1 ϫ phosphate-buffered saline, and 10 6 cpm of 35 Slabeled hRXR␣ was loaded into each well and incubated for 90 min at 4°C in 30 l of binding buffer (20 mM HEPES, pH 7.8, 130 mM KCl, 1 mM EDTA, 1 mM ␤-mercaptoethanol, 0.05% Nonidet P-40, and 20% glycerol). Unbound hRXR␣ was removed by four washes with ice-cold binding buffer, and the specifically adsorbed 35 S-labeled hRXR␣ was released by incubation with 50 l of 0.1% SDS, 0.4 M HCl. Radioactivity was quantified by scintillation counting. Protein assays were carried out according to Bradford (20) using bovine serum albumin as a standard.
Statistical Analysis-All incubations or assays were performed at least in triplicate. Measured values were used to calculate a mean Ϯ S.E., and groups of data were compared using a two-sided analysis of variance test followed by a Dunnett test. Significance was defined as p Ͻ than 0.01. Calculations were carried out using the Prism software (GraphPad Inc., San Diego, CA).

hRAR␣ Is Phosphorylated by PKC␣ and PKC␥ on Serine
Residues in Vitro-We reported previously that one or two amino acids in the hRAR␣ sequence located outside of the LBD were phosphorylated by PKCs purified from rat brain (14). We thus determined which isoforms of PKC were present in these commercial preparations by Western blotting analysis using a panel of isozyme-selective monoclonal antibodies. We found that rat brain extracts contained exclusively the PKC␣ and ␥ isoforms, with trace amounts of the isoform (Fig. 1A). Using this protein kinase mixture in an in vitro phosphorylation reaction with purified His 6 -tagged hRAR␣ as a substrate, we found that this receptor was a substrate for PKCs, as reported previously (Fig. 1B). PKCs were able, in these conditions, to display autophosphorylating activity as documented previously (for review, see Ref. 21), but no kinase activity was detected in purified hRAR␣ preparations. The PKC-catalyzed hRAR␣ phosphorylation was inhibited by GF109203X, a specific but general inhibitor of PKCs, and by an inhibitor specific for ␣, ␤, and ␥ isoforms (Gö6976; Ref. 22). Gö6976 strongly inhibited PKC-directed phosphorylation of hRAR␣, but phosphate incorporation was still detectable, suggesting that PKC may display some weak phosphorylating activity in these conditions. From these experiments, we conclude that PKC␣ and/or PKC␥ is the most active isozyme phosphorylating hRAR␣ in vitro.
Phosphorylated hRAR␣ was gel purified and submitted to acid hydrolysis to determine which type of amino acid was phosphorylated in these conditions. Products were analyzed by thin layer chromatography and identified by comigration with phosphoamino acid standards (Fig. 1C). Only phosphoserine was detected in these conditions, suggesting that hRAR␣ is phosphorylated by PKC only on serine(s) located outside of the LBD.
Identification of Serine Residues Phosphorylated in Vitro by PKC␣ and PCK␥-To identify hRAR␣ amino acids phosphorylated by PKC␣ and ␥ isoforms further, we submitted the purified, denatured, phosphorylated receptor to trypsin digestion. Proteolysis products were fractionated by reverse phase high performance liquid chromatography, and two labeled peptides (Fig. 2B) were reproducibly detected in these conditions, albeit in varying ratio. 2 These peptides were NH 2 -terminally sequenced by the Edman reaction. Peak I corresponded to the sequence (K) 148 CFEVGMS, whereas peak II was identified as peptide (R) 139 CQYCRLQ. The occurrence of lysine (K) and of arginine (R) upstream of the cleavage site in the primary sequence of hRAR␣ identified these peptides as products of trypsin cleavage. These results indicated, however, that partial tryptic digests were generated in our conditions but nevertheless mapped phosphorylated serine(s) to the same region of the DBD, spanning the T and A boxes. A PKC consensus phosphorylation site is found at position 157. Because partial proteolysis was likely, Ser-154 remained a possible target for PKCcatalyzed phosphorylation, although this amino acid is not surrounded by amino acids usually defining a PKC-directed phosphorylation site (RXXS/TXR). Thus we converted Ser-157 and Ser-115, Ser-154, Ser-232, and Ser-388 into an alanine (A) to determine whether these serine residues could be minor phosphorylation sites undetected in our phosphopeptide mapping experiments. His-tagged wt and mutated hRAR␣ were expressed in bacteria, purified by immobilized metal affinity chromatography, and tested for their ability to serve as a substrate for PKC. Phosphorylation reactions were carried out as described above, using an identical amount of each receptor derivative, and products were subjected to SDS-PAGE. Receptor concentrations were controlled by silver staining (Fig. 3, left panels) of gels, which were then dried and autoradiographed (Fig. 3, right panels). We found that only mutation of Ser-157 severely compromised PKC-catalyzed phosphorylation, whereas all other substitutions designed to inactivate other consensus sequences had no significant effect on the phosphate content of hRAR␣. Residual phosphate incorporation was, however, observed with the S157A mutant, suggesting that other sites may represent poor substrates for PKC. Alternatively, this residual activity might result from contaminating, unidentified protein kinases in the rat brain extract. Thus Ser-157 appeared to be the major target for PKC in vitro, and these data support both peptide mapping results and our previous conclusion that putative phosphorylation site(s) lie outside of the LBD (14). Ser-157 is located in the DBD of the receptor, in a region lying COOH-terminally of the second zinc finger motif. Sequence alignment between RAR and RXR isoforms showed that Ser-157 is found only in RAR␣ and RAR␤ (Fig. 4) and is located in the so-called T box region, which has been reported 2 A. Tahayato, P. Formstecher, and P. Lefebvre, unpublished data.

FIG. 1. Phosphorylation of hRAR␣ by PKCs in vitro. Panel A,
identification of PKC isoforms purified from rat brain. 1 ng of purified rat brain PKC (Calbiochem) was fractionated by 8% SDS-PAGE and blotted on a polyvinylidene difluoride membrane. Immunodetection of each isoform was carried out using specific monoclonal antibodies as described under "Experimental Procedures." Molecular masses are indicated on the left. Panel B, hRAR␣ is phosphorylated by PKC␣ and PKC␥ in vitro. His 6 -tagged hRAR␣ was purified to near homogeneity from E. coli extracts using NiTA affinity chromatography and incubated with purified rat brain PKCs and [␥-32 P]ATP, in the presence of a PKC inhibitor (100 nM GF109203X) or an inhibitor specific for ␣, ␤, and ␥ isoforms (100 nM Gö6976). Phosphorylated proteins were analyzed by 8% SDS-PAGE, and gels were either silver stained (left panel) or autoradiographed (right panel). Panel C, PKC␣ and PKC␥ phosphorylate hRAR␣ on serine residues. hRAR␣ was phosphorylated by rat brain PKCs and gel purified as in panel B. The polypeptide was submitted to total acid hydrolysis, and reaction products were separated by thin layer chromatography and identified by comigration with phosphoamino acid standards.

FIG. 2.
Identification of PKC phosphorylation sites along the hRAR␣ primary sequence. hRAR␣ was used as a substrate for purified PKCs and gel purified as above. The denaturated polypeptide was then digested with TPCK-treated trypsin, and cleaved peptides were fractionated on a C18 reverse phase HPLC column. Peptides were identified by monitoring OD at 225 nm (panel A), and phosphorylated species were localized by quantifying radioactivity in each fraction (panel B). Fractions containing phosphorylated peptides (I and II) were lyophilized and subjected to automatic Edman degradation. The first radioactive peak (panel B) corresponds to unbound radioactivity, and the dotted lines indicate the acetonitrile concentration in percent (right axis).
to be important for dimerization and DNA binding activities of several nuclear receptors (23,24).
Converting Serine 157 into Aspartic Acid Decreases Transcriptional Activity of hRAR␣-To examine the role of this phosphorylatable amino acid in regulating the transcriptional activity of hRAR␣, we constructed eukaryotic expression vectors encoding Ser to Ala or Gly, and Ser to Asp hRAR␣ mutants, either to prevent constitutive phosphorylation by PKC (Ser to Ala/Gly mutants) or to mimic the net charge increase brought by phosphorylation (Ser to Asp mutants). As for in vitro experiments, we deliberately mutated all potential phosphorylation sites for PKC (Ser-115, Ser-157, Ser-232, Ser-388, and Ser-452) and compared the activity of Ser to Ala and of Ser to Asp mutants with that of the wt hRAR␣. HeLa cells express low levels of RARs and RXRs which are able to activate prototypical retinoid response elements such as the direct repeat found in the RAR␤2 gene promoter (DR5). To avoid any interference with endogenous receptors, we used the luciferase reporter gene driven by the synthetic, palindromic thyroid response element TREpal, which is also activated by retinoids but necessitates high levels of receptors which are only achieved upon transfection of RAR and/or RXR expression vectors (17,25). We thus monitored the transcriptional activity of Ser to Ala and of Ser to Asp mutants in the presence of hRXR␣ in this system and compared them with the activity of wt hRAR␣, using atRA as an inducer at a 10 Ϫ6 M final concentration (Fig. 5). The basal level of luciferase activity measured in the presence of wt hRAR␣ or mutant receptors did not fluctuate significantly, suggesting that these receptors have a similar affinity for nuclear corepressors. In vitro protein-protein interaction assays revealed that all of these receptors bound to the nuclear corepressor SMRT with a wt affinity. 3 The addition of atRA to the culture medium leads typically to a 5-10-fold increase in luciferase activity: only two mutations affected the ability of hRAR␣ to respond to atRA, S115G and S157D, although both receptors were expressed at levels comparable to that of wt hRAR␣. 3 S115G activity decreased by 30% compared with wt hRAR␣. Because Ser-115 is not phosphorylated by PKCs (Fig. 3), and S115D displayed a wt activity, it is likely that the observed decreased activity results from structural alterations of the DBD and not from direct phosphorylation. Ser-115 is indeed located between the two zinc fingers of hRAR␣, and mutations at this position in the hVDR sequence 3 M.-H. Delmotte and P. Lefebvre, unpublished data.

FIG. 3. Mutation of serine 157 prevents phosphorylation of hRAR␣ by PKCs.
Wild-type or mutated hRAR␣ was overexpressed in E. coli and purified by NiTA chromatography. Each receptor was submitted to a phosphorylation reaction in the presence of [␥-32 P]ATP and PKCs from rat brain. Samples were resolved by 8% SDS-PAGE, gels were silver stained (left panels, Proteins) and autoradiographed (right panels, 32

P).
Positions of molecular mass markers are indicated on the left, and the hRAR␣ position is indicated by an arrowhead (right). Note that electrophoresis conditions were slightly different for mutants S157A and S388A, explaining the observed different mobility for labeled polypeptides of ϳ25 and 18 kDa. (S51) were proposed to alter the ␣-helical structure of this region (8). A decrease of at least 50% was observed in atRAinduced luciferase activity in cells overexpressing S157D, whereas S157A retained wt activity, consistent with our working hypothesis of a PKC-mediated inactivation of hRAR␣. The latter observation also suggests that no phosphorylation of Ser-157 occurs in nonstimulated cells, in agreement with a previous report (27). We also note that this decrease in activity is similar to that observed upon chronic treatment of cells with TPA (14), and therefore identify Ser-157 as a functional substrate for PKCs.
Phosphorylation of Serine 157 Decreases the Ability of hRAR␣ to Heterodimerize with RXR-The effects of the Ser-157 mutation on RAR properties were investigated: the ability of the mutated receptor to bind natural and synthetic ligands, its subcellular localization in response to ligand as well as its half-life, its interaction with nuclear corepressor (SMRT) and coactivators (SRC-1, RIP140) were tested. None of these assays revealed functional alterations of the receptor. 3 The ability of the receptor to interact with RXR in a ligand-dependent manner in the absence of DNA was quantified in glutathione Stransferase pull-down experiments (Fig. 6). A Sepharose GST-RXR affinity matrix was used as a bait for 35 S-labeled wt hRAR␣ and receptor mutants S157A and S157D, in the presence or not of 1 M atRA. Similar amounts of receptors were used in each condition and high salt washes allowed to monitor the ligand-induced stabilization of RAR-RXR interaction. Heterodimer formation was stimulated 3-fold for wt hRAR␣ and the S157A mutant (Fig. 6A). S157D displayed a slightly but significantly decreased constitutive dimerization (in the absence of atRA) and a 2-fold reduction in its ability to dimerize with RXR in the presence of atRA. This result was confirmed using an independent assay in which phosphorylated or native His-tagged RAR was immobilized on a hydrophobic matrix (8). 35 S-Labeled RXR was then incubated, in the presence of 1 M atRA, with this matrix, and specifically bound receptors were quantified (Fig. 6B). RAR phosphorylation by PKC, in conditions where more than 80% of the receptor was phosphorylated by PKCs, led to a 50% decrease in RXR retention on this matrix, in agreement with results generated by the glutathione S-transferase pull-down assay. Thus to a 50% decreased transcriptional activity is associated with a decreased ability to form a dimer with hRXR␣, strongly suggesting that PKC-catalyzed post-translational modification of hRAR␣ is controlling the ability of RAR to heterodimerize with RXR.

DISCUSSION
Modulation of the phosphorylation state plays an important role in the reversible control of the activity of many proteins, including nuclear receptors. In the case of hRAR␣1, heterodimerization properties have been shown to be altered by protein phosphatases 1 and 2A (15) and the NH 2 -terminal function AF-1 to be activated through the proline-directed Cdk7 (27). We now show that dimerization of hRAR␣ with hRXR␣ can be modulated by PKC-dependent phosphorylation. Several potential phosphorylation sites by PKCs are located along the hRAR␣ sequence. Phosphopeptide mapping studies as well as mutagenesis of hRAR␣ identified Ser-157 as the major target for PKC␣ and PKC␥ isoforms (Figs. 2 and 3). This amino acid is located in the T box of hRAR␣, which, together with the A box, are part of a larger region (hinge domain) that is organized into an ␣-helical structure in the RXR⅐T 3 R DBD dimer (23). Molecular modeling of RXR⅐RAR dimers suggested that the T box, forming a loop interrupted by an ␣-helical turn perpendicular to the DNA axis, engages dimerization contacts with the RXR DBD. Interestingly, Ser-157 neighbors amino acids that, in RXR⅐T 3 R and RXR⅐VDR dimer structures, establish salt bridges with the RXR DBD (Asp-69 and Glu-69, respectively). Thus phosphorylation of Ser-157 may introduce strong conformational constraints on this dimerization region and, as observed, decrease the relative affinity of RAR DBD for RXR DBD and inhibit dimerization. Alternatively, phosphorylation of Ser-157 may perturb the relative orientation of RAR DBD and LBD as suggested for VDR (24) and in turn modify the orientation of the strong dimerization interface located in the hRAR␣ LBD (8). Protein kinases C are a family of Ser/Thr kinases whose activities are involved in the regulation of complex biological responses such as differentiation, proliferation, and apoptosis. They are organized in two domains, a COOHterminal catalytic domain and an NH 2 -terminal regulatory domain. Their tissue distribution is in most cases ubiquitous, with the exception of PKC␥, which is expressed specifically in the brain and spinal cord. Response to agonists (hormones, neurotransmitters, growth factors) is mediated through a transient short term production of DAG, essentially through phosphatidylinositol 4,5-bisphosphate hydrolysis. A more sustained response may be observed with phosphatidylcholine hydrolysis or phorbol ester treatment. The basic model of PKC activation suggests that PKC translocation to the plasma membrane is the main event regulating this signaling cascade. However, PKC is often found, irrespective of its activation state, in the particulate fraction (membranes, organelles, and nuclei) of cul- FIG. 5. Comparison of the transcriptional activities of hRAR␣ mutants in HeLa cells. HeLa cells were cotransfected with (TREpal)TATA 3 Luc, pSG 5 wt hRAR␣, or mutants and pSG 5 hRXR␣. Basal and ligand-induced activities of the (TREpal) 3 luciferase reporter gene are represented as bar chart for each mutant, in the presence (atRA) or absence (basal) of 1 M atRA. Each mutant activity was assayed when overexpressed together with hRXR␣, thus reflecting heterodimeric RAR activity. The average luciferase activity from at least six experiments (Ϯ S.E.) is given for each mutant, with the level of luciferase activity of wt hRAR␣ observed in the presence of 1 M atRA taken as 100%. * p ϭ 0.01; **p ϭ 0.04. tured cells and in nuclei themselves. HeLa cells were found to express ␣, ␥, , , , and isozymes, among which ␣ and ␥ isozymes appeared to be sensitive to TPA-induced down-regulation. 3 Both PKC␣ and ␥ can be found in nuclei of HeLa cells. 3 A prolonged activation of PKCs by phorbol esters or repeated DAG stimulation leads to cellular depletion in PKCs by inducing a high rate of proteolytic clipping of DAG-dependent kinases, concomitantly with the nuclear translocation of the catalytic domain (PKM), which displays unregulated kinase activity (28). We observed that the catalytic domain of PKC␣ and ␥ (PKM) was almost exclusively nuclear and that PKM accumulated in nuclei upon TPA treatment, to reach a 3-5-fold higher concentration than in control cells. 3 However, this treatment, much like PKC␣ overexpression, also led to strong alteration of the cellular morphology as well as perturbation of the cell cycle, and therefore made a direct correlation between PKC activation/translocation and hRAR␣ inhibition very difficult to establish. However, in these types of experiments, S157A displayed a very low sensitivity to TPA treatment, in opposition to wt hRAR␣. 3 Several nuclear proteins are potential or known substrates for PKCs: LIM-containing proteins (29), p53 (30), WT1 (31), c-myc (32), and nuclear receptors such as T 3 R (6), VDR (7), RAR (14 and this report) and RXR. 2 PKC activation has been reported to inhibit VDR-controlled transcription (33) through a mechanism involving direct phosphorylation of the DBD (8). RAR-mediated transactivation has also been reported to be strongly potentiated by short term phorbol ester treatment in T cells (34). However, similar experiments carried out in HeLa cells demonstrated a strong inhibition of the atRAinduced transcriptional response, in agreement with the proposed inhibition of the retinoid pathway by the AP-1 signaling module. 4 It is therefore likely that the reported potentiation arises from an AP-1 independent, cell-type specific indirect mechanism.
atRA has been described as an effector able to modulate PKC isozymes expression (35)(36)(37)(38) and/or PKC subcellular repartition in several cell lines (39,40). However, our data neither provided evidence for significant variation of PKC isoform content in atRA-treated cells, nor we were able to document alteration of the cellular partitioning of expressed intact isoforms by indirect immunofluorescence. 3 This set of data clearly suggests that interferences between the PKC and retinoid signaling pathways are cell-specific and diverse and that, in this respect, one must distinguish between short term events (AP-1 induction, PKC activation) and long term events (alteration of PKC expression rate). Short term TPA treatment of HeLa cells may activate several signaling pathways involving PKCs, MAP kinases, and c-Jun NH 2 -terminal kinase (JNK). PKC down-regulation has been reported to block completely JNK activation by phorbol esters in HeLa cells (41), and MAP kinase overexpression in COS cells did not modify the phosphorylation pattern of hRAR␣ (27). No increase in MAP kinase activity was detected in TPA-treated HeLa cells, 3 thus ruling out a potential involvement of these two protein kinases in the observed inactivation of hRAR␣. Furthermore, additional cross-talk processes can be also considered based on the nongenomic effects of steroid and of others ligands for nuclear receptors. For example, vitamin D3 has been shown to activate the MAP kinase signaling cascade through activation of PKC (42) and is able to induce subcellular redistribution of calcium-dependent PKCs. Tamoxifen, a therapeutic/chemopreventive nonsteroidal antagonist of estrogen receptor, induces PKC translocation to the membrane and down-regulation of this enzyme through oxidative stress (43). Finally, retinoids, estrogens, and vitamin D3 treatment of cells modulate the rate of expression of PKC isoforms (26, 44 -46), underlining a complex network of intertwined, and often antagonistic, signaling pathways eventually controlling cell differentiation, growth, and death.
In conclusion, our findings establish that PKC isoforms are able to phosphorylate hRAR␣ and control its dimerization properties, very likely through T box conformational alteration. The observed inhibition may be of importance in differentiation and proliferation processes during which sustained activation of PKCs is necessary to induce a full biological response to external stimuli.
FIG. 6. hRAR␣ heterodimerization with hRXR␣ is inhibited by phosphorylation of serine 157. Panel A, 35 S-Labeled hRAR␣ polypeptides were produced by in vitro coupled transcription-translation, incubated for 2 h with dimethyl sulfoxide or 1 M atRA, and adsorbed on a Sepharose-GST-hRXR␣ affinity matrix. Bound receptors were resolved by SDS-PAGE and quantified by PhosphorImager analysis. Results are shown as the average of three independent experiments (Ϯ S.D.) and are presented with a typical autoradiogram. Results from analysis were normalized for translation efficiency using the protein input lane as a reference (which represents 20% of total input) and expressed as the ratio of bound receptor to 20% input protein. Panel B, purified wt hRAR␣ was submitted to a phosphorylation reaction in the presence of rat brain purified PKCs, ATP in the presence or not of a specific inhibitor (100 nM GF109203X) and adsorbed to microtiter wells. After extensive washing to remove unbound ATP, 10 6 cpm of 35 Slabeled hRXR␣ was added to the wells and incubated for 2 h with 1 M atRA. Bound material was quantified by scintillation counting, and the results are expressed as means Ϯ S.D. of three independent experiments.