Ligand-induced conformational changes in the human retinoic acid receptor detected using monoclonal antibodies.

The mechanism by which the naturally occurring ligand for a nuclear hormone receptor regulates transcription remains largely unknown. One approach combines the specificity of monoclonal antibodies, which recognize a three-dimensional epitope, with ligand binding. Using purified retinoic acid receptor gamma D and E domains, a panel of six unique monoclonal antibodies were isolated and characterized using solid-state receptor binding and retinoic acid receptor (RAR)-RXR heterodimer supershift formation. Three antibodies are specific for RARgamma (mAbI, mAbII, and mAbV) and four recognize a three-dimensional epitope (mAbI, mAbIV, mAbV, and mAbVI). Three antibodies (mAbIII, mAbV, and mAbVI) dissociate from the receptor in electrophoretic mobility shift assays upon the addition of retinoic acid. In particular, the binding characteristics of mAbIII, whose epitope was mapped to a region identified as an omega-loop (amino acids 207-222), suggest a model for ligand binding to the receptor. In this model, ligand binding causes a positioning of helix 12 into a favorable conformation for interaction with the transcriptional machinery. The Omega-loop then closes in order to stabilize this "active" position. The results reported here also suggest that a region of the hinge or D domain of the receptor (amino acids 156-188), an area that can play a role in protein-protein interactions, may also be important in ligand-induced functional changes.

Retinoic acid (t-RA), 1 one of the biologically active derivatives of vitamin A (retinol) is an important regulator of cell growth and differentiation in both the adult and developing embryo (1,2). The effects of t-RA are mediated through two types of nuclear retinoic acid receptors: RAR, which binds all-trans-RA (t-RA), as well as the isomer 9-cis -RA and RXR, which binds only 9-cis-RA (3). Both RAR and RXR have three subtypes, ␣, ␤, and ␥, each of which have different isoforms (4). RAR and RXR have been shown in vitro to bind cooperatively as heterodimers to retinoic acid response elements (RAREs) (5).
The retinoic acid receptors belong to the type II subfamily of the steroid/thyroid hormone superfamily of nuclear receptors which include thyroid hormone receptors and vitamin D 3 receptor (6) and share a modular structure consisting of six domains (A-F) to which several specific functions have been assigned (4). The E or ligand binding domain has also been shown to have ligand dependent transactivation (AF-2) and dimerization functions as well (7)(8)(9). The D domain, or hinge region, has a function which is largely unknown and consists of three subregions (D1, D2, and D3), of which D2 is the least conserved among subtypes.
Several studies have shown that conformational changes occur in RARs and other nuclear receptors upon ligand binding (10,11). Further, ligand-induced conformational changes have also been hypothesized as prerequisites for transcriptional activation (12,13). Several recently discovered receptor interacting proteins, including some that serve as activators of the AF-2 function (14,15) and others that act as silencing mediators or co-repressors (16 -19), are thought to act through protein-protein interactions dependent upon ligand-induced conformational changes.
Recently, the x-ray crystal structure of a liganded RAR␥ ligand binding domain was reported by Renaud et al. (20). They propose an ⍀-loop structure, which may play a role in ligandinduced conformational changes that impact transcriptional activity. These results support earlier reports, using site-directed mutagenesis, which verify critical amino acid contact points within the ligand binding domain in close proximity to the ⍀-loop (21)(22)(23)(24).
In the present study, we have isolated and characterized a series of monoclonal antibodies raised against purified DE␥ protein (24) that recognize conformational epitopes. mAbI and mAbV are specific for hRAR␥ and recognize elements both in the N-terminal end of the D domain and the C-terminal end of the E domain, thereby placing regions of these domains within close proximity in three-dimensional space. Another antibody, mAbIII, has been mapped to the ⍀-loop region of domain E. These antibodies predict elements of three-dimensional structure that are consistent with the x-ray crystal structure of RAR␥ (20) and are also in good agreement with the recently published structure of RXR␣, which shares considerable sequence homology and secondary structural characteristics (25). mAbIII is of particular interest, since it is shown here to dissociate from the receptor upon binding of t-RA and is therefore sensitive to conformational changes within the receptor. This action of mAbIII suggests a model of the conformational events that occur with ligand binding that may explain the transcriptional activity of t-RA.

EXPERIMENTAL PROCEDURES
Materials-Expression vector pET-15b, and host strain Escherichia coli BL21(DE3) were purchased from Novagen, Inc. pSG5/hRAR␣, ␤, and ␥ and hRXR␣ were gifts from Dr. Pierre Chambon. Taq polymerase, PCR buffers, and deoxynucleoside triphoshates were obtained from Perkin-Elmer Corp. Amplification was performed in an Ericomp Easy-cycler™. Oligonucleotides used in PCR reactions and for electrophoretic shift assay (EMSA) were synthesized by Genosys Biotechnologies, Inc. pMal-C and pIH902 vectors were purchased from New England Bio-Labs, Inc. All cell culture reagents were purchased from Life Technologies, Inc. All other chemicals were purchased from Sigma unless otherwise noted. All experiments involving retinoic acid were performed in subdued light.
Generation of Monoclonal Antibodies to hRAR/DE␥-BALB/C mice were immunized subcutaneously once with a 1:1 mixture of hRAR/DE␥ in complete Freund's adjuvant followed by a second injection in incomplete Freund's adjuvant 2 weeks later (50 g of RAR/animal). Purification of hRAR/DE␥ using nickel chelation chromatography was as described previously (24). Two weeks after the second immunization, 25 g of hRAR/DE␥ in PBS was given intravenously. A mouse demonstrating high titer was sacrificed and its spleen fused 3 days later. The fusion was performed by standard methods (26). Cell culture supernatants of hybrid cells were assayed for anti-RAR antibodies by the antibody capture enzyme-linked immunosorbent assay (ELISA) described below. Twelve positive masterwells were then analyzed by EMSA. On the basis of these results, mAbs were subcloned by limiting dilution from three masterwells. After two rounds of subcloning, six stable cell lines were established. Hybridomas were maintained in Iscove's modified Dulbecco's medium supplemented with L-glutamine (292 g/ml), 15% fetal bovine serum, 1 ϫ penicillin-streptomycin, and 1 ϫ hypoxanthine thymidine supplement. For dilution cloning, 10% hybridoma cloning supplement (Boehringer Mannheim) was added. Monoclonal antibodies were purified using an ImmunoPure IgG purification kit (Pierce) as described by the manufacturer.
Preparation of Receptor Proteins for Use in ELISA and EMSA-Construction of pET-15b expression vectors for full-length RAR␣, ␤, and ␥ and RXR␣ was by PCR amplification from pSG5/hRAR␣, ␤, ␥, or hRXR␣. For RAR␣, ␤, and ␥, forward primers were designed so that an NdeI site at the 5Ј end of each receptor placed an ATG start codon downstream and in frame with the N-terminal polyhistidine tail. Reverse primers for each receptor contained a stop codon, immediately followed by a BamHI site. For RXR␣, the method was the same except XhoI sites at the 5Ј and 3Ј end were used. Verification of the constructs was by DNA sequencing according to the method of Sanger (27), and the integrity of the expressed proteins was confirmed by ligand binding assays (data not shown). Expression in E. coli BL21(DE3) and preparation of crude soluble extracts was as described previously (24), except that cells were grown in superbroth (Quality Biological, Inc.) containing 50 g/ml carbenicillin.
The maltose-binding protein-RAR fusions described in Fig. 2 and used for epitope mapping were prepared using appropriate restriction enzyme sites located within pSG5/hRAR␥ and ligated into either the pMal-C or pIH902 expression vectors. Crude receptor protein extracts were made from E. coli containing these fusions according to the manufacturer's instructions (New England Biolabs).
Epitope Mapping-Epitope mapping was done concurrently by two methods using the MBP/hRAR␥ protein fusions to detect regions to which the monoclonal antibodies bound. The first method was the antibody capture ELISA as described below. The second method involved immunoprecipitation as follows. 10 g of purified mAb was incubated with 100 l of MBP/hRAR␥ protein fusion extracts on ice for 2 h in 50 mM Tris-HCl, pH 8.0. 100 l of 10% v/v protein A-agarose was added to each reaction and allowed to incubate, rotating at 4°C for 1 h. The resin was collected by centrifugation at 14,000 rpm for 1 min at 4°C and washed three times with 0.5 ml of lysis buffer (50 mM Tris-HCl, pH 8.0). The antibody-antigen complexes were removed from the resin by adding 50 l of 1 ϫ Laemmli buffer in the absence of reducing agents and heating at 85°C for 10 min, followed by centrifugation at 14,000 rpm for 2 min. The supernatants were then analyzed on 10% SDS-polyacrylamide gel electrophoresis (29) stained with rapid Coomassie stain (Diversified Biotech).
Determination of Ligand Binding Affinity in the Presence of mAbs-Ligand binding affinity was determined using methods previously described (24). In a given binding assay, 300 ng of purified hRAR/DE␥ receptor protein was incubated with a 10-fold excess of each of the purified mAbs in the presence of 1 nM [ 3 H]t-RA. The apparent K d was calculated from the IC 50 derived from competition for binding with unlabeled t-RA.
Antibody Capture ELISA-Purified DE␥ was diluted to 2 g/ml and crude extracts of RAR␣, ␤, and ␥ and RXR␣ and MBP/hRAR␥ protein fusions were diluted 1:10 with 50 mM Carbonate/Bicarbonate buffer, pH 9.6. Diluted antigen (50 l/well) was placed into 96-well microtiter plates (Fisher). The plates were sealed and incubated at 37°C for 2 h and then at 4°C overnight. The plates were washed three times with PBS containing 0.05% Tween 20. The wells were blocked with 200 l of PBS containing 5% nonfat dry milk for 30 min at room temperature and washed three times as before with PBS/Tween 20. The wells were then incubated with 50 l of cell culture supernatants, diluted if necessary in 50 mM Carbonate/Bicarbonate buffer for 1 h at room temperature, and washed three times as described previously. The wells were next incubated with 50 l/well horseradish peroxidase-conjugated goat antimouse antibody (Southern Biotechnology Associates, Inc.) diluted 1:4,000 in 50 mM Carbonate/Bicarbonate buffer for 30 min at room temperature and then washed three times as before. Finally, each well was incubated with 50 l of substrate buffer (95 mM NaOAc⅐3H 2 O, 3.8 mM citric acid monohydrate, 1.4 mM urea H 2 O 2 , pH 5.5) containing 0.1 mg/ml 3,3Ј,5,5Ј-tetramethylbenzidine at room temperature until blue color developed. The substrate was prepared as follows: 4.9 g of NaOAc⅐3H 2 O was dissolved in 360 ml of water. The pH adjusted to 5.5 with 0.42 g of citric acid monohydrate dissolved in 20 ml of water, after which was added 49.4 mg of urea H 2 O 2 . The reaction was stopped with 50 l/well of 3 N H 2 SO 4 and the absorbance at 450/630 read. Isotype analysis was accomplished using mouse monoclonal antibody isotyping reagents (Sigma) in the antibody capture ELISA described above. To determine if the epitopes were conformational, the antibody capture ELISA described above was repeated using antigen (purified and crude extracts) that had been heated in a 95°C water bath for 15 min.
Electrophoretic Mobility Shift Assay-Receptor protein (1-2 l of E. coli expressed RAR or RXR extracts or 1 l of a 1:1:6 mixture of RAR extract:RXR extract:binding buffer) was incubated with 1 g of poly(dI⅐dC) (Pharmacia Biotech Inc.) and binding buffer for 15 min on ice. The binding buffer contained 10 mM Tris-HCl, pH 8.0, 0.1 mM EDTA, 0.4 mM dithiothreitol, 5% glycerol, and 150 mM KCl in the final reaction volume. In experiments where t-RA was added, 1 l of a 1:1:6 mixture of RAR extract:RXR extract:binding buffer incubated with t-RA dissolved in Me 2 SO was further incubated overnight at 4°C so that the final ligand concentration in the protein-DNA binding reactions ranged from 0 to 1 M. The concentration of Me 2 SO in the initial 18-h incubations of receptor and ligand did not exceed 2% (v/v). The samples were then incubated for 15 min at room temperature after the addition of 1 l of 33 P-labeled, double-stranded oligonucleotide probe consisting of the DR5 ␤RARE sequence 5Ј-GGT AGG GTT CAC CGA AAG TTC ACT CG-3Ј (28) prepared as follows. Each synthesized complementary oligonucleotide was labeled with an equimolar quantity of [␥-33 P]ATP (Du-Pont, specific activity 2,000 Ci/mmol) using polynucleotide kinase (Pharmacia) in labeling buffer (0.1 M dithiothreitol, 0.1 M MgCl 2 , 0.5 M Tris-HCl, pH 9.0) for 30 min at 37°C. The reaction was stopped by addition of 100 mM EDTA. Annealing of the complementary oligonucleotides was accomplished by slow cooling (3-4 h) in a water bath from 95°C to room temperature. The labeled probe was diluted so that 1 l equaled 100,000 cpm.
Antibody-induced supershifts were done by incubation of reaction mixtures (see above) for an additional 15 min at room temperature with 5 l of hybridoma supernatants or, in some cases, 1 l of purified mAb. The final reaction volume was 20 l. Protein-DNA complexes were resolved on 5% nondenaturing polyacrylamide gels (prerun 1 h) in 0.5 ϫ TBE at 12.5 V/cm at 8°C for approximately 2.5 h. The gels were dried onto filter paper, and the radioactivity was visualized either on Kodak X-Omat film or on a Molecular Dynamics PhosphorImager (model 445SI).

RESULTS
Immunization of BALB/C mice with purified hRAR/DE␥ (24) resulted in hybridomas exhibiting a range of binding activities as measured by ELISA and EMSA. The supernatants from 12 masterwells were initially characterized by EMSA on the basis of their ability to supershift recombinant RAR-RXR heterodimers on a labeled DR5 ␤RARE (results not shown). Three masterwells, which tested positive for supershifting, were then subjected to two rounds of subcloning by limiting dilution. Selection of positive clones was by ELISA using purified hRAR/ DE␥ as the antigen. Further characterization of these positive clones resulted in the establishment of six distinct monoclonal antibody-producing cell lines (Table I).
Characterization of mAbs to Purified hRAR/DE␥-The characterization of the six unique mAbs using ELISA and EMSA is shown in Table I. Each antibody was found to be of the IgG1 class using isotype-specific reagents. Additionally, none of the antibodies cross-react with RXR␣ in the ELISA assay.
Three antibodies, mAbI, mAbII, and mAbV, were found to be specific for the RAR␥ receptor by ELISA. Two of these, mAbI and mAbV, were also specific for hRAR␥ by EMSA. mAbII and mAbIV could only be detected by ELISA, most probably because they recognize an epitope that is involved in DNA binding or heterodimerization and therefore were unable to supershift these complexes. Further, mAbI can only be detected by ELISA in the truncated form. This may be due to recognition of an epitope protected in the full-length receptor or may simply reflect the sensitivity of the assay when using a purified target.
Differences were also noted in the degree of antibody binding to the full-length receptor protein as assessed between the two assays for mAbVI, which gave a stronger response for RAR␣ and ␤ in ELISA but a slightly weaker response for RAR␥ as compared with the EMSA. mAbIII had an equally strong response in both assays. The differences between assays may be reflective of the different forms of the antigen that are presented in each, i.e. a monomer in the plate-based assay and a heterodimer complex with DNA in the gel shift assay. Fig. 1 shows a representative EMSA for mAbI demonstrating its specificity for hRAR␥. Taken together, these data suggest that mAbI, mAbII, and mAbV are specific for hRAR␥, while mAbIII, mAbIV, and mAbVI recognize all three receptor subtypes. Interestingly, only mAbIII retains recognition for all three receptor subtypes in both assays.
Conformational Epitope Determination-As pointed out above, the antigen used to produce the antibodies in this report was purified unliganded hRAR/DE␥ expressed in E. coli with binding characteristics comparable to the native protein (24). Since the antigen was present in its native conformation, it is possible that at least some of the antibodies isolated would recognize a three-dimensional determinant. To identify which of the six antibodies fit this criterion, each was tested in a solid state ELISA for binding to native versus denatured protein (Table II). After heat denaturation, only mAbII and mAbIII were able to recognize hRAR/DE␥ in both the native and denatured form reflecting recognition by a primary amino acid sequence epitope. The binding of all other antibodies was nearly eliminated when heat-denatured hRAR/DE␥ was used. These results suggest that mAbI, mAbIV, mAbV, and mAbVI recognize a three-dimensional epitope and that two of these, mAbI and mAbV, recognize hRAR␥-specific sequences or conformations.
Epitope Mapping-In order to identify the regions within hRAR/DE␥ that are recognized by each antibody, particularly in the case of those that recognize a conformational epitope, a series of E. coli expressed, truncated MBP/hRAR␥ protein fusions ( Fig. 2A) were used in immunoprecipitation reactions. Each construct is named for the hRAR␥ amino acid numbers it contains.
FIG. 1. mAbI is specific for RAR␥. Electrophoretic mobility shift assays using RAR-RXR heterodimers and a 33 P-labeled DR5 ␤RARE in the presence or absence of mAbI. Labeled RARE alone is shown in the first lane. Lanes 2, 4, and 6 contain RAR-RXR heterodimers in the absence of mAbI. Lanes 3, 5, and 7 contain RAR-RXR heterodimers in the presence of purified mAbI. The mobility of each complex is indicated by the arrow. and (354 -421) are necessary for antibody recognition. The binding of mAbII, mAbIV, and mAbVI was mapped to amino acids (156 -188) only, and mAbIII was mapped to amino acids (207-222) using the same strategy. This mapping was confirmed using ELISA in which the plates were coated with crude extracts containing the protein fusions. Identical results were achieved except in the case of mAbI and mAbV, where no binding could be detected (results not shown). Inspection of Table I and Fig. 2B indicates that mAbI and mAbV have very similar properties and may actually recognize the same epitope. Both mAbs demonstrated a weak receptor interaction in ELISA compared to a strong interaction in the EMSA and gave nearly identical results with regard to their specificity, ability to bind only native protein, and map locations. In order to determine if these two mAbs were unique, an EMSA was done with various combinations of two antibodies added simultaneously (Fig. 3). The results of this experiment suggest that mAbI and mAbV, while mapped to the same general region, recognize different epitopes since their simultaneous addition results in an increased mobility shift beyond that of either antibody alone. This observation also occurred when mAbI was added simultaneously with mAbVI, indicating that the epitopes contained within the mapped region common to them were different as well. The addition of mAbV with mAbVI, however, did not lead to an enhanced supershift. Since these antibodies are unable to bind concurrently, they must compete for and therefore contain a shared epitope on the receptor. The complete epitopes of mAbV and mAbVI cannot be identical, however, since mAbV requires an additional region for binding (amino acids 354 -421).
Effect of mAb on Ligand Binding-In order to measure the effect of the mAbs on the receptor's ability to bind ligand, the apparent K d of binding to t-RA in the presence of a 10-fold excess of mAb was determined by competition binding experiments. The results given in Table III indicate that no significant change in binding affinity occurred for any of the antibodies tested.
Effect of Ligand Binding on mAb Epitope Recognition-Since the mAbs were shown to have no effect on ligand binding affinity themselves, they became ideal "reporter" molecules with which to moniter ligand-induced surface changes that potentially affected residues within an epitope. In order to prove this hypothesis, EMSA antibody supershifts were performed with receptor that had been incubated with increasing amounts of ligand. Fig. 4A shows the effect of concentrations up to 1 M t-RA on mAbI, mAbV, and mAbVI supershifts of an RAR␥-RXR␣ heterodimer complexed to the ␤RARE as described under "Experimental Procedures." At 10 Ϫ8 M t-RA, there is a clearly diminished supershift with mAbV and mAbVI, while no effect is observed for mAbI, even at the highest concentration of t-RA. Fig. 4B shows the same experiment using mAbIII. The supershifted (upper) bands disappear with increasing t-RA concentration, while the heterodimer (lower) band reappears in a reciprocal fashion. Quantitation of these bands using phosphoimagery is shown in Fig. 4C. The density of the supershifted (upper) band and heterodimer (lower) band is equivalent at approximately 3.5 nM, close to the reported K d for binding of t-RA to hRAR␥ (3).

DISCUSSION
Investigation into the nature of ligand-induced conformational changes and the protein-protein interactions or noninteractions that follow will provide a greater understanding of retinoid action. Primary amino acid sequence has been used to identify non-homologous regions between receptor subtypes, and attempts to raise antibodies against synthetic peptides that recognize these regions have been very successful in selection for receptor specificity (30 -33). The primary goal of this study, however, has been the discovery of antibodies which recognize three-dimensional epitopes that may be sensitive to FIG. 4. The binding of mAbIII, mAbV, and mAbVI but not mAbI to RAR␥-RXR␣ heterodimers is diminished by the addition of increasing concentrations of t-RA. The mobility of each complex is as indicated by the arrows. A, EMSA assays were carried out using RAR␥-RXR␣ heterodimers and a 33   structural changes that occur upon ligand binding. To accomplish this, mice were immunized against the native, unliganded D and E domains of human RAR␥ purified as described previously (24). The resulting monoclonal antibodies (mAbI-VI) were isolated and further characterized. Fig. 5 provides a summary of the primary amino acid sequence for RAR␥ and the regions to which the mAbs in this study were mapped. The helices determined by crystallographic studies for hRAR␥ are noted (20) including the helix previous to helix 1, which is predicted after Garnier et al. (34).
Although the precise recognition sequence for mAbI-mAbVI has not been determined, these antibodies are shown to bind to three distinct regions within the receptor. All mAbs except mAbIII recognize sequence elements in region 1. mAbII and mAbIV are similar in that neither is able to bind an RAR␥-RXR␣ heterodimer/DNA complex. However, the recognized epitopes are clearly distinct since mAbII recognizes RAR␥ only, while mAbIV cross-reacts with all three receptor subtypes. Region 1 also contains amino acids termed the H-box by Predki et al. (35) (amino acids 156 -171). These authors provide evidence that this region may mediate the structural arrangement of heterodimeric partners on RAREs and suggest it functions by direct protein-protein contact with the RXR DNA binding domain. The H-box is contained within the D1 region of the receptor, which is almost completely conserved between RAR␣, ␤, and ␥ and also contains a predicted helical region (34). This observation is consistent with those reported here, since mAbIV was found to recognize all three receptor subtypes and does not bind to denatured receptor (Table II). mAbII, which is RAR␥-specific and recognizes a non-conformational epitope might overlap from this region into D2 (amino acids 171-191), as this region is quite divergent between RAR subtypes and contains a random stretch of amino acids between the predicted ␣-helix and helix 1 (Fig. 5).
Two additional antibodies (mAbI and mAbV) were also found to map to region 1. Epitope mapping of these mAbs using the truncated receptors ( Fig. 2A) led to the interesting finding that the N-terminal region of the D domain (amino acids 156 -188) as well as the C-terminal region of the E domain (amino acids 354 -421) were both required for antibody recognition. Further, these antibodies were unable to recognize receptor protein upon heat denaturation (Table II). This is consistent with the idea that these mAbs recognize a conformational epitope and suggests that region 1 and region 3 are in close proximity in the three-dimensional structure of the native receptor, a finding in good agreement with the recently published x-ray crystallographic structures of RAR␥ (20) and RXR␣ (25), in which helix 1 and helix 9 are in close proximity to one another. Inspection of the primary sequence of the mapped regions and those corresponding to the region immediately preceding helix 1 and the end of helix 9 revealed the presence of a basic amino acid cluster of lysine (amino acids 166 -169) and arginine (amino acids 366 -369), respectively. Inasmuch as the convergence of the N-terminal end of D and C-terminal end of E is immediately adjacent to the DNA binding domain, the presence of the basic amino acid clusters in these regions suggests the possibility of an interaction with the phosphodiester backbone of DNA upstream or downstream of the RARE.
Finally, mAbVI also maps to region 1 and shares a common epitope with mAbV. mAbVI differs from mAbV in that it is not RAR␥-specific, but the specificity of mAbV could be conferred by the area in region 3 to which it and not mAbVI binds. Interestingly, these mAbs are shown to dissociate from the RAR␥-RXR␣ heterodimer/DNA complex upon addition of t-RA, a result not observed with mAbI (Fig. 4A). That these mAbs are sensitive to ligand-induced conformational changes suggests a role for protein-protein interactions in this region. The model proposed by Renaud et al. (20) suggests a major conformational change upon ligand binding, tending toward more compactness of the molecule. Such a conformational change might also preclude the binding of an interacting factor with the receptor in this region. Region 1 contains amino acids 183-188, which are homologous to the first 6 amino acids in the CoR-box of RAR␣, FIG. 6. A model of ligand-induced conformational changes detected using mAbIII. mAbIII is mapped to amino acids 207-222, a region identified as an ⍀-loop. In the unliganded receptor, this region is free to bind to mAbIII. Upon addition of t-RA, mAbIII dissociates from the receptor allowing helix 12 to form a salt bridge with helix 4/5. The ⍀-loop repositions to stabilize this structure. The relative position of the ␣ helices is modified from Ref. 20. Secondary structure is after that given in Ref. 20 for RAR␥ except the helix previous to H1, which is predicted by modeling (34). The shaded rectangles indicate the position of ␣ helices, and unshaded rectangles indicate the position of a ␤ sheet. Vertical arrows indicate the position of the D (1-3) and E domains of hRAR␥. Horizontal arrows indicate the regions to which the mAbs were mapped. Asterisk indicates the first residue detected in the x-ray crystal structure (20). an 18-amino acid region proposed by Kurokawa et al. (19) to bind the nuclear receptor co-repressor, N-CoR.
Only one mAb was isolated that mapped to region 2 (mAbIII). mAbIII recognizes both native and denatured receptor in the random coil region of amino acids 207-222 (Table II). Region 2 has been identified as an ⍀-loop (20) from the x-ray crystal structure and is located between helix 1 and 3. Fig. 4 (B and C) shows that the antibody supershift caused by interaction with mAbIII can be effectively dissociated from the RAR-RXR heterodimer with increasing concentrations of t-RA. Quantitation of the supershifted and heterodimer bands by phosphoimagery show an inverse relationship with the point of intersection at about 3.5 nM, close to the K d for t-RA binding (3).
These results provide direct evidence to support the model proposed by Renaud et al. (20), which is based on a comparison of the x-ray crystal structures derived for holoE-RAR␥ and apoE-RXR␣. Since mAbIII was made using apoDE-RAR␥ as antigen, it most probably recognizes the unliganded conformation as depicted in Fig. 6. When t-RA is bound, helix 12 flips up to cover the ligand binding pocket making a salt bridge between a lysine residue (K264) on helix 4 and a glutamic acid (E414) on helix 12. The ⍀-loop closes over helix 12, thus stabilizing the liganded or "active" conformation and allowing interaction between the AF2 domain contained within helix 12 and the transcriptional machinery. The residues recognized by mAbIII are then unavailable for antibody binding and mAbIII dissociates ( Fig. 4C and 6).
The panel of six distinct mAbs raised against native hRAR/ DE␥ described here have been characterized and shown to be useful probes for providing structural and functional information regarding RAR␥, its interactions with ligand and possibly other factors. In order to support the use of these antibodies as "reporters" of conformational changes on the surface of the receptor protein, it was important to show that the antibodies themselves did not induce structural alterations which affected ligand binding affinity (Table III). The lack of any change in ligand binding affinity induced by the binding of antibodies was not surprising in light of the x-ray crystal structure showing that t-RA binds in the interior of the receptor (20). As a result, several of the mAbs could be used to detect ligand-induced conformation changes. The most striking result involves the binding properties of mAbV and mAbVI, since both of these antibodies dissociate from the receptor in the presence of ligand yet their recognition epitope is located at some distance from the ligand binding pocket. This result suggests that the ligand binding signal can be transduced to other parts of the protein and ultimately affect residues on the protein's surface which may be spatially distinct from the actual ligand binding site. Current experiments are aimed at testing this hypothesis.