The High Affinity Ligand Binding Conformation of the Nuclear 1,25-dihydroxyvitamin D3 Receptor is Functionally Linked to the Transactivation Domain 2 (AF-2)

 1996 Oxford University Press Nucleic Acids Research, 1996, Vol. 24, No. 22 4513–4518

The high affinity ligand binding conformation of the

nuclear 1,25-dihydroxyvitamin D

₃

receptor is functionally

linked to the transactivation domain 2 (AF-2)

Sepideh Nayeri

,

Jean-Pierre Kahlen and Carsten Carlberg*

Clinique de Dermatologie, Hôpital Cantonal Universitaire, CH-1211 Genève 14, Switzerland

Received July 24, 1996; Revised and Accepted October 9, 1996

ABSTRACT

The nuclear receptor for 1,25-dihydroxyvitamin D3 (VD), VDR, is a transcription factor that mediates all genomic actions of the hormone. The activation of VDR by ligand induces a conformational change within its ligand binding domain (LBD). Due to the lack of a crystal structure analysis, biochemical methods have to be applied in order to investigate the details of this receptor–ligand interaction. The limited protease digestion assay can be used as a tool for the determination of a functional dissociation constant (K_df) of VDR with any potential ligand. This method provided with the natural hormone VD two protease-resistant fragments of the VDR LBD and with the 20-epi conformation of VD, known as MC1288, even an additional fragment of intermediate size. These frag-ments were interpreted as different receptor conforma-tions and their decreasing size was found to be associated with decreasing ligand binding affinity. A critical amino acid for VDR’s high ligand binding conformation has been identified by C-terminal recep-tor truncations and point mutations as phenylalanine 422. This amino acid appears to directly contact the ligand and belongs to the ligand-inducible activation function-2 (AF-2) domain. Moreover, functional assays supported the observation that high affinity ligand binding is directly linked to transactivation function. INTRODUCTION

The biological active form of vitamin D3, 1,25-dihydroxyvitamin

D3 (VD), is a structurally very flexible seco-steroid that has

important regulatory functions on calcium homeostasis, but also on cellular growth, differentiation and apoptosis (1). VD mediates its genomic actions through the activation of its specific nuclear receptor VDR, which is a member of the large family of structurally-related transcription factors, called the nuclear recep-tor superfamily (2). While other nuclear receptors, e.g., retinoid receptors, have several subtypes and various isoforms, only one form of human VDR is known, i.e. all direct gene regulatory effects of VD are mediated by one single receptor type. However,

different VDR conformations may explain the functional pleio-tropy of VD.

Most nuclear receptors, including VDR, are exclusively localized in the nucleus and most of the small lipophilic ligands enter the nucleus passively by diffusion. It is well accepted that ligand binding induces a conformational change in the LBD of the nuclear receptor, which transforms it into an activated state. Recent crystal structure and subsequent computer modeling analysis of the human retinoid X receptor α (RXRα) (3), which is a nuclear receptor for 9-cis retinoic acid (RA), of the human all-trans RA receptor γ (RARγ) (4) and of the rat thyroid hormone receptor β (TRβ) (5) provided important details and principles of this interaction. Several hydrophobic amino acids of the LBD form the inner surface of the ligand binding pocket, but for each of the three analyzed receptors only three amino acid residues were identified that directly contact the ligand. For VDR, so far, no crystal structure has been reported and the amino acids that are critical for ligand binding have not been determined.

It was suggested that fitting of the ligand into its binding pocket mediates a structural change of the LBD that appears to result in the exposure of a small, evolutionary well conserved region to the surface of the receptor (3). These few amino acids are located close to the C-terminus of the receptor and are referred to as the activation function-2 (AF-2) domain (6,7). The AF-2 domain provides an interface for the interaction of the nuclear receptor with a co-activator or a co-repressor protein that mediates the signal ‘activation’ or ‘repression’, respectively, to the basal transcriptional machinery (8–10).

A prerequisite for this ligand-activated protein–protein interac-tion is the fixainterac-tion of the nuclear receptor on DNA in the vicinity of a transcriptional start site. As monomer, most nuclear receptors do not have sufficient affinity to DNA, thus they form homo- and heterodimers on directly repeated (DR), palindromic and inverted palindromic (IP) arrangements of two hexameric binding motifs with specific numbers of spacing nucleotides (11,12). These DNA sequences are called response elements. VDR can form homodimers (13) and also heterodimers with RAR and TR (14), but in most cases the receptor is found as a heterodimer with RXR (15). Natural VD response elements have been found to have either DR3-, DR4-, DR6- or IP9-type structures (12).

In vitro VDR binds to appropriate response elements in the

absence of ligand, but in vivo footprinting experiments (16)

suggested that, for the binding to chromosomal DNA, VDR needs to pre-form a complex with its ligand. This leads to the question, whether the binding of the ligand influences the dimerization and DNA binding properties of VDR, i.e. its conformation, or vice versa. Ligand binding is mostly studied by traditional competition assays using radiolabelled ligand, but this method does not visualize any conformational changes of the receptor. An attractive alternative is the limited protease digestion assay, which is based on the principle that the conformational change of the LBD could hide a cutting site for a protease and create, under limited reaction conditions, a protease-resistant receptor frag-ment. This method has mainly been developed in order to demonstrate the physical interaction of a receptor with its ligand (17–20) and is also suitable for the screening of potential nuclear receptor ligands (21).

In this report the limited protease digestion assay has been applied to investigate different VDR conformations. Three conformations could be discriminated, but only one of them was found to be the high ligand binding conformation of the receptor. C-terminal truncations and point mutations of VDR identified the phenylalanine at position 422 as a critical amino acid for the interaction with the ligand.

MATERIALS AND METHODS Compounds

VD and MC1288 (20-epi VD) were kindly provided by L. Binderup (LEO Pharmaceutical Products, Denmark). Both compounds were dissolved in 2-propanol at 4 mM. Dilutions were performed in ethanol (final concentration of ethanol in the cell culture medium: 0.1 %).

DNA constructs

The cDNA for human VDR and human RXRα were subcloned into the expression vector pSG5 (Stratagene) (13). For the point mutations V421A, F422A, G423A, F422*, G423* and N424* (V, valine; A, alanine; F, phenylalanine; G, glycine; *, stop; N, aspa-ragine) linearized VDR wild type cDNA was used as template for two separate PCR reactions with the profile 1 min at 94
C, 1 min at 60
C and 0.5 min at 72
C for 30 cycles. For the generation of the point mutation V421A the two primer pairs PST (CAACAC-ACTGCAGACGTACA) / V421A– (TCTCATTGCCAAACGC-TTCGAGCACAAGG) and V421A+ (CCTTGTGCTCGAAG-CGTTTGGCAATGAGA) / XBA (TTTGAGTGAGCTGATAC-CGC) were used. The respective specific primers for the five other point mutations have been F422– (ATCTCATTGCCAGC-CACTTCGAGCACAA) and F422A+ (TTGTGCTCGAAGTG-GCTGGCAATGAGAT), G423A– (AGGAGATCTCATTGGC-AAACACTTCGAGC) and G423+ (GCTCGAAGTGTTTGCC-AATGAGATCTCCT), F422*– (GATCTCATTGCCTCACAC-TTCGAGCA) and F422*+ (TGCTCGAAGTGTGAGGCAAT-GAGATA), G423*– (GGAGATCTCATTTCAAAACACTTCG-AG) and G423*+ (CTCGAAGTGTTTTGAAATGAGATCTCC) and N424*– (TCAGGAGATCTCTTAGCCAAACACTTC) and N424*+ (GAAGTGTTTGGCTAAGAGATCTCCTGA). The PCR products were purified and for each point mutation the respective fragments were mixed at equal molar amounts, alkali-denatured and used as a template for a second round of PCR using the primer pair PST/XBA. The reaction products were purified, digested with PstI and XbaI and fused with the PstI/XbaI-digested

original pSG5-VDR construct. The sequence of the entire replaced region was confirmed by sequencing.

The DNA templates for the transcription of C-terminal truncated VDR were generated by PCR using the T7 promoter containing primer VDR1 (TAATACGACTCACTATAGGGCC-ATGGAGGCAATGGCGGCCA) and the primers VDR1269 (GCCAAACACTTCGAGCACAA), VDR1266 (AAACACTT-CGAGCACAAG) and VDR1263 (CACTTCGAGCACAAG-GG), respectively. The PCR profile was 1 min at 94
C, 1 min at 55
C and 2.5 min at 72
C for 40 cycles. Blunt-ended PCR products were generated by incubation with Pfu-polymerase (Stratagene) for 30 min at 72
C and subsequently purified.

The subcloning of the rat atrial natriuretic factor (ANF) VD response element into the XbaI site of pBLCAT2 (22) in front of the thymidine kinase (tk) promoter to drive the expression of the chloramphenicol acetyl transferase (CAT) reporter gene has already been described (23); for the response element core sequence see Figure 5.

Limited protease digestion assay

Linearized cDNA of the wild type VDR and of the six point mutations and the PCR-generated truncated VDR templates were used for in vitro transcription as recommended by the supplier (Promega). Ten micrograms of in vitro transcribed VDR RNA were mixed with 175 µl rabbit reticulocyte lysate (Promega), 100 U RNasin, 20 µl [35_{S]methionine (1000 Ci/mmol) and}

20 µM amino acid mixture (minus methionine) in a total volume of 250 µl and incubated at 30
C for 1 h (total protein

concentration 60 µg/µl). One microlitre of in vitro translated protein, 5.5 µl 50 mM Tris, pH 7.9 and 1 µl ligand were preincubated for 30 min. Then 2.5 µl of trypsin (Promega; cleaves the peptide bond after lysine or arginine) was added to a final concentration of 27 or 50 µg/ml and the mixtures were incubated for 2–30 min at room temperature, as indicated. The digestion reactions were stopped by adding 10 µl protein gel loading buffer (0.25 M Tris, pH 6.8, 20% glycerol, 5% mercaptoethanol, 2% SDS, 0.025% bromophenol blue) and the samples were denatured at 95
C for 5 min, electrophoresed through a 15% SDS-polyacrylamide gel (acrylamide/N,N′-methylene-bisacrylamide weight ratio 33:1) using the Mini-PROTEAN electrophoresis system (Biorad), electrotransferred to a nitrocellulose filter, air-dried and autoradiographed overnight. The protease-sensitive VDR fragments were localized, excised from the filter and radioactivity was measured by scintillation counting.

Transfection and CAT assays

COS-7 (SV40-transformed African Green monkey kidney) cells were seeded into 6-well plates (1–1.5 × 105 _{cells/well) and grown}

overnight in phenol red-free RPMI supplemented with 10% charcoal-treated fetal calf serum (FCS). Liposomes were formed by incubating 1 µg of the reporter plasmid, each 0.3 µg of pSG5-based expression vectors for VDR and RXRα and 1 µg of the reference plasmid pCH110 (Pharmacia) with 15 µg

N-[1-(2,3-dioleoyloxy)propyl]-N,N,N-trimethylammonium

methyl-sulfate (DOTAP, Boehringer Mannheim) for 15 min at room temperature in a total volume of 100 µl. After dilution with 0.9 ml phenol red-free RPMI, the liposomes were added to the cells. Phenol red-free RPMI (500 µl) supplemented with 30% charcoal-treated FCS was added 4–8 h after transfection. At this time 100 nM VD, 1 µM 9-cis RA, the combination of both or solvent

4515

Nucleic Acids Research, 1994, Vol. 22, No. 1Nucleic Acids Research, 1996, Vol. 24, No. 22 4515

was also added. The cells were harvested 40 h after onset of the stimulation and CAT-assays were performed as described (24). The CAT activities were normalized to β-galactosidase activity and induction factors were calculated as the ratio of CAT activity of ligand-stimulated cells to that of mock-induced controls. RESULTS

VDR belongs to those nuclear receptors that, when bound by a ligand, undergo a conformational change that hides one or several cutting sites of the serine endopeptidase trypsin (25,26). In both studies the high amount of 5 µl in vitro translated VDR per assay was used, the trypsin concentrations were 20 and 27 µg/ml and the incubation times were 10 and 15 min, respectively. Figure 1 shows the results of a limited protease digestion assay using in

vitro translated, [35_{S]methionine labelled VDR, which had been}

preincubated with 1 µM VD or MC1288 (20-epi VD). Since we had recently observed (27) that decreasing amounts of VDR improved the resolution of the main protease-resistant fragment (band 1), a variation of the trypsin concentration and incubation time parameters was evaluated with 1 µl in vitro translated VDR. With VD two protease-resistant VDR fragments (bands 1 and 3) could be detected, whereas with MC1288 an additional fragment (band 2) was observed. The intensity of these fragments was found to depend on the assay conditions: trypsin concentration and mainly increasing incubation time. This analysis led to the modification of the former assay conditions (26,27) to 1 µl in

vitro translated VDR per assay, 27 µg/ml trypsin and 30 min incubation time for all subsequent studies.

For the assessment of the next parameter, ligand concentration, limited protease digestion assays were performed in the presence of increasing concentrations of VD and MC1288 (Fig. 2A and B). In confirmation of an earlier report (26) increasing ligand concentrations resulted in a higher amount of ligand-occupied VDR and, therefore, in an increasing total fragment quantity. At

Figure 1. Parameter variation in the limited protease digestion assay. One microlitre of in vitro synthesized [35_{S]methionine-labelled full length VDR was}

preincubated with 1 µM VD or MC1288. Trypsin was added to a final concentration of 27 or 50 µg/ml and the mixtures were incubated for 2–30 min at room temperature. Samples were electrophoresed through a 15% SDS-polyacrylamide gel, electrotransferred to a nitrocellulose filter, air-dried and autoradiographed. Up to three protease-resistant VDR fragments (bands 1–3) were resolved. This series of experiments led to the modification of the limited protease digestion assay to the novel conditions of 1 µl VDR, 27 µg/ml trypsin and 30 min incubation time.

high, saturating ligand concentrations (Fig. 1) all protease-resis-tant fragment quantities showed a comparable intensity, whereas at lower ligand concentrations the relative amount of VDR fragments differed (Fig. 2). With both ligands the intensity of the third VDR fragment (band 3) increased only at a ligand concentration of >100 nM. The second fragment, which was observed only in the presence of MC1288 (Fig. 2B), augmented at a ligand concentration of >1 nM. The first fragment (band 1) was protease-resistant even at a low concentration characteristic for each ligand. All VDR fragments were excised, scintillation counted and normalized for the background of the solvent control. At each concentration the total amount of radioactivity of the protease-resistant fragments were presented in Scatchard plots (Fig. 2C and D). From these plots the functional dissociation constant (Kdf value), i.e. the ligand concentration that provides

50% of protease-resistant VDR fragments, was calculated as 0.56 nM for VD and 3.7 nM for MC1288. It cannot be excluded

Figure 2. Functional ligand binding to wild type VDR. One microlitre of in vitro synthesized [35_{S]methionine-labelled wild type VDR was preincubated with the}

indicated concentrations of VD (A) or MC1288 (B). Trypsin was added to a final concentration of 27 µg/ml and the mixtures were incubated for 30 min at room temperature. Samples were electrophoresed through a 15% SDS-polyacrylamide gel, electrotransferred to a nitrocellulose filter, air-dried and autoradiographed. FL indicates full-length receptor. The resistant VDR fragments (marked 1–3) were cut out from the filter and their radioactivity was measured in a scintillation counter. Scatchard plots of the sum of the two resistant fragments of VD (C) and the three resistant fragments of MC1288 (D) are presented and the respective Kdf values were

Figure 3. Functional ligand binding to truncated VDR. The [35_{S]methionine-labelled truncated receptor proteins VDR (1–423), VDR (1–422) and VDR (1–421) were}

obtained by in vitro transcription/translation of respective PCR-generated DNA templates (A). One microlitre of the three VDR truncations and wild type VDR (1–427) were each preincubated with saturating concentrations of VD and MC1288 (both 10 µM) or with solvent, and limited protease digestion was performed as described in Figure 2. Scatchard plots of assays with VDR (1–423) and VDR (1–422) preincubated with graded concentrations of VD and MC1288 are shown and the respective Kdf values were calculated (B).

that the ligands bind to the receptor without causing functional effects; therefore the Kdf value may not be identical to a Kd value

obtained by traditional ligand binding assays.

In previous reports, where the limited protease digestion assay was applied under conditions that did not allow the separation of multiple protease resistant VDR fragments, the Kdf values for VD

and MC1288 were determined as 0.9 (26) and 0.41 nM (27), respectively. The second VDR fragment that was observed in the modified version of the limited protease digestion assay created a Kdf value for MC1288 that is apparently higher than the recently

reported value. Consequently, the Kdf values for the first and

second VDR fragments, which were obtained by incubation with MC1288, were determined separately as 0.14 and 23 nM, respectively (data not shown). This confirmed the visual impres-sion that the first VDR fragment may represent a high affinity conformation of the receptor.

In order to map this putative high affinity conformation, C-terminal truncated VDR proteins were generated by in vitro transcription/translation of respective PCR-generated templates. VDR proteins, were truncated by the last four (1–423), five (1–422) and six (1–421) amino acids, which were preincubated with VD or MC1288. Figure 3A shows the protease-resistant VDR fragments obtained. VDR (1–423) showed with both ligands the same pattern as the wild type VDR, but the truncation of one further amino acid [VDR (1–422)] provided, even at saturating ligand concentrations, a clear decline of the intensity of the first VDR fragment. Finally, the truncation of a further amino acid [VDR (1–421)] resulted in the nearly complete disappearance of the first VDR fragment. These visual impressions were confirmed by the determination of the respective Kdf values (Fig. 3B). With

VDR (1–423) Kdf values of 3.9 and 4.5 nM were determined for

VD and MC1288, whereas VDR (1–422) provided clearly lower functional affinities of 420 and 34 nM, respectively.

These results suggested that the amino acids glycine 423 and, in particular, phenylalanine 422, play an important role in the high affinity conformation of VDR. By site-directed mutagenesis of

the wild type VDR cDNA the point mutations F422A and F422* were generated [F422* provides the same VDR truncation as VDR (1–421)]. In comparison with wild type VDR, both point mutations were tested at saturating concentrations of VD and MC1288 for their sensitivity to trypsin (Fig. 4A). Both point mutations showed a complete disappearance of the first VDR fragment. The Kdf values of F422A were determined as 250 and

32 nM for VD and MC1288, respectively, and those of F422* were 1200 and 21 nM (Fig. 4B).

In order to test the functionality of the two VDR point mutations F422* and F422A and of four related point mutations, which were V421A, G423A, G423* and N424*, COS-7 cells were transfected with respective VDR expression vector con-structs, an expression vector for RXR and the potent DR3-type VD response element of the rat ANF promoter (23) in a CAT reporter gene construct; subsequently the cells were stimulated with saturating concentration of ligand (Fig. 5). The seven different VDR–RXR heterodimers differed clearly in their response to VD. Wild type VDR mediated a 6.1-fold stimulation of gene activity by VD, which was almost reached only by the mutant G423A (4.9-fold). The two mutants N424* (2.9-fold), G423* (2.2-fold) and F422A (1.9-fold) showed a clearly reduced and the mutants V421A (1.5-fold) and F422* (1.2-fold) no reasonable inducibility by VD. Moreover, the VDR-RXR hetero-dimers showed either an ∼2.3-fold (wild type, F422A, G423A and F422*) or no (V421A, G423* and N424*) inducibility by the RXR ligand 9-cis RA. With the exception of the F422* containing complex, the co-stimulation with both ligands provided for the remaining six heterodimeric complexes a superactivation that was slightly synergistic compared with the inducibility with the individual ligands.

DISCUSSION

The interaction of VD with VDR induces a functionally important conformational change within the receptor and is the central

4517

Nucleic Acids Research, 1994, Vol. 22, No. 1Nucleic Acids Research, 1996, Vol. 24, No. 22 4517

Figure 4. Functional ligand binding to point-mutated VDR. The [35_{S]methionine-labelled point-mutated receptor proteins VDR F422A and VDR F422* were obtained}

by in vitro transcription/translation of templates generated by site-directed mutagenesis of wild type receptor (VDR wt) (A). They were preincubated with saturating concentrations of VD and MC1288 (both 10 µM) or solvent, and limited protease digestion was performed as described in Figure 2. Scatchard plots of assays with VDR F422A and VDR F422* preincubated with graded concentrations of VD and MC1288 are shown and the respective Kdf values were calculated (B).

Figure 5. Functional activity of point-mutated VDR. COS-7 cells were transfected with CAT reporter constructs containing the DR3-type rat ANF VD response element (its core sequence is given below) fused to the tk promoter and expression vectors for human RXRα and the indicated VDR construct. The cells were treated with either 100 nM of VD, 1 µM 9-cis RA alone or in combination or with solvent. CAT activities were determined 40 h later and fold induction was calculated in comparison with solvent-induced controls. Each column represents the mean of triplicates; the bars indicate standard deviations.

molecular mechanism of nuclear VD signalling. The fact that VDR undergoes a conformational change could already be deduced from limited protease digestion studies with other nuclear receptors (17–20) and was recently experimentally demonstrated for the first time (25,26). The limited protease digestion assay enables the direct visualization and quantification of VDR-VD analogue interactions and has, therefore, a great advantage compared with the traditional indirect measurement by competition studies. It has been shown that biologically potent VD analogues have a higher functional affinity to VDR than the natural hormone (27). In contrast, a comparable competition study provided misleading results (25).

In this report, the conditions of the limited protease digestion assay have been modified in order to demonstrate the existence of at least three different VDR conformations. It is obvious that each of the three VDR fragments that were obtained by trypsin digestion contained a large proportion of the C-terminally located LBD. They were referred to as functional LBD cores. In order to determine the borderlines of these LBD cores, N- and C-terminal truncations of the VDR have been performed. Preliminary results from N-terminal VDR truncations indicate that the functional LBD core of all three protease-resistant fragments starts around the amino acid position 130 (S.N. and C.C., unpublished results). This region belongs to the flexible hinge region between the LBD and the DNA binding domain. It can therefore be assumed that the estimated size differences between the trypsin-resistant VDR fragments of ∼2 (band 1 to band 2) and ∼4 kDa (band 1 to band 3), which corresponds to 18 and 36 amino acids, are mainly related to a different length of their C-terminus.

One main result of this report is that the three functional LBD cores clearly differ in their Kdf values. As expected, the smallest

LBD core showed only very low functional affinity for ligand (Kdf

> 100 nM). Assessments of physiological importance of this low affinity VDR conformation are currently in progress. The LBD core of intermediate size was only observed with the synthetic analogue 20-epi VD (MC1288), but not with the natural ligand VD. However, studies with other VD analogues showed that some, but not all, 20-epi VD analogues (27) and, interestingly, also 20-methyl VD analogues (28) induce, or at least recognize, this VDR conformation. Moreover, under defined conditions the limited protease digestion assay allows the observation of all three VDR conformations even in the absence of ligand (S.N. and C.C., unpublished results). This suggests that the second VDR con-formation is not only artificially induced by a synthetic ligand, but may also have a natural role like, e.g., the contact to some VD metabolites. In the case of MC1288 the Kdf value of the second

VDR conformation has been determined to be in the order of 20 nM. This explains why the C-terminal truncations VDR (1–422) and VDR F422*, and the point mutation VDR F422A, that

all suppress the formation of the first, high affinity ligand binding conformation, show only very low affinity for VD (Kdf values

between 250 and 1200 nM), but still a relatively high affinity for MC1288 (Kdf values between 21 and 34 nM).

This study has shown that the modification of the conditions of the limited protease digestion assay provides an apparently higher total Kdf value than the previous conditions (26,27) for those VD

analogues that also induce the second VDR conformation. This problem can be solved by the individual analysis of the different conformations. It is obvious that the functional characterization of VD analogues is of immense importance for the application of VD signalling in clinical therapy.

Further investigations, in particular on the physiological import-ance of the second, medium affinity conformation of VDR, are in progress. It is known that not only the ligand, but also the DNA (the response element) and proteins (mainly the heterodimeric partner) influence the conformation of VDR. It can be assumed that in the complex in vivo situation VDR forms more than just the three in

vitro observed conformations (12,15). Therefore, it is likely that those VD analogues, which, like MC1288, bind in vitro to a third VDR conformation, also have more possibilities of interactions with VDR in vivo compared with the natural ligand VD. This may result in an overall greater functional potential of such analogues, but maybe also in a decreased specificity.

The fine mapping of the first VDR conformation with the limited protease digestion assay showed that the last four C-terminal amino acids of the receptor do not significantly contribute to ligand binding. But the truncation of glycine 423 diminished and the further truncation of phenylalanine 422 totally abolished VDR’s high ligand binding conformation. In the functional reporter gene assay the mutations F422A and V421A, but not the mutation G423A, decreased the ligand inducibility of VDR by a factor of >5. Moreover, the truncation of the last four and five amino acids (N424* and G423*) already clearly reduced and the truncation of the last six amino acids (F422*) completely abolished the inducibility of VDR by VD in the functional assay. This suggests that amino acid F422 is of central importance for the high affinity ligand binding conformation of VDR and that this amino acid may directly contact the ligand. Moreover, the point mutation V421A indicated that valine 421 is important for the functionality of the VDR, but since by the truncation F422* VDR already lost high affinity VD binding and VD inducibility, it is more likely that this amino acid is only involved in the transactivation process by directly contacting co-factors. In analogy to retinoid receptors (7) both V421 and F422 belong to the AF-2 domain of VDR. This indicates that high affinity ligand binding and the contact to co-factors appears to be mediated by the same (F422) or at least neighboured (F422 and V421) amino acids. The central importance of amino acid F422 in both high affinity ligand binding and transactivation is emphasized by the observation that, of the six tested VDR mutations, F422* was the only one that completely lost inducibility by VD alone and also showed no synergistic effect in co-stimulation with both ligands. A similar example has recently been reported for 9-cis RA and RARα (20,29): isoleucine 410 of human RARα, which belongs

to the RARα’s AF-2 domain, was found to be critical for the high affinity binding of the pan-agonist 9-cis RA, but, interestingly, not for the binding of all-trans RA. Both findings suggest the attractive principle that high affinity ligand binding and transac-tivation may be mediated by only one amino acid or a small

distinct region of only a few amino acids. It is likely that this principle is true for a subgroup of the nuclear receptor superfamily that probably contains the more condensed members of the superfamily. Further investigations with other nuclear receptors, but most importantly a crystal structure of ligand-bound VDR are necessary in order to confirm this hypothesis.

ACKNOWLEDGEMENTS

We would like to thank J.-H. Saurat for discussions, L. Binderup for MC1288 and M. Schräder and C. Danielsson for critical reading of the manuscript. This work was supported by the Swiss Cancer League and the LEO Research Foundation.

REFERENCES

1 Walters, M. R. (1992) Endocr. Rev., 13, 719–764.

2 Mangelsdorf, D. J., Thummel, C., Beato, M., Herrlich, P., Schütz, G., Umesono, K., Blumberg, B., Kastner, P., Mark, M., Chambon, P. and Evans, R. M. (1995) Cell, 83, 835–839.

3 Bourguet, W., Ruff, M., Chambon, P., Gronemeyer, H. and Moras, D. (1995) Nature, 375, 377–382.

4 Renaud, J.-P., Rochel, N., Ruff, M., Vivat, V., Chambon, P., Gronemeyer, H. and Moras, D. (1995) Nature, 378, 681–689.

5 Wagner, R. L., Apriletti, J. W., McGrath, M. E., West, B. L., Baxter, J. D. and Fletterick, R. J. (1995) Nature, 378, 690–697.

6 Danielian, P. S., White, R., Lees, J. A. and Parker, M. G. (1992) EMBO J., 11, 1025–1033.

7 Durand, B., Saunders, M., Gaudon, C., Roy, B., Losson, R. and Chambon, P. (1994) EMBO J., 13, 5370–5382.

8 Cavaillès, V., Dauvois, S., L’Horset, F., Lopez, G., Hoare, S., Kushner, P. J. and Parker, M. G. (1995) EMBO J., 14, 3741–3751.

9 Le Dourain, B., Zechel, C., Garnier, J.-M., Lutz, Y., Tora, L., Pierrat, B., Heery, D., Gronemeyer, H., Chambon, H., Chambon, P. and Losson, R. (1995) EMBO J., 14, 2020–2033.

10 vom Baur, E., Zechel, C., Heery, D., Heine, M. J. S., Garnier, J. M., Vivat, V., Le Dourain, B., Gronemeyer, H., Chambon, P. and Losson, R. (1996)

EMBO J., 15, 110–124.

11 Yu, V. C., Näär, A. M. and Rosenfeld, M. G. (1992) Curr. Opin. Biotech., 3, 597–602.

12 Carlberg, C. (1995) Eur. J. Biochem., 231, 517–527.

13 Carlberg, C., Bendik, I., Wyss, A., Meier, E., Sturzenbecker, L. J., Grippo, J. F. and Hunziker, W. (1993) Nature, 361, 657–660.

14 Schräder, M., Müller, K. M. and Carlberg, C. (1994) J. Biol. Chem., 269, 5501–5504.

15 Carlberg, C. (1996) Endocrine, 4, 91–105.

16 Breen, E. C., van Wijen, A. C., Lian, J. B., Stein, G. S. and Stein, J. L. (1994) Proc. Natl. Acad. Sci. USA, 91, 12902–12906.

17 Allan, G. F., Leng, X., Tsai, S. Y., Weigel, N. L., Edwards, D. P., Tsai, M.-J. and O’Malley, B. W. (1992) J. Biol. Chem., 267, 19513–19520. 18 Leid, M. (1994) J. Biol. Chem., 269, 14175–14181.

19 Keidel, S., LeMotte, P. and Apfel, C. (1994) Mol. Cell. Biol., 14, 287–298. 20 Tate, B. F., Allenby, G., Janocha, R., Kazmer, S., Speck, J., Sturzenbecker, L. J., Abarzua, P., Levin, A. A. and Grippo, J. F. (1994) Mol. Cell. Biol., 14, 2323–2330.

21 Leng, X., Tsai, S., O’Malley, B. W. and Tsai, M.-J. (1993) J. Steroid

Biochem. Molec. Biol., 46, 643–661.

22 Luckow, B. and Schütz, G. (1987) Nucleic Acids Res., 15, 5490. 23 Kahlen, J.-P. and Carlberg, C. (1996) Biochem. Biophys. Res. Commun.,

218, 882–886.

24 Pothier, F., Ouellet, M., Julien, J.-P. and Guérin, S. L. (1992) DNA Cell

Biol., 11, 83–90.

25 Peleg, S., Sastry, M., Collins, E. D., Bishop, J. E. and Norman, A. W. (1995) J. Biol. Chem., 270, 10551–10558.

26 Nayeri, S., Danielsson, C., Kahlen, J.-P., Schräder, M., Mathiasen, I. S., Binderup, L. and Carlberg, C. (1995) Oncogene, 11, 1853–1858. 27 Nayeri, S., Mathiasen, I. S., Binderup, L. and Carlberg, C. (1996) J. Cell.

Biochem., 62, 325–333.

28 Danielsson, C., Nayeri, S., Wiesinger, H., Thieroff-Ekerdt, R. and Carlberg, C. (1996) J. Cell. Biochem., 63, in press.