A Rationale for Treatment of Hereditary Vitamin D-resistant Rickets with Analogs of 1α,25-Dihydroxyvitamin D3

Hereditary vitamin D-resistant rickets (HVDRR) is caused by heterogeneous inactivating mutations in the vitamin D receptor (VDR). Treatment of HVDRR patients with high doses of oral calcium and supraphysiologic doses of 1α,25-dihydroxyvitamin D3 (1,25D3) has had limited success. In this study we explored the use of vitamin D analogs as a potential therapy for this disorder. The rationale for the use of vitamin D analogs is that they bind the VDR at different amino acid residues than 1,25D3, and their ability to modulate VDR functions differs from that of the natural hormone. In this report, we examined the VDR from three HVDRR patients with mutations in the ligand-binding domain of the VDR (histidine 305 to glutamine, arginine 274 to leucine, and phenylalanine 251 to cysteine) for their responses to two vitamin D analogs, 20-epi-1,25D3 and 1β-hydroxymethyl-3-epi-16-ene-26a,27a-bishomo-25D3(JK-1626-2). Our results reveal that vitamin D analogs partially or completely restore the responsiveness of the mutated VDR. Analog treatment seemed to be more successful when the mutation affects the amino acids directly involved in ligand binding rather than amino acids that contribute to a functional VDR interface with dimerization partners or coactivators of transcription.

Hereditary vitamin D-resistant rickets (HVDRR) 1 is an autosomal recessive disorder characterized by end-organ resistance to 1␣,25-dihydroxyvitamin D 3 (1,25D 3 ) (1,2). Clinically, the syndrome is recognized by severe early onset rickets with bowing of the lower extremities, short stature, and often alopecia (2)(3)(4)(5). Resistance to 1,25D 3 in HVDRR leads to impaired intestinal calcium absorption. This results in a series of metabolic abnormalities including frank hypocalcemia, secondary hyperparathyroidism, elevated alkaline phosphatase levels, hypophosphatemia, and markedly increased 1,25D 3 levels (2). Treatment of HVDRR patients with high doses of oral calcium and supraphysiologic doses of 1,25D 3 has had limited success. More aggressive therapy to bypass the defect in intestinal calcium absorption is long term intravenous infusion of calcium that restores the serum calcium levels to normal and reverses the rickets in some cases (2,(5)(6)(7). However, this approach has the usual complications associated with intravenous therapy, and its use has been limited because of its complexity.
The resistance to 1,25D 3 in HVDRR is caused by heterogeneous mutations in the nuclear receptor for vitamin D (2,8,9). Therefore, this disease emphasizes the importance of vitamin D receptor (VDR)-mediated action of 1,25D 3 in skeletal development and bone mineralization as well as in the development of hair follicles. Like other members of the nuclear receptor superfamily, the VDR is activated by binding to its ligand (1,25D 3 ), and its action is mediated through distinct functional domains for DNA binding, ligand binding, and transcriptional activation (10,11). HVDRR patients with mutations in the DNA binding domain of VDR do not respond to 1,25D 3 treatment mainly because of structural disruption of VDR binding to its DNA response elements (VDREs) with a total loss of transcriptional regulation (2). Conversely, some patients harboring mutations in the ligand-binding domain (LBD) have been partially responsive to high doses of calcium and 1,25D 3 , presumably because the excess active metabolite was able to override the insensitivity caused by the ligand-binding defect (2).
In the present study we considered vitamin D analogs as an alternative treatment to large doses of 1,25D 3 for the following reasons. First, some analogs of vitamin D are significantly more potent transcriptionally than the natural hormone. This enhanced transcriptional potency is proposed to be mediated through differential modulation of VDR functions (11,12). For example, side chain-modified analogs such as 20-epi-1,25D 3 (20E-1,25D 3 ) have 100 -1000-fold greater transcriptional potency than 1,25D 3 , and this is associated with enhanced potency to heterodimerize with the retinoid X receptor (RXR) and to interact with the transcription coactivator vitamin D receptor interacting protein 205 (12)(13)(14). The proposed mechanism for this enhanced potency is that in which the analog changes the conformation of the VDR to provide a more effective interface for interaction with RXR and transcription coactivators. Second, even analogs that are less potent transcriptionally than 1,25D 3 may contact different amino acid residues in the ligand-binding pocket such that they have the potential to bind and activate VDRs with mutations at the usual contact points for the hormone (15,16).
To determine whether analogs can bind and restore transcriptional activity of VDR in HVDRR patients, we selected three HVDRR mutations that were restricted to the LBD of the * This work was supported by National Institutes of Health Grants DK 42482 (to D. F.), CA 44530 (to G. H. P), and DK 50583 (to S. P.). The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

1,25D
Site-directed Mutagenesis-To create the desired mutations in the VDR we used the Quick Change site-directed mutagenesis kit (Stratagene, La Jolla, CA) or the Gene Editor kit (Promega, Madison, WI). Sequence changes were confirmed by DNA sequencing.
Receptor Binding and Competition Assays-To assess the relative affinities of 1,25D 3 and the analogs for WT and mutated VDR in vitro, whole-cell homogenates from COS-1 cells transfected with VDR expression plasmids were prepared in KTED (10 mM Tris-HCl, pH 7.4, 1.5 mM EDTA, 0.3 M KCl, and 1 mM dithiothreitol) as described previously (15). The homogenates were then aliquoted into tubes containing 0.2 pmol of [ 3 H]1,25D 3 and increasing concentrations of nonradioactive ligand. The mixtures were incubated on ice for 3-4 h, after which free ligand was separated from bound by hydroxyapatite. The bound ligand was released from the hydroxyapatite by ethanol extraction, and the radioactivity was measured by scintillation counting. The results of the competition assays were plotted as the inverse value of the percentage of maximal binding against competitor concentration by the method of Wecksler and Norman (25).
Transfections and Transcriptional Assays-CV-1 monkey kidney cells, were plated in 35-mm dishes at a density of 3 ϫ 10 5 cells/dish. Cells were transfected by the DEAE-dextran method with 2 g/dish of WT or mutant VDR expression vectors and a reporter construct containing the human osteocalcin VDRE (ocVDRE) linked to the thymidine kinase promoter and the growth hormone reporter gene (ocVDRE/TK-GH) (12). In some experiments a reporter gene containing the rat 24-hydroxylase promoter ligated to the luciferase gene (2 g/dish) was used (26). The medium was collected 2 days after transfection, and growth hormone was measured using a radioimmunoassay as described by the manufacturer (Nichols Institute, San Juan Capistrano, CA). For measurements of luciferase activity, the transfected cells were lysed 2 days after transfection using passive lysis buffer (Promega) and assayed using the luciferase assay reagent according to manufacturer instructions (Promega). Light units were measured with a TD 20/20 luminometer (Turner Designs, Sunnyvale, CA) and normalized for protein concentration in each cell lysate (19).
Northern Blot Analyses-Skin fibroblasts from normal subjects or from HVDRR patients were grown to confluence and then treated with 1,25D 3 or with an analog for 6 h in medium containing 1% fetal bovine serum. RNA was prepared using Trizol reagent (Life Technologies, Inc.). Total RNA (5 g) was separated by formaldehyde-agarose gel electrophoresis, transferred to Hybond membranes (Amersham Pharmacia Biotech), and immobilized on the membranes by ultraviolet irradiation. The membranes were then hybridized with 24-hydroxylase and L7 ribosomal RNA probes labeled by the random primer method (27). L7 has been shown in multiple experiments to be unaffected by 1,25D 3 treatment and therefore was used as a control for loading and transfer.
Protease Sensitivity Assays-WT and mutant VDRs were synthesized and labeled in vitro with [ 35 S]methionine (1000 Ci/mmol) using the TNT coupled transcription/translation system (Promega). The translated receptor preparations were incubated with 1,25D 3 or analogs for 10 min at ambient temperature. Next, trypsin was added to a concentration of 20 g/ml, and the mixtures were incubated for 10 min. The digestion products were then separated by SDS polyacrylamide gel electrophore-sis and detected by autoradiography.
Glutathione S-transferase (GST) Pull-down Assays-To determine the potency of ligands to induce interaction of VDR with RXR␣ or steroid receptor coactivator-1 (SRC-1) we used GST pull-down assays (28). A binding reaction containing 11 l of PBSDP buffer (phosphatebuffered saline, 1 mM dithiothreitol, and 10 mM phenylmethylsulfonyl fluoride), 3 l of 35 S-labeled VDR, and 1 l of ethanol or ligand (in ethanol) were incubated at ambient temperature for 10 min. Then, 3-5 g of purified GST fusion protein (either GST-RXR␣ or GST-SRC-1) and 20 l of glutathione-Sepharose beads (equilibrated in PBSDP buffer) were added, and the volume was brought up to 100 l with NETND buffer (20 mM Tris-HC, pH 7.8, 100 mM NaCl, 1 mM EDTA, 0.2% Nonidet P-40, and 1 mM dithiothreitol). The mixtures were incubated at 4°C for 1 h, and then the beads were washed once with NETND and twice with PBSDP. The bound proteins were eluted from the beads by boiling in Laemmli buffer for 3 min and analyzed by SDS-polyacrylamide gel electrophoresis and autoradiography.

RESULTS
Restoring the Transcriptional Activities of the HVDRR Mutant H305Q by Using a Side Chain-modified Analog-The H305Q mutation in the VDR was reported as the cause of HVDRR in two siblings of Turkish origin (18,29). Histidine 305 is a proposed contact point for the 25-OH group on the side chain of 1,25D 3 (21). We used the analog 20E-1,25D 3 ( Fig. 1) to determine whether it could effect the transcriptional activity of this mutant VDR for two reasons. First, 20E-1,25D 3 is 100 times more potent than 1,25D 3 in transactivation assays (12). Second, 20E-1,25D 3 has a side chain modification (a 20-epi stereochemistry) that would be expected to position the 25-OH group in contact with amino acids other than the glutamine 305 residue in the mutant VDR. In [ 3 H]1,25D 3 competition assays, the average ED 50 for binding to the WT VDR by 1,25D 3 or by 20E-1,25D 3 was 0.5 and 0.3 nM, respectively. There was only a slight decrease in affinity of the natural hormone (ED 50 ϭ 1 nM) and no decrease in the affinity of 20E-1,25D 3 for the H305Q mutant. This was consistent with the reported findings in which only a moderate binding deficiency was found in cultured fibroblasts from these patients (18). Synthetic Analogs Restore Defective Vitamin D Receptor Action through the H305Q mutant was decreased only 4-fold compared with the WT VDR (ED 50 of 0.008 nM for the WT and 0.03 nM for the mutant). Therefore, although the potency of the analog was ϳ100-fold greater than 1,25D 3 in WT VDR, the potency of the analog to induce transcription through the H305Q VDR was ϳ1000-fold greater than that of the natural hormone.
We next examined the potency of the 20E-1,25D 3 analog to induce 24-hydroxylase gene transcription in the fibroblasts of the patient. As shown in Fig. 2B, in normal fibroblasts 24hydroxylase mRNA expression was induced at concentrations as low as 0.01 nM for 20E-1,25D 3 compared with 1 nM for 1,25D 3 . In cells from a patient with the H305Q mutation, weak induction of 24-hydroxylase mRNA was detected only at concentrations as high as 100 nM 1,25D 3 . In contrast, induction of 24-hydroxylase mRNA was detected at concentrations as low as 0.1 nM 20E-1,25D 3 . These results show that the transcriptional activity of the 20E-1,25D 3 analog is at least a 100-fold greater than that of the natural hormone in the fibroblasts from the H305Q patient. The 10-fold difference observed in the transactivation assays in CV-1 cells and the mRNA induction in the cells of the patient may be caused by a greater sensitivity of the reconstituted system or to differences in half-life of the gene products in the two assay systems.
Different Potencies of 1,25D 3 and 20E-1,25D 3 to Induce a Transcriptionally Active Conformation in the H305Q Mutant VDR-Because the loss of 1,25D 3 -mediated transcriptional activity by the H305Q mutant was not proportional to the apparent reduction in affinity for 1,25D 3 measured by the competi-tion assays, we considered the possibility that another aspect of ligand interaction with this mutant VDR was altered. We (16) and others (30) have shown that the transactivation potency of an analog of 1,25D 3 may be determined by its ability to induce VDR to assume a protease-resistant conformation. To examine whether this aspect of ligand interaction with the receptor indeed was impaired by the H305Q mutation, we performed quantitative protease sensitivity assays. As shown in Fig. 3A, 1,25D 3 stabilized the conformation of WT VDR at concentrations as low as 0.1 nM, the principal protease-resistant fragment being 34 kDa. In contrast, the potency of 1,25D 3 to stabilize the conformation of H305Q was significantly lower such that only at 10 nM was a detectable amount of protease-resistant product observed, and the main stabilized fragment was 28 kDa. When the 20E-1,25D 3 analog was used as the ligand it was much better at stabilizing the H305Q mutant VDR than 1,25D 3 . Furthermore, there was no difference in its potency to stabilize the conformation of the mutant VDR compared with the WT VDR, although the size of the main trypsin-resistant fragment was 28 kDa for the mutant and 34 kDa for the WT VDR (Fig. 3A).
The protease sensitivity assays demonstrated that there were qualitative changes in the mode of 1,25D 3 and 20E-1,25D 3 interaction with the H305Q mutant. These changes suggested that the conformations induced by these ligands were different in the WT and mutant VDRs, although the potency of the 20E-1,25D 3 analog to stabilize the mutant VDR against trypsin digestion was similar to that for WT VDR. Therefore, we next wanted to determine whether this apparent change in confor- The results are expressed as percentages of the maximal 1,25D 3 -induced reporter gene expression in the WT VDR-transfected cells. Each transfection experiment was performed in duplicate, and each titration was performed 3-6 times. In typical experiments, maximal 1,25D 3 -induced reporter gene expression was 10 -15-fold over vehicle-treated controls. B, induction of 24-hydroxylase (24OHase) mRNA in normal fibroblasts and fibroblasts from an HVDRR patient with the H305Q mutant VDR was assessed by Northern blot. The cells were treated with the indicated doses of 1,25D 3 or 20E-1,25D 3 (shown above the panels). Expression of the L7 mRNA served as control. mation compromised the ability or the potency of the analog to induce interaction of the mutant VDR with transcription coactivators. To elucidate this point we performed quantitative pull-down assays using in vitro translated VDR and the nuclear receptor-interacting domain of the SRC-1 fused to GST-SRC-1 (28). As shown in Fig. 3B, the ED 50 for 1,25D 3 to induce interaction with GST-SRC-1 was 0.2 nM for the WT VDR and 30 nM for the H305Q mutant. In contrast, the ED 50 for the analog was 0.3 nM for both the WT VDR and the H305Q mutant, although maximal ligand-induced interaction of H305Q with GST-SRC-1 did not reach the levels of WT VDR. These results suggest that both ligands stabilized a conformation of the mutant VDR in which the SRC-1-interacting interface of the VDR is efficiently exposed. The greater potency of the analog to stabilize H305Q VDR conformation and to induce SRC-1 interaction with the mutant VDR correlates well with its greater transcriptional potency in CV-1 cells and in the fibroblasts of the patient.
Restoring the Transcriptional Activities of the HVDRR Mutant R274L by Using an A Ring-modified Analog-The R274L mutation in the VDR was identified in a severely debilitated child who later died of disease-related complications (17). Arginine 274 is a contact point for the 1␣-OH group of 1,25D 3 (21). The original findings on this mutant VDR showed little detectable 1,25D 3 binding, and it was transcriptionally active only at micromolar concentrations of 1,25D 3 (17). We hypothesized that a transcriptionally potent analog with a modification at the 1␣ position may be able to overcome the ligand-binding and transactivation defects of the R274L VDR. We have shown that the analog JK-1626-2 ( Fig. 1) is as transcriptionally potent as 1,25D 3 , although it has two modifications at the A ring (1␤hydroxymethyl and 3␣-hydroxyl groups) (16). Therefore, we considered this analog to be a candidate to restore the transcriptional activity of the R274L mutant VDR. To this end we measured transcriptional activities in CV-1 cells cotransfected with expression vectors bearing the VDR and the ocVDRE-GH reporter gene. As shown in Fig. 4, A and B, the R274L mutation diminished 1,25D 3 -mediated transcription such that it was only active at 1 M 1,25D 3 . In contrast, the analog JK-1626-2 exhibited a transcriptional potency of ED 50 ϭ 2 nM with the WT VDR and ED 50 ϭ 10 nM with the R274L mutant VDR.
Because fibroblasts from the patient with the R274L mutation were unavailable, we performed additional transcriptional assays with CV-1 cells using a 24-hydoxylase promoter-luciferase reporter construct (26). Fig. 4C shows the potencies of 1,25D 3 (ED 50 ϭ 1 nM) and JK-1626-2 (ED 50 ϭ 5 nM) to induce transcription from this promoter. The potency of 1,25D 3 to induce transcription through the 24-hydroxylase promoter was diminished by the R274L mutation (i.e. detectable only at 1 M), results which are similar to those obtained with the ocV-DRE-GH reporter. However, the analog JK-1626-2 exhibited significant transcriptional activity at 100 nM and reached a maximal level of WT VDR-mediated transcription at 1 M (Fig.  4D). These results support our hypothesis that the A ringmodified analog would transactivate the R274L mutant more effectively than the natural hormone despite having a transcriptional potency very similar to 1,25D 3 in the WT VDR. However, the loss (20 -40-fold) of the ability of the analog to induce transcription through the 24-hydroxylase promoter was significantly greater than its loss (5-fold) of ability to induce transcription through the ocVDRE-TK promoter. These results suggest that R274L-analog complexes have a preference for the ocVDRE over the 24-hydroxylase VDREs. Alternatively, it is possible that the mutant VDR-analog complexes interact more effectively with the basal transcription apparatus of the thymidine kinase promoter than with the natural promoter of the 24-hydroxylase gene.
Different Potencies of 1,25D 3 and JK-1626-2 to Stabilize the Conformation of the R274L VDR-We wished to determine whether the greater potency of the JK-1626-2 analog to trans- FIG. 3. Ligand potencies to stabilize WT VDR or H305Q VDR in a proteaseresistant conformation and to induce binding to SRC-1. A, protease sensitivity assays. In vitro translated 35 S-labeled WT or H305Q VDRs were incubated with or without 1,25D 3 and the 20E-1,25D 3 analog before digestion with trypsin. The protease-resistant fragments were separated by SDS-polyacrylamide gel electrophoresis and detected by autoradiography. B, GST pull-down assays contained 35 S-labeled VDR, 1,25D 3 , or the 20E-1,25D 3 analog, GST-SRC-1, and glutathione-Sepharose beads. The bound proteins were eluted from the packed beads, analyzed by SDS-polyacrylamide gel electrophoresis and autoradiography, and quantified by densitometric scanning. The results were expressed as a percentage of maximal 1,25D 3 -induced binding of 35 Slabeled WT VDR to GST-SRC-1.
activate the R274L mutant VDR was indeed because of better receptor binding. The [ 3 H]1,25D 3 binding activity of R274L is poor, and therefore competition assays with the analog cannot be used to assess its relative affinity for this mutant VDR. Because radiolabeled JK-1626-2 is not available for direct binding analysis, we used the quantitative protease sensitivity assay to assess the ability of the analog to bind VDR. We rationalized that although the inability of a ligand to stabilize receptor conformation does not necessarily reflect lack of affinity for VDR, the ability to stabilize conformation does reflect effective binding to VDR. Fig. 5 shows that the analog stabilized the conformation of WT VDR at a concentration as low as 0.1 nM and induced a proteolytic pattern identical to that induced by 1,25D 3 . The R274L mutation diminished the ability of 1,25D 3 to stabilize the VDR conformation such that only weak ligand-mediated resistance to trypsin was detected at 1 M. In contrast, JK-1626-2 maintained a significant ability to stabilize VDR conformation even at 100 nM. These results suggest that the binding of JK-1626-2 to R274L VDR is at least 100-fold greater than that of 1,25D 3 , although the mutation probably somewhat decreased the affinity of the analog for the mutant VDR.
We also examined the relationship between stabilization of R274L VDR conformation and the availability of a functional interface for interaction with coactivators using GST-pull down assays (data not shown). We found that 1,25D 3 and JK-1626-2 had similar potencies to induce interaction of the WT VDR with GST-SRC-1 (ED 50 ϭ 0.5-1 nM). Although 1,25D 3 could not induce interaction of the R274L mutant with SRC-1 even at 1 M 1,25D 3 , the analog maintained a significant ability to induce interaction with GST-SRC-1, which was evident at 10 nM. However, the efficacy of the JK-1626-2 to induce this interaction with the R274L mutant was lower than with the WT VDR.
Failure to Restore the E1 Domain-dependent Functions of the HVDRR Mutant F251C-We have described recently an HVDRR patient with an F251C mutation who was unrespon-sive to treatment with large doses of 1,25D 3 (19,20). Phenylalanine 251 is in the E1 domain of VDR in a loop connecting helices 3 and 4 (21). This domain has been implicated in regulating RXR heterodimerization and coactivator interaction but not in ligand binding (23). To determine whether a functional mutation within the LBD but not at a ligand contact point can be alleviated by an analog of 1,25D 3 , we used the analog 20E-1,25D 3 . We examined the transcriptional activity of the F251C mutant VDR in CV-1 cells using both the ocVDRE-GH and the 24-hydroxylase-luciferase reporters (Fig. 6A). Neither 1,25D 3 nor 20E-1,25D 3 were able to restore transcriptional activity to the F251C mutant VDR using the ocVDRE reporter gene even at concentrations as high as 1 M. However, the 20E-1,25D 3 analog was able to partially restore some activity through the 24-hydroxylase promoter at concentrations of 100 -1000 nM. No induction of transcription by 1,25D 3 was observed at these concentrations, although transactivation of the 24-hydroxylase promoter by 1000 nM of 1,25D 3 could be detected in COS-7 cells (19). In cells from the patient with the F251C mutation, there was no detectable induction of 24-hydroxylase mRNA by 1,25D 3 at concentrations up to 100 nM, but there was a significant induction of this mRNA at concentrations of 10 -100 nM of the 20E-1,25D 3 analog (Fig. 6B). It should be noted that a weak stimulation of 24-hydroxylase mRNA could be detected in the cells of the patient by 1000 nM 1,25D 3 (19).
We next determined whether the loss of transcriptional activity of the F251C mutant is associated with impairment of ligand-receptor interaction. Competition assays performed at 4°C using homogenates from transfected COS-1 cells revealed that the affinities of 1,25D 3 for F251C and WT VDR were similar (Fig. 7A), although maximal binding activity was lower (25-50% of wild-type VDR, data not shown). Likewise, protease sensitivity assays showed that there was no change in the potency of 1,25D 3 to stabilize the conformation of the mutant VDR. Furthermore, the conformation of the mutant VDR, as judged by this assay, was not different from that of WT VDR. However, maximal intensity of the 34-kDa ligand-dependent trypsin-resistant fragment generated from F251C was 50% of that generated from WT VDR (Fig. 7B). These results confirmed earlier functional studies (22) showing that the E1 domain does not contribute to ligand-binding affinity even though this domain resides in the LBD. Furthermore, our protease sensitivity assays suggested that phenylalanine 251 does not contribute significantly to the conformation of the ligand-binding domain, although the F251C mutation does have some effect on the stability of the VDR.
To determine whether loss of transcriptional activity of the F251C receptor is caused by a loss of ability to dimerize and/or to interact with transcription coactivators, we compared the abilities of F251C VDR and WT VDR to interact with GST-RXR␣ or GST-SRC-1 in a ligand-dependent manner. As shown in Fig. 7C, there was a significant decrease in the ability of F251C VDR to interact with the LBD of RXR␣ in a ligand-dependent or independent fashion. Although the potencies of 20E-1,25D 3 analogs to induce RXR dimerization with WT VDR have been reported to be significantly greater than that of the natural hormone (12,28,31), the F251C mutant had the same dimerization ability whether it was bound to 1,25D 3 or to 20E-1,25D 3 . Furthermore, neither 1,25D 3 nor 20E-1,25D 3 were able to significantly induce interaction of F251C VDR with GST-SRC-1. Taken together these results suggest that the F251C mutation diminishes the ability of VDR to interact with RXR␣ and with the transcription coactivators from the p160 family and that this probably causes the loss of transcriptional activity. That the 20E-1,25D 3 analog had some ability to induce transcriptional activity of the F251C VDR through the 24hydroxylase promoter in the cells of the patient and in the transfected CV-1 cells is of interest. The findings suggest that VDR-mediated transcription of the 24-hydroxylase gene may be modulated by coactivators that are different from the p160 family and that they may be preferentially recruited by the 20E-1,25D 3 analog (14,32).

Comparing the Preferences of Side Chain-modified and A Ring-modified Analogs for the HVDRR Mutants-
The rationale we used to restore the transcriptional activities the HVDRR mutants R274L and H305Q by vitamin D analogs seems to be supported by the results of the experiments described above. However, it is necessary to determine whether the analogs restored these activities merely because of their increased potency or because they specifically bypassed the defective amino acid residue in the mutant VDR. To that end, we performed experiments to examine the ability of 20E-1,25D 3 to restore the binding and transcriptional activity of R274L and the ability of the A ring-modified analog JK-1626-2 to bind and restore transcription of the HVDRR mutant H305Q. We also tested whether Ro25-5318, another side chain-modified analog, could better restore functional activity to a mutated 25-OH contact point than to a 1-OH contact point. Fig. 8 summarizes our binding analysis using the protease sensitivity assay as a parameter for ligand ability to bind and induce a functional conformation of the VDR. These experiments show that the side chain-modified analog 20E-1,25D 3 had a greater binding preference for the H305Q mutant than for the R274L mutant. The other side chain-modified analog, Ro25-5318, also had a binding preference for the H305Q. Furthermore, Ro25-5318 was able to stabilize a wild-type protease sensitivity profile of the H305Q VDR, suggesting it restored the analog-occupied mutant VDR to a WT conformation. In con- trast, the A ring-modified analog JK-1626-2 had a strong binding preference for the R274L mutant, whereas it bound poorly to the H305Q mutant.
To further examine the relationship of the binding preferences of the analogs and their ability to discriminate the VDR mutants in terms of transcriptional activity, we used the osteocalcin VDRE-growth hormone reporter, the three VDR mutants, and all four ligands in transfection assays. Table I summarizes these data and shows that of the four ligands tested, the most potent to restore H305Q transcription were the two side chain-modified analogs, 20E-1,25D 3 and Ro25-5318, whereas the hormone and the A ring-modified analog JK-1626-2 retained only 1.2 and 0.4% of their wild-type activity, respectively. Although the potency of the 20E-1,25D 3 analog to induce transcription through the H305Q VDR was the highest, this activity represented only 6% of its potency to induce transcription through the WT VDR. In contrast, the fluorinated side chain analog Ro25-5318 had transcription potency similar to that of 1,25D 3 through the WT VDR but maintained 43% of this potency through the H305Q VDR. These results, taken together with the protease sensitivity assays, suggest that Ro25-5318 might be a better match for the H305Q mutant than 20E-1,25D 3 .
When a similar analysis was performed with the R274L mutant VDR, the A ring analog had the greatest potency to induce transcription through this mutant. Furthermore, although the 20E-1,25D 3 analog still had significant ability to transactivate this mutant (ED 50 ϭ 30 nM) when calculated as a percentage of its WT activity (0.01%), it was not greater than the residual activity of the natural hormone (0.04%) or the side chain analog Ro25-5318 (0.03%).

DISCUSSION
The study presented here demonstrates that vitamin D analogs may be an effective treatment for a subset of patients with a rare but devastating inherited disorder, HVDRR. Analogs of 1,25D 3 have been developed in the past 20 years primarily to counter hypercalcemic effects of the natural hormone while maintaining various therapeutic properties including bone protective and antiproliferative actions in malignant cells (33,34). The analogs are being used also as molecular probes to understand the mechanism of action of proteins associated with the vitamin D endocrine system such as the P450 enzymes, vitamin D-binding protein, and the nuclear VDR (35). Studies focusing on the mechanism of action of these analogs with respect to VDR activation have led to the realization that they have the potential to interact with the receptor at amino acid contact points that differ from those utilized by the natural ligand, 1,25D 3 . Furthermore, these analogs have the potential to shape the functional surfaces of the LBD of the receptor such that it interacts with dimerization partners and coactivators of transcription differently from the 1,25D 3 -bound receptor (12,14,16). The most obvious application of these findings is to determine whether they apply to a disease or diseases with functional defects in VDR action, hence, our choice to study HVDRR.
The selection of analogs for these studies out of a pool of several hundred compounds became somewhat easier as a result of molecular modeling of the LBD of VDR. Models have been either constructed theoretically using coordinates of other ligand-binding nuclear receptors (36,37) or directly by x-ray crystallography of the hormone-occupied VDR LBD (21). These models elucidate the contact points of the 1,25D 3 in the binding pocket and the nature of amino acid residues that contribute to the functional surface of the receptor. We used this information to select candidate mutations for analog treatment and the analogs that may be able to restore the functions of these mutants.
Our experiments strongly support the concept that analogs of 1,25D 3 are capable of using alternative contact points in the LBD and show for the first time that this alternative interaction can restore the function of mutant VDRs. By using an analog with a 20E-1,25D 3 side chain, which places the 25-OH group in the northwest orientation (as opposed to the northeast orientation in the natural hormone), or an analog that had hexafluoride substitutions at positions 26 and 27 of the side chain, we were able to restore functions of a VDR that had a mutation at its 25-OH contact point, H305Q. However, from the findings with this mutant we made two additional observations. First, although histidine 305 is a contact point for the 25-OH group according to the x-ray crystallography, it is not essential for high affinity binding of 1,25D 3 . However, it has been reported that the analog 1␣-OH-D 3 , which lacks this group, does not have a significant affinity for VDR and does not have significant activity in vitro (although it is used as a prohormone in vivo) (34,35). Therefore, we propose that the 25-OH group is essential for 1,25D 3 binding but that the hormone may use an alternative contact point in the H305Q VDR, as do the analogs. The difference is that the presumed new contact of the natural hormone with the mutant VDR fails to stabilize a functional conformation of the protein, whereas the analogs do so more successfully.
The second observation is that the VDR conformations of the 1,25D 3 -or 20E-1,25D 3 -bound WT and H305Q receptors are probably different, as has been demonstrated by the protease sensitivity assays, although the 20E-1,25D 3 analog had the same potency to stabilize both conformations. This change in conformation did not have a significant effect on the potency of the 20E-1,25D 3 -occupied H305Q VDR to interact with the transcription coactivator SRC-1 in vitro, but in cultured cells there was a 4 -10-fold decrease in its potency to induce transcription of either the ocVDRE-TK-GH reporter or the 24-hydroxylase gene. These results imply that the conformation of the 20E-1,25D 3 -bound H305Q is associated with somewhat impaired transcription. This conclusion is supported by our findings that another side chain analog (Ro25-5318) induced a WT conformation of the H305Q VDR (Fig. 8) and maintained 43% of WT potency to induce transcription of the ocVDRE-GH transgene ( Table I) and 100% of WT potency to induce the 24-hydroxylase mRNA in the patient cells (data not shown). Taken together, these results suggest that perhaps it is not the p160 coactivator-binding site that is compromised in the 20E-1,25D 3 -occupied H305Q VDR but another functional interface.
Our results with R274L, a VDR mutated at a contact point for another essential functional group of 1,25D 3 , that is the 1␣-OH group, were simpler to interpret. The data confirmed both the importance of the 1␣-OH group for binding to VDR and the importance of R274 in this binding. The results also clearly showed that it is possible to restore the activity of R274L VDR with an analog (JK-1626-2), the 1-OH group of which is replaced with the 1␤-hydroxymethyl group. Furthermore, the limited structure-activity analyses performed in this study clearly show that the A ring-modified analog had a greater potency to bind and transactivate this mutant than the two side chain analogs used in this study. Our results also showed that of JK-1626-2 apparently either used an alternative contact point for the substituted 1-OH group or made contact with the substituted leucine. We hypothesize that the latter possibility was unlikely, because in additional experiments (data not shown) we replaced the arginine 274 with a less hydrophobic residue, alanine, and found that the potencies of JK-1626-2 to bind the R274A and R274L VDRs were similar. Further evidence that JK-1626-2 uses an alternative contact point or points is that mutation of serine 275 diminished binding of JK-1626-2 but not the binding of 1,25D 3 (data not shown). However, because we did not use molecular modeling to dock JK-1626-2 in the VDR binding pocket, we cannot make a conclusive statement about the interactions of the modified A-ring of JK-1626-2, especially because it also has a side chain modification.
This study also attempted to test the hypothesis that analogoccupied VDR might assume a functional conformation that is different from the functional conformation of 1,25D 3 -occupied VDR by using the F251C mutant VDR. The functional interface of the LBD of VDR includes binding sites for transcriptional coactivators and for bridging factors such as vitamin D receptor-interacting protein 205 and several regions that regulate heterodimerization such as E1 and heptad 9 (14,28,30,38,39). An effect of the analog on the functional interface may be to expose or hide an amino acid residue that may therefore be included or excluded from the functional interface, because they are found in the 1,25D 3 -VDR complexes. These changes may improve or diminish the binding of common coactivators to the receptor or, alternatively, may change the preference of the binding site for individual dimerization partners or coactivators. Although 1,25D 3 and 20E-1,25D 3 have been shown to induce different conformational changes in VDR (12,30), they failed to induce the F251C mutant to bind SRC-1 or to induce transcription of the ocVDRE-TK-GH reporter. Interestingly, the analog 20E-1,25D 3 exhibited a significant ability to induce transcription through the 24-hydroxylase promoter and to upregulate 24-hydroxylase mRNA in the cells of the patient. This may imply that the transcriptional activity of the VDR through the ocVDRE reporter gene may utilize primarily a SRC-like coactivator that is necessary for both 1,25D 3 -and 20E-1,25D 3induced transcription. However, for transactivation of the 24hydroxylase promoter, another coactivator or bridging factor may be used (14,40,41), and that factor is preferentially recruited by the 20E-1,25D 3 -bound VDR. However, our results do not reveal whether the binding site for the putative factor is compromised by the F251C mutation or located outside of the E1 domain.
In conclusion, our studies demonstrated that analogs of 1,25D 3 may restore the function of defective mutant VDR in HVDRR. Our results suggest that use of analogs to treat HVDRR will be more successful if the mutation is at a ligand contact point rather than a contact point for a coactivator of transcription. However, we cannot exclude the possibility that TABLE I Summary of transcriptional activities of 1,25D 3 and its analogs using WT VDR and the HVDRR mutants Transcriptional activation of a reporter gene containing the osteocalcin VDRE was examined in CV-1 cells cotransfected with the reporter and the indicated VDR expression plasmid. Results are expressed in ED 50 (the effective dose required to reach 50% of maximal transcriptional activity). The ED 50 values shown represent the mean Ϯ S.E. of 2-5 transfection experiments. Numbers in parentheses represent the percentage of the potency of each ligand to induce transcription of the transgene through the WT VDR. NR, ED 50 not reached; ND, not determined. other mutations in the LBD that alter the interaction of VDR with proteins indirectly may also be rescued by vitamin D analogs. These biochemical and cellular studies form the basis for considering clinical trials using vitamin D analogs to treat selected patients with HVDRR.