|Home | About | Journals | Submit | Contact Us | Français|
Structure-functional characterization of vitamin D receptor (VDR) requires identification of structurally distinct areas of VDR-ligand-binding domain (VDR-LBD) important for biological properties of 1α,25-dihydroxyvitamin D3 (1,25(OH)2D3). We hypothesized that covalent attachment of the ligand into VDR-LBD might alter ‘surface structure’ of that area influencing biological activity of the ligand. We compared anti-proliferative activity of three affinity alkylating derivatives of 1,25(OH)2D3 containing an alkylating probe at 1,3 and 11 positions. These compounds possessed high-affinity binding for VDR; and affinity labeled VDR-LBD. But, only the analog with probe at 3-position significantly altered growth in keratinocytes, compared with 1,25(OH)2D3. Molecular models of these analogs, docked inside VDR-LBD tentatively identified Ser237 (helix-3: 1,25(OH)2D3-1-BE), Cys288 (β-hairpin region: 1,25(OH)2D3-3-BE,) and Tyr295 (helix-6: 1,25(OH)2D3-11-BE,) as amino acids that are potentially modified by these reagents. Therefore, we conclude that the β-hairpin region (modified by 1,25(OH)2D3-3-BE) is most important for growth-inhibition by 1,25(OH)2D3, while helices 3 and 6 are less important for such activity.
1α,25-Dihydroxyvitamin D3 (1,25(OH)2 D3), the dihydroxylated metabolite of vitamin D3 serves multiple functions. Its biological properties include calcium and phosphorus-homeostasis, growth and maturation-control of a broad range of malignant cells, and immune-regulation. As a result the therapeutic potential of 1,25(OH)2 D3 in a broad range of diseases, including mineral homeostatic diseases such as renal osteodystrophy, proliferative diseases such as psoriasis and cancer, and immune-deficiency diseases, such as type I diabetes mellitus is well-recognized [1–7]. However, inherent toxicity of the parent hormone (hypercalcemia, hypercalciuria), particularly at pharmacological doses, has largely precluded its general use as a therapeutic agent. This limitation has spawned a strong interest in developing analogs of 1,25(OH)2D3 that retain intact beneficial effects but display reduced toxicity. Several such analogs have shown promise, but have displayed only a moderate effect clinically in non-toxic doses [8–12]. It is amply clear that rational development of such analogs will require proper understanding of their mechanism of action at the molecular level.
According to current dogma, 1,25(OH)2D3 binds to its nuclear receptor, vitamin D receptor (VDR) in target cells with high specificity; allosterically promoting heterodimerization with the retinoid X receptor (RXR), and binding of the VDR-1,25(OH)2D3-RXR complex to vitamin D response elements (VDREs) in the vitamin D–regulated genes [e.g., osteopontin, osteocalcin, 1α,25-dihydroxyvitamin D3-24-hydroxylase (CYP-24)] and recruitment of co-activators to initiate transcription and translation [13, 14]. In an alternative proposal apo-VDR, bound to co-repressors, remains transcriptionally inactive till 1,25(OH)2D3 binds to initiate the multi-step transcriptional process. The most important among all the steps in this transcriptional process is highly specific interaction between 1,25(OH)2D3 and VDR. Therefore, structure-functional knowledge, from the sides of both VDR and 1,25(OH)2D3 is crucial for a complete understanding of the molecular mechanism of this hormone, as well as development of new generations of 1,25(OH)2D3-based drugs for various diseases with high efficacy and low toxicity.
Recently reports describing the x-ray crystal structures of VDR-LBD, bound to its natural ligand (1,25(OH)2D3) and several 1,25(OH)2D3-analogs have been published [15–18]. These studies have provided structural evidence for the crucial role played by helix 12 (H-12) in the C-terminal region of VDR in co-activator recruitment and ligand-activated transcriptional process . In addition, a recent study has indicated important role played by H-3 in the transcriptional process and ligand-related activities . Beyond this, there is very little information available about possible role of other structurally distinct areas/features in VDR-LBD that may play an important role in the VDR- 1,25(OH)2D3-mediated transcriptional process.
Mutation in a protein introduces structural/conformational perturbation that is sometimes translated into changes in biological properties related to this protein. We hypothesized that structural perturbation could also be brought about by covalently attaching appendages that mimic the actual ligand to specific areas of the VDR-LBD (chemical modification by “mutated” ligand). Such a change might influence the signal transduction process (as reflected in the biological activity of the ligand) by altering the ‘surface structure’, micro-environment and polarity of that area of the protein, and that may result in differential recruitment of co-activators for transcription. This way we might be able to identify yet unrecognized structural motifs inside VDR-LBD for the control of the transcriptional process.
Affinity labeling is a classical biochemical method to study interaction between a ligand and its receptor . Such an interaction leads to covalent and specific labeling of the ligand-binding domain by a ligand-analog. In recent times this method has been extended to proteomic studies to investigate protein-protein interaction in complex cellular systems as well.
We have previously demonstrated that 1α,25-dihydroxyvitamin D3-3-bromoacetate (1,25(OH)2D3-3-BE), an affinity labeling analog of 1,25(OH)2D3 containing a reactive electrophilic group at the 3-position, covalently labels a single cysteine residue (Cys288) in VDR-LBD . We argued that affinity labeling analogs of 1,25(OH)2D3 with reactive affinity probe attached to various parts of the parent molecule could potentially attach 1,25(OH)2D3 to different parts of the VDR-LBD, and cause perturbation maximally localized in that area; and such a process might be reflected in the biological activities of these analogs. Therefore, in the present study we compared the anti-proliferative activity of three affinity labeling analogs of 1,25(OH)2D3 containing affinity probe at the 1, 3 and 11 positions of the parent molecule. Furthermore, we computationally docked these compounds inside the VDR-LBD crystal coordinates to identify areas of the protein that may be specifically labeled by these compounds, and compared cellular activity of each compound with the parent hormone, 1,25(OH)2D3 in normal human keratinocytes. This communication reports results of these studies, and their implications.
1α,25-Dihydroxyvitamin D3-1α-(2)-bromoacetate (1,25(OH)2D3-1-BE) and 1α,25-dihydroxy[26(27)-3H]vitamin D3-1α-(2)-bromoacetate (3H-1,25(OH)2D3-1-BE, specific activity 175 Ci/mmol) and 1α,25-dihydroxyvitamin D3-3β-(2)-bromoacetate (1,25(OH)2D3-3-BE) and 1α,25-dihydroxyvitamin D3-3β-[1-14C]-(2)-bromoacetate (14C-1,25(OH)2D3-3-BE, specific activity 18.65 mCi/mmol) were synthesized by previously published procedures . 1α,25-Dihydroxyvitamin D3-6-propoxy-(2)-bromoacetate (1,25(OH)2D3-6-BE) was obtained by a method described earlier by us . 1α,25-Dihydroxyvitamin D3-11α-hydroxy-(2)-bromoacetate (1,25(OH)2D3-11-BE) and 1α,25-dihydroxyvitamin D3-11α-hydroxy-[1-14C]-(2)-bromoacetate (14C-1,25(OH)2D3-11-BE, specific activity 18.65 mCi/mmol) were synthesized according to our published procedure .
Full-length recombinant VDR was expressed in E. coli as a GST-fusion protein, and purified according to Swamy et al. .
These assays were carried out by a standard procedure. Typically 50 ng of reVDR was incubated with 3H-1,25(OH)2D3 (4000 cpm, sp. activity 120 Ci/mmol, Amersham) in the presence of increasing concentrations of 1,25(OH)2D3 or analogs (44.7 fmol - 2.4 pmol) in VDR assay buffer (50 mM Tris HCl, 150 mM NaCl, 1.5 mM EDTA, 10 mM sodium molybdate, 5 mM DTT and 0.1% Triton X 100, pH 7.4) for 15 hours at 4°C. Rat liver nuclear extract was included in the binding assays to provide the nuclear accessory factor/s . After the incubation Dextran-coated charcoal was added to remove unbound 3H-1,25(OH)2D3 and the radioactivity in the supernatants, after centrifugation, was determined by liquid scintillation counting. Assays were carried out in triplicate.
Samples of reVDR (5 μg) were incubated with 2,000 cpm (0.07 nmol) of 14C-1,25(OH)2D3-3-BE or 14C-1,25(OH)2D3-11-BE or 10,000 cpm of 3H-1,25(OH)2D3-1-BE (0.06 fmol) in the presence or in the absence of 1,25(OH)2D3 (1 μg, 2.4 nmol) in 50 mM Tris HCl buffer, pH 7.4 containing 5 mM DTT for 2 hours at 4°C, and reaction was terminated by boiling with SDS-PAGE sample buffer for 5 minutes. The samples were analyzed by SDS-PAGE, followed by radioactive scanning (3H-containing samples) and phosphorimaging (14C-containing samples).
Briefly, 3T3 cells were plated at 104 cells per 35 mm tissue-culture dish, and were irradiated lethally after two days with a 60Co source (5000 rads). Keratinocytes were obtained from neonatal foreskin after overnight trypsinization at 4°C and treatment with 0.2% EDTA. The cells, in serum-free medium, were plated on lethally-irradiated 3T3 cells. Each experiment was performed on primary or secondary kerationocyte cultures obtained from different skin samples. The serum-free medium consisted of MCDB 153 medium (Sigma Chemical Co.) with additives and calcium (0.15 mM). The cells were grown to 50–60% confluence, when medium was removed and replaced with one ml of fresh medium containing either ethanol (0.1% v/v) or various doses of 1,25(OH)2D3 or analogs.
Cells, grown to approximately 50% confluence in MCDB medium were incubated at 37°C with various concentrations (10−10 M and 10−8M) of either 1,25(OH)2D3 or an analog (in one μL of ethanol per mL of medium) for 24 hours. Each experiment was carried out in triplicate. Control experiments were carried out by incubating cells with ethanol for the same period of time. After the incubation, medium was removed from each plate and was replaced with 3H-thymidine (one μCi, Sigma-Aldrich Chemical Co., St. Louis, MO) in one mL of fresh medium. The cells were incubated at 37°C for three (3) hours followed by the removal of the medium and washing with salinated phosphate buffer. The cells were cooled on ice, and one ml of ice-cold perchloric acid solution (5%) was added to each dish followed by incubation on ice for 15 minutes. After the incubation, aqueous medium was removed, and the cells were washed with one mL of ice-cold perchloric acid solution, and finally replaced with one mL of fresh perchloric acid solution. The cells were incubated in a shaking water bath at 70°C for 15 minutes. After this period, medium from each dish was removed, mixed with 10 mL of liquid scintillation cocktail, and counted for radioactivity. Results of this experiment are reported as percentage of cpm for ethanol control for each compound and at each dose level. These results (Figure 4) are representative of same experiments carried out twice. Statistical analysis was done by student’s t test.
The VDR-LBD structure, taken directly from the co-crystallized protein-ligand complex (Protein Data Bank, PDB ID: 1DB1) was prepared by removing any water molecules, adding hydrogen atoms and assigning Kollman partial charges using Autodock tools before performing actual docking with AutoDock3 suite. Dockings were performed using the Lamarcian Genetic Algorithm (LGA) with default GA parameters using 60×60×60 grid map calculated by Autogrid. During the run the bromoacetate and hydroxyl groups on the ligand structure were flexible whereas rest of the steroid-derived skeleton was kept rigid in the original conformation. 50 GA runs were performed on each analog and results were crosschecked with multiple Simulated Annealing (SA) dockings each with 10(100) runs, varying number of cycles, step parameters and temperature factors, both for partially flexible and full-flexible performed SA dockings were carried out. The cluster conformations with lowest docked energy (defined as intermolecular interaction energy plus torsional free energy) were manually inspected and superimposed. The amino acid residues in VDR-LBD in close proximity of the carbon atom bearing Br (atom) were identified.
Affinity alkylating derivatives of naturally occurring bio-molecules constitute high-affinity substrate/ligand-analogs that can provide valuable information about the three-dimensional geometries of the substrate/ligand binding pockets of their cognate proteins including identity of key amino acid residues (contact points) and orientation of the substrate/ligand inside the pocket. The affinity labeling process involves interaction between a nucleophilic amino acid residue in the substrate/ligand binding pocket with an electrophile in the affinity reagent that lies in close enough proximity to form a covalent bond. In summary, the technique of affinity/photoaffinity labeling, coupled with mutational analysis and functional assay provide a dynamic picture of the binding event between a ligand/substrate with its cognate receptor/enzyme .
For the past several years our laboratory has focused on the development of affinity and photaffinity labeling derivatives of 1,25(OH)2D3 to probe VDR-LBD and obtain structure-functional information about VDR-1,25(OH)2D3 interaction [22, 28–31]. In this effort we synthesized 1,25(OH)2D3-3-BE, an affinity labeling derivative of 1,25(OH)2D3, and demonstrated that this compound specifically labels a single Cysteine residue (Cys288) in VDR-LBD . Functional importance of this residue was confirmed by mutation and 1,25(OH)2D3-binding analysis. Additionally we identified several other contact points inside VDR-LBD using Cys288 as the ‘docking point’ for the 3-hydroxyl group in 1,25(OH)2D3 (vide infra). One criticism that is usually leveled at the affinity labeling technique is that the alkylation process could be random, i.e. irrelevant amino acid residue/s in- and/or outside the actual ligand-binding/substrate-binding pocket could be alkylated by these analogs. Our experience, however, is quite different; and we observed that 1,25(OH)2D3-3-BE specifically labeled a single cysteine residue (Cys288) out of three (3) inside the ligand-binding pocket of VDR strongly suggesting that VDR has a very tight binding pocket in the vicinity of the A-ring (of 1,25(OH)2D3). Interestingly Cys288 is one of the conserved Cys residues in the VDR-LBD among other steroid hormone receptors; and prior to our report Nakajima et al. demonstrated a crucial role of Cys288 towards ligand binding .
In addition to the demonstration of specific labeling (of VDR-LBD) by 1,25(OH)2D3-3-BE, we established that labeling by 1,25(OH)2D3-3-BE is rapid and quantitative, and labeling process leads to relative stabilization of the holo-VDR-osteocalcin vitamin D responsive element (VDRE) complex. In addition, we observed that 1,25(OH)2D3-3-BE has a considerably stronger antiproliferative activity compared with 1,25(OH)2D3 in human keratinocytes . So, we asked the question of whether or not other affinity labeling derivatives of 1,25(OH)2D3 containing affinity probes attached at different parts of 1,25(OH)2D3, that can potentially cross-link different sites in the VDR-LBD can have differential cellular effects compared with 1,25(OH)2D3.
As a part of an earlier study we synthesized 1,25(OH)2D3-1-BE, 1,25(OH)2D3-6-BE and 1,25(OH)2D3-11-BE (Figure 1). Competitive radio-ligand binding assays of four affinity analogs with affinity probe at 1, 3, 6 and 11 positions with reVDR demonstrated half-maximal binding concentrations of 1,25(OH)2D3-1-BE, 1,25(OH)2D3-3-BE,1,25(OH)2D3-11-BE and 1,25(OH)2D3 (control) are 0.52, 0.18, 0.52 and 0.0015 nmol respectively. By contrast, 1,25(OH)2D3-6-BE showed no specific binding to VDR (Figure 2). Therefore, we used 1,25(OH)2D3-1-BE, 1,25(OH)2D3-3-BE,1,25(OH)2D3-11-BE for subsequent experiments.
Incubation of VDR samples with 3H-1,25(OH)2D3-1-BE, 14C-1,25(OH)2D3-3-BE and 14C-1,25(OH)2D3-11-BE shows that all these compounds covalently label the protein (Figure 3). However, carrying out the incubation with an excess of 1,25(OH)2D3 significantly reduced labeling in each case indicating specific labeling of VDR-LBD by these compounds (Figure 3). It should be noted although excess of 1,25(OH)2D3 reduced the intensity of labeling in each case, it did not obliterate cross-linking of the compounds to VDR-LBD. A possible explanation would be that if interaction between the affinity reagent and VDR-LBD is rapid and irreversible, as we have observed ), covalently labeled VDR should accumulate with time, even when 1,25(OH)2D3-X-BE (x = 1, 3, 11) has to compete with 1,25(OH)2D3 to occupy the binding site in VDR-LBD. Furthermore, as seen in Figure 3, decrease of labeling (decrease in the intensity of the labeled band) in the presence of a constant amount of 1,25(OH)2D3 varies with different affinity reagents indicating various degrees of competition between 1,25(OH)2D3 and its affinity derivatives. Overall, these results indicated that all these analogs specifically and covalently labeled the 1,25(OH)2D3-binding site in the VDR-LBD.
In our next experiment antiproliferative activities of 1,25(OH)2D3-1-BE, 1,25(OH)2D3-3-BE and 1,25(OH)2D3-11-BE were compared with 1,25(OH)2D3 at two dose levels. Results of these assays demonstrated that 10−8M of 1,25(OH)2D3 exhibits approximately 20% growth inhibition (Figure 4). At the same dose 1,25(OH)2D3-3-BE inhibits the growth by approximately 45%. At 10−10M dose level 1,25(OH)2D3 practically has no effect on the growth, while an equmolar amount of 1,25(OH)2D3-3-BE inhibits the growth by approximately 25%. Interestingly, at both doses effects of 1,25(OH)2D3-1-BE and 1,25(OH)2D3-11-BE are not significantly different from that of 1,25(OH)2D3.
We have demonstrated that 1,25(OH)2D3-1-BE, 1,25(OH)2D3-3-BE and 1,25(OH)2D3-11-BE bind VDR with high and almost equal affinity (Figure 2), and specifically label VDR-LBD (Figure 3). Therefore, the growth assay data suggest that simple perturbation of specific areas of VDR-LBD by cross-linking of these analogs might not contribute significantly towards differential growth-inhibitory activity of these compounds in keratinocytes.
Construction of molecular models of these affinity labels, and docking them inside VDR-LBD, based on crystal coordinates identified Ser237, Cys288 and Tyr295 as amino acids that are potentially modified by 1,25(OH)2D3-1-BE, 1,25(OH)2D3-3-BE and 1,25(OH)2D3-11-BE respectively (Figures 5–7). Ser237, present in Helix 3 has been implicated in ligand-binding by others and us [20, 22]. However, according to our growth inhibition assays 1,25(OH)2D3-1-BE is similar to 1,25(OH)2D3 in antiproliferative property in keratinocytes. A similar conclusion is drawn about helix 6 (containing Tyr295). However, the β-hairpin region, containing Cys288 (contact point for 1,25(OH)2D3-3-BE) is different, and its perturbation lead to modulation of cellular activity. It should be emphasized that Cys288 was implicated to be crucial for ligand-binding in an earlier study by a different group . In addition, in our earlier study we mutated Trp286 and Met284, two neighboring amino acid residues of Cys288 (Figure 8). Mutation of Trp286 to Ala or Phe almost completely removed 1,25(OH)2D3-binding, while conversion of Met284 to Ser or Ala reduced such binding by approximately 70% . Met284, Trp286 and Cys288 are constituent amino acid residues in the unstructured β-hairpin region. Therefore, we concluded that the β-hairpin region is important for growth-inhibitory property of 1,25(OH)2D3, while helices 3 and 6 are less important for such activity.
In recent years our laboratory and others have reported strong agonistic activity of the 3-bromoacetate derivative of 1,25(OH)2D3 and 25-hydroxyvitamin D3, the pre-hormonal precursor of 1,25(OH)2D3 [33–35]. Previously such activities have been solely attributed to the enhanced catabolic stability of these compounds compared with 1,25(OH)2D3, as illustrated by the cartoon in Figure 8. However, our current studies demonstrate that mere catabolic stability of these compounds (by affinity labeling) may not contribute equally towards their cellular activity; and a combination of enhanced catabolic stability and specific area of perturbation (by covalent attachment of the ligand) might be responsible for the enhanced antiproliferative activity.
In conclusion, majority of structure-function studies have identified helix 12 as the most important structural region of VDR-LBD for the biological activities of 1,25(OH)2D3 and its analogs. Results of the current study strongly suggest that the β-hairpin region of VDR-LBD might also contribute significantly towards such activities. The 3-bromoacetate derivative of 1,25(OH)2D3 and 25-hydroxyvitamin D3 have been projected to have strong potential in several malignancies [33–35]. Therefore, information presented in this communication will be extremely important in providing molecular basis of their action, and develop next generation compounds with enhanced pharmacological properties.
This work was supported by grants (to RR) from Department of Defense, Prostate Cancer Research Program (PC 051136), National Cancer Institute of the National Institutes of Health (1R41 CA126317-01A1 and 1R21 CA127629-01A2), and Community Technology Fund, Boston University. The authors would also like to thank Dr. Sandor Vajda, College of Engineering, Boston University.
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.