Studies on the Mechanism of Action of Synthetic Ligands of the Vitamin D and Estrogen receptor

July 28, 2011

Both the estrogen receptor (ER) and vitamin D receptor (VDR) belong to the same superfamily of nuclear receptors and both receptors are activated by interaction with their ligands.

Also available in Dutch


Both the estrogen receptor (ER) and vitamin D receptor (VDR) belong to the same superfamily of nuclear receptors and both receptors are activated by interaction with their ligands. Via a complex cascade of events already discussed in Chapter 1 of the thesis this interaction will eventually lead to regulation of target gene transcription. Conformational changes and phosphorylation of the receptor, as well as interaction with dimer-partners, co-factors, basal transcription factors, and specific sites in the DNA are major events in this cascade.

1,25-Dihydroxyvitamin D3 (1,25-(OH)2D3) and 17-oestradiol (E2), the natural ligands for VDR and ER, respectively, have potential therapeutic properties. However, their use as therapeutic agents is limited because of certain side effects (Chapter 1, Sections 1.3.1 and 1.6.1 of the thesis). The search for compounds with a more beneficial therapeutic profile initialized the development of synthetic ligands for VDR (1,25-(OH)2D3 analogs) and ER (SERMs). In case of VDR ligands this means that the dissociation between calcemic effects and effects on cell proliferation and differentiation is enlarged, in case of ER ligands this means that the beneficial agonistic effects of E2 (e.g. maintenance of bone mass, improved cognitive function) are mimicked, while undesirable effects (e.g. increased risk for breast and uterine cancer) are antagonized. Indeed, several of the newly synthesized compounds have improved characteristics compared to the natural ligands. An explanation for the changed biological profile is not completely clear and obviously not similar for all synthetic ligands. It probably lies in specific interactions with one or more of the events in the cascade mentioned above (see also Sections 1.3.3. and 1.6.3. of the thesis).

The aim of the studies presented in this thesis was to extend the knowledge on the mechanism of action of the VDR and ER, and their modified ligands in particular. Therefore we evaluated the differences in effect of 1,25-(OH)2D3 and a selection of its analogs on differentiation and growth of osteoblast-like cells and on in vitro bone resorption (Chapter 2 of the thesis), and investigated the role of VDR affinity (Chapter 2), conformational change of the VDR (Chapter 3), VDR stability (Chapter 3), VDRE binding (Chapter 4), and metabolism of 1,25-(OH)2D3 analogs (Chapter 5). In Chapter 6 the effect of a selection of SERMs on ER conformation was studied.

The first part of this thesis focusses on ligand interaction with the VDR. Strikingly, most investigations studying 1,25-(OH)2D3 analogs do not use cells of bone origin - a classic target tissue - but rather nonclassic target cells like breast cancer cells and cells of the immune system. However, in our view also the effects on bone should be evaluated to establish an accurate biological profile of the analogs. Therefore, the rat osteoblast-like cell line ROS 17/2.8 and the human osteoblast-like cell line MG-63 was used as a model to evaluate the effect of several side chain-modified 1,25-(OH)2D3 analogs on growth and differentiation of these cells and to compare their potency with that of 1,25-(OH)2D3. Furthermore, isolated long bones from fetal mice were used to study the capacity of 1,25-(OH)2D3 and the analogs to release calcium from bone.

All 1,25-(OH)2D3 analogs examined in this thesis exhibited agonistic activity in all responses tested: like 1,25-(OH)2D3 the analogs stimulated the synthesis of the bone matrix proteins osteocalcin and type I procollagen. Furthermore, the analogs mimicked the natural ligand in their effect on osteoblast-like cell growth and in their stimulatory effect on in vitro bone resorption (Chapter 2 of the thesis). Some analogs (OCT, MC903, CB966) exhibited an in vitro biological potency comparable to that of 1,25-(OH)2D3, whereas others (KH1049, KH1060, EB1089) were much more potent than 1,25-(OH)2D3. This observation is supported by almost every in vivo and in vitro study published so far: 1,25-(OH)2D3 analogs are super, full, or partial agonists of 1,25-(OH)2D3.

We found that the potency of the 1,25-(OH)2D3 analogs was related to the biological response. In some responses their potency (compared to 1,25-(OH)2D3) was only moderately increased, while in other responses the increase in potency was much more striking. In Table 2.3 of the thesis this differential effect on gene regulation is expressed as the ED50 ratio for stimulation of in vitro bone resorption and stimulation of extracellular bone matrix protein synthesis. It clearly demonstrates that for EB1089 and KH1060 the ED50 ratio is increased compared to the ED50 ratio for 1,25-(OH)2D3. The finding that for some analogs the ED50 ratio between bone formation parameters (stimulation of osteocalcin and type I collagen synthesis) and bone resorption parameters (stimulation of calcium release) is increased compared to the ED50 ratio for 1,25-(OH)2D3 might support a therapeutic potential of these analogs in the treatment of metabolic bone diseases, i.e. osteoporosis. A relationship between increased bone turnover and the risk for metastasis of cancer to bone has been reported. The in vivo potency of these analogs to induce bone resorption must therefore be examined with great care.

The studies presented in Chapter 2 of the thesis show cell type-specific differences in sensitivity for 1,25-(OH)2D3 and some of the analogs. For instance, in MG-63 cells maximal stimulation of osteocalcin production by EB1089, KH1049, and KH1060 was almost two-fold increased compared to 1,25-(OH)2D3 and the analogs with moderate activity (OCT, MC903, CB966), while no difference in maximal stimulation was observed in ROS 17/2.8 cells. However, in general, MG-63 cells were considerably less sensitive than ROS 17/2.8 cells. For instance, in MG-63 cells the ED50 for osteocalcin synthesis induced by 1,25-(OH)2D3, OCT, CB966, or EB1089 was reached at 100-600 times higher concentrations than in ROS 17/2.8 cells. Cell type-specific differences in VDR content and/or function might underlie this phenomenon. We found that MG-63 cells have a much lower VDR content compared to ROS 17/2.8 cells and do not show homologous VDR up-regulation (data not shown). In addition, cell type-specific differences in presence and/or distribution of cofactors might play a role. We also observed that the ED50’s for both KH1049 and KH1060 (both 20-epi analogs) were comparable between the two cell lines, indicating that for these analogs cell type-specific differences might be of less importance. Another possibility is that KH1049 and KH1060 are metabolized more efficiently by ROS 17/2.8 cells, (or less efficiently by MG-63 cells) in relation to 1,25-(OH)2D3 and the other analogs. This could lead to relatively high concentrations of KH analogs in MG-63 cells or relatively low concentrations of these compounds in ROS 17/2.8 cells.

With most cell types 1,25-(OH)2D3 has an inhibitory effect on growth; a feature that gave rise to the thought to use 1,25-(OH)2D3 or its analogs for the treatment of hyperproliferative diseases. Likewise, with the human osteosarcoma cell line MG-63 and the human breast cancer cell lines MCF-7 and ZR75-1 our laboratory observed dose-dependent inhibition of proliferation by 1,25-(OH)2D3 and its analogs (Chapter 2 of the thesis). However, the rat osteblast-like ROS 17/2.8 cells were stimulated in their growth. Interestingly, the analogs exhibited a similar order of potency in their growth inhibitory activity (in MG-63 cells) and in their growth stimulatory activity (in ROS 17/2.8 cells).

One of the key events in the cascade leading to regulation of gene transcription is the ligand-induced conformational change of the VDR. To study the effects of 1,25-(OH)2D3 and KH1060 on VDR conformation partial proteolytic digestion analysis was used. The method is based on the idea that accessibility of potential protease cleavage sites within a receptor molecule alters as a result of ligand-induced changes in receptor conformation. Consequently, changes in receptor conformation will lead to changes in the protease digestion pattern of the receptors. We found that in the absence of ligand in vitro synthesized VDR is rapidly degraded by proteases. In contrast, 1,25-(OH)2D3 and the analogs changed the VDR conformation, resulting in enhanced protease resistance of distinct VDR fragments. In general 1,25-(OH)2D3 analogs can be regarded as agonists, although the potency by which they mimic the natural ligand differs between tissues, cells, and responses studied. This probably explains why we could not establish qualitative differences in the VDR protease digestion profiles of 1,25-(OH)2D3 and the analogs: the sizes of the protease resistant VDR fragments were not distinct, only the intensity of the preserved fragments was different.

Interestingly, we and others found that treatment of unliganded in vitro synthesized VDR with low trypsin concentrations resulted in preservation of VDR fragments with a similar size as the fragments observed with liganded VDR (Chapter 3 of the thesis). When protease digestion was intensified (by increasing the trypsin concentration or the trypsin incubation time) these fragments disappeared, whereas the fragments of liganded VDR were still protected. 

We observed that 1,25-(OH)2D3 analogs and metabolites with strong biological activity are also strong inducers of a VDR conformation with increased protease resistance and vice versa (Chapters 2, 3, and 4 of the thesis). However, one should be cautious in overestimating the quantitative differences obtained by limited protease digestion analysis. For instance, the strong biological effect of EB1089 (an analog with normal configuration at C-20) is not reflected by a strong increase in the intensity of protease resistant VDR fragments (data not shown). One can not exclude the possibility that some potent 1,25-(OH)2D3 analogs might induce a conformational change of the VDR that does not lead to increased protease resistance but even to increased protease sensitivity.

Crystal structure analysis revealed that liganded nuclear receptors adopt a more compact structure, obviously less sensitive to protease activity. In cells this property could lead to increased VDR stability. Indeed, our studies presented in Chapters 3 and 5 of the thesis show that 1,25-(OH)2D3, analogs and specific metabolites of the analog KH1060 prolong VDR half-life in ROS 17/2.8 cells.

The ligand-induced conformational change of the VDR probably facilitates dimerization with RXR and enhances binding to specific regulatory elements (VDREs) in the vicinity of 1,25-(OH)2D3 target genes. In Chapter 4 of the thesis the potency of KH1060 was compared with the potency of 1,25-(OH)2D3 to induce binding of the VDR to VDREs. Both in vitro synthesized VDR/RXR as well as nuclear extracts of ROS 17/2.8 cells were used to investigate VDR binding to VDREs from three different target genes: rat osteocalcin, human osteocalcin and mouse osteopontin. Both 1,25-(OH)2D3 and KH1060 induced VDR binding to these VDREs in a ligand dose-dependent manner, independent of the source of VDRs (in vitro synthesized or nuclear extracts). All three VDREs are of the DR3 type: two hexameric half-sites separated by three nucleotides, and despite of minor differences in nucleotide sequence (see Table 4.1 of the thesis) large differences in VDR-VDRE affinity were observed. The VDR bound with lowest affinity to the rat osteocalcin VDRE, and with highest affinity to the mouse osteopontin VDRE. We further elaborated on this interesting finding and produced hybrid VDREs in which the rat osteocalcin VDRE half-sites were replaced by one or both mouse osteopontin VDRE half-sites. Gel mobility shift assays performed with these VDREs clearly showed that the proximal half-site of the mouse osteopontin VDRE (i.e. the OC/OP VDRE) was mainly responsible for the increased affinity. Transfection studies in ROS 17/2.8 cells revealed that 1,25-(OH)2D3-induced transcription of the OC/OP VDRE-driven reporter gene was more effective than via the rat osteocalcin VDRE-driven reporter. Further substitution studies revealed that in particular the third and/or fourth nucleotides (both thymidine) determined the difference in VDR-VDRE affinity.

We found that KH1060 was a stronger inducer of VDR/RXR binding to the mouse osteopontin VDRE than 1,25-(OH)2D3. However, surprisingly, when we studied binding of in vitro synthesized VDR/RXR to the human and rat osteocalcin VDRE KH1060 exhibited no increased potency compared to 1,25-(OH)2D3. In contrast, when VDR protein was extracted from ligand-treated cells KH1060 was a stronger inducer of VDR/RXR-VDRE binding than 1,25-(OH)2D3 with all VDRE types tested. So, this indicates that the KH1060-induced conformational change of the VDR does not necessarily result in enhanced VDRE binding, but in addition depends on the VDRE nucleotide sequence and furthermore underlines the importance of a cellular context (e.g. presence of cofactors) to obtain optimal VDR-VDRE binding. Therefore, it seems justified to speculate that KH1060 changes the VDR conformation in such a way that, compared to the 1,25-(OH)2D3-induced VDR conformation, cofactor binding is facilitated or that other KH1060-specific cofactors are involved and that this will lead to increased VDR-VDRE binding.

Formation of biologically active metabolites might be a cause for concern. Obviously, from a therapeutic point of view rapidly acting analogs with inactive metabolites are much better to control than analogs that are metabolized into one or more active metabolites. For each of these metabolites the mechanism of action and possible side effects should be established before safe clinical use can be guaranteed. Chapter 5 of the thesis focusses on the potential contribution of metabolites of KH1060 to the biological potency of the parent compound. Earlier was shown that in vitro KH1060 is metabolized into at least 22 different compounds. The metabolites 24a-OH-KH1060, 26-OH-KH1060, and 26a-OH-KH1060 were potent inducers of osteopontin mRNA expression, rat osteocalcin VDRE-driven reporter gene transcription, and active stimulators of osteocalcin synthesis (Chapter 5 of the thesis). The same metabolites were also active inducers of VDR binding to rat osteocalcin and rat cytochrome P450 (CYP24) VDREs (Chapter 5). Next, we investigated whether the biological potency of these metabolites was correlated with their potency to induce a conformational change of the VDR with enhanced protease resistance. Indeed, partial protease digestion analysis showed that the biological active metabolites changed the VDR conformation resulting in enhanced protease resistance, whereas a metabolite with weak biological activity was inactive in the protease digestion analysis. A corresponding pattern was observed studying the effect of the metabolites on VDR stability in ROS 17/2.8 cells.

The second part of the thesis focusses on ligand interaction with the ER. We studied the effect of E2 and several synthetic ER ligands on the conformation of both ER subtypes: ERa and ER. One of the tested ER ligands, tamoxifen, is a SERM with tissue-dependent partial agonistic/antagonistic activities: it mimics E2 with anabolic effects on bone, protective effects on the cardiovascular system, and beneficial effects on cognitive function, whereas it counteracts the stimulatory effect of E2 on breast tumors (See Section B.5.2. of the thesis). An explanation for the tissue-dependent partial agonistic/antagonistic activation of SERMs is still unknown.

Using partial protease digestion analysis we showed that tamoxifen changed the conformation of both ERa and ERb differently from that induced by E2. Also other partial agonists/antagonists (4-hydroxytamoxifen, idoxifen, and the raloxifene derivative LY 117,018-HCl) had a similar effect as tamoxifen. Like other investigators we could not discriminate between different SERMs based on the protease digestion profiles. Other techniques, like phage enzyme-linked immunoassays have shown to be more suitable for this purpose. Furthermore, it is difficult to translate the intensity of the preserved fragments obtained with the partial protease digestion analysis into a functional receptor affinity as was done for VDR analogs. Therefore, protease resistance (as a tool to study ligand-induced conformational changes of a receptor) can only be used to calculate a functional receptor affinity when the investigated ligands exert agonistic activities. However, for nuclear hormone receptors like the progesterone receptor and ER (partial) agonistic/antagonistic ligands are known. Interaction with these ligands lead to a conformational change of the receptor that is clearly distinguishable by partial digestion analysis from the conformational change induced by ligands with pure agonistic action.