2025 Volume 73 Issue 1 Pages 1-17
2α-Functionalization of 1α,25-dihydroxyvitamin D3 (active vitamin D3) A-ring enhances binding affinity for the vitamin D receptor (VDR) and prolongs the half-life in target cells due to gaining resistance to CYP24A1-dependant metabolism. The wide variety of modified A-ring precursor enynes for Trost coupling with CD-ring bromoolefin were synthesized from d-glucose. The A-ring modification provided potent, selective biological activities without calcemic side-effects in vivo; for example, 2α-(3-hydroxypropyl)-19-nor-1α,25-dihydroxyvitamin D3 (MART-10) exhibits potent antitumor activity (0.3µg/kg/d, twice/week for 3 weeks) in nude mice inoculated with BxpC-3 cancer cells, 2α-[2-(tetrazol-2-yl)ethyl]-1α,25-dihydroxyvitamin D3 (AH-1) shows better bone-forming effects (0.02µg/kg/d, 5d/week for 4 weeks) in ovariectomized (OVX) rats as an osteoporosis model than natural active vitamin D3, and NS-74c exhibits potent VDR-antagonistic activity (IC50 7.4pM) in HL-60 culture cells. The A-ring modification was also applicable to the synthesis of stable 14-epi-19-nortachysterols, and their novel VDR binding mode was confirmed by X-ray co-crystallographic analysis. 25-Hydroxyvitamin D3 has two independent target molecules: VDR and a sterol regulatory element-binding protein (SREBP)/SREBP cleavage-activating protein (SCAP) complex, and 25-hydroxyvitamin D3 shows SREBP/SCAP inhibitory activity. The VDR-silent vitamin D analog KK-052 with selective SREBP/SCAP inhibitory activity in vivo was developed. A chemical library of side-chain fluorinated vitamin D analogs is currently under construction, and some analogs have shown potent anti-inflammatory activity and therapeutic effects on psoriasis model mice.
The biosynthetic precursor of cholesterol, 7-dehydrocholesterol, is converted to previtamin D3 by a sunlight-dependent (UVB, 290–315nm) electrocyclic reaction in the plasma membrane of skin cells.1) The photoproduct previtamin D3 isomerizes to vitamin D3 due to the body temperature, thermodynamically, via a [1,7]-sigmatropic rearrangement consisting of an equilibrium at a ratio of approx. 5:95. Previtamin D3 can absorb solar UVB photons and isomerize into lumisterol and tachysterol2) (Chart 1). Vitamin D3 can enter the circulation by docking with vitamin D-binding protein (DBP) or can absorb solar photons to isomerize to 5,6-trans-vitamin D3, suprasterol I, and suprasterol II.3) These photochemical reactions proceed in the skin through all nonenzymatic processes.
Some of these analogs were studied in this review.
Vitamin D3 is itself biologically inert and activated by sequential steps of 25-hydroxylation by hepatic microsomal CYP2R1 and mitochondrial CYP27A1 to produce the major circulating 25-hydroxyvitamin D3 [25(OH)D3] by docking with DBP, and then 1α-hydroxylation by renal mitochondrial CYP27B1, which is tightly regulated by the need for calcium and phosphorus in the blood, to produce the most physiologically active hormonal form of 1α,25-dihydroxyvitamin D3 [1α,25(OH)2D3]4–8) (Chart 2).
1α,25(OH)2D3 is the most potent vitamin D receptor (VDR) agonist in humans and regulates cellular proliferation and differentiation, apoptosis, and immune responses, in addition to its major classical roles in calcium and phosphorus homeostasis and bone mineralization through VDR, which is a ligand-dependent transcription factor; therefore, 1α,25(OH)2D3, even the weak ligand 25(OH)D3, and their synthetic analogs may have therapeutic applications not only for bone diseases, but also cancers, autoimmune diseases, cardiovascular disorders, and infectious diseases.9–12) A wide variety of vitamin D analogs have been synthesized by our group, and the representative structures mentioned in this review are summarized in Table 1, including synthetic manipulation and characteristic biological activities in vitro and in vivo. The synthesis of each analog, the VDR binding mode, CYP24A1-dependant metabolism, and biological activity, including important therapeutic effects, if available, are described.
 |
In the late 1990s, Takayama’s laboratory of Teikyo University found that introducing some specific functional groups, such as a methyl or 3-hydroxypropyl group, to the A-ring 2α-position of 1α,25(OH)2D3 enhanced binding affinity for VDR beyond the natural ligand, 1α,25(OH)2D3.13–16) Construction of the triene system of 2α-modified 1α,25(OH)2D3 analogs was performed by the Trost coupling reaction between an A-ring precursor enyne and CD-ring bromoolefine. The first synthesis of the 2α-methylated A-ring precursor was started from methyl (R)-3-hydroxy-2-methylpropionate,13,14) and the 3-hydroxypropyl version was synthesized from D-xylose.15,16) 2α-Methyl- and 2α-(3-hydroxypropyl)-1α,25(OH)2D3 showed 4- and 3-times greater VDR binding affinity than that of 1α,25(OH)2D3, respectively, and to visualize the effect of the 2α-(3-hydroxypropoxy) group, a novel synthetic approach using d-glucose as a chiral source of the A-ring precursor was developed. Briefly, methyl 2,3-anhydro-4,6-O-benzylidene-α-d-mannopyranoside prepared from commercially available methyl α-d-glucopyranoside was reacted with 1,3-propanediol under basic conditions and gave an epoxide ring-opening adduct in good yield. Subsequent synthetic manipulation led to the enyne (R=OCH2CH2CH2OH) for the Trost coupling reaction17) (Chart 3). This epoxide ring-opening reaction was good for many kinds of nucleophiles, such as Grignard reagents, and various enyne A-ring precursors for the synthesis of 2α-modified-1α,25(OH)2D3 were obtained.17–26) The representative product 2α-(3-hydroxypropoxy)-1α,25-dihydroxyvitamin D3 (O2C3) exhibited 1.8-times greater VDR binding affinity, and its binding mode was analyzed by X-ray co-crystallography.
When 1α,25(OH)2D3 binds to hVDR, six hydrogen bonds are formed between the three hydroxy groups of 1α,25(OH)2D3 and six amino acid residues (1α-OH with Ser237 and Arg274, 3β-OH with Ser278 and Tyr143, and 25-OH with His-305 and His-397) of the ligand binding domain (LBD) of human VDR (hVDR) (Chart 4). One of the most important hydrogen bonds connects the 1α-OH group to the Arg274 residue of LBD.27) The binding affinity of 1α,25(OH)2D3 to hVDR is 500–1000 times stronger than that of 25(OH)D3, which cannot form a hydrogen bond with Arg274, but 25(OH)D3 is still a weak VDR-agonist that benefits human health.12)
Left: Co-crystalline analysis with O1C3 proved the same binding mode as for O2C3.28) Right: The A-Ring part is magnified. W1–W3 indicate water molecules in the hVDR LBD-[1α,25(OH)2D3] complex,27) but W1 and W2 disappeared from the O2C3 complex, and the ω-hydroxy group occupied the position of W2. PDB number for the VDR complex with O2C3 is 2HAR, and for O1C3 Complex, 2HB7.28)
X-ray co-crystallographic analyses revealed that the 2α-methyl group of 2α-methyl-1α,25(OH)2D3 filled a hydrophobic cavity in LBD surrounded by Phe150, Leu233, and Ser237, and the binding affinity was raised up to 4-times using additional weak van der Waals contacts compared with 1α,25(OH)2D3.28) On the other hand, 2α-(3-hydroxypropyl)-1α,25(OH)2D3 (O1C3)16) and 2α-(3-hydroxypropoxy)-1α,25(OH)2D3 (O2C3),17) which have 3- and 1.8-times higher VDR binding affinity than 1α,25(OH)2D3, respectively, showed the ω-OH group of the 2α-substituent and 1α-OH formed pincer-type hydrogen bonds with Arg27428) (Chart 4). O1C3 and O2C3 have greater biological activities in vitro and in vivo,29,30) including transactivation of target genes, induction of HL-60 cell differentiation, and elevation of the rat serum calcium concentration, than those of the natural hormone 1α,25(OH)2D3.16,19,31)
1-3. Resistance against CYP24A1-Dependent Metabolism on O2C3CYP24A1 is an essential inactivation enzyme for both 1α,25(OH)2D3 and 25(OH)D3, and its expression is induced by the binding of 1α,25(OH)2D3 to VDR. CYP24A1 is responsible for the degradation and inactivation of 1α,25(OH)2D3 itself and 25(OH)D332,33) (Chart 5). This event is important physiologically for hormone deactivation to maintain calcium and phosphorus homeostasis; otherwise, elevated plasma levels of 1α,25(OH)2D3 would lead to hypercalcemia, hypercalciuria, and kidney stones. However, the 2α-substituted active vitamin D3 analogs O1C3 and O2C3 were resistant to metabolism by human CYP24A1, and the metabolism proceeded with only a few percent of 1α,25(OH)2D334,35) (Table 2). The survival of vitamin D analogs from CYP24A1-dependent metabolism may show a long half-life in target cells, prolonging their biological effects and benefitting therapy for vitamin D-related disorders. Basic biological activities of 2α-substituted 1α,25(OH)2D3 analogs, including O1C3 and O2C3, are summarized in Table 3.
| Substrate | kcat/Km | Relative kcat/Km (%)a) |
|---|---|---|
| 1α,25(OH)2D3 | 42 | 100 |
| O1C3 | 1.7 | 4.0 |
| O2C3 | 1.2 | 2.9 |
a) The kcat/Km value is an indicator of enzyme susceptibility.
 |
Epidemiologic research has shown that low plasma 25(OH)D3 levels increase the risk of cancer.36) The potent cellular modulatory properties of 1α,25(OH)2D3 suggest that it can be used as a therapeutic agent for cancer.11,12) However, a level above the physiological concentration causes serious hypercalcemia, and safer analogs of 1α,25(OH)2D3 that have less calcemic effects but retain anti-proliferative activity are attractive. 1α,25-Dihydroxy-19-norvitamin D3 was synthesized by Prof. DeLuca’s group at the University of Wisconsin in 1990, and they found that it induced the differentiation of human leukemia HL-60 cells in vitro with potency comparable to that of 1α,25(OH)2D3, but with little or no calcemic effect in animals.37) Deletion of the 10(19)-exomethylene group from 1α,25(OH)2D3 reduces the VDR binding affinity due to the loss of hydrophobic interaction with LBD of VDR, and the binding affinity of 19-nor-1α,25(OH)2D3 is only 30 and 17% of 1α,25(OH)2D3 for porcine VDR38) and calf thymus VDR,39) respectively. The 2α-(3-hydroxypropyl) group was introduced to the 19-norvitamin D3 skeleton to form pincer-type hydrogen bonds to Arg274 (Chart 4), as shown in the case of O2C3 and O1C3 to improve VDR binding affinity (Table 3). We expected the new compound to not exhibit a calcemic side-effect in vivo like 19-nor-1α,25(OH)2D3.
2-1. Synthesis of MART-10: 2α-(3-Hydroxypropyl)-1α,25-dihydroxy-19-norvitamin D32α-(3-Hydroxypropyl)-1α,25-dihydroxy-19-norvitamin D3 (MART-10) and its 2β-epimer MART-11 were synthesized using Julia–Kocienski olefination between an A-ring ketone starting from (–)-quinic acid and a CD-ring sulfone for C5-C6 double bond formation39,40) (Chart 6).
HPLC separation of C2-epimers yielded MART-10 and MART-11 at almost 1 : 1 ratio.
The 19-norvitamin D3 derivative MART-10 showed the same level of VDR binding affinity as natural 1α,25(OH)2D3 due to the additional hydrogen bond formation to Arg274 using the introduced 2α-(3-hydroxypropyl) group, but its 2β-counterpart MART-11 exhibited a much weaker affinity for VDR39) (only 3% of 1α,25(OH)2D3, Table 4). MART-10 exhibited potent antitumor activity, which was approx. 100- to 1000-times more active than 1α,25(OH)2D3 in prostate cancer,41–45) liver cancer,46,47) pancreatic cancer,48,49) breast cancer,50–53) anaplastic thyroid cancer,54) head and neck squamous carcinoma,55,56) cholangiocarcinoma,57,58) and neuroendocrine tumor59,60) cells in culture at 10–9–10–8 M level. MART-10 induced upregulation of p21 and p27 to arrest the cell cycle in the G0/G1 phase to inhibit cancer cell growth.47) MART-10 was an effective anti-angiogenesis agent and much more potent than 1α,25(OH)2D3 in vivo without inducing hypercalcemia.61,62) In a xenograft animal model inoculated with BxpC-3 human pancreatic carcinoma cells, MART-10 had a 10-fold greater antitumor effect than that of 1α,25(OH)2D3 without raising serum calcium. This in vivo experiment showed marked tumor regression with a dose of only 0.3µg/kg/d, twice/week for 3 weeks.48) The higher biological activity of MART-10 is due to tighter binding to VDR, which is similar to that of 1α,25(OH)2D3, and the 2α-(3-hydroxypropyl) group enhances the binding affinity for VDR and leads to increased resistance to CYP24A1 degradation inside cells (Table 4, kcat/Km). CYP24A1 is less than 1% effective for MART-10 metabolism compared with 1α,25(OH)2D3, which results in a longer MART-10 half-life in cells.34) MART-10 is approx. 100-fold more active than 1α,25(OH)2D3 in inhibiting PC-3 cell invasion since MART-10 exhibited more potent downregulation of matrix metalloproteinase-9 (MMP-9) expression at translational levels.43) As MMP-9 is an enzyme involved in the cell invasion pathway, the greater downregulation of MMP-9 activity may be responsible for the more potent anti-invasion effect observed in the presence of MART-10.42) Another unique property of MART-10 is that it has a lower binding affinity for DBP than 1α,25(OH)2D3.41) This lower binding affinity results in a higher concentration of bioavailable MART-10 in circulation for translocation to various target tissues42) (Table 4). We found that MART-10 was either non-calcemic or less calcemic than 1α,25(OH)2D3 in vivo.43,48) MART-11 with a 2β-(3-hydoxypropyl) group is also biologically active but less potent than MART-10.39,45) We also investigated transcriptional coactivators, such as hTIF-2 and hSRC-1, that MART-10 is likely to recruit after binding to VDR.63)
| VDR binding | Anti-proliferation | Anti-invasion | CYP24A1 kcat/Kma) | DBP binding | |
|---|---|---|---|---|---|
| 1α,25(OH)2D3 | 1 | 1 | 1 | 1 | 1 |
| MART-10 | 1 | 1000 | 100 | 1/500 | 1/25 |
| MART-11 | 0.03 | — | – | – | – |
a) The kcat/Km value is an indicator of enzyme susceptibility. —: not available.
In summary, MART-10 can exist in its free form in blood circulation with a long half-life due to low DBP binding affinity and insensitivity to CYP24A1-dependent metabolism, and binds to VDR tightly, triggers cell cycle arrest in the G0/G1 phase to inhibit cancer cell growth, and induces potent downregulation of MMP-9 to repress cancer cell invasion, and importantly, without inducing hypercalcemia in vivo.
The first specific human VDR antagonist TEI-9647 and its (23R)-epimer TEI-9648 were discovered by the Teijin research group in 1999 by modifying the side chain of a 1α,25-dihydroxyvitamin D3-26,23-lactone metabolite derived from 1α,25(OH)2D3 through CYP24A1-dependent metabolism64) (Chart 7). TEI-9647 has an α-methylene-γ-butyrolactone part on the side chain, which is a dehydrated form of the natural lactone metabolite, and TEI-9647 with the 23S-configration is more potent (IC50 6.3–8.3nM) than TEI-9648 (IC50 119nM) with the 23R-configration (Chart 7). The exomethylene lactone structure is indispensable for the antagonistic activity of TEI-9647.65,66) The LBD of human VDR contains two cysteine residues, Cys403 and Cys410, which are close to the CD-ring side chain of 1α,25(OH)2D3 in the hVDR-ligand complex based on X-ray studies.27) These two cysteines are important for the antagonistic effect of TEI-9647 on VDR.66) As shown in Chart 7, the nucleophilic thiol groups of Cys403 and Cys410 can attack the α-methylene-γ-lactone of TEI-9647 via 1,4-addition to give a corresponding cysteine-covalent adduct according to matrix-assisted laser desorption/ionization-time of flight (MALDI–TOF) MS.67) When the exomethylene moiety is located at a more favorable position to react with Cys403 or Cys410 after ligand binding, the new TEI-9647 analog should exhibit stronger VDR antagonistic activity. We synthesized almost 150 analogs of TEI-9647 based on the 2α-functionalization concept (section 1)68) to potentiate VDR binding affinity69) and an additional C24-modification70–78) to improve the position of the exomethylene group in LBD for enhancing the reaction with Cys residues.
TEI-9647 is a stronger antagonist than TEI-9648, and modification at R1–R3 may improve the VDR-antagonism.
We investigated C2α and C24-double modification of TEI-9647 and identified a potent compound, (23S,24S)-25-dehydro-2α-(3-hydroxypropoxy)-24-propyl-1α-hydroxyvitamin D3-26,23-lactone (NS-74c), which showed 850-fold higher antagonistic activity (IC50=7.4pM) than that of the original TEI-9647 (IC50=6.3nM).71) The synthetic route is shown in Chart 8. The lactone CD-ring was synthesized using Oshima's chromium-mediated syn-selective allylation to the CD-ring side-chain aldehyde followed by spontaneous lactonization, and the (23S,24S)-isomer was coupled with the A-ring precursor enyne prepared from d-glucose to yield NS-74c.
Based on the above synthetic strategy, a total of approx. 150 kinds of TEI-9647 and TEI-9648 analogs were obtained.71) The relative VDR binding affinity compared with 1α,25(OH)2D3 and VDR antagonistic activity of the 10 representative analogs are listed in Table 5. Interestingly, potencies between VDR binding affinity and VDR antagonistic activity were not parallel, and R2 and R3 were sensitive to the antagonism as expected.
 |
VDR antagonists may serve as potent therapeutic agents against some diseases caused by VDR-hypersensitivity to 1α,25(OH)2D3, such as Paget’s bone disease, the most extreme example of disordered bone remodeling and second most common bone disease after osteoporosis in Anglo–Saxons.79)
Nitrogen atoms of an azole ring, instead of the hydroxy group of O1C3 and O2C3, were subsequently used to create additional pincer-type hydrogen bonds with the VDR Arg274 residue. We studied the effects of azole rings, such as tetrazole, triazole, and imidazole, at the 2α-position of 1α,25(OH)2D3 on binding to hVDR and biological activities in vitro and in vivo. The length of an alkyl linker at the C2α-position was decided as an ethyl group for effective pincer-type hydrogen bond formation, and X-ray co-crystallographic analysis of the hVDR-(synthetic-ligand: AH-1, see below) complex showed the expected hydrogen bond networks (Chart 9).
The A-ring part is magnified. PDB number is 4ITE.
Synthesis of the A-ring precursor of AH-1 was started from methyl α-d-glucopyranoside and tetrazole was introduced to the 2α-ethyl terminal by the Mitsunobu reaction to yield a desired 2-alkylated enyne as the major product with its 1-alkylated isomer (Chart 10). The chemical structures of these two isomeric enynes were determined by comparing the 1H- and 13C-NMR chemical shifts of the correlated methylene and methine H and C atoms. The major product enyne was connected to the CD-ring bromoolefin by Trost coupling, as shown in Chart 3, to afford AH-1.80) Based on this synthetic route, 1,2,3-, 1,2,4-triazoles and the imidazole ring were incorporated into the C2α-position to yield 6 kinds of 2α-heteroarylethylated 1α,25(OH)2D3 including regio-isomers (Chart 9); however, those other than AH-1 were weak VDR-binders and showed poor effects on osteocalcin promoter transactivation.80)
Among six 2α-heteroarylethylated active vitamin D3, 2α-[2-(tetrazol-2-yl)ethyl]-1α,25(OH)2D3 (AH-1) exhibited a potent binding affinity for hVDR [67% of 1α,25(OH)2D3] and increased transactivation activity in human osteosarcoma (HOS) cells (EC50 10pM) compared with that of 1α,25(OH)2D3 (EC50 26pM). AH-1 was evaluated regarding its in vivo therapeutic effect using ovariectomized (OVX) rats as an osteoporosis model. AH-1 was safe and increased the bone mineral density (BMD) of the spine (L4–L5) at a low dose of 0.02µg/kg/d, 5 times a week for 4 weeks without significant side effects of hypercalcemia, and it had a much more favorable therapeutic effect on osteoporosis than that of 1α,25(OH)2D3, tending to elevate the serum Ca2+ level80–82) (Table 6).
| dose (mg/kg/d) |
BMD (g/cm2)a) |
serum Ca2+ (mg/dL) |
|
|---|---|---|---|
| Control (sham) | – | 0.231–0.245 | 9.71 |
| Control (OVX) | – | 0.209 | 9.14–9.25 |
| 1α,25(OH)2D3 | 0.025 | 0.218 | 10.1 |
| AH-1 | 0.020 | 0.223 | 9.66 |
a) Bone mineral density (BMD) is measured using a dual X-ray absorption meter.
We considered that this potent bone-forming activity in vivo would be caused by the long half-life of AH-1 because AH-1 had a 2α-substituent like O1C3 or O2C3, being resistant to CYP24A1-deactivating metabolism. However, CYP24A1 metabolized AH-1 with 31% efficiency of 1α,25(OH)2D3, which suggested that AH-1 was not as resistant to CYP24A1-dependent metabolism as O1C3 and O2C3 (approx. 2–4%).34,35) According to HPLC analysis, the major metabolite of AH-1 by CYP24A1 was (24R)-24-hydroxy-AH-1, which was synthesized separately.83) The (24R)-24-hydroxy-AH-1 metabolite showed similar VDR binding affinity and human leukemia cell (HL-60 cell) differentiation activity to those of AH-1.84) In contrast, 1α,25(OH)2D3 was metabolized by multistep monooxygenation reactions of CYP24A1 (Chart 5), and the end-products lactone and calcitroic acid lost their binding affinity for VDR. Therefore, the greater therapeutic effects of AH-1 compared with those of 1α,25(OH)2D3 in vivo using OVX rats may be due to the higher VDR binding affinity of the AH-1 CYP24A1-dependant metabolite, (24R)-24-hydroxy-AH-1.84)
Interestingly, AH-1 showed potent chemotherapeutic effects on VDR(Arg270Leu) hereditary vitamin D-dependent rickets model rats, in which the mutated rat VDR(Arg270Leu) was an ortholog of human VDR(R274L) isolated from patients with hereditary vitamin D-dependent rickets (VDDR) type II. VDDR-II patients are barely responsive to physiological doses of 1α,25(OH)D3 since there are no Arg274 residues in the LBD of VDR to form strong hydrogen bonds with the 1α-OH group of 1α,25(OH)D3. AH-1 had much higher affinity even for rat VDR(Arg270Leu) than both 25(OH)D3 and 1α,25(OH)2D3. Marked osteogenic activity of AH-1 was observed in VDR(Arg270Leu) rats. A 40-fold lower dose of AH-1 than that of 25(OH)D3 was effective in resolving symptoms of rickets in VDR(Arg270Leu) rats. Therefore, AH-1 may be promising in therapy for VDDR-II with VDR(Arg270Leu).85,86)
4-3. How about Propyl Linker?We also synthesized four 2-[3-(tetrazolyl)propyl]-1α,25-dihydroxy-19-norvitamin D3 analogs, but these 19-nor analogs with a propyl group instead of the ethyl group as a linker had weak transactivation activity through hVDR in HOS cells (EC50 7.3nM, when 1α,25(OH)2D3 0.23nM).87) These analogs were used as synthetic chemical probes that differentiate vitamin D activities between VDR activation and sterol regulatory element-binding protein (SREBP) inhibitory activity (see Section 7).88)
As shown in Chart 1, tachysterol is produced by previtamin D3 photolysis in the skin. We attempted to study the synthesis and evaluate the biological activity of 1α,25-dihydroxy-19-nortachysterol; however, synthesized 1α,25-dihydroxy-19-nortachysterol was unstable after deprotection and gradually decomposed even under neutral conditions at room temperature, and so it was impossible to evaluate its biological activity. To stabilize the tachysterol skeleton, C14 was epimerized to construct a cis-CD-ring system since cis-hydrindane shows a predilection for the C8,9-endo double bond rather than the C7,8-exo double bond,89) and we considered that 14-epimerization was essential to obtain stable tachysterol analogs to test their biological activity. It is known that 14-epimerization of vitamin D3 shifts its previtamin D3 form in the thermodynamic equilibrium at a ratio of approx. 95:5, which is reverse in Chart 1.90–93)
We identified proton-catalyzed isomerization of synthesized 14-epi-1α,25-dihydroxy-19-norprevitamin D3 skeletons (cis-form) to the stable 14-epi-1α,25-dihydroxy-19-nortachysterols (trans-form).94) We synthesized 14-epi-1α,25-dihydroxy-19-nortachysterol and its 2-substituted analogs using a Stille coupling reaction between the A-ring precursor vinylstannane and CD-ring triflate, i.e., the A-ring precursor enynes were converted to the corresponding vinylstannanes, Stille coupling was applied to connect each vinylstannane and CD-ring triflate, and the coupled products were deprotected (Chart 11). The target molecules of 14-epi-1α,25-dihydroxy-19-nortachysterol, its 2-exomethylene-, 2-(3-hydroxypropyl)-, and 2-(3-hydroxypropoxy)-substituted analogs were obtained.94–99) The 2-(3-hydroxypropyl)-substituted analog was separated by HPLC for the 2α- and 2β-diastereomers.96) For further modification of the A-ring part, regioselective hydrogenation of the 2-exomethylene analog at the terminal alkene was accomplished with Wilkinson’s catalyst to give 2-methyl substituted diastereomers, which were separated by HPLC.94)
After the final deprotection, HPLC separation for 2α- and 2β-isomers was performed, if necessary.
VDR binding affinity of the seven synthesized 14-epi-1α,25-dihydroxy-19-nortachysterol analogs was tested, and the data are shown in Chart 12 with their structures. For 14-epi-19-nortachysterol, the 2α-long substituent, 3-hydroxypropyl, and 3-hydroxypropoxy groups, disturbed binding to hVDR, and the 2β-(3-hydroxypropyl) A-ring showed higher binding affinity (48%) than its 2α-counterpart (5.6%) and the non-substituted compound (15%). This was opposite to the effects of 1α,25(OH)2D3 on VDR binding, including O1C3 and O2C3. The 2α-terminal hydroxy group of O1C3 forms a hydrogen bond network with Arg274 and consequently causes higher binding affinity.28) Also, we identified 14-epi-19-nortachysterol binding configurations in LBD of hVDR by X-ray co-crystallographic analysis,94) and the X-ray data of hVDR-[2α- and 2β-14-epi-(3-hydroxypropyl)-1α,25-dihydroxy-19-nortachysterol] complexes96) showed a C5,6-s-trans and C7,8-s-trans triene configuration in LBD with an unprecedented binding mode for seco-steroid analogs94) (Chart 13). In this binding mode, the 2α-(3-hydroxypropyl) group was less effective for constructing a hydrogen bond network with Arg274 than the 2β-counterpart, and 2α-(3-hydroxypropyl) and 2α-(3-hydroxypropoxy) substituents led to steric hindrance in LBD of hVDR96) (Chart 13).
PDB accession numbers are 5YT2 for 2α and 5YSY for 2β.
1α,15α,25-Trihydroxyvitamin D3 and its analogs were synthesized for the first time to investigate the effects of the C15-functionalized CD-ring on biological activity concerning the agonistic positioning of helix-3 and helix-12 of VDR. Although oogoniol and pavoninin were known as 15-hydroxysteroid natural products, the C15-functionalized B-seco-steroid was unidentified.
The known ketone was converted to the 15α,16α-epoxyhydrindan derivative whose acetate was crystallized to confirm the absolute configuration by X-ray crystallographic analysis (Chart 14). The subsequent Wittig reaction gave ethylidene, and a side-chain was successfully introduced via 1,4-addition of the magnesium cyanocuprate reagent from 5-bromo-2-methyl-2-pentanol to ethylidene epoxide. The 15α-hydroxy-16-ene-CD-ring was obtained with the natural (20R)-configuration as a major product (20R:20S=11:1). After hydrogenolysis, protected 8-oxo CD-ring was coupled with A-ring phosphine oxide to give 1α,15α,25-trihydroxyvitamin D3100) (Chart 14). With our synthetic route, the new C15-substituted 16-ene-active vitamin D3 analogs were also available, without the hydrogenolysis step, to compare biological activity.100,101)
The 16-ene-derivatives were also synthesized by skipping the hydrogenation step.
The 15-substituent interfered with binding to VDR (R=H: 13%, CH3: 1.5%, when 1α,25(OH)2D3 100%); however, osteocalcin promoter transactivation activity in HOS cells was equipotent to 1α,25(OH)2D3100) (Table 7). The 16-ene-CD-ring improved the binding (R=H: 65%).100,101) The planar Δ16-sp2 carbons may cancel the hindrance of the C15-substituent. Combination of A-ring 2α-methylation and 16-ene-15-hydroxylation resulted in almost 3-times higher VDR binding affinity (R=H: 278%) and greater osteocalcin promoter transactivation activity than 1α,25(OH)2D3 in HOS cells (EC50 0.037nM).101) X-ray co-crystallographic analysis of 15α-methoxy-1α,25(OH)2D3 proved that 0.6Å shifts of the CD-ring and shrinkage of the side chain were necessary to maintain the position of the 25-hydroxy group for appropriate interaction with His305 and His397 residues of LBD of VDR.100)
| VDR binding affinity (%) |
Osteocalcin transactivation activity in HOS/SF cells (EC50, nM) |
|
|---|---|---|
| 1α,25(OH)2D3 | 100 | 0.094 |
| R=H | 13.3 | 0.12 |
| R=CH3 | 1.5 | 0.12 |
| 16-ene, R=H | 65 | 0.024 |
| 16-ene, 2α-CH3, R=H | 278 | 0.037 |
In 2017, it was reported that 25(OH)D3 inhibited the activation of sterol regulatory element-binding proteins (SREBPs), a family of master transcription factors for lipogenesis, by interacting with SREBP cleavage-activating protein (SCAP), a specific escort protein for SREBP.102) The 25(OH)D3 binding to SCAP induced degradation of both SCAP and SREBP to block the expression of SREBP-controlled lipogenic genes, and this effect was independent of VDR. A low µM dose of 25(OH)D3 is required to inhibit cellular lipogenesis by limiting the expression of SREBP-responsive genes in cultured cells. However, 25–100nM of 25(OH)D3 is observed in healthy individual blood. 25(OH)D3 is a weak ligand for VDR, and a level exceeding the physiological concentration of 25(OH)D3 is harmful as it can cause VDR-mediated excess calcemic action such as hypercalcemia; therefore, VDR-silent vitamin D analogs with potent SREBP inhibitory activity are attractive for developing therapeutic candidates of fatty liver diseases and cancers, which involve abnormal or increased lipid metabolism.
About 90 A-ring-replaced vitamin D molecules, including suprasterols (Chart 1), were synthesized and evaluated regarding VDR-silent and SREBP/SCAP inhibitory activity in cultured CHO-K1 cells, and 14 selected compounds with excellent selectivity of SREBP over VDR were tested in mice103) (Chart 15). Four side-chain fluorinated compounds (red asterisk in Chart 15) significantly suppressed the expression of two representative SREBP-response genes, ACC1 and FASN, compared with the vehicle control; however, the upper two side-chain hexafluoro-derivatives (with a red asterisk) showed elevated levels of serum calcium, indicating that these compounds or their metabolites act on VDR in vivo.
Four side-chain fluorinated compounds (with a red asterisk) significantly suppressed the expression of two representative SREBP-response genes, ACC1 and FASN, in the mouse liver.
Finally, KK-052 was prioritized for long-term testing of fatty liver prevention effects in vivo. Six-week-old ob/ob mice were treated with 10mg/kg KK-052, 10mg/kg 25(OH)D3, or vehicle control 5 times per week for 4 weeks. All ob/ob mice treated with 25(OH)D3 developed hypercalcemia and were euthanized at 1 week, while ob/ob mice treated with KK-052 appeared healthy until the end of the experiments. Histology of the liver of vehicle-treated ob/ob mice revealed marked liver steatosis, and such conditions were reduced by treatment with KK-052 without inducing hypercalcemia. The SREBP inhibitory activity of KK-052 with IC50 0.7µM was attributable to its ability to decrease the levels of endogenous SREBP and SCAP, resembling the action of 25(OH)D3.103,104)
7-2. Improved Synthesis of KK-052Tetrazole 1-substituted KK-052 was the minor product of the original synthesis using the Mitsunobu reaction (Chart 16A), and improved scalable synthesis of KK-052 was demonstrated, in which [3+2] cycloaddition between benzimidoyl chloride and the azide ion produced the tetrazole ring effectively105) (Chart 16B). Recently, more potent KK-052-related compounds with selective SREBP inhibitory activity in vitro were synthesized.106) Further in vitro experiments advanced knowledge of VDR-silent vs. SREBP inhibitory activity involving vitamin D analogs.88)
KK-052 contains a 24,24-difluoro group, without which it is not effective in vivo. Introduction of the fluorine atom to the vitamin D3 side-chain has been tested since the late 1970s,107–109) and the most successful example is 26,26,26,27,27,27-hexafluoro-1α,25-dihydroxyvitamin D3 (falecalcitriol), which has been clinically applied to treat secondary hyperparathyroidism. The introduced fluorine affects VDR binding affinity and efficacy for CYP24A1-dependant metabolism.
A new synthetic method to gain (22R)-22-fluoro-, (22S)-22-fluoro-, 22,22-difluoro-,110) (23R)-23-fluoro-, (23S)-23-fluoro-,111) 23,23-difluoro-,112) (24R)-24-fluoro-, (24S)-24-fluoro-,113,114) 24,24-difluoro-,115) 26,27-difluoro-, 26,26,27,27-tetrafluoro-,116) and 26,26,26,27,27,27-hexafluoro-CD-rings117) was established, and these were connected to the A-ring part for fluorinated 25(OH)D3 synthesis (Chart 17). Their VDR binding affinity, osteocalcin promoter transactivation ability in HOS cells, and resistance to CYP24A1-dependant metabolism were analyzed (Table 8). The fluorinated position, stereochemistry, and numbers of introduced fluorine atoms were sensitive to their biological activity. Interestingly, C22- and C23-fluorination afforded some weak VDR-ligands; however, 26,26,26,27,27,27-hexafluoro-25(OH)D3 was the most potent ligand.
| Ligand | VDR binding affinity (%) | Osteocalcin transactivation activity in HOS/SF cells (EC50, nM) |
Metabolic Activities of hCYP24A1 (nmol/min/nmol-P450) |
|---|---|---|---|
| 25(OH)D3 | 100 | 319 | 4.56±0.38a) 6.2±1.5 |
| (22R)-F-25(OH)D3 | <0.5 | 464 | 8.2±0.9 |
| (22S)-F-25(OH)D3 | 15 | 229 | 9.3±1.1 |
| 22,22-F2-25(OH)D3 | 1 | 302 | 4.4±0.5 |
| (23R)-F-25(OH)D3 | 310 | 117 | 4.8±1.5 |
| (23S)-F-25(OH)D3 | 7.7 | 283 | 2.7±0.27 |
| 23,23-F2-25(OH)D3 | 8.2 | 253 | 0.77±0.15 |
| (24R)-F-25(OH)D3 | 73 | 190 | 1.6±0.5 |
| (24S)-F-25(OH)D3 | 64 | 178 | 4.8±1.5 |
| 24,24-F2-25(OH)D3 | 180 | 89.0 | 0.53±0.12 |
| 26,27-F2-25(OH)D3 | 95 | 294 | 3.16±0.15b) |
| 26,27-F4-25(OH)D3 | 710 | 179 | 1.82±0.14b) |
| 26,27-F6-25(OH)D3 | 5700 | 16.6 | 0.59±0.10b) |
b) Based on a).
Further derivatization of the A-ring part with fluorinated side-chains led to safe and markedly potent anti-inflammatory vitamin D analogs that improve symptoms in psoriasis,118,119) inflammatory bowel disease, and diabetic retinopathy model mice. These data will be published elsewhere in due course.
We have confirmed selective VDR-mediated actions, especially derivatives that do not increase blood calcium levels as a side-effect, and non-VDR actions (SREBP/SCAP inhibitory effect) by derivatizing vitamin D3 in vitro and in vivo, and the results of the research may lead to the development of vitamin D analogs that can be applied to cancer, hepatic steatosis, type 2 diabetes, obesity, inflammatory disease, psoriasis, and bone disorders such as Pagetʼs bone disease, osteoporosis, rickets, and osteomalacia including high-quality bone formation. We have endeavored to design drug-like molecules according to therapeutic purposes and perform synthetic research to realize the molecular construction. We have also conducted basic drug discovery research based on the vitamin D skeleton with many collaborators. We are continuing our research on derivatives for treating neurodegenerative disorders and ophthalmological diseases. The possibilities of the derivatization of vitamin D for disease treatment, including unknown new target proteins, are endless.
I would like to express my sincere gratitude to all my excellent collaborators, including those cited in the references, for their guidance and support. In particular, I would like to express my deepest appreciation to Professor Toshiyuki Sakaki (currently Professor Emeritus and Special Research Professor) of Toyama Prefectural University, Professor Motonari Uesugi of Kyoto University, Professor Tai C. Chen (currently Professor Emeritus) of Boston University School of Medicine (especially for MART-10 studies), Teijin Institute for Bio-medical Research Teijin Pharma Limited, and Professor Toshio Okano (currently Professor Emeritus) of Kobe Pharmaceutical University, and Pharmaceutical Chemistry, Faculty of Pharmaceutical Sciences of Teikyo University, with whom I have worked together on research for many years, and I had a great opportunity to advance the frontiers of vitamin D research field with them. This study was supported by Grants-in-Aid from the Japan Society for the Promotion of Science (Nos. 13672230, 15590021, 17590012, 19590016, 21590022, 24590022, 15K07869, 18K06556, and 23K06029) and the Uehara Memorial Foundation. I am also grateful to AMED-CREST, AMED (in Prof. Uesugi’s group, October 2014–March 2020) for the KK-052 research.
The author declares no conflict of interest.
This review of the author’s work was written by the author upon receiving the 2024 Pharmaceutical Society of Japan Award for Divisional Scientific Contribution.
