Differential Metabolic Stability of 4α,25- and 4β,25-Dihydroxyvitamin D3 and Identification of Their Metabolites

Vitamin D3 (1) is metabolized by various cytochrome P450 (CYP) enzymes, resulting in the formation of diverse metabolites. Among them, 4α,25-dihydroxyvitamin D3 (6a) and 4β,25-dihydroxyvitamin D3 (6b) are both produced from 25-hydroxyvitamin D3 (2) by CYP3A4. However, 6b is detectable in serum, whereas 6a is not. We hypothesized that the reason for this is a difference in the susceptibility of 6a and 6b to CYP24A1-mediated metabolism. Here, we synthesized 6a and 6b, and confirmed that 6b has greater metabolic stability than 6a. We also identified 4α,24R,25- and 4β,24R,25-trihydroxyvitamin D3 (16a and 16b) as metabolites of 6a and 6b, respectively, by HPLC comparison with synthesized authentic samples. Docking studies suggest that the β-hydroxy group at C4 contributes to the greater metabolic stability of 6b by blocking a crucial hydrogen-bonding interaction between the C25 hydroxy group and Leu325 of CYP24A1.

In 2011, Thummel et al. found a novel vitamin D metabolite, 4β,25-dihydroxyvitamin D 3 (6b), which is formed from 2 by hydroxylation at C4β [12]. The conversion of compound 2 to compound 6 may be involved in drug-induced osteomalacia. Notably, in vitro studies revealed that CYP3A4 generated 4α,25-dihydroxyvitamin D 3 (6a) and 6b at a ratio of 1:2. Interestingly, however, analysis of human serum showed the presence of only 6b, with no detection of 6a. This intriguing result led us to propose that differential metabolic stability between 6a and 6b might explain this discrepancy. We hypothesized that a difference in metabolic stability between 6a and 6b would account for this finding. Here, we synthesized 6a and 6b, and evaluated their metabolic stability in the presence of CYP24A1, which plays a major role in vitamin D metabolism. We also identified metabolites of 6 generated by CYP24A1 by means of HPLC comparison with synthesized authentic samples. D3 (6b), which is formed from 2 by hydroxylation at C4β [12]. The conversion of compound 2 to compound 6 may be involved in drug-induced osteomalacia. Notably, in vitro studies revealed that CYP3A4 generated 4α,25-dihydroxyvitamin D3 (6a) and 6b at a ratio of 1:2. Interestingly, however, analysis of human serum showed the presence of only 6b, with no detection of 6a. This intriguing result led us to propose that differential metabolic stability between 6a and 6b might explain this discrepancy. We hypothesized that a difference in metabolic stability between 6a and 6b would account for this finding. Here, we synthesized 6a and 6b, and evaluated their metabolic stability in the presence of CYP24A1, which plays a major role in vitamin D metabolism. We also identified metabolites of 6 generated by CYP24A1 by means of HPLC comparison with synthesized authentic samples. Scheme 1. Metabolic pathways of vitamin D3 (1).
(S)-3,4-bis((tert-butyldimethylsilyl)oxy)butyl benzoate (S1): To a solution of diol 9 [13] (300 mg, 1.4 mmol) in CH 2 Cl 2 (7 mL) was added 2,6-lutidine (0.44 mL, 3.7 mmol) and tertbutyldimethylsilyl triflate (0.82 mL, 3.6 mmol) at 0 • C. The resulting mixture was allowed to warm to room temperature and stirred for 30 min. The reaction mixture was quenched with H 2 O, and the aqueous layer was extracted with CH 2 Cl 2 three times. The combined organic layer was washed with brine, dried over MgSO 4 , filtered, and concentrated in vacuo. The residue was purified by flash chromatography on silica gel (n-hexane/ethyl acetate = 20:1) to give the protected product S1 (627 mg, 99%) as a colorless oil. [ (S)-3-((tert-butyldimethylsilyl)oxy)-4-hydroxybutyl benzoate (10): To a solution of diol S1 (627 mg, 1.4 mmol) in MeOH/THF = 1:1 (17 mL) was added (±)-CSA (40 mg, 0.17 mmol) at room temperature. The resulting mixture was stirred at the same temperature for 1 h. The reaction mixture was quenched with sat. NaHCO 3 aq, and the aqueous layer was extracted with CH 2 Cl 2 three times. The combined organic layer was washed with brine, dried over MgSO 4 , filtered, and concentrated in vacuo. The residue was purified by flash chromatography on silica gel (n-hexane/ethyl acetate = 10:1) to give 10 (77 mg, 54%) as a colorless oil. [ (3S,4R)-3-((tert-butyldimethylsilyl)oxy)-4-hydroxy-6-(trimethylsilyl)hex-5-yn-1-yl benzoate (11b): To a solution of oxalic acid dichloride (0.48 mL, 5.6 mmol) in CH 2 Cl 2 (7 mL) was added dimethyl sulfoxide (1.0 mL, 14.0 mmol) at -78 • C. The resulting mixture was stirred at the same temperature for 10 min, then a solution of 10 in CH 2 Cl 2 (0.4 M, 7 mL, 2.78 mmol) and triethylamine (3.9 mL, 27.8 mmol) were added dropwise at the same temperature. The resulting mixture was allowed to warm to room temperature and stirred for 1 h. The reaction mixture was quenched with H 2 O, and the aqueous layer was extracted with CH 2 Cl 2 three times. The combined organic layer was washed with brine, dried over MgSO 4 , filtered, and concentrated in vacuo. The residue was purified by flash chromatography on silica gel (n-hexane/ethyl acetate = 1:1) to give the aldehyde as a yellow oil. The crude residue was used in the subsequent step without purification. To a solution of trimethylsilylacetylene (1.0 mL, 7.2 mmol) in THF (12 mL) was added n-butyllithium (2.6 M in hexane; 2.3 mL, 6.0 mmol) at -78 • C. The resulting mixture was stirred at the same temperature for 30 min, then a solution of the aldehyde in THF (0.3 M, 9.3 mL, 2.78 mmol) was added dropwise at the same temperature. The resulting mixture was stirred at the same temperature for 15 min. The reaction mixture was quenched with sat. NH 4 Cl aq, and the aqueous layer was extracted with ethyl acetate three times. The combined organic layer was washed with brine, dried over MgSO 4 , filtered, and concentrated in vacuo. The residue was purified by flash chromatography on silica gel (n-hexane/ethyl acetate = 30:1) to give 11b (910 mg, 90%) as a colorless oil. (S)-3-((tert-butyldimethylsilyl)oxy)-4-oxo-6-(trimethylsilyl)hex-5-yn-1-yl benzoate (S2): To a solution of 11b (100 mg, 0.24 mmol) and pyridine (0.38 mL) in CH 2 Cl 2 (19 mL) was added Dess-Martin periodinane (404 mg, 0.95 mmol) at 0 • C. The resulting mixture was allowed to warm to room temperature and stirred for 30 min. The reaction mixture was quenched with sat. Na 2 S 2 O 3 aq, and the aqueous layer was extracted with CH 2 Cl 2 three times. The combined organic layer was washed with brine, dried over MgSO 4 , filtered, and concentrated in vacuo. The residue was purified by flash chromatography on silica gel (n-hexane/ethyl acetate = 10:1) to give S2 (94 mg, 94%) as a colorless oil. [ (3S,4S)-3-((tert-butyldimethylsilyl)oxy)-4-hydroxy-6-(trimethylsilyl)hex-5-yn-1-yl benzoate (11a): To a solution of S2 (94 mg, 0.22 mmol) in THF (11 mL) was added L-selectride (1.0 M in THF, 0.40 mL, 0.40 mmol) at -78 • C. The resulting mixture was stirred at the same temperature for 1 h. The reaction mixture was quenched with sat. NH 4 Cl aq, and the aqueous layer was extracted with ethyl acetate three times. The combined organic layer was washed with brine, dried over MgSO 4 , filtered, and concentrated in vacuo. The residue was purified by flash chromatography on silica gel (n-hexane/ethyl acetate = 10:1) to give 11a (94 mg, 99%) as a colorless oil. (3S,4S)-3,4-bis((tert-butyldimethylsilyl)oxy)-6-(trimethylsilyl)hex-5-yn-1-yl benzoate (S3a): To a solution of 11a (115 mg, 0.27 mmol) in CH 2 Cl 2 (3 mL) was added 2,6-lutidine (98 µL, 0.82 mmol) and tert-butyldimethylsilyl triflate (94 µL, 0.41 mmol) at 0 • C. The resulting mixture was allowed to warm to room temperature and stirred for 30 min. The reaction mixture was quenched with H 2 O, and the aqueous layer was extracted with CH 2 Cl 2 three times. The combined organic layer was washed with brine, dried over MgSO 4 , filtered, and concentrated in vacuo. The residue was purified by flash chromatography on silica gel (n-hexane/ethyl acetate = 40:1) to give S3a (143 mg, 98%) as a colorless oil. (3S,4R)-3,4-bis((tert-butyldimethylsilyl)oxy)-6-(trimethylsilyl)hex-5-yn-1-yl benzoate (S3b): To a solution of 11b (264 mg, 0.63 mmol) in CH 2 Cl 2 (6 mL) was added 2,6-lutidine (0.22 mL, 1.9 mmol) and tert-butyldimethylsilyl triflate (0.22 mL, 0.94 mmol) at 0 • C. The resulting mixture was allowed to warm to room temperature and stirred for 30 min. The reaction mixture was quenched with H 2 O, and the aqueous layer was extracted with CH 2 Cl 2 three times. The combined organic layer was washed with brine, dried over MgSO 4 , filtered, and concentrated in vacuo. The residue was purified by flash chromatography on silica gel (n-hexane/ethyl acetate = 40:1) to give S3b (334 mg, 99%) as a colorless oil. (3S,4S)-3,4-bis((tert-butyldimethylsilyl)oxy)hex-5-yn-1-ol (12a): To a solution of S3a (143 mg, 0.27 mmol) in MeOH (0.9 mL) was added K 2 CO 3 (148 mg, 1.1 mmol) at room temperature. The resulting mixture was stirred at the same temperature for 1 h. The reaction mixture was quenched with H 2 O, and the aqueous layer was extracted with ethyl acetate three times. The combined organic layer was washed with brine, dried over MgSO 4 , filtered, and concentrated in vacuo. The residue was purified by flash chromatography on silica gel (n-hexane/ethyl acetate = 30:1) to give 12a (68 mg, 71%) as a colorless oil.  (3S,4R)-3,4-bis((tert-butyldimethylsilyl)oxy)hex-5-yn-1-ol (12b): To a solution of S3b (206 mg, 0.39 mmol) in MeOH (1.3 mL) was added K 2 CO 3 (213 mg, 1.5 mmol) at room temperature. The resulting mixture was stirred at the same temperature for 1 h. The reaction mixture was quenched with H 2 O, and the aqueous layer was extracted with ethyl acetate three times. The combined organic layer was washed with brine, dried over MgSO 4 , filtered, and concentrated in vacuo. The residue was purified by flash chromatography on silica gel (n-hexane/ethyl acetate = 30:1) to give 12b (104 mg, 76%) as a colorless oil. , and iodine (308 mg, 1.2 mmol) at -20 • C. The resulting mixture was stirred at the same temperature for 30 min. The resulting mixture was allowed to warm to room temperature and stirred for 20 min. The reaction mixture was quenched with sat. Na 2 S 2 O 3 aq, and the aqueous layer was extracted with ethyl acetate three times. The combined organic layer was washed with brine, dried over MgSO 4 , filtered, and concentrated in vacuo. The residue was purified by flash chromatography on silica gel (n-hexane) to give S4b (149 mg, 84%) as a yellow oil. (4S,5R)-4,5-bis((tert-butyldimethylsilyl)oxy)hept-6-ynenitrile (13a). To a solution of S4a (54 mg, 0.11 mmol) in DMF (0.3 mL) was added sodium cyanide (8.4 mg, 0.17 mmol) at room temperature. The resulting mixture was allowed to warm to 90 • C and stirred for 20 min. The reaction mixture was quenched with sat. NaHCO 3 aq, and the aqueous layer was extracted with ethyl acetate three times. The combined organic layer was washed with brine, dried over MgSO 4 , filtered, and concentrated in vacuo. The residue was purified by flash chromatography on silica gel (n-hexane/ethyl acetate = 80:1) to give 13a (42 mg, 99%) as a yellow oil. [ (4S,5R)-4,5-bis((tert-butyldimethylsilyl)oxy)hept-6-ynenitrile (13b): To a solution of S4b (133 mg, 0.28 mmol) in DMF (0.7 mL) was added sodium cyanide (21.0 mg, 0.43 mmol) at room temperature. The resulting mixture was allowed to warm to 90 • C and stirred for 20 min. The reaction mixture was quenched with sat. NaHCO 3 aq, and the aqueous layer was extracted with ethyl acetate three times. The combined organic layer was washed with brine, dried over MgSO 4 , filtered, and concentrated in vacuo. The residue was purified by flash chromatography on silica gel (n-hexane/ethyl acetate = 80:1) to give 13b (105 mg, 99%) as a yellow oil. 14 mL, 0.14 mmol) at 0 • C. The resulting mixture was stirred at the same temperature for 30 min. To the reaction mixture was added 2-propanol (0.098 mL), silica gel (200 mg), and water (1 mL) at room temperature. The resulting mixture was stirred at the same temperature for 30 min. The reaction mixture was filtered through a pad of Celite and concentrated in vacuo. The residue was purified by flash chromatography on silica gel (n-hexane/ethyl acetate = 10:1) to give the aldehyde as a yellow oil. The crude residue was used in the subsequent step without purification. To a solution of methyltriphenylphosphonium iodide (240 mg, 0.39 mmol) in THF (0.6 mL) was added sodium bis(trimethylsilyl)amide (1.9 M in THF, 0.29 mL, 0.56 mmol) at 0 • C. The resulting mixture was stirred at the same temperature for 30 min, then a solution of the aldehyde in THF (0.2 M, 0.55 mL, 0.11 mmol) was added dropwise at the same temperature. The resulting mixture was allowed to warm to room temperature and stirred for 1 h. The reaction mixture was quenched with sat. NH 4 Cl aq, and the aqueous layer was extracted with n-hexane three times. The combined organic layer was washed with brine, dried over MgSO 4 , filtered, and concentrated in vacuo. The residue was purified by flash chromatography on silica gel (n-hexane) to give 8a (27 mg, 64%) as a colorless oil.  (1 mL) at room temperature. The resulting mixture was stirred at the same temperature for 30 min. The reaction mixture was filtered through a pad of Celite and concentrated in vacuo. The residue was purified by flash chromatography on silica gel (n-hexane/ethyl acetate = 10:1) to give the aldehyde as a yellow oil. The crude residue was used in the subsequent step without purification. To a solution of methyltriphenylphosphonium iodide (189 mg, 0.47 mmol) in THF (0.5 mL) was added sodium bis(trimethylsilyl)amide (1.9 M in THF, 0.23 mL, 0.44 mmol) at 0 • C. The resulting mixture was stirred at the same temperature for 30 min, then a solution of the aldehyde in THF (0.2 M, 0.45 mL, 0.089 mmol) was added dropwise at the same temperature. The resulting mixture was allowed to warm to room temperature and stirred for 1 h. The reaction mixture was quenched with sat. NH 4 Cl aq, and the aqueous layer was extracted with n-hexane three times. The combined organic layer was washed with brine, dried over MgSO 4 , filtered, and concentrated in vacuo. The residue was purified by flash chromatography on silica gel (n-hexane) to give 8b (23 mg, 70%) as a colorless oil. , and triethylamine (0.8 mL) in toluene (0.8 mL) was added tetrakis(triphenylphosphine)palladium(0) (9 mg, 0.0080 mmol) at room temperature. The resulting mixture was allowed to warm to 90 • C and stirred for 2 h. The reaction mixture was concentrated in vacuo. The residue was purified by flash chromatography on silica gel (n-hexane) to give the coupling product (39 mg, 65%) as a yellow oil. The crude residue was used in the subsequent step without purification. To a solution of the coupling product (9 mg, 0.012 mmol) in THF (0.6 mL) was added tetra-n-butylammonium fluoride (1.0 M in THF, 0.12 mL, 0.12 mmol) at room temperature. The resulting mixture was stirred at the same temperature for 12 h. The reaction mixture was quenched with sat. NaHCO 3 aq, and the aqueous layer was extracted with ethyl acetate three times. The combined organic layer was washed with brine, dried over MgSO 4 , filtered, and concentrated in vacuo.  13

Metabolism of 6a and 6b by CYP24A1
The metabolism of 6a and 6b by CYP24A1 was examined using recombinant human CYP24A1 as described in previous studies [7,15]. Briefly, the reaction mixture containing 20 nM CYP24A1, 2.0 µM bovine adrenodoxin, 0.2 µM bovine adrenodoxin reductase, 5 µM each substrate, 1 mM NADPH, and 1 mM ethylenediaminetetraacetic acid (EDTA) in 8 of 13 100 mM Tris-HCl (pH 7.4) was incubated at 37 • C for 15 or 30 min. The metabolites were extracted with 4 volumes of chloroform/methanol (3:1) and analyzed by reversed-phase HPLC under the same conditions followed in our previous study [16]. 16a and 16b and Identification of New Metabolites 4α,24R,25-trihydroxyvitamin D 3 (16a): To a solution of CD-ring 15 [17] (9.6 mg, 0.016 mmol), 8a (5 mg, 0.014 mmol), and triethylamine (0.11 mL) in toluene (0.11 mL) was added tetrakis(triphenylphosphine)palladium(0) (3.2 mg, 0.0016 mmol) at room temperature. The resulting mixture was allowed to warm to 90 • C and stirred for 2 h. The reaction mixture was concentrated in vacuo. The residue was purified by flash chromatography on silica gel (n-hexane) to give the coupling product (11 mg, 88%) as a yellow oil. The crude residue was used in the subsequent step without purification. To a solution of the coupling product (11 mg, 0.012 mmol) in THF (0.1 mL) was added tetra-n-butylammonium fluoride (1.0 M in THF, 144 µL, 0.144 mmol) at room temperature. The resulting mixture was stirred at the same temperature for 12 h. The reaction mixture was quenched with sat. NaHCO 3 aq, and the aqueous layer was extracted with ethyl acetate three times. The combined organic layer was washed with brine, dried over MgSO 4 , filtered, and concentrated in vacuo. The residue was purified by flash chromatography on silica gel (methanol/chloroform 1:15) to give 16a (4.9 mg, 94%) as a yellow oil.  4β,24R,25-trihydroxyvitamin D 3 (16b): To a solution of CD-ring 15 (9.6 mg, 0.016 mmol), 8b (5 mg, 0.014 mmol), and triethylamine (0.11 mL) in toluene (0.11 mL) was added tetrakis(triphenylphosphine)palladium(0) (3.2 mg, 0.0016 mmol) at room temperature. The resulting mixture was allowed to warm to 90 • C and stirred for 2 h. The reaction mixture was concentrated in vacuo. The residue was purified by flash chromatography on silica gel (n-hexane) to give the coupling product (8.5 mg, 68%) as a yellow oil. The crude residue was used in the subsequent step without purification. To a solution of the coupling product (8.5 mg, 0.0096 mmol) in THF (0.1 mL) was added tetra-n-butylammonium fluoride (1.0 M in THF, 115 µL, 0.115 mmol) at room temperature. The resulting mixture was stirred at the same temperature for 12 h. The reaction mixture was quenched with sat. NaHCO 3 aq, and the aqueous layer was extracted with ethyl acetate three times. The combined organic layer was washed with brine, dried over MgSO 4 , filtered, and concentrated in vacuo. The residue was purified by flash chromatography on silica gel (methanol/chloroform 1:15) to give 16b (3.5 mg, 84%) as a yellow oil.

Docking Study
The initial structure of the human CYP24A1 was obtained from the AlphaFold protein structure database [18] (AF-Q07973-F1-model_v2.pdb) and refined for docking simulations using the Protein Preparation Wizard script within Maestro (Schrödinger, LLC, New York, NY, USA). For all compound molecules, ionization and energy minimization were performed using the OPLS3e force field in the LigPrep script in Maestro. These minimized structures were used as input structures for the docking simulations using the Glide SP docking [19,20] program (Schrödinger, LLC, New York, NY, USA), with a grid box defined by a potential binding site position from SiteMap [21,22] (Schrödinger, LLC, New York, NY, USA). In docking simulations, we also introduced hydrogen-bonding constraints between the sidechain of The395 and any polar atoms in the compound molecules, because this hydrogen-bonding formation is a key interaction in known complexes of CYP24A1 bound to the vitamin D 3 analogs [23]. After the docking simulations were completed, the lowest distances between the Fe atom of the HEM molecule and the C24 of the compounds from 100 poses on a binding site were selected.

Docking Study
The initial structure of the human CYP24A1 was obtained from the AlphaFold protein structure database [18] (AF-Q07973-F1-model_v2.pdb) and refined for docking simulations using the Protein Preparation Wizard script within Maestro (Schrödinger, LLC, New York, NY, USA). For all compound molecules, ionization and energy minimization were performed using the OPLS3e force field in the LigPrep script in Maestro. These minimized structures were used as input structures for the docking simulations using the Glide SP docking [19,20] program (Schrödinger, LLC, New York, NY, USA), with a grid box defined by a potential binding site position from SiteMap [21,22] (Schrödinger, LLC, New York, NY, USA). In docking simulations, we also introduced hydrogen-bonding constraints between the sidechain of The395 and any polar atoms in the compound molecules, because this hydrogen-bonding formation is a key interaction in known complexes of CYP24A1 bound to the vitamin D3 analogs [23]. After the docking simulations were completed, the lowest distances between the Fe atom of the HEM molecule and the C24 of the compounds from 100 poses on a binding site were selected.
The syntheses of 4α,25-dihydroxyvitamin D 3 (6a) and 4β,25-dihydroxyvitamin D 3 (6b) are illustrated in Scheme 3A. The primary hydroxy alcohol 10 was synthesized in 54% yield by the reaction of 9, derived from L-(-)-malic acid, with TBSOTf followed by selective deprotection of the primary TBS ether in the presence of camphorsulfonic acid (CSA). After oxidation of the primary hydroxy group in 10 under Swern conditions, the resulting aldehyde was reacted with TMS acetylene in the presence of n-butyllithium to stereoselectively give 4β-11b in 90% yield (2 steps), where the stereochemistry at C4 is controlled according to the Felkin-Anh model (Scheme 3B). The 4α-isomer 11a was stereoselectively obtained by oxidation of 11b with Dess-Martin periodinane followed by reduction of the resulting ketone with a bulky reducing agent, L-selectride. The alcohols 11a and 11b were then converted to 6a and 6b, respectively, as follows. The secondary hydroxy groups in 11a and 11b were protected as TBS ethers in the presence of TBSOTf and 2,6-lutidine, and the benzoyl and TMS groups were removed with potassium carbonate to give alcohols 12a and 12b in 70% and 76% yields, respectively. Treatment of 12a and 12b with iodine and PPh 3 followed by reaction with sodium cyanide gave the corresponding nitriles 13a and 13b. Reduction of the nitriles 13 with DIBAL-H afforded the corresponding aldehydes, which were then reacted with Wittig reagent 14 to give the A-ring synthons, enynes 8a and 8b, in 63% and 70% yield, respectively. Coupling of 8a and 8b with the CD-ring synthon, bromoolefin 7, was performed in the presence of a Pd catalyst, and the coupling products were deprotected with TBAF to remove the silyl ether groups, affording vitamin D metabolites 6a and 6b in 50% yield and 65% yield, respectively. and PPh3 followed by reaction with sodium cyanide gave the corresponding nitriles 13a and 13b. Reduction of the nitriles 13 with DIBAL-H afforded the corresponding aldehydes, which were then reacted with Wittig reagent 14 to give the A-ring synthons, enynes 8a and 8b, in 63% and 70% yield, respectively. Coupling of 8a and 8b with the CD-ring synthon, bromoolefin 7, was performed in the presence of a Pd catalyst, and the coupling products were deprotected with TBAF to remove the silyl ether groups, affording vitamin D metabolites 6a and 6b in 50% yield and 65% yield, respectively.

Metabolism of 6a and 6b by CYP24A1
A reconstituted system containing human CYP24A1, adrenodoxin reductase, and adrenodoxin was employed to examine the metabolism of 6a and 6b. The conversion ratios of the substrate to its metabolites were obtained from the peak area ratio in HPLC chromatograms after 15 min incubation of 6 with CYP24A1 ( Figure S1). The 6a was

Metabolism of 6a and 6b by CYP24A1
A reconstituted system containing human CYP24A1, adrenodoxin reductase, and adrenodoxin was employed to examine the metabolism of 6a and 6b. The conversion ratios of the substrate to its metabolites were obtained from the peak area ratio in HPLC chromatograms after 15 min incubation of 6 with CYP24A1 ( Figure S1). The 6a was sequentially metabolized by CYP24A1 to produce multiple metabolites, as well as 1,25D 3 and 25D 3 [6,7]. The total conversion ratio of 6a to the multiple metabolites was 38.2%, whereas the conversion ratio of 6b was only 9.8%. These results suggest that the difference in serum concentrations between 6a and 6b can be explained by the difference in their metabolic stability, as we had hypothesized. Specifically, the high resistance of 6b to metabolism by CYP24A1 may explain why 6b is detected in serum, whereas 6a is not.

Identification of 16a and 16b as Metabolites of 4,25-(OH) 2 -D 3 (6)
Next, we set out to identify the major metabolites of 6a and 6b generated by CYP24A1. Since CYP24A1 is known to hydroxylate C24 of 2 with R-stereochemistry [7], we hypothesized that 6 would undergo 24R-hydroxylation by CYP24A1 to afford 4,24R,25trihydroxyvitamin D 3 (7). To test this hypothesis, we synthesized the 4α compound 16a and the 4β compound 16b and compared their HPLC retention times with the metabolites of 6a and 6b produced by CYP24A1.
Compounds 16a and 16b were synthesized by a similar procedure to that described for the 4-hydroxylated metabolites of 6 (Scheme 4). Specifically, the CD ring 15 [17] bearing an R-hydroxy group at C24 was reacted with the A-ring precursor enyne 8a or 8b in the presence of a palladium catalyst to give the coupling product; deprotection of the silyl ether using TBAF afforded 16a and 16b in 80% and 58% yields, respectively. those of the metabolites produced from 6 by CYP24A1. The retention times of 16a and 16b were consistent with those of the main metabolites produced by CYP24A1 from 6a and 6b (peak A and peak B in Figure 1A,B), respectively. In HPLC analyses on a chiral column (SUMICHIRAL OA-7000), the retention times of 16 matched those of the metabolites generated from 6 ( Figure S2), indicating that the main metabolites produced by CYP24A1 from 6a and 6b are 16a and 16b, respectively. The conversion ratios to 16a and 16b from 6a and 6b were 25.6 and 7.0 %, respectively (Table 1). Then, we compared the HPLC retention times of the synthesized 16a and 16b with those of the metabolites produced from 6 by CYP24A1. The retention times of 16a and 16b were consistent with those of the main metabolites produced by CYP24A1 from 6a and 6b (peak A and peak B in Figure 1A,B), respectively. In HPLC analyses on a chiral column (SUMICHIRAL OA-7000), the retention times of 16 matched those of the metabolites generated from 6 ( Figure S2), indicating that the main metabolites produced by CYP24A1 from 6a and 6b are 16a and 16b, respectively. The conversion ratios to 16a and 16b from 6a and 6b were 25.6 and 7.0 %, respectively (Table 1). the 4β compound 16b and compared their HPLC retention times with the metabolites of 6a and 6b produced by CYP24A1.
Compounds 16a and 16b were synthesized by a similar procedure to that described for the 4-hydroxylated metabolites of 6 (Scheme 4). Specifically, the CD ring 15 [17] bearing an R-hydroxy group at C24 was reacted with the A-ring precursor enyne 8a or 8b in the presence of a palladium catalyst to give the coupling product; deprotection of the silyl ether using TBAF afforded 16a and 16b in 80% and 58% yields, respectively.
Then, we compared the HPLC retention times of the synthesized 16a and 16b with those of the metabolites produced from 6 by CYP24A1. The retention times of 16a and 16b were consistent with those of the main metabolites produced by CYP24A1 from 6a and 6b (peak A and peak B in Figure 1A,B), respectively. In HPLC analyses on a chiral column (SUMICHIRAL OA-7000), the retention times of 16 matched those of the metabolites generated from 6 ( Figure S2), indicating that the main metabolites produced by CYP24A1 from 6a and 6b are 16a and 16b, respectively. The conversion ratios to 16a and 16b from 6a and 6b were 25.6 and 7.0 %, respectively (Table 1).

Docking Study
To investigate the reason for the difference in metabolic stability between 6a and 6b, docking studies with CYP24A1 and 6a and 6b were carried out. In the docking of 6a with CYP24A1, hydrogen bonds are formed between the C25 and C4α hydroxy groups of 6a and Leu325 and Thr395 of CYP24A1, resulting in the formation of a stable complex in which the side chain of the CD-ring in 6a is located close to the heme iron (Figure 2A). In contrast, in the docking of 6b with CYP24A1, hydrogen bonds were formed between the C4β and C3 hydroxy groups of 6b and the carboxylic acid of the side chain in heme and Thr395 of CYP24A1 ( Figure 2B). In this case, the side chain of the CD ring in 6b is located away from the heme iron, and the hydrogen bond between the C25 hydroxy group and the Leu325, which plays an important role in the metabolism of vitamin D by CYP24A1 [24], is not formed. This may explain why 6b is less susceptible to metabolism by CYP24A1.

Docking Study
To investigate the reason for the difference in metabolic stability between 6a and 6b, docking studies with CYP24A1 and 6a and 6b were carried out. In the docking of 6a with CYP24A1, hydrogen bonds are formed between the C25 and C4α hydroxy groups of 6a and Leu325 and Thr395 of CYP24A1, resulting in the formation of a stable complex in which the side chain of the CD-ring in 6a is located close to the heme iron (Figure 2A). In contrast, in the docking of 6b with CYP24A1, hydrogen bonds were formed between the C4β and C3 hydroxy groups of 6b and the carboxylic acid of the side chain in heme and Thr395 of CYP24A1 ( Figure 2B). In this case, the side chain of the CD ring in 6b is located away from the heme iron, and the hydrogen bond between the C25 hydroxy group and the Leu325, which plays an important role in the metabolism of vitamin D by CYP24A1 [24], is not formed. This may explain why 6b is less susceptible to metabolism by CYP24A1.

Conclusions
4α,25-Dihydroxyvitamin D 3 (6a) and 4β,25-dihydroxyvitamin D 3 (6b) are both produced from 25-hydroxyvitamin D 3 (2) by CYP3A4, but 6b is detectable in serum, whereas 6a is not. Our findings show that 6a is a better substrate of CYP24A1 than 6b, and thus the greater metabolic stability of 6b can account for its presence in human serum. Furthermore, major metabolites of 6a and 6b generated by CYP24A1 were identified as 4,24R,25trihydroxyvitamin D 3 (16a, 16b) by HPLC, by comparison with synthesized authentic samples. Docking studies indicated that while 6a can form a hydrogen bond between the hydroxy group at C25 and Leu325 in CYP24A1, the presence of the 4β-hydroxy group in 6b prevents the formation of this hydrogen bond, which plays a crucial role in the metabolic activity of CYP24A1.

Conflicts of Interest:
The authors declare no conflict of interest.