Synthesis of C2-Alkoxy-Substituted 19-Nor Vitamin D3 Derivatives: Stereoselectivity and Biological Activity

The active form of vitamin D3 (D3), 1a,25-dihydroxyvitamn D3 (1,25D3), plays a central role in calcium and bone metabolism. Many structure–activity relationship (SAR) studies of D3 have been conducted, with the aim of separating the biological activities of 1,25D3 or reducing its side effects, such as hypercalcemia, and SAR studies have shown that the hypercalcemic activity of C2-substituted derivatives and 19-nor type derivatives is significantly suppressed. In the present paper, we describe the synthesis of 19-nor type 1,25D3 derivatives with alkoxy groups at C2, by means of the Julia–Kocienski type coupling reaction between a C2 symmetrical A ring ketone and a CD ring synthon. The effect of C2 substituents on the stereoselectivity of the coupling reaction was evaluated. The biological activities of the synthesized derivatives were evaluated in an HL-60 cell-based assay. The a-methoxy-substituted C2α-7a was found to show potent cell-differentiating activity, with an ED50 value of 0.38 nM, being 26-fold more potent than 1,25D3.


Introduction
The active form of vitamin D 3 (D 3 ), 1a,25-dihydroxyvitamin D 3 (1,25D 3 ), is involved in various physiological activities, including calcium metabolism, cell differentiation, and immunomodulation, via binding to the vitamin D receptor (VDR) [1][2][3][4]. Various derivatives of 1,25D 3 have been synthesized and their structure-activity relationships (SARs) have been investigated, with the aim of separating the diverse biological activities of 1,25D 3 or reducing side effects, such as hypercalcemia (Figure 1). It has been found that substituents at C2 have significant effects on the binding affinity to VDR, as well as on calcium metabolism and cell-differentiating activity. For example, a derivative of ED-71 (C2β-7a') bearing a hydropropoxy group at C2, shows bone formation activity comparable to that of 1,25D 3 despite its weak binding affinity for VDR [5,6]. Furthermore, 19-nor type derivatives (7b') have been reported to show suppressed hypercalcemic activity, while retaining cell differentiation-inducing activity [7]. Thus, there is considerable interest in the synthesis and biological activities of 19-nor type derivatives bearing a substituent at C2.
thesized, but few 19-nor type derivatives with alkoxy substituents at C2 have been reported, and their biological activities have been little investigated [11][12][13][14]. A series of C2alkoxy-substituted 19-nor type derivatives with α-benzyloxy or epoxy groups has been synthesized and evaluated. In the case of the α-benzyloxy-substituted derivative (C2α-7g'), the HL-60 cell differentiation activity was decreased to ca 1/10th of that of 1,25D3 [10]. The 2α-epoxy-substituted derivative (C2α-7h') showed a low VDR binding affinity, only1/25th of that of 1,25D3, and the affinity of the 2β-derivative was even lower [15]. In this study, we synthesized 19-nor D3 derivatives with a methoxy, benzyloxy, or pnitrophenoxy group at C2 by means of Julia-type coupling, between a C2-symmetric A ring ketone and a CD ring synthon. In the coupling reaction, we observed interesting effects of substituents at C2, on the diastereoselectivity of the coupling reaction, and these effects are discussed in terms of the transition state of the coupling reaction [16]. The VDR binding affinity and HL-60 cell differentiation-inducing activity were evaluated. Among these compounds, the 2α-methoxy-substituted derivative showed ca 26-fold more potent cell differentiation-inducing activity than 1,25D3, while its affinity for VDR was similar to that of 1,25D3.
In this study, we synthesized 19-nor D 3 derivatives with a methoxy, benzyloxy, or p-nitrophenoxy group at C2 by means of Julia-type coupling, between a C2-symmetric A ring ketone and a CD ring synthon. In the coupling reaction, we observed interesting effects of substituents at C2, on the diastereoselectivity of the coupling reaction, and these effects are discussed in terms of the transition state of the coupling reaction [16]. The VDR binding affinity and HL-60 cell differentiation-inducing activity were evaluated. Among these compounds, the 2α-methoxy-substituted derivative showed ca 26-fold more potent cell differentiation-inducing activity than 1,25D 3 , while its affinity for VDR was similar to that of 1,25D 3 .
General procedure for the synthesis of (3R,5R)-3,5-bis((tert-butyldimethylsilyl)oxy)-4-R-cyclohexan-1-one (5). Pyridinium p-toluenesulfonate (15 eq) at room temperature was added to a solution of the crude product 4 in MeOH (0.012 M). After being stirred for 15 min, the mixture was poured into saturated aqueous NaHCO 3 and extracted with AcOEt. The combined organic layer was washed with brine, dried over MgSO 4 and concentrated in vacuo. The residue was purified by column chromatography on silica gel (n-hexane/ethyl acetate = 10:1) to yield a crude product.
Sodium periodate (6 eq) at room temperature was added to a solution of the crude product in MeOH/distilled water (7:1, 0.02 M). After being stirred for 2 h, the mixture was poured into water and extracted with AcOEt. The combined organic layer was washed with brine, dried over MgSO 4 and concentrated in vacuo. The residue was purified by column chromatography on silica gel (n-hexane/ethyl acetate = 20:1) and afforded 5 as a colorless oil.
Cell culture: Human promyelocytic leukemia cells (HL-60) were grown in RPMI supplemented with 10% heat-inactivated FBS in a CO 2 incubator. The population doubling time of HL-60 cells was approximately 37 h under the conditions employed, and cells were passaged at 2.0 × 10 5 cells/mL every 4 days. Cell viability was determined by trypan blue staining and was always more than 90%.

Vitamin D Receptor Binding Analysis
The 1,25D 3 and the synthesized derivatives were dissolved in DMSO (10 mM stock solutions) and diluted as required with DMSO and the VDR Red Assay Buffer. VDR and fluoromone in the VDR Red Assay Buffer solution (VDR 7.2 nM, fluoromone 2.0 nM, and 5.0 mM DTT, 15 µL) were added to the vitamin D 3 derivatives (78 nM, DMSO 10%, 15 µL) in the VDR Red Assay Buffer solution and incubated for 2 h in a 384-well microplate at room temperature in the dark. Fluorescence polarization was measured on a SPARK-TKIS multimode plate reader (Tecan) using a 535 nm excitation filter (25 nm bandwidth) and a 590 nm emission filter (20 nm bandwidth). This experiment was performed three times for each vitamin D 3 derivative. The VDR binding activity ratio was calculated as 9DmP sample /DmP 1,25D3 × 100), where DmP = (mP without sample)-(mP with 39 nM vitamin D 3 derivative or 1,25D 3 ).

Evaluation of the Differentiation-Inducing Activity by NBT Staining
HL-60 cells (2.0 × 10 5 cells/mL) were cultured with or without vitamin D derivatives (1 × 10 −6 -10 −11 M) in 200 µL of RPMI (0.1% DMSO) in a 96-well plate for 5 days in a CO 2 incubator. Then, PMA (160 nM) and NBT (0.1%) were added and incubation was continued for 30 min in a CO 2 incubator. The NBT-stained and non-stained cells were counted on a hemocytometer. The differentiation-inducing activity was calculated as (NBT-stained cell number)/(total cell number), in the presence of various concentrations of vitamin D derivatives or 1,25D 3 . EC 50 values are given as the average of triplicate assays [21,22].

Chemistry
For the synthesis of the C2-alkoxy-substituted 19-nor type D 3 derivatives, we firstly synthesized the A ring synthon ketones 5 bearing alkoxy-type substituents at C2, and examined the Julia-Lythgoe olefination reaction with the CD ring synthon 6. Thus, the ketones 5a-c bearing a methoxy, benzyloxy or 4-NO 2 -phenyloxy group were synthesized from diol 2 derived from (-)-quinic acid (1), as follows (Scheme 1) [23]. The diol 2 was reacted with p-anisaldehyde dimethyl acetal, using camphor sulfonic acid as a catalyst, to give benzylidene acetal 3, and the resulting secondary alcohol was converted into a methoxy, benzyloxy or p-nitrophenoxy group by the reaction with methyl iodide, benzyl iodide, or 4-fluoro-p-nitrobenzene, respectively, to give ethers 4a-c. the deprotection of the benzylidene acetal in 4 was conducted with PPTS, followed by oxidative cleavage of the resulting diol with sodium periodate to obtain A ring ketones 5a-c in 20-60% yield from 4. Biomolecules 2022, 12, x FOR PEER REVIEW 7 of 12 Scheme 1. Synthesis of A ring ketones 5a-c.
The C2 alkoxy-substituted ketones 5a-c were subjected to the Julia-Kocienski coupling reaction with the CD ring synthon, sulfone 6, affording two diastereomers at C2. The selectivity was determined by 1 H NMR [9,24] after the deprotection of the silyl ethers and MOM group at C25, and the results are summarized in Table 1. In the case of 5a, with the methoxy group at C2, C2β-7a was obtained preferentially in a ratio of 1:3.5 (entry 1), while no selectivity was observed in the case of 5b (entries 2). Interestingly, only one diastereomer, C2α-7c, was obtained in the case of 5c with a p-NO2-phenyl ether group (entry 3) [25] [See supplementary materials]. Table 1. Diastereomeric ratio of coupling product 7 in Julia-type coupling of ketone 5 with sulfone 6.

Scheme 1. Synthesis of A ring ketones 5a-c.
The C2 alkoxy-substituted ketones 5a-c were subjected to the Julia-Kocienski coupling reaction with the CD ring synthon, sulfone 6, affording two diastereomers at C2. The selectivity was determined by 1 H NMR [9,24] after the deprotection of the silyl ethers and MOM group at C25, and the results are summarized in Table 1. In the case of 5a, with the methoxy group at C2, C2β-7a was obtained preferentially in a ratio of 1:3.5 (entry 1), while no selectivity was observed in the case of 5b (entries 2). Interestingly, only one diastereomer, C2α-7c, was obtained in the case of 5c with a p-NO 2 -phenyl ether group (entry 3) [25] [See Supplementary Materials]. These results suggest that the substituent at C2 influences the diastereoselectivity in the Julia-type coupling reaction, and therefore conformational analysis of the A ring ketones 5 bearing substituents at C2 was carried out by 1 H NMR. Regardless of the substituent at C2, the ketones appear to preferentially take a particular chair conformation in the conformational equilibrium, as judged from the 1 H NMR data (only small coupling constants of H 2 were observed; JH2-H1 = 2.1-2.4 Hz, JH2-H3 = 2.1-2.5 Hz). Furthermore, the nuclear Overhauser effect (nOe) was observed between the H 2 atom, and H 1 and H 3 , respectively. Thus, the ketones adopted the chair conformation A with axial substituents at C1 and C2, as shown in Scheme 2. On the other hand, the A ring of the C2-substituted VD3 These results suggest that the substituent at C2 influences the diastereoselectivity in the Julia-type coupling reaction, and therefore conformational analysis of the A ring ketones 5 bearing substituents at C2 was carried out by 1 H NMR. Regardless of the substituent at C2, the ketones appear to preferentially take a particular chair conformation in the confor-mational equilibrium, as judged from the 1 H NMR data (only small coupling constants of H 2 were observed; J H2-H1 = 2.1-2.4 Hz, J H2-H3 = 2.1-2.5 Hz). Furthermore, the nuclear Overhauser effect (nOe) was observed between the H 2 atom, and H 1 and H 3 , respectively. Thus, the ketones adopted the chair conformation A with axial substituents at C1 and C2, as shown in Scheme 2. On the other hand, the A ring of the C2-substituted VD 3 derivative is known to take a chair conformation, as shown in F, in which the C2 substituent occupies the equatorial position [9,24,26]. Thus, in the coupling reaction, the addition of sulfonate occurred from the a-face of the ketone to avoid steric hindrance with the C3 substituent, generating C. The BT (benzothiazole) group in the CD ring is transferred to the resulting hydroxyl group, and the ring conformation flips to E from D. Finally, the olefin is formed by the elimination of SO 2 and BTO anion to generate F. Basically, in the Julia-type olefination reaction, two types of elimination pathways can occur, i.e., syn-periplanar elimination and anti-periplanar elimination, and the geometric isomerism of the generated olefin depends on these elimination processes. The differences in the stereoselectivity of the coupling products of the Julia-type olefination with ketones 5, can be attributed to the differences in the stability of the corresponding transition states (TS). There are four possible transition states resulting from differences in the mode of elimination (anti/syn) and the stereochemistry at the C6 position (S/R), and the TSs for 5a with a methoxy group at C2 are shown in Figure 2a. Among them, TS-syn-6R-1, which leads to the C2α-product, would be unstable due to steric repulsion between the TBS group and the CD ring moiety. Therefore, the reaction would proceed through the three remaining states (TS-anti-6S-1, TS-syn-6S-1 and TS-anti-6R-1), affording the C2β product preferentially. On the other hand, in the case of 5c with a p-NO2-PhO group, this substituent is sterically hindered and would tend to push the TBS group towards the CD ring moiety, leading to steric repulsion between the TBS group and CD ring moiety, unfavorably impacting on the TSs of TS-anti-6S-2, TS-syn-6S-2 and TSsyn-6R-2. Therefore, TS-anti-6R-2 would be the preferred transition state to afford C2α-7c (Figure 2b). The differences in the stereoselectivity of the coupling products of the Julia-type olefination with ketones 5, can be attributed to the differences in the stability of the corresponding transition states (TS). There are four possible transition states resulting from differences in the mode of elimination (anti/syn) and the stereochemistry at the C6 position (S/R), and the TSs for 5a with a methoxy group at C2 are shown in Figure 2a. Among them, TS-syn-6R-1, which leads to the C2α-product, would be unstable due to steric repulsion between the TBS group and the CD ring moiety. Therefore, the reaction would proceed through the three remaining states (TS-anti-6S-1, TS-syn-6S-1 and TS-anti-6R-1), affording the C2β product preferentially. On the other hand, in the case of 5c with a p-NO 2 -PhO group, this substituent is sterically hindered and would tend to push the TBS group towards the CD ring moiety, leading to steric repulsion between the TBS group and CD ring moiety, unfavorably impacting on the TSs of TS-anti-6S-2, TS-syn-6S-2 and TS-syn-6R-2. Therefore, TS-anti-6R-2 would be the preferred transition state to afford C2α-7c (Figure 2b).

Evaluation of the VDR Binding Affinity of 7a-c
The VDR binding affinity of C2-alkoxy-substituted 19-nor D3 derivatives 7a-c was evaluated by the VDR competition assay with fluorescence polarization (PolarScreen™, Invitrogen) [27], and the results are summarized in Table 2. In the case of C2α-7a (2α-OMe), the binding affinity (mP value) was found to be 100.8% when the value for1,25D3 was normalized to 100%. On the other hand, C2β-7a (2β-OMe) showed a lower VDR binding affinity than C2α-7a (47.5%). In the case of 7b, the binding affinities of C2α-7b (2α-OBn) and C2b-7b (2β-OBn) were 2.5% and 0.4%, respectively, and almost no binding was observed in the case of C2α-7c (2α-p-NO2-PhO). These results suggest that a smaller alkoxy substituent at C2 with α-stereochemistry tends to show a higher VDR binding affinity [8]. From these results, it is considered that the smaller substituent with the alpha con-figuration at the C2 position would construct an appropriate interaction between the substituent and the ligand-binding domain (LBD) in the VDR, and shows a stronger VDRbinding affinity [28].

Evaluation of the VDR Binding Affinity of 7a-c
The VDR binding affinity of C2-alkoxy-substituted 19-nor D 3 derivatives 7a-c was evaluated by the VDR competition assay with fluorescence polarization (PolarScreen™, Invitrogen) [27], and the results are summarized in Table 2. In the case of C2α-7a (2α-OMe), the binding affinity (mP value) was found to be 100.8% when the value for1,25D 3 was normalized to 100%. On the other hand, C2β-7a (2β-OMe) showed a lower VDR binding affinity than C2α-7a (47.5%). In the case of 7b, the binding affinities of C2α-7b (2α-OBn) and C2b-7b (2β-OBn) were 2.5% and 0.4%, respectively, and almost no binding was observed in the case of C2α-7c (2α-p-NO 2 -PhO). These results suggest that a smaller alkoxy substituent at C2 with α-stereochemistry tends to show a higher VDR binding affinity [8]. From these results, it is considered that the smaller substituent with the alpha con-figuration at the C2 position would construct an appropriate interaction between the substituent and the ligand-binding domain (LBD) in the VDR, and shows a stronger VDR-binding affinity [28].

Evaluation of the HL-60 Cell Differentiation-Inducing Activity of 7a-c
It is known that 1,25D 3 induces the differentiation of the HL-60 cells into macrophages. Thus, the differentiation-inducing activities of the new 19-nor type derivatives, 7a-c, were evaluated by the NBT reduction method [21,22], and the EC 50 values and ratios normalized to 1,25D 3 are summarized in Table 2. Surprisingly, the differentiation activity ratio of C2α-7a was ca 26-fold greater than that of 1,25D 3 , even though the VDR binding affinity was similar to that of 1,25D 3 (entry 1). The cell differentiation-inducing activities of C2β-7a and C2α-7b were similar or lower than those of 1,25D 3 (1.41-and 0.30-fold, respectively). On the other hand, C2β-7b and C2α-7c did not show significant cell differentiation-inducing activities (ratios of 0.08-and 0.02-fold, respectively).
Thus, the 19-nor derivative with a methoxy substituent at C2α showed characteristic potent cell differentiation-inducing activity. Since the activities of the corresponding methyl-and hydroxy-substituted 19-nor D 3 derivatives were reported to be 0.02-and 1-fold, respectively [8,9,15], the presence of a small alkoxy group appeared to have a substantial impact on the cell differentiation-inducing activity in the D 3 derivatives. These results provide a new insight into the relationship between the structure and activity of D 3 analogs, especially for the 19-nor type analogs, and should be helpful in the design of more potent derivatives, with better a separation of the D 3 activities associated with binding to the VDR.

Conclusions
In summary, we have investigated the stereoselectivity in the Julia-type coupling reaction of the A ring ketones 5 and the CD ring synthon 6. In this reaction, the steric size of the substituent at C2 of 5 affects the stereoselectivity, and the C2β product 7 is preferentially obtained when the substituent at C2 is small, while the C2α product is generated when the substituent is large. These results can be explained in terms of the stability of the transition state, which is affected by steric repulsion involving the TBS ether group on the A ring, the substituent at C2 and the CD ring moiety. Among the C2-alkoxy-substituted 19-nor type D 3 derivatives 7a-c, C2α-7a showed 26-fold more potent cell differentiation-inducing activity than 1,25D 3 (1).