Seco-B-Ring Steroidal Dienynes with Aromatic D Ring: Design, Synthesis and Biological Evaluation

Continuing our structure-activity studies on the vitamin D analogs with the altered intercyclic seco-B-ring fragment, we designed compounds possessing dienyne system conjugated with the benzene D ring. Analysis of the literature data and the docking experiments seemed to indicate that the target compounds could mimic the ligands with a good affinity to the vitamin D receptor (VDR). Multi-step synthesis of the C/D-ring building block of the tetralone structure was achieved and its enol triflate was coupled with the known A-ring fragments, possessing conjugated enyne moiety, using Sonogashira protocol. The structures of the final products were confirmed by NMR, UV and mass spectroscopy. Their binding affinities for the full-length human VDR were determined and it was established that compound substituted at C-2 with exomethylene group showed significant binding to the receptor. This analog was also able to induce monocytic differentiation of HL-60 cells.


Introduction
Calcitriol (1; 1α,25-dihydroxyvitamin D 3 ; 1α,25-(OH) 2 D 3 ; Figure 1) is the most active metabolite of the vitamin D 3 [1], representing its hormonal form [2]. Numerous studies demonstrated that this active form of vitamin D 3 is responsible for calcium and phosphorus homeostasis, and, additionally, plays an important role in cell proliferation and differentiation as well as apoptosis and immunomodulation [3][4][5]. These activities are primarily mediated through the vitamin D receptor (VDR) [6], a member of a nuclear receptor superfamily [7,8] acting as a ligand-activated transcription factor. It has been established that calcitriol binds to the VDR, and then heterodimerizes with retinoid X receptor (RXR). Upon recruitments of specific coactivators, the formed complex influences the target genes by binding to vitamin D receptor elements (VDRE) [9,10]. Antiproliferative and prodifferentiating effects of calcitriol on various types of malignant cells could be used for the treatment of cancer [11]. However, usefulness of calcitriol as a cancer chemopreventive agent is significantly limited by its strong calcemic effect that could result in hypercalcemia, mineralization of the internal organs, blood vessels and cutaneous tissues. This fact has stimulated broad synthetic efforts in the pharmaceutical companies and academic institutions directed to the development of calcitriol analogs characterized by clear dissociation between the antiproliferative and calcemic activities [12][13][14]. Skeletal modifications in the synthesized vitamin D compounds focused mainly on their A ring and steroidal side chain at C-17. Structural changes in the triene system were significantly less common and involved, for example, hydrogenation of C(10)=C (19) bond [15,16], inverted configuration of C(5)=C(6) [17] or C(7)=C (8) [18][19][20] bond and deletion of an exomethylene unit at C-10 [21,22]. The intercyclic 5,7-diene fragment was modified by its substitution at C-6 [23][24][25][26][27][28][29][30][31][32] and C-7 [32,33]. In 2011, an interesting study was published by Kittaka et al. [34], indicating that more drastic seco-B-ring modification can still result in compounds of potent VDR binding affinity. Thus, the Kittaka group described that 14-epi-19-nortachysterol derivative 2 has a binding ability to the hVDR decreased only seven times compared to that of calcitriol; its 2-methylene substituted analog 3 was even more active. Interestingly, also in the case of compound 4 with the conjugated dienyne fragment, only two-fold reduction of the binding activity has been reported relative to the natural hormone 1 [34].
Taking the above-described facts into consideration, we have decided to broaden the structureactivity studies in this area and synthesize similar seco-B steroidal compounds 5 and 6 characterized by a presence of an aromatic D ring. It should be added that the first report on a new class of vitamin D analogs with a benzene D ring was published by Mouriño et al. in 2010 [35]. However, the synthesized compound 7 slowly equilibrated into a more stable previtamin form. Obviously such isomerization could not occur in our designed compounds 5 and 6, and their polyunsaturated system, located in all steroidal rings, should be stable. We performed molecular studies with docking these target compounds into the ligand binding domain of the hVDR and both ligands 5 and 6 seemed to show all expected, favorable interactions with the respective amino acids (vide infra). Skeletal modifications in the synthesized vitamin D compounds focused mainly on their A ring and steroidal side chain at C-17. Structural changes in the triene system were significantly less common and involved, for example, hydrogenation of C(10)=C (19) bond [15,16], inverted configuration of C(5)=C(6) [17] or C(7)=C(8) [18][19][20] bond and deletion of an exomethylene unit at C-10 [21,22]. The intercyclic 5,7-diene fragment was modified by its substitution at C-6 [23][24][25][26][27][28][29][30][31][32] and C-7 [32,33]. In 2011, an interesting study was published by Kittaka et al. [34], indicating that more drastic seco-B-ring modification can still result in compounds of potent VDR binding affinity. Thus, the Kittaka group described that 14-epi-19-nortachysterol derivative 2 has a binding ability to the hVDR decreased only seven times compared to that of calcitriol; its 2-methylene substituted analog 3 was even more active. Interestingly, also in the case of compound 4 with the conjugated dienyne fragment, only two-fold reduction of the binding activity has been reported relative to the natural hormone 1 [34].
Taking the above-described facts into consideration, we have decided to broaden the structure-activity studies in this area and synthesize similar seco-B steroidal compounds 5 and 6 characterized by a presence of an aromatic D ring. It should be added that the first report on a new class of vitamin D analogs with a benzene D ring was published by Mouriño et al. in 2010 [35]. However, the synthesized compound 7 slowly equilibrated into a more stable previtamin form. Obviously such isomerization could not occur in our designed compounds 5 and 6, and their polyunsaturated system, located in all steroidal rings, should be stable. We performed molecular studies with docking these target compounds into the ligand binding domain of the hVDR and both ligands 5 and 6 seemed to show all expected, favorable interactions with the respective amino acids (vide infra).

Chemistry
We considered Sonogashira coupling of the known A-ring fragments 8 and 9, prepared in the Norman's and our laboratory [36,37], with the C/D-ring triflate 10 as a convenient route to the target compounds 5 and 6. As a starting material served commercially available 5-bromotetralone (11; Scheme 1), this was first converted to the corresponding cyano compound 12 using the procedure described by Tschaen et al. [38]. The cyanation method involving the use of CuCN and FeCl 3 [39] proved to be less efficient. Diisobutylaluminum hydride reduction of 12 gave the racemic hydroxy aldehyde 13, and its further synthetic transformations were executed on mixtures of compounds being epimeric at C-5.

Chemistry
We considered Sonogashira coupling of the known A-ring fragments 8 and 9, prepared in the Norman's and our laboratory [36,37], with the C/D-ring triflate 10 as a convenient route to the target compounds 5 and 6. As a starting material served commercially available 5-bromotetralone (11; Scheme 1), this was first converted to the corresponding cyano compound 12 using the procedure described by Tschaen et al. [38]. The cyanation method involving the use of CuCN and FeCl3 [39] proved to be less efficient. Diisobutylaluminum hydride reduction of 12 gave the racemic hydroxy aldehyde 13, and its further synthetic transformations were executed on mixtures of compounds being epimeric at C-5.  Silylation of 13 and the following Horner-Wadsworth-Emmons reaction of the protected hydroxy aldehyde 14 with ethyl (triphenylphosphoranylidene)acetate furnished the unsaturated ester 15. The observed value (15.8 Hz) of the vinylic protons coupling in its 1 H NMR spectrum proved E-configuration of the newly introduced double bond. Reduction of the ester 15 provided the allylic alcohol 16 that was then subjected to Sharpless asymmetric epoxidation using the (−)-diisopropyl D-tartrate to generate the epoxide 17 of the desired configuration [40]. Regioselective ring opening of this epoxide with lithium methylcyanocuprate afforded the mono-protected triol 18. Its subsequent cleavage with sodium periodate gave the aldehyde 19 that was reduced with sodium borohydride to the monoprotected diol 20. This compound was tosylated and the formed tosylate 21 subjected to alkylation with the Grignard reagent, generated from the chloro ether A. This process was carried out in the presence of CuI and resulted in the formation of the product 23 with the desired steroidal side chain. Subsequent oxidation of the hydroxyl group with Dess-Martin periodinane afforded the tetralone derivative 24. Its enolate form, generated with lithium diisopropylamide, was treated with N-phenyltrifluoromethanesulfonimide and the resulting enol triflate 10 was a building block suitable for the planned coupling with the acetylenic compounds 8 and 9.
Sonogashira reaction of the synthesized enol triflate 10 with the corresponding A-ring fragments 8 and 9 provided the expected dienyne products 25 and 26, albeit in rather low yield (Scheme 2). Despite our attempts, the outcome of this process could not be improved. Final desilylation of the protected compounds 25 and 26 afforded the target compounds 5 and 6. All spectroscopic data fully supported the structures of the synthesized compounds. Silylation of 13 and the following Horner-Wadsworth-Emmons reaction of the protected hydroxy aldehyde 14 with ethyl (triphenylphosphoranylidene)acetate furnished the unsaturated ester 15. The observed value (15.8 Hz) of the vinylic protons coupling in its 1 H NMR spectrum proved E-configuration of the newly introduced double bond. Reduction of the ester 15 provided the allylic alcohol 16 that was then subjected to Sharpless asymmetric epoxidation using the (−)-diisopropyl Dtartrate to generate the epoxide 17 of the desired configuration [40]. Regioselective ring opening of this epoxide with lithium methylcyanocuprate afforded the mono-protected triol 18. Its subsequent cleavage with sodium periodate gave the aldehyde 19 that was reduced with sodium borohydride to the monoprotected diol 20. This compound was tosylated and the formed tosylate 21 subjected to alkylation with the Grignard reagent, generated from the chloro ether A. This process was carried out in the presence of CuI and resulted in the formation of the product 23 with the desired steroidal side chain. Subsequent oxidation of the hydroxyl group with Dess-Martin periodinane afforded the tetralone derivative 24. Its enolate form, generated with lithium diisopropylamide, was treated with N-phenyltrifluoromethanesulfonimide and the resulting enol triflate 10 was a building block suitable for the planned coupling with the acetylenic compounds 8 and 9.
Sonogashira reaction of the synthesized enol triflate 10 with the corresponding A-ring fragments 8 and 9 provided the expected dienyne products 25 and 26, albeit in rather low yield (Scheme 2). Despite our attempts, the outcome of this process could not be improved. Final desilylation of the protected compounds 25 and 26 afforded the target compounds 5 and 6. All spectroscopic data fully supported the structures of the synthesized compounds.

Docking Studies
The docking simulations of the synthesized compounds 5 and 6 to the ligand binding domain (LBD) of the VDR were performed using Molegro Virtual Docker (release 4.0, CLC bio, Qiagen, Aarhus, Denmark). The LBD was extracted from crystalline hVDR (LBD)-1 complex (PDB Code: 1DB1) [41]. After docking the energy-minimized structures of the ligands 5 and 6 into VDR, we carefully analyzed the calculated complexes, taking into consideration their energies, number of hydrogen bonds and orientation of the ligand with respect to the Trp-286 aromatic rings. It was established by mutation [42] and NMR [43] experiments that this unique residue is important for ligand's anchoring in the binding pocket and transcription of genes controlled by VDR.
The docking studies show that both compounds 5 and 6 anchor the receptor similarly to calcitriol (Figure 2a,b). The notable difference consists in the "shift" of both ligands in the binding pocket [35], resulting in close proximity of their D ring to the parallel-oriented tryptophan rings. Such orientation of both aromatic fragments and the distance between them (ca. 4 Å) allows for their mutual π-π interaction [44,45]. Hydroxyl groups of the steroidal ligands create six hydrogen bonds with the same receptor residues as it was found in the crystalline complex hVDR-1 [41].

Docking Studies
The docking simulations of the synthesized compounds 5 and 6 to the ligand binding domain (LBD) of the VDR were performed using Molegro Virtual Docker (release 4.0, CLC bio, Qiagen, Aarhus, Denmark). The LBD was extracted from crystalline hVDR (LBD)-1 complex (PDB Code: 1DB1) [41]. After docking the energy-minimized structures of the ligands 5 and 6 into VDR, we carefully analyzed the calculated complexes, taking into consideration their energies, number of hydrogen bonds and orientation of the ligand with respect to the Trp-286 aromatic rings. It was established by mutation [42] and NMR [43] experiments that this unique residue is important for ligand's anchoring in the binding pocket and transcription of genes controlled by VDR.
The docking studies show that both compounds 5 and 6 anchor the receptor similarly to calcitriol (Figure 2a,b). The notable difference consists in the "shift" of both ligands in the binding pocket [35], resulting in close proximity of their D ring to the parallel-oriented tryptophan rings. Such orientation of both aromatic fragments and the distance between them (ca. 4 Å) allows for their mutual π-π interaction [44,45]. Hydroxyl groups of the steroidal ligands create six hydrogen bonds with the same receptor residues as it was found in the crystalline complex hVDR-1 [41].

Biological Evaluation: Binding to the Vitamin D Receptor
The affinities of the synthesized compounds 5 and 6 to VDR were assessed by a fluorescence polarization (FP)-based competition assay. The VDR affinities of compounds were checked using a wide range of concentrations and compared to that of calcitriol. Dose-response curves were plotted using GraphPad Prism software (version 6.04, GraphPad Software, Inc., San Diego, CA, USA), and half maximal inhibitory concentration (IC50) values were calculated from these dose-response curves. The binding affinities of compounds were compared to that of calcitriol, and they are presented in Table 1. Compound 5, unsubstituted at C-2, was practically devoid of binding affinity to the VDR, whereas its analog 6 with the 2-exomethylene group was twenty times less potent than calcitriol.

Biological Evaluation: Differentiation of HL-60 Cells
HL60 cells were used to determine prodifferentiating activities of new analogs [46]. After initial screening, the concentration ranges were established for each compound. Compound 5 was tested at concentrations from 3 × 10 −8 M to 10 −6 M, compound 6 at concentrations from 6.25 × 10 −8 M to 10 −6 M, while calcitriol was applied at concentrations from 10 −10 M to 10 −7 M. The cells were exposed to the compounds for 96 h and then the expression of differentiation markers CD14 and CD11b was studied using flow cytometry. CD14 is an antigen characteristic for monocytes and macrophages [47], while CD11b is an integrin present on monocytes, macrophages and granulocytes [48]. Percentages of CD14-and CD11b-positive cells were read out using Becton Dickinson Accuri C6 software (Becton Dickinson, San Jose, CA, USA). Half maximal effective concentrations (EC50) values were estimated from dose-response curves plotted with GraphPad Prism software. Compound 5 was not active in inducing differentiation at all. Compound 6, however less active than calcitriol, was able to significantly upregulate CD14 expression (Table 2) and CD11b (Table 3).

Biological Evaluation: Binding to the Vitamin D Receptor
The affinities of the synthesized compounds 5 and 6 to VDR were assessed by a fluorescence polarization (FP)-based competition assay. The VDR affinities of compounds were checked using a wide range of concentrations and compared to that of calcitriol. Dose-response curves were plotted using GraphPad Prism software (version 6.04, GraphPad Software, Inc., San Diego, CA, USA), and half maximal inhibitory concentration (IC 50 ) values were calculated from these dose-response curves. The binding affinities of compounds were compared to that of calcitriol, and they are presented in Table 1. Compound 5, unsubstituted at C-2, was practically devoid of binding affinity to the VDR, whereas its analog 6 with the 2-exomethylene group was twenty times less potent than calcitriol.

Biological Evaluation: Differentiation of HL-60 Cells
HL60 cells were used to determine prodifferentiating activities of new analogs [46]. After initial screening, the concentration ranges were established for each compound. Compound 5 was tested at concentrations from 3 × 10 −8 M to 10 −6 M, compound 6 at concentrations from 6.25 × 10 −8 M to 10 −6 M, while calcitriol was applied at concentrations from 10 −10 M to 10 −7 M. The cells were exposed to the compounds for 96 h and then the expression of differentiation markers CD14 and CD11b was studied using flow cytometry. CD14 is an antigen characteristic for monocytes and macrophages [47], while CD11b is an integrin present on monocytes, macrophages and granulocytes [48]. Percentages of CD14-and CD11b-positive cells were read out using Becton Dickinson Accuri C6 software (Becton Dickinson, San Jose, CA, USA). Half maximal effective concentrations (EC 50 ) values were estimated from dose-response curves plotted with GraphPad Prism software. Compound 5 was not active in inducing differentiation at all. Compound 6, however less active than calcitriol, was able to significantly upregulate CD14 expression (Table 2) and CD11b (Table 3).

Chemistry
Melting points (uncorrected) were determined on a SMP10 Stuart Scientific capillary melting point apparatus (Sunnyvale, CA, USA). Optical rotations were measured in chloroform using a Perkin-Elmer model 343 polarimeter (Shelton, CT, USA) at 24 • C. Ultraviolet (UV) absorption spectra were obtained on a Shimadzu UV-1800 UV spectrophotometer (Kyoto, Japan) in absolute ethanol. Nuclear magnetic resonance spectra were recorded in CDCl 3 solutions using Bruker AVANCE 300 MHz (Karlsruhe, Germany) and Bruker AVANCE 500 MHz instruments. Chemical shifts (δ) are reported in parts per million relative to (CH 3 ) 4 Si (δ 0.00) or solvent signal as an internal standard. Signals in 1 H NMR spectra are described using the following abbreviations: s-singlet, d-doublet, t-triplet, q-quartet, quint-quintet, sext-sextet, m-multiplet, br-broad, narr-narrow. High-resolution mass spectra were recorded on LCT time-of-flight(TOF) or Mass Quattro LC spectrometers using electrospray ionization (ESI) technique.
Reactions were usually carried out with magnetic stirring. All reactions involving moisture-or oxygen-sensitive compounds were carried out under dry argon atmosphere. Reaction temperatures refer to external bath temperatures. Tetrahydrofuran was distilled from Na/benzophenone; dichloromethane and toluene were distilled from P 2 O 5 , whereas pyridine, diisopropylamine, diethylamine and triethylamine were distilled from CaH 2 . The organic extracts were dried over anhydrous MgSO 4 , filtered and concentrated using a rotary evaporator at a water aspirator pressure (20-30 mmHg).
To a solution of the silylated tosylate (4.00 g, 10.75 mmol) in DMF (60 mL), LiCl (2.26 g, 53.76 mmol) was added and the mixture was heated at 80 • C for 24 h with stirring. The solvent was evaporated, the water was added to the residue and the mixture was extracted with ethyl acetate. The combined organic phases were dried (MgSO 4 ) and concentrated. The residue was purified by column chromatography over silica using hexane/diethyl ether (96:4) to give the chloro compound A (1.48 g, 58%) as a colorless oil. 1  The chloro compound A (244 mg, 1.04 mmol) was added dropwise to a vigorously stirred mixture of magnesium turnings (1.30 g) in anhydrous THF (4 mL) under argon. The reaction mixture was heated to reflux and second portion of chloride A (488 mg, 2.08 mmol) was added dropwise via a reflux condenser. After 10 min, a third portion of chloride A (488 mg, 2.08 mmol) was added and the mixture was refluxed for 30 min. The resulted solution of the Grignard reagent was diluted with THF (10 mL), cooled to −40 • C, and transferred to a cooled solution (−40 • C) of the tosylate 21 (615 mg, 1.30 mmol) in anhydrous THF (16 mL). After 15 min, a solution of CuI (0.986 g, 5.19 mmol) in anhydrous THF (8 mL) was added. The reaction mixture was stirred at −40 • C for 4 h. Then, it was quenched by addition of saturated NH 4 Cl and extracted with ethyl acetate. The combined organic phases were dried (MgSO 4 ) and concentrated. The residue was purified by flash chromatography over silica using hexane/diethyl ether (98:2) to afford diether 22 (476 mg, 72%) as a colorless oil. 1 (47 mL), tetrabutylammonium fluoride (1.0 M in THF; 18.9 mL, 18.9 mmol) was added at room temperature under argon. The stirring was continued for 30 min, brine was added, and the mixture was extracted with ethyl acetate. The organic extracts were dried (MgSO 4 ) and evaporated. The residue was purified by column chromatography over silica using hexane/ethyl acetate (9:1 ≥ 1:1) to afford a protected alcohol 23 (345 mg, 93%) as a colorless oil. 1  (1 R)-5-[1 ,5 -Dimethyl-5 -(triethylsilyl)oxy-hexyl]-1,2,3,4-tetrahydronaphthalen-1-one (24). Dess-Martin periodinane (563 mg, 1.33 mmol) was added to a solution of alcohol 23 (0.345 g, 0.89 mmol) in anhydrous methylene chloride (35 mL). The mixture was stirred at room temperature for 1 h under argon. Then, reaction was quenched with saturated Na 2 S 2 O 3 and saturated NaHCO 3 . The mixture was extracted with methylene chloride, and the organic phase was dried (MgSO 4 ) and evaporated. The residue was purified by column chromatography over silica using hexane/ethyl acetate (8:2) to afford ketone 24 (237 mg, 69%) as a colorless oil. Trifluoro-methanesulfonic acid (1 R)-5-[1 ,5 -dimethyl-5 -(triethylsilyl)-oxy-hexyl]-3,4-dihydronaphthalen-1-yl ester (10). To a solution of i-Pr 2 NH (69 µL, 0.49 mmol) in dry THF (1.5 mL) at −78 • C, n-BuLi (1.6 M solution in hexanes; 300 µL, 0.48 mmol) was dropwise added. The mixture was stirred at −78 • C for 1 h, and then a solution of the ketone 24 (116 mg, 0.30 mmol) in dry THF (0.9 mL) was added. The reaction mixture was stirred for 10 min, and a solution of N-phenyltrifluoromethanesulfonimide (160 mg, 0.45 mmol) in dry THF (0.9 mL) was transferred via cannula. The reaction mixture was stirred at −78 • C for 1 h and at room temperature for 30 min. Water was added and the mixture was extracted used for finding the global minimum structures was analogous to that described previously by us for the vitamin D side chain conformers [49] and involved the Conformational Search module. The calculated global minimum conformers were next energy-minimized using PCModel (release 9.0) program (Serena Software, Bloomington, IN, USA). Then, they were docked into the ligand binding pocket of the vitamin D receptor, extracted from crystalline hVDR (LBD)-1 complex (Protein Data Bank, Code: 1DB1), using Molegro Virtual Docker (release 4.0) program (CLC bio, Qiagen, Aarhus, Denmark).

Conclusions
Two compounds, characterized by a presence of dienyne moiety conjugated with the aromatic D ring, were successfully synthesized using convergent strategy. Despite the polyunsaturated nature of their structures, these compounds, contrary to the classical vitamin D analogs, cannot undergo undesired thermal isomerization to previtamin D forms. Compound with a 2-exomethylene substituent exhibited moderate affinity to the VDR predicted by molecular docking experiments. The reason of drastically lower binding activity of its counterpart unsubstituted at C-2 is not clear, but this fact remains in agreement with the literature data indicating that a presence of such A-ring methylene moiety can significantly increase the VDR affinity of calcitriol analogs [34,50]. One possible explanation of this effect comes from the examination of crystal structures of the VDR bound to differently C(2)-substituted (2α-methyl, -ethyl, -propyl, etc.) vitamin D analogs. It was found that a 2α-methyl substituent, present in the VDR superagonists characterized by high binding affinity, provides additional van der Waals contacts, while being small enough not to destroy the water network present in the channel located near C-2 [51]. It can be, therefore, possible that the 2-methylene moiety exerts an analogous effect, interacting with the receptor in a similar manner. The new class of compounds, presented in this work, can be further modified and optimized in a search for potential VDR ligands exhibiting selective biological activities.