Design and Synthesis of Novel Breast Cancer Therapeutic Drug Candidates Based upon the Hydrophobic Feedback Approach of Antiestrogens

Based upon hydrophobic feedback approaches, we designed and synthesized novel sulfur-containing ERα modulators (4 and 5) as breast cancer therapeutic drug candidates. The tetrahydrothiepine derivative 5a showed the highest binding affinity toward ERα because of its high hydrophobicity, and it acted as an agonist toward MCF-7 cell proliferation. The corresponding alkylamino derivative 5d maintained high binding affinity to ERα and potently inhibited MCF-7 cell proliferation (IC50: 0.09 μM). Docking simulation studies of compound 5d with the ERα BD revealed that the large hydrophobic moiety of compound 5d fit well into the hydrophobic pocket of the ERα LBD and that the sulfur atom of compound 5d formed a sulfur–π interaction with the amino acid residue His524 of the ERα LBD. These interactions play important roles for the binding affinity of compound 5d to the ERα LBD.


Introduction
17β-Estradiol (E2) is an endogenous female hormone and plays an important role in the female reproductive system. Estrogens including E2 influence the growth, differentiation, and functioning of many target tissues mediated by the binding to estrogen receptor α (ERα) [1][2][3]. Although activation of ERα is essential for the maintenance of homeostasis of female functions, they can sometimes induce and cause progression of estrogen-dependent breast cancers [4][5][6]. Several antiestrogens have been developed to prevent and control hormone-responsive breast cancer [7][8][9][10]. Tamoxifen has been used worldwide for more than 40 years for the treatment of estrogen-dependent breast cancer ( Figure 1) [11][12][13]. Interestingly, tamoxifen acts as either an agonist or an antagonist depending on tissue type; tamoxifen exhibits antiestrogenic action in breast cancer and hot flashes, and estrogenic action in bone and cholesterol metabolism, and is a selective estrogen receptor modulator (SERM) [11][12][13]. The dimethylaminoethyl chain of tamoxifen is attached to a benzene ring, where it plays a strategic role in the expression of antiestrogenic activity. The triphenylethylene moiety of tamoxifen plays an important role in controlling ERα binding affinity. The triphenylethylene moiety is a promising structure for the development of an ER ligand. The triphenylethylene structure exhibits geometric isomers E and Z that are caused by the key alkylamino chain, and the asymmetric hydrophobic part and the isomers easily isomerize between the E and Z forms. Therefore, the isomerization often become a big problem in the synthesis, purification, and preservation of an either isomer. Indeed, Figure 1. Structures of estrogen receptor (ER) modulators and newly designed sulfur-containing ER antagonist candidates (compounds 4a-4d and 5a-5d).

Chemistry
Synthesis of the dihydrothiophene derivatives (4a-4d) is summarized in Scheme 1. Commercially available 2-bromo-4′-methoxyacetophenone (6) was reacted with sodium sulfide (Na2S) to form the sulfide compound 7 in 95% yield [24]. Intramolecular McMurry coupling of 7 resulted in formation of the dihydrothiophene derivative 8 in 60% yield, which was demethylated with BBr3 to form the target compound 4a in 97% yield [25]. Although the dihydrothiophene derivative 8 was treated with 1 equimolar amount of m-chloroperbenzoic acid (m-CPBA) to form the sulfoxide compound 9 in 70% yield, followed by demethylation with BBr3 to form the decomposition On the other hand, we found that a simple 1,2-bis(4-hydroxyphenyl)-o-carborane, BE360 (1) exhibited high binding affinity toward ERα and estrogenic action in bone without showing estrogenic action in the uterus of ovariectomized (OVX) and orchidectomized (ORX) mice ( Figure 1) [22]. That is, compound 1 acted as an agonist in bone and an antagonist in female reproductive tissues, despite lack of an alkylamino chain. Based upon the hydrophobic feedback approaches, we focused on three-dimensional hydrocarbon units as new hydrophobic pharmacophores to design and synthesize BE1060 (2) and BE1054 (3) (Figure 1) [23]. Although the tetramethylcyclohexene-based bisphenol compound 3 acted as a partial agonist, similar to compound 1, the bicyclo[2,2,2]octene-based bisphenol compound 2 showed potent estrogenic activity. Unlike the triphenylethylene structure, a vicinal bisphenol structure does not have a geometric isomer, even with the key alkylamino chain. We were interested in the effects of ring size and sulfur atom in the hydrophobic structure on ER activity. Furthermore, to understand the effects of sulfur, sulfone, and sulfoxide on the biological activity, we designed dihydrothiophene and tetrahydrothiepine derivatives (4a-4d and 5a-5d) as novel ER antagonist candidates. Herein, we describe the facile synthesis and biological activities of the designated compounds 4a-4d and 5a-5d ( Figure 1).

Chemistry
Synthesis of the dihydrothiophene derivatives (4a-4d) is summarized in Scheme 1. Commercially available 2-bromo-4 -methoxyacetophenone (6) was reacted with sodium sulfide (Na 2 S) to form the sulfide compound 7 in 95% yield [24]. Intramolecular McMurry coupling of 7 resulted in formation of the dihydrothiophene derivative 8 in 60% yield, which was demethylated with BBr 3 to form the target compound 4a in 97% yield [25]. Although the dihydrothiophene derivative 8 was treated with 1 equimolar amount of m-chloroperbenzoic acid (m-CPBA) to form the sulfoxide compound 9 in 70% yield, followed by demethylation with BBr 3 to form the decomposition product of compound 9, it did not result in the formation of compound 4b. Thus, we carried out oxidation of compound 4a with m-CPBA to form the desired compound 4b in 73% yield. When compound 8 was treated with two equimolar amounts of m-CPBA, the corresponding sulfone compound 10 was obtained and was then demethylated with BBr 3 to form compound 4c in quantitative yield over two steps. Mitsunobu reaction of compound 4a with 3-dimethylamino-1-propanol resulted in formation of compound 4d with a dimethylaminopropyl chain in 37% yield [26].
Molecules 2019, 24, x FOR PEER REVIEW 3 of 12 product of compound 9, it did not result in the formation of compound 4b. Thus, we carried out oxidation of compound 4a with m-CPBA to form the desired compound 4b in 73% yield. When compound 8 was treated with two equimolar amounts of m-CPBA, the corresponding sulfone compound 10 was obtained and was then demethylated with BBr3 to form compound 4c in quantitative yield over two steps. Mitsunobu reaction of compound 4a with 3-dimethylamino-1propanol resulted in formation of compound 4d with a dimethylaminopropyl chain in 37% yield [26]. Scheme 2 summarizes the synthesis of tetrahydrothiepine derivatives (5a-5d). A starting material (12) for the seven-membered ring derivative 5 was synthesized by Friedel-Craft acylation of anisole with acyl chloride 11, followed by sulfide formation with Na2S to form compound 13 in 84% yield over two steps [24]. The tetrahydrothiepine derivatives (5a-5d) were synthesized in the same manner as those of compounds 4a-4d. Briefly, intramolecular McMurry coupling of compound 13, followed by demethylation, resulted in formation of the sulfide compound 5a in a 66% yield [25]. Compound 5a was transformed into the sulfoxide compound 5b by oxidation with one equimolar amount of m-CPBA in a 30% yield. Moreover, compound 5a was reacted with 3-dimethylamino-1propanol under Mitsunobu reaction conditions to form compound 5d in a 28% yield [26]. Treatment of compound 14 with two equimolar amounts of m-CPBA followed by demethylation resulted in the formation of the sulfone compound 5c in a 49% yield over two steps. Scheme 2 summarizes the synthesis of tetrahydrothiepine derivatives (5a-5d). A starting material (12) for the seven-membered ring derivative 5 was synthesized by Friedel-Craft acylation of anisole with acyl chloride 11, followed by sulfide formation with Na 2 S to form compound 13 in 84% yield over two steps [24]. The tetrahydrothiepine derivatives (5a-5d) were synthesized in the same manner as those of compounds 4a-4d. Briefly, intramolecular McMurry coupling of compound 13, followed by demethylation, resulted in formation of the sulfide compound 5a in a 66% yield [25]. Compound 5a was transformed into the sulfoxide compound 5b by oxidation with one equimolar amount of m-CPBA in a 30% yield. Moreover, compound 5a was reacted with 3-dimethylamino-1-propanol under Mitsunobu reaction conditions to form compound 5d in a 28% yield [26]. Treatment of compound 14 with two equimolar amounts of m-CPBA followed by demethylation resulted in the formation of the sulfone compound 5c in a 49% yield over two steps.

Biological Evaluation
The binding affinities of the synthesized bisphenols (compounds 4a-4c and 5a-5c) toward ERα were evaluated by means of a competitive binding assay using [6, H] 17β-estradiol and human recombinant ERα [27][28][29]. Figure 2A,B shows dose-response curves for competitive binding of the synthesized compounds 4a-4c and 5a-5c to ERα, respectively. Compound 4a, containing a sulfide group, moderately bound to ERα and the binding affinity was about 100 times lower than that of E2. The sulfoxide (4b) and sulfone (4c) compounds did not bind to the ERα ligand-binding domain (LBD), even at concentrations 1000 times higher than that of E2. The binding affinity of the tetrahydrothiepine derivative 5 toward ERα showed a similar tendency to that of compound 4, and the sulfide derivative 5a bound to ERα more potently than compound 4a. The tamoxifen-inspired derivatives 4d and 5d, both containing a dimethylaminopropyl group, showed similar binding affinities to compounds 4a and 5a, respectively. With the binding affinity of E2 to ERα taken as 100, the relative binding affinity (RBA) values of compounds 4d and 5d were 4.0 and 37.1, respectively (Table 1). It is suggested that the binding affinities of compounds 4 and 5 depend on the hydrophobicity of the ring structures and the polarity of the sulfur atoms, which can be understood from the CLogP values of compounds 4a-4d and 5a-5d in Table 1.

Biological Evaluation
The binding affinities of the synthesized bisphenols (compounds 4a-4c and 5a-5c) toward ERα were evaluated by means of a competitive binding assay using [6,7-3 H] 17β-estradiol and human recombinant ERα [27][28][29]. Figure 2A,B shows dose-response curves for competitive binding of the synthesized compounds 4a-4c and 5a-5c to ERα, respectively. Compound 4a, containing a sulfide group, moderately bound to ERα and the binding affinity was about 100 times lower than that of E2. The sulfoxide (4b) and sulfone (4c) compounds did not bind to the ERα ligand-binding domain (LBD), even at concentrations 1000 times higher than that of E2. The binding affinity of the tetrahydrothiepine derivative 5 toward ERα showed a similar tendency to that of compound 4, and the sulfide derivative 5a bound to ERα more potently than compound 4a. The tamoxifen-inspired derivatives 4d and 5d, both containing a dimethylaminopropyl group, showed similar binding affinities to compounds 4a and 5a, respectively. With the binding affinity of E2 to ERα taken as 100, the relative binding affinity (RBA) values of compounds 4d and 5d were 4.0 and 37.1, respectively (Table 1). It is suggested that the binding affinities of compounds 4 and 5 depend on the hydrophobicity of the ring structures and the polarity of the sulfur atoms, which can be understood from the CLogP values of compounds 4a-4d and 5a-5d in Table 1. and 5a-5c (B) toward ERα. All binding assays were performed in duplicate (n = 2). IC50 value of E2 for ERa is 4 nM. Table 1. Biological profiles of compounds 4a-4d and 5a-5d.
Compound Ring  a Relative binding affinity (RBA) values of each compound were estimated from the sigmoidal dose-response curves in Figure 2A,B. The binding affinity of E2 was taken as 100. b EC 50 and IC 50 values of the test compounds were estimated from the sigmoidal dose-response curves using GraphPad Prism 7 software. c E max indicates the maximal efficacy of each compound relative to that of E2, which was taken as 100. d CLogP values of each compound were estimated using ChemBioDraw Ultra 12.0. e "NT" indicates not tested.
Next, estrogenic and antiestrogenic activities of compounds 4a-4d and 5a-5d, except for 4b and 5b (poor binding affinity toward ERa), were evaluated by means of a cell proliferation assay using the human breast cancer cell line MCF-7, which shows ER-dependent growth [27][28][29]. Table 1 summarizes the EC 50 , IC 50 , and E max values estimated from the MCF-7 cell proliferation assays. E max represents the maximal efficacy of each compound toward cell proliferation and is an index of ER partial agonistic activity. E max was estimated relative to that of E2, which was taken as 100. Although the binding affinity of compound 4a was~10 times higher than that of compound 4c, both compounds showed estrogenic activity with similar EC 50 values. Based on the E max values of compounds 4a and 4c, we suggest that the complex formed between compound 4c and the ERα LBD is more active for transcription than that formed between compound 4a and the ERα LBD. Compound 5a, which exhibited the most potent binding affinity to ERα LBD, also showed the most potent MCF-7 cell proliferation activity, with an EC 50 value of 0.048 µM. As expected from the weak binding affinity to ERα LBD, compound 5c showed weak estrogenic activity. On the other hand, the tamoxifen-inspired derivatives 4d and 5d, containing alkylamino chains, potently inhibited cell proliferation of MCF-7 cells induced by 0.1 nM of E2, with IC 50 values of 0.41 and 0.09 µM, respectively.

Docking Study
To understand the docking modes of the sulfur-containing ligands 4d and 5d with the ERα LBD, docking simulation studies of both compounds toward the human ERα LBD (PDB: 1ERR) were carried out using an automatic docking program (Discovery Studio 2018/CDOCKER) [30]. The docking mode of compound 4d was superimposed onto that of compound 5d at the ERα LBD ( Figure 3A,B). The docking mode of compound 4d toward the ERα LBD was similar to that of compound 5d. Each phenolic hydroxy group of compounds 4d and 5d formed hydrogen bonds with the amino acid residues Glu353 and Arg394 ( Figure 3C), and each dimethylaminopropyl chain of compounds 4d and 5d was found to be located within the same space of the ERα LBD ( Figure 3A). The hydrophobic moiety of compound 5d was spread widely across the hydrophobic pocket of the ERα LBD and the sulfur atom of the tetrahydrothiepine ring formed a sulfur-π interaction with the amino acid residue His524 ( Figure 3B,C). Docking scores (CDOCKER interaction energy) of compounds 4d and 5d toward the ERα LBD were 51.1 and 54.0, respectively, and the order of docking score was correlated with the results of the binding assay. Therefore, we attributed the potent binding affinity of compound 5d toward the ERα LBD to both hydrophobic interactions and sulfur-π interactions. moiety of compound 5d was spread widely across the hydrophobic pocket of the ERα LBD and the sulfur atom of the tetrahydrothiepine ring formed a sulfur-π interaction with the amino acid residue His524 ( Figure 3B,C). Docking scores (CDOCKER interaction energy) of compounds 4d and 5d toward the ERα LBD were 51.1 and 54.0, respectively, and the order of docking score was correlated with the results of the binding assay. Therefore, we attributed the potent binding affinity of compound 5d toward the ERα LBD to both hydrophobic interactions and sulfur-π interactions.

General Consideration
Melting points were determined with an MP-J3 micro melting point apparatus (Yanaco, Kyoto, Japan) and were not corrected. 1 H NMR and 13 C NMR spectra were recorded with JNM-LA-400 spectrometer (JEOL, Tokyo, Japan). Chemical shifts for 1 H NMR spectra were referenced to tetramethylsilane (0.0 ppm) as an internal standard. Chemical shifts for 13 C NMR spectra were

Competitive Binding Assay Using Human ER
The ligand binding activity of human estrogen receptor α (ER α) was determined by a nitrocellulose filter binding assay method. ER α was diluted with a binding assay buffer (20 mM Tris-HCl pH 8.0, 0.3 M NaCl, 1 mM EDTA pH 8.0, 10 mM 2-mercaptoethanol, 0.2 mM phenylmethylsulfonyl floride) and incubated with 4 nM [6, 7-3 H]-17β-estradiol in the presence or absence of an unlabeled competitor at 4 • C for 18 h. The incubation mixture was absorbed by suction onto a nitrocellulose membrane that had been soaked in binding assay buffer. The membrane was washed two times with buffer (20 mM Tris-HCl pH 8.0, 0.15 M NaCl) and with 25% ethanol in distilled water. Radioactivity that remained in the membrane was measured in Atomlight by using a liquid scintillation counter.

MCF-7 Cell Proliferation Assay
The human breast adenocarcinoma cell line MCF-7 was routinely cultivated in DMEM supplemented with 10% FBS and 100 IU/mL penicillin and 100 mg/mL streptomycin at 37 • C in a 5% CO 2 humidified incubator. Before an assay, MCF-7 cells were switched to DMEM (low glucose phenol red-free supplemented with 5% FBS, and 100 IU/mL penicillin and 100 mg/mL streptomycin). Cells were trypsinized from the maintenance dish with phenol red-free trypsin-EDTA and seeded in a 96-well plate at a density of 2000 cells per final volume of 100 µL DMEM supplemented with 5% stripped FBS and 100 IU/mL penicillin and 100 mg/mL streptomycin. After 24 h, the medium was removed and 90 µL of fresh medium and 10 µL of drug solution, supplemented with serial dilutions of test compounds or DMSO as dilute control in the presence or absence of 0.1 nM E2, were added to triplicate microcultures. Cells were incubated for four days. At the end of the incubation time, number of cells was counted by using the WST-8, which was added to microcultures 10 µL each, and they were incubated for 2-4 h. The absorbance at 450 nm was measured. This parameter relates to and number of living cells in the culture.

Docking Simulation Study
Three-dimensional (3D) structures of protein-ligand complexes were predicted using the Discovery Studio 2018/CDOCKER software (BIOVIA) with default settings. The 3D structures of ER used in this study were retrieved from the RCSB Protein Data Bank (PDB ID: 1ERR). Missing hydrogen atoms in the crystal structure were computationally added and the center of the active site was defined as the center of raloxifen in 1ERR. The conformations of 4d and 5d were optimized using the CHARMm force field, and the docking simulations of 4d and 5d with 1ERR were performed using the CDOCKER protocol. The docking poses for 4d and 5d with the highest CDOCKER ENERGY were selected for the discussions of 10 docking modes, respectively.

Conclusions
Novel sulfur-containing ERα modulators (compounds 4 and 5) as potential breast cancer therapeutic drug candidates were designed and synthesized based upon the hydrophobic feedback approach for the simple bisphenols 1-3 developed in our previous studies. Tetrahydrothiepine derivatives (compounds 5a-5d) showed higher binding affinity toward ERα than the corresponding dihydrothiophene derivatives (compounds 4a-4d) because of the hydrophobicity of the sulfur-containing ring structure. Although the bisphenol derivatives 4a, 4c, 5a, and 5c showed estrogenic activity toward the MCF-7 breast cancer cell line, the corresponding alkylamino derivatives (compounds 4d and 5d) acted as ER antagonists. In particular, compound 5d showed the most potent antiestrogenic activity among the tested compounds with an IC 50 value of 0.09 µM. A sulfur-containing structure might be a promising scaffold for antiestrogen discovery, owing to ease of synthesis, binding modes toward the ERα LBD, and availability of various substituted derivatives.