Core Structure–Activity Relationship Studies of 5,7,20-O-Trimethylsilybins in Prostate Cancer Cell Models

Silibinin, also known as silybin, is isolated from milk thistle (Silybum marianum). Silibinin has been demonstrated to be a good lead compound due to its potential to prevent and treat prostate cancer. Its moderate potency and poor pharmacokinetic profile hindered it from moving forward to therapeutic use. Our research group has been working on optimizing silibinin for the potential treatment of castration-resistant prostate cancer. Our previous studies established 5,7,20-O-trimethylsilybins as promising lead compounds as they can selectively suppress androgen receptor (AR)-positive LNCaP cell proliferation. Encouraged by the promising data, the present study aims to investigate the relationships between the core structure of 5,7,20-O-trimethylsilybin and their antiproliferative activities towards AR-positive (LNCaP) and AR-negative prostate cancer cell lines (PC-3 and DU145). The structure–activity relationships among the four different core structures (including flavanonol-type flavonolignan (silibinin), flavone-type flavonolignan (hydnocarpin D), chalcone-type flavonolignan, and taxifolin (a flavonolignan precursor) indicate that 5,7,20-O-trimethylsilybins are the most promising scaffold to selectively suppress AR-positive LNCaP prostate cancer cell proliferation. Further investigation on the antiproliferative potency of their optically enriched versions of the most promising 5,7,20-O-trimethylsilybins led to the conclusion that (10R,11R) derivatives (silybin A series) are more potent than (10S,11S) derivatives (silybin B series) in suppressing AR positive LNCaP cell proliferation.

Six 5,7,20-O-trimethylhydnocarpin D derivatives (17)(18)(19)(20)(21)(22) were evaluated for their antiproliferative activity towards AR-positive (LNCaP) and AR-negative (PC-3 and DU145) prostate cancer cell lines by WST-1 cell proliferation assay. As demonstrated in Table 2  1 IC 50 is the compound concentration effective in inhibiting 50% cell proliferation measured by WST-1 cell proliferation assay after 3-day exposure. The data were presented as the mean ± SD from n = 3. 2 Human androgen receptor-negative prostate cancer cell line derived from bone metastasis of prostate tumor. 3 Human androgen receptor-negative prostate cancer cell line derived from brain metastasis of prostate tumor. 4 Human androgen receptor-positive prostate cancer cell line derived from lymph node metastasis of prostate tumor.

Synthesis, Structure Characterization, and Antiproliferative Evaluation of Chalcone-Type Flavonolignans
So far, only one chalcone-type flavonolignan was prepared from silibinin (1) [20], and none were isolated from nature. Chalcone has been evidenced as a privileged scaffold in the field of drug design and drug discovery due to its robust medicinal properties [21]. Certain natural or synthetic chalcones have been revealed to have appreciable in vitro antiproliferative activity in prostate cancer cell models at sub-micromolar to micromolar concentration [22][23][24][25][26][27][28][29]. The in vivo antitumor efficacy of some chalcones has also been confirmed by the animal experimental data [26,28]. Some chalcones have been demonstrated to boost prostate cancer cell apoptosis mediated by TRAIL (tumor necrosis factor-related apoptosis-inducing ligand) [30]. The anti-prostate cancer activity is associated with different mechanisms of action, including inhibition of 5α-reductase, androgen receptor translocation, and sexual hormone synthesis [31]. It is thus intriguing to investigate the capability of chalcone-type flavonolignans in suppressing prostate cancer cell proliferation, especially in comparison with that of 5,7,20-O-trimethylsilibins. As shown in Figure 4 and Scheme 3, a chalcone-type flavonolignan (23) was synthesized from 23-O-TBS-5,7,20-O-trimethylsilybin (12) through metal-free iodine-mediated deoxygenation of 3-OH followed by opening ring C (breaking O-1 and C-2 bond) by treating with triphenylphosphine (PPh 3 ), imidazole, and iodine [20]. Removing the TBS group in 23 gave chalcone-type flavonolignan 24 with one phenolic hydroxyl at C-8a and aliphatic hydroxyl at C-23. At this point, the chemical manipulation of 8a-OH and 23-OH furnished an additional eleven chalcone-type flavonolignans ( Figure 4 and Scheme 3).
Chalcone-type flavonolignans are structurally distinguished from silibinin and hydnocarpin D because of the trans α,β-unsaturated ketone derived from the ring C opening. The chemical structure of chalcone-type flavonolignan 23 was elucidated based on its NMR and HRMS. The downshifts of H-3 at 4.42 ppm and H-2 at 4.92 ppm to 7.79 ppm (d, J = 15.9 Hz) and 7.71 ppm (d, J = 15.9 Hz) imply the existence of the characteristic trans α,β-unsaturated ketone, suggesting the formation of the chalcone scaffold.   Chalcone-type flavonolignans are structurally distinguished from silibinin and hydnocarpin D because of the trans α,β-unsaturated ketone derived from the ring C opening. The chemical structure of chalcone-type flavonolignan 23 was elucidated based on its NMR and HRMS. The downshifts of H-3 at 4.42 ppm and H-2 at 4.92 ppm to 7.79 ppm (d, J = 15.9 Hz) and 7.71 ppm (d, J = 15.9 Hz) imply the existence of the characteristic trans α,β-unsaturated ketone, suggesting the formation of the chalcone scaffold.
Thirteen chalcone-type flavonolignans were synthesized for the evaluation of their antiproliferative activities towards AR-positive (LNCaP) and AR-negative (PC-3 and DU-   Chalcone-type flavonolignans are structurally distinguished from silibinin and hydnocarpin D because of the trans α,β-unsaturated ketone derived from the ring C opening. The chemical structure of chalcone-type flavonolignan 23 was elucidated based on its NMR and HRMS. The downshifts of H-3 at 4.42 ppm and H-2 at 4.92 ppm to 7.79 ppm (d, J = 15.9 Hz) and 7.71 ppm (d, J = 15.9 Hz) imply the existence of the characteristic trans α,β-unsaturated ketone, suggesting the formation of the chalcone scaffold.
Thirteen chalcone-type flavonolignans were synthesized for the evaluation of their antiproliferative activities towards AR-positive (LNCaP) and AR-negative (PC-3 and DU- Thirteen chalcone-type flavonolignans were synthesized for the evaluation of their antiproliferative activities towards AR-positive (LNCaP) and AR-negative (PC-3 and DU-145) prostate cancer cell lines by WST-1 cell proliferation assay (Table 3).  1 IC 50 is the compound concentration effective in inhibiting 50% cell proliferation measured by WST-1 cell proliferation assay after 3-day exposure. The data were presented as the mean ± SD from n = 3. 2 Human androgen receptor-negative prostate cancer cell line derived from bone metastasis of prostate tumor. 3 Human androgen receptor-negative prostate cancer cell line derived from brain metastasis of prostate tumor. 4 Human androgen receptor-positive prostate cancer cell line derived from lymph node metastasis of prostate tumor.

Antiproliferative Evaluation of Taxifolin Derivatives
Taxifolin ( Figure 1) is the biogenetical flavonoid precursor for silybin A and silybin B [32]. The AR-positive LNCaP prostate cancer cells were reported to possess a similar sensitivity to taxifolin and silibinin [5]. It is thus interesting to evaluate the antiproliferative activity of trimethyltaxifolin (36) and its 3-O-(thio)carbamoyl derivatives (37-39) ( Figure 5) on AR-positive and AR-negative prostate cancer cell models. The four taxifolin derivatives were synthesized according to the procedures reported in our earlier publication (Scheme 4) [16]. As summarized in Table 4, all four derivatives (36-39) cannot suppress cell proliferation in either AR-positive or AR-negative prostate cancer cell models up to 25 µM, suggesting the lignan portion is imperative to the impressive antiproliferitive potency of silibinin derivatives in the AR-positive LNCaP cell model. This is also in agreement with the notation that flavonolignans are generally more potent than their respective flavonoid precursor in prostate cancer cell models [14].     1 IC50 is the compound concentration effective in inhibiting 50% cell proliferation measured by WST-1 cell proliferation assay after 3-day exposure. The data were presented as the mean ± SD from n = 3. 2 Human androgen receptor-negative prostate cancer cell line derived from bone metastasis of prostate tumor. 3 Human androgen receptor-negative prostate cancer cell line derived from brain metastasis of prostate tumor. 4 Human androgen receptor-positive prostate cancer cell line derived from lymph node metastasis of prostate tumor.

Synthesis and Antiproliferative Evaluation of Optically Enriched 5,7,20-O-Trimethylsilybins
The different diastereomeric isomers of silibinin and derivatives have been evidenced to have different biological effects [33,34]. It is thus necessary to further evaluate the antiproliferative potency of the optically pure version for optimal derivatives. The fact that 5,7,20-O-trimethylsilybin (5) and its derivatives (6)(7)(8)(9)(10)(11)16) possess the optimal antiproliferative potency and selectivity in the AR-positive LNCaP cells spurred us to further investigate the antiproliferative potency and selectivity of optically enriched 5,7,20-O-  1 IC 50 is the compound concentration effective in inhibiting 50% cell proliferation measured by WST-1 cell proliferation assay after 3-day exposure. The data were presented as the mean ± SD from n = 3. 2 Human androgen receptor-negative prostate cancer cell line derived from bone metastasis of prostate tumor. 3 Human androgen receptor-negative prostate cancer cell line derived from brain metastasis of prostate tumor. 4 Human androgen receptor-positive prostate cancer cell line derived from lymph node metastasis of prostate tumor.
trimethylsilybin A (5A) and B (5B) and their derivatives. Silybin A (1A), silybin B (1B), 23-O-acetylsilybin A (40A), and 23-O-acetylsilybin B (40B) were prepared from diastereomeric silibinin employing the selective transesterification of silibinin (1) and stereoselective alcoholysis of 23-O-acetylsilybin (40) based on the reported procedure [2,35]. Novozym 435 was used as a biocatalyst to discriminate the diastereoisomers. The optically enriched versions of derivatives 6-11 were synthesized from the optically enriched 5,7,20-Otrimethysilybin A (5A) and B (5B) as illustrated in Figure 6 and Scheme 5.  To confirm the optical purity of silybin A (1A) and B (1B) isolated from diastereomeric silibinin (1), the specific rotation values were measured in acetone and compared to the reported values (Table 5) [35]. Our experimental specific rotation value for silybin A (1A) is [α]D 22 = +15.14 (c 0.7, acetone) and for silybin B (1B) is [α]D 22 = +2.28 (c 1.97, acetone); by comparing the data reported by previous studies, Silybin A (1A) and B (1B) were confirmed to be separated successfully (Table 5) (Figure 7). The H-3 assignment was supported by the following critical correlation peaks in the set of 2D NMR spectra for 5 (refer to Figures S169- To confirm the optical purity of silybin A (1A) and B (1B) isolated from diastereomeric silibinin (1), the specific rotation values were measured in acetone and compared to the reported values ( ; by comparing the data reported by previous studies, Silybin A (1A) and B (1B) were confirmed to be separated successfully (Table 5) Figure 8). The H-3 signal assignment of 6 has been confirmed in our previous research according to the critical HMBC correlations from H-3 to H-4, carbonyl carbon from the dimethyl carbamoyl group, and C-14 [16]. The fact that (10R,11R) derivatives (5,7,20-O-trimethylsilybin A series) and their (10S,11S) counter partners have slightly different chemical shifts at H-3 is probably caused by the different conformation of ring C and different orientations of H-3 about the aromatic ring B (Figure 9). The antiproliferative potency and selectivity of ten pairs of diastereomers were evaluated on AR-positive (LNCaP and 22Rv1) and AR-negative (PC-3 and DU145) human prostate cancer cell lines. The IC50 values are summarized in Table 6 and reveal that (10R,11R) derivatives (silybin A series) possess a significantly greater antiproliferative potency and higher selectivity than (10S,11S) derivatives (silybin B series) towards LNCaP prostate cancer cell lines. Derivatives 8A and 41A were identified as the optimal derivatives with an IC50 value of 0.07 M in the LNCaP cell model. However, these derivatives cannot suppress 22Rv1 prostate cancer cell proliferation up to a 10 M concentration, revealing that the derivatives of 5,7,20-O-trimethylsilybin very likely bind to the ligandbinding domain on AR to exhibit antiproliferative activity in LNCaP cells.      The antiproliferative potency and selectivity of ten pairs of diastereomers were evaluated on AR-positive (LNCaP and 22Rv1) and AR-negative (PC-3 and DU145) human prostate cancer cell lines. The IC 50 values are summarized in Table 6 and reveal that (10R,11R) derivatives (silybin A series) possess a significantly greater antiproliferative potency and higher selectivity than (10S,11S) derivatives (silybin B series) towards LNCaP prostate cancer cell lines. Derivatives 8A and 41A were identified as the optimal derivatives with an IC 50 value of 0.07 µM in the LNCaP cell model. However, these derivatives cannot suppress 22Rv1 prostate cancer cell proliferation up to a 10 µM concentration, revealing that the derivatives of 5,7,20-O-trimethylsilybin very likely bind to the ligand-binding domain on AR to exhibit antiproliferative activity in LNCaP cells. 1 IC 50 is the compound concentration effective in inhibiting 50% cell proliferation measured by WST-1 cell proliferation assay after 3-day exposure. The data were presented as the mean ± SD from n = 3. 2 Human AR-negative prostate cancer cell line derived from bone metastasis of prostate tumor. 3 Human AR-negative prostate cancer cell line derived from brain metastasis of prostate tumor. 4 Human AR-positive prostate cancer cell line derived from lymph node metastasis of prostate tumor. 5 Human AR-positive prostate cancer cell line derived from a castration-resistant xenograft.

General Procedures
HRMS were obtained on a Thermo Scientific Q-Exactive mass spectrometer with electrospray ionization (ESI). NMR spectra were obtained on a Bruker Fourier 300 spectrometer in CDCl 3 . The chemical shifts are given in ppm referenced to the respective solvent peak, and coupling constants are reported in Hz. All reagents and solvents were purchased from commercial sources and were used without further purification. Silica gel column chromatography was performed using silica gel (32-63 µm). Preparative thinlayer chromatography (PTLC) separations were carried out on thin layer chromatography plates loaded with silica gel 60 GF254 (EMD Millipore Corporation, MA, USA). 5,7,20-O-Trimethylsilybin (5, HPLC purity 96.3%) was synthesized from silibinin (>98%, purchased from Fischer Scientific, Waltham, MA, USA) using the procedure previously described by us [36]. The synthesis and physical data of compounds 36-39 have been included in our recent publication [16].  [2,35]. The HPLC purity analyses were performed on an Agilent Hewlett-Packard 1100 series HPLC DAD system using a 5 µM C18 reversed phase column (4.6 × 250 mm) and a diode array detector. Solvent A is methanol and solvent B is 5% methanol in DI water. All testing samples were run 30 min of 35-100% A in B, with 20 min gradient. The flow rate is 1 mL/min.

Synthesis of 3-O-Mesyl-5,7,20-O-Trimethylsilybin (15)
HF·Py (1.4 mL, 1.6 M) was added dropwise into the solution of 14 (71.6 mg, 0.1 mmol) in THF (1 mL, 0.1M) in a 5 mL plastic reaction vial at 0 • C. The reaction mixture was stirred for 1 h and then transferred into a saturated sodium bicarbonate solution (20 mL). The resulting mixture was extracted with ethyl acetate (10 mL × 3). The organic layers were combined, dried over anhydrous sodium sulfate, and concentrated in vacuo to produce the alcohol as a yellow foam in 55% yield. 1 13

Synthesis of 5,7,20-O-Trimethylhydnocarpin D (17)
To a solution of 18 (237 mg, 0.38 mmol, 1.0 eq) in THF (3.8 mL) at 0 • C was added dropwise HF·Py (1.6 M, 5.4 mL) via a needle. The reaction solution was stirred for 3 h at room temperature before being transferred into a separatory funnel with saturated sodium bicarbonate (50 mL), which was extracted with ethyl acetate (15 mL × 3). The extracts were dried over anhydrous sodium sulfate and concentrated. The crude product was purified by PTLC eluting with hexane:ethyl acetate (3:4, v/v) to furnish the alcohol in 89% as a yellow solid. 1 13   The reaction solution was stirred overnight at 60 • C prior to being quenched with brine (20 mL). The subsequent mixture was extracted with EtOAc (10 mL × 3). The organic layers were combined, dried over anhydrous sodium sulfate, and concentrated in vacuo. The crude product was subjected to PTLC purification eluting with EtOAc to furnish the desired product as a yellow syrup in 5% yield.
Method 2: NaH (60% in mineral oil, 6.2 mg, 0.15 mmol) was added into the solution of 17 (52 mg, 0.10 mmol) in THF (1.0 mL) at 0 • C. The mixture was stirred for 30 min before adding methanesulfonyl chloride (0.02 mL, 0.26 mmol). The reaction was then refluxed overnight before adding water (20 mL) to quench the reaction. The subsequent mixture was extracted with EtOAc (10 mL × 3), the organic layers were dried over anhydrous sodium sulfate, and the organic solvents were removed in vacuo. The crude product was purified twice via PTLC eluting with DCM: MeOH (95:5, v/v) and pure EtOAc, respectively, to produce the desired product as a yellow foam in 25% yield. 1 13 4 mL), NaH (60%, 15 mg, 0.36 mmol) at 0 • C was added, and the reaction mixture was stirred for 30 min before adding dimethylcarbamoyl chloride (0.033 mL, 0.36 mmol). The reaction mixture was then refluxed overnight under argon before adding water (20 mL) to quench the reaction. The subsequent mixture was extracted with EtOAc (10 mL × 3), the EtOAc extracts were dried over anhydrous sodium sulfate, and the organic solvents were removed in vacuo. The crude product was subjected to a three-time PTLC purification eluting with EtOAc:hexane (7:3, v/v), DCM:MeOH (95:5, v/v) and pure EtOAc, respectively, to provide the desired product as a yellow foam in 46%.
Method 2: Triethylamine (0.027 mL, 0.20 mmol), dimethylcarbamyl chloride (0.018 mL, 0.20 mmol), and DMAP (12 mg, 0.10 mmol) were added to the solution of 17 (50 mg, 0.10 mmol) in DCM (1.0 mL), and the reaction solution was refluxed overnight before being quenched with saturated ammonium chloride (20 mL). The resulting mixture was then extracted with ethyl acetate (10 mL × 3), the combined extracts were dried over anhydrous sodium sulfate, and the organic solvents were removed in vacuo. The crude product was sequentially purified via PTLC twice eluting with DCM:MeOH (97:3, v/v) and pure EtOAc, respectively, to produce 20 as a yellow foam in 8% yield. 1 13   NaH (60% in mineral oil, 6 mg, 0.15 mmol) was added to the solution of 17 (49 mg, 0.10 mmol) in THF (1.0 mL) at 0 • C, and the suspension was for 30 min before adding diethylcarbamoyl chloride (0.02 mL, 0.15 mmol). The reaction mixture was refluxed overnight under argon prior to being quenched with DI water (20 mL). The resulting mixture was extracted with EtOAc (10 mL × 3), the combined organic layers were dried over anhydrous sodium sulfate, and the organic solvents were removed in vacuo. The crude product was subjected to two sequential PTLC purification eluting with DCM:MeOH (95:5, v/v) and pure EtOAc, respectively, to produce the desired product as a yellow foam in 43% yield. 1 13   NaH (60% in mineral oil, 6 mg, 0.15 mmol) was added to the solution of 17 (50 mg, 0.10 mmol) in THF (1.0 mL) at 0 • C, and the suspension was stirred for 30 min before adding dimethylthiocarbamoyl chloride (18 mg, 0.15 mmol). The reaction was allowed to proceed with refluxing overnight under argon prior to being quenched with DI water (20 mL). The subsequent mixture was extracted with EtOAc (10 mL × 3), the organic extracts were combined and dried over with sodium sulfate, and the organic solvents were removed under vacuum. The crude product was sequentially purified twice via PTLC. The first PTLC used EtOAc:hexane (7:3, v/v) as eluent, while the second PTLC developed twice sequentially eluting with DCM:MeOH (95:5, v/v) and hexane:ethyl acetate (4:1, v/v) to furnish the desired thiocarbonate as a yellow foam in 18% yield. 1

Synthesis of Chalcone-Type Flavonolignan 23
Iodine (220 mg, 0.87 mmol) was added to the solution of imidazole (71 mg, 1.0 mmol) and triphenylphosphine (250 mg, 0.95 mmol) in DCM (3.8 mL) at 0 • C, and the reaction mixture was stirred for 10 min before adding the solution of 12 (302 mg, 0.47 mmol) in DCM (0.95 mL). The reaction was allowed to proceed with heating at 50 • C overnight prior to being quenched with a saturated sodium thiosulfate solution (20 mL). The subsequent mixture was extracted with ethyl acetate (10 mL × 3), the organic extracts were combined and dried over anhydrous sodium sulfate, and the organic solvents were removed in vacuo. The crude product was subjected to PTLC purification eluting with hexane:ethyl acetate (1:1, v/v) to produce chalcone 23 as a yellow oil in 43% yield. 1 13

Synthesis of Chalcone-Type Flavonolignan 24
HF·Py (1.6 M, 1.35 mL) was added dropwise to the solution of 23 (60 mg, 0.10 mmol) in THF (1.0 mL) at 0 • C. The reaction solution was stirred for 3 h at room temperature, and a saturated sodium bicarbonate solution (50 mL) was used to quench the reaction. The subsequent mixture was extracted with ethyl acetate (15 mL × 3), and the combined extracts were dried over anhydrous sodium sulfate and concentrated. PTLC purification of the crude product, using hexane: ethyl acetate (3:4, v/v) as eluent, produced chalcone 24 as a yellow solid in 84% yield. 1 13

Synthesis of Chalcone-Type Flavonolignan 25
NaH (60% in mineral oil, 7 mg, 0.18 mmol) was added to the solution of 24 (45 mg, 0.09 mmol) in THF (0.09 mL) at 0 • C, and the reaction mixture was stirred for 30 min before adding diethylcarbamoyl chloride (0.02 mL, 0.18 mmol). The reaction was allowed to proceed with stirring overnight at room temperature before being quenched with DI water (20 mL). The resulting mixture was extracted with ethyl acetate (20 mL × 3), and the combined extracts were dried over anhydrous sodium sulfate and concentrated in vacuo. The crude product was subjected to three sequential PTLC purifications, eluting with DCM:MeOH (97:3, v/v), EtOAch:hexane (7:3, v/v) and EtOAc:hexane (3:2, v/v), respectively, to afford chalcone 25 as a yellow solid in 54% yield. 1

Synthesis of Chalcone-Type Flavonolignan 26
Method 1: NaH (60% in mineral oil, 7 mg, 0.18 mmol) was added to the solution of 24 (45 mg, 0.09 mmol) in THF (0.09 mL) at 0 • C, and the reaction suspension was stirred for 30 min before adding diethylcarbamoyl chloride (0.02 mL, 0.18 mmol). The reaction was allowed to proceed with stirring at room temperature overnight prior to being quenched with DI water (20 mL). The subsequent mixture was extracted with ethyl acetate (20 mL × 3), the combined extracts were dried over anhydrous sodium sulfate, and the organic solvents were concentrated in vacuo. The crude product was purified via PTLC eluting with DCM:MeOH (95:5, v/v) to produce 26 as a yellow wax in 39% yield.
Method 2: Triethylamine (0.06 mL, 0.44 mmol), diethylcarbamyl chloride (0.06 mL, 0.44 mmol), and DMAP (14 mg, 0.11 mmol) were sequentially added to the solution of 24 (58 mg, 0.11 mmol) in DCM (1.1 mL). The reaction mixture was stirred at room temperature under argon for 4 h, and then saturated ammonium chloride (20 mL) was added to quench the reaction. The resulting mixture was extracted with ethyl acetate (10 mL × 3), the combined extracts were dried over anhydrous sodium sulfate, and the organic solvents were removed in vacuo. PTLC purification of the crude product, eluting with EtOAc:hexane (3:2, v/v), yielded 26 as a yellow wax in 32% yield. 1

Synthesis of Chalcone-Type Flavonolignan 27
NaH (60% in mineral oil, 7 mg, 0.18 mmol) was added to the solution of 24 (45 mg, 0.09 mmol) in THF (0.09 mL) at 0 • C. After stirring the reaction mixture for 30 min, methanesulfonyl chloride (0.02 mL, 0.18 mmol) was added. The reaction mixture was continued to stir overnight at room temperature, and then DI water (20 mL) was added to quench the reaction. The resulting mixture was extracted with ethyl acetate (20 mL × 3), the combined extracts were dried over anhydrous sodium sulfate, and the organic solvents were removed in vacuo. The crude product was sequentially subjected to twice PTLC purification eluting with DCM:MeOH (97:3, v/v) and EtOAc:hexane (7:3, v/v), respectively to produce 27 as a yellow solid in 33% yield. 1

Synthesis of Chalcone-Type Flavonolignan 28 and 29
NaH (60% in mineral oil, 11 mg, 0.28 mmol) was added to the solution of 24 (67 mg, 0.13 mmol) in THF (0.13 mL) at 0 • C, and the mixture was stirred for 30 min before adding methansulfonyl chloride (0.02 mL, 0.26 mmol). The reaction mixture was then stirred at room temperature overnight. DI water (20 mL) was then added to quench the reaction. The subsequent mixture was extracted with ethyl acetate (20 mL × 3), the combined extracts were dried over anhydrous sodium sulfate, and the organic solvents were removed in vacuo. The crude product was purified twice via PTLC using DCM:MeOH (95:5, v/v) and EtOAc:hexane (7:3, v/v) as eluent to produce chalcone-type flavonolignans 28 and 29.

Cell Culture
All prostate cancer cell lines were originally purchased from American Type Culture Collection (ATCC). The PC-3, LNCaP, and 22Rv1 prostate cancer cells were routinely cultured in RPMI-1640 medium supplemented with 10% FBS and 1% penicillin/streptomycin. The DU145 prostate cancer cell line was routinely cultured in Eagle's Minimum Essential Medium (EMEM) supplemented with 10% FBS and 1% penicillin/streptomycin. Cultures were maintained in a high humidity environment supplemented with 5% carbon dioxide at a temperature of 37 • C.

WST-1 Cell Proliferation Assay
PC-3, DU145, and LNCaP cells were placed in 96-well plates at a density of 3200 each well in 200 µL of culture medium. The 22Rv1 cells were placed in 96-well plates at a density of 6400 each well in 200 µL of culture medium. The cells were cultured for 16 h and then treated with enzalutamide or silibinin (both as positive controls) or synthesized derivatives at five different doses for 3 days. Equal treatment volumes (200 µL) of DMSO (0.25%) in medium were used as vehicle control. The cells were further cultured in a CO 2 incubator at 37 • C for three days. Ten µL of the premixed WST-1 cell proliferation reagent (Takara Bio USA, Inc., San Jose, CA, USA) was added to each well. After mixing gently for 1 min on an orbital shaker, the cells were cultured for an additional 3 h at 37 • C. The absorbance of each well was measured using a microplate reader (Synergy HT, BioTek) at a wavelength of 430 nm. The IC 50 value is the concentration of each derivative that inhibits cell proliferation by 50% under the experimental conditions and is the average from triplicate determinations that were reproducible and statistically significant. For calculating each IC 50 value, a linear proliferative inhibition was made based on at least five dosages for each compound.

Conclusions
By comparing with 5,7,20-O-hydnocarpin Ds, chalcone-type flavonolignans, and taxifolin derivatives, 5,7,20-O-trimethylsilybin and its derivatives possess the highest potency and selectivity towards AR-positive LNCaP prostate cancer cell line. Our data indicate that 5,7,20-O-trimethylsilybins are the most promising scaffold for AR modulation among the four and that the appropriate modification of the alcoholic hydroxyl group at C-3 and/or C-23 can retain or even enhance the antiproliferative potency in the ARpositive LNCaP cell model. The most promising 5,7,20-O-trimethylsilybins were further studied on the antiproliferative potency of their optically pure versions. To this end, our data show that (10R,11R) derivatives (silybin A series) are more potent than (10S,11S) derivatives (silybin B series) in the AR-positive LNCaP cell model. The detailed structureactivity relationships among four core structures and diastereomers were illustrated in Figure 10. Two (10R,11R) derivatives 8A and 41A were established as the optimal lead compounds because they can selectively inhibit AR-positive LNCaP cell proliferation with IC 50 value of 0.07 µM. The fact that 8A and 41A cannot suppress AR-null PC-3 and DU145 prostate cancer cell proliferation suggests their antiproliferative activity in the AR-positive LNCaP cell model may be associated with AR. Even though both LNCaP and 22Rv1 are AR-positive prostate cancer cell lines, they bear one critical difference. LNCaP possesses full-length AR that contains four domains: N-terminal domain, DNA-binding domain, C-terminal ligand binding domain (LBD), and the flexible hinge region. In contrast, 22Rv1 contains both full-length AR and LBD-truncated AR-V7. (10R,11R) Derivatives 8A and 41A cannot suppress 22Rv1 prostate cancer cell proliferation up to 10 µM concentration, suggesting that they are very likely binding to the ligand-binding domain on AR to exhibit antiproliferative activity in LNCaP cells. This assumption needs to be confirmed by further investigation on the optimal compounds 8A and 41A. Our findings warrant the further