Total Synthesis and Metabolic Stability of Hispidulin and Its d-Labelled Derivative

Abstract

Hispidulin is a naturally occurring flavone known to have various Central nervous system (CNS) activities. Proposed synthetic approaches to synthesizing hispidulin have proven unsatisfactory due to their low feasibility and poor overall yields. To solve these problems, this study developed a novel scheme for synthesizing hispidulin, which had an improved overall yield as well as more concise reaction steps compared to previous methods reported. Additionally, using the same synthetic strategy, d-labelled hispidulin was synthesized to investigate its metabolic stability against human liver microsome. This work may produce new chemical entities for enriching the library of hispidulin-derived compounds.

1. Introduction

Flavonoids, a group of polyphenolic compounds, occur ubiquitously in plants. These compounds have generally been categorized into several classes, including chalcones, flavonols, flavones, isoflavones, flavan-3-ols, neoflavones, flavanones and anthocyanins, based on their chemical structures [1]. In addition to the diverse structures, they also have many biological properties, such as antioxidative [2], anti-inflammatory [3], antimicrobial [4], anticonvulsant [5], antidepressant [6] and anticancer [7] activities. These facts indicate the important role of flavonoid compounds.

Our recent clinical case report described a patient with a refractory chronic motor tic disorder who dramatically improved after taking the leaf juice of a local herb, Clerodendrum inerme (L.) Gaertn (CI) [8]. The ethanol extract of CI leaf alleviated methamphetamine-induced hyperlocomotion (MIH) as a mouse model of motor tic [9]. Using a bioassay-guided purification procedure in this animal model, we further isolated the constituents from CI leaf extract to identify a flavone hispidulin (6-methoxy-4′,5,7-trihydroxyflavone) to be the main active ingredient (Figure 1) [9]. Hispidulin significantly attenuated MIH and even in amounts up to 100 mg/kg was incapable of affecting spontaneous locomotor activity or performance in mice [9]. Importantly, hispidulin had no hit on human ether-à-go-go-related gene (hERG) channels, an undesirable target in drug development [10].

Hispidulin is widely distributed in Asteraceae [11,12,13,14] and some of the Lamiaceae [15]. Further assays for screening towards 92 target proteins associated with Central nervous system (CNS) diseases indicated that hispidulin formed strong bonds with GABA_A receptors (IC₅₀ = 0.73−1.78 μM) and inhibited catecholamine-O-methyl-transferase (COMT) (IC₅₀ 1.32 μM) [10]. By acting as a positive allosteric modulator (PAM), it enhanced cerebellar α₆GABA_A receptor activity. Studies revealed some C6-substituted flavones are identified to be PAMs of GABA_A receptors [16,17]. The results of a structure–activity relationship (SAR) study indicated the C6 substituent in flavones greatly contributes to activity [10].

Because of its interesting biological activity, several research groups have developed synthetic approaches to hispidulin [18,19,20,21,22,23]. For example, Shen and coworkers developed a strategy for semisynthesis of hispidulin in seven reaction steps by using a naturally occurring scutellarin (Scheme 1) [22]. Although this method is concise and has an overall yield as high as 10.7% [22], the researchers showed that, upon the large-scale synthesis of compound 1, the protection of the catechol moiety of scutellarein using dichlorodiphenylmethane at 175 °C failed [20,23]. A seven-step synthesis route developed by Lin and coworkers solved this problem but reduced the overall yield to 7.1% (Scheme 1) [20]. Zhang and coworkers then developed a scheme that only required four reaction steps (Scheme 1). Nonselective MOM (methoxymethyl) protection of scutellarein caused the overall yield of the synthesis of hispidulin to be decreased (6.3%) [23]. Despite the satisfactory overall yield of these strategies, the tedious purification procedure required to isolate scutellarin from plants limits their use for large-scale preparation of hispidulin.

Kavvadias and coworkers developed a method synthesizing hispidulin from 2,4,6-trihydroxyacetophenone compound 6 in nine reaction steps (Scheme 2). However, the overall yield of this method is very limited (1.1%) [18]. We recently developed a feasible and reproducible approach for synthesizing hispidulin (Scheme 3) [21]. This method slightly improved the overall yield due to the low yield of Friedel–Crafts acetylation of compound 8 as well as unsatisfactory regioselective MOM protection of compound 9. These facts motivated us to investigate an efficient and high-yield route to synthesize hispidulin.

Deuterium is a stable isotope of hydrogen. Because deuterium has a stronger chemical bond with carbon than hydrogen, deuterium-labelled compounds can affect the absorption, distribution, metabolism and toxicology of their counterpart compound [24,25]. The Food and Drug Administration (FDA) recently approved the first deuterated drug, deutetrabenazine, for treating involuntary writhing movements or chorea in Huntington’s disease [26]. Deuterium incorporation of tetrabenazine markedly increased its half-life and area under the curve (AUC) in plasma [27]. Study showed that the pig caecal microflora metabolized hispidulin to scutellarein due to O-demethylation at the C6 position [28]. This result suggested that replacement of OCH₃ of this position of hispidulin using OCD₃ may reduce its metabolic liability. Therefore, this study replaced C6-OCH₃ in hispidulin with C6-OD₃ to investigate whether such modification allows enhancement of the metabolic stability. Notably, introducing deuterium into hispidulin leads to a new chemical entity (NCE) that meets the criteria for a 505(b) (2) patent application [27].

The new hispidulin synthesis scheme developed in this study is more feasible compared to all methods previously reported. In particular, it had the highest overall yield, which may help to synthesize C6-OCH₃-containing hispidulin derivatives. The same strategy can also be used to synthesize d-labelled hispidulin. The microsome stability of hispidulin and its deuterium counterpart were compared.

2. Results and Discussion

2.1. Retrosynthetic Analysis of Hispidulin

Figure 2 shows the retrosynthetic analysis to successfully synthesize hispidulin. Hispidulin can be produced from compound I via debenzylation and oxidative cyclization. Compound I is derived from compound II and commercially available compound III, which are used to conduct Claisen–Schmidt condensation and deprotection of MOM (methoxymethoxy). Compound II in turn can be prepared from compound IV via methylation. Compound IV is considered to be the key intermediate that possesses two different protecting groups as well as acetyl and hydroxy moieties. Selective Bayer–Villiger reaction of compound V provided compound IV. Compound V is prepared from compound VI via Stille coupling. Benzylation and regioselective iodination of compound VII gives compound VI. Starting from commercially available compound 7, Compound VII is synthesized via selective MOM protection.

2.2. Synthesis of Hispidulin

Scheme 4 shows a synthetic approach originally developed for hispidulin. Using compound 6 as the starting material, selective MOM protection towards its C4- and C6-OH gave compound 11. First, N-bromosuccinimide (NBS) was used for bromination of C3 of compound 11; however, only undesirable C3 and C5 dibromination occurred. Alternatively, compound 11, when reacted with I₂ in the presence of Lewis acid CF₃CO₂Ag, provided selectively iodinated product 12 in the high yield (90%). The experimental result resulted from the difference of steric hindrance between C3 and C5 upon using the more bulky reagent I₂. Benzylation of compound 12 generated compound 13. Next, several reported oxidation methods were used [29,30,31] in attempts to convert the iodide group of compound 13 into phenol. These oxidation catalysts included CuI, Cu and Pd, as shown in

Table 1. Attempts towards the oxidation of aryl halide 13.

Entry	Catalyst	Ligand	Base	Solvent	T (°C)	Yield (%)
1	CuI	1,10-Phenanthroline	KOH	DMSO–H2O	100	trace
2	Cu	-	NaOH	H2O	100	trace
3	Cu	-	NaOH	DMSO–H2O	100	trace
4	Pd2dba3	tBuXPhos	KOH	Dioxane–H2O	100	27
5	Pd2dba3	tBuXPhos	KOH	Dioxane–H2O	70	14
6	Pd2dba3	tBuXPhos	KOH	Dioxane–H2O	100 1	26
7	Pd2dba3	tBuXPhos	KOH	Dioxane–H2O	100 2	23
8	Pd2dba3	tBuXPhos	KOH	DMF	100 1	trace

¹ Microwave reaction: 100 W, 20 min; ² microwave reaction: 120 W, 20 min.

. Reaction using Pd₂dba₃ coupled with the ligand 2-di-tert-butylphosphino-2′,4′,6′-triisopropylbiphenyl (tBuXPhos) gave compound 14, but its yield was poor (27%). Despite the use of microwave irradiation, the yields were incapable of being improved. Instead, Claisen–Schmidt condensation of compound 13, reacted with 4-(benzyloxy)benzaldehyde prior to selective deprotection of MOM group in the presence of HCl, gave chalcone 15. The selective deprotection reaction can be manipulated by decreasing the reaction time to avoid further MOM deprotection of C4. Oxidative cyclization of compound 15 in the presence of catalytic I₂ provided flavone 16. Notably, excess I₂ caused an undesirable retro-aldol reaction that gave C4- and C6-deprotected compound 13. Treatment of catalytic Pd(PPh₃)₄ and tributyl(1-ethoxyvinyl)tin converted compound 16 to compound 17 through the Stille reaction. Baeyer–Villiger oxidation followed by basic hydrolysis of compound 17 by using H₂O₂ or m-chloroperoxybenzoic acid (MCPBA) [32] failed to give expected product 18.

Thus, we developed a new synthesis (Scheme 5) for hispidulin using the retrosynthetic analysis as shown in Figure 2. First, 2,4,6-trihydroxybenzaldehyde 7 reacted with MOMCl gave the bis(methoxymethoxy)-protected compound 20. Regioselective iodination of compound 20 prior to reaction with BnBr produced compound 21. The chemical structure of compound 21 was confirmed by rotating frame nuclear Overhauser effect spectroscopy (ROESY) spectrum. Figure 3 shows the key correlation of H-5 (δ_H 6.84) to H-2′″ (δ_H 3.52), H-2″″ (δ_H 3.54), H-1′″ (δ_H 5.29) and H-1″″ (δ_H 5.32); and H-1′ (δ_H 10.30) to H-2″″ (δ_H 3.54), H-1″ (δ_H 4.99) and H-1″″ (δ_H 5.32) (Supplementary Materials). Stille coupling of compound 21 produced compound 22.

Table 2. Optimization of reaction condition for Stille coupling of compound 21.

Entry	Catalyst	Solvent	Yield (%)	Reaction Time (h)
1	Pd(PPh3)4	Dioxane	68	30
2	Pd(PPh3)4	Toluene	70	24
3	PdCl2(PPh3)2	Dioxane	73	20
4	PdCl2(PPh3)2	Toluene	83	10
5	Pd(dppf)Cl2	Dioxane	34	78
6	Pd(dppf)Cl2	Toluene	53	43

shows how tributyl(1-ethoxyvinyl)tin was used to optimize Stille coupling. First, catalyst Pd(PPh₃)₄ was used in dioxane at 100 °C. The reaction had a satisfactory yield (68%), but the reaction time was up to 30 h. Replacement of the solvent by toluene led to the reaction time decreasing to 24 h. Further experiments using palladium catalysts such as PdCl₂(PPh₃)₂ and Pd(dppf)Cl₂ in dioxane or toluene showed that PdCl₂(PPh₃)₂ significantly improved the yield and decreased the reaction time. In particular, PdCl₂(PPh₃)₂ coupled with toluene not only gave the highest yield, but also had the lowest reaction time. Baeyer–Villiger oxidation and basic hydrolysis of compound 22 afforded compound 14. Methylation of compound 14 using CH₃I gave compound 23a. Claisen–Schmidt condensation of compound 23a with 4-(benzyloxy)benzaldehyde prior to MOM deprotection produced chalcone 24a. Cyclization of compound 24a in the presence of catalytic I₂ generated flavone 25a. Although debenzylation of compound 25a catalyzed by 10% Pd–C did not yield hispidulin, a reaction using BCl₃ at −80 °C successfully converted compound 25a into hispidulin. The chemical structure of hispidulin was identified as judged by 2D-NMR analyses. Figure 4 shows the key correlation in the ROESY spectrum of hispidulin that 5-OH (δ_H 13.07) was correlated to 6-OMe′″ (δ_H 3.74) and H-8 (δ_H 6.59) was correlated to H-2′ and H-6′ (δ_H 7.92). Additionally, the HMBC spectrum showed that 5-OH (δ_H 13.07) correlated to C-5 (δ_C 152.8), C-6 (δ_C 131.4), C-7 (δ_C 157.3), C-9 (δ_C 152.4) and C-10 (δ_C 104.1); H-8 (δ_H 13.07) correlated to C-6 (δ_C 131.4), C-7 (δ_C 157.3), C-9 (δ_C 152.4) and C-10 (δ_C 104.1); 6-OMe-H (δ_H 3.74) correlated to C-6 (δ_C 131.4) (Supplementary Materials). The ¹H- and ¹³C-NMR data of synthesized hispidulin were similar to those of hispidulin previously isolated (Supplementary Materials) [33].

2.3. Synthesis of d-Hispidulin

The synthesis of d-hispidulin is described in Scheme 5. Methylation of compound 14 using CD₃I gave compound 23b. The same method to hispidulin was used to synthesize d-labelled hispidulin starting from compound 23b. Due to the absence of a proton signal of the CD₃O group in the ¹H-NMR spectra of the d-containing intermediate compounds 23b, 24b, 25b and d-hispidulin, these compound structures were identified depending on the ¹³C-NMR spectra without ¹H decoupling and the mass technique. The ¹³C-NMR spectra revealed a characteristic multiplet splitting pattern of the ¹³C signal for the CD₃O group in compounds 23b, 24b, 25b and d-hispidulin. The mass spectra also supported chemical structures of these d-labelled compounds. All synthesized compounds had an estimated purity of at least 98% as determined by HPLC analysis (Supplementary Materials).

2.4. Comparison of Hispidulin Synthesis Methods

Strategies for synthesizing hispidulin can be classified as semisynthesis and total synthesis strategies. The starting material used in most semisynthetic methods is scutellarin, which is a natural product.

Table 3. Comparison of hispidulin synthesis methods.

Research Group	Synthesis Route	Reaction Steps	Overall Yield (%)
Shen and coworkers	Semisynthesis	Seven	10.7
Lin and coworkers	Semisynthesis	Seven	7.1
Zhang and coworkers	Semisynthesis	Four	6.3
Kavvadias and coworkers	Total synthesis	Nine	1.1
Chao and coworkers	Total synthesis	Ten	1.6
This study	Total synthesis	Eight	26.9

shows that the semisynthesis routes had fewer reaction steps compared to the total synthesis methods; however, they need tedious isolation procedures for scutellarin, which limits the scale for further chemical modification. Furthermore, their overall yields are only 6.3–10.7% [20,22,23]. For total synthesis, Kavvadias and coworkers developed a nine-step synthesis approach. The starting material used in this method is commercially available 2,4,6-trihydroxyacetophenone. Although this method solves the issue of the source for starting material, its drawback is low overall yield [18]. We previously developed a feasible route of hispidulin synthesis that has an overall yield comparable to that of the method developed by Kavvadias and coworkers [21]. This present study further made the reaction steps more concise. In particular, the synthetic scheme showed the highest overall yield of all approaches to synthesize hispidulin.

2.5. Human Liver Microsome Stability

Metabolic stability is associated with susceptibility of compounds to biotransformation. Metabolic half-life (t_1/2) and intrinsic clearance (CL_int) was compared between hispidulin and d-hispidulin by testing these synthesized compounds in a human liver microsome stability assay. The study revealed that FDA-approved deuterated agent deutetrabenazine had a t_1/2 (8.6 h) superior to tetrabenazine (4.8 h). In addition to t_1/2, the AUC of deutetrabenazine (542 ng·hr/mg) was also higher than that of its counterpart compound (261 ng·hr/mg) [27]. In contrast, the experimental results indicated hispidulin and d-hispidulin had no significant difference in t_1/2 and CL_int (

Table 4. Human liver microsome stability of hispidulin and d-hispidulin.

Compound	Half-Life (min)	CLint 1 (mL/min/mg Protein)
Hispidulin	46	0.0298
d-Hispidulin	43	0.0325
Testosterone	19	0.0727

¹ Intrinsic clearance (CL_int) was calculated based on CL_int = k/P, where k is the elimination rate constant and P is the protein concentration in the incubation.

), which suggested that the C6-OMe of hispidulin is resistant to be modified by the human liver microsome. The metabolic site of hispidulin in the human liver microsome is worthy of further study.

The experiments showed that this method of synthesizing hispidulin and its d-labelled derivative is highly feasible. Specifically, it increases overall yield compared to previous methods. The t_1/2 and intrinsic clearance of these two compounds were identified as well. Overall, this synthetic route can be applied to produce 6-OMe-containing hispidulin derivatives as new chemical entities for investigating their biological activities.

3. Experimental Section

3.1. General Information

The NMR spectra (¹H- and ¹³C-NMR, ROESY, HSQC and HMBC) were obtained with a Bruker AV500 using standard pulse programs. The MS data were recorded with a Finnigan Mat TSQ-7000 mass spectrometer (HR–ESI–MS) (Thermo, Ringoes, NJ, USA). The HPLC was performed on a C₁₈ column (150 mm × 4.6 mm, Ascentis) by using an L-2130 pump (Hitachi, Ibaraki, Japan) and a UV/vis L-2420 detector (Hitachi, Ibaraki, Japan). The column chromatography was performed on silica gel (70-230 mesh, Merck, Darmstadt, Germany). All TLC analyses were performed on silica gel plates (KG60-F254, Merck, Darmstadt, Germany). Reagents and materials were used without further purification and chemicals were purchased from ACROS (Geel, Belgium). Dry dichloromethane was distilled from CaH₂ under nitrogen atmosphere. MOMCl was acquired from TCI (Tokyo, Japan), and 2,4,6-trihydroxybenzaldehyde was purchased from Alfa Aesar (Heydham, UK).

3.2. Chemistry

2-Hydroxy-4,6-bis(methoxymethoxy)benzaldehyde (20). To a solution of compound 7 (10 g, 64.9 mmol) in CH₂Cl₂ (200 mL) was added DIPEA (28.3 mL, 162.2 mmol). The resulting mixture was stirred for 10 mins in an ice-bath under N₂. MOMCl (10.8 mL, 142.7 mmol) was added dropwise to the reaction mixture by addition funnel. The reaction mixture was warmed to room temperature (rt) and stirred for 3 h. The mixture was concentrated in vacuo. The residue was diluted with EtOAc (100 mL) and washed with distilled H₂O (3 × 80 mL). The organic layer was dried over Na₂SO₄, filtered and removed in vacuo. The residue was purified by silica gel chromatography (EtOAc:n-hexane = 1:9) to give compound 20 (12.3 g, 80%), a light-yellow microcrystalline powder; ¹H-NMR (CDCl₃, 300 MHz) δ 12.29 (1H, s), 10.16 (1H, s), 6.25 (1H, d, J = 2.1 Hz), 6.23 (1H, d, J = 2.1 Hz), 5.24 (2H, s), 5.18 (2H, s), 3.51 (3H, s), 3.47 (3H, s); ¹³C-NMR (CDCl₃, 125 MHz) δ 192.1, 165.6, 165.5, 161.2, 106.9, 96.6, 94.6, 94.1, 94.0, 56.6, 56.5; HR–ESI–MS m/z 243.0858 [M + H]⁺ (calcd. for C₁₁H₁₅O₆, 243.0863).

2-Benzyloxy-3-iodo-4,6-bis(methoxymethoxy)benzaldehyde (21). To a mixture of compound 20 (7.2 g, 29.6 mmol) and CF₃CO₂Ag (7.8 g, 35.5 mmol) in CH₂Cl₂ (150 mL) at 0 °C was added I₂ (8.3 g, 32.5 mmol) in CH₂Cl₂ (200 mL) dropwise by an addition funnel over 2 h. The reaction mixture was warmed to rt and stirred for 3 h. Then, saturated Na₂S₂O_3(aq) (50 mL) was added and the mixture was washed with distilled H₂O (3 × 100 mL). The organic layer was dried over Na₂SO₄, filtered and removed in vacuo. The residue was dissolved in DMF (100 mL) and then K₂CO₃ (7.2 g, 51.8 mmol) was added. Benzyl bromide (460 μL, 3.9 mmol) was added dropwise to the mixture by an addition funnel at 0 °C. The resulting solution was warmed to rt and stirred for 3 h. The mixture was diluted with EtOAc (300 mL) and washed with distilled H₂O (3 × 00 mL). The organic layer was dried over Na₂SO₄, filtered and removed in vacuo. The residue was purified by silica gel chromatography (EtOAc:n-hexane = 1:5) to give compound 21 (12.1 g, 89%), a white microcrystalline powder; ¹H-NMR (CDCl₃, 300 MHz) δ 10.30 (1H, s), 7.66 (2H, dd, J = 1.8, 8.1 Hz), 7.45–7.36 (3H, m), 6.84 (1H, s), 5.32 (2H, s), 5.29 (2H, s), 4.99 (2H, s), 3.54 (3H, s), 3.52 (3H, s); ¹³C-NMR (CDCl₃, 125 MHz) δ 187.2, 161.9, 161.8, 161.6, 136.1, 129.0, 128.5, 128.4, 115.5, 98.2, 95.2, 94.9, 78.5, 56.7; HR–ESI–MS m/z 459.0294 [M + H]⁺ (calcd. for C₁₈H₂₀O₆I, 459.0299).

3-Acetyl-2-benzyloxy-4,6-bis(methoxymethoxy)benzaldehyde (22). To a mixture of compound 21 (5 g, 10.9 mmol) and PdCl₂(PPh₃)₂ (766 mg, 1.1 mmol) in toluene (200 mL) was added tributyl(1-ethoxyvinyl)tin (5.5 mL, 16.4 mmol). The resulting solution was heated to 100 °C and stirred for 12 h. After cooling to rt, the reaction mixture was acidified with 1 M HCl (50 mL) and stirred for 30 min. The mixture was diluted with EtOAc (200 mL) and washed with distilled H₂O (3 × 100 mL). The organic layer was dried over Na₂SO₄, filtered and concentrated in vacuo. The residue was purified by silica gel chromatography (EtOAc:n-hexane = 1:4) to give compound 22 (3.5 g, 83%), a yellow microcrystalline powder; ¹H-NMR (CDCl₃, 300 MHz) δ 10.37 (1H, s), 7.48 (2H, dd, J = 1.8, 7.8 Hz), 7.41–7.33 (3H, m), 6.80 (1H, s), 5.29 (2H, s), 5.24 (2H, s), 4.96 (2H, s), 3.53 (3H, s), 3.49 (3H, s), 2.43 (3H, s); ¹³C-NMR (CDCl₃, 125 MHz) δ 200.9, 187.3, 162.1, 159.3, 158.4, 136.1, 129.0, 128.8, 128.6, 128.5, 128.4, 121.8, 114.3, 97.5, 95.0, 94.5, 79.1, 56.8, 32.5; HR–ESI–MS m/z 375.1432 [M + H]⁺ (calcd. for C₂₀H₂₃O₇, 375.1438).

2-Benzyloxy-3-hydroxy-4,6-bis(methoxymethoxy)acetophenone (14). To a solution of 70% MCPBA (6.8 g, 27.6 mmol) in dry CH₂Cl₂ (100 mL) at 0 °C was added compound 22 (3.4g, 9.2 mmol) in CH₂Cl₂ (100 mL) dropwise by an addition funnel over 1 h. The resulting solution was warmed to rt and stirred for 8 h. Then, saturated Na₂S₂O_3(aq) (30 mL) was added and the reaction mixture was washed with distilled H₂O (3 × 100 mL). The organic layer was dried over Na₂SO₄, filtered and concentrated in vacuo. The residue was dissolved in MeOH (80 mL) and 10% NaOH_(aq) (60 mL) was added. The resulting solution was stirred at rt for 1.5 h. The mixture was diluted with EtOAc (250 mL) and washed with distilled H₂O (3 × 100 mL). The organic layer was dried over Na₂SO₄, filtered and concentrated in vacuo. The residue was purified by silica gel chromatography (EtOAc:n-hexane = 1:4) to give compound 14 (2.3 g, 68%), a light-yellow oil; ¹H-NMR (CDCl₃, 300 MHz) δ 7.44 (2H, dd, J = 1.8, 8.1 Hz), 7.40–7.32 (3H, m), 6.76 (1H, s), 5.61 (1H, s), 5.21 (2H, s), 5.08, (2H, s), 5.06 (2H, s), 3.53 (3H, s), 3.46 (3H, s), 2.45 (3H, s); ¹³C-NMR (CDCl₃, 125 MHz) δ 201.3, 146.6, 146.2, 143.2, 136.9, 134.9, 128.5, 128.3, 122.1, 100.4, 96.0, 95.8, 76.3, 56.6, 56.3, 32.6; HR–ESI–MS m/z 363.1431 [M + H]⁺ (calcd. for C₁₉H₂₃O₇, 363.1438).

2-Benzyloxy-3-methoxy-4,6-bis(methoxymethoxy)acetophenone (23a). To a mixture of compound 14 (587 mg, 1.6 mmol) and K₂CO₃ (1.1 g, 8.1 mmol) in acetone (20 mL) was added CH₃I (0.5 mL, 8.1 mmol). The resulting solution was heated to 56 °C and stirred for 5 h. The reaction mixture was concentrated in vacuo, diluted with EtOAc (100 mL) and washed with distilled H₂O (3 × 50 mL). The organic layer was dried over Na₂SO₄, filtered and concentrated in vacuo. The residue was purified by silica gel chromatography (EtOAc:n-hexane = 1:5) to give compound 23a (558 mg, 92%), a white microcrystalline powder; ¹H-NMR (CDCl₃, 300 MHz) δ 7.42 (2H, dd, J = 2.0, 8.3 Hz), 7.40–7.32 (3H, m), 6.76 (1H, s), 5.23 (2H, s), 5.11 (2H, s), 5.07 (2H, s), 3.85 (3H, s), 3.53 (3H, s), 3.46 (3H, s), 2.38 (3H, s); ¹³C-NMR (CDCl₃, 125 MHz) δ 200.6, 151.8, 149.4, 149.2, 137.9, 136.7, 128.1, 128.0, 127.7, 121.4, 99.6, 95.0, 94.9, 76.0, 60.8, 56.0, 55.9, 32.1; HR–ESI–MS m/z 377.1587 [M + H]⁺ (calcd. for C₂₀H₂₅O₇, 377.1595).

2-Benzyloxy-3-[²H₃]-methoxy-4,6-bis(methoxymethoxy)acetophenone (23b). Following the procedure as described for compound 23a, reaction of compound 14 (1.8 g, 4.8 mmol), K₂CO₃ (3.3 g, 24.2 mmol), CD₃I (1.5 mL, 24.2 mmol) in acetone (40 mL) gave compound 23b (1.7 g, 93%), a white microcrystalline powder; ¹H-NMR (CDCl₃, 300 MHz) δ 7.42 (2H, dd, J = 2.1, 8.4 Hz), 7.39–7.31 (3H, m), 6.76 (1H, s), 5.23 (2H, s), 5.11 (2H, s), 5.07 (2H, s), 3.53 (3H, s), 3.46 (3H, s), 2.38 (3H, s); ¹³C-NMR (CDCl₃, 125 MHz) δ 200.6, 151.8, 149.4, 149.2, 137.8, 136.7, 128.1, 128.0, 127.7, 121.4, 99.6, 95.0, 94.9, 76.0, 56.0, 55.9, 32.1; HR–ESI–MS m/z 380.1775 [M + H]⁺ (calcd. for C₂₀H₂₂D₃O₇, 380.1783).

(E)-1-(2-Benzyloxy-4,6-dihydroxy-3-methoxyphenyl)-3-(4-benzyloxyphenyl)prop-2-en-1-one (24a). To a solution of 23a (545 mg, 1.45 mmol) and 4-benzyloxybenzaldyde (615 mg, 2.9 mmol) in EtOH (20 mL) was added KOH (813 mg, 14.5 mmol) in EtOH–H₂O (3 mL:3 mL) dropwise by addition funnel at 0 °C over 30 min. The resulting solution was warmed to rt and stirred for 24 h. The mixture was diluted with distilled H₂O (50 mL) and washed with EtOAc (3 × 50 mL). The organic layer was dried over Na₂SO₄, filtered and concentrated in vacuo. MeOH–THF (14.5 mL:14.5 mL) and 12 M HCl (0.7 mL) was added to the residue at 0 °C. The resulting solution was warmed to rt and stirred for 8 h. The reaction mixture was diluted with distilled H₂O (50 mL) and washed with EtOAc (3 × 50 mL). The organic layer was dried over Na₂SO₄, filtered and concentrated in vacuo. The residue was purified by silica gel chromatography (EtOAc:n-hexane = 1:4) to give 24a (641 mg, 92%), a microcrystalline powder; ¹H-NMR (DMSO-d₆, 300 MHz) δ 12.92 (1H, s), 10.52 (1H, s), 7.59 (2H, s), 7.48–7.45 (2H, m), 7.44–7.37 (5H, m), 7.32–7.29 (5H, m), 6.94 (2H, d, J = 8.8 Hz), 6.21 (1H, s), 5.16 (2H, s), 5.06 (2H, s), 3.77 (3H, s); ¹³C-NMR (DMSO-d₆, 125 MHz) δ 192.0, 160.2, 159.6, 157.7, 153.1, 143.0, 136.7, 136.6, 134.4, 130.3, 128.5, 128.4, 128.3, 128.2, 128.0, 127.8, 127.4, 124.6, 115.2, 109.0, 99.7, 75.7, 69.4, 60.7; HR–ESI–MS m/z 483.1795 [M + H]⁺ (calcd. for C₃₀H₂₇O₆, 483.1802).

(E)-1-(2-Benzyloxy-4,6-dihydroxy-3-[²H₃]methoxyphenyl)-3-(4-benzyloxyphenyl)prop-2-en-1-one (24b). According to the procedure as described for compound 24a, reaction of compound 23b (1.0 g, 2.6 mmol), 4-benzyloxybenzaldyde (1.1 g, 5.3 mmol) and KOH (1.5 g, 26.4 mmol) in EtOH (60 mL) followed by treatment of 12 M HCl (1.3 mL) and MeOH–THF (26.4 mL:26.4 mL) gave compound 24b (1.0 g, 81%), a yellow microcrystalline powder; ¹H-NMR (DMSO-d₆, 300 MHz) δ 12.95 (1H, s), 10.54 (1H, s), 7.59 (2H, s), 7.48–7.45 (2H, m), 7.44–7.36 (5H, m), 7.33–7.29 (5H, m), 6.94 (2H, d, J = 8.8 Hz), 6.22 (1H, s), 5.16 (2H, s), 5.06 (2H, s); ¹³C-NMR (DMSO-d₆, 125 MHz) δ 192.0, 160.2, 159.7, 157.7, 153.2, 143.0, 136.7, 136.6, 134.4, 130.3, 128.5, 128.4, 128.3, 128.2, 128.0, 127.8, 127.4, 124.6, 115.2, 109.0, 99.7, 75.7, 69.4; HR–ESI–MS m/z 486.1983 [M + H]⁺ (calcd. for C₃₀H₂₄D₃O₆, 486.1990).

4′-Benzyloxy-6-methoxy-5-benzyloxy-7-hydroxyflavone (25a). To a solution of compound 24a (598 mg, 1.2 mmol) in dry DMSO (100 mL) was added I₂ (32 mg, 0.1 mmol) in DMSO (3 mL) dropwise by syringe. The resulting solution was heated to 120 °C, and stirred for 2 h. After cooling to rt, saturated Na₂S₂O_3(aq) (10 mL) was added to the reaction mixture. The mixture was diluted with EtOAc (100 mL) and wash with distilled H₂O (3 × 50 mL). The organic layer was dried over Na₂SO₄, filtered and concentrated in vacuo. The residue was purified by silica gel chromatography (EtOAc:n-hexane = 1:3) to give 25a (555 mg, 93%), a yellow microcrystalline powder; ¹H-NMR (DMSO-d₆, 300 MHz) δ 10.76 (1H, s), 7.98 (2H, d, J = 8.9 Hz), 7.61 (2H, d, J = 7.0 Hz), 7.48 (2H, d, J = 7.0 Hz), 7.44–7.34 (6H, m), 7.18 (2H, d, J = 8.9 Hz), 6.91 (1H, s), 6.67 (1H, s), 5.22 (2H, s), 5.00 (2H, s), 3.75 (3H, s); ¹³C-NMR (DMSO-d₆, 125 MHz) δ 175.8, 160.9, 160.1, 156.1, 153.8, 150.7, 139.6, 137.6, 136.6, 128.5, 128.3, 128.1, 128.0, 127.8, 123.4, 115.3, 111.4, 106.0, 100.1, 75.6, 69.5, 60.9; HR–ESI–MS m/z 481.1637 [M + H]⁺ (calcd. for C₃₀H₂₅O₆, 481.1646).

4′-Benzyloxy-6-[²H₃]methoxy-5-benzyloxy-7-hydroxyflavone (25b). Following the procedure as described for compound 25a, reaction of compound 24b (849 mg, 1.7 mmol) with I₂ (44 mg, 0.2 mmol) in DMSO (120 mL) gave compound 25b (664 mg, 79%), a yellow microcrystalline powder; ¹H-NMR (DMSO-d₆, 300 MHz) δ 10.75 (1H, s), 7.98 (2H, d, J = 8.9 Hz), 7.61 (2H, d, J = 6.8 Hz), 7.48 (2H, dd, J = 1.8, 8.4 Hz), 7.44–7.34 (6H, m), 7.18 (2H, d, J = 8.9 Hz), 6.91 (1H, s), 6.67 (1H, s), 5.22 (2H, s), 5.01 (2H, s); ¹³C-NMR (DMSO-d₆, 125 MHz) δ 175.8, 160.9, 160.1, 156.1, 153.7, 150.7, 139.5, 137.6, 136.6, 128.5, 128.3, 128.1, 128.0, 127.8, 123.4, 115.3, 111.4, 106.0, 100.1, 75.6, 69.5; HR–ESI–MS m/z 484.1827 [M + H]⁺ (calcd. for C₃₀H₂₂D₃O₆, 484.1834).

Hispidulin. To a solution of compound 25a (301 mg, 0.6 mmol) in dry CH₂Cl₂ (40 mL) was added 1 M BCl₃ (2.5 mL, 2.5 mmol) in dry CH₂Cl₂ (5 mL) dropwise by syringe at −78 °C over 20 min. The resulting solution was stirred for 1 h. The reaction mixture was diluted with distilled H₂O (50 mL) and washed with EtOAc (3 × 50 mL). The organic layer was dried over Na₂SO₄, filtered and concentrated in vacuo. The residue was purified by silica gel chromatography (EtOAc:n-hexane = 1:2) to give hispidulin (160 mg, 85%), a yellow microcrystalline powder; ¹H-NMR (DMSO-d₆, 500 MHz) δ 13.07 (1H, s, 5-OH), 10.73 (1H, s, 7-OH), 10.38 (1H, s, 4′-OH), 7.92 (2H, d, J = 8.9 Hz, H-2′, H-6′), 6.92 (2H, d, J = 8.9 Hz, H-3′, H-5′), 6.77 (1H, s, H-3), 6.59 (1H, s, H-8), 3.74 (3H, s, 6-OMe); ¹³C-NMR (DMSO-d₆, 125 MHz) δ 182.2 (C-4), 163.9 (C-2), 161.2 (C-4′), 157.3 (C-7), 152.8 (C-5), 152.4 (C-9), 131.4 (C-6), 128.5 (C-2’, C-6′), 121.2 (C-1), 116.0 (C-3′, C-5′), 104.1 (C-10), 102.4 (C-3), 94.3 (C-8), 60.0 (6-OMe); HR–ESI–MS m/z 301.0702 [M + H]⁺ (calcd. for C₁₆H₁₃O₆, 301.0707).

d-Hispidulin. Following the procedure as described for hispidulin, reaction of compound 25b (536 mg, 1.1 mmol) in CH₂Cl₂ (75 mL) with 1 M BCl₃ (4.4 mL, 4.4 mmol) in CH₂Cl₂ (8.8 mL) gave d-hispidulin (268 mg, 80%), a yellow microcrystalline powder; ¹H-NMR (DMSO-d₆, 500 MHz) δ 13.07 (1H, s, 5-OH), 10.70 (1H, s, 7-OH), 10.36 (1H, s, 4′-OH), 7.91 (2H, d, J = 8.9 Hz, H-2′, H-6′), 6.92 (2H, d, J = 8.9 Hz, H-3′, H-5′), 6.76 (1H, s, H-3), 6.58 (1H, s, H-8); ¹³C-NMR (DMSO-d₆, 125 MHz) δ 182.1 (C-4), 163.8 (C-2), 161.2 (C-4’), 157.2 (C-7), 152.8 (C-5), 152.4 (C-9), 131.3 (C-6), 128.5 (C-2′, C-6′), 121.2 (C-1), 116.0 (C-3′, C-5′), 104.1 (C-10), 102.4 (C-3), 94.2 (C-8); HR–ESI–MS m/z 304.0888 [M + H]⁺ (calcd. for C₁₆H₁₀D₃O₆, 304.0895).

3.3. Human Liver Microsome Stability Assay

Mixed-gender human liver microsomes (Lot # 1210347) were purchased from XenoTech. The reaction mixture minus NADPH was prepared as described below. The test compounds were added into the reaction mixture at a final concentration of 1 μM. A separate reaction with the control compound, testosterone, was run simultaneously with the reactions with the test compounds. An aliquot of the reaction mixture (without cofactor) was equilibrated in a shaking water bath at 37 °C for 3 min. After addition of cofactor to initiate the reaction, the mixture was incubated in a shaking water bath at 37 °C. Aliquots (100 μL) were withdrawn at 0, 10, 20, 30 and 60 min for the test compounds and testosterone. The reaction was terminated by immediately combining the tested compounds and testosterone samples with 400 μL of ice-cold 50/50 acetonitrile (ACN)/H₂O containing 0.1% formic acid and internal standard. The samples were then mixed and centrifuged to precipitate proteins. All samples were assayed by LC-MS/MS using electrospray ionization. The peak area response ratio (PARR) to internal standard was compared to the PARR at time 0 to determine the percent remaining at each time point. The values for half-life (t_1/2) and intrinsic clearance (CL_int) of the tested compounds were determined by Absorption System Corp. Half-life calculated using GraphPad software was fitted to a single-phase exponential decay equation.