**1. Introduction**

Flavonoids are polyphenolic secondary metabolites that are common in the plant kingdom and are ingested by humans in their food [1]. Flavonoids are grouped into various classes based on structure. These classes are: anthocyanidins, chalcones, flavanones, flavones, flavonols, isoflavonoids, and biflavonoids [2]. Many pharmacological benefits have been ascribed to flavonoids, including antioxidant, anti-inflammatory, anti-cancer, antiviral, and hepatoprotective effects [3,4].

Having flavonoids in your diet may reduce the risk of atherosclerosis, cardiovascular disease, diabetes mellitus, osteoporosis, and certain cancers [4,5]. Because of flavonoids' benefits and wide distribution, their intake has risen steadily in recent years in the West and Asia. Daily intake of flavonoids has been estimated at 100 mg/day in the Asian population because of the high consumption of soy products [6,7]. On the other hand, daily intake of flavonoids has been estimated to be in the range of 20–50 mg/day in Western populations [8]. Further intake of flavonoids through dietary supplements and plant extracts with prescribed drugs is common. The vast body of literature describes the significant interactions between flavonoid herbs and therapeutic drugs.

Several flavonoids are substrates for cytochrome P450 (P450) and uridine 5-diphosphoglucuronosyl transferase (UGT) enzymes [2], suggesting that flavonoids could inhibit the activities of these enzymes. A number of studies have demonstrated that flavonoids are potent inhibitors of CYP1A2, CYP3A, and UGT1A1 in vitro [5,8]. For example, the flavone tangeretin competitively inhibits the activity of CYP1A2 with a *K*i value as low as 68 nM in human liver microsomes (HLMs) [9]. It also inhibits UGT1A1-mediated estradiol glucuronidation with an IC50 value of 1 μM [10]. The flavonols quercetin and kaempferol inhibit the metabolism of nifedipine and felodipine by CYP3A4 in HLMs at concentrations larger than 10 μM. [11]. Animal studies show that oral quercetin increases the bioavailability of oral doxorubicin [12]. These results can be attributed to the reduced first-pass metabolism of doxorubicin due to quercetin-induced inhibition of CYP3A and/or enhanced doxorubicin absorption in the gastrointestinal tract via quercetin-induced inhibition of P-glycoprotein (P-gp). Surya Sandeep et al. (2014) reported that naringenin significantly increases the bioavailability of orally administered felodipine, a P-glycoprotein and CYP3A4 substrate drug, in rats, through the inhibition of intestinal P-gp and CYP3A4 [13]. Alnaqeeb et al. (2019) reported that quercetin and guava leaf extracts in combination with warfarin exert a greater increase on warfarin's *C*max and International Normalized Ratio values than when used alone, indicating the inhibition of CYP2C8, 2C9 and 3A4, major warfarin-metabolizing enzymes [14]. Biflavonoids, formed by the covalent bond between two monoflavonoids, are a subclass of flavonoid [15]. They are secondary metabolites, but are limited to several species in plants such as *Ginkgo biloba*, *Selaginella* species, *Hypericum perforatum*, and *Garcinia kola* [16]. Befitting their status as flavonoids, they have anti-cancer, anti-microbial, antiviral, and anti-inflammatory properties [16]. In contrast to the extensive studies on drug interaction with flavonoids, data on the inhibitory effects of biflavonoids on P450 and UGT enzymes are rare, though biflavonoids are taken in the form of dietary supplements (e.g., *Ginkgo biloba* extract [17]). The inhibitory potential of amentoflavone, the major biflavonoid in *Cupressus funebris*, against P450 and UGT enzymes was only recently reported [18,19].

In this study, we evaluate the inhibitory effects of five biflavonoids—selamariscina A, amentoflavone, robustaflavone, cupressuflavone, and taiwaniaflavone (Figure 1)—on nine P450 enzymes using HLMs. We further investigate the ability of selamariscina A, which most strongly inhibited CYP2C8 and CYP2C9 activities, to inhibit six UGT isoforms. Furthermore, the inhibition mechanism and kinetic parameters (*K*i) were determined for selamariscina A and compared with those of montelukast, a well-known selective CYP2C8 inhibitor [20].

**Figure 1.** Chemical structures of biflavonoids from *Selaginella tamariscina*: selamariscina A (**A**), amentoflavone (**B**), robustaflavone (**C**), cupressuflavone (**D**) and taiwaniaflavone (**E**).

#### **2. Materials and Methods**

#### *2.1. Chemicals and Reagents*

We purchased acetaminophen, alamethicin, amodiaquine, bupropion, chenodeoxycholic acid, chlorzoxazone, dextromethorphan, estrone glucuronide, glucose-6-phosphate (G6P), glucose- 6-phosphate dehydrogenase (G6PDH), hydroxybupropion, magnesium chloride (MgCl2), N-acetylserotonin, β-nicotinamide adenine dinucleotide phosphate (NADP+), N-desethylamodiaquine, omeprazole, phenacetin, trifluoperazine, trimipramine, and uridine 5-diphosphoglucuronic acid (UDPGA) from Sigma-Aldrich (St. Louis, MO, USA). 4-Hydroxydiclofenac, 4-hydroxytolbutamide, 5-hydroxyrosiglitazone, coumarin, diclofenac, midazolam, montelukast, mycophenolic acid, rosiglitazone, and tolbutamide came from Toronto Research Chemicals (Toronto, ON, Canada). We obtained 1-hydroxymidazolam from Cayman Chemical (Ann Arbor, MI, USA), while 7-ethyl-10-hydroxycomptothecine (SN-38) was provided by Santa Cruz Biotechnology (Dallas, TX, USA). All solvents were LC–MS grade (Fisher Scientific, Pittsburgh, PA, USA). All the other reagents were of analytical or LC–MS grade and are commercially available. We purchased the pooled human liver microsomes (XTreme 200) from XenoTech (Lenexa, Kansas City, KS, USA). In this study, we used selamariscina A, amentoflavone, robustaflavone, cupressuflavone, and taiwaniaflavone identified from Selaginella tamariscina (Beauv.), which were collected at Yen Tu Mountain, Uong Bi town, Quang Nihn province, Vietnam. The information regarding the identification of their chemical structures was described in our previously published paper [21,22].

We isolated selamariscina A, amentoflavone, robustaflavone, cupressuflavone, and taiwaniaflavone from Selaginella tamariscina (Beauv.), which were collected at Yen Tu Mountain, Uong Bi town, Quang Nihn province, Vietnam. The five compounds were purified and examined by HPLC to ge<sup>t</sup> 95% purity. Their chemical structures were identified by analyzing their NMR data, which were in good agreemen<sup>t</sup> with those published in a previous report [21,22].

#### *2.2. Inhibitory E*ff*ect of Five Biflavonoids against Human Cytochrome P450 Activity*

The inhibitory potential of the five biflavonoids on the metabolism of nine P450 probe substrates was evaluated using previously developed methods with minor modifications [23,24]. Biflavonoids were dissolved in methanol. The final concentration of methanol in the incubation mixture was 1.0% (*v*/*v*). We used these P450 probe substrates: phenacetin for CYP1A2, coumarin for CYP2A6, bupropion for CYP2B6, amodiaquine for CYP2C8, diclofenac for CYP2C9, omeprazole for CYP2C19, dextromethorphan for CYP2D6, chlorzoxazone for CYP2E1 and midazolam for CYP3A (Table 1). The incubation mixtures containing pooled human liver microsomes (HLMs, XTreme 200, XenoTech), P450 probe substrates, and inhibitor (0~20 μM) were pre-incubated at 37 ◦C for 5 min. The concentration range of the inhibitor varied (0, 0.002, 0.005, 0.02, 0.05, and 0.2 μM for CYP2C8; 0, 0.02, 0.05, 0.2, 0.5, and 2 μM for CYP2C9; 0, 0.5, 2, 5, 10, and 20 μM for other P450 isoforms). After pre-incubation, a reduced nicotinamide adenine dinucleotide phosphate (NADPH) generation system containing 1 unit/ml G6PDH, 1.3 mM β- nicotinamide adenine dinucleotide phosphate (β- NADP+), 3.3 mM MgCl2, and 3.3 mM G6P was added to initiate a reaction, and further incubated for 10 min at 37 ◦C. The reaction was stopped by adding 50 μL of ice-cold acetonitrile containing 7 nM trimipramine (internal standard, IS). After centrifugation at 18,000 g (5 min, 4 ◦C), aliquots of supernatants were analyzed by LC–MS/MS (Shimadzu LCMS 8060 system, Shimadzu, Kyoto, Japan). All microsomal incubations were conducted in triplicate.

**Table 1.** Selected reaction monitoring (SRM) condition for the major metabolites of the nine cytochrome P450 probe substrates and internal standard (IS).


ESI: Electrospray ionization (ESI) interface to generate protonated ions [M+H]<sup>+</sup> or deprotonated ion [M−H]<sup>−</sup>.

#### *2.3. Kinetic Characterization of Five Biflavonoids on CYP2C8 in Human Liver Microsomes*

To determine the inhibition mechanism and constants (Ki values) of the five biflavonoids against CYP2C8 activity, different concentrations of biflavonoids (0, 0.002, 0.005, 0.02, 0.05, and 0.2 μM for selamariscina A; 0, 0.05, 0.02, 0.05, 0.2 and 0.5 μM for the other four biflavonoids) were added to reaction mixtures containing different concentrations of amodiaquine (0.1, 0.4 and 1 μM). The other conditions were the same as in the cytochrome P450 inhibition study.

#### *2.4. Kinetic Characterization of Selamariscina A on Five P450 Enzymes in Human Liver Microsomes*

We used HLMs to determine the mechanisms and constants (Ki values) for selamariscina A inhibition of CYP1A2, CYP2B6, CYP2C8, CYP2C9 and CYP3A. The selamariscina A (0~50 μM) was added into the reaction mixtures, each of which contained concentrations of phenacetin (20, 50, and 100 μM), bupropion (20, 50, and 100 μM), amodiaquine (0.1, 0.4, and 1 μM), rosiglitazone (2, 5, and 10 μM), diclofenac (1, 4, and 10 μM), tolbutamide (50, 100, and 200 μM), and midazolam (0.5, 2, and 5 μM). The substrates were used at concentrations approximately near to their respective Km values [25–27]. The concentration range of selamariscina A varied (0, 0.002, 0.005, 0.02, 0.05, and 0.2 μM for CYP2C8; 0, 0.05, 0.02, 0.05, 0.2, and 0.5 μM for CYP2C9; 0, 0.2, 0.5, 2, 5, and 20 μM for CYP3A; 0, 0.5, 2, 5, 20, and 50 μM for CYP1A2 and CYP2B6). The other conditions were the same as in the cytochrome P450 inhibition study.

#### *2.5. Time-Dependent Inhibition Assay*

The time-dependent inhibition of selamariscina A against CYP2C8 and CYP2C9 enzymes was evaluated using an IC50 shift method. Selamariscina A was pre-incubated at six concentrations (0, 0.002, 0.005, 0.02, 0.05, and 0.2 μM) with HLMs in the presence of an NDAPH generation system for 30 min at 37 ◦C. The reaction was initiated by adding 1 μM amodiaquine or 10 μM diclofenac and further incubated for 10 min. Incubation was terminated by adding 50 μL of ice-cold acetonitrile containing 7 nM trimipramine. After centrifugation, aliquots of supernatants were analyzed by LC–MS/MS.

#### *2.6. Inhibitory E*ff*ect of Selamariscina A against Human UGT Activity*

The ability of selamariscina A to inhibit the metabolism of six UGT enzyme probe substrates was evaluated using previously developed methods with minor modifications [28]. The microsomal incubation was performed by dividing the non-interactive substrate cocktail sets (set A included SN-38 for UGT1A1, CDCA for UGT1A3 and TFP for UGT1A4 while set B included N-SER for UGT1A6, MPA for UGT1A9 and NX for UGT2B7) (Table 2). In brief, HLMs (0.25 mg/mL) were activated by incubation in the presence of alamethicin (25 μg/mL) for 15 min on ice. After the addition of UGT probe substrates and inhibitor (0, 0.5, 2, 5, 20 and 50 μM), the incubation mixtures were pre-incubated at 37 ◦C for 5 min. After pre-incubation, 5 mM UDPGA was added to initiate a reaction, and further incubated for 60 min at 37 ◦C. The reaction was stopped by adding 50 μL of ice-cold acetonitrile containing 250 nM estrone glucuronide (IS). After centrifugation at 18,000 g (5 min, 4 ◦C), aliquots of supernatants were analyzed by LC–MS/MS. All microsomal incubations were conducted in triplicate.

**Table 2.** Selected reaction monitoring (SRM) condition for the major metabolites of the six uridine 5-diphosphoglucuronosyl transferase (UGT) enzyme substrates and internal standard (IS).

