*2.8. Data Analysis*

We analyzed the data with Shimadzu LabSolution LC–MS software. The IC50 values were calculated by WinNonlin software (Pharsight, Mountain View, CA, USA). The type of inhibition and the apparent kinetic parameters for inhibitory activity ( *K*i) were determined by following several criteria: visual inspection of Dixon plots, Lineweaver–Burk double reciprocal plots, and secondary plots of Lineweaver–Burk plots versus biflavonoid concentrations, the size of the sum of squares of the residuals, Akaike Information Criteria values, the S.E. and 95% confidence interval of the parameter estimates from the nonlinear regression analysis [29] using the WinNonlin software. The models tested included competitive, competitive partial, noncompetitive, noncompetitive partial, uncompetitive, uncompetitive partial, and mixed-type inhibition.

#### **3. Results and Discussion**

#### *3.1. Inhibition of Cytochrome P450 Enzymes Activities by Five Biflaovnoids*

The inhibitory potential of the five biflavonoids against cytochrome P450 enzyme activity was evaluated using HLMs (Table 3). We found selamariscina A, amentoflavone, robustaflavone, cupressuflavone, and taiwaniaflavone strongly inhibit CYP2C8 activity with respective IC50 values of 0.019, 0.084, 0.083, 0.083, and 0.12 μM. They also show strong inhibition on CYP2C9 activity with IC50 values of 0.047, 0.15, 0.15, 0.21, and 0.20 μM. Their inhibition of the other seven P450 isoforms was much lower (IC50 ≥ 1.2 μM) than on CYP2C8 and CYP2C9 (IC50 ≤ 0.21 μM). The IC50 value of the inhibition of diclofenac hydroxylase activity by amentoflavone that we found (0.15 μM) is 4.3 times higher than the 0.035 μM reported by von Moltke et al. (2004) [19]. The reason could be differences in incubation conditions, such as CYP2C19 probe substrate and concentrations (diclofenac 10 μM versus S-mephenytoin 25 μM) [30]. The inhibitory potential (IC50 = 1.3 μM) of amentoflavone on CYP3A was similar to the previously reported value (IC50 = 4.3 μM) [19].

**Table 3.** Inhibitory effects of five biflavonoids and montelukast against nine cytochrome P450 isoforms.


As the five flavonoids strongly inhibited microsomal CYP2C8 activity, we sought to clarify the mechanism of inhibition. The Lineweaver–Burk plots, Dixon plots and secondary reciprocal plots indicated that selamariscina A, amentoflavone, robustaflavone, cupressuflavone, and taiwaniaflavone noncompetitively inhibited CYP2C8-mediated amodiaquine N-deethylation activity with Ki values of 0.018, 0.083, 0.084, 0.103, and 0.142 μM, respectively (Table 4), which are lower than those of the well-known CYP2C8 inhibitors zafirlukast (0.39 μM) [31] and quercetin (4.72 μM) [32].

**Table 4.** *K*i values for inhibition of CYP2C8-catalyzed amodiaquine *N*-deethylation in human liver microsomes by five biflavonoids.


a Values represent the average ± standard error in triplicate.

Of the five biflavonoids, selamariscina A most strongly inhibited CYP2C8-mediated amodiaquine N-deethylation activity with a Ki value of 0.018 μM, which is similar to the IC50 value of the known strong CYP2C8 inhibitor montelukast (0.020 μM) [31]. Further, its inhibitory potential against CYP2C8 was much stronger than other known CYP2C8 inhibitors axitinib (Ki = 0.17 μM [33]), clotrimazole (IC50 = 0.78 μM [31]), felodipine (IC50 = 1.20 μM [31]), nilotinib (Ki = 0.10 μM [33]), and quercetin (Ki = 1.56 μM [34]). We further evaluated the inhibition mechanism of selamariscina A, which showed the strongest CYP2C8 inhibition, for the other P450 enzymes. The inhibitory potential (Ki) of selamariscina A against P450 enzyme activities was in the order: CYP2C8 > CYP2C9 > CYP1A2 > CYP3A > CYP2B6 (Table 5, Figure 2 and Figure S1). To determine whether inhibition by selamariscina A was substrate-specific, we evaluated its inhibitory e ffects on CYP2C8-mediated rosiglitazone 5-hydroxylation. We found that it showed strong inhibition with a Ki value of 0.010 μM in a substrate-independent manner. Selamariscina A also inhibited CYP2C9-mediated diclofenac and tolbutamide hydroxylation with Ki values of 0.032 and 0.065 μM, respectively, in a substrateindependent manner. Its inhibitory potential against CYP2C9 was much stronger than other known CYP2C8 inhibitors sulfaphenazole (Ki = 0.12–0.7 μM [35]), fluvoxamine (Ki = 0.63–16 μM [35]), fluconazole (Ki = 0.28 μM [36]), and fluoxetine (Ki = 19–87 μM [35]). Its inhibitory potential for CYP2C8 and CYP2C9 was much stronger than other P450s.

**Table 5.** *K*i values for the inhibition of CYP1A2-catalyzed phenacetin *O*-deethylation, CYP2B6-catalyzed bupropion hydroxylation, CYP2C8-catalyzed amodiaquine *N*-deethylation, CYP2C8-catalyzed rosiglitazone 5-hydroxylation, CYP2C9-catalyzed diclofenac 4-hydroxylation, CYP2C9-catalyzed tolbutamide 4-hydroxylation, and CYP3A-catalyzed midazolam 1-hydroxylation in human liver microsomes by selamariscina A.


aValues represent the average ± standard error in triplicate.

In addition, several P450 inhibitors including azamulin, clopidogrel, methoxalene, and ticlopidine [37–39] have been shown to be time-dependent inhibitors of cytochrome P450. We investigated the e ffect of incubation time on IC50 values of selamariscina A using the CYP2C8 substrate amodiaquine and the CYP2C9 substrate diclofenac. The inhibitory potential of selamariscina A against CYP2C8-mediated amodiaquine O-deethylase activity and CYP2C9-mediated diclofenac hydroxylase activity in HLMs pre-incubated in the presence of an NADPH generation system (IC50 values of 0.031 and 0.092 μM, respectively) was a bit weaker than in untreated HLMs (IC50 values of 0.019 and 0.054 μM, respectively). This suggests that selamariscina A is not a time-dependent inhibitor (data are not shown).

**Figure 2.** Representative Dixon plots obtained from a kinetic study of CYP1A2-catalyzed phenacetin *O*-deethylation (**A**), CYP2B6-catalyzed bupropion hydroxylation (**B**), CYP2C8-catalyzed amodiaquine N-deethylation (**C**), CYP2C8-catalyzed rosiglitazone 5-hydroxylation (**D**), CYP2C9-catalyzed diclofenac 4-hydroxylation (**E**), and CYP3A-catalyzed midazolam 1-hydroxylation (**F**) in the presence of different concentrations of selamariscina A in pooled human liver microsomes (XTreme 200, XenoTech). Each data point shown represent the mean ± standard error in triplicate for the samples.

#### *3.2. Inhibition of UGT Enzymes Activities by Selamariscina A*

The inhibitory potential of selamariscina A against uridine 5-diphosphoglucuronosyl transferase (UGT) activity was evaluated using HLMs (Table 6). Selamariscina A inhibited UGT1A1 and UGT1A4 activity with IC50 values of 1.7 and 7.7 μM, respectively. However, its inhibition of UGT1A1 and UGT1A4 isoforms was much weaker than that of CYP2C8 (IC50 = 0.019 μM). The inhibitory potential of selamariscina A for UGT1A3, UGT1A6, UGT1A9, and UGT2B6 was negligible (IC50 > 40 μM). The IC50 value of the inhibition of UGT1A1 activity by amentoflavone found in our study (1.7 μM) is similar to its previously reported value (IC50 = 0.78 μM) [18].

**Table 6.** Inhibitory effects of selamariscina A against six uridine 5-diphosphoglucuronosyl transferase (UGT) isoforms.


\* SN-38: 7-Ethyl-10-hydroxy camptothecin; a values represent the average ± standard error in triplicate.

#### *3.3. Comparison of the Selectivity of Selamariscina A and Montelukast for CYP2C8 Inhibition*

Montelukast has been used to inhibit CYP2C8 in reaction-phenotyping studies [20]. We re-evaluated its inhibitory potential against the nine P450 isoforms in this study using HLMs (XTreme 200, XenoTech). Montelukast strongly inhibited CYP2C8 activity with an IC50 value of 0.52 μM, but it showed weak inhibition on the other eight P450 enzymes (IC50 > 9.73 μM) (Table 3). The IC50 value for the CYP2C8 isoform (IC50 = 0.52 μM at 0.25 mg/mL microsomal protein concentration) was similar to previously reported values (IC50 = 0.18 μM at 0.3 mg/mL microsomal protein concentration) [20]. However, montelukast showed more than 25 times weaker inhibition than selamariscina A (IC50 = 0.019 μM at 0.25 mg/mL microsomal protein concentration). At 0.5 μM selamariscina A concentration, approximately 25 times greater than the Ki value, selamariscina A was found to inhibit CYP2C8 and CYP2C9 by 92.8% and 88.6% respectively, and only slightly affected the enzyme activities of the other P450 isoforms (Figure 3). Selamariscina A at 0.5 μM concentration inhibited none of the other P450 isoform-specific activities above 21.8% in HLMs, indicating that selamariscina A could be used as a selective CYP2C8 and CYP2C9 inhibitor in P450 phenotyping studies. Montelukast at 0.5 μM concentration, a well-known selective CYP2C8 inhibitor [20], showed moderate inhibition on CYP2C8 by 52.7% in pooled HLMs. At 5 μM concentration, montelukast inhibited CYP2C8 by 86.1% in HLMs; however, it also inhibited CYP2C9 and CYP2B6 activities by 31.0% and 20.4%, respectively in pooled HLMs. Montelukast (5 μM) showed negligible inhibition on the other six P450 isoforms. Selamariscina A could be useful as a strong CYP2C8 and CYP2C9 inhibitor in P450 reaction-phenotyping studies.

**Figure 3.** Inhibitory effects of selamariscina A (0.5 μM, ) and montelukast (0.5 μM, ; 5 μM, ) on the enzymatic activities of nine P450 isoforms in pooled human liver microsomes (0.25 mg/mL, XTreme 200, XenoTech). Phenacetin (100 μM), coumarin (5 μM), bupropion (50 μM), amodiaquine (1 μM), diclofenac (10 μM), omeprazole (20 μM), dextromethorphan (5 μM), chlorzoxazone (50 μM), and midazolam (5 μM) were used as the respective substrates of P450s 1A2, 2A6, 2B6, 2C8, 2C9, 2C19, 2D6, 2E1, and 3A. The data are means of the average ± standard error in triplicate.

#### *3.4. Evaluation of Drug Interaction Potential of Selamariscina A*

It was estimated that an in vivo interaction potential via the inhibition of P450 would likely occur if the ratio of inhibitor *C*max/*K*<sup>i</sup> exceeded one and would be possible if it was between 0.1 and 1.0 [40,41]. Based on amentoflavone's maximum concentrations (0.041 and 0.063 μM) in rat blood after a single oral dose of *Selaginella doderleinii* Hieron extracts (200 mg/kg; contents: 103.82 mg/g amentoflavone, 37.52 mg/g robustaflavone, 44.4 mg/g 2,"3-dihydro-3,3"-biapigenin, 53.4 mg/g <sup>3</sup>,3"-binaringenin, and 35.1 mg/g delicaflavone) [42] and *Selaginella doderleinii* Hieron extracts (600 mg/kg) [43], the respective values of *C*max/*K*<sup>i</sup> were 0.49 and 0.76 from the data of pooled HLMs (*K*i = 0.083 μM), indicating that amentoflavone has possible drug interaction potential with CYP2C8 substrate drugs [44]. Recently, nanotechnology-based delivery systems such as liposomes have been developed for improving oral bioavailability [42]. The values of *C*max (0.22 μM) of amentoflavone after administration of liposome-based *Selaginella doderleinii* Hieron extracts (200 mg/kg) were 5.4 times higher than those of the control [42], resulting in a *C*max/*K*<sup>i</sup> value of 2.65, indicating that amentoflavone has drug interaction potential. In the case of selamariscina A, the present study provides the first published data on its pharmacokinetics in animals and humans. Therefore, it is difficult to estimate the drug interaction potential of selamariscina A for humans. However, selamariscina A might have drug interactions with CYP2C8 substrate drugs such as cerivastatin [45], paclitaxel [46], and rosiglitazone [47] because its CYP2C8 inhibitory potential was more than 4.5 times stronger than that of amentoflavone. Therefore, in vivo studies are necessary to determine whether drug interactions between selamariscina A and CYP2C8 or CYP2C9 substrates have clinical relevance.
