Ginseng has been used as a traditional Asian medicine in Korea, Japan, and China for thousands of years (Xiang et al. 2008; Liu et al. 2021; Li et al. 2022). Ginsenosides which are main active constituents of ginseng, have a variety of pharmacological properties such as anti-aging (Park et al. 2022), anti-cancer (Hwang et al. 2022), anti-diabetic (Park et al. 2019), anti-hyperlipidemic (Kwak et al. 2010), anti-inflammatory (Cho et al. 2023), and anti-obesity activities (Yoon et al. 2021).

The inhibition of cytochrome P450 (P450) by ginsenosides has been evaluated for a long time and is considered to be a main cause for ginseng-drug interactions (Dong et al. 2017; Seong et al. 2018; Kim et al. 2020; Xie and Wang 2023). However, unlike P450, studies on the ability of ginsenosides to inhibit uridine 5’-diphospho-glucuronosyltransferase (UGT) enzymes are lacking. Fang et al. (2013) reported that 100 µM of the ginsenosides Rc and F2 inhibited > 69% of UGT1A9- and UGT1A6-mediated 4-methylumbelliferone glucuronidation activities, respectively. Ginsenoside Rh2 also inhibited UGT1A6, UGT2B7, and UGT2B17 activities with inhibition values of > 75%, whereas ginsenoside Rg3 inhibited UGT1A6, UGT1A7, UGT1A9, UGT2B7, and UGT2B15 activities with inhibition values of > 85% in recombinant UGT isoforms (Fang et al. 2013). Kim et al. (2016) reported stereoselective inhibition of UGT1A1-mediated SN38 glucuronidase and UGT1A3-mediated chenodeoxycholic acid glucuronidase activities by ginsenoside Rh2 and Rg2. Ginsenoside 20 (S)-Rh2 inhibited UGT1A1 and UGT1A3 activities with IC₅₀ values of 44.8 and 37.9 µM, respectively, whereas 20 (R)-Rh2 showed negligible inhibition of the UGT1A1 enzyme (Kim et al. 2016). Contrary to the ginsenoside Rh2, UGT1A1 activity was inhibited by the S enantiomer of ginsenoside Rg2. Our group also reported that ginsenoside Rc selectively inhibited UGT1A9-catalyzed mycophenolic acid and the propofol glucuronidation activities with K_i values of 3.31 and 2.83 µM, respectively (Lee et al. 2019).

In this study, we evaluated the inhibitory potential of 6 protopanaxadiol (PPD)- and 5 protopanaxatriol (PPT)-type ginsenosides against the activity of 6 UGT isoforms in human liver microsomes (HLMs). We also elucidated the ginsenoside structure and UGT inhibition relationship based on the IC₅₀ values obtained.

Materials

7-Ethyl-10-hydroxy-campothesin (SN-38) was obtained from Santa Cruz Biotechnology (Dallas, TX). N-Acetylserotonin, alamethicin, chenodeoxycholic acid, estrone-β-D-glucuronide (EG), mycophenolic acid, naloxone, trifluoperazine, and uridine 5’-diphosphoglucuronic acid (UDPGA) were purchased from Sigma-Aldrich (St. Louis, MO). Ginsenoside F1 (98.2%), F2 (98.4%), Rb1 (99.0%), Rb2 (98%), Rd (98.9%), Re (98%), Rf (95.0%), Rg1 (99.7%), Rg3 (98.5%), Rh1 (98%), and Rh2 (98%) were obtained from Ambo Institute (Daejeon, Korea). Pooled HLMs (XTreme 200) were acquired from XenoTech (Lenexa, KS). All the other solvents and reagents were of analytical or liquid chromatography (LC)-mass spectrometry (MS) grade.

Inhibitory potential of 11 ginsenosides against the activity of 6 UGTs

The inhibitory effects of the six protopanaxadiol (PPD)-type ginsenosides (ginsenosides F2, Rb1, Rb2, Rd, Rg3, and Rh2, Fig. 1) and five protopanaxatriol (PPT)-type ginsenosides (ginsenosides F1, Re, Rf, Rg1, and Rh1, Fig. 2) were determined using our UGT cocktail assays and pooled HLMs incubated in the presence or absence of each of the ginsenoside. Each ginsenoside (0-50 M) was screened for the inhibition of HLM UGT1A1, UGT1A3, UGT1A4, UGT1A6, UGT1A9, and UGT2B7 activities. Our previously developed UGT isoform-selective substrate cocktail method (Joo et al. 2014) was used to estimate the IC₅₀ (concentration of the inhibitor causing 50% inhibition of the original enzyme activity) values. In brief, the HLMs were diluted in 100 mM Tris HCl buffer (pH 7.4) to a concentration of 0.25 mg/mL and then incubated in the presence of alamethicin (25 µg/mL) for 15 min on ice. After the addition of the UGT isoform-selective substrates (SN-38 for UGT1A1, chenodeoxycholic acid for UGT1A3, trifluoperazine for UGT1A4, N-acetylserotonin for UGT1A6, mycophenolic acid for UGT1A9, and naloxone for UGT2B7) and pre-incubation for 5 min, the reaction was started by adding UDPGA (5 mM). After incubation at 37℃ for 50 min, the reaction was terminated by adding 50 µL ice-cold acetonitrile containing EG as an internal standard (IS) and sample was centrifuged at 10,000 × g at 4℃ for 5 min. All evaluations were performed in triplicate.

Figure 1. Chemical structures of the protopanaxadiol-type ginsenosides. (A) Ginsenosides Rb1, (B) Rb2, (C) Rd, (D) Rg3, (E) Rh2, and (F) F2.

Figure 2. Chemical structures of the protopanaxatriol-type ginsenosides. (A) Ginsenoside Rh1, (B) Rg1, (C) Re, (D) Rf, and (E) F1.

LC-MS/MS analysis

All metabolites and the IS were separated on a Kinetex XB-C18 column (100 × 2.10 mm, 2.6 µm, 100 Å; Phenomenex, Torrance, CA, USA) and analyzed by the LC-triple quadrupole mass spectrometry (MS/MS) system (Shimadzu LCMS 8060, Shimadzu, Kyoto, Japan) as previously described (Lee et al. 2019). The mobile phase consisted of 0.1% formic acid in water (A) and 0.1% formic acid in acetonitrile (B) and was set as 0%→30% B (0-1 min), 30%→50% B (1-5 min), 50%→0% B (5-5.1 min) and 0% B (5.1-8 min). The total run time was 8 min, and the flow rate was 0.2 mL/min. Electrospray ionization was performed in positive-ion mode at 4000 V or in negative-ion mode at −3500 V. The optimum operating conditions were determined as follows: vaporizer temperature, 300℃; capillary temperature, 350℃; collision gas (argon) pressure, 1.5 mTorr. Quantitation was conducted in selected reaction monitoring (SRM) modes with the precursor-to-product ion transition for each metabolite (Table 1, Fig. 3).

Figure 3. SRM chromatograms from the analysis of human liver microsomal incubated with the UGT substrate cocktails: SN-38 glucuronide (A), chenodeoxycholic acid 24-glucuronide (B), trifluoperazine N-glucuronide (C), N-acetylserotonin glucuronide (D), mycophenolic acid 7-O-glucuronide (E), naloxone 3-glucuronide (F), estrone-β-D-glucuronide (G).

Table 1 . Selected reaction monitoring (SRM) condition for the major metabolites of the six UGT enzyme substrates and internal standard (IS).

UGT enzyme	Substrates	Concentration (µM)	Metabolites	SRM transition (m/z)	Polarity	Collision energy (eV)
1A1	SN-38	0.5	SN-38 glucuronide	569 > 393	ESI+	30
1A3	Chenodeoxycholic acid (CDCA)	2	CDCA-24 glucuronide	567 > 391	ESI−	20
1A4	Trifluoperazine (TFP)	0.5	TFP N-glucuronide	584 > 408	ESI+	30
1A6	N-Acetylserotonin (N-SER)	1	N-SER glucuronide	395 > 219	ESI+	10
1A9	Mycophenolic acid (MPA)	0.2	MPA 7-O-glucuronide	495 > 319	ESI−	25
2B7	Naloxone (NX)	1	NX 3-glucuronide	504 > 310	ESI+	30
IS	EG	0.25		445 > 269	ESI−	35

Data analysis

IC₅₀ values were calculated from the following equation using WinNonlin software (Pharsight, Mountain View, CA):

where A is the maximum activity and I is the inhibitor concentration (Kim et al. 2006).

Inhibition of UGT activities by six protopanaxadiol-type ginsenosides

Six PPD-type ginsenosides had no inhibitory effect on the UGT1A1-, UGT1A4-, and UGT1A6-mediated glucuronidation (IC₅₀ > 50 µM, Table 2). Ginsenosides Rg3 and Rh2 weakly inhibited the UGT1A3-mediated chenodeoxycholic acid glucuronidation (IC₅₀ ≅ 35 µM), whereas the other four PPD-type ginsenosides showed negligible inhibition of the UGT1A3 activity (IC₅₀ > 50 µM) (Table 2). Ginsenoside Rg3 moderately inhibited the UGT2B7-mediated naloxone glucuronidation (IC₅₀ = 9.6 µM), whereas the other PPD-type ginsenosides showed negligible inhibition of the UGT2B7 activity (IC₅₀ > 50 µM) (Table 2). Ginsenosides Rg3, Rb1, Rb2, and Rd showed weak inhibition of the UGT1A9-mediated mycophenolic acid glucuronidation (IC₅₀ = 10-25 µM).

Table 2 . Inhibitory effects of 11 ginsenosides on six uridine 5’-diphospho-glucuronosyltransferase (UGT) isoforms activity in human liver microsomes.

	Ginsenoside	IC50 (µM)
		UGT1A1	UGT1A3	UGT1A4	UGT1A6	UGT1A9	UGT2B7
Protopanaxadiol-type ginsenosides	Rb1	> 50	> 50	> 50	> 50	21.3	> 50
	Rb2	> 50	> 50	> 50	> 50	22.7	> 50
	Rd	> 50	> 50	> 50	> 50	17.4	> 50
	Rg3	> 50	34.8	> 50	> 50	10.2	9.6
	Rh2	> 50	34.6	> 50	> 50	> 50	> 50
	F2	> 50	> 50	> 50	> 50	> 50	> 50
Protopanaxatriol-type ginsenosides	Rh1	> 50	> 50	> 50	> 50	> 50	> 50
	Rg1	> 50	> 50	37.1	> 50	> 50	> 50
	Re	> 50	> 50	> 50	> 50	> 50	> 50
	Rf	> 50	> 50	> 50	> 50	> 50	> 50
	F1	> 50	> 50	> 50	> 50	> 50	> 50

Inhibition of UGT activities by five protopanaxatriol-type ginsenosides

Five PPT-type ginsenosides had no inhibitory effect on UGT1A1-, UGT1A3-, UGT1A6-, UGT1A9-, and UGT2B7-mediated glucuronidation (IC₅₀ > 50 µM, Table 2). Ginsenoside Rg1 weakly inhibited the UGT1A4-trifluoperazine glucuronidation (IC₅₀ = 37.1 µM), whereas the other PPT-type ginsenosides showed negligible inhibition of the UGT1A4 activity (IC₅₀ > 50 µM) (Table 2).

This study investigated the inhibitory effects of 11 ginsenosides on the activity of six UGT isoforms in HLMs. Among the 11 ginsenosides tested, ginsenoside F1, F2, Re, Rf, and Rh1 displayed negligible inhibition of activity of the six UGT tested (IC₅₀ > 50 µM, Table 2). These observations were similar to those reported in previous studies. Kim et al. (2016) reported that the inhibition of the activities of UGT1A1, UGT1A3, UGT1A4, UGT1A6, UGT1A9, and UGT2B7 by ginsenoside Rh1 was negligible in HLMs (IC₅₀ > 50 µM). Fang et al. (2013) reported that ginsenoside F1 and Re had a negligible inhibitory potential on the activity of six UGT isoforms (< 53% inhibition at 100 µM concentration). However, Fang et al. (2013) also reported that ginsenoside F2 had a moderate inhibitory potential on the activity of UGT1A6-catalyzed 4-methylumbelliferone glucuronidase (11.8% inhibition at 100 µM concentration) although they showed weak or negligible inhibitory potential against UGT1A1, UGT1A9, and UGT2B7 (< 63% inhibition at 100 µM concentration). These differences could have originated from the difference in the UGT1A6 substrate (N-acetylserotonin vs 4-methylumbelliferone) and enzyme source (HLMs vs recombinant enzymes) used. For example, ketoprofen inhibited UGT2B7-mediated naloxone and zidovudine glucuronidation with an IC₅₀ value of > 200 µM (Joo et al. 2015) and 40 µM (Mano et al. 2007), respectively, in a substrate dependent manner.

Among the six UGT isoforms tested, the ginsenosides Rh2 and Rg1 showed a weak inhibitory effect only on the activities of UGT1A3 and UGT1A4, respectively (IC₅₀ = 34.6 and 37.1 µM, respectively). These values were similar to those reported by Kim et al. (2016) (IC₅₀ = 37.9 µM). Ginsenoside Rb1, Rb2, and Rd showed a moderate inhibitory effect only on the activity of UGT1A9 among the six UGT isoforms tested (IC₅₀ = 17.4-22.7 µM). Fang et al. (2013) reported the negligible inhibitory potential of three ginsenosides on the activity of UGT1A9-mediated 4-methylumbelliferone glucuronidase (< 61% inhibition at 100 µM concentration). These differences could also be explained by the difference in the UGT1A9 substrate (mycophenolic acid vs 4-methylumbelliferone) and enzyme source (HLMs vs recombinant enzymes) used. The ginsenoside Rg3 inhibited the activities of UGT1A9 and UGT2B7 with IC₅₀ values of 10.2 and 9.6 µM, respectively. Fang et al. (2013) also reported the inhibitory effect of the ginsenoside Rg3 against the activity if the UGT1A9 and UGT2B7 isoforms.

The present study found that all ginsenosides tested had negligible inhibitory effects on the activities of UGT1A1- and UGT1A6-mediated glucuronidation (IC₅₀ > 50 µM, Table 2) and this result was similar to that of previous findings which reported that ginsenosides Rg3, Rh1, and Rh2 had weak inhibitory effects on the activities of UGT1A1 and UGT1A6 in HLMs (IC₅₀ > 45 µM) (Kim et al. 2016). Fang et al. (2013) also reported that the ginsenosides F1, F2, Rb1, Rb2, Rd, Re, Rf, Rg1, Rg3, Rh1, and Rh2 exhibited weak inhibition of UGT1A1-mediated 4-methylumbelliferone glucuronidation in recombinant UGT isoforms (< 65% inhibition at 100 µM concentration).

Interestingly, UGT2B7-mediated naloxone glucuronidase activity was inhibited only by the ginsenoside Rg3, which is PPD-type ginsenoside and has two sugar moieties at the carbon-3 position of the ginsenoside structure (IC₅₀ = 9.6 µM) (Fig. 4). The inhibition of UGT2B7 by the ginsenoside Rg3 was also reported by Kim et al. (2016) and the inhibitory potential was greater than our data (IC₅₀ = 23.1 µM). These differences could have originated from the difference in the UGT1A6 substrate (naloxone vs zidovudine) used. Except for PPT, an aglycone form of the PPT-type ginsenosides Rc, Rg2, and Rh1 showed negligible inhibition of the UGT2B7 enzyme (IC₅₀ > 100 µM) (Kim et al. 2016; Lee et al. 2019). In addition, PPD and compound K also showed negligible inhibition of the UGT2B7 enzyme (IC₅₀ > 73.2 µM) (Jang et al. 2021).

Figure 4. Inhibitory effects of 11 ginsenosides (25 µM) on the enzyme activities of UGT2B7-mediated naloxone glucuronidation in pooled human liver microsomes (0.25 mg/mL). The data are shown as means of experiments performed in triplicate (n = 3).

The UGT1A9-mediated mycophenolic acid glucuronidase activity was inhibited by the ginsenoside Rb1, Rb2, Rd, and Rg3, which are PPD-type ginsenosides that have two sugar moieties at the carbon-3 position of the ginsenoside structure (IC₅₀ = 10.2-22.7 µM) regardless of the presence or absence of a sugar group at the carbon-20 position of ginsenoside structure (Fig. 5). The ginsenoside Rc, which is also a PPD-type ginsenoside that has two sugar moieties at the carbon-3 position of the ginsenoside structure, also inhibited the activity of UGT1A9 (IC₅₀ = 6.34 µM) (Lee et al. 2019). The UGT1A9 inhibition by the ginsenoside Rg3 (IC₅₀ = 10.2 µM) was also reported by Kim et al. (2016) and the inhibitory potential was similar to our data (IC₅₀ = 15.1 µM). PPT and compound K, which are PPD-type ginsenosides that do not have two sugar moieties at the carbon-3 position of ginsenoside structure and PPT-type ginsenosides PPT, Rg2, and Rh1, also showed negligible inhibition against the UGT1A3 enzyme (IC₅₀ > 100 µM) (Kim et al. 2016; Jang et al. 2021).

Figure 5. Inhibitory effects of 11 ginsenosides (25 µM) on the enzyme activities of UGT1A9-mediated mycophenolic acid glucuronidation in pooled human liver microsomes (0.25 mg/mL). The data are shown as means of experiments performed in triplicate (n = 3).

UGT1A3-mediated chenodeoxycholic acid glucuronidase activity was only inhibited by ginsenoside Rg3 and Rh2 which are PPD-type ginsenosides having sugar group at not carbon-20 but carbon-3 position of ginsenoside structure (IC₅₀ = 34.6-34.8 µM) (Fig. 6). The inhibition of UGT1A3 by ginsenoside Rg3 was also reported by Kim et al. (2016) and the inhibitory potential was similar to our data (IC₅₀ = 18.6-20.9 µM). Except for PPT, which is an aglycone form of PPT-type ginsenosides, the PPT-type ginsenosides Rc, Rg2, and Rh1 also showed negligible inhibition against the UGT1A3 enzyme (IC₅₀ > 100 µM) (Kim et al. 2016; Lee et al. 2019).

Figure 6. Inhibitory effects of 11 ginsenosides (25 µM) on the enzyme activities of UGT1A3-mediated chenodeoxycholic acid glucuronidation in pooled human liver microsomes (0.25 mg/mL). The data are shown as means of experiments performed in triplicate (n = 3).

In conclusion, our results indicated that the ginsenosides have structure-specific UGT inhibitory effects. PPD-type ginsenoside have two sugar moieties at the carbon-3 but not at the carbon-20 position of the ginsenoside structure (e.g. ginsenoside Rg3) and inhibited the activity of UGT2B7. In addition, the PPD-type ginsenosides have two sugar moieties at the carbon-3 position of ginsenoside regardless of the presence or absence of the sugar group at the carbon-20 position of the ginsenosides (e.g., ginsenoside Rb1, Rb2, Rd, and Rg3) and inhibited UGT1A9 activity. The UGT1A3 activity was only inhibited by ginsenoside Rg3 and Rh2, which are PPD-type ginsenosides that have a sugar group at carbon-3 but not at the carbon-20 position of the ginsenosides. The PPT-type ginsenosides showed minimal or negligible inhibition against the six evaluated UGT isoforms. These results suggest that ginsenosides have a structure-specific inhibitory effect on UGTs in HLMs. Additionally, ginsenosides with these structures may have clinically relevant pharmacokinetic drug interactions with other co-administered drugs metabolized by UGT2B7, UGT1A9, and/or UGT1A3.

This work was supported by the National Research Facilities and Equipment Center (NFEC) grant financed by Ministry of Education, Government of South Korea (2019R1A6C1010001).

The authors declare that they have no conflict of interest.