Inhibitory Effects of Baicalein Derived from Japanese Traditional Herbal Medicine on SN-38 Glucuronidation

Purpose: The chemotherapeutic agent irinotecan is hydrolyzed to its active form SN-38 by human carboxyesterases, but SN-38 is converted into the inactive form SN-38G by hepatic UDPglucuronosyltransferases (UGTs). The aim of the present study was to evaluate the inhibitory effects of two -glucuronidase-treated Japanese traditional herbal medicines (kampo), Hange-Shashin-To (TJ-14) and SaireiTo (TJ-114) on SN-38 glucuronidation, and the deglycosylation of baicalin (BG) and glycyrrhizic acid (GL) derived from TJ-14 and TJ-114 to form their respective aglycones, baicalein (BA) and glycyrrhetinic acid (GA). Methods: The inhibitory effects of -glucuronidase-treated TJ-14 and TJ-114 on SN-38 glucuronidation by human liver microsomes were examined. BA and GA, which were enzymatically converted from BG and GL present in TJ-14 and TJ-114, were examined in the same manner. Furthermore, the enzymatic activities were measured by using recombinant UGT1A1 and UGT1A9 isoforms instead of human liver microsomes. BA, GA, SN-38, and their glycosides/glucuronides were analyzed with an LC-MS system. Results: As regards the linear initial reaction rate, SN-38 glucuronidation by human liver microsomes was significantly inhibited by the addition of -glucuronidase-untreated TJ-14 and TJ-114, but was more strongly inhibited by the addition of -glucuronidase-treated TJ-14 and TJ-114. The results of LC-MS analysis and pharmacokinetic studies suggested that BA is the main inhibitor of SN-38 glucuronidation. In the Dixon plot, BA showed competitive inhibition of SN-38 glucuronidation, and the inhibition constant was 8.70 ± 3.24 M. Previous reports, studies of recombinant UGT isoforms indicated that SN-38 glucuronidation was mainly catalyzed by UGT1A1. Conclusions: These findings strongly suggested that SN-38 glucuronidation is inhibited by BA. BA could act as a pharmacokinetic regulating factor associated with SN-38 glucuronidation. This article is open to POST-PUBLICATION REVIEW. Registered readers (see “For Readers”) may comment by clicking on ABSTRACT on the issue’s contents page. _______________________________________________________________________________________

The inactive form, SN-38G, is excreted mainly into bile via such transporters as breast cancer resistance protein (BCRP), multidrug resistanceassociated protein (MRP), and P-glycoprotein (11,12). SN-38G is deconjugated by -glucuronidase present in intestinal microflora and regenerated to SN-38. As a result, SN-38 causes gastrointestinal toxicity by inducing mucosal damage. The major dose-limiting toxicity after administration of CPT-11 is severe myelosuppression (13) and late-onset diarrhea (14). In many cases, neutropenia caused by myelosuppression converges in a relatively short time and is treated with granulocyte colonystimulating factor (G-CSF) (15,16). On the other hand, late-onset diarrhea continues to be a serious, debilitating, and life-threatening condition that leads to dehydration (17).
The Japanese traditional herbal medicine (kampo) Hange-Shashin-To (TJ-14) (17) and antibiotics, such as cefixime (18), are orally administered to ameliorate gastrointestinal toxicity because they inactivate bacterial -glucuronidase, thereby suppressing the formation of toxic SN-38 from SN-38G (19). Kampo is prepared by combining multiple crude drugs and thus it contains many substances. As kampo is a combination of many substances, its effect is a combination of various interactions of the constituent substances (20). TJ-14 is commonly used to treat gastrointestinal disorders, such as acute and chronic gastrointestinal catarrh, fermentative diarrhea, and acute gastroenteritis. TJ-14 is composed of seven medicinal herbs, one of which is Scutellariae radix (skullcap), the active ingredient of which is baicalin (BG). BG is the glucuronide of baicalein. A glucuronide is also a glycoside and the most specific term in the field. BG has one glucuronic acid molecule and is a natural glucuronide conjugate. The mechanism of action of TJ-14 toward SN-38-induced late-onset diarrhea is considered to be bacterial -glucuronidase inhibition by BG. As a result, BG alleviates SN-38-induced mucosal damage (21) and enzymatically hydrolyzed BA is absorbed by small intestinal epithelial cells through passive diffusion. TJ-14 also contains Glycyrrhizae radix (licorice). Glycyrrhizic acid (GL) is one of the active ingredients of licorice; it has two glucuronic acid molecules and is the glucuronide of glycyrrhetinic acid (GA). Orally administered GL is enzymatically hydrolyzed via 3monoglucuronyl-glycyrrhetinic acid (3MGA) to more bioactive GA by bacterial -glucuronidase.
Then, the deglycosylated forms, GA and BA, are absorbed into bloodstream and converted via 3MGA into GL and BG by hepatic UGTs (22,23). On the other hand, SN-38 deconjugated by bacterial glucuronidase in intestine is taken up by the organic anion transporting polypeptide (OATP) 2B1 located at the apical membrane of enterocytes (24,25) and absorbed into bloodstream. Based on the above findings, we hypothesized that BA and GA might influence hepatic UGT-mediated SN-38 glucuronidation. In this study, we examined whether the inhibitory effects of BA and GA originating from TJ-14 on UGT activity could be a pharmacokinetic regulating factor associated with SN-38-induced late-onset diarrhea in vitro. In addition to TJ-14, another Japanese traditional herbal medicine, Sairei-To (TJ-114), which is composed of Scutellariae radix and Glycyrrhizae radix and used to treat severe diarrhea, was also examined. Composition of Japanese Traditional Herbal Medicine Japanese traditional herbal medicines TJ-14 and TJ-114 were spray-dried hot-water extracts of a mixture of herbal crude drugs. According to the manufacturer's information, the total daily dose of 7.5 g of TJ-14 (three times a day) consists of 4.5 g of water-extracted herbal powder and 3.0 g of additives.

-Glucuronidase Treatment of Japanese Traditional Herbal Medicine
The deglycosylation of BG to BA and GL via 3MGA to GA in TJ-14 and TJ-114 was conducted according to a previous report (26) with minor modifications. Ground TJ-14 and TJ-114 were added to 50 mM sodium acetate buffer (pH 5.0). The 5.0 w/v% suspensions were incubated without or with 200 units/mL -glucuronidase for 8 hr at 37 °C. A portion of the reaction solutions were passed through a solidphase extraction cartridge (Sep-Pak C18 Plus Short Cartridge, Waters, Milford, MA, U.SA.) and elution was carried out with methanol. The eluates were diluted to 0.5 w/v% solutions using 50 vol% methanol and filtered through 0.45 m membrane filter. The filtrate diluted to one-tenth its original concentration was injected into a liquid chromatograph-mass spectrometer (LC-MS) system.

Glucuronidation of SN-38
The inhibitory effects of -glucuronidase-treated TJ-14 and TJ-114 on SN-38 glucuronidation in human liver microsomes were examined. Ten L of a 0.5 w/v% solution of -glucuronidase-treated or untreated TJ-14 or TJ-114 was added into a microtube and the solution was evaporated under reduced pressure to give a residue. The residue was added to an incubation mixture composed of 50 mM Tris-HCl buffer (pH 7.4) containing 10 mM MgCl2, 0.2 mg protein/mL human liver microsomes, and 25 μg/mL alamethicin, and the mixture was kept on ice for 10 min. Pre-incubation was carried out after the addition of SN-38 methanolic solution for 5 min at 37 °C. The final concentration of methanol was adjusted to 1.0 vol%. The reaction was initiated by adding 5 mM UDPGA in a final volume of 100 L, followed by incubation for 0-60 min at 37 °C. The reaction was terminated by adding 100 L of icecold acetonitrile to the reaction mixture. Precipitated proteins were removed by centrifugation for 5 min at 5,000 rpm and the supernatant was filtered through 0.45 m membrane filter. The filtrate was injected into a LC-MS system.
The inhibitory effects of BA and GA on SN-38 glucuronidation were examined in the same manner as described above except for the addition of glucuronidase-treated or untreated TJ-14 or TJ-114. Furthermore, except for the addition of rUGT1A1 or rUGT1A9 instead of human liver microsomes, the incubation was performed in the same manner as that described above.

DATA ANALYSIS
Kinetic parameters (Vmax, Km, and Ksi (substrate inhibition constant)) were estimated by fitting the Michaelis-Menten and substrate inhibition equations to the substrate concentrations and the initial reaction rates. Data analysis was performed with Enzyme Kinetics Module of SigmaPlot 13 (Systat Software, San Jose, CA, U.S.A.). The goodness of fit was evaluated on the basis of R 2 values, AICc (Akaike's Information Criterion with adjustment for small sample size), and Sy.x (standard error of the regression). For the test of the difference between two independent groups, the Student's t test was conducted after the dispersion test by the F test. The test of the difference between three independent groups was processed using the Dunnett method and the Tukey method after the dispersion test. RESULTS

Time Course and Kinetic Profiles of SN-38 Glucuronidation in Human Liver Microsomes
In order to clarify the kinetic characterization of SN-38 glucuronidation, the effect of incubation time on SN-38 glucuronidation were examined. Human liver microsomes were incubated with 10 M SN-38 for 60 min (Figure 1). SN-38 glucuronidation activity increased in a time-dependent manner and the increase was linear up to 20 min. Then, the effect of substrate concentration on SN-38 glucuronidation in human liver microsomes was examined. Human liver microsomes were incubated with 0-100 M SN-38 for 20 min (Figure 2). SN-38 glucuronidation increased in a concentration-dependent manner up to 10 M and the increase was linear up to 2.0 M (inset in Figure 2 Table 1 shows the contents of five active ingredients BG, BA, GL, 3MGA, and GA in glucuronidase-treated or untreated TJ-14 and TJ-114. The results indicated that deglycosylation by glucuronidase markedly increased BA obtained from BG but slightly increased GA from GL. Accordingly, it was considered that SN-38 glucuronidation is strongly inhibited by aglycone BA rather than glycosides BG and GL, as shown in Figure 3.  Then, the Dixon plots were analyzed to clarify the types of inhibition and the inhibition constant (Ki) of BA in human liver microsomes. In Figure 5, the concentrations of SN-38 were set at 0.5, 0.75 (lower than Km value), and 1.5 M (Km value). BA showed competitive inhibition of SN-38 glucuronidation and the corresponding Ki value was 8.70 ± 3.24 M. SN-38 Glucuronidation by rUGT1A1 and rUGT1A9 SN-38 is mainly converted into inactive SN38G by UGT1A1 (6) and UGT1A9 (27) in the liver. Therefore, we investigated the kinetic profiles of SN-38 glucuronidation by rUGT1A1 and rUGT1A9. Recombinant UGT1A1 or rUGT1A9 was incubated with 0-75 M SN-38 for 20 min (Figure 6). SN-38 glucuronidation increased in a concentrationdependent manner up to 20 M and the increase was linear up to 5.0 M. In the same way as human liver microsomes at concentrations higher than 20 M, the kinetic profiles of SN-38 glucuronidation indicated substrate inhibition (10). The enzyme kinetic parameters of Km, Vmax, CLint, and Ksi by

Inhibitory
Effect of BA on SN-38 Glucuronidation by rUGT1A1 The inhibitory effect of BA on SN-38 glucuronidation by rUGT1A1 was investigated (Figure 7). Co-incubation with BA concentrationdependently decreased SN-38 glucuronidation significantly; the addition of 5 M and 10 M BA reduced SN-38 glucuronidation by 42% and 68%, respectively. On the other hand, SN-38 glucuronidation by rUGT1A9 showed very low activity; hence, SN-38G could not be detected when BA was added to the incubation mixture.

DISCUSSION
The kinetic profiles of SN-38 glucuronidation in human liver microsomes were clarified. SN-38 glucuronidation increased in a time-dependent ( Figure 1) and concentration-dependent manner and showed the highest activity (250 pmol/mg protein/min) at approximately 20 M (Figure 2). However, at concentrations higher than 20 M, the activity gradually decreased, indicating substrate inhibition. Some studies of SN-38 glucuronidation in human liver microsomes reported the Michaelis-Menten type profile (6,(27)(28)(29), whereas Wu et al. (10) reported substrate inhibition profiles in human liver microsomes and rUGT1A1, which were similar to our results. Luukkanen et al. (30) pointed out that the bisubstrate (drug and UDPGA) kinetics of seven recombinant human UGTs 1A1, 1A3, 1A6, 1A7, 1A8, 1A9, and 1A10 indicated substrate inhibition. However, they concluded that these substrate inhibitions did not occur under physiological conditions. We speculate that the difference in kinetic profiles among research groups is probably due to differences in incubation conditions, including the lot number of human liver microsomes, the incubation time, and the ratio of UDPGA to SN-38. In particular, human liver microsomes may be affected by not only UGT contents in donor liver, but also the difference in abundance ratio of such genotypes as UGT1A1*1 and UGT1A1*28 (31). From our results, the Ksi/Km ratio of SN-38 glucuronidation in human liver microsomes was 5.4 (see Figure 2). Consequently, if the concentration of SN-38 in the reaction were set to a value lower than Km, a reaction system that does not undergo substrate inhibition would be realized. From the above, the concentration of SN-38 in the reaction was set to 1.5 M, which was within the linear initial reaction rate and six times lower than the Km value.
Such glycosides as BG and GL, which contain a glucuronic acid moiety, are considered natural prodrugs in herbal medicine. Orally administered glycosides are hydrolyzed by -glucuronidase in the intestinal microflora to form more bioactive aglycones that are absorbed and exhibit various effects. Hence, the inhibitory effects of glucuronidase-treated or untreated TJ-14 and TJ-114 on SN-38 glucuronidation in human liver microsomes were examined. As shown in Figure 3, the addition of 0.05 w/v% (final concentration) each of -glucuronidase untreated TJ-14 and TJ-114 to the incubation mixture significantly inhibited SN-38 glucuronidation by 46% and 52% as residual activity, respectively. Katoh et al. (32) reported that when 0.1 w/v% each of TJ-14 and TJ-114, the amount of which was twice ours, was added to the incubation mixture, the residual activities were approximately 20% and 30%, respectively. Their results coincided with ours. We therefore suggest that TJ-14 and TJ-114 contain active components that inhibit SN-38 glucuronidation. However, because kampo is characterized by its complex composition, it may be difficult to demonstrate its mechanism or mode of action (20). On the assumption that SN-38G was hydrolyzed by -glucuronidase from intestinal microflora and converted into the aglycone, glucuronidase-treated TJ-14 and TJ-114 were added to the incubation mixture. As a result, the residual activities of TJ-14 and TJ-114 were 13% and 28%, respectively. Our results indicate that the inhibitory effect on SN-38 glucuronidation is enhanced in deglycosylated Japanese traditional herbal medicine. Katoh et al. (32) reported that such aglycones as BA and GA exert a greater inhibitory effect on SN-38 glucuronidation than their respective glycosides BG and GL.
The contents of the active ingredients in 5.0 w/v% suspensions of -glucuronidase-treated or untreated TJ-14 and TJ-114 are indicated in Table 1 Table 1 were diluted to 100-fold their amounts by treatment with solid-phase extract and added to the reaction mixture. Therefore, the final concentrations of TJ-14 and TJ-114 in Figure 3 were 0.05 w/v%. When intestinal volume was estimated to be 500 mL (33), the concentrations of TJ-14 and TJ-114 were approximately 0.5 w/v%, and this value was further multiplied by 10 as the dilution coefficient. Thus, BA derived from TJ-14 and TJ-114 corresponded to the addition of approximately 21 M and 17 M, respectively, to the incubation mixture. Both concentrations almost agreed with 14 M, which is the IC50 of BA, as shown in Figure 4. However, conflicting results were obtained between GA contents treated with -glucuronidase corresponding to less than 0.4 M in Table 1 and the IC50 of 15 M in Figure 4. SN-38 glucuronidation is mainly inhibited by BA in human liver microsomes. In order to clarify the type of inhibition and the Ki value of BA, a Dixon plot was drawn ( Figure 5). BA showed competitive inhibition and the corresponding Ki value was 8.7 M. Previous reports of human (32) and rat (34) liver microsomes yielded three times (26.9 M, mixed type) and one hundred times (970 M, mixed type) higher Ki values than ours. This may indicate species difference in the SN-38 glucuronidation inhibition process.
According to Hanioka et al. (35), SN-38 glucuronidation was catalyzed by UGT1A1, UGT1A3, UGT1A6, and UGT1A9 as well as by human liver microsomes. Among these UGT isoforms, UGT1A1 showed the highest activity for SN-38 glucuronidation at both low (1 µM) and high (200 µM) substrate concentrations. The ranking in order of activity at low and high substrate concentrations was UGT1A1 > UGT1A9 > UGT1A6 > UGT1A3 and UGT1A1 > UGT1A3 > UGT1A6 ≥ UGT1A9. Based on this report, the kinetic profiles of SN-38 glucuronidation catalyzed by rUGT1A1 and rUGT1A9 are shown in Figure 6. Both isoforms showed a substrate inhibition profile similar to human liver microsomes, but the CLint value of rUGT1A1 was approximately 25 times higher than that of rUGT1A9. These results suggest that SN-38 glucuronidation in human liver is catalyzed primarily by UGT1A1. However, even if standardization is carried out by the negative control microsomes, each UGT expression level in the recombinant system is different from that in human liver microsomes. Therefore, care is necessary to compare with glucuronidation activities of each rUGT isoform. This UGT1A1-catalyzed SN-38 glucuronidation was inhibited in a concentrationdependent manner by the addition of BA (Figure 7), suggesting that BA is strongly involved in the blockade of this pathway. The maximum plasma concentrations of GA and BA in humans are 0.10-0.43 M (36-38) and 0.49 M (39), respectively. In the present study, the Ki value of BA in human liver microsomes was 8.7 M (Figure 5), which was higher than the plasma concentration. Although BA concentration in liver may be higher than that in plasma, it is very difficult to compare liver concentration due to lack of information.
Hirose et al. (40) reported the relationship between SN-38G/SN-38 concentration ratio and neutropenia in 70 Japanese colorectal cancer patients. The SN-38G/SN-38 concentration ratio was lower in patients who were homozygous for UGT1A1*6, heterozygous for UGT1A1*6 or UGT1A1*28, or double heterozygotes than in patients with wild-type genes. It has been proposed that BG from TJ-14 and TJ-114 inhibits bacterial -glucuronidase (17), and the uptake of SN-38 mediated by OATP2B1 in the apical side of intestinal epithelial cells (24,25), thereby reducing the risk of SN-38-induced lateonset diarrhea. On the other hand, the deglycosylated form, BA, may raise plasma SN-38 levels by inhibiting hepatic UGT1A1 activity as well as UGT1A1*6 or UGT1A1*28, and consequently decrease neutrophil count. Further research is required to prove this speculation.
We summarize our findings as follows: when TJ-14 and TJ-114 are orally administered, BG derived from Scutellariae radix is hydrolyzed to aglycone BA by bacterial -glucuronidase in the intestinal tract. As the deglycosylation of GL derived from Glycyrrhizae radix is inhibited by BG, a sufficient amount of GA could not be produced. These results led us to suggest that GA does not contribute to SN-38 glucuronidation. BA is transported to liver via bloodstream.
CPT-11 is hydrolyzed by carboxyesterase and converted into SN-38, and SN-38 glucuronidation is mainly catalyzed by UGT1A1 in the liver. Our results strongly suggest that SN-38 glucuronidation is inhibited by BA.