Screening of Furanocoumarin Derivatives in Citrus Fruits by Enzyme- Linked Immunosorbent Assay

974 Biol. Pharm. Bull. 27(7) 974—977 (2004) Vol. 27, No. 7

Tetsuya SAITA,* Hiroshi FUJITO, and Masato MORI Department of Pharmacy, Saga University Hospital; 5–1–1 Nabeshima, Saga 849–8501, Japan. Received January 30, 2004; accepted March 16, 2004

This paper reports a sensitive and specific enzyme-linked immunosorbent assay for screening of furanocoumarin derivatives as cytochrome P450 3A4 inhibitors in citrus fruits. Anti-6,7-dihydroxybergamottin antibody was obtained by immunizing rabbits with 6,7-dihydroxybergamottin conjugated with bovine serum albumin using the N-succinimidyl ester method. An enzyme marker was similarly prepared by coupling 6,7-dihydroxybergamottin with b-D-galactosidase. The enzyme-linked immunosorbent assay is capable of detecting as lit- tle as 800 pg/ml of 6,7-dihydroxybergamottin and 4 ng/ml of bergamottin. Cross-reactivity data showed that the antibody well recognizes both the furanocoumarin and 6,7-dihydroxy-3,7-dimethyloct-2-enyloxy moieties of the 6,7-dihydroxybergamottin, and is thus specific to the structure of furanocoumarin derivatives containing geranyloxy side chain as the cytochrome P450 3A4 inhibitors in grapefruit juice. The antibody was, therefore, used for screening a large number of citrus fruits for furanocoumarin derivatives such as 6,7-dihydroxybergamottin. Fifteen citrus fruits were examined and significant reactivity was observed in 8 of these: red pummelo, sweetie, melogold, banpeiyu pummelo, hassaku, sour orange, lime and natsudaidai. This enzyme-linked immunosorbent assay may be a powerful tool for screening for furanocoumarin derivatives as cytochrome P450 3A4 inhibitors in grapefruit juice. Key words furanocoumarin derivative; enzyme-linked immunosorbent assay; screening; citrus fruit; grapefruit juice

Grapefruit juice (GFJ) has been reported to increase the Besides the Rutaceae family (like grapefruit), many plants oral availability of many clinically important drugs.1—5) The of other families such as Umbelliferae, Leguminosae and interaction is caused by the inhibition of their first pass me- Moraceae also contain furanocoumarin derivatives.11) Many tabolisms mainly catalyzed by an intestinal cytochrome P450 of these plants are used as common vegetables12) or tradi- (CYP) 3A4.6,7) Recently, six furanocoumarin derivatives, tional medicines.13) Thus it is possible that furanocoumarin bergamottin, 6,7-epoxybergamottin, 6,7-dihydroxyberg- derivatives contained in these vegetables or crude drugs also amottin (DHB) and three furanocoumarin dimers (Fig. 1) change the pharmacokinetics of a drug. These findings re- were isolated from GFJ as inhibitors of CYP3A4 and are quire us to develop a simple screening method for the fura- now suggested to be clinically active constituents.8—10) nocoumarin derivatives. Their determination in plants has usually been made by thin layer or column chromatography, however, it is very difficult to find these derivatives by chromatography alone. Therefore, we wanted to develop a simple and specific enzyme-linked immunosorbent assay (ELISA) for screening of furanocoumarin derivatives in plants. The methodology for the antibody production, the labeling of furanocoumarin derivative with b-D-galactosidase (b-Gal) to act as a tracer, the characterization of antibody specificity, and the technique developed for the screening of furanocoumarin derivatives in citrus fruits by ELISA is described here.

MATERIALS AND METHODS

Reagents Bergamottin and DHB were purchased from Ultraﬁne, Ltd. (Manchester, England). b-Gal (EC 3.2.1.23) from Escherichia coli and 4-methylumbelliferyl-b-D-galactopyranoside were obtained from Boehringer Mannheim (Mannheim, Germany). All other reagents and solvents were of the highest grade commercially available. Samples of Citrus Fruits Citrus fruits were obtained from local commercial sources and information on them is shown in Table 1. All fruits were squeezed and ﬁltered, and the juices were stored at 20 °C until needed for the study. The fruit juice samples were diluted 50, 250 and 1250 times with 0.25 M phosphate buffer (pH 7.4) containing 10 mM eth- Fig. 1. Chemical Structures of Furanocoumarin Derivatives in GFJ ylenediaminetetraacetate, 0.1% bovine serum albumin (BSA)

Table1. Species and Origin of Citrus Fruits Used in the Present Study

Citrus fruit Species Origin

Grapefruit (white) C. paradisi MACF. California U.S.A. Red pummelo C. grandis OSBECK California U.S.A. Sweetie C. grandis OSBECKC. paradisi MACF. Israel Melogold C. grandis OSBECKC. paradisi MACF. California U.S.A. Banpeiyu pummelo C. grandis OSBECK Kumamoto Japan Hassaku C. hassaku HORT. ex TANAKA Saga Japan Sour orange C. aurantium L. Saga Japan Lime C. aurantifolia SWINGLE Mexico Natsudaidai C. natsudaidai HAYATA Kumamoto Japan Navel orange C. sinensis L. OSBECK var. brasiliensis TANAKA Australia Sweet orange C. sinensis OSBECK Australia Yuzu C. junos SIEB. ex TANAKA Miyazaki Japan Iyokan C. iyo HORT. ex TANAKA Ehime Japan Satsuma mandarin C. unshu MARCOVITCH Fukuoka Japan Ponkan C. reticulata BLANCO Oita Japan Dekopon C. reticulata BLANCO(C. unshu MARCOVITCHC. sinensis L. OSBECK) Oita Japan

and 0.1% NaN3 (buffer A), then used to screen fura- female rabbits were each given multiple subcutaneous injec- nocoumarin derivatives using ELISA. tions over sites along both sides of their backs. Booster injec- Preparation of the Immunogen A solution of DHB tions were then given three times at biweekly intervals, using (4.5 mg, 12 mmol) and succinic anhydride (12 mg, 120 mmol) one-half the amount of the dose of the first immunization. in pyridine (300 ml) was stirred overnight at 60 °C. After re- The rabbits were bled from an ear vein 10 weeks after immu- moving pyridine by passing nitrogen through the reaction nization began. The sera (10 ml) were separated by centrifu- mixture, 1 ml of ethyl acetate and 1 ml of H2O were added to gation and heated at 55 °C for 30 min. Fractions of IgG were the residue, and the mixture was shaken vigorously. The separated from the sera with 50% saturated ammonium sul- ethyl acetate layer was washed with saturated sodium chlo- fate and chromatographed on a column of DEAE-Sephacel ride, dried over anhydrous sodium sulfate, and evaporated to (2.123 cm) using 17.5 mM phosphate buffer (pH 6.8) as an give DHB hemisuccinate (3.4 mg) as a white solid. The re- eluant.14) The fraction passed through the column was sulting DHB hemisuccinate was used without further purifi- lyophilized and used as anti-DHB IgG for ELISA. cation for preparing the conjugates with BSA and b-Gal, re- Enzyme Labeling DHB was labeled by binding to b- spectively, as the immunogen and the tracer in the ELISA. Gal, essentially by the same principle as used for the prepara- The yield of DHB hemisuccinate was tentatively estimated to tion of the DHB immunogen. In brief, a solution of b-Gal be 60% according to HPLC measurements of the quantity of (156 mg, 0.28 nmol) in 1 ml of 0.1 M phosphate buffer (pH nonreacted DHB. 7.0) was mixed with a solution of succinimidyl DHB 1-Ethyl-3,3-dimethylaminopropyl-carbodiimide hydrochlo- hemisuccinate (approximately 0.2 mg, 0.35 mmol) in 50 ml of ride (EDPC) (3 mg, 15.6 mmol) and N-hydroxysuccinimide DMF and incubated at room temperature for 1 h. The mixture (1.8 mg, 15.6 mmol) were added to a solution of DHB was chromatographed on a Sepharose 6B column (2.040 hemisuccinate (approximately 3.4 mg, 7.2 mmol) in 95% diox- cm) using 20 mM phosphate buffer (pH 7.0) containing 0.1 M ane (0.5 ml), and the resulting solution was allowed to stand NaCl, 1 mM MgCl2, 0.1% BSA, and 0.1% NaN3 (buffer B) to at room temperature for 2 h. After the addition of H2O, the remove any remaining small molecules. Four-milliliter frac- resulting mixture was extracted with ethyl acetate. The ethyl tions were collected, and fractions 15 to 17, representing the acetate layer was dried over anhydrous sodium sulfate, and main peak showing enzyme activity, were combined and evaporated to give succinimidyl DHB hemisuccinate as a used as a label in the ELISA. white solid. A solution of BSA (10 mg, 0.15 mmol) in 0.1 M ELISA Method ELISA is based on the principle of phosphate buffer (pH 7.0, 1 ml) was mixed with a solution of competition between enzyme-labeled and unlabeled drugs succinimidyl DHB hemisuccinate (approximately 3.8 mg, for an immobilized antibody, followed by measurement of 6.7 mmol) in N,N-dimethylformamide (DMF, 300 ml) and in- the marker enzyme activity of the immunocomplex bound to cubated at room temperature for 2 h. The reaction mixture the solid phase. Briefly, the wells in microtiter plates (Nunc F was chromatographed on a Sephadex G-100 column (2.8 Immunoplates I, Nunc, Reskilde, Denmark) were coated by 42 cm) with an eluant of 0.1 M phosphate buffer (pH 7.0) loading 150 ml of anti-DHB IgG (0.5 mg/ml) in 10 mM containing 3 M urea. Then the purified conjugate was exam- Tris–HCl buffer (pH 8.5) containing 10 mM NaCl and 10 mM ined spectrophotometrically and estimated to contain approx- NaN3 and allowed to stand for 1 h at 37 °C. After the plates imately 3.1 molecules of DHB per BSA molecule, assuming had been washed twice with buffer A, they were incubated the molar extinction coefficients of DHB to be 5300 at with 200 ml of 10 mM Tris–HCl buffer (pH 8.5) containing

280 nm and 14300 at 310 nm, and those of BSA to be 43600 10 mM NaCl, 10 mM NaN3 and 2% BSA for 20 min at 37 °C at 280 nm. to prevent nonspecific adsorption. The anti-DHB IgG-coated Antibody Production in Rabbits An aliquot, containing wells were then filled with 50 ml of either sample, or buffer A about 1 mg of the DHB–BSA conjugate, was emulsified with as a control, followed immediately by 50 ml of the pooled an equal volume of Freund’s complete adjuvant. Two white DHB–b-Gal conjugate (diluted 1 : 2000 in buffer A). The 976 Vol. 27, No. 7 wells were then incubated for 3 h at room temperature and washed once again thoroughly with buffer A. The activity of the enzyme conjugate bound to each well was then measured by the addition of 125 ml of 0.1 mM 4- methylumbelliferyl-b-D-galactopyranoside in buffer B, followed by incubation of the wells at 37 °C for 30 min. The enzyme reaction was stopped by the addition of 75 ml of 0.5 M glycine–NaOH buffer (pH 10.3) to each well, and the fluorescence due to resulting 4-methylumbelliferone was measured at wavelengths of 355 nm for excitation and 460 nm for emis- sion using a fluorescence microplate reader (Fluoroskan As- cent, Labsystems, Helsinki, Finland).

RESULTS AND DISCUSSION

As shown in Fig. 1, six furanocoumarin derivatives isolated from GFJ contain a geranyloxy side chain(s). Berg- amottin and DHB are the main furanocoumarin derivatives, all of which reduce CYP3A4 activities through both compet- itive and mechanism-based inhibition mechanisms.6,15,16) Ohta et al.17) reported that the presence of 5-geranyloxyfuranocoumarin moiety is essential for the CYP3A4 inhibition. Therefore, to develop an ELISA for screening of furanocoumarin derivatives as the CYP3A4 inhibitors, it is nec- essary to produce an anti-furanocoumarin derivative antibody that recognizes the 5-geranyloxyfuranocoumarin moiety of the furanocoumarin derivatives. DHB has two hydroxyl Fig. 2. Preparation of the Immunogen for Furanocoumarin Derivatives groups at the terminal of the geranyl group. Ohta et al.17) reported that the hydroxy group at the C-6 position (Fig. 2) is readily available for introducing a reactive group into the molecule. Therefore, we chose this site to introduce the car- boxylic group. DHB immunogen was prepared using the N- hydroxysuccinimide ester method as shown in Fig. 2. The DHB–BSA conjugate, whose DHB/BSA molar ratio is about 3, induced the formation of specific antibodies in each of the two rabbits immunized. The DHB–b-Gal conjugate was also prepared by essentially the same procedure. The conjugate thus obtained was stable in eluted buffer (pH 7.0) at 4 °C for more than 6 months, during which no loss of enzyme activity or immunoreactivity was seen. Using anti-DHB antibody and DHB–b-Gal as a tracer, an ELISA for DHB and bergamottin was developed (Fig. 3). The sensitivities of DHB and bergamottin in the ELISA were shown to be 800 pg/ml with DHB and 4 ng/ml with berg- Fig. 3. Dose–Response Curves for DHB and Bergamottin with Anti-DHB amottin (amount of each compound at the B/B0 values of IgG 80%), respectively. Therefore, this ELISA may be sensitive The curves show the amount (percentage) of bound enzyme activity for various enough to screen DHB and bergamottin. doses of DHB and bergamottin (B) as a ratio of that bound using DHB–b-Gal alone (B ). , DHB; , bergamottin. The antibody specificity was determined by the cross-re- 0 activity with other types of furanocoumarin derivatives, which were defined as the percentage of each compound to derivatives isolated from GFJ have not yet been confirmed. DHB in the concentrations required for 50% replacement of However, it is estimated that they show similar cross-reaction bound b-Gal activity. The anti-DHB antibody showed 17.2% to bergamottin or DHB judging from the specificity of the cross-reaction with bergamottin and 0.28% with 5-methoxy- anti-DHB antibody. Therefore, this ELISA may be specific psoralen; no detectable cross-reaction, however, was found enough to screen furanocoumarin derivatives containing ger- with 8-methoxypsoralen (Table 2). These findings suggest anyloxy side chain as the inhibitor of CYP3A4 in GFJ. that the antibody well recognizes both the furanocoumarin Fifteen citrus fruit juices were examined for detection of and 6,7-dihydroxy-3,7-dimethyloct-2-enyloxy moieties of furanocoumarin derivatives using the ELISA for DHB (Table DHB. Thus, the anti-DHB antibody recognizes 5-geranyl- 3). Values of the immunoreactivity showed the amount (per- oxyfuranocoumarin moiety that is essential for the CYP3A4 centage) of bound enzyme activity for test samples (B) as a inhibition. The cross-reactivities of the other furanocoumarin ratio of that bound using DHB–b-Gal alone (B0). Therefore, July 2004 977

Table2. Specificity of Anti-DHB IgG furanocoumarin derivatives from these fruits, however, this result suggests the possibility that they do contain a small Compound % cross-reaction (50%) amount of the derivatives. On the other hand, no detectable immunoreaction was found in the other citrus fruits, and nat- urally they have not been reported to contain furanocoumarin DHB 100.0 derivatives. These results indicate that the ELISA for DHB may be a powerful tool for screening for these derivatives as CYP3A4 inhibitors. Guo et al.10) reported that the inhibitory potency of GFJ samples tended to be higher with GFJ con- Bergamottin 17.2 taining a higher amount of furanocoumarin derivatives. This raises the possibility that the citrus fruits showing a stronger immunoreactivity by the ELISA show a stronger drug inter- 5-Methoxypsoralen 0.28 action. This is the first report of screening for furanocoumarin derivatives in citrus fruits by ELISA. Besides the Rutaceae 8-Methoxypsoralen 0.06 family, many plants of other families such as Umbelliferae, Leguminosae and Moraceae also contain furanocoumarin derivatives.11) We are now trying to screen furanocoumarin de- Table3. Immunoreactivity of Furanocoumarin Derivatives in Citrus Fruits rivatives using ELISA from a large number of crude drugs. with Anti-DHB IgG REFERENCES Immunoreactivity (B/B , %) 0 1) Bailey D. G., Spence J. D., Munoz C., Arnold J. M. O., Lancet, 337, Citrus fruit 268—269 (1991). Dilution (fold) 2) Ducharme M. P., Warbasse L. H., Edwards D. J., Clin. Pharmacol. 50 250 1250 Ther., 57, 485—491 (1995). 3) Bailey D. G., Arnold J. M. O., Spence J. D., Br. J. Clin. Pharmacol., Grapefruit (white) 2.8 4.5 10.3 46, 101—110 (1998). Red pummelo 5.9 7.3 13.5 4) Dresser G. K., Bailey D. G., Carruthers S. G., Clin. Pharmacol. Ther., Sweetie 5.6 8.7 13.8 68, 28—34 (2000). Melogold 4.8 9.3 17.1 5) Christensen H., Asberg A., Holmboe A.-B., Berg K. J., Eur. J. Clin. Banpeiyu pummelo 8.5 17.3 28.3 Pharmacol., 58, 515—520 (2002). Hassaku 8.4 17.6 29.5 6) Schmiedlin-Ren P., Edwards D. J., Fitzsimmons M. E., He K., Lown K. Sour orange 12.5 22.8 42.3 K., Woster P. M., Rahman A., Thummel K. E., Fisher J. M., Hollem- Lime 17.2 26.5 46.0 berg P. F., Watkins P. B., Drug Metab. Dispos., 25, 1228—1233 Natsudaidai 17.6 28.4 50.9 (1997). Navel orange 29.1 50.3 70.3 7) Lown K. S., Bailey D. G., Fontana R. J., Janardan S. K., Adair C. H., Sweet orange 56.1 78.6 93.3 Fortlage L. A., Brown M. B., Guo W., Watkins P. B., J. Clin. Invest., Yuzu 75.3 92.3 98.0 99, 2545—2553 (1997). Iyokan 97.9 104.6 102.3 8) Edwards D. J., Bellevue F. H., Woster P. M., Drug Metab. Dispos., 24, Satsuma mandarin 99.0 100.9 100.2 1287—1290 (1996). Ponkan 96.0 99.0 99.3 9) Fukuda K., Ohta T., Oshima Y., Ohashi N., Yoshikawa M., Yamazoe Dekopon 93.0 95.7 96.7 Y., Pharmacogenetics, 7, 391—396 (1997). 10) Guo L.-Q., Fukuda K., Ohta T., Yamazoe Y., Drug Metab. Dispos., 28, 766—771 (2000). 11) Pathak M. A., Daniels F., Fitzpatrick T. B., J. Invest. Dermatol., 39, the higher B/B0 value suggests a lower content, and the lower B/B value means a higher content of DHB, bergamottin 225—239 (1962). 0 12) Beier R. C., Rev. Environ. Contam. Toxicol., 113, 47—137 (1990). and/or the related furanocoumarin derivatives in the test sam- 13) Namba T., “Coloured Illustrations of Wakan-Yaku, (The Crude Drugs ple. Eight (red pummelo, sweetie, melogold, banpeiyu pum- in Japan, China and the Neighbouring Countries),” Hoikusha Publish- melo, hassaku, sour orange, lime and natsudaidai) out of the ing, Osaka, 1980, in Japanese with English index. 15 samples tested showed a significant immunoreactivity, in- 14) Onoue K., Yagi Y., Pressman D., J. Immunol., 92, 173—184 (1964). dicating the existence of furanocoumarin derivatives. Five 15) Bellevue F. H., Woster P. M., Edwards D. J., He K., Hollenberg P. F., 10) 10) 10) Bioorg. Med. Chem. Lett., 7, 2593—2598 (1997). (red pummelo, sweetie, banpeiyu pummelo, sour or- 16) He K., Iyer K. R., Hayes R. N., Sinz M. W., Woolf T. F., Hollenberg P. 18) 19) ange and lime ) of the 8 citrus fruits showing a significant F., Chem. Res. Toxicol., 11, 252—259 (1998). immunoreactivity, have already been reported to contain fu- 17) Ohta T., Nagahashi M., Hosoi S., Tsukamoto S., Bioorg. Med. Chem., ranocoumarin derivatives as the CYP3A4 inhibitors in GFJ. 10, 969—973 (2002). 18) Malhotra S., Bailey D. G., Paine M. F., Watkins P. B., Clin. Pharma- Therefore, it is estimated that melogold, hassaku and natsu- col. Ther., 69, 14—23 (2001). daidai also contain the furanocoumarin derivatives. Navel or- 19) Bailey D. G., Dresser G. K., Bend J. R., Clin. Pharmacol. Ther., 73, ange, sweet orange and yuzu showed a slight immunoreac- 529—537 (2003). tion. So far, there has been no report of the isolation of the