Screening Method for the Discovery of Potential ... - ACS Publications

Sep 3, 2005 - Test extracts found to form adducts with Keap1 were then incubated with the ... As a result, Nrf2 is sequestered in the cytoplasm by Kea...
0 downloads 0 Views 167KB Size
Download PDF
Anal. Chem. 2005, 77, 6407-6414

Screening Method for the Discovery of Potential Cancer Chemoprevention Agents Based on Mass Spectrometric Detection of Alkylated Keap1 Guowen Liu,† Aimee L. Eggler,† Birgit M. Dietz,† Andrew D. Mesecar,† Judy L. Bolton,† John M. Pezzuto,‡ and Richard B. van Breemen*,†

Department of Medicinal Chemistry and Pharmacognosy, University of Illinois College of Pharmacy, 833 South Wood Street, Chicago, Illinois 60612-7231, and Medicinal Chemistry and Molecular Pharmacology, School of Pharmacy, Nursing and Health Sciences, Purdue University, 1330 Heine Pharmacy Building, West Lafayette, Indiana 47907

Natural products are important sources of drugs such as cancer chemopreventive agents, but most assays for the discovery of compounds in natural product extracts are low throughput and provide little information about lead compounds in these complex mixtures. The induction of enzymes such as quinone reductase, glucuronyl transferases, glutathione S-transferases, and sulfotransferases can protect cells against the toxic and neoplastic effects of carcinogens. An increase in the concentration of Nrf2 in the nucleus of a cell upregulates the antioxidant response element and induces the expression of these chemopreventive enzymes. Based on the hypothesis that ubiquitination and proteosome-mediated degradation of Nrf2 in the cytoplasm decreases upon the covalent modification of 1 or more of the 27 cysteine sulfhydryl groups on Keap1 (a protein that sequesters Nrf2 in the cytoplasm) and results in higher Nrf2 levels both in the cytoplasm and in the nucleus, a high-throughput mass spectrometry-based screening assay was designed to detect alkylation of sulfhydryl groups of human Keap1. As an initial high-throughput screening step, matrixassisted laser desorption time-of-flight mass spectrometry was used to determine whether incubation of Keap1 with a botanical sample produced adducts of Keap1. Test extracts found to form adducts with Keap1 were then incubated with the alternative biological nucleophile glutathione and characterized using LC-UV-MS-MS. After validation of the assay using two model alkylating agents, fractions of an extract of hops (Humulus lupulus L.) from the brewing industry were screened, and several compounds were detected as potential chemopreventive agents. Two of these electrophilic hops constituents were identified as xanthohumol and xanthohumol D. In a subsequent cell-based assay, xanthohumol and xanthohumol D were confirmed to be potent inducers of quinone reductase, and reaction with Keap1 was also confirmed. Therefore, this new mass spectrometric screening assay was dem10.1021/ac050892r CCC: $30.25 Published on Web 09/03/2005

© 2005 American Chemical Society

onstrated to facilitate the discovery of chemoprevention agents in complex natural product mixtures. Many carcinogens are activated via phase I reactions to electrophilic intermediates that can attack macromolecules such as proteins and DNA. These reactive metabolites can be deactivated through either phase II conjugation reactions, such as enzymatic sulfation and glucuronidation, conjugation with glutathione (GSH), or also reduction. Therefore, cells, tissues, organs, etc., that express high levels of these detoxification enzymes are more resistant to damage by electrophilic compounds.1,2 One strategy for cancer chemoprevention focuses on the use of natural or synthetic agents to modulate the metabolism and disposition of endogenous and environmental carcinogens through upregulation of phase II enzymes.3-5 Since detoxification enzymes are not necessarily expressed or function at maximal capacity, their induction should be an effective strategy for cancer chemoprevention. In support of this hypothesis, overexpression of the phase II enzyme GST P1 in human prostate cells prevented cytotoxicity and DNA damage by the heterocyclic amine carcinogen PhIP.6 In addition, deficiency of quinone reductase I increased the susceptibility of mice to benzo[a]pyrene-induced skin carcinogenesis.7 Nuclear accumulation of the transcription factor Nrf2 is essential for the antioxidant responsive element (ARE)-mediated induction of genes coding for phase II detoxifying and antioxidative stress enzymes. Under basal conditions, Nrf2 is associated with the cytoplasmic actin-binding protein Keap1. As a result, Nrf2 is sequestered in the cytoplasm by Keap1.8-10 * To whom correspondence should be addressed. Telephone: (312)-996-9353. Fax: (312) 996-7107. E-mail: [email protected]. † University of Illinois. ‡ Purdue University. (1) Talalay, P. Biofactors 2000, 12, 5-11. (2) Talalay, P.; De Long, M. J.; Prochaska, H. J. Proc. Natl. Acad. Sci. U.S.A. 1988, 85, 8261-8265. (3) Wattenberg, L. W. Cancer Res. 1985, 45, 1-8. (4) Hong, W. K.; Sporn, M. B. Science 1997, 278, 1073-1077. (5) Talalay, P. Proc. Am. Philos. Soc. 1999, 143, 52-72. (6) Nelson, C. P.; Kidd, L. C.; Sauvageot, J.; Isaacs, W. B.; De Marzo, A. M.; Groopman, J. D.; Nelson, W. G.; Kensler, T. W. Cancer Res. 2001, 61, 103109. (7) Long, D. J., 2nd. Cancer Res. 2000, 60, 5913-5915.

Analytical Chemistry, Vol. 77, No. 19, October 1, 2005 6407

There are 27 cysteine residues in human Keap1, and alkylation of one or more of these cysteine sulfhydryl groups leads to nuclear accumulation of Nrf2 through mechanisms that are still only partially understood11-15 but appear to at least involve regulation of Nrf2 ubiquitination.11,15 Cysteine residues in Keap1 protein appear to function as sensors for induction of the ARE. Three cysteine to serine mutants of Keap1 affect Nrf2 nuclear accumulation and ARE signaling by electrophiles (C151, C273, C288).11 In addition, the ability of agents to modify Keap1 correlates with their ability to induce phase II enzymes.12 Furthermore, this function is consistent with the observation that, among nine different structural classes of monofunctional phase II inducers, all are electrophilic and share the capacity to react with sulfhydryl groups.2 Therefore, to facilitate the search for chemoprevention agents that prevent sequestration of Nrf2 in the cytoplasm, we designed a screening assay for the discovery of xenobiotic compounds that modify Keap1. Currently, in vitro screening assays of quinone reductase induction,16,17 ornithine decarboxylase inhibition,18,19 and cyclooxygenase inhibition20 are popular for bioassay-guided fractionation of natural product extracts for the discovery of potential chemoprevention agents. However, assays based on fractionation require multiple iterations to isolate active constituents and must be accompanied by conventional structure elucidation analyses such as NMR, spectrophotometry, and mass spectrometry. These procedures are low throughput and labor intensive. To enhance the throughput of these chemopreventive agent discovery assays, our new mass spectrometry-based assay provides, in the first step, identification of extracts that contain compounds that react with Keap1. Since our in vitro assay uses purified, recombinant human Keap1 protein, it is quicker and therefore higher in throughput than cell-based assays such as the quinone reductase induction assay. In a second level of analysis, structural data are obtained by using LC-UV-MS-MS in a variation of our previously published assay for the identification of compounds that react with glutathione.21 Recently, some of the flavonoids and chalcones contained in hops were reported to have potential anticancer or cancer chemoprevention activities.22,23 In particular, xanthohumol, a major (8) Hayes, J. D.; McMahon, M. Cancer Lett. 2001, 174, 103-113. (9) Itoh, K.; Wakabayashi, N.; Katoh, Y.; Ishii, T.; Igarashi, K.; Engel, J. D.; Yamamoto, M. Genes Dev. 1999, 13, 76-86. (10) Zipper, L. M.; Mulcahy, R. T. J. Biol. Chem. 2002, 277, 36544-36552. (11) Zhang, D. D.; Hannink, M. Mol. Cell Biol. 2003, 23, 8137-8151. (12) Dinkova-Kostova, A. T.; Holtzclaw, W. D.; Cole, R. N.; Itoh, K.; Wakabayashi, N.; Katoh, Y.; Yamamoto, M.; Talalay, P. Proc. Natl. Acad. Sci. U.S.A. 2002, 99, 11908-11913. (13) Kwak, M. K.; Wakabayashi, N.; Kensler, T. W. Mutat. Res. 2004, 555, 133148. (14) Wakabayashi, N.; Dinkova-Kostova, A. T.; Holtzclaw, W. D.; Kang, M. I.; Kobayashi, A.; Yamamoto, M.; Kensler, T. W.; Talalay, P. Proc. Natl. Acad. Sci. U.S.A. 2004, 101, 2040-2045. (15) Zhang, D. D.; Lo, S. C.; Cross, J. V.; Templeton, D. J.; Hannink, M. Mol. Cell. Biol. 2004, 24, 10941-10953. (16) Prochaska, H. J.; Santamaria, A. B. Anal. Biochem. 1988, 169, 328-336. (17) Su, B. N.; Cuendet, M., Farnsworth, N. R.; Fong, H. H. S.; Pezzuto, J. M. and Kinghorn, A. D. Planta Med. 2002, 68, 1125-1128. (18) Lee, S. K.; Pezzuto, J. M. Arch. Pharm. Res. 1999, 22, 559-564. (19) Chang, L. C.; Song, L. L.; Jung Park, E.; Luyengi, L.; Lee, K. J.; Farnsworth, N. R.; Pezzuto, J. M.; Kinghorn, A. D. J. Nat. Prod. 2000, 63, 1235-1238. (20) Su, B. N.; Jung Park, E.; Vigo, J. S.; Graham, J. G.; Cabieses, F.; Fong, H. H. S.; Pezzuto, J. M.; Kinghorn, A. D. Phytochemistry 2003, 63, 335-341. (21) Johnson, B. M.; Bolton, J. L.; van Breemen, R. B. Chem. Res. Toxicol. 2001, 14, 1546-1551.

6408

Analytical Chemistry, Vol. 77, No. 19, October 1, 2005

prenylated flavonoid in hops and beer, was shown to be a cancer chemoprevention agent through the induction of quinone reductase.24 Humulus lupulus L. (hops) is a climbing perennial plant native to Eurasia and North America. The best known use of the female flowers or strobiles of this plant is in the brewing of beer.25 Historically, hops strobiles have also been used as a mild sedative to treat insomnia in Europe, China, India, and North America.26 Therefore, as an application of our new assay, fractions of spent hops pellets from the brewing industry were screened for potential cancer chemoprevention agents. EXPERIMENTAL PROCEDURES Materials and Reagents. Recombinant human Keap1 protein containing a histidine tag was expressed in and purified from Escherichia coli. The details of the cloning, expression, and purification of Keap1 are published elsewhere.27 GSH, menadione, 1-chloro-2,4-dinitrobenzene (CDNB), 3,5-dimethoxy-4-hydroxycinnamic acid, and R-cyano-4-hydroxycinnamic acid were purchased from Aldrich-Sigma (Milwaukee, WI). Sequencing grade trypsin was purchased from Promega (Madison, WI). Double-deionized water was prepared using a Millipore (Bedford, MA) Milli-Q system. Methanol and acetonitrile (Optima grade) were purchased from Fisher Scientific (Hanover Park, IL). Spent Nugget hops pellets (plant material remaining after supercritical CO2 extraction of pelletized strobuli of H. lupulus cv. Nugget) were obtained from Yakima Chief (lot PE-MANUO04; Sunnyside, WA). A methanol extract of spent hops pellets, a CHCl3 partition of the methanol extract, and authentic standards of the hops compounds xanthohumol, isoxanthohumol, and xanthohumol D were obtained as a gift from Dr. Luke R. Chadwick and Dr. Guido F. Pauli of the University of Illinois College of Pharmacy and were prepared as described previously.28 Sample Preparation. Menadione (a Michael addition acceptor) and CDNB (an electrophile) were used as model compounds for method development and validation. Stock solutions of 10 mM of each compound were prepared in methanol. Menadione and CDNB at 200 µM were incubated individually for 2 h at 37 °C with 5 µM Keap1 and 200 µM tris(2-carboxyethyl)phosphine to keep the reactive cysteines in the reduced and reactive state or with 1 mM GSH in a total volume of 200 µL of 25 mM Tris-HCl buffer (pH 8.0). Alternatively, a methanol extract of spent hops pellets or a chloroform partition of the methanol extract were incubated at 100 µg/mL with 5 µM Keap1 in a total volume of 100 µL of Tris-HCl buffer. An identical control experiment was carried out except that the Keap1 protein was incubated with the methanol solvent only instead of a methanolic solution of menadione, xanthohumol, or a hops extract. For additional characteriza(22) Henderson, M. C.; Miranda, C. L.; Stevens, J. F.; Deinzer, M. L.; Buhler, D. R. Xenobiotica 2000, 30, 235-251. (23) Miranda, C. L.; Stevens, J. F.; Helmrich, A.; Henderson, M. C.; Rodriguez, R. J.; Yang, Y. H.; Deinzer, M. L.; Barnes, D. W.; Buhler, D. R. Food Chem. Toxicol. 1999, 37, 271-285. (24) Miranda, C. L.; Aponso, G. L.; Stevens, J. F.; Deinzer, M. L.; Buhler, D. R. Cancer Lett. 2000, 149, 21-29. (25) Stevens, J. F.; Taylor, A. W.; Clawson, J. E.; Deinzer, M. L. J. Agric. Food Chem. 1999, 47, 2421-2428. (26) Fussel, A.; Wolf, A.; Brattstrom, A. Eur. J. Med. Res. 2000, 5, 385-390. (27) Eggler, A. L.; Liu, G.; Pezzuto, J. M.; van Breemen, R. B.; Mesecar, A. D. Proc. Natl. Acad. Sci. U.S.A. 2005, 102, 10070-10075. (28) Chadwick, L. R.; Nikolic, D.; Burdette, J. E.; OverK, C. R.; Bolton, J. L.; van Breemen, R. B.; Frohlich, R.; Fong, H. H. S.; Farnsworth, N. R.; Pauli, G. F. J. Nat. Prod. 2004, 67, 2024-2032.

tion of alkylated Keap1, 50 µg of untreated Keap1 or Keap1 that had been incubated with CDNB were digested at 37 °C overnight using 2 µg of trypsin in Tris-HCl buffer (pH 8.0) in a total volume of 100 µL. Quinone Reductase Induction Assay. Hepa 1c1c7 murine hepatoma cells were supplied by Dr. J. P. Whitlock, Jr., of Stanford University (Stanford, CA). The cells were maintained in minimum essential medium supplemented with 1% penicillin-streptomycin, 10% fetal bovine serum (Atlanta Biologicals, Atlanta, GA) and incubated in 5% CO2 at 37 °C. Induction of quinone reductase activity was assessed using Hepa 1c1c7 murine hepatoma cells as described previously with minor modifications.16 Briefly, Hepa lclc7 cells were seeded in 96-well plates at a density of 1.25 × 104 cells/mL in 190 µL of media. After incubating for 24 h, the test samples (compounds or hops extracts) were added to each well, and the cells were incubated for an additional 48 h. Next, the cell culture medium was decanted, and the cells were incubated at 37 °C for 10 min with 50 µL of 0.8% digitonin and 2 mM EDTA at pH 7.8. The plates were agitated on an orbital shaker (100 rpm) for 10 min at room temperature, and then 200 µL of a solution containing bovine serum albumin (0.67 mg/mL), 3-(4,5-dimethylthiazo-2-yl)-2,5-diphenyltetrazolium bromide (MTT) (25 mM), 0.5 M Tris-HCl, 0.01% Tween 20, 5 µM FAD, 1 mM glucose 6-phosphate, 30 µM NADP, 2 units/mL glucose-6-phosphate dehydrogenase, and 50 µM menadione were added to each well. After 5 min, the absorbance of each well was measured at 595 nm. The specific activity of quinone reductase was determined by measuring NADPH-dependent menadiol-mediated reduction of MTT to blue formazan. Induction of quinone reductase activity was determined by comparing the specific activities of quinone reductase in sample-treated and solvent-treated cells (control). The inducer concentration of each isolated compound or extract required to double quinone reductase activity (defined as the CD value) was determined. Matrix-Assisted Laser Desorption/Ionization Time-ofFlight (MALDI TOF) Mass Spectrometry. In preparation for analysis using MALDI TOF mass spectrometry, a 1.0-µL aliquot of a matrix solution consisting of 3,5-methoxy-4-hydroxycinnamic acid (for proteins) or R-cyano-4-hydroxycinnamic acid (for peptides) in acetonitrile/water/trifluoroacetic acid (50:49.9:0.1, v/v/ v) at a concentration of 10 mg/mL was mixed with 1.0 µL of the intact Keap1 protein solution or its tryptic digest. Immediately before analysis, 0.5 µL of the mixture was spotted onto the MALDI sample stage and air-dried. Positive ion MALDI TOF mass spectra were acquired over the range m/z 50 000-90 000 for proteins or m/z 800-4000 for peptides using an Applied Biosystems (Foster City, CA) Voyager DE-Pro MALDI TOF mass spectrometer, which was operated in linear mode for proteins or reflector mode for peptides. A total of 600 laser shots were acquired and signal averaged per mass spectrum. After time-delayed extraction, the ions were accelerated to 20 kV for TOF mass spectrometric analysis. LC-UV-MS-MS. After incubation of test compounds or hops extracts with GSH, the reaction product mixtures were analyzed using high performance liquid chromatography with in-line ultraviolet absorbance and tandem mass spectrometric detection (LC-UV-MS-MS) using a Thermo Electron (San Jose, CA) Surveyor HPLC system equipped with a Phenomenex (Torrance, CA)

C18 column (5 µm, 250 × 2.0 mm), a Thermo Electron UV absorbance photodiode array detector, and a Thermo Electron TSQuantum triple quadrupole mass spectrometer. A 20-µL aliquot of each GSH reaction mixture was injected onto the LC-UV-MSMS per analysis. For analyses using positive ion electrospray mass spectrometry, the mobile phase consisted of a 15-min linear gradient from 95% water (containing 0.1% formic acid) to 80% acetonitrile. For negative ion electrospray MS-MS, a 15-min gradient was used from 95% water to 80% methanol. The column eluate was diverted to waste during the first 5 min so that salts and unreacted GSH did not enter the mass spectrometer. The flow rate of the mobile phase was 200 µL/min. As each GSH conjugate eluted from the HPLC column, the UV spectrum was recorded over the range 200-500 nm. Next, collision-induced dissociation (CID) and tandem mass spectrometry were used with precursor ion scanning for the selective detection of GSH adducts, and then CID of the GSH adduct ions was used with product ion scanning to obtain structural information. Specifically, positive ion electrospray with CID and precursor ion scanning of the GSH b1-type fragment ion at m/z 130 was used to identify peaks corresponding to GSH adducts as described in our previous publication.21 Negative ion electrospray, CID, and precursor ion scanning were used to identify the deprotonated molecules of GSH adducts that fragmented to form ions of m/z 272 as described by Dieckhaus et al.29 After the molecular weight of each GSH adduct was determined using LC-UV-MS-MS with precursor ion scanning, product ion MS-MS analyses were carried out on each adduct to obtain structural information. The collision energy for all MS-MS experiments was 22 eV in positive mode and 20 eV in negative ion mode. Argon was used for CID at a pressure of 1.5 mTorr. RESULTS AND DISCUSSION CDNB, an electrophile containing a chloride leaving group, and menadione, a Michael addition acceptor, were chosen as model compounds for alkylation of the sulfhydryl groups of Keap1 or GSH via either nucleophilic substitution or addition reactions. Under the reaction conditions described above, incubation of Keap1 with either CDNB or menadione resulted in an increase in the molecular weight of the protein consistent with an average of 11 molecules of CDNB or 10 molecules of menadione per molecule of Keap1 (see mass spectra in Figure 1). Human Keap1 contains 27 cysteine residues, and since none are known to participate in intramolecular disulfide linkages, most of these sulfhydryl side chains are probably available for reaction with electrophiles. Therefore, CDNB and menadione probably alkylated cysteine sulfhydryl groups of Keap1. To confirm that cysteine residues were the primary sites of alkylation, unmodified Keap1 and Keap1 that had been incubated with CDNB were digested with trypsin, and the tryptic peptides were analyzed using MALDI TOF mass spectrometry. As shown in Figure 2, new peptide signals were detected in the tryptic digest of Keap1 that had been incubated with CDNB. As expected, these new signals corresponded to cysteine-containing peptides that had been alkylated by CDNB. After detecting Keap1 modified by the electrophiles menadione or CDNB using MALDI TOF mass spectrometry, the utility of (29) Dieckhaus, C. M.; Fernandez-Metzler, C. L.; King, R.; Krolikowski, P. H.; Baillie, T. A. Chem. Res. Toxicol. 2005, 18, 630-638.

Analytical Chemistry, Vol. 77, No. 19, October 1, 2005

6409

Figure 1. Positive ion MALDI-TOF mass spectra of Keap1 following incubation with (A) CDNB, (B) menadione, (C) a chloroform partition of a methanol extract of spent hop pellets, or (D) a methanol extract of spent hop pellets. The dashed lines indicate Keap1 incubation with buffer (control), and the solid lines represent Keap1 incubated with the test compound or hops extract.

Figure 2. Positive ion MALDI TOF mass spectra of tryptic peptides of Keap1 obtained using a matrix consisting of 10 mg/mL R-cyano-4hydroxycinnamic acid in 50% methanol with 0.1% TFA. (A) Tryptic peptides of untreated Keap1 and (B) tryptic digest of Keap1 after incubation with a 20-fold excess of CDNB for 2 h at room temperature. Note that the new peaks, labeled with an asterisk (/), correspond to cysteinecontaining peptides modified by CDNB.

this approach for the screening of complex natural product extracts for the presence of electrophilic chemoprevention agents was investigated by screening hops extracts. Two hops preparations were screened, a methanol extract of spent hops pellets and a chloroform partition fraction of the methanol extract. After incubation of each hops extract with Keap1, positive ion MALDI TOF mass spectrometry was used to test for any increase in the mass of Keap1. As shown in Figure 1 (C and D), the mass of Keap1 increased 541 ( 25 Da compared to control as a result of incubation with the chloroform partition of the hops methanolic extract. In comparison, a much smaller mass increase of only 143 ( 23 Da was observed for Keap1 incubated with the methanol extract. These results suggest that the chloroform partition of the hops extract contained either more concentrated or more reactive 6410

Analytical Chemistry, Vol. 77, No. 19, October 1, 2005

alkylating agents and, as a result, should be more effective than the methanol extract in upregulating the ARE. The utility of this Keap1 screening assay in determining whether botanical extracts, such as hops fractions, contain electrophilic species that can upregulate the ARE was confirmed by measuring the induction of quinone reductase in a cell-based system. The chloroform partition of the methanol extract showed a CD (concentration required to double quinone reductase activity) of 3.24 ( 0.33 µg/mL whereas the CD of the methanol extract exceeded 20 µg/mL (for additional information regarding quinone reductase induction by hops extracts, see Dietz et al.30). (30) Dietz, B. M.; Kang, Y. H.; Liu G.; Eggler, A. L.; Yao, P.; Chadwick, L. R.; Pauli, G. F.; Farnsworth, N. R.; Mesecar, A. D.; van Breemen, R. B.; Bolton, J. L. Chem. Res. Toxicol. 2005, 18, 1296-1305.

Therefore, both the Keap1 alkylation assay and the quinone reductase induction assay indicated that the chloroform partition of the methanolic hops extract should be more effective in chemoprotection than the methanolic extract. These data are consistent with the hypothesis that alkylation of Keap1 by electrophilic compounds is indicative of the ability of these compounds to upregulate the ARE and induce enzymes such as quinone reductase. Although screening complex mixtures to determine whether they contain compounds that can alkylate Keap1 is a rapid first step toward discovering new cancer chemoprevention agents, subsequent analyses are still necessary to identify these electrophilic constituents. The elucidation of the structures of the Keap1 adducts would be slow and labor intensive due to the complexity and high molecular weight of the protein. Therefore, after the initial screening using Keap1 and MALDI TOF mass spectrometry, a simpler approach was used to identify the electrophilic species contained in mixtures based on our previously published toxicity screening assay for the identification of electrophilic metabolites that form GSH conjugates.21 Since monofunctional phase II inducers such as those in hops react with sulfhydryl groups in Keap1, they should also react with the sulhydryl group of GSH. The only modification of our previous screening assay based on GSH alkylation that was implemented was the elimination of the metabolic activation step, since the electrophilic compounds CDNB, menadione, and those in hops reacted with Keap1 without metabolic activation. Highly reactive compounds (such as the quinone imine metabolites of acetaminophen31) are hepatotoxic and can be deadly at high concentrations. However, the weak electrophiles that are the targets of this Keap1 screening assay should not be toxic at therapeutic or dietary concentrations. Strongly electrophilic compounds are unlikely to exist in natural product extracts such as these hops preparations. If such cytotoxic species had been present, they would have been quenched through reaction with weak nucleophiles in the extract. To test the performance of the GSH-based secondary screening assay, CDNB and menadione were incubated separately with GSH, and then LC-MS-MS precursor ion scanning was used to screen for GSH adducts. Precursor ions were detected corresponding to GSH adducts that fragmented to form product ions of m/z 130 during positive ion electrospray21 or to form product ions of m/z 272 in negative ion mode.29 Both of these product ions are characteristic of GSH adducts. Subsequent product ion LC-MSMS analyses of the protonated or deprotonated GSH adducts were then carried out to characterize the GSH conjugates of each model electrophile. For example, the negative ion product ion tandem mass spectrum of menadione is shown in Figure 3. Although the GSH adducts studied in this investigation formed product ions of m/z 272 that were more abundant than the positively charged GSH fragment ions of m/z 130 (which is consistent with the report of Dieckhaus et al.29), both negative and positive ion product ion tandem mass spectra were used to provide complementary structural information concerning the GSH conjugates. Next, the hops chloroform partition was analyzed using the LC-MS-MS screening assay for GSH conjugates. As shown in

Figure 4, the deprotonated molecules of GSH conjugates were detected at m/z 660 (one peak) and 676 (multiple peaks). Analysis of the same reaction mixture using positive ion electrospray confirmed these results. Most of the known monofunctional phase II inducers form GSH conjugates through addition reactions.32 Assuming that the electrophilic hops constituents also reacted with GSH through addition reactions, the molecular masses of these compounds before conjugation were predicted to be 354 and 370 Da. Therefore, negative ion electrospray LC-MS analysis was carried out to determine whether any of these precursor compounds were present in the chloroform partition of the methanol extract of spent hops. As shown in Figure 5, the deprotonated molecules of two isomers of m/z 353 were detected at retention times of 25.9 and 30.3 min, and at least nine compounds of m/z 369 were detected eluting from 22.1 to 26.9 min. To ascertain which of these peaks

(31) Bulera, S. J.; Birge, R. B.; Cohen, S. D.; Khairallah, E. A. Toxicol. Appl. Pharmacol. 1995, 134, 313-320.

(32) Talalay, P.; Dinkova-Kostova, A. T.; Holtzclaw, W. D. Adv. Enzyme Regul. 2003, 43, 121-134.

Figure 3. Negative ion electrospray product ion tandem mass spectrum of the GSH adduct of menadione. The base peak of the mass spectrum corresponds to the GSH product ion of m/z 272. Note that no reduced form of the menadione-glutathione conjugate was detected at m/z 478.

Figure 4. Negative ion electrospray LC-MS-MS analyses of the chloroform partition of a methanol extract of spent hops after incubation with GSH for 2 h at 37 °C. (A) Precursor ion chromatogram showing compounds that fragmented to form the GSH product ion of m/z 272. Note that peak eluting at 6.6 min corresponds to a GSH dimer. (B) Computer-reconstructed precursor ion chromatogram showing signals for m/z 272 that were formed from m/z 660. Note that one intense peak was detected at 24.6 min. (C) Computerreconstructed precursor ion chromatogram showing signals for m/z 272 that were formed from m/z 676. In this chromatogram, multiple peaks were detected with one intense peak eluting at 23.0 min.

Analytical Chemistry, Vol. 77, No. 19, October 1, 2005

6411

Figure 5. Negative ion electrospray LC-MS selected ion chromatograms, (A) m/z 353 and (B) 369, of the chloroform partition of a methanol extract of spent hops after incubation with vehicle (solid line) or Keap1 (dashed line). Peaks labeled with an asterisk (*) represent compounds that reacted with Keap1 so that the intensity of these peaks diminished after incubation.

represent electrophilic compounds that might be responsible for induction of the ARE, an aliquot of the chloroform partition was incubated with Keap1, analyzed using LC-MS, and compared with an aliquot incubated only with solvent (see Figure 5). The intensities of the peaks corresponding to the electrophilic compounds were diminished in the chromatogram of the Keap1incubated fraction, since the majority of these compounds had been consumed through formation of covalent adducts with Keap1. Specifically, the compound of m/z 353 eluting at 30.3 min (but not the compound eluting at 25.9 min) was identified as a Keap1 alkylating agent. Among the compounds forming deprotonated molecules of m/z 369, those eluting at 22.1, 22.8, 23.4, 26.2, and 26.9 min alkylated Keap1 (see labeled peaks in Figure 5).

The NAPRALERT33 database was searched for hops constituents with molecular masses of 354 Da, and xanthohumol and isoxanthohumol were identified as candidates for the unknown electrophilic compounds eluting at 25.9 and 30.3 min in Figure 5.34,35 By comparing the tandem mass spectra (Figure 6), UV spectra (Figure 6), and HPLC retention times of the hops constituents weighing 354 Da with xanthohumol and isoxanthohumol standards, the peak eluting at 25.9 min was identified as isoxanthohumol, and the peak with a retention time of 30.3 min was identified as the electrophile xanthohumol. To confirm that xanthohumol but not isoxanthohumol could alkylate Keap1, each compound was incubated with Keap1 and then analyzed using MALDI TOF mass spectrometry. The structures of xanthohumol and isoxanthohumol and the mass spectra of Keap1 after incubation with these compounds are shown in Figure 7. Upon incubation with xanthohumol, the mass of Keap1 increased by more than 2000 Da, indicating alkylation of multiple cysteine sulfhydryl groups by xanthohumol. In contrast, the mass of Keap1 did not change significantly following incubation with isoxanthohumol. As additional confirmation that xanthohumol but not isoxanthohumol are electrophilic, each compound was incubated with GSH. Only xanthohumol reacted with GSH, and an adduct was detected at m/z 660 (data not shown). Comparing the structures of xanthohumol and isoxanthohumol (see Figure 7), only xanthohumol has a Michael addition center for reaction with GSH or the sulfhydryl groups of Keap1, which can explain the selective reactivity of xanthohumol. The significance of natural products such as xanthohumol that can alkylate Keap1 is that they might function as chemoprotective agents by inducing protective enzymes regulated by the ARE. To verify that xanthohumol can induce such protective enzymes, a quinone reductase induction assay was carried out using Hepa

Figure 6. Negative ion electrospray product ion tandem mass spectra of the deprotonated molecules of m/z 353 (A-D) and the corresponding UV spectra (E-H) for the constituents in the chloroform partition of a methanol extract of spent hops eluting at 25.9 and 30.3 min in Figure 5 and the reference compounds isoxanthohumol and isoxanthohumol. (A, E) Isoxanthohumol standard; (B, F) hops constituent with a retention time of 25.9 min; (C, G) xanthohumol standard; (D, H) hops constituent with a retention time of 30.3 min. Based on these data and coelution of the standards with the unknowns, the unknown hops constituents were identified as isoxanthohumol (retention time 25.9 min) and xanthohumol (retention time 30.3 min). 6412 Analytical Chemistry, Vol. 77, No. 19, October 1, 2005

Figure 7. Positive ion MALDI-TOF mass spectrum of Keap1 incubated with (A) xanthohumol or (B) isoxanthohumol. Note that xanthohumol alkylated Keap1 multiple times to increase the mass of the protein by almost 2000 Da, whereas no reaction occurred during incubation with isoxanthohumol (the small change in mass of Keap1 is approximately equal to attachment of a sodium cation).

Figure 8. UV spectra (A, B) and negative ion electrospray product ion tandem mass spectra (C-H) of the reference compound xanthohumol D and constituents in the chloroform partition of a methanol extract of spent hops forming deprotonated molecules of m/z 369. (A, C) Xanthohumol D standard; (B, D) hops constituent eluting at 26.9 min (see Figure 5); (E-H) hops constituents with retention times (see Figure 5) of 22.1, 22.8, 23.4 and 26.2, respectively. Note that the concentrations of the compounds eluting at 22.1, 22.8, 23.4, and 26.2 min were insufficient to obtain UV spectra. The compound eluting at 26.9 min was identified as xanthohumol D.

1c1c7 murine hepatoma cells. In this cell-based system, xanthohumol was found to be a potent quinone reductase inducer with a CD value of 1.7 ( 0.7 µg/mL.30 These results are consistent with a literature CD value of 2.1 µM for xanthohumol.24 As expected, isoxanthohumol was not active in the quinone reductase induction assay and did not double the quinone reductase activity even at concentrations as high as 50 µg/mL.30 In addition to the detection of a peak of m/z 353 in Figure 5 (later identified as xanthohumol) corresponding to a hops (33) Loub, W. D.; Farnsworth, N. R.; Soejarto, D. D.; Quinn, M. L. J. Chem. Inf. Comput. Sci. 1985, 25, 99-103. (34) Verzele, M. S., J.; Fontijn, F.; Anteunis, M. Bull. Soc. Chim. Belg. 1957, 66, 452-475. (35) Gerhaeuser, C. A., A. P.; Klimo, K.; Knauft, J.; Frank, N.; Becker, H. Phytochem. Rev. 2003, 1, 369-377.

constituent that can alkylate Keap1, several peaks of m/z 369 were detected corresponding to additional hops compounds that can alkylate Keap1. The UV spectra and negative ion product ion tandem mass spectra of these compounds are shown in Figure 8. Based on searches of the botanical natural products literature using resources such as NAPRALERT, four isomeric hops constituents were identified that are consistent with these peaks of m/z 369 in Figure 5 including xanthohumol D, xanthohumol B, xanthohumol I, and trans-hydroxyxanthohumol. Like xanthohumol, these compounds are Michael acceptors that can alkylate Keap1. The compound corresponding to the most intense of these peaks of m/z 369 (retention time of 26.9 min) in Figure 5 was identified as xanthohumol D by comparison of its UV spectrum, tandem mass spectrum, and HPLC retention time with an Analytical Chemistry, Vol. 77, No. 19, October 1, 2005

6413

authentic standard (see Figure 8). Next, xanthohumol D was assayed for induction of quinone reductase using Hepa 1c1c7 murine hepatoma cells to verify that it can upregulate enzymes of the ARE. In this assay, the CD value of xanthohumol D was 7.4 ( 0.7 µM, which was not as potent as xanthohumol. No standards corresponding to other potential hops constituents such as xanthohumol B and xanthohumol I were available to confirm the structures of the Keap1 alklyating agents eluting at 22.1, 22.8, 23.8, and 26.2 min (see Figure 5). However, the HPLC retention times and tandem mass spectra shown in Figures 5 and 8 will be helpful for the identification of these compounds when standards become available. CONCLUSIONS A screening assay has been developed for the identification of potential cancer chemoprevention agents in complex mixtures such as natural product extracts. This assay is based on the mass spectrometric identification of compounds that alkylate Keap1 leading to the upregulation of the ARE by Nrf2. This assay was validated using a conventional quinone reductase induction assay in Hepa 1c1c7 murine hepatoma cells, which is a much slower assay (days compared to minutes) than this new mass spectrometrybased approach. Although electrophiles such as carbocations, quinones, quinone methides, quinone imines, epoxides, and Michael acceptors should all react with Keap1 and potentially be cancer chemoprevention agents, the most reactive of these compounds should be expected to cause toxicity (especially at high doses). Therefore,

6414

Analytical Chemistry, Vol. 77, No. 19, October 1, 2005

only weak electrophiles such as the Michael acceptor xanthohumol should show cancer chemoprevention activity over a wide range of doses without showing toxicity. Candidates for such cancer chemoprevention agents might therefore be expected to exist in biological systems without reacting with water or other weak nucleophiles. Furthermore, the screening assay described here has been designed to identify chemoprevention agents in natural product extracts that do not require metabolic activation. ACKNOWLEDGMENT This work was supported by grants P01 CA48112 from the National Cancer Institute and P50 AT00155 provided jointly by the National Center for Complementary and Alternative Medicine (NCCAM), the Office of Dietary Supplements (ODS), the Office for Research on Women’s Health (ORWH), and the National Institute of General Medicine (NIGMS) of the National Institutes of Health (NIH). The contents of this paper are solely the responsibility of the authors and do not necessarily represent the official views of the NCCAM, NCI, ODS, ORWH, NIGMS, or the NIH. The authors thank Dr. Lucas R. Chadwick and Dr. Guido F. Pauli for the hops extract and standard compounds of xanthohumol, isoxanthohumol, and xanthohumol D.

Received for review May 20, 2005. Accepted August 4, 2005. AC050892R