Cytochrome P450 3A4-Mediated Bioactivation of Raloxifene

R. Scott Obach, Amit S. Kalgutkar, John R. Soglia and Sabrina X. Zhao. Can In Vitro ... Hong Liu, Zhihui Qin, Gregory R. J. Thatcher and Judy L. Bolto...
0 downloads 0 Views 104KB Size
Chem. Res. Toxicol. 2002, 15, 907-914

907

Cytochrome P450 3A4-Mediated Bioactivation of Raloxifene: Irreversible Enzyme Inhibition and Thiol Adduct Formation Qing Chen,† Jason S. Ngui,† George A. Doss,† Regina W. Wang,† Xiaoxin Cai,† Frank P. DiNinno,‡ Timothy A. Blizzard,‡ Milton L. Hammond,‡ Ralph A. Stearns,† David C. Evans,† Thomas A. Baillie,† and Wei Tang*,† Departments of Drug Metabolism and Medicinal Chemistry, Merck Research Laboratories, Rahway, New Jersey 07065 Received February 6, 2002

Raloxifene is a selective estrogen receptor modulator which is effective in the treatment of osteoporosis in postmenopausal women. We report herein that cytochrome P450 (P450)3A4 is inhibited by raloxifene in human liver microsomal incubations. The nature of the inhibition was irreversible and was NADPH- and preincubation time-dependent, with KI and kinact values estimated at 9.9 µM and 0.16 min-1, respectively. The observed loss of P450 3A4 activity was attenuated partially by glutathione (GSH), implying the involvement of a reactive metabolite(s) in the inactivation process. Subsequently, GSH adducts of raloxifene were identified in incubations with human liver microsomes; substitution with GSH occurred at the 5- or 7-position of the benzothiophene moiety or at the 3′-position of the phenol ring, with the 7-glutathionyl derivative being most abundant based on LC/MS and NMR analyses. These adducts are postulated to derive from addition of GSH to raloxifene arene oxides followed by dehydration and aromatization. Alternatively, raloxifene may be oxidized to an extended quinone intermediate, which then is trapped by GSH conjugation. The bioactivation of raloxifene most likely is catalyzed by P450 3A4, since the formation of GSH adducts was almost abolished when liver microsomes were pretreated with ketoconazole or with an inhibitory anti-P450 3A4 IgG. The GSH adducts also were detected in incubations of raloxifene with rat or human hepatocytes, while the corresponding N-acetylcysteine adducts were identified in the bile and urine from rats treated orally with the drug at 5 mg/kg. Taken together, these data indicate that P450 3A4-mediated bioactivation of raloxifene in vitro is accompanied by loss of enzyme activity. The significance of these findings with respect to the clinical use of raloxifene remains to be determined.

Introduction Estrogens are endogenous hormones and also are the most frequently prescribed medication in the United States for the treatment of menopausal symptoms and for the prevention and management of osteoporosis (1). In addition, data have been obtained to suggest that estrogens may produce desirable effects on the cardiovascular and central nervous systems, including reduced risks of stroke and delayed onset of Alzheimer’s disease (2, 3). However, overall health benefits from the so-called “estrogen replacement therapy” remain controversial, particularly because of an observed increase in the incidence of endometrial and breast cancer following longterm therapy (4). The effect of estrogens is believed to be mediated through interactions with the estrogen receptor (ER),1 which is a member of the nuclear receptor * Correspondence should be addressed to this author at the Department of Drug Metabolism, Merck & Co., P.O. Box 2000, RY800-B211, Rahway, NJ 07065. Tel: (732) 594-4501; Fax: (732) 594-4820. † Department of Drug Metabolism. ‡ Department of Medicinal Chemistry. 1 Abbreviations: CID, collision-induced dissociation; ER, estrogen receptor; GSH, reduced glutathione; IgG, immunoglobulin G; LC/MS/ MS, liquid chromatography-tandem mass spectrometry; NAC, Nacetylcysteine; P450, cytochrome P450.

superfamily. Two subtypes of ER have been discovered, one of which is termed ER-R, expressed in major female organs such as uterus and mammary gland (5). The other subtype is named ER-β and can be found abundantly in testis and prostate (5). Both receptors also are present in bone, liver, and the cardiovascular and central nervous systems (5). To minimize the effect on endometrial and breast tissues but maintain health benefits, selective ER modulators may be attractive alternatives to nonselective therapies (4). Among therapeutically effective ER modulators, raloxifene (Figure 1) has been approved by the Food and Drug Administration for the treatment of osteoporosis (6). The metabolism of raloxifene in laboratory animals was shown to lead primarily to the formation of phenolic glucuronide derivatives which, for example, accounted for approximately 50% of the dose recovered in rat bile (7). These glucuronic acid conjugates also were the major circulating metabolites in human subjects (8). On the other hand, mono- and dihydroxylated derivatives, which apparently bound to the ER, were identified in incubations of raloxifene with rat liver microsomes (9). In addition, the drug was shown to be associated with decreases in cytochrome P450 (P450) aromatase activities in human colon carcinoma cells (10). Preliminary studies

10.1021/tx0200109 CCC: $22.00 © 2002 American Chemical Society Published on Web 06/07/2002

908

Chem. Res. Toxicol., Vol. 15, No. 7, 2002

Figure 1. Structures of raloxifene and its thioether adducts. The identities of specific aromatic protons are discussed in the text relative to NMR data presented in Table 2.

in our laboratory indicated that raloxifene is a substrate of P450 3A4. In this report, we describe irreversible inhibition of P450 3A4 by raloxifene and the concomitant formation of glutathione (GSH) and N-acetylcysteine (NAC) adducts during raloxifene metabolism. The inhibitory potency of the drug on P450 3A4 was evaluated in human liver microsomal incubations, whereas the thioether adducts were identified in microsomal and hepatocyte incubations as well as in vivo in rats. The results are discussed in the context of metabolic activation of raloxifene, mechanisms of the drug-mediated P450 inhibition, and potential clinical implications.

Materials and Methods Materials. N-Acetylcysteine (NAC), glucose 6-phophate, glucose-6-phosphate dehydrogenase, reduced glutathione (GSH), NADP, NADPH, quinidine sulfate, and trifluoroacetic acid were purchased from the Sigma Chemical Co. (St. Louis, MO). Ketoconazole was obtained from Janssen Biotech (Olen, Belgium), and testosterone and 6β-hydroxytestosterone were from Steraloids (Wilton, NH). Raloxifene was extracted from Evista tablets using methanol. BondElut C18 solid-phase extraction cartridges were obtained from Varian Chromatography Systems (Walnut Creek, CA). [phenyl-3,5-3H]Raloxifene was synthesized at Merck Research Laboratories according to a published procedure (11); the specific activity was 25.79 Ci/mmol, and the radiochemical purity was 99%. Recombinant P450 2D6 and 3A4, coexpressed with NADPHP450 oxidoreductase in baculovirus-insect cells, were from Gentest Co. (Woburn, MA). Monoclonal inhibitory antibodies against human hepatic P450 2D6 and 3A4 were prepared in mice by immunization with the corresponding P450 enzymes (12). Instrumentation. Liquid chromatography-tandem mass spectrometry (LC/MS/MS) was carried out on a Perkin-Elmer Sciex API 3000 tandem mass spectrometer (Toronto, Canada) interfaced to an HPLC system consisting of a Perkin-Elmer Series 200 quaternary pump and a Series 200 autosampler (Norwalk, CT). Turbo IonSpray was utilized for ionization with positive ion detection. The source temperature was set at 150 °C, ionization voltage at 5 kV, and orifice potential at 50 V. Collision-induced dissociation (CID) of MH+ ions was achieved with nitrogen as the collision gas with a collision energy at 35 eV. Chromatography was performed on a DuPont Zorbax (Wilmington, DE) Rx-C8 column (4.6 × 250 mm, 5 µm), and samples were delivered at a flow rate of 1 mL/min with a 1:25 split to the ion source. The mobile phase consisted of acetonitrile-water containing 10% methanol and 0.05% trifluoroacetic

Chen et al. acid and was programmed for a linear increase from 10% to 70% acetonitrile during a 30 min period. Radioactivity was monitored by a Packard Flow Scintillation Analyzer (Meriden, CT). NMR spectra were recorded on a Varian Inova600 spectrometer operated at 600 MHz. Isolated metabolites of raloxifene were dissolved in deuterated acetonitrile. Chemical shifts are expressed as parts per million relative to tetramethylsilane. Incubations with Human Liver Microsomes or Recombinant P450 3A4. Human liver samples from four male and five female donors were obtained from the Pennsylvania Regional Tissue Bank (Exton, PA). The medical history of the donors has been reported elsewhere (13). Liver microsomes were isolated from individual livers by differential centrifugation (14). Aliquots from each preparation then were pooled on the basis of equivalent protein concentrations to yield a representative pool of human liver microsomes. The activity of P450 3A4 in individual liver microsomal preparations and in the pooled sample was estimated based on the 6β-hydroxylation of testosterone. Briefly, testosterone in methanol was added to human liver microsomes suspended in 100 mM phosphate buffer (pH 7.4) containing 1 mM EDTA and 6 mM magnesium chloride. The final substrate concentration was 50 µM, and the incubation volume was 1 mL. Incubations were performed in the presence of an NADPH-generating system, consisting of 10 mM glucose 6-phosphate, 0.7 unit of glucose-6-phosphate dehydrogenase, and 1 mM NADP, at 37 °C for 10 min. The formation of 6β-hydroxytestosterone was analyzed based on an HPLC assay described previously (15). For studies of raloxifene metabolism, human liver microsomes or recombinant P450 was suspended in phosphate buffer (pH 7.4) containing 0.5 mM GSH (or NAC). The final concentration of P450 was 0.24 nmol/mL for human liver microsomes and 0.07 nmol/mL for the recombinant enzymes. [phenyl-3,5-3H]Raloxifene in methanol was added to reach final concentrations ranging from 0 to 1 mM. The final concentration of methanol in incubation media was 0.2% (v/v), and the total incubation volume was 1 mL. Incubations were performed in the presence of an NADPH-generating system at 37 °C for 20 min, and the reaction was quenched by adding 60 µL of 10% aqueous trifluoroacetic acid. In experiments involving selective P450 inhibitors, ketoconazole or quinidine in methanol was added to microsomal suspensions containing GSH and incubated for 10 min. Controls contained the same amount of organic solvent but lacked inhibitors. Raloxifene and an NADPH-generating system then were added, and incubations (final volume 1 mL) were performed for an additional 20 min at 37 °C. The reaction was quenched by adding 60 µL of 10% aqueous trifluoroacetic acid. Immunoinhibition experiments followed a protocol similar to that described above. Briefly, monoclonal antibodies against P450 2D6 or 3A4 (0.5-2 mg of IgG/nmol of CYP) were preincubated with human liver microsomes for 15 min at room temperature. Control incubations contained ascites from untreated animals. Raloxifene was added to reach a final concentration of 50 µM and incubated in the presence of an NADPHgenerating system for an additional 20 min. The reaction was quenched with 10% aqueous trifluoroacetic acid. All experiments were performed in duplicate. Irreversible Inhibition of P450 3A4. Human liver microsomes were preincubated with raloxifene at 37 °C for 5-30 min in the presence of an NADPH-generating system and the presence or absence of GSH. The concentration of raloxifene ranged from 1 to 50 µM while that of GSH was 5 mM. Controls contained the same amount of raloxifene but lacked NADPH. The preincubation mixture then was diluted 10-fold with 0.1 mM phosphate buffer (pH 7.4), and testosterone 6β-hydroxylase activity was determined as described in the preceding section, except that the substrate concentration was 250 µM. Aliquots of the preincubation mixture (200 µL) were dialyzed overnight at 4 °C in a Slide-A-Lyzer mini-dialysis unit, consisting of a regenerated cellulose membrane with a molecular mass cutoff of 7 kDa, from the Pierce Chemical Co. (Rockford, IL).

Bioactivation of Raloxifene

Chem. Res. Toxicol., Vol. 15, No. 7, 2002 909

The dialysis was performed against 0.1 M phosphate buffer (pH 7.4) which was exchanged 3 times over the course of the experiment. Testosterone 6β-hydroxylase activity in the mixture then was determined and compared with that estimated before the dialysis. Protein concentrations were determined using a Pierce bicinchoninic (BCA) protein assay kit. The maximum rate constant for inactivation of the enzyme (kinact) and the inactivation constant (KI) were calculated based on the equation:

k)

kinact × I0 KI + I0

where k is the initial rate constant for inactivation of P450 3A4 and I0 is the initial inhibitor concentration (16, 17). The partition coefficient, which represents overall turnover of the inhibitor per inactivating process, was estimated from the ratio of the initial rate of disappearance of raloxifene to the rate of enzyme inactivation at 2 µM of the inhibitor (17). The concentrations of P450 enzymes in preincubation mixtures were measured based on their reduced carbon monoxidedifferential photometric spectra (18). All experiments were performed in duplicate. Incubations with Human Hepatocytes. A liver sample from a 27-year-old female donor, who died from drug overdose, was obtained from the Pennsylvania Regional Tissue Bank (Exton, PA). Hepatocytes were isolated according to a two-step perfusion procedure and exhibited a viability of greater than 80% as determined by the trypan blue exclusion test. Cryopreserved human hepatocyte samples from two male and two female donors were obtained from In Vitro Technologies (Baltimore, MD). These hepatocytes exhibited viabilities of 70-80% based on testing with trypan blue. Incubations were performed with the hepatocytes suspended in Krebs-bicarbonate buffer followed by addition of raloxifene in dimethyl sulfoxide. The final concentration of raloxifene in the suspension was 25 µM, and that of dimethyl sulfoxide was 0.1% (v/v). Incubations proceeded for 2 h at 37 °C, and reactions were quenched with 10% aqueous trifluoroacetic acid. Animal Experiments. Experiments were performed according to procedures approved by the Merck Institutional Animal Care and Use Committee. Two male Sprague-Dawley rats, purchased from Harlan Laboratories (Indianapolis, IN) and weighing 270-360 g, were allowed free access to commercial rat chow and water. The animals were anesthetized with Nembutal and their bile ducts cannulated with PE-10 tubing. Control bile was collected before treatment. An aqueous PEG400ethanol solution (PEG/ethanol/water, 20/10/70, v/v/v) of [phenyl3,5-3H]raloxifene was administered at 5 mg/kg by oral gavage, and bile and urine were collected for a period of 24 h. Isolation and Identification of GSH or NAC Adducts. Samples from in vitro incubations or specimens of bile and urine from rats treated with raloxifene were acidified to pH 2 and applied to a C18 extraction cartridge which had been prewashed with methanol and water. The cartridge was washed consecutively with water and methanol. The methanol eluate was evaporated to dryness under a stream of nitrogen and the residue reconstituted in 300 µL of 60% aqueous acetonitrile containing 0.05% trifluoroacetic acid. An aliquot (10-100 µL) of the resulting samples was injected onto a Zorbax Rx-C8 column for analysis by LC/MS/MS and isolation of metabolites. Identification of the metabolites was based on the product ion spectra obtained upon CID of the MH+ ions and, in some cases, was based on selected reaction monitoring of three mass transitions, namely, m/z 779 f 650, 779 f 506, and 779 f 112, with dwell times set at 400 ms per channel. The amount of the adducts formed was estimated based on radioactivity measurements.

Results Irreversible Inhibition of Human Hepatic P450 3A4. The activity of P450 3A4 in human liver microsomes

Figure 2. Preincubation time- and NADPH-dependent inhibition of P450 3A4 by raloxifene in human liver microsomal incubations. The inhibition was attenuated partially by the addition of GSH to preincubation media.

was estimated based on the rate of testosterone 6βhydroxylation under defined conditions. Raloxifene was evaluated initially for competitive inhibition of P450 3A4, and the IC50 value was determined to be 12 µM. Preincubation time-dependent inhibition of P450 3A4 then was investigated. The experiment was performed such that human liver microsomes were preincubated with raloxifene in the presence of NADPH followed by a 10-fold dilution of the incubation volume prior to assessment of the activity of testosterone 6β-hydroxylase. This approach was adopted to minimize contributions to enzyme inhibition from competitive components. With this experimental paradigm, it was found that preincubation of raloxifene with human liver microsomes led to decreased activity of P450 3A4, which was both NADPH- and preincubation time-dependent (Figure 2). This inhibition of P450 3A4 by raloxifene was further determined to be irreversible in nature. In particular, the activity of testosterone 6β-hydroxylase in liver microsomes could not be restored following an overnight dialysis when preincubations were performed in the presence of NADPH, whereas similar dialysis of preincubation mixtures lacking NADPH resulted in quantitative recovery of the enzyme activity (Table 1). The KI and kinact values for raloxifene-mediated inhibition of P450 3A4 were determined to be 9.9 µM and 0.16 min-1, respectively, while the partition coefficient was 1.8. However, despite the fact that decreases in P450 3A4 activities were apparent, no loss of spectrally detectable P450 was observed after treatment of human liver microsomes or recombinant P450 3A4 with raloxifene in the presence of NADPH (data not shown). The irreversible inhibition of P450 3A4 by raloxifene also was investigated in the presence of GSH. In this case, the degree of inhibition of testosterone 6β-hydroxylase in human liver microsomes following 20 min preincubations with raloxifene was reduced from 80% in the absence of GSH to 65% in the presence of the thiol tripeptide (Figure 2). Further work therefore was directed toward the identification of potential GSH and NAC adducts of raloxifene. Formation of GSH and NAC Adducts. Three components with MH+ ions at m/z 779 were detected by LC/

910

Chem. Res. Toxicol., Vol. 15, No. 7, 2002

Chen et al.

Table 1. Effect of Dialysis on Testosterone 6β-Hydroxylase Activity in Human Liver Microsomesa testosterone 6β-hydroxylase activity incubation mixture components

% of controls % of controls before dialysis after dialysis

methanol - NADPH

100

92

methanol + NADPH

85

66

raloxifene + methanol - NADPH

100

104

raloxifene + methanol + NADPH

24

9

a

Human liver microsomes (1 mg of protein/mL) were suspended in phosphate buffer (0.1 M, pH 7.4) containing EDTA. Raloxifene in methanol was added to reach a final concentration of 50 µM. Controls had the same amount of methanol (0.2%, v/v). Reactions were initiated by adding raloxifene and NADPH and conducted for 10 min. Dialysis was performed for 16 h at 4 °C. The activity of testosterone 6β-hydroxylase was determined for the same sample before and after dialysis.

Figure 3. Product ion mass spectra of (A) 7-glutathionylraloxifene, (B) 5-glutathionyl-raloxifene, and (C) 3′-glutathionylraloxifene derived from incubations of raloxifene with human liver microsomes. The spectra were obtained by CID of the MH+ ions at m/z 779. Fragmentation patterns are discussed in the text.

MS analysis of human liver microsomal incubations containing raloxifene and GSH and were arbitrarily assigned as M-GS-1, M-GS-2, and M-GS-3. Subsequent CID of the MH+ at m/z 779 in each case produced fragment ions at m/z 704 and 650, corresponding to losses of the elements of glycine and pyroglutamate, respectively (Figure 3), which are characteristic for GSHconjugated metabolites (19). The fragment ion at m/z 761 was derived from loss of water from the MH+ species, whereas the fragment ion at m/z 112 (protonated vinylpiperidine) was indicative of the presence of an intact

ethylpiperidine moiety in the conjugate. The fragment ion at m/z 506 was assigned as a cleavage adjacent to the thioether moiety with charge retention on the raloxifene residue (Figure 3). These GSH adducts were isolated and subjected to analysis by NMR. The proton NMR data are summarized in Table 2, together with the corresponding raloxifene data for reference. In the spectrum of raloxifene, the doublet signals at 6.66 and 7.21 ppm were assigned to the two pairs of equivalent protons on the phenol moiety (b and a), while the doublets at 6.78 and 7.67 ppm were assigned to the two pairs of protons on the O-alkylated phenol ring (d and c). A doublet of doublets at 6.86 ppm was attributed to proton 5 on the benzothiophene moiety, resulting from the coupling with proton 4 and an additional long-distance and weaker coupling interaction with proton 7. Consequently, both protons 4 and 7 were represented by doublet signals with different coupling constants (Table 2). In the case of M-GS-1, signals corresponding to proton 7 were absent while the signal for proton 5 collapsed from a doublet of doublets to a single doublet (Table 2). The metabolite was identified on this basis as 7-glutathionyl-raloxifene. For M-GS-2, the signals from proton 5 were absent, whereas singlets at 7.38 and 7.46 ppm were consistent with a para relationship between protons 4 and 7, and the metabolite therefore was assigned as 5-glutathionyl-raloxifene (Table 2). For M-GS-3, all signals associated with the benzothiophene (protons 4, 5, and 7) and O-alkylated phenol moieties (c and d) remained intact, but signals corresponding to protons a and b were replaced by those representing e, f, and g (Table 2). These features led to the assignment of this metabolite as 3′-glutathionylraloxifene. The GSH adducts of raloxifene also were detected in incubations with rat or human hepatocytes based on selected reaction monitoring LC/MS/MS analysis. Three characteristic mass transitions were used for metabolite detection, corresponding to the loss of pyroglutamate (m/z 779 f 650), cleavage of the cysteinyl thioether bond with charge retention on the aromatic moiety (m/z 779 f 506), and O-dealkylation forming a putative protonated Nvinylpiperidine (m/z 779 f 112). These transitions coincided with the HPLC retention times of 5- and 7-glutathionyl-raloxifene and 3′-glutathionyl-raloxifene (Figure 4). The amount of total adducts formed was approximately 4.4 pmol/million cells in rat hepatocyte suspensions and ranged from 2.7 to 13.6 pmol/million cells in human hepatocyte suspensions, with 7-glutathionylraloxifene being the most abundant conjugate formed (data not shown). The corresponding NAC adducts of raloxifene were detected by LC/MS/MS in the bile and urine from rats dosed orally with the drug, and these adducts were arbitrarily assigned as M-NAC-1 and M-NAC-2, with their MH+ ions at m/z 635. In each case, CID of the MH+ ion resulted in a fragment ion at m/z 593, corresponding to the loss of the elements of ketene, and a fragment ion at m/z 506, which was assigned as raloxifene plus a thiol substituent (Figure 5). The fragment ion at m/z 112 was again considered to derive from the ethylpiperidine moiety (Figure 5). These NAC adducts were isolated and subjected to analysis by NMR. Similar to its counterpart M-GS-1, the proton spectrum of M-NAC-1 was characterized by the absence of signals corresponding to proton 7 and a single

Bioactivation of Raloxifene

Chem. Res. Toxicol., Vol. 15, No. 7, 2002 911 Table 2. Proton NMR Data for Raloxifene and Its Thiol Adductsa chemical shifts (ppm), multiplicity, J (Hz), integral

compoundb

H4

H5

H7

Ha

raloxifene

7.38 d J ) 8.6, 1H 7.39 d J ) 8.7, 1H 7.46 s 1H

6.86 dd J ) 8.6, 2.2, 1H 6.98 d J ) 8.7, 1H -

7.32 d J ) 2.2, 1H -

7.51 d J ) 8.8, 1H 7.38 d J ) 8.6, 1H 7.47 s 1H

6.89 dd J ) 8.8, 2.3, 1H 6.93 d J ) 8.8, 1H -

7.33 d J ) 2.2, 1H -

7.21 m J ) 8.6, 2H 7.19 m J ) 8.6, 2H 7.19 m J ) 8.7, 2H -

6.66 m J ) 8.6, 2H 6.65 m J ) 8.6, 2H 6.65 m J ) 8.7, 2H -

7.47 d J ) 8.8, 1H

6.89 dd J ) 8.8, 2.2, 1H

7.33 d J ) 2.2, 1H

7.21 m J ) 8.6, 2H 7.19 m J ) 8.6, 2H -

6.65 m J ) 8.6, 2H 6.65 m J ) 8.6, 2H -

M-GS-1 M-GS-2c M-GS-3 M-NAC-1 M-NAC-2c (1) M-NAC-2 (2)

7.38 s 1H

7.39 s 1H

Hb

Hc

Hd

7.67 m J ) 8.9, 2H 7.68 m J ) 8.8, 2H 7.69 m J ) 8.8, 2H 7.65 m J ) 8.9, 2H 7.69 m J ) 8.9, 2H 7.70 m J ) 8.9, 2H 7.67 m J ) 8.9, 2H

6.78 m J ) 8.9, 2H 6.85 m J ) 8.8, 2H 6.85 m J ) 8.9, 2H 6.81 m J ) 8.9, 2H 6.85 m J ) 8.9, 2H 6.86 m J ) 8.9, 2H 6.83 m J ) 8.8, 2H

He

Hf

Hg

-

-

-

-

-

-

-

-

-

6.70 d J ) 8.5, 1H -

7.11 dd J ) 8.5, 2.3, 1H -

7.27 d J ) 2.3, 1H -

-

-

-

6.70 d J ) 8.4, 1H

7.12 dd J ) 8.4, 2.1, 1H

7.26 d J ) 2.2, 1H

a Spectra were acquired at 600 MHz in CD CN for raloxifene and in CD CN-D O (3:1, v/v) for the thioether adducts. b M-GS-1 3 3 2 corresponds to 7-glutathionyl-raloxifene; M-GS-2, 5-glutathionyl-raloxifene; and M-GS-3, 3′-glutathionyl-raloxifene. M-NAC-1 corresponds to 7-(N-acetylcysteinyl)-raloxifene, while M-NAC-2 consists of a mixture of 5-(N-acetylcysteinyl)-raloxifene (1) and 3′-(N-acetylcysteinyl)raloxifene (2). c Assignments for protons 4 and 7 are interchangeable.

Figure 4. Detection of the GSH adducts of raloxifene in (A) rat and (B) human hepatocyte suspensions. Three characteristic mass transitions, namely, m/z 779 f 650, 779 f 506, and 779 f 112, were utilized for detection of the metabolites.

doublet from proton 5 (Table 2). This metabolite therefore was identified as 7-(N-acetylcysteinyl)-raloxifene. The spectrum of M-NAC-2 exhibited a pattern consistent with that of a mixture of 5-(N-acetylcysteinyl)-raloxifene and 3′-(N-acetylcysteinyl)-raloxifene (Table 2). Quantitatively, approximately 70% of the administered radioactivity was recovered in rats treated with [phenyl-

3,5-3H]raloxifene at 5 mg/kg over a period of 24 h; 97% of the radioactivity was found in the bile and 3% in the urine. The NAC adducts collectively accounted for 1% of the recovered dose. Human P450 Enzymes That Catalyze the Oxidative Metabolism of Raloxifene. When oxidative metabolism of raloxifene was examined in incubations with

912

Chem. Res. Toxicol., Vol. 15, No. 7, 2002

Chen et al.

Figure 5. Product ion mass spectra of (A) M-NAC-1 and (B) M-NAC-2 in the bile from rats treated orally with raloxifene at 5 mg/kg. The spectra were obtained by CID of the MH+ ions at m/z 635. Fragmentation patterns are discussed in the text.

recombinant P450 1A2, 2C9, 2D6, 2E1, and 3A4, turnover of the parent compound was observed only with recombinant P450 2D6 and 3A4 (data not shown). Bioactivation of raloxifene then was evaluated based on GSH adduct formation. The adducts were identified only in incubations with recombinant P450 2D6 and 3A4, with the reaction rates estimated at 0.5 and 1.1 pmol min-1 (nmol of P450)-1, respectively. In human liver microsomes, the turnover of raloxifene and the formation of GSH adducts decreased to less than 15% of control values in incubations containing ketoconazole, a selective P450 3A4 inhibitor (Table 3). Inhibition of raloxifene metabolism also was apparent when human liver microsomes were pretreated with an inhibitory monoclonal anti-P450 3A4 IgG. Although an approximately 20% decrease in the metabolism of raloxifene was associated with the addition of 50 µM quinidine to human liver microsomal incubations, inhibition was not observed at lower quinidine concentrations, nor with an anti-P450 2D6 IgG (Table 3). The effect of quinidine at the higher concentration may be attributed to its interference with P450 3A4-mediated processes; inhibition of testosterone 6β-hydroxylase activity was shown in human liver microsomal incubations containing 50 µM quinidine (20). Further studies were performed with individual human liver preparations. A good correlation (r2 > 0.9) was obtained between the extent of GSH adducts formed and the activity of testosterone 6β-hydroxylase in liver microsomal preparations (Figure 6). This result, together

Table 3. Effect of Anti-P450 IgG or Selective Chemical Inhibitors on the Metabolism of Raloxifene in Incubations with Human Liver Microsomesa IgG or inhibitor

concentration

anti-2D6 antibody quinidine quinidine anti-3A4 antibody anti-3A4 antibody ketoconazole ketoconazole

0.5-2 (mg of IgG/ nmol of P450) 1-10 (µM) 50 (µM) 0.5 (mg of IgG/ nmol of P450) 2 (mg of IgG/ nmol of P450) 1 (µM) 10 (µM)

formation of metabolism of raloxifene GSH adducts (% of control)b (% of control)b 100

100

95-99 83 30

100 100 13

20

4

9 0

15 5

a Raloxifene and GSH in phosphate buffer were added to human liver microsomes (1 mg of protein/mL) suspended in phosphate buffer (0.1 M, pH 7.4) containing EDTA. The final concentration of raloxifene was 50 µM and GSH 5 mM. Anti-P450 2D6 IgG or anti-P450 3A4 IgG was preincubated with human liver microsomes for 20 min at room temperature. Control incubations contained preimmune IgG. Ketoconazole or quinidine was added in methanol solution. Controls had the same amount of methanol (0.2%, v/v). Reactions were initiated by adding raloxifene and NADPH and proceeded for additional 10 min. The formation of GSH adducts was analyzed by LC/MS/MS. b The control value for the rate of raloxifene metabolism in human liver microsomal incubations was estimated to be 444 pmol min-1 (mg of protein)-1; the formation of raloxifene GSH adducts was 96 pmol min-1 (mg of protein)-1.

with the data derived from studies with recombinant P450 enzymes, chemical inhibitors, and inhibitory antibodies, suggests strongly that P450-mediated bioactiva-

Bioactivation of Raloxifene

Chem. Res. Toxicol., Vol. 15, No. 7, 2002 913 Scheme 1. Proposed Mechanisms for the Cytochrome P450 3A4-Mediated Bioactivation of Raloxifene, Leading to the Formation of 7-Glutathionyl-raloxifenea

Figure 6. Correlation between the extent of formation of GSH adducts and the activity of P450 3A4 (testosterone 6β-hydroxylation) in human liver microsomes. The substrate concentration was 10 µM.

tion of raloxifene in human liver tissues (as measured by the formation of GSH adducts) is catalyzed primarily by P450 3A4.

Discussion Raloxifene is a selective ER modulator and has been approved for the treatment of osteoporosis in postmenopausal women (6, 21). In a recent study, raloxifene was shown to be associated with inhibition of P450 aromatase (10). In the present account, we demonstrate that raloxifene is an irreversible inhibitor of P450 3A4 and that metabolism of the drug results in the formation of reactive species which may be trapped by GSH. It was observed initially in this study that preincubation of raloxifene with human liver microsomes led to decreases in P450 3A4 activity. This enzyme inhibition was characterized by a dependence upon NADPH, the cofactor required for P450-catalyzed reactions, and was enhanced with increasing preincubation time. The loss of P450 3A4 activity could not be restored by dialysis of preincubation mixtures fortified with NADPH, although the competitive inhibitory effect of the drug was eliminated nearly completely by the dialysis processes. These data collectively suggest that the inhibition is irreversible in nature and possibly is due to the formation of reactive metabolites of raloxifene. Such metabolites could either alkylate the heme moiety of P450 3A4, form a metabolicintermediate (MI) complex, or, alternatively, bind covalently to the apoprotein, thereby impairing enzyme function. It is unlikely, however, that the inactivation of P450 3A4 by raloxifene occurs via modification of the heme moiety, since spectrally detectable P450 is not perturbed by preincubation of the drug with human liver microsomes or with recombinant P450 3A4. Subsequent studies of the kinetics of raloxifene-mediated P450 3A4 inactivation indicated that the dissociation constant (KI) for the drug-enzyme complex (9.9 µM) and the partition coefficient (1.8) both are much lower than the corresponding values (46 µM and 9) reported for gestodene, a known mechanism-based inactivator of P450

aSimilar mechanisms can be depicted for the formation of the 5′- and 3′-glutathionyl derivatives. Whether these reactive intermediates are responsible for the inactivation of P450 3A4 remains to be established.

3A4 (22). Therefore, raloxifene appears to be effective in inactivating P450 3A4, as judged by overall metabolic turnover of the inhibitor per inactivation. This effectiveness might result from a tight complex between raloxifene and P450 3A4, that would lead to an immediate attack of the drug-derived reactive species to the enzyme. On the other hand, the rate of enzyme inactivation induced by raloxifene (kinact 0.16 min-1) is only half the rate (kinact 0.39 min-1) produced by gestodene (22). A possible scenario might be that turnover of raloxifene to reactive, P450 3A4 inhibitory products occurs more slowly and, as a consequence, the resulting enzyme inactivation also is slower than in the case with gestodene. Further evidence implicating reactive metabolites in the raloxifene-mediated inactivation of P450 3A4 includes an apparent attenuation of the degree of enzyme inactivation in preincubation mixtures fortified with GSH, and the identification of raloxifene-derived GSH adducts in human liver microsomal incubations. A common structural feature of those thioether adducts is the attachment of GSH to the phenolic moieties of raloxifene, with conjugation occurring primarily at the benzothiophene portion of the drug. A possible mechanism for the metabolic activation of raloxifene would involve initial P450-catalyzed epoxidation of the phenol residues to form arene oxide intermediates. Nucleophilic attack on these reactive species by GSH, followed by dehydration and aromatization, would result in the formation of the identified adducts (Scheme 1). Alternatively, raloxifene could be oxidized to an extended quinone intermediate, which then would be trapped by GSH conjugation (Scheme 1). This latter mechanism for raloxifene bioactivation would involve an intermediate that had several equally reactive centers for GSH conjugation, whereas the empirical results indicated that addition of GSH to raloxifene occurred preferentially at the 7-position of the benzothiophene moiety. Nevertheless, it is conceivable that one or more of the putative arene oxide or extended quinone reactive intermediate(s) are capable of covalently modifying the P450 3A4 apoprotein. Whether the resulting drug-protein complexes would eventually lead to the observed inactivation of the enzyme remains to be established.

914

Chem. Res. Toxicol., Vol. 15, No. 7, 2002

The oxidative bioactivation of raloxifene in human liver most likely is catalyzed by P450 3A4, since the GSH adducts were identified in incubations of the drug with the recombinant enzyme. The formation of the adducts was nearly completely inhibited when human liver microsomes were pretreated with ketoconazole, a selective inhibitor of P450 3A4, or with a monoclonal anti-P450 3A4 IgG. In addition, the extent of formation of the adduct correlated linearly with P450 3A4 activities in nine individual human liver samples. Although raloxifene metabolism was observed with recombinant P450 2D6, the reaction in human liver microsomes was not affected by the presence of quinidine, nor by an inhibitory antiP450 2D6 IgG. These data collectively suggest that P450 3A4-mediated epoxidation (and/or quinone formation) is rate-limiting in the process of raloxifene bioactivation and subsequent conjugation with GSH and associated P450 3A4 inactivation. Further studies are in progress to investigate the nature of a possible covalent modification of the enzyme by raloxifene. It has been reported that conjugation with glucuronic acid represents a major pathway for the clearance of raloxifene in humans (7, 8). In the present study, the GSH adducts of raloxifene were detected by LC/MS in rat hepatocyte suspensions, while the NAC adducts were identified in the bile and urine from treated rats. The NAC adducts most likely are derived from the corresponding GSH adducts via the mercapturic acid pathway, and may be taken as a reflection of the formation in vivo of arene oxide (and/or extended quinone) intermediates. The GSH adducts also were identified in five individual human hepatocyte preparations, implying that bioactivation of raloxifene may occur in humans via a process similar to that observed in rats. The clinical significance of raloxifene bioactivation and consequent irreversible inhibition of P450 3A4 requires further investigation, although arene oxides of carbamazepine and phenytoin have been implicated as possible culprits in drug-related idiosyncratic toxicities (23, 24). In a recent report, raloxifene was implicated in a case of hepatitis in a 49-year-old woman who had jaundice accompanied by elevated liver enzymes following oral administration of the drug for 30 days (23). The toxicity was suggested to result from an immune mechanism based on observations of skin rash and eosinophilia (25), whereas the role of chemically reactive metabolites of raloxifene in the process remains unknown. In summary, raloxifene appears to cause irreversible inhibition of P450 3A4 in vitro, possibly through metabolism to one or a number of electrophilic arene oxide (and/ or extended quinone) metabolites which covalently modify the enzyme apoprotein. These reactive species may be trapped by GSH in the form of thioether conjugates, which are formed in the rat as well as in human liver preparations. Collectively, these findings may be important in the rational design of new selective ER modulators with decreased propensity to undergo metabolic activation in humans.

Acknowledgment. We thank Ms. Deborah Newton (Merck Research Laboratories) for isolation of human liver microsomes and determination of P450 activities and Dr. Frank Tang (Merck Research Laboratories) for synthesis of tritium-labeled raloxifene. We also thank Dr. Anthony Lu (Rutgers University, New Brunswick, NJ) for constructive suggestions.

Chen et al.

References (1) Felson, D. T., Zhang, Y., Hannan, M., Kiel, D. P., Wilson, P. W. F., and Anderson, J. J. (1993) The effect of postmenopausal estrogen therapy on bone density in elderly women. N. Engl. J. Med. 329, 1141-1146. (2) Nabulsi, A. A., Folsom, A. R., White, A., Patsch, W., Heiss, G., Wu, K. K., and Szklo, M. (1993) Association of hormone-replacement therapy with various cardiovascular risk factors in postmenopausal women. N. Engl. J. Med. 328, 1069-1075. (3) Tang, M.-X., Jacobs, D., Stern, Y., Marder, K., Schofield, P., Gurland, B., Andrews, H., and Mayeux, R. (1996) Effect of oestrogen during menopause on risk and age at onset of Alzheimer’s disease. Lancet 348, 429-432. (4) Cosman, F., and Lindsay, R. (1999) Selective estrogen receptor modulators: Clinical spectrum. Endocr. Rev. 20, 418-434. (5) Muramatsu, M., and Inoue, S. (2000) Estrogen receptors: How do they control reproductive and nonreproductive functions? Biochem. Biophys. Res. Commun. 270, 1-10. (6) Clemett, D., and Spencer, C. M. (2000) Raloxifene: A review of its use in postmenopausal osteoporosis. Drugs 60, 379-411. (7) Lindstrom, T. D., Whitaker, N. G., and Whitaker, G. W. (1984) Disposition and metabolism of a new benzothiophene antiestrogen in rats, dogs and monkeys. Xenobiotica 14, 841-847. (8) Knadler, M. P., Lantz, R. J., Gillespie, T. A., Allerheiligen, S. R. B., and Henry, D. P. (1995) The disposition and metabolism of 14C-labeled raloxifene in humans. Pharm. Res. 12, S-372. (9) Lim, H. K., Stellingweif, S., Sisenwine, S., and Chen, K. W. (1999) Rapid drug metabolite profiling using fast liquid chromatography, automated multiple-stage mass spectrometry and receptor-binding. J. Chromatogr., A 831, 227-241. (10) Fiorelli, G., Picariello, L., Martineti, V., Tonelli, F., and Brandi, M. L. (1999) Estrogen synthesis in human colon cancer epithelial cells. J. Steroid Biochem. Mol. Biol. 71, 223-230. (11) Dodge, J. A., Stocksdale, M. G., and Jones, C. D. (1994) A chemical probe for the estrogen receptor: Synthesis of the 3H-isotopomer of raloxifene. J. Labeled Comput. Radiopharm. 36, 43-49. (12) Mei, Q., Tang, C., Assang, C., Slaughter, D., Rodrigues, D. A., Rushmore, T. H., and Shou, M. (1999) Role of a potent inhibitory monoclonal antibody to cytochrome P450 3A4 in assessment of human drug metabolism. J. Pharmacol. Exp. Ther. 291, 749759. (13) Tang, W., Stearns, R. A., Wang, R. W., Chiu, S.-H. L., and Baillie, T. A. (1999) Roles of human hepatic cytochrome P450s 2C9 and 3A4 in the metabolic activation of diclofenac. Chem. Res. Toxicol. 12, 192-199. (14) Raucy, J. L., and Lasker, J. M. (1991) Isolation of P450 enzymes from human liver. Methods Enzymol. 206, 557-587. (15) Wang, R. W., Newton, D. J., Liu, N., Atkins, W. M., and Lu, A. Y. H. (2000) In vitro drug-drug interaction patterns are substrate dependent. Drug Metab. Dispos. 28, 360-366. (16) Waley, S. G. (1985) Kinetics of suicide substrates: Practical procedures for determining parameters. Biochem. J. 227, 843849. (17) Silverman, R. B. (1988) Mechanism-based enzyme inactivation: Chemistry and enzymology, Vols. 1 and 2, CRC Press, Boca Raton, FL. (18) Omura, T., and Sato, R. (1964) The carbon monoxide-binding pigment of liver microsomes. I. Evidence for its hemoprotein nature. J. Biol. Chem. 239, 2370-2378. (19) Baillie, T. A., and Davis, M. R. (1993) Mass spectrometry in the analysis of glutathione conjugates. Biol. Mass Spectrom. 22, 319325. (20) Newton, D. J., Wang, R. W., and Lu, A. Y. H. (1995) Cytochrome P450 inhibitors: Evaluation of specifities in the in vitro metabolism of therapeutic agents by human liver microsomes. Drug Metab. Dispos. 23, 154-158. (21) Balfour, J. A., and Goa, K. L. (1998) Raloxifene. Drugs Aging 12, 335-341. (22) Guengerich, F. P. (1990) Mechanism-based inactivation of human liver microsomal cytochrome P-450 IIIA4 by gestodene. Chem. Res. Toxicol. 3, 363-371. (23) Park, B. K., Pirmohamed, M., and Kitteringham, N. R. (1992) Idiosyncratic drug reactions: A mechanistic evaluation of risk factors. Br. J. Clin. Pharmacol. 34, 377-395. (24) Madden, S., Maggs, J. L., and Park, B. K. (1996) Bioactivation of carbamazepine in the rat in vivo: Evidence for the formation of reactive arene oxide(s). Drug Metab. Dispos. 24, 469-479. (25) Vilches, A. R., Perez, V., and Suchecki, D. E. (1998) Raloxifeneassociated hepatitis. Lancet 352, 1524-1525.

TX0200109