240
Chem. Res. Toxicol. 1989, 2, 240-246
Inhibition of Hepatic Drug Metabolism by Phenothiazine Tranquilizers: Quantitative Structure-Activity Relationships and Selective Inhibition of Cytochrome P-450 Isoform-Specific Activities Michael M u r r a y
Liver Research Unit, Department of Medicine, University of Sydney, Westmead Hospital, Westmead, N S W 2145, Australia Received January 4, 1989 Twelve phenothiazine tranquilizers were investigated for the capacity t o inhibit rat hepatic microsomal cytochrome P-450(P-450) isoform-specific drug oxidation in vitro. All congeners were substituted in the 2- (carbocyclic) and 10- (thiazinyl nitrogen of the thiazine ring) positions. Cytochrome P-450PB-B-mediated 7-pentylresorufin 0-depentylase and P-450BNF-B-mediated 7-ethylresorufin 0-deethylase activities were effectively inhibited by most of the compounds. Structure-activity correlations revealed the apparent importance of the lipophilicity of the 2-substituent, and the negative effect of flexibility in the 10-position substituent, on anti-P-450 PB-B potency. On the other hand, inhibition of P-450BNF-B activity was promoted by bulkiness and branching within the 10-substituent and the shape/bulk of the 2-group. From this analysis it is likely that the active site of P-450PB-B is relatively small with a t least one lipophilic region t h a t may be involved in substrate and inhibitor binding. T h e active site of P-450BNF-B is relatively large, and steric properties, rather than lipophilic character, appear to determine inhibition by phenothiazines. Derivatives with piperidinyl and piperazinyl ring systems in the 10-position were relatively active inhibitors of P-450PCN-E (or an immunochemically related form of P-450)that catalyzes androst-4-ene-3,17-dione 6P-hydroxylation, and P-450UT-Amediated 16a-hydroxylase activity. In contrast, steroid 7a-hydroxylation (P-450UT-F) and N-nitrosodimethylamine N-demethylation (P-45Oj) were refractory to inhibition. From these studies it is apparent that phenothiazines are inhibitors of nonuniform potency toward different P-450sand that different structural features contribute to inhibition of individual P-450s.Studies of this type add t o the knowledge of active site topologies of P-450s,and isozyme-selective inhibition data will ultimately facilitate the prediction of toxic pharmacokinetic interactions between drugs.
Introduction A wide variety of lipophilic molecules are converted to more polar metabolites by cytochrome P-450 (P-450),'the catalytic component of the hepatic microsomal mixedfunction oxidase (MFO) system. Numerous studies have established that the P-450s are a family of closely related, distinct hemoproteins ( 1-4).2 Since ferroprotoporphyrin IX is the prosthetic heme group in all P-450s, the different MFO activity profile displayed by each form of P-450 is most likely a consequence of different substrate binding domains within apoprotein molecules. Most recently, it has been demonstrated that certain microsomal MFO reactions are attributable to individual P-450s. Thus, selection of appropriate MFO activities for measurement can provide information regarding the function of specific Abbreviations: P-450, cytochrome P-450; MFO, mixed-function oxidase; androstenedione, androst-4-ene-3,17-dione; PROD, 7-pentylresorufin (7-pentoxy-3H-phenoxazin-3-one) 0-depentylase; EROD, 7-ethylresorufin (7-ethoxy-3H-phenoxazin-3-one) 0-deethylase; QSAR, quantitative structure-activity relationship. Apparently equivalent forms of cytochrome P-450 mentioned in this paper include the following: P-450 BNF-B (24) and P-45Oc (39), and P-450 IA1 is the corresponding gene designation (40);P-450 PB-B (24), P-450b (39),and P-450 PB-4 ( 4 ) ,gene designation P-450 IIBl (40);P-450 UT-A (24),P-450h ( 2 ) ,P-450 2~ ( 4 ) ,P-450 RLM5 (3),and P-45018, (41), gene designation P-450 IICll (40);P-450 PCN-E (24), P-45Op (42), and P-450 2a ( 4 ) , a member of the P-450 IIIA subfamily (40); P-45Oj (43), P-450ac (44),P-450 RLM6 (45),gene designation P-450 IIEl (40);P-450 ISF-G (24), P-450d (46), gene designation P-450 IA2 (40); and P-450 UT-F (24) is apparently identical with P-450a (39) and P-450 3 ( 4 ) ,gene designation P-450 IIAl (40).
P-450s in intact microsomal fractions. A range of compounds are effective inhibitors of the MFO system. In particular, some nitrogen heterocycles, such as the imidazoles and benzimidazoles, are among the most potent MFO inhibitors yet described (5-8). Many such compounds are also used in therapeutic situations and have the capacity to modulate the metabolism of coadministered drugs. The structural requirements for effective MFO inhibition have usually been found to include lipophilic character, although steric and electronic factors may also be important (5, 7). Most structure-inhibition studies, however, have employed MFO reactions that are not specific for individual P-450s. It was the purpose of this study to investigate the P-450 isoform specificity of inhibition by a range of phenothiazine antipsychotic agents that are in clinical use. Structure-activity relationships were developed for the inhibition of two P-450s in terms of conventional physicochemical parameters and in terms of molecular descriptors related to molecular connectivity and shape.
Experimental Procedures Chemicals. Promazine hydrochloride, chlorpromazine hydrochloride, trifluoperazine dihydrochloride, promethazine hydrochloride, thioridazine hydrochloride, perphenazine, and ethopropazine hydrochloride were obtained from Sigma Chemical Co., St. Louis, MO. Pericyazine and prochlorperazine maleate were gifts from May and Baker Australia Pty. Ltd., West Footscray, Victoria. Thiopropazate hydrochloride was the gift of Searle
0893-228~/89/2702-0240$01.50/0 0 1989 American Chemical Society
Chem. Res. Toxicol., Vol. 2, No. 4, 1989 241
Inhibition of Cytochromes P-450 by Phenothiazines Research, Crows Nest, NSW, Australia, fluphenazine hydrochloride was provided by E. R. Squibb and Sons Pty. Ltd., St. Leonards, NSW, Australia, and thiethylperazine dimaleate was generously provided by Sandoz Pty. Ltd., North Ryde, NSW, Australia. [4-l4C]Androst-4-ene-3,l7-dione (androstenedione; sp act. 59 mCi/mmol) was purchased from Amersham Australia (Sydney, NSW). Testosterone, unlabeled androstenedione and its 68- and 16a-hydroxy metabolites, and all biochemicals were obtained from Sigma. 7a-Hydroxyandrostenedione was provided by the MRC Steroid Collection, Queen Mary's College, London, U.K. 168Hydroxyandrostenedione was prepared enzymatically as described elsewhere (9, 10). 7-Pentylresorufin and 7-ethylresorufin were purchased from Pierce Chemicals, Rockford, IL. Resorufin, isoniazid, 8-naphthoflavone, and N-nitrosodimethylamine were obtained from Aldrich Chemicals, Milwaukee, WI. Solvents and other reagents were from Ajax, Sydney, Australia, and were a t least analytical grade. Animals. Male Wistar rats (250-350 g) were used in all experiments. Animals were either untreated or were induced with phenobarbital (100 mg/kg ip in saline on 3 consecutive days), p-naphthoflavone (40mg/kg ip in corn oil on 3 consecutive days) or isoniazid (0.1% in drinking water for 10 days). Hepatic microsomal fractions were prepared 48 h after the final exposure t o inducer and were stored as frozen aliquots (-70 OC) in 20% glycerol containing 1 mM EDTA and 50 mM potassium phosphate, p H 7.4, until required for use. Androstenedione Hydroxylase Activity. The assay of androstenedione hydroxylation was performed as described elsewhere (11). Androstenedione metabolites were resolved by TLC (silica gel 60, F254 type, 20 X 20 cm X 0.25 mm, heated a t 100 "C for 15 min before use; E. Merck, Darmstadt, West Germany) in the solvent system CHCl,/ethyl acetate 1:2 (developed twice with drying between runs) of Waxman e t al. (12). Zones that comigrated with authentic standards were scraped into vials for scintillation spectrometry (Econofluor, New England Nuclear Corp., Boston, MA). Other MFO Assays. 7-Pentyl- and 7-ethylresorufin O-dealkylations were monitored essentially by the continuous spectrofluorometric procedure of Prough e t al. (13). Substrate and protein concentrations were 10 pM and 100 pg/mL, respectively, in the case of pentylresorufin metabolism, and 2.5 pM and 50 wg/mL, for ethylresorufin 0-deethylation. N-Nitrosodimethylamine N-demethylase activity WBS estimated in hepatic microsomes from isoniazid-induced rats by using substrate and protein concentrations of 4 mM and 1.7 mg/mL, respectively (14). Formaldehyde was quantified by the Nash procedure (15). Inhibitors were added directly to microsomal incubations in either potassium phosphate buffer (0.1 M, p H 7.4), absolute ethanol, or 50% ethanol in Nfl-dimethylformamide; solvent was added to control incubations (concentration 0.5%). Less than 10% inhibition was observed with these solvents. No inhibitors were studied a t concentrations greater than 100 pM. 150 values were determined from plots of log inhibitor concentration versus percent inhibition, using a t least four different inhibitor concentrations and a t least in duplicate ( r > 0.95 in each 150 determination). The extent of inhibition was usually in the range 10-90% * Physicochemical Parameter Values. Substituent hydrophobic (*) and molar refractivity (MR) constants were taken from the literature (16, 17) or were derived by using the addivity principle, assuming a a increment of 0.50/CH2 unit and an MR increment of 5.65/CHz unit; MR values were scaled by 0.1 to yield more manageable coefficients. Molecular connectivity indexes ("xVt)were calculated by the methods of Kier and Hall (18). In the expression M ~ v t ,m refers to the order of the function (number of bonds over which the connectivity calculation is performed), t is the type of function (path, cluster, or path/cluster), and v refers to the use of valence 6's in the computation of x indexes. Valence 6's (hV) were obtained from the literature (19). As an example, the first-order valence molecular connectivity expression is
Table I. Structures of Phenothiazine Tranquilizers Used in the Present Study
no. comDd I promazine I1 chlorpromazine I11 promethazine IV ethopropazine V trifluoperazine VI prochlorperazine VI1 fluphenazine
R, H C1 H H CF3 C1 CF3
VI11 perphenazine
C1
IX X XI
SCH, SCzH5 C1
thioridazine thiethylperazine thiopropazate
XI1 pericyazine
CN
where n is the number of single bond path lengths (or subgraphs) in the structure and i and j correspond to the individual vertexes (atoms) that form the subgraph. x indexes of different types and orders were calculated in analogous fashion. In this study the phenothiazine nucleus was common to all inhibitors and was not incorporated into x calculations. Molecular shape indexes ("K,) were calculated by the method of Kier (20). In this expression m refers to the order of the function (the number of bonds over which the function is calculated) and a refers to the use of modified atom counts (to account for non-sp3 hybridized atoms in the structure). K indexes were used t o calculate E (a molecular graph based substituent steric effect parameter) from the expression 2 ' K , - OK, - ,K, (21). Statistics. Regression equations were determined by using the STATVIEW program package on the Apple Macintosh I1 computer in the Liver Research Unit a t Westmead Hospital.
Results Inhibition of Hepatic Microsomal 7-Pentylresorufin 0 -Depentylase Activity by Phenothiazine Drugs. The series of phenothiazine tranquilizers shown i n Table I proved t o be q u i t e potent inhibitors of micros o m a l 7-pentylresorufin 0 - d e p e n t y l a s e ( P R O D ) activity i n phenobarbital-induced rat liver. The majority of comp o u n d s exhibited I,,s i n the range 1.5-15 pM, w i t h thioridazine (IX) proving exceptionally potent (I, = 0.37 pM, Table 11) and pericyazine proving quite nonpotent (I, = 53 pM). In general, inhibition p o t e n c y was g r e a t e s t i n t h o s e c o m p o u n d s w i t h flexible (dimethy1amino)propyl g r o u p s i n t h e R2 position (I and 11). T h o s e c o m p o u n d s w i t h piperazinyl-containing R2 g r o u p s tended to be less potent inhibitors and, although increasing the lipophilicity of the R1g r o u p d i d not m a r k e d l y e n h a n c e PROD inhibition, t h e hydrophilic R1-cyano group of pericyazine clearly resulted i n a weak inhibitor. T h e b e s t q u a n t i t a t i v e structure-activity relationships (QSAR) obtained for PROD inhibition b y phenothiazines a r e shown i n Table IV. The best single-variable equation was eq 2 in which 51% of the data variance was explained i n t e r m s of a three-bond path connectivity f u n c t i o n describing the R2 substituent. T h u s , increasing n u m b e r s of possible three-bond paths i n this part of the phenothiazine detracts from PROD inhibition potency. Similar equations involving t h e molar refractivity (MR) and single-bond path connectivity index (Ix")of the R2 g r o u p also had negative coefficients and accounted for a substantial proportion of
242 Chem. Res. Toxicol., Vol. 2, No. 4, 1989
Murray
Table 11. Inhibition of Microsomal Cytochrome P-450 Isozyme-Specific Pathways of D r u g Oxidation by Phenothiazine Tranquilizers4
phenothiazine promazine chlorpromazine promethazine ethopropazine trifluoperazine prochlorperazine fluphenazine perphenazine thioridazine triethylperazine thiopropazate pericyazine
PRODb 1.5 1.8 4.0 7.5 5.8 8.6 9.3 15 0.37 3.7 12 53
I,, fiM androst-4-ene-3,17-dione hydroxylase 6P 7a 16a 16B
EROD 69 9.1 5.5 4.9 1.1
5.0 2.6 7.4 8.6 5.5 6.1 11
-
-C
-
-
66 59 67 62 52 53 67 -
88 64 73 53 100 66 -
-
95 -
-
-
69 47 53 39 43 94 48 -
Cytochrome P-450 isozymes reflected by the oxidase activity: PROD, P-450 PB-B; EROD, P-450 BNF-B; androst-4-ene-3,17-dione so-, 7a-, and l6a-hydroxylases, P-450 PCN-E (or immunochemically similar forms), P-450 UT-Fand P-450 UT-A, respectively (4, 12,22, and 24). Uninhibited reaction rates: PROD, 2.2 nmol of resorufin/(min.mg of protein); EROD, 2.5 nmol of resorufin/(min.mg of protein); androstenedione 60-, 7a-, 16a-, and 16B-hydroxylase activities were 2.3,0.28,1.4, and 0.19 nmol of hydroxy metabolite/(min-mg of protein). PROD, 7-pentylresorufin 0-depentylase in hepatic microsomes from phenobarbital-induced rats; EROD, 7-ethylresorufin 0-deethylase in hepatic microsomes from p-naphthoflavone-inducedrats. Less than 50% inhibition observed at drug concentrations of 100 fiM. (I
Table 111. QSAR Parameters Used i n t h e Derivation of Rearession Eauationsa promazine chlorpromazine promethazine ethopropazine trifluoperazine prochlorperazine fluphenazine perphenazine thioridazine thiethylperazine thiopropazate pericyazine a
0 0.71 0 0 0.88 0.71 0.88 0.71 0.61 1.07 0.71 -0.57
2.804 2.804 2.806 3.936 4.637 4.637 5.207 5.207 4.144 4.637 6.285 4.280
2.210 2.210 2.060 3.289 4.028 4.028 4.714 4.714 3.868 4.028 5.594 4.023
0.605 0.605 0.881 1.435 2.070 2.070 2.356 2.356 2.254 2.070 2.652 2.031
0.224 0.224 0.806 0.737 0.652 0.652 0.671 0.671 0.570 0.652 0.730 0.518
0 1.29 0 0 3.79 1.29 3.79 1.29 2.35 3.35 1.29 1.29
0
1.57 0 0 6.68 1.57 6.68 1.57 2.06 2.92 1.57 1.20
Parameters were taken from the literature or calculated according to literature methods, as described under Experimental Procedures.
Table IV. Regression Equations of t h e Inhibition of Microsomal 7-Pentyl- a n d 7-Ethylresorufin 0-Dealkylase Activities by Phenothiazines eq N r r2, % 5 F P Inhibition of 7-Pentylresorufin 0-Depentylase +6.181 11 0.61 37 1 -0.235 0.36 5
