Human cytochromes P450

Molecular Aspects of Medicine 20 (1999) 1±137 www.elsevier.com/locate/mam

Human cytochromes P450 Julia A. Hasler a,*, Ronald Estabrook b, Michael Murray c, Irina Pikuleva d, Michael Waterman d, Jorge Capdevila e, Vijakumar Holla e, Christian Helvig e, John R. Falck b, Geo€rey Farrell f, Laurence S. Kaminsky g, Simon D. Spivack g, Eric Boitier h, Philippe Beaune h a Department of Biochemistry, University of Zimbabwe, 167 Mt. Pleasant, Harare, Zimbabwe Department of Biochemistry, University of Texas, South Western Medical Centre, 5323 Harry Hines Blvd, Dallas, TX 75235, USA c School of Physiology and Pharmacology, University of New South Wales, Sydney, NSW 2052, Australia d Department of Biochemistry, Vanderbilt University School of Medicine, Nashville, TN 37232-0146, USA e Departments of Medicine and Biochemistry, Vanderbilt University Medical School, Nashville, USA f University of Sydney at Westmead Hospital, Westmead Hospital, Westmead, NSW 2145, Australia g NY State Department of Health, Wadsworth Center, P.O. Box 509, Albany, NY 12201-0509, USA h INSERM U 490, Toxicologie Mol eculaire, Universit e Ren e Descartes, 45 rue des Saints-P eres, F-75270 Paris, Cedex 06, France b

Abstract The cytochrome P450 proteins (CYPs) are a family of haem proteins resulting from expression of a gene super-family that currently contains around 1000 members in species ranging from bacteria through to plants and animals. In humans, about 40 di€erent CYPs are present and these play critical roles by catalyzing reactions in: (a) the metabolism of drugs, environmental pollutants and other xenobiotics; (b) the biosynthesis of steroid hormones; (c) the oxidation of unsaturated fatty acids to intracellular messengers; and (d) the stereo- and regio-speci®c metabolism of fat-soluble vitamins. This review deals with aspects of cytochrome P450s of relevance to human physiology, biochemistry, pharmacology and medicine. Topics reviewed include: pharmacogenetics of CYPs, induction and inhibition of these haem proteins, their role in metabolism of endogenous compounds such as steroids and eicosanoids, the e€ect of disease on CYP function, CYPs and cancer, and CYPs as targets of antibodies in immunemediated diseases. Ó 1999 Elsevier Science Ltd. All rights reserved. Keywords: Cytochrome P450; Xenobiotics; Drug metabolism; Genetic polymorphisms; Steroid hormone synthesis; Fatty acid epoxidation; Eicosaniod metabolism; Disease; Chemical carcinogenesis; Cancer; Autoantibodies

*

Corresponding author. E-mail: hasler@gira€e.icon.co.zw

0098-2997/99/$ - see front matter Ó 1999 Elsevier Science Ltd. All rights reserved. PII: S 0 0 9 8 - 2 9 9 7 ( 9 9 ) 0 0 0 0 5 - 9

2

J.A. Hasler et al. / Molecular Aspects of Medicine 20 (1999) 1±137

Contents 1.

An 1.1. 1.2. 1.3. 1.4. 1.5. 1.6. 1.7. 1.8.

introduction to the cytochrome P450s . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5 What are P450s? . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5 History and spectrophotometric characterization . . . . . . . . . . . . . . . . . . . . . . 6 Biochemical properties . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7 What do P450s do? . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8 The nomenclature of the P450 super-family. . . . . . . . . . . . . . . . . . . . . . . . . . 9 The key role of P450s in human biology and medical therapeutics . . . . . . . . . 11 The future . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12

2.

Pharmacogenetics of cytochromes P450 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12 2.1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12 2.2. The CYP2D6 (debrisoquine hydroxylase) polymorphism . . . . . . . . . . . . . . . . 14 2.3. The CYP2C19 (S-mephenytoin hydroxylase) polymorphism . . . . . . . . . . . . . . 20 2.4. Polymorphisms in other cytochromes P450 (CYPs 1A1, 1A2, 2A6, 2C9, and 2E1) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21

3.

Induction and inhibition of CYPs and implications for medicine . . . . . . . . . . . . . . 3.1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.1.1. Cytochrome P450 enzymes: functional importance in drug metabolism. . . 3.1.2. Complexity of the CYP system in human tissues . . . . . . . . . . . . . . . . . . 3.1.3. Factors in¯uencing CYP function . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.2. Impact of CYP induction on drug therapy . . . . . . . . . . . . . . . . . . . . . . . . . . 3.2.1. Studies in nonhuman species and their impact on human drug metabolism 3.2.2. Molecular mechanisms of CYP induction . . . . . . . . . . . . . . . . . . . . . . . 3.2.3. Approaches to the study of CYP induction in human systems . . . . . . . . . 3.3. Pharmacokinetic drug interactions involving CYP inhibition. . . . . . . . . . . . . . 3.3.1. CYP inhibition mechanisms: relationship to enzymic function . . . . . . . . . 3.3.2. Reversible inhibition of CYP enzymes . . . . . . . . . . . . . . . . . . . . . . . . . . 3.3.3. Irreversible inhibition of CYP enzymes . . . . . . . . . . . . . . . . . . . . . . . . . 3.3.4. Structural features of drugs that inactivate CYPs by suicide processing . . 3.3.5. Structural features of drugs that inactivate CYPs by MI-complexation. . . 3.4. Clinical impact of understanding CYP induction and inhibition . . . . . . . . . . . 3.5. Acknowledgements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

4.

Cytochromes P450 in synthesis of steroid hormones, bile acids, vitamin D3 and cholesterol. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33 4.1. Steroid hormone biosynthesis. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33 4.1.1. Overview . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33 4.1.2. Adrenal cortex . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34 4.1.2.1. Pathways in the adrenal cortex . . . . . . . . . . . . . . . . . . . . . . . . . . . 34 4.1.2.2. Regulation in the adrenal cortex . . . . . . . . . . . . . . . . . . . . . . . . . . 36 4.1.3. Testis (Leydig cells) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 37 4.1.4. Ovary (Theca and Granulosa) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 38 4.1.5. Steroidogenesis in nonsteroidogenic cells . . . . . . . . . . . . . . . . . . . . . . . . 38 4.1.6. Congenital adrenal hyperplasia . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39

24 24 24 24 25 26 26 26 28 28 28 29 29 30 31 32 33

J.A. Hasler et al. / Molecular Aspects of Medicine 20 (1999) 1±137

4.2. 4.3. 4.4. 4.5. 4.6.

Bile acid synthesis . . . . . Vitamin D3 metabolism . Cholesterol biosynthesis. Conclusion . . . . . . . . . . Acknowledgments . . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

3

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

39 40 41 41 42

5.

Microsomal cytochrome P450 and eicosanoid metabolism. . . . . . . . . . . . . . . . . . . 42 5.1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42 5.2. NADPH-independent metabolism of eicosanoids . . . . . . . . . . . . . . . . . . . . . . 44 5.3. NADPH-dependent metabolism of eicosanoids . . . . . . . . . . . . . . . . . . . . . . . 45 5.4. Metabolism of arachidonic acid: the ``P450 arachidonic acid monooxygenase pathway'' . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 46 5.4.1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 46 5.4.2. Reactions catalyzed . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 47 5.4.2.1. AA /ÿ1 hydroxylase reaction . . . . . . . . . . . . . . . . . . . . . . . . . . . . 48 5.4.2.2. The arachidonic acid epoxygenase reaction . . . . . . . . . . . . . . . . . . . 49 5.4.3. Functional signi®cance of the AA monooxygenase metabolites . . . . . . . . 52 5.5. Conclusion, unresolved issues, future perspectives . . . . . . . . . . . . . . . . . . . . . 54 5.6. Acknowledgments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 54

6.

E€ects of disease on expression and regulation of CYPs . . . . . . . . . . . . . . . . . . . . 6.1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6.2. E€ects of liver disease . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6.2.1. Changes in total hepatic P450 levels . . . . . . . . . . . . . . . . . . . . . . . . . . . 6.2.2. Changes in expression of individual CYPs . . . . . . . . . . . . . . . . . . . . . . . 6.2.2.1. CYP1A . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6.2.2.2. CYP2C . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6.2.2.3. CYP3A . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6.2.2.4. CYP2E1 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6.2.2.5. Other CYPs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6.3. What is the mechanism for altered expression of hepatic CYPs in liver disease? 6.3.1. Changes in protein turnover . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6.3.2. Changes in mRNA levels . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6.3.3. Animal models of cirrhosis. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6.3.4. Portosystemic shunting (portal bypass) . . . . . . . . . . . . . . . . . . . . . . . . . 6.3.5. Liver regeneration . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6.3.6. Bile duct ligated rats . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6.4. E€ects of infections, vaccinations and in¯ammatory mediators . . . . . . . . . . . . 6.4.1. Endotoxin and TNF-a . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6.4.2. Interleukins. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6.4.3. Interferons . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6.4.4. E€ects on enzyme induction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6.4.5. Clinical relevance . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6.5. Thyroid disease. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6.6. Pituitary disease, and growth hormone therapy . . . . . . . . . . . . . . . . . . . . . . . 6.7. Other diseases. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6.8. Summary: clinical signi®cance and future directions . . . . . . . . . . . . . . . . . . . .

55 55 55 55 57 57 58 58 59 59 60 60 60 61 64 64 65 65 66 66 66 67 67 67 68 69 70

4

J.A. Hasler et al. / Molecular Aspects of Medicine 20 (1999) 1±137

7.

Cytochromes P450 and cancer . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7.1. Bioactivation of carcinogens . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7.1.1. Xenobiotics. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7.1.2. Endogenous compounds. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7.2. Susceptibility to cancer . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7.2.1. Lung cancer . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7.2.1.1. CYP1A1 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7.2.1.2. CYP2D6 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7.2.1.3. CYP2E1 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7.2.1.4. Other CYPs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7.2.2. Upper aerodigestive cancer. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7.2.3. Liver cancer . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7.2.4. Gastric cancer . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7.2.5. Breast cancer . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7.3. Metabolism of anticancer drugs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7.4. Expression of CYPs in tumors and chemosensitivity of cancers. . . . . . . . . . . . 7.5. Gene therapy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

8.

Cytochromes P450 as targets to autoantibodies in immune mediated diseases . . . . . 84 8.1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 84 8.2. Diseases in which autoantibodies against a CYP were characterized and in which no toxic agent was obviously implicated . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 85 8.2.1. Autoimmune polyglandular syndrome type 1 . . . . . . . . . . . . . . . . . . . . . 85 8.2.2. Autoimmune hepatitis type 2 (AIH2) . . . . . . . . . . . . . . . . . . . . . . . . . . 88 8.2.3. Genetics of AIH2 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 90 8.3. Diseases in which autoantibodies against a CYP were identi®ed and in which a drug was identi®ed as causative agent . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 91 8.3.1. Dihydralazine-induced hepatitis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 91 8.3.2. Tienilic acid-induced hepatitis. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 92 8.3.3. Halothane-induced hepatitis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 93 8.3.4. Hepatitis induced by other drugs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 94 8.3.4.1. Anticonvulsants . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 94 8.3.4.2. Miscellaneous other drugs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 95 8.4. Hypotheses for the appearance and role of autoantibodies in autoimmune hepatitis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 95 8.4.1. Modi®ed proteins fool the self-tolerance immune system . . . . . . . . . . . . . 95 8.4.2. Molecular mimicry . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 97 8.5. Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 99

70 71 71 75 76 76 76 78 79 79 79 80 80 80 80 82 84

References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 100

J.A. Hasler et al. / Molecular Aspects of Medicine 20 (1999) 1±137

5

1. An introduction to the cytochrome P450s 1.1. Introduction The biochemistry of mammalian cellular metabolism is a complex array of enzymes and metabolic products whose purpose is to generate energy for life and protect ®delity for DNA replication. For humans these activities occur in a hostile environment where continued exposure to a wide variety of foreign chemicals (xenobiotics) is commonplace. Plants and animals that serve as food sources consumed by humans are rich in their diversity of chemical constituents; the current life style of humans fosters the use of disposable items leading to pollution; industrial advances are often associated with a proliferation of new synthetic chemicals; and the growth and acceptance of the pharmaceutical industry has introduced the general use of therapeutic drugs and other xenobiotics, to name but a few examples. As eukaryotic cells evolved and the phylogenetic diversity of organisms developed, mechanisms have emerged which serve to protect animals (and plants) against chemical insults. One of these protective systems that we recognize today contains as its centerpiece a unique hemeprotein, called cytochrome P450 (P450). Members of this super-family of oxygen-reacting hemeproteins are remarkable by their promiscuity of action and their widespread distribution in biology. The P450 proteins are an expression of a gene super-family (CYP's) that currently contains more than 40 di€erent members in humans (and 1000 in biology). In humans (and most other mammals) the P450s play critical roles by catalyzing reactions functional in: (a) the biosynthesis of steroid hormones; (b) the metabolism of xenobiotics to reactive metabolites (free radicals) that interact with cellular macromolecules (DNA, RNA, proteins) or undergo detoxication by reaction with cellular constituents such as glutathione; (c) the oxidation of unsaturated fatty acids to intracellular messengers; (d) the stereo- and regio-speci®c metabolism of fatsoluble vitamins; and on and on. The purpose of this brief chapter is to present an introductory overview of the P450 hemeproteins with emphasis on their essential role in human tissues. 1.2. What are P450s? P450s are intracellular hemeproteins that ``activate'' molecular oxygen for the oxidative metabolism of a great variety of lipophilic organic chemicals. In eukaryotic cells the P450s exist as membrane-bound hemeproteins, each containing about 500 amino acids with iron-protoporphoryrin IX as the prosthetic group. What sets the P450s apart from other cellular hemeproteins is the role of a thiol-group from a cysteine of the protein which serves as a ligand to the heme-iron. Most hemeproteins in mammals (e.g., hemoglobin, cytochrome b, peroxidases) have a nitrogen from the imidazole group of histidine which serves as a similar ligand. The role of the thiolgroup as a ligand alters the electron density of the resonant porphyrin ring of the heme thereby providing an electronic center for the activation of molecular oxygen.

6

J.A. Hasler et al. / Molecular Aspects of Medicine 20 (1999) 1±137

1.3. History and spectrophotometric characterization The cytochrome P450s were ®rst recognized by Martin Klingenberg (Klingenberg, 1958) who was studying the spectrophotometric properties of pigments in a microsomal fraction prepared from rat livers. When a reducing agent (sodium dithionite) was added to diluted microsomes previously gassed with carbon monoxide, a unique spectral absorbance band with a maximum at about 450 nm appeared (Fig. 1). This absorbance band with a maximum at 450 nm is unique amongst hemeproteins and serves as the signature of P450 proteins (the name P450 is derived from the property of this pigment with an absorbance band maximum at 450 nm). This optical absorbance characteristic proves to be a fortuitous property useful in spectrophotometrically identifying and quantifying P450s since no other mammalian hemeproteins (except nitric oxide synthase) signi®cantly absorb light at this wavelength.

Fig. 1. Spectrophotometric identi®cation of P450s. Microsomes were prepared from livers of rats treated with phenobarbital, 3-methylcholanthrene, or untreated. Microsomes were diluted in 0.1 M potassium phosphate bu€er to a concentration of 1 mg of protein per ml. The diluted samples were reduced with a small amount of sodium dithionite, divided equally into two spectrophotometer cuvettes and the contents of one cuvette gassed for 30 s with a stream of carbon monoxide gas. The di€erence spectrum was recorded using an Aminco-Chance DW2 spectrophotometer.

J.A. Hasler et al. / Molecular Aspects of Medicine 20 (1999) 1±137

7

P450 remained a spectrophotometric curiosity until Omura and Sato (1962, 1964a,b) characterized the pigment as a hemeprotein. This ®nding was soon followed by the report by Estabrook et al. (1963) demonstrating the role of adrenal cortex P450 in the 21-hydroxylation of progesterone using the classic photochemical action spectrum technique developed by Otto Warburg (1949). By applying the same methods, Cooper et al (1965) provided evidence that P450's present in liver microsomes play a central role in the metabolism of drugs and other xenobiotics. Descriptions of these seminal studies are presented in historical reviews by Cooper (1973) and Estabrook (1995). 1.4. Biochemical properties Today we have a vast knowledge of the P450s. Approximately 1000 genes for di€erent P450s have been cloned, sequenced and expressed in heterologous expression systems (see Nelson et al., 1996 and the Internet Homepages of Nelson, 1999 and Degtiarenko, 1999). Many of the reactions catalyzed by P450s have been characterized showing the great diversity of action of these remarkable catalysts. Recently Coon et al. (1996) have identi®ed (Fig. 2) at least 40 di€erent types of reactions catalyzed by P450s (e.g., hydroxylation of aromatic and aliphatic chemicals, the N- and O-dealkylation of secondary and tertiary amines and O-methyl derivatives, the b-scission of hydroperoxides etc.). The number of chemicals that can serve as substrates metabolized by P450s is enormous and is certainly greater than 1000. The P450s are widely distributed in Nature with di€erent P450s present in plants, insects, some bacteria, yeast, and mammals. One can summarize the known properties of mammalian P450s as follows: (a) P450 proteins contain approximately 500 amino acids. A cysteine molecule located near the carboxy-terminus of the protein provides the essential thiol-ligand for the heme iron. The signature sequence for most P450's containing this cysteine is FxxGxxxCxG (Fig. 2) . The amino-terminus of the protein is rich in hydrophobic amino acids and is believed to act as a domain for binding the protein to membranes.

Fig. 2. The diversity of P450s. Modi®ed from Coon et al. (1996).

8

J.A. Hasler et al. / Molecular Aspects of Medicine 20 (1999) 1±137

(b) The P450s catalyze the NADPH and oxygen dependent oxidative transformation of a large number of di€erent chemical compounds (Fig. 2). In general a speci®c P450 will catalyze the metabolism of a limited number of chemical structures (such as steroids and fatty acids) while other P450s have a broad substrate speci®city suggesting a role for a unique ``active site geometry'' for a P450. (c) In tissues such as liver, intestine and the cortex of the adrenal gland, the concentration of P450s greatly exceeds the concentration of other hemeproteins (such as mitochondrial cytochromes). Indeed, in rat liver the concentration of P450s can be as high as 50 lM (i.e. 50 nmole per gram wet weight of tissue), thereby representing more than one percent of the protein of the liver; in the endoplasmic reticulum of the liver cell P450s can make up over twenty percent of the protein of this membrane fraction. (d) P450s are distributed in almost every organ of the human body ± although the type of P450 in a tissue appears to be speci®c. (e) The cellular expression of many P450s is regulated by transcription factors which become activated during exposure to various chemicals. The ability of a chemical to serve as an ``inducer'' is generally linked to a Family of P450s, e.g., polycyclic aromatic hydrocarbons will induce one type of P450s while barbituates will induce a di€erent type of P450. 1.5. What do P450s do? The P450s are members of the class of enzymes called oxygenases. Speci®cally, the P450s are monooxygenases (Hayaishi, 1962) or mixed function oxidases (Mason, 1957). In most instances mammalian P450 enzymes catalyze reactions for the oxidative conversion of a chemical following the equation illustrated in Fig. 3. Two electrons originating from NADPH are transferred to the hemeprotein by a ¯avoprotein (or a ¯avoprotein/iron sulfur protein) in the presence of an organic chemical and molecular oxygen. The organic chemical is oxidized and one atom of molecular

Fig. 3. The equation for P450-dependent mixed-function oxidase (oxygenase) reactions and the two types of electron transport carrier systems functional with di€erent P450s depending on their sub-cellular localization.

J.A. Hasler et al. / Molecular Aspects of Medicine 20 (1999) 1±137

9

oxygen is incorporated into the chemical product. In general, P450s undergo a cyclic series of reactions where: (a) ferric form of the hemeprotein initially reacts with a molecule of a chemical to form a complex; (b) the ferric P450-substrate complex is reduced by an electron transported from NADPH; (c) the ferrous-substrate complex reacts with molecular oxygen to form a ternary complex of ferrous P450-substrateoxygen. Carbon monoxide can compete with molecular oxygen resulting in the formation of a carbon monoxide complex of ferrous P450. As discussed above carbon monoxide reacts with reduced P450 producing an absorbance band with a maximum at 450 nm in the optical spectrum (the ®ngerprint characteristic of a P450, see above); (d) the ternary complex of ferrous P450-organic chemical substrate-molecular oxygen is further reduced by a second electron transferred from NADPH. This generates a two electron reduced intermediate (that has not yet been fully identi®ed) where a molecular rearrangement occurs and the chemistry of oxygen incorporation into the substrate chemical results; (e) the complex of ferric P450 and oxidized chemical product dissociate regenerating the uncomplexed ferric P450 which can participate in the metabolism of another molecule of chemical. In simplist terms ± the P450s serve as the common interaction site for the chemical substrate to be oxidized, electrons donated from reduced pyridine nucleotides (NADPH or NADH), atmospheric oxygen, and protons contributed by the solvent so that chemistry involving hydrogen abstraction, oxygen activation, and speci®c stereo- and regio-oxygenation can occur. Central to the operation of the P450 cycle is the need to provide electrons from NADPH. As shown in Fig. 3, mammalian tissues have two types of electron transport systems operative for di€erent P450s. One type is located in the mitochondria of some cells and consists of an FAD-containing reductase and an iron-sulfur protein (called ferredoxin or adrenodoxin). This mini-electron transport system is similar to that found in bacteria where a P450 may be functioning to break down chemicals as an energy source for growth. The second type is associated with the endoplasmic reticulum of many cell types where a unique ¯avoprotein containing both FAD and FMN as cofactors functions. The latter type is frequently referred to as the microsomal-type of P450 system. 1.6. The nomenclature of the P450 super-family Currently about forty di€erent P450s have been identi®ed in humans although recent results characterizing the human genome suggests there are at least ten additional P450s in humans that remain to be identi®ed. Some examples of reactions catalyzed by human P450s are identi®ed in Fig. 4. Included are reactions required for the (a) conversion of cholesterol to androgens, estrogens, gluco- and mineralocorticoids; (b) for the synthesis and degradation of prostaglandins and other unsaturated fatty acids; (c) the conversion of vitamins to their active forms; (d) the metabolism of cholesterol to bile acids; (e) and a large number of reactions involved in the metabolism of xenobiotics. Clearly the P450's play a central role in cellular metabolism and the maintenance of cellular homeostasis. When studies of P450 were ®rst undertaken in the early 1960s it was not recognized that so many di€erent P450s existed. The advent of molecular biology

10

J.A. Hasler et al. / Molecular Aspects of Medicine 20 (1999) 1±137

Fig. 4. Tissue and sub-cellular distribution of some human P450s and examples of reactions catalysed.

provided techniques that permitted one to identify and characterize the plethora of P450s that we now know. A major milestone in research on P450s was the ®rst cloning of a P450 by Fujii-Kuryama and his colleagues (Fujii-Kuriyama et al., 1982). In the 1980's a burst of new information was published identifying the unique sequences and enzymatic properties of individual mammalian P450s. This deluge of information continues today as interest in P450s present in plants, insects, and other areas of biology attract those working in this area of science. The proliferation of the P450s identi®ed the need for developing a logical means of categorizing these hemeproteins so that a common and readily understood language was available to permit scienti®c communication. Nebert et al. (1987) published a recommended nomenclature for the P450 gene super-family. This classic article brought order from the jargon-driven chaos of trivial names that previously confused most workers. The P450s were catagorized by Families and sub-Families based on the principle that ``a P450 protein sequence from one gene family is de®ned as usually having < 40% resemblance to that from any other family''. Sub-Families grouped together proteins of < 60% sequence similarity. When these principles are applied to the known human P450s one obtains a structured relationship as shown in Fig. 5. Examination of this method of categorizing P450s shows that there are at least sixteen families of P450s and many sub-Families. Of interest to note is the observation that over half (nine) Families of human P450s are associated with cholesterol and steroid hormone metabolism. Four Families of P450s are located with mitochondria and therefore use the mini-electron transport chain containing an iron sulfur protein. However, the largest number of human P450s are in the Families CYP 1 to CYP4 with CYP2 containing as many as eight sub-Families. Fig. 5 also indicates the primary substrates for the P450s associated with a speci®c family. It is proposed that there are about ten P450s in human tissue that have yet to be identi®ed. Understanding the role of these unknown P450s remains as a challenge for the future.

J.A. Hasler et al. / Molecular Aspects of Medicine 20 (1999) 1±137

11

Fig. 5. Division of the currently known 40 human P450 enzymes into families.

1.7. The key role of P450s in human biology and medical therapeutics The P450s participate in many key roles essential for maintaining the health and well being of an individual. Currently greatest research emphasis focuses on the role P450s play in the metabolism of drugs and other xenobiotics. The early work of James and Elizabeth Miller focused attention of the ability of some chemicals to cause cancer. Central to these studies of chemical carcinogenesis is the hypothesis that a covalent binding of chemicals to cellular macromolecules, such as DNA, RNA, and proteins, is a key step in the multi-stage process leading to tumor formation. In many instances the formation of chemicals with the high electrophilicity required to form adducts with DNA occurred during oxidative metabolism by P450s. The generation of the diol-epoxide of polycyclic aromatic hydrocarbons (benzo(a)pyrene) and the metabolism of nitrosamines are examples where P450s of the CYP1A family are key to the generation of reactive metabolites. A second pathway by which P450s can contribute to the alteration of cellular macromolecules leading to a toxic response or to cellular damage is via the formation of highly reactive oxygen free radical species, such as superoxide or hydroxyl radicals that may arise as by-products of P450 reactions. A major challenge today is to develop methods that can be used to predict the generation of toxic (or carcinogenic) metabolites formed during metabolism by P450s. Of great interest for humans is understanding the varied response to a therapeutic drug. It has been recognised for many years that some individuals are rapid metabolizers of some drugs, while other individuals are slow metabolizers. The varied response is due, in many instances, to di€erences in the amount of P450 expressed in the intestine and the liver. The in¯uence of diet, prior exposure to other drugs which may serve as inducers, and personal habits such as smoking and alcohol consumption, are causes for changing the level of expression of P450s. More recently, in this

12

J.A. Hasler et al. / Molecular Aspects of Medicine 20 (1999) 1±137

era of genetic medicine, the presence of polymorphic forms of P450s with modi®ed activities, have been identi®ed and may serve as the basis of di€erences in the reponse to a drug. The magnitude of this problem is illustrated in a recent article in the Journal of the American Medical Association (Lazarou et al., 1998) where the authors estimate that, in 1994, 106,000 people died (0.32% of the hospitalized population) and another 1.3 million Americans were injured (6.7%) as the result of adverse reactions to properly prescribed medications. 1.8. The future Clearly, the P450 enzymes are important to human health. We are still at the beginning in our understanding of their many roles in human homeostasis. A challenge for the future will be to better de®ne their role in metabolism and to develop methods for determining their levels of expression and in vivo activities. There are a large number of reviews written about P450s and each year brings a new group of meetings and symposia discussing advances in P450 research. The present volume contains a number of excellent articles representative of the current interest and research on speci®c types of P450s and their functions. 2. Pharmacogenetics of cytochromes P450 2.1. Introduction Metabolism of most drugs in¯uences their pharmacological and toxicological e€ects. Drugs particularly a€ected are those with a narrow therapeutic window and which are subject to considerable ®rst pass metabolism (Thummel et al., 1997). Much of the interindividual and interethnic di€erences in e€ects of drugs is now attributable to genetic di€erences in their metabolism. Mutations in a gene coding for a drug metabolising enzyme can give rise to enzyme variants with higher, lower, or no activity (functional mutations) or may have no e€ect on enzyme activity. If the mutant allele occurs with a frequency of at least 1% in the normal population and causes a di€erent drug response or phenotype, this phenomenon is termed a pharmacogenetic polymorphism (Meyer, 1994). Polymorphisms are detected by observing di€erences between individuals or populations in the function of the enzyme (phenotype) and/or in the gene for the enzyme (genotype). In phenotyping studies in vivo, typically a probe drug is administered to subjects and the metabolism of the drug is assessed by determination of parent drug to metabolite ratios in the urine or blood. Alternatively, if tissue samples are available, phenotyping can include microsomal enzyme activity determinations for speci®c substrates, CYP enzyme protein levels detected by immunological methods and CYP mRNA levels measured by various nucleic acid hybridisation methods or by reverse transcription coupled with the polymerase chain reaction (RT-PCR). In genotyping studies, mutations in genes are usually detected by use of polymerase chain reaction (PCR) methods, restriction fragment length

J.A. Hasler et al. / Molecular Aspects of Medicine 20 (1999) 1±137

13

polymorphism (RFLP) analysis, or single-strand conformation polymorphism (SSCP) analysis, followed by direct sequencing. Genetic mutations characterised include deletion of a whole gene, point mutations within genes, deletion or insertions of fragments of DNA within genes, and multiple copies of genes, leading to absent, de®cient or enhanced enzyme activity (Linder et al., 1997; Hong and Yang, 1997). In some cases, however, genetic variants have been identi®ed which have no functional signi®cance and this highlights the importance of being able to link a genetic variant with a variant phenotype. Apparently inconsistent results between phenotyping and genotyping studies may arise as a result a lack of complete speci®city of some of the probe drugs used e.g. there are suggestions of a lack of speci®city of chlorzoxazone as a probe of CYP 2E1 and tolbutamide as a probe for CYP 2C9 (Daly, 1995). A common stategy for assessing the possible functional signi®cance of mutations identi®ed in CYPs involves studies of the metabolism of speci®c substrates in vitro using the mutated cDNA in a heterologous expression system. Polymorphisms of drug metabolism divide a population into at least two phenotypes, extensive and poor metabolisers (Meyer, 1994) giving rise to a bimodal frequency distribution for markers of the extent of metabolism. Polymorphisms have been detected in many drug metabolising enzymes, including the cytochromes P450 (CYP), at both genotypic and phenotypic levels. While some allelic variants, such as the CYP2D6*5 (gene deletion), are common to all populations studied (i.e., African, Caucasian and Oriental), others seem to be characteristic for a particular population e.g., CYP2D6*10 in Orientals (Johansson et al., 1991, 1994; Yokota et al., 1993) and CYP2D6*17 in Africans (Masimirembwa et al., 1996a). The type and prevalence of allelic variants present in an individual or in a population will in¯uence the pharmacological and toxicological e€ects of drugs, toxins and carcinogens leading to interindividual and interethnic di€erences in e€ects of drugs and other xenobiotics (Kalow and Bertilsson, 1994). This becomes important for individual drug therapy, for clinical trials that are appropriate for a particular ethnic group and also in searching for possible relationships between genotypes and susceptibility to cancer and other diseases linked to xenobiotic metabolising enzymes. The possible relationship of genetic variants of cytochromes P450 and cancer susceptibility is reviewed in Chapter 7. The cytochromes P450 for which phenotypic and/or genotypic polymorphisms have been described include CYPs 1A1, 1A2, 2A6, 2C9, 2C19, 2D6, and 2E1. CYP 3A4 is subject to wide inter-individual variation but no data exist to support a possible genetic polymorphism for this cytochrome P450. Rather induction and inhibition of this enzyme is suggested to be responsible for the variations observed (See Chapter 3). Table 1 lists characterisitics of the major human cytochrome P450s and important substrates for each of these enzymes. It would appear from the numbers and frequencies of the di€erent functional alleles so far identi®ed for di€erent cytochrome P450s that CYP2D6 is the least conserved whilst others, such as CYP2E1 are functionally well conserved (Hu et al., 1997). The two most well studied and best characterised CYP polymorphisms are those for debrisoquine 4-hydroxylase (CYP2D6) and S-mephenytoin hydroxylase (CYP2C19) which will be dealt with here in some detail whilst the polymorphisms of the other CYPP450s will be covered in outline.

14

J.A. Hasler et al. / Molecular Aspects of Medicine 20 (1999) 1±137

Table 1 Characteristics of the major human cytochrome P450s (Linder et al., 1997; Daly, 1995; Thummel et al., 1997; Gonzalez, 1997) Cyt. P450

Approx. % of liver

Polymorphic CYP

Signi®cant for:

Representative substrates

First pass metabolism of drugs

Metabolism of carcinogens

CYP1A1

ÿ

Yes

No

Yes

CYP1A2

13

Yes

Yes

Yes

CYP2A6 4 9 CYP2C9 > > > > = 18 > > > > ; CYP2C19

Yes Yes

No Yes

Yes No

Yes

Yes

No

CYP2D6

2

Yes

Yes

No

CYP2E1

7

Yes

No

Yes

CYP3A4

29

No

Yes

Yes

Carcinogenic polycyclic aromatic hydrocarbons, e.g., benz[a]pyrene Arylamines, nitrosamines, a¯atoxin B1 , ca€eine, paracetamol, theophylline, imipramine, ¯uvoxamine Coumarin, nicotine Tolbutamide, ibuprofen, mefenamic acid, tetrahydrocannabinol, losartan, diclofenac7 S-mephenytoin, amitriptyline, diazepam, omeprazole, proguanil, hexobarbital, propranolol, imipramine Debrisoquine, metoprolol, sparteine, propranolol, encainide, codeine, dextromethorphan, clozapine, desipramine, haloperidol, amitriptyline, imipramine Ethanol, nitrosamines, paracetamol, chlorzoxazone, halothane Erythromycin, ethinyl estradiol, nifedipine, triazolam, cyclosporine, amitriptyline, imipramine, a¯atoxin B1

2.2. The CYP2D6 (debrisoquine hydroxylase) polymorphism CYP2D6 has been associated with the metabolism of over 50 clinically important drugs (Fromm et al., 1997). The capacity to metabolise probe drugs (e.g., debrisoquine to 4-hydroxydebrisoquine) was found in many studies to be bimodally distributed in Europeans and other Caucasian populations and the enzyme responsible for this metabolism was subsequently found to be CYP2D6 (Meyer, 1994). Phenotyping investigations in Caucasians have consistently shown that 5±10% of the

J.A. Hasler et al. / Molecular Aspects of Medicine 20 (1999) 1±137

15

population are poor metabolisers (PMs) with high metabolic ratios for probe drugs, whilst in Orientals only 0±1% of populations are PMs (Meyer, 1994; Bertilsson and Dahl, 1996; Fromm et al., 1997). Studies in African populations have, however, shown a great variation in prevalence of PMs ranging from 0% to 19% (cf. Masimirembwa and Hasler, 1997) presumably an indication of the ethnic heterogeneity on the continent. In general, black Africans (mostly west, central and southern African people) showed a 0±2% prevalence of PMs. Besides variation in the proportion of poor metabolisers in di€erent ethnic groups, the mean value of metabolic ratios in extensive metabolisers also varies between di€erent populations studied. The mean value of metabolic ratios in Oriental extensive metabolisers is higher than that in Caucasians, indicating a lower CYP2D6 activity in Orientals (Bertilsson et al., 1992). This tendency of Oriental populations to have higher metabolic ratios has been termed the ``right shift'' in the frequency distribution and is also characteristic of most African populations phenotyped. This can be illustrated by the median values of metabolic ratios for debrisoquine which is about 0.6 in Caucasians and 1.0 in Orientals. In African populations, the median varies from 0.8 to 1.5, averaging 1.0 with the exception of Egyptians with a value of 0.6 (Masimirembwa and Hasler, 1997). Another interethnic di€erence apparent is the 1000-fold variation (0.01±10.0) in metabolic ratio in Caucasian extensive metabolisers with a group of people with very low metabolic ratios. Among Caucasian extensive metabolisers, this subgroup of ultrarapid metabolisers (MRs < 0.2) exists but this phenomenon has not been observed in Orientals and most African populations. Interestingly, Ethiopians and Egyptians have a subgroup of ultrarapid metabolisers (Mahgoub et al., 1979; Aklillu et al., 1996) (see below). The metabolism of the CYP2D6 probe drugs, sparteine, metoprolol, and debrisoquine cosegregate in Caucasian and Oriental populations. In some African populations, poor correlations have been reported (Masimirembwa and Hasler, 1997; Droll et al., 1998) and have been attributed to possible poor subject compliance during phenotyping, interfering drugs/dietary components and/or the existence of CYP2D6 variants with di€ering catalytic activity towards these compounds. The molecular genetic basis for the di€erent CYP2D6 phenotypes has been characterised very well for Caucasians and Orientals, and to a lesser extent for African populations with the discovery of at a number of allelic variants of CYP2D6 (Daly et al., 1996), the most common of which are listed in Table 2. Fifty three CYP2D6 alleles have been characterised in a European population (Marez et al., 1997). Using a combination of restriction fragment length polymorphism (RFLP) analysis and allele-speci®c ampli®cation by the polymerase chain reaction (PCR), over 95% of the Caucasians carrying variant alleles can be identi®ed (Broly et al., 1991; Marez et al., 1997). These genotyping techniques have been used in many population studies to characterise the frequency of known allelic variants (Table 2). The molecular genetic basis for the poor metaboliser status has been shown to be the occurrence of combinations of any of a number of defective alleles, the most common being the CYP2D6*4 (formerly CYP2D6B) bearing a splice-site mutation. Other less frequent detrimental alleles are the CYP2D6*3 (formerly CYP2D6A) with

5, 1, 2, 6, 9,

A,G,A del C®T C®T C®T G®C

ex 5, A del in3/ex4, G ® A gene deletion

ex ex ex ex ex

as above

wild type ex 6, C ® T ex 9, G ® T

Mutations of variants

2.0 23.0 5.0 5±10

1.5 5.0 0.0

1.0±2.0

Caucasian

0.0 0.8 5.7 0±1

0.0 50.0 0.0

0.0

Oriental

3.3

0.0 8.0

2.2

Canadian Inuit

3.5 1.0 1±2

3.0 3.0

10.4

Saudi Arabian

0.24±0.6 7.3±8.5 6.0±6.9 1.9

0.7 5.0 26.0

2.0

African American

0.0 2.0 4.0 2.0

0.0 5.0 34.0

1.0

Zimbabwean

0.0 1.2 3.3 2.0

0.0 8.6 9.0

16.0

Ethiopian

ex ˆ exon, del ˆ deletion, in ˆ intron. Data compiled from: Aklillu et al., 1996; Evans et al., 1993; Masimirembwa et al., 1993, 1996a; Jurima±Romet et al., 1998; McLellan et al., 1997; Nowak et al., 1997; Leathart et al., 1998.

a

Absent CYP2D6*3 CYP2D6*4 CYP2D6*5 % of PM

Decreased CYP2D6*9 CYP2D6*10 CYP2D6*17

Increased (CYP2D6*2)n >1

Normal CYP2D6*1 CYP2D6*2

CYP2D6 activity

Table 2 The major allelic variants of CYP2D6 and their frequency of distribution (as percentages) in some di€erent ethnic groupsa

16 J.A. Hasler et al. / Molecular Aspects of Medicine 20 (1999) 1±137

J.A. Hasler et al. / Molecular Aspects of Medicine 20 (1999) 1±137

17

a frame shift mutation and the CYP2D6*5 (formerly CYP2D6D) being a gene deletion (Gaedigk et al., 1991; Kagimoto et al., 1990). A number of rare variant alleles have been described e.g. CYP2D6*9 (formerly CYP2D6C) bearing a three base pair deletion leading to the production of a low activity enzyme but not resulting in the poor metaboliser status and found in Caucasians (Tyndale et al., 1991). The ultrarapid metabolisers in Caucasians have been found to have haplotypes with two or more functional CYP2D6 genes of the CYP2D6*2 (CYP2D6L) variant (Dahl et al., 1995; Johansson et al., 1993). The CYP2D6*2 variant has about the same activity as the wildtype CYP2D6*1 but has a tendency to duplicate or amplify. Subjects with the duplicated or ampli®ed gene have very high CYP2D6 activity and accounted for 40% of the ultrarapid metabolisers in a Swedish study (Dahl et al., 1995). The CYP2D6 locus in Orientals has a number of di€erences from that of Caucasians. First, the Orientals do not have the CYP2D6*3 and CYP2D6*4 mutant variants, hence the low prevalence of poor metabolisers (Table 2). The CYP2D6*5 exists at a frequency similar to that in Caucasians and Africans (Table 2) and accounts for the few poor metabolisers in Orientals (Johansson et al., 1991). A CYP2D6 mutant variant causing a 34Pro ® Ser amino acid change was discovered in Chinese and Japanese subjects. Subsequent studies showed that this variant, CYP2D6*10 (formerly CYP2D6J,Ch1) was the basis of diminished CYP2D6 activity in Oriental populations with the mutant gene resulting in the production of a low activity and unstable enzyme (Yokota et al., 1993; Johansson et al., 1994). The CYP2D6*10 exists at a high allele frequency of 50% in Oriental populations compared to its low frequency of 5% in Caucasians (Armstrong et al., 1994; Johansson et al., 1994) hence explaining the higher median of metabolic ratios in the former populations. No duplicated or ampli®ed haplotypes of the CYP2D6*2 gene were found in Oriental populations. In African populations, fewer genotyping studies have been done. Genotyping of Zimbabweans (Masimirembwa et al., 1993; Masimirembwa et al., 1996b), Ethiopians (Aklillu et al., 1996) and Tanzanians (Dandara et al,. 1998; Wennerholm et al., 1998) indicate the absence of the detrimental CYP2D6*3 allele and low prevalence of the CYP2D6*4 hence explaining the low prevalence of poor metabolisers in these populations (Table 2). The CYP2D6*5 variant occurred at a frequency similar to that found in all other ethnic groups genotyped so far. This implies that this allelic variant predates the evolutionary split of Caucasoid, Oriental and African populations. This is in contrast to the CYP2D6*3 and*4 which appear to be Caucasian speci®c. An exception to the general ethnic pattern for these alleles is found in African-Americans with the CYP2D6*3 allele being present at a frequency of 0.6% and CYP2D6*4 at 7.3% (Leathart et al., 1998) presumeably due to intermarriage with Caucasians. The CYP2D6*10 which explains the ``right shift'' in metabolic ratios (CYP2D6 low activity) in Chinese was found at a very low frequency of 5% in the Zimbabweans (Masimirembwa et al., 1996b) and Tanzanians (Dandara et al., 1998; Wennerholm et al., 1998) although it was found at a higher frequency (12.5%) in a study of Ghanaians (Droll et al., 1998) but the number of individuals in this latter study was low (n ˆ 21). Phenotyping and CYP2D6 gene sequencing investigations led to

18

J.A. Hasler et al. / Molecular Aspects of Medicine 20 (1999) 1±137

the discovery of a mutant variant of the CYP2D6 gene in Zimbabweans designated CYP2D6*17 (CYP2D6Z). The variant has three mutations causing 107Thr ® Ile, 206Arg ® Cys, and 486Ser ® Thr amino acid changes in CYP2D6. The CYP2D6*17 occurred at a high frequency of 34%, and showed a gene-dose correlation with diminished CYP2D6 activity with the 107Thre ® Ile appearing to be the principal mutation of this variant (Masimirembwa et al., 1996a). This allelic variant has also been found in the South African Venda and in Tanzanians at an allele frequency of 18±20% (Dandara et al., 1998; Wennerholm et al., 1998), in Ghananians at 15% (Droll et al., 1998) and in African-Americans at 26% (Leathart et al., 1998). In vitro studies with the CYP2D6*17 heterologously expressed in yeast showed that this variant has a low anity for CYP2D6 substrates (Oscarson et al., 1997). The CYP2D6*17 was not found in Swedish and Chinese subjects analysed (Masimirembwa et al., 1996a) and only one allele out of 1344 was found in another study of Europeans (Marez et al., 1997). Since the Zimbabwean population is genetically related to most central and southern African populations, the CYP2D6*17 variant was postulated to be the basis of diminished debrisoquine hydroxylase activity in these populations (Masimirembwa et al., 1996a,b). The presence of the CYP2D6*17 allele is associated with higher metabolic ratios for debrisoquine in phenotyping studies (Masimirembwa et al., 1996a,b; Leathart et al., 1998). Phenotyping results in most African populations have not shown any ultrarapid metabolisers. Genotyping of Zimbabweans, on the other hand showed two subjects having the duplicated CYP2D6*2 but these also carried the detrimental CYP2D6*4 mutation in the locus (Masimirembwa et al., 1993, 1996b). In a genotyping study done in Ghanaians, none had the duplicated or ampli®ed gene status (Asante-Poku et al., 1996). Phenotyping and genotyping studies in Ethiopians showed the most interesting admixture of CYP2D6 mutant alleles observed in one population so far (Aklillu et al., 1996). Firstly, the phenotyping studies showed a low prevalence of poor metabolisers which was accounted for by the infrequent occurrence of the CYP2D6*4 variant and absence of the CYP2D6*3 variant. Secondly, most of the subjects phenotyped as extensive metabolisers showed a distribution of debrisoquine metabolic ratios similar to that of Zimbabweans and other African populations, that is, higher mean metabolic ratios compared to Caucasians. Most of the subjects with diminished CYP2D6 activity were accounted for by the CYP2D6*10 and the CYP2D6*17 (9% allele frequency in each case) (Aklillu et al., 1996). Thirdly, like Egyptians and Caucasian populations, the phenotyping results showed a subgroup of ultra-rapid metabolisers (7%) with MR below 0.2. These subjects had multiple copies (3±5) of the CYP2D6*2 gene hence explaining the high CYP2D6 activity. It must be noted, however, that some subjects carrying the duplicated CYP2D6*2 and no other known detrimental mutations had relatively high MRs (Aklillu et al., 1996). Saudi Arabians, neighbours to Ethiopians, have also been shown to have a high frequency (18%) of duplicated CYP2D6*2 alleles (McLellan et al., 1997). The clinical impact of polymorphisms of CYP2D6 has been reviewed a number of times with special attention paid to cardiovascular and neuroactive drugs (DeVane, 1994; Bertilsson and Dahl, 1996, Fromm et al., 1997) because so many CYP2D6 substrates belong to these classes of drugs. In general, there is a 2±5 times di€erence

J.A. Hasler et al. / Molecular Aspects of Medicine 20 (1999) 1±137

19

between poor and extensive metabolisers in the capacity to metabolise CYP2D6 substrates; hence PMs obtain the same steady-state serum levels as EMs at doses which are 50±80% lower than those given to EMs (Brùsen, 1993). In panel studies with haloperidol and zuclopenthixol, PMs experienced more side e€ects than EMs at the assumed subtherapeutic doses leading to the need for administering half the dose to PMs in these studies. At these halved doses, the PMs still attained plasma concentrations comparable to the EMs taking a double dose (Dahl et al., 1991; Llerena et al., 1992). In panel studies with subtherapeutic doses of perphenazine, PMs experienced more adverse e€ects than EMs and at that dose, PMs had already attained drug plasma concentrations within the therapeutic range (Dahl-Puustinen et al., 1989). In a retrospective study of the CYP2D6 status in patients who had experienced acute neuroleptic-induced adverse e€ects, an overrepresentation of PMs was found (Spina and Caputi, 1994). It is known that antipsychotics have a very narrow therapeutic window and that usually relatively low concentrations are required for a therapeutic e€ect; higher concentrations only increase the risk of extrapyrimidal side e€ects without further increase in antipsychotic eciency (Hansen et al., 1981). If not identi®ed, PMs of debrisoquine may thus be treated with unnecessarily high doses of neuroleptics, resulting in an increased risk of disturbing side e€ects which may in turn lead to decreased patient compliance with the treatment (Shah, 1993). PMs are therefore at increased risk for the accumulation of drugs, resulting in drug-related toxicity or lack of therapeutic ecacy in cases where a pro-drug is metabolised to a pharmacologically active metabolite. As previously noted, there is a wide range of CYP2D6 activity amongst the EMs which could call for clinical drug dose adjustments. On one extreme, are the ultrarapid metabolisers and on the other, a subgroup of subjects with diminished (but not de®cient) CYP2D6 activity. The EMs as a whole are prone to possible drug-drug interactions involving the co-administration of drugs that may be substrates or inhibitors of CYP2D6. PMs are not subject to these interactions since they lack the enzyme altogether. Coadministration of an inhibitor of CYP 2D6, such as quinidine, and a psychoactive drug which is a CYP2D6 substrate would result in the patient experiencing side e€ects due to drug accumulation similiar to what is observed for PMs. Regarding relatively slow extensive metabolisers, we have noted already that Orientals and most Africans will, on average, metabolise CYP2D6 substrates at a slower rate compared to Caucasians as a result of mutations which lead to the ``right shift'' in the frequency distribution of metabolic ratios for probe drugs. From empirical observations of drug ecacy and side e€ects, clinicians in Oriental populations have been prescribing antipsychotics at lower doses than in Caucasians and it is now postulated that, besides possible di€erences in other pharmacokinetic parameters or receptor-drug interactions, the di€erence in metabolism of the drugs is the basis of lower prescriptions in the Chinese (Lin et al., 1991). As a result of such observations, the Japanese regulatory authorities do not depend on pharmacokinetic studies done in Caucasians to optimise drug doses but insist that they be done with Japanese subjects (Shah, 1993).

20

J.A. Hasler et al. / Molecular Aspects of Medicine 20 (1999) 1±137

No such studies or observations have been documented in native African populations. If the CYP2D6*17 variant is widespread in Africa, there may be a need to prescribe neuroleptics and antidepressants at lower doses on the basis that Africans metabolise these drugs at a lower rate than Caucasians. The increasing knowledge of the CYP2D6 phenotype and genotype status in di€erent populations should be used as a guide to carry out pharmacokinetic studies with a view to optimising drug doses. Many subjects with very high CYP2D6 activity as indicated by metabolic ratios of < 0.2, have a duplicated or ampli®ed mutant allele resulting in two or more functional genes and hence enhanced metabolic capacity. The clinical consequence of ultrarapid metaboliser status is that such patients will not attain therapeutic serum concentration of neuroleptics and antidepressants if prescribed at doses given to normal EMs. This is evident in case reports from Huddinge Hospital, Sweden (Bertilsson et al., 1993); one patient having a MR of 0.07 had to be treated with 500 mg nortriptyline daily (3±5 times the recommended dose) to attain therapeutic plasma levels whilst another patient had to be treated with 300 mg clomipramine daily (recommended dose is 25±150 mg) for successful therapy. On genetic analysis, both patients were found to have a duplicated CYP2D6*2 allele (Bertilsson et al., 1993). 2.3. The CYP2C19 (S-mephenytoin hydroxylase) polymorphism The genetic polymorphism of S-mephenytoin was discovered in the early 1980s (Kupfer and Preisig, 1984; Wilkinson et al., 1989). It was shown that the de®cient metabolism of this drug was inherited as an autosomal recessive trait distinct from the debrisoquine hydroxylation polymorphism. The polymorphic enzyme catalyses the enantiomer-selective 40 -hydroxylation of S-mephenytoin. The PM phenotype can be determined by the hydroxylation index in urine after a standard dose of racemic mephenytoin or measurement of the ratio of excretion of the S- and R-enantiomers. The enzyme was identi®ed and designated as CYP2C19 and its gene located on chromosome 10 (Goldstein and de Morais, 1994). CYP2C19 also metabolises hexobarbital, propanolol, omeprazole, impramine, and diazepam to varying extents (Bertilsson et al., 1996). Importantly in tropical countries, CYP2C19 metabolises proguanil to the active antimalarial metabolite cycloguanil (Ward et al., 1991). In phenotyping studies, the de®ciency in S-mephenytoin metabolism was observed in 2±6% of Caucasians (Bertilsson et al., 1996) and in 14±22% of Orientals (Wilkinson et al., 1989). On phenotyping 191 unrelated African-Americans (from middle Tennessee, USA), two (1.05%) were found to be PM (Edeki et al., 1996) while 4 out of 103 Zimbabweans (4%) and 8 out of 251 Tanzanians (7.5%) were observed to be PMs (Masimirembwa et al., 1995a; Herrlin et al., 1998). The S-mephenytoin hydroxylation phenotype has also been investigated among Ethiopians with 6 out of 114 subjects (5.2%) found to have S/R ratios indicative of the PM status (Persson et al., 1996). The principal molecular defect in poor metabolisers of mephenytoin is a mutation which creates an aberrant splice site which produces a premature stop codon and hence an inactive protein lacking the haem-binding region. This variant was

J.A. Hasler et al. / Molecular Aspects of Medicine 20 (1999) 1±137

21

designated CYP2C19*2 and was found to account for 83% of PMs in Caucasians and 75% of Japanese PMs (de Morais et al., 1994a). In Zimbabweans and Ethiopians, this variant accounted for 75% of the PM status (Masimirembwa et al., 1995a; Persson et al., 1996). A second mutant variant was discovered, CYP2C19*3, in Japanese PMs not carrying the CYP2C19*2 (de Morais et al., 1994b). The mutation in the CYP2C19*3 variant introduces a premature stop codon to produce a polypeptide containing only the ®rst 4 exons of the gene. The CYP2C19*3 accounted for the other 25% of PMs in Orientals but was absent or very rare in Caucasians, in African-Americans and in Zimbabweans implying that there were other yet to be discovered mutations to explain the other 20±25% of PMs in these populations (Edeki et al., 1996; Masimirembwa et al., 1995a). Recently, further defective allelic variants have been characterised, namely CYP2C19*4 (Ferguson et al., 1998) and CYP2C19*5A and 5B (Xiao et al., 1997; Ibeanu et al., 1998). In the study of Ibeanu et al. (1998), of 37 Caucasian poor metabolisers, the CYP2C19*2 allele accounted for 86.5% of the defective alleles, while CYP2C19*3, CYP2C19*4, and CYP2C19*5 accounted for 1.4%, 2.7% and 1.4% respectively. In Tanzanians, the allele frequencies of CYP2C19*2 and CYP2C19*3 were 17.9% and 0.6% respectively, which is similar to what is seen in Caucasians, although there was a lower rate of metabolism of mephenytoin and omeprazole in relation to the CYP2C19 genotype compared to Caucasian and Oriental populations (Herrlin et al., 1998). The clinical implications of the CYP2C19 status have not been as much explored as those of CYP2D6 possibly because fewer drugs are metabolised by CYP2C19 compared to CYP2D6. This is, however, an important polymorphism clinically as some of its substrates (Table 1) are widely used drugs and because up to 20% of people in Oriental populations lack the enzyme. Ghoneim et al. (1981) noted that Orientals had a lower clearance of diazepam than Caucasians. This can be explained by the high frequency of PMs and subjects heterozygous for mutant CYP2C19 alleles (Kalow and Bertilsson, 1994). A gene-dose e€ect has been observed in Zimbabweans and African-Americans where there were signi®cant di€erences in CYP2C19 activity between subjects homozygous wildtype and those heterozygous for the CYP2C19*2, the latter having higher S/R ratios (Edeki et al., 1996; Masimirembwa et al., 1995a). The high frequency of the PM status and low activity heterozygotes in Orientals could, therefore, explain why physicians in Asia routinely prescribe smaller doses of diazepam for Chinese than for Caucasians (Kumana et al., 1987). 2.4. Polymorphisms in other cytochromes P450 (CYPs 1A1, 1A2, 2A6, 2C9, and 2E1) CYP1A1, an inducible CYP, is important for conversion of carcinogenic polycyclic aromatic hydrocarbons to epoxides. A phenotypic polymorphism in inducibility was ®rst described in 10% of Caucasians who showed much higher CYP1A1 activity in lymphocytes after exposure to inducer than was observed in the rest of the study group (Kellerman et al., 1973). No clear molecular mechanism to explain this observation has been determined although it was suggested that it might arise from polymorphisms of the Ah receptor gene (Fujii-Kuriyama et al., 1995). A possible association of the high CYP1A1 inducibilty phenotype and genetic polymorphisms

22

J.A. Hasler et al. / Molecular Aspects of Medicine 20 (1999) 1±137

within the CYP1A1 promoter region and within the Ah receptor gene is still being investigated (Daly, 1995; Gonzalez, 1997; Smart and Daly, 1998). In genotyping studies, two closely linked polymorphisms of the CYP1A1 gene have been wellstudied in Caucasian and Oriental populations, being the 30 -¯anking region Msp1 site (m2 allele) and the exon 7 Ile-Val substitution (Val allele). The m2 allele is associated with increased inducibility in some studies whilst the Val allele appears to result in higher enzyme activity in vitro (Hayashi et al., 1991a, 1992). The m2 alleleVal allele combination is common in Japanese with an allele frequency of 31% but much less frequent in European populations (less than 12%) and not found in some African populations studied (Hayashi et al., 1991a; Gonzalez, 1997; Masimirembwa et al., 1998). Interestingly, a recent study shows a very high frequency of the m2 allele-Val allele combination of 75% in Mapuche South Amerindians of Chile but the functional signi®cance of this remains to be elucidated (Mu~ noz et al., 1998). There have been many studies of a possible relationship between these alleles and susceptibility to cancers (See Chapter 7). The hepatic CYP1A2 is another inducible enzyme closely related to CYP1A1 which is not expressed in the liver. A possible polymorphism for CYP1A2 has been suggested by a number of studies in which ca€eine is used as a phenotypic probe of enzyme activity. In particular, studies have indicated a trimodal distribution in Caucasians (Butler et al., 1992) and a bimodal distribution in Japanese (Nakajima et al., 1994). In the latter study, however, no genetic polymorphism has been detected in the CYP1A2 gene despite sequencing of DNA from individuals at extremes of the distribution of the CYP1A2 metabolic index (Nakajima et al., 1994). The use of ca€eine as a probe of CYP1A2 activity, however, is problematic as di€erent ratios of metabolites are used by di€erent workers and there is controversy over the interpretation of the studies (Notarianni et al., 1995). CYP2A6 is known to catalyse the 7-hydroxylation of coumarin and some evidence of a bimodality of its metabolism in vivo has been obtained from phenotyping studies (Cholerton et al., 1992; Rautio et al., 1992). After the organisation and structure of the CYP2A6 gene cluster was characterised (Ho€man et al., 1995), three alleles of the CYP2A6 gene were found, i.e., wild-type, CYP2A6*1, and variants, CYP2A6*2 (CYP2A6v1) and CYP2A6*3 (CYP2A6v2) (Fernandez-Salguero et al., 1995). CYP2A6*2 has a point mutation in codon 160 leading to a Leu-His amino acid change and CYP2A*3 has several alterations in exons 3, 6 and 8 generated apparently by gene conversion between CYP2A6 and CYP2A7. Both variants are considered to be inactive enzymatically (Yamano et al., 1990; Fernandez-Salguero et al., 1995) although the exact functional signi®cance of the two mutant alleles has not yet been determined. Similarly to other polymorphic cytochromes P450, the frequency of the variants appears to vary between ethnic groups (Fernandez-Salguero et al., 1995). Recently, a deletion mutation for CYP2A6 has been characterised in 2 out of 20 Japanese individuals for which liver samples were available (Nunoya et al., 1998). CYP2C9 is important in the metabolism of a number of drugs with narrow therapeutic indices, e.g. S-warfarin, tolbutamide and phenytoin (Goldstein and de Morais, 1994a). For CYP2C9, a number of genetic variants have been characterised

J.A. Hasler et al. / Molecular Aspects of Medicine 20 (1999) 1±137

23

including the two most common variants which produce intact enzyme with reduced enzymatic activity arising from amino acid substitutions that are at positions critical for activity. The wild-type CYP2C9 has Arg144 and Ile359 , with variants being CYP2C9*2 (with Cys144 ) and CYP2C9*3 (with Leu359 ) (Stubbins et al., 1996). It has been demonstrated in vitro in studies with heterologously expressed cDNA that there is decreased metabolic capacity for S-warfarin of CYP2C9*3 with con¯icting results seen for CYP2C9*2 (Kaminsky et al., 1993; Rettie et al., 1994; Haining et al., 1996; Sullivan-Klose et al., 1996). Investigations of the relationship between phenotype and the CYP2C9*3 variant show that there is signi®cantly reduced metabolic capacity for S-warfarin, tolbutamide, losartan and phenytoin compared to the wildtype CYP2C9*1 (reviewed in Takahashi et al., 1998). Further, a study has shown a dangerously exacerbated therapeutic response to normal doses of racemic warfarin in a patient homozygous for the CYP2C9*3 variant (Steward et al., 1997). A recent study showing very good correlation between in vitro enzyme activity in a yeast heterologous expression system and in vivo metabolism in humans suggests that the in vtro system could be very useful in predicting e€ects of the CYP2C9*3 variant in metabolism in vivo of other drug substrates (Takahashi et al., 1998). The clinical signi®cance of CYP2C9*2 is still not clear. Allele frequencies for these two variants are of the order of 6±12% for Caucasian populations (Sullivan-Klose et al., 1996; Stubbins et al., 1996; Yasar et al., 1998; Ackermann et al.,1998). In studies of Chinese and Japanese populations, CYP2C9*2 was not detected while CYP2C9*3 occurred at frequencies around 2% (Wang et al., 1995; Nasu et al., 1997). CYP2E1 is an ethanol inducible enzyme important for the metabolism of ethanol, paracetamol, N-nitrosamines and a number of organic solvents (Guengerich et al., 1991). Phenotyping studies with chlorzoxazone show a 4±5 fold variation in clearance of the drug in humans (Daly, 1995) and a 50-fold variation in the expression of CYP2E1 (Stephens et al., 1994) but there is no evidence of a bimodal distribution consistent with a metabolic polymorphism. A number of genetic polymorphisms in the CYP2E1 gene have been reported most of which are located in introns and appear to have no functional signi®cance (Daly, 1995; Grove et al., 1998). Two genetic polymorphisms of interest, however, in the CYP2E1 gene have been characterised on the basis of RFLP analysis (Hayashi et al., 1991b; Persson et al., 1993). A single point mutation (c2 allele) results in loss of an RsaI restriction site in the 50 ¯anking region of the CYP2E1 gene and leads to an increase in transcriptional activity, mRNA and protein expression in vitro compared to the wild-type (c1 allele) (Tsutsumi et al., 1994). A DraI restriction site in intron 6 de®nes allele C. The RsaI c2 allele is present at around 2% in Caucasian populations and around 20±28% in Oriental populations (Ingelman-Sundberg et al., 1992; Stephens et al., 1994). The clinical signi®cance of these mutations is not clear but many studies have investigated a role in susceptibilty to cancers (see Chapter 6) and in alcoholic liver disease. In the latter case, results have been con¯icting in both Japanese and Caucasian studies. In a recent large study investigating the e€ect of both alcohol dehydrogenase 3 (ADH3) and CYP2E1 polymorphisms, patients with the c2 allele had a signi®cant excess of the ADH3*2 allele (coding for the slower c2 ADH subunit) compared with patients and controls homozygous for the c1 allele (Grove et al., 1998). Patients with

24

J.A. Hasler et al. / Molecular Aspects of Medicine 20 (1999) 1±137

the c2 allele had increased risk of developing alcoholic liver disease at lower levels of alcohol consumption and presented 7 years earlier than patients homozygous for the c1 allele. Grove et al. (1998) conclude that the available data now strongly suggest that the c2 RsaI allele, although rare, is associated with an increased risk of advanced alcoholic liver disease in individuals who abuse alcohol. A further two functional allelic variants of the CYP2E1 gene (CYP2E1*2 coding an Arg76-His change) and CYP2E1*3 coding for a Val389-Ile change) have been described but only CYP2E1*2 a€ected activity of CYP2E1 in a COS-1 cell heterologous expression system (Hu et al., 1997). These variants were found to be rare in human populations investigated in the latter study. 3. Induction and inhibition of CYPs and implications for medicine 3.1. Introduction 3.1.1. Cytochrome P450 enzymes: functional importance in drug metabolism Most therapeutic agents are lipophilic and require metabolic processing before they can be eliminated from the body. In the absence of metabolic oxidation and conjugation drugs would be cleared more slowly, leading to their accumulation and toxicity. The cytochromes P450 (CYP) have a pivotal role in the oxidative conversion of drugs to polar products before elimination. An unusual property of hepatic CYPs is their catalytic versatility in the oxidation of many substrates. To some extent this is attributable to the multiplicity of the CYP system but individual CYPs usually oxidize a wide range of exogenous and endogenous substrates. 3.1.2. Complexity of the CYP system in human tissues CYPs are distributed ubiquitously but hepatic CYPs are signi®cant in drug oxidation. As outlined in Chapter 1, the present nomenclature groups CYPs on the basis of amino acid similarity (Nelson et al., 1996). CYPs involved in hepatic drug oxidation belong to three families (CYPs 1±3). CYP1A2 participates in the oxidation of paracetamol (acetaminophen) and theophylline in human liver (Gu et al., 1992; Patten et al., 1993). 2C subfamily CYPs, including CYP2C8 and CYP2C9, metabolize tolbutamide, phenytoin, torsemide and diclofenac (Guengerich, 1995). CYP2C19 and CYP2D6 are, respectively, the polymorphic S-mephenytoin and debrisoquine 4-hydroxylases. The list of other drugs substrates for these enzymes is expanding and includes omeprazole (Andersson et al., 1990), moclobemide (Gram et al., 1995) and proguanil (Birkett et al., 1994) (CYP2C19) and b-adrenoreceptor antagonists, sparteine and codeine (Distelrath and Guengerich, 1984; Koymans et al., 1992) (CYP2D6). Individuals who possess defective CYP2C19 or 2D6 alleles are termed poor metabolizers. The extensive metabolizer phenotype, by contrast, is catalytically competent. CYP2E1 oxidizes small molecules, including solvents and nitrosamines (Guengerich et al., 1991), and participates in the metabolism of general anesthetics such as sevo¯urane and halothane. Indeed CYP2E1 may convert halothane to a reactive metabolite that binds to hepatic proteins and has been implicated

J.A. Hasler et al. / Molecular Aspects of Medicine 20 (1999) 1±137

25

in halothane hepatotoxicity (Spracklin et al., 1997). There are at least three functionally important members of the CYP3A subfamily expressed in human liver. CYP3A4 is probably the major drug oxidase, and is active on cyclosporin, macrolide antibiotics, anticancer agents like taxol among others. This enzyme represents about half of the total human hepatic CYP (Shimada et al., 1994b) and CYP3A5 is expressed in 20% of the population (Wrighton et al., 1989; Guengerich, 1995). 3.1.3. Factors in¯uencing CYP function In most cases drugs are oxidized to several metabolites involving di€erent CYPs. However, if a drug is oxidized extensively by a single CYP, inhibition or induction would have a major impact on therapy with the drug. It is appreciated, for example, that the CYP2D6 poor metabolizer phenotype is more susceptible to toxicity during therapy with the antihypertensive agent debrisoquine (Eichelbaum, 1982). It has been shown that extensive metabolizer individuals are converted transiently to the poor metabolizer phenotype by administration of CYP2D6 inhibitors, e.g., selective serotonin-reuptake inhibitors (Brùsen et al., 1993) and antiparasitic agents (Masimirembwa et al., 1995b). An interesting corollary is that, if the drug interacts selectively with CYP2D6, then poor metabolizers are at lower risk. Most products of CYP oxidation are inactive, but there are exceptions. Some drug metabolites are pharmacologically active. For example, diazepam is oxidized by CYPs to nordiazepam and oxazepam (Fig. 6), which retain sedative properties (Greenblatt et al., 1980). It has also been found that some drugs are prodrugs that are activated by CYPs, e.g., proguanil is converted to the active agent cycloguanil (Fig. 6) in part by CYP2C19; this step is inhibited by the gastric proton pump inhibitor omeprazole (Funck-Brentano et al., 1997). In further instances, the products of CYP action on some drugs can produce reactive metabolites with cytotoxic properties. Examples include paracetamol (acetaminophen), whose toxicity is increased by CYP following oxidation to the N-acetylquinone imine that binds to

Fig. 6. Role of CYP in the conversion of the inactive prodrug proguanil to the active antimalarial agent cycloguanil, and in the conversion of the benzodiazepine diazepam to its active metabolite nordiazepam and oxazepam.

26

J.A. Hasler et al. / Molecular Aspects of Medicine 20 (1999) 1±137

tissue macromolecules (Patten et al., 1993), and dihydralazine, which is activated by CYP to a metabolite that acts as a hapten and against which an immune response is mounted, resulting in autoimmune hepatic injury (Bourdi et al., 1994). This chapter concentrates on the inhibition and induction of CYP-dependent processes in the clinical setting. These are processes encountered commonly during therapy that produce, respectively, signi®cant decreases or increases in the rate of drug biotransformation and elimination. In considering the impact of these processes it is essential to be aware that CYPs can generate drug metabolites with unusual biological properties. 3.2. Impact of CYP induction on drug therapy 3.2.1. Studies in nonhuman species and their impact on human drug metabolism CYP induction was ®rst described for polycyclic aromatic hydrocarbons (PAH), such as the carcinogen 3-methylcholanthrene, and sedative-hypnotics, such as phenobarbitone (Conney et al., 1956, 1960). PAH increased the metabolism of azo dyes and barbiturates markedly increased drug elimination. Over the last four decades many chemicals with similar e€ects on drug biotransformation have been identi®ed. Induction appears to be an adaptive response of the organism to an environment containing foreign chemicals that might otherwise produce toxicity. Induction involves the synthesis of new enzymes. The puri®cation of individual CYPs and the development of monospeci®c anti-CYP antibodies have enhanced the understanding of the induction properties of drugs toward CYPs (Dannan et al., 1983; Thomas et al., 1983; Waxman et al., 1985). Although most work in this area has been done in rodents, and the rat is highly responsive to induction, similar processes have been observed in patients during drug treatment and have clinical signi®cance. Enhanced awareness of the complexity of CYP genes has facilitated molecular interpretations of the mechanisms by which chemicals increase CYP expression. 3.2.2. Molecular mechanisms of CYP induction The induction of most CYPs occurs at the level of gene transcription, but CYP gene products are also stabilized under certain conditions, which also leads to increased hepatic CYP content. Perhaps the largest body of mechanistic information centers around the induction of CYPs 1A by the Ah (aromatic hydrocarbon) receptor (Poland et al., 1976). Bulky planar chemicals bind with high anity to this intracellular protein which undergoes activation to a DNA-binding state. Subsequent work in cell lines established the role for an auxiliary protein, the aromatic hydrocarbon receptor nuclear translocator (Arnt), which mediates the interaction of the PAH-liganded Ah receptor with DNA (Ho€man et al., 1991). In rodents the classical CYP inducer phenobarbital upregulates CYPs from the 2B, 2C and 3A subfamilies (Waxman et al., 1985). A receptor for phenobarbital has not been identi®ed and the mechanism of induction remains a subject of continuing research. To some extent species di€erences in the potency of induction

J.A. Hasler et al. / Molecular Aspects of Medicine 20 (1999) 1±137

27

by phenobarbital and other inducers have complicated the issue, e.g., 1,4-bis[2(3,5-dichloropyridyloxy)]-benzene induces murine CYPs 2B but is inactive in the rat (Poland et al., 1981). There are also species di€erences in some of the molecular processes of apparent importance in induction. In rat liver phosphorylation/dephosphorylation in¯uence the extent of induction by phenobarbital (Sidhu and Omiecinski, 1998). In bacteria, a speci®c gene response element termed a barbie box, has been implicated in the induction of CYPs by barbiturates (He and Fulco, 1991). Similar elements are present in the regulatory regions of mammalian CYP genes but do not appear to be involved directly in induction (He and Fulco, 1991). Recently, the participation of a phenobarbital-responsive module, a large region of DNA containing several important arrangements of nucleotides, in induction of the mouse CYP2B10 gene has been established (Honkakoski and Negishi, 1997). Induction of CYP3A4 is highly signi®cant for drug therapy because of the role of this enzyme in the metabolism of most drugs. Apart from phenobarbital, other CYP3A inducers include rifampicin, carbamazepine, macrolide antibiotics and dexamethasone. Early studies indicated the induction of CYPs 3A in experimental animals by glucocorticoids (Schuetz et al., 1984). However, the potency of nonsteroidal inducers of CYP3A, such as rifampicin, troleandomycin and metyrapone, argued against a role for the glucocorticoid receptor (Quattrochi et al., 1995; Wright et al., 1996). Most recently, Kliewer et al. (1998) cloned and characterized a pregnane receptor (PXR) that bound both dexamethasone and pregnenolone 16a-carbonitrile and activated transcription of rat CYPs 3A. Apart from transcription, drugs like troleandomycin that bind tightly to CYPs 3A and increase CYP3A expression in hepatocytes by blocking protein degradation (Watkins et al., 1986). CYP2C9 is also responsive to rifampicin, although less so than CYP3A4 (Guengerich, 1995). The mechanism of drugs-activated CYP2C9 transcription is unclear but liver-speci®c trans-acting factors control basal transcription (Ibeanu and Goldstein, 1995). Rifampicin-responsive elements in the CYP2C9 promoter have not been identi®ed. Interestingly, however, a transcriptional silencer is present toward the 50 end of the ®rst exon of CYP2C9 (Xiang et al., 1998). This silencer may modulate the response of CYP2C9 to the induction by drugs like rifampicin. CYP2E1 is a toxicologically important enzyme whose regulation is complex (Koop and Tierney, 1990). The mRNA is stabilized in liver of diabetic and fasting rats (transcription rate is unchanged), and the protein is stabilized against proteolysis by CYP2E1 substrates. The rodent enzyme is responsive to agents like ethanol; anecdotal evidence suggests that human CYP2E1 may be regulated similarly (Tsutsumi et al., 1989). Thus, ethanol metabolism and CYP2E1 protein expression is greater in hepatic microsomes obtained from alcoholics. Apart from hepatic CYPs, recent work has indicated that CYP3A4 is expressed at signi®cant levels in the colon (Watkins et al., 1987; Lown et al., 1997). This enzyme is important in drug oxidation soon after absorption from the small intestine (Lown et al., 1997; Holtbecker et al., 1996). Thus, inhibition and induction processes involving extrahepatic CYPs are possible factors that may a€ect the total clearance of some drugs.

28

J.A. Hasler et al. / Molecular Aspects of Medicine 20 (1999) 1±137

3.2.3. Approaches to the study of CYP induction in human systems Because of species di€erences in CYP induction, it would be preferable to evaluate the induction potential of drugs directly in humans. Several approaches to overcome the obvious constraints have included the use of cultured primary human hepatocytes and liver-derived cell lines. The emergence of matrigel, a tumor-derived basement membrane substitute, has provided impetus for the use of primary hepatocyte culture in induction studies (Kocarek et al., 1993). Problems associated with the older culture methods, including loss of di€erentiated function (especially CYP expression), have precluded the study of the more important human CYPs. CYP regulatory studies on matrigel are feasible, at least over a limited time period whilst the cells remain viable (Li et al., 1997a,b). Optimization of time in culture for each CYP is necessary before induction data can be derived. Perhaps the major obstacle is the availability of tissue for research purposes, given ethical considerations and the requirement of most tissue for liver transplantation. Liver-derived cell lines, such as HepG2 and HUH-7 cells, have also been evaluated as systems for the study of CYP regulation (Schuetz et al., 1995). However, there is usually low level expression of only a few CYPs, although those that are expressed are responsive to exogenous chemicals (Dawson et al., 1985). Unfortunately, the absence of most drug-metabolizing CYPs from these cells diminishes their value in human induction studies. 3.3. Pharmacokinetic drug interactions involving CYP inhibition Many drugs and chemicals inhibit CYP activity in vitro due to the broad substrate speci®city of these enzymes. In clinical practice inhibitory processes are quite common, especially during multiple drug therapy. Commonly observed pharmacokinetic e€ects include impaired drug clearance and prolongation of drug action. These processes are short-lived, because di€usion of the drug away from the CYP restores function. In contrast, a number of drugs exert more profound e€ects on the pharmacokinetic behaviour of coadministered drugs in patients. Macrolide antibiotics, such as troleandomycin, and acetylenic steroids, such as ethynyl estradiol, undergo biotransformation to reactive metabolites that do not escape the active site of the CYPs in which they are generated (Murray and Reidy, 1990). Covalent modi®cation leads to prolonged or irreversible CYP dysfunction. To avoid adverse e€ects, the prediction of drugs with a high risk of pharmacokinetic interactions is essential, as is the identi®cation of the target CYPs. 3.3.1. CYP inhibition mechanisms: relationship to enzymic function The determinants of CYP substrate speci®city reside within the polypeptide chains, which are encoded by individual genes that are under distinct regulatory control (Nelson et al., 1996). CYPs possess a ferroprotoporphyrin IX heme prosthetic group which activates oxygen for substrate metabolism. To appreciate the process of CYP inactivation by drugs it is necessary to consid er the CYP reaction cycle (Chapter 1). Brie¯y, microsomal CYP is in the substrate-free, ferric form. Drugs bind to ferric CYP and the resultant bimolecular complex accepts an electron

J.A. Hasler et al. / Molecular Aspects of Medicine 20 (1999) 1±137

29

from NADPH delivered by the coenzyme NADPH-CYP-reductase (Ortiz de Montellano, 1995). Molecular oxygen is coordinated to the ferrous CYP-substrate complex and a second electron from NADPH/NADPH-CYP-reductase or NADH/ NADH-cytochrome b5 reductase enters the cycle. Water is released and a highly reactive oxygen atom, possibly an oxyperferryl species that is radical in nature, remains coordinated to the CYP heme, ready for substrate oxidation. Inhibition processes may be reversible or irreversible (Murray and Reidy, 1990). Reversible inhibition occurs when a drug or metabolite attains concentrations near the CYP that are sucient for impairment of enzyme activity. Di€usion of the inhibitor restores enzymic function so that inhibition is transient. In contrast, irreversible inhibition occurs when the inhibitory agent is converted to a reactive metabolite that binds covalently to the CYP heme or apoprotein. Removal of this species is not usually possible and the loss of enzyme activity is permanent. 3.3.2. Reversible inhibition of CYP enzymes The most common type of inhibition is due to the reversible interaction of a drug or a stable metabolite with a CYP enzyme. This process is probably responsible for most pharmacokinetic drug-drug interactions in vivo. Essentially, reversible processes would occur when a drug or stable metabolite inhibits an early step in the CYP reaction cycle, usually substrate binding or oxygen coordination to the heme. Thus, substrate turnover is decreased. Certain drug metabolites may be formed at rates that are suciently high to enable their accumulation within liver. Metabolites such as N-desmethyldiltiazem (the major metabolite of diltiazem) and nor¯uoxetine (the major metabolite of ¯uoxetine) are more potent than the parent drugs against CYPs (Montamat and Abernethy, 1987; von Moltke et al., 1996; Sutton et al., 1997). This complicates the interpretation of drug-drug interactions because those individuals with higher hepatic concentrations of the CYPs that generate inhibitory metabolite are probably at a greater risk during therapy. Some inhibitory interactions may be bene®cial. Diltiazem, which inhibits CYP3A4, has been used in conjunction with cyclosporin A to impede its elimination (Smith et al., 1994). This has an economic bene®t because the dose interval of the expensive immunosuppressant can be increased. Replacement of diltiazem by CYP3A4 inhibitors without major pharmacological e€ects may be a viable strategy. 3.3.3. Irreversible inhibition of CYP enzymes The term mechanism-based inhibition (Rando, 1984) indicates that catalysis (at least one complete CYP reaction cycle) is required for the inhibition of enzyme activity. There are two basic types of mechanism-based processes that lead to irreversible enzyme inhibition: autocatalytic inactivation and metabolite intermediate (MI)-complexation. Autocatalytic inactivation, or suicide processing, occurs when a reactive drug metabolite binds to CYP and alters its structure irreversibly, resulting in loss of function. Because only one or a few CYPs generate the reactive metabolite much greater selectivity of inhibition is observed than with reversible inhibitors. Inactivation is incomplete: there is usually partitioning between the formation of

30

J.A. Hasler et al. / Molecular Aspects of Medicine 20 (1999) 1±137

isolable metabolites and CYP destruction (Ortiz de Montellano et al., 1979; Kunze and Trager, 1993). The other category of mechanism-based inhibition is metabolite-intermediate (MI) complexation of CYPs which occurs when drug metabolites are formed that bind tightly to the CYP heme (Murray and Reidy, 1990). The distinction between autocatalytic inactivation and MI-complexation is that, in the latter, the hemeprotein is rendered catalytically inert but it is not destroyed. Some generalizations can be made about the arrangements of atoms in drugs that favour autocatalytic inactivation or MI-complexation of CYPs. 3.3.4. Structural features of drugs that inactivate CYPs by suicide processing Drugs containing terminal ole®nic or acetylenic substituents (carbon±carbon double or triple bonds at the end of a functional group) are likely to be suicide substrates of CYPs. The classical inactivator is allylisopropylacetamide (AIA) and the de®nitive studies of Ortiz de Montellano and coworkers established the mechanism by which CYP converts AIA to a destructive metabolite that modi®es CYP heme and activity (Ortiz de Montellano et al., 1979). Drugs containing unsaturated functional groups are uncommon but include gestodene (Guengerich, 1990a; Fig. 7), ethinyl estradiol (Guengerich, 1988b), ethchlorvynol (Ortiz de Montellano et al., 1982) and secobarbital (Lunetta et al., 1989). Apart from unsaturated carbon±carbon bonds, other substituents can also be activated by CYPs. For example, spironolactone is oxidized at the side chain sulfur atom to a reactive intermediate that inactivates the cytochrome and elicits hemeprotein adduct formation (Decker et al., 1989). CYP inactivation has also been observed with tienilic acid (a uricosuric; Lopez-Garcia et al., 1993), chloramphenicol

Fig. 7. Structures of mechanism based inhibitors of CYP. Nortriptyline and orphenadrine generate MI complexes with CYPs in mammalian liver and gestodene and furafylline are autocatalytic inactivators of CYPs.

J.A. Hasler et al. / Molecular Aspects of Medicine 20 (1999) 1±137

31

(an antibacterial; Halpert, 1981), methoxsalen (an antipsoriatic agent; Tinel et al., 1987), clorgyline (an antidepressant; Sharma et al., 1996) and furafylline (an antiasthma drug; Kunze and Trager, 1993; Fig. 7). The versatility of CYP in substrate oxidation is probably responsible for the range of drugs that are activated to destructive metabolites. 3.3.5. Structural features of drugs that inactivate CYPs by MI-complexation Some of the structural requirements of drugs that form MI-complexes with CYPs have been explored. Possibly the most signi®cant class of MI-complex forming drugs is the alkylamines, a common functional group present in several therapeutic classes, including antidepressants, antibiotics and antihistamines. MIcomplex formation is dependent on the oxidation of the alkylamino substituent to its putative reactive metabolite, the nitroso analogue, which is thought to be the CYP-binding reactive species (Mansuy et al., 1978; Lindeke et al., 1979; Bast and Noordhoek, 1982). Perhaps the best documented group of MI-complex forming alkylamine drugs is the macrolide antibiotics, of which troleandomycin and erythromycin (14-membered macrocycles) appear the most important members. Adverse e€ects with coadministered drugs that are metabolized by CYP3A4 include jaundice with estrogen-containing oral contraceptives and toxicity with theophylline (Periti et al., 1992). Because CYP3A4 is important in the oxidation of most drugs and constitutes 50± 60% of the total CYP in human liver, the e€ect of long term inhibition of this enzyme is considerable. CYP inhibition may occur during high dose therapy with some of the newer macrolides, such as josamycin (a 16-membered macrocycle), roxithramycin and clarithromycin (both possess 14-membered rings), but MI-complexation has not been observed (Azanza et al., 1992; Gharbi-Benarous et al., 1991; Ohmori et al., 1993). The macrolides spiramycin, rokitamycin, dithromycin and azithromycin, are not associated with signi®cant drug interactions (Periti et al., 1992). More serious drug±drug interactions are elicited by MI-complex forming macrolides, which are relatively hydrophobic drugs in which there is also steric hindrance around the alkylamino moiety (Sartori et al., 1989). Some alkylamine drugs form MI-complexes with hepatic CYPs in experimental animals, but it is unclear whether human CYPs are a€ected similarly. For example, orphenadrine (Fig. 7) elicits an MI-complex with the major phenobarbital-inducible CYPs 2B in rat liver after in vivo administration and impairs its own elimination after multiple dose regimen in patients (Labout et al., 1982; Reidy et al., 1989) but whether this is due to MI-complexation has not been clari®ed. Indeed, the selectivity of CYP inhibition by orphenadrine appears to be di€erent in rat and human liver (Guo et al., 1997). Similarly, tricyclic antidepressants such as desipramine and nortriptyline (Murray, 1992; Fig. 7) and serotonin reuptake inhibitors like sertraline and ¯uoxetine (Bensoussan et al., 1995) generate MI-complexes in rat liver, but not necessarily in human liver. Caution is required in the interpretation of inhibitor studies conducted in nonhuman species. Some alkylamine drugs do not form MI-complexes with CYPs. The reason is unclear but some of these drugs may be metabolized along pathways that do not lead

32

J.A. Hasler et al. / Molecular Aspects of Medicine 20 (1999) 1±137

to complexation or intermediary metabolites may be unstable. Steric factors may prevent nitroso metabolite formation or its approach to the CYP heme group. Apart from alkylamines, MI-complexes with CYPs are also produced by benzodioxoles, which are insecticide synergists; some derivatives were used previously as ¯avouring agents (Wilkinson et al., 1984). MI-complex formation between these chemicals has been studied in rodent liver, but evidence for similar inhibition of human CYPs has not been presented. Interestingly, the anticonvulsant stiripentol contains a benzodioxole nucleus and inhibits drug metabolism by rat and human CYPs (Mesnil et al., 1988; Tran et al., 1997). Hydrazines like iproniazid (monoamine oxidase inhibitor) and isoniazid (tuberculostatic) also generate MI-complexes with CYPs, at least in rodent liver (Hines and Prough, 1980; Muakkassah et al., 1982; Moloney et al., 1984). These complexes are similar to those generated by alkylamines in that they are stable when the CYP is in the ferrous form. Evidence that dissociation of these complexes can occur in vivo has not been provided. Benzodioxole metabolite complexes with ferric CYP are stable in vivo and in vitro, except in the presence of CYP substrates (Murray et al., 1983; Wilkinson et al., 1984; Murray et al., 1986b). 3.4. Clinical impact of understanding CYP induction and inhibition A test of the signi®cance of the body of information on CYPs is its relevance to human drug therapy. Early observations that drug metabolism in nonhuman species is inducible by barbiturates and inhibited by SKF-525A and other chemicals are applicable to humans. However, species di€erences in CYP regulation and function have been delineated. Furthermore, the individual variation in human CYP expression is considerable and the polymorphic distribution of some CYPs, such as CYP2C19 and CYP2D6, underscore the complexities of human drug metabolism. The impact of CYP polymorphisms on human drug metabolism are becoming clearer and it is appreciated that several CYPs may contribute to multiple pathways of the metabolism of certain drugs. An underexplored area is the impact of the in vivo induction and inhibition of CYPs on overall drug metabolism when multiple oxidative pathways for the drug exist. If there are multiple pathways of drug metabolism it may be that inhibition or induction of CYPs involved in one pathway has little impact. Perhaps problems only occur when the dominant metabolic pathway is inhibited or induced. An intrinsic problem of the CYP system, and which is responsible for the multitude of pharmacokinetic drug interactions, is its capacity to bind and oxidize many drugs. Polypharmacy is a fundamental feature of the management of certain diseases or conditions, including asthma, diabetes, epilepsy, psychoses and HIV infection. Drug combinations should be selected to minimise the likelihood of CYP inhibition or induction. More study is required to understand the relationships between drug structure and induction or inhibition because it may be possible to select pharmacologically equivalent drugs with less impact on biotransformation. For example, use of the anticonvulsant oxycarbamazepine instead of carbamazepine, or rifabutin in place of rifampicin for treatment of tuberculosis may be preferred because of lesser

J.A. Hasler et al. / Molecular Aspects of Medicine 20 (1999) 1±137

33

impact on CYP3A4 levels in human liver. Speci®c quinolone antibacterials or macrolide antibiotics could be selected because of greater understanding of the chemical features associated with CYP1A2 inhibition or MI-complexation of CYP3A4. These considerations are as important for e€ective therapy as the appreciation of CYP phenotypes. Given the high incidence of iatrogenic disease due to inappropriate prescribing and leading to hospitalization, with the attendent costs to the community, a case can be made for the use of noninvasive screening tests to evaluate the biotransformation capability of an individual, perhaps before introduction of certain forms of drug therapy (Frye et al., 1997). 3.5. Acknowledgements The support of the National Health and Medical Research Council and NSW State Cancer Council is gratefully acknowledged by author, M. Murray (Chapter 3). 4. Cytochromes P450 in synthesis of steroid hormones, bile acids, vitamin D3 and cholesterol 4.1. Steroid hormone biosynthesis 4.1.1. Overview All steroid hormones are derived from cholesterol and are produced primarily in a very selected set of tissues, the adrenal cortex, the gonads, and the placenta (Waterman and Keeney, 1996). Fig. 8 shows that the ®rst step in all steroidogenic pathways begins with the conversion of cholesterol to pregnenolone by cholesterol

Fig. 8. Outline of steroidogenic pathways in adrenal cortex and gonads. All enzymes are discussed in the text except 17b-HSD (17b-hydroxysteroid dehydrogenase) and 5a-RED (5a-reductase).

34

J.A. Hasler et al. / Molecular Aspects of Medicine 20 (1999) 1±137

side chain cleavage cytochrome P450 (P450scc the product of CYP11A1). Cellspeci®c expression of di€erent steroidogenic P450s controls the production of di€erent steroid hormones in di€erent cell types; aldosterone (adrenal glomerulosa), cortisol and corticosterone (adrenal fasciculate/reticularis), testosterone (Leydig cells in testis), estrogens (ovarian granulosa and theca cells). In addition to cell-speci®c distribution of steroidogenic P450s, subcellular distribution of these enzymes is a key feature of steroidogenic pathways. As seen in Fig. 9, certain steps occur in mitochondria and others in the endoplasmic reticulum. In addition to cell-speci®c and subcellular distribution of steroidogenic P450s, species-speci®c variation in distribution and activities of these enzymes is also found. 4.1.2. Adrenal cortex 4.1.2.1. Pathways in the adrenal cortex. The adrenal cortex is the best studied of the steroidogenic tissues, and provides the best examples of species-speci®c variations of both distribution and activities of steroidogenic P450s. In addition to describing the human steroidogenic enzymes in the adrenal cortex and important features of their activities, species variations will be summarized. As indicated, all steroidogenic pathways begin in the mitochondrion with the conversion of cholesterol to pregnenolone by P450scc. This enzyme is localized to the inner mitochondrial membrane, facing the mitochondrial matrix (Simpson, 1979). While activity of microsomal P450s requires NADPH-cytochrome P450 reductase (P450 reductase), mitochondrial P450s are reduced by a mini-electron transport chain in the mitochondrial matrix consisting of a ¯avoprotein (ferredoxin reductase) and a 2Fe±2S protein (ferredoxin) (Omura et al., 1966). These enzymes resemble the electron transport system associated with most bacterial P450s and because they were ®rst

Fig. 9. General scheme of steroidogenesis in human adrenal cortex, emphasizing the subcellular localization of the enzymes.

J.A. Hasler et al. / Molecular Aspects of Medicine 20 (1999) 1±137

35

studied in the adrenal cortex, are often called adrenodoxin reductase and adrenodoxin. The conversion of cholesterol to pregnenolone involves cleavage of a carbon± carbon bond converting a C27 sterol to a C21 sterol. Like other P450 reactions which result in carbon±carbon bond cleavage, this is a three step reaction requiring 3 moles of O2 and NADPH. First, hydroxylation occurs at C22, then at C20, with bond cleavage being the third step (Lambeth et al., 1982). Pregnenolone leaves the mitochondrion by passive di€usion and ®nds its way to the endoplasmic reticulum (ER) where it can encounter a variety of fates (Simpson and Waterman, 1995; Kagawa and Waterman, 1995). In glomerulosa cells which constitute the outer zone of the adrenal cortex just under the capsule, it is converted to progesterone by 3b-hydroxysteroid dehydrogenase (3b-HSD) which is then converted to 21-hydroxyprogesterone (deoxycorticosterone) by 21-hydroxylase cytochrome P450 (P450c21 product of CYP21). Deoxycorticosterone leaves the vicinity of the ER and is taken up by mitochondria where it is converted to corticosterone by 11b-hydroxylase cytochrome P450 (P450c11 product of CYP11B1). Like P450scc, P450c11 is located in the inner mitochondrial, membrane and requires adrenodoxin and adrenodoxin reductase for function. Another mitochondrial P450, aldosterone synthase cytochrome P450 (P450aldo product of CYP11B2) converts corticosterone to the major mineralocorticoid aldosterone, in a two step reaction involving 18-hydroxylation of corticosterone. In aldosterone biosynthesis in the adrenal glomerulosa we encounter the ®rst examples of both cell-speci®c and species-speci®c properties of steroidogenesis, P450aldo is found only in the glomerulosa and not in the fasiculata/reticularis of the adrenal cortex. As noted below, 17a-hydroxylase/17,20-lyase cytochrome P450 (P450c17 product of CYP17) is absent from the glomerulosa. The absence of P450c17 and the presence of P450aldo, provides the enzymatic pathway in the glomerulosa necessary for aldosterone biosynthesis. Humans, rats and mice contain two mitochondrial P450s necessary for biosynthesis of aldosterone from deoxycorticosterone (P450c11 and P450aldo). Mitochondria of the adrenal glomerulosa in cows contain only one of these enzymes, P450c11, which has the capacity for both 11b-hydroxylation and aldosterone biosynthesis (Ogishima et al., 1989; Mathew et al., 1990). The reason for this species variation is unknown. The inner zones of the adrenal cortex, fasciculata/reticularis, produce glucocorticoids. Pregnenolone can undergo two fates in these cells, progesterone formation via 3b-HSD or 17a-hydroxypregnenolone (17OH Preg) formation via P450c17 (both in the ER). To produce cortisol it is necessary that progesterone be converted to 17ahydroxyprogesterone (170H Prog) via P450c17 or that 17OH Preg be converted to 17OH Prog by 3b-HSD. 17OH Prog is then converted to deoxycortisol by P450c21. P450c21 will only hydroxylate D4 sterols (from 3bHSD) and not D5 sterols. P450c17 will not hydroxylate 21-hydroxylated sterols. Thus the two paths outlined above are the only ones leading to the major glucocorticoid in humans, cortisol (Simpson and Waterman, 1995; Kagawa and Waterman, 1995). Deoxycortisol leaves the vicinity of the ER, traveling passively to the inner mitochondrial membrane where it is converted to cortisol by P450c11. Recall that P450aldo is not present in the fasiculata/reticularis, assuring that cortisol and not aldosterone is the major product.

36

J.A. Hasler et al. / Molecular Aspects of Medicine 20 (1999) 1±137

P450c17 has two distinct enzymatic activities, 17a-hydroxylase and 17,20-lyase (Zuber et al., 1986). This second activity converts C21 sterols to the C19 androgens. Lyase activity on 170H Preg leads to dehydroepiandrosterone (DHEA) and on 170H Prog leads to androstenedione. DHEA can be converted to androstenedione by 3b-HSD. This second activity of P450c17 is essential for production of testosterones and estrogens in the gonads, but also occurs in the adrenal producing the adrenal androgens. An interesting species-speci®c variation is also found in glucocorticoid biosynthesis. While most species including humans produce cortisol as the primary glucocorticoid, rodents do not (Le Goascogne et al., 1991; Perkins and Payne, 1988). Rather the primary glucocorticoid in rodents is corticosterone. This is because P450c17 is not expressed in adult rodent adrenal. It has been found that P450c17 is present in fetal mouse adrenal, but disappears before birth remaining absent throughout life (Keeney et al., 1995a). Thus the absence of P450c17 and P450aldo in the rodent fasiculata/reticularis leads to the production of corticosterone as the glucocorticoid. 4.1.2.2. Regulation in the adrenal cortex. Study of the regulation of steroidogenesis in the adrenal cortex has led to important insight into the biochemistry of these pathways. The peptide hormone from the anterior pituitary, ACTH, is the key regulator of adrenal steroidogenesis, having e€ects which are both acute and chronic, as seen in Fig. 10 (Waterman and Keeney, 1996; Simpson and Waterman, 1995; Kagawa and Waterman, 1995). These are separated from one another temporally. The acute response occurs very rapidly, within seconds or minutes, and has been found to result in the rapid mobilization of cholesterol from lipid stores into the inner mitochondrial membrane in the vicinity of P450scc. Steroidogenic mitochondria contain two pools of cholesterol, that which in¯uences membrane ¯uidity found in all mitochondrial and the second which is steroidogenic. ACTH regulates levels of this latter pool via cAMP (Jefcoate et al., 1992). Upon binding of ACTH to its adrenocortical cell surface receptor, adenylate cyclase is rapidly activated and intracellular levels of cAMP increase. Phosphorylation by protein kinase A (PKA) activates cholesterol ester hydrolase generating free cholesterol in the lipid vesicles in the adrenal cortex (Jefcoate et al., 1992). Cholesterol is rapidly transported to the outer mitochondrial membrane via an unknown path. Sterol carrier protein 2 (SCP2)

Fig. 10. Outline of the acute and chronic peptide hormone actions in steroidogenic tissues.

J.A. Hasler et al. / Molecular Aspects of Medicine 20 (1999) 1±137

37

has been suggested to be the cholesterol carrier (Jefcoate et al., 1992), but this role for SCP2 remains to be positively established. The acute response of steroidogenesis is inhibited by the protein synthesis inhibitor cycloheximide. A protein called steroidogenic acute regulatory protein (StAR) has recently been discovered to be crucial for the movement of cholesterol from the outer membrane of the mitochondrion to the inner membrane where P450scc resides (Clark et al., 1994). How StAR functions to serve this role is a point of great debate, but it apparently does not do so by binding cholesterol itself and directly transporting it across the intermembrane space. The synthesis of StAR is rapid, and is the locus of the CHX inhibitory site of the acute response of ACTH (King et al., 1995). The chronic action of ACTH regulates the synthesis of steroidogenic enzymes in the adrenal cortex at the transcriptional level (Waterman and Keeney, 1996; Simpson and Waterman, 1995; Kagawa and Waterman, 1995). Transcriptional regulation of steroidogenesis is multifactorial, involving developmental, tissue speci®c, cAMP-independent and cAMP-dependent mechanisms. Developmental and tissue speci®c transcriptional regulation involves an orphan nuclear receptor called steroidogenic factor-1 (SF-1, see below). cAMP-independent regulation involves the action of cytokines and growth factors. cAMP-dependent regulation is the major transcripitonal regulatory mechanism throughout life which maintains optimal steroidogenic capacity in the adrenal in response to ACTH. In the testis and ovary LH and FSH are the peptide hormones assuring optimal steroidogenic capacity (Fig. 10). The chronic action of ACTH is also mediated by cAMP, but does not involve traditional cAMP regulatory transcriptional systems (CRE and CREB). Each steroidogenic CYP gene has its own novel cAMP response sequence (CRS) and associated trans-acting factors (Waterman, 1994). 4.1.3. Testis (Leydig cells) Two steroidogenic P450s are required for testosterone biosynthesis, P450scc and P450c17. P450scc functions just as described for the adrenal cortex, except LH is the peptide hormone from the anterior peptide which regulates steroidogenesis. P450c17 produces the C19 androgen intermediates in testosterone biosynthesis which are converted to testosterone by nonP450 enzymes. Since there is no glucocorticoid production in the testis, P450c17 serves solely to produce androgens in Leydig cells. Here again we ®nd an interesting species-speci®c variation. In rodents, where CYP17 is not expressed in the adrenal, the puri®ed enzyme as represented by studies of rat P450c17 not only readily catalyzes 17a-hydroxylation reactions but also 17,20-lyase reactions (Fevold et al., 1989). In humans, where 17a-hydroxylation is essential for production of cortisol in the adrenal, the puri®ed enzyme has very low 17,20-lyase activity, particularly 17OH Prog lyase activity (Imai et al., 1993). There is only a single CYP17 gene indicating that this single human enzyme which does not favor androgen production in the adrenal cortex, somehow does so in Leydig cells. The lyase activity of human P450c17 has been found to require an accessory ER protein, cytochrome b5 , for production of physiological androgen levels in the testis (Katagiri et al., 1995). The ratio of b5 /microsomal P450 in the human testis is 10/1 while it is 0.3/1 in rat testis. Thus high levels of cytochrome b5 in human testis assure adequate

38

J.A. Hasler et al. / Molecular Aspects of Medicine 20 (1999) 1±137

17,20 lyase activity for testosterone biosynthesis (Mason et al., 1973). The rat enzyme has evolved independent of a need for cytochrome b5 because of its absence in the adrenal cortex and therefore lack of need to participate in glucocorticoid biosynthesis. 4.1.4. Ovary (Theca and Granulosa) Steroidogenesis in the ovary introduces a new microsomal P450, aromatase cytochrome P450 (P450arom product of CYP19) (Mahendroo et al., 1993). Here we also encounter a fascinating di€erentiation system which alters cellular P450 pro®les. Steroidogenic cells, theca and granulosa, are located within the ovarian follicle. Thecal cells line the follicle and the granulosa cells surround the ovum within the follicular ¯uid. Both cell types contain P450scc and 3b-HSD and are able to produce progesterone (Hsueh and Villig, 1995). The thecal cells also contain P450c17 and can produce androgens, particularly androstenedione. These are released and taken up by the glomerulosa cells where P450arom (located in glomerulosa cells but not theca cells) converts them to estrogens (Hsueh and Villig, 1995). Following ovulation, the ovarian follicle di€erentiates into the corpus luteum. The thecal and granulosa cells di€erentiate into leuteal cells which in the cow produce large quantities of progesterone. P450c17 disappears from the theca-derived leuteal cells and levels of P450scc, adrenodoxin, adrenodoxin reductase and 3b-HSD all increase substantially in leuteal cells derived from both types of follicular steroidogenic cells leading to progesterone production (Rogers et al., 1986a,b; Rogers et al., 1987). This di€erentiation response is mediated by LH. Normal follicular steroidogenesis is regulared by another peptide hormone from the anterior pituitary, FSH. 4.1.5. Steroidogenesis in nonsteroidogenic cells Space does not permit discussion of the many observations of steroidogenic enzymes in nonsteroidogenic tissues. Most of this work has been carried out in rodents and it remains to be established whether similar patterns of expression of steroidogenic enzymes occur in human nonsteroidogenic tissues, beyond the well-known expression of P450arom in adipose tissue. In the rat, steroidogenic enzymes are expressed in the brain (Le Goascogne et al., 1987), liver (Vianello et al., 1997), stomach (Vianello et al., 1997; Le Goascogne et al., 1995), and intestine (Dalla Valle et al., 1995; Keeney et al., 1995b). The physiological role of these enzymes in all cases is unknown, remaining the subject of speculation. Unlike the steroidogenic cells described above, levels of expression of steroidogenic enzymes in nonsteroidogenic cells is generally low and P450scc activity is not associated with StAR, meaning that steroidogenic pools of cholesterol are probably not present in these cells. Finally, transcriptional regulation of steroidogenic CYP genes is di€erent between steroidogenic and nonsteroidogenic tissues. Parker and colleagues (Ikeda et al., 1993) and Morohashi and Omura (Morohashi et al., 1992) found that all steroidogenic genes including the CYP genes contain binding sites for an orphan nuclear receptor now generally called steroidogenic factor 1 (SF-1). This transcription factor has been found to have two roles; being essential for adrenal and gonad organogenesis and also for transcription of steroidogenic enzymes (Luo et al., 1994). While SF-1 is

J.A. Hasler et al. / Molecular Aspects of Medicine 20 (1999) 1±137

39

thought to be important for expression of steroidogenic genes, it is not present in nonsteroidogenic tissues. Therefore expression of steroidogenic CYP genes in nonsteroidogenic tissues is SF-1-independent (Keeney et al., 1995b). 4.1.6. Congenital adrenal hyperplasia Congenital adrenal hyperplasia is a group of diseases arising from mutations in steroidogenic genes. The molecular bases of these diseases is very broad, resulting from single nucleotide changes leading to premature stop codons or amino acid changes altering protein structure and/or function, deletion of small or large portions of genes and duplication of coding sequences as a result of aberrant recombination (New, 1995). Mutations in each of the CYP genes in the steroidogenic pathways except P450scc are known. Most are very rare, occurring in one or a few families, the exception being 21-hydroxylase (P450c21) de®ciency which is quite common in certain ethnic populations such as Ashkenazic Jews. 21-Hydroxylase de®ciency is much more common because both a functional gene (CYP21B) and a pseudogene (CYP21A) are present in close proximity in the MHC locus on chromosome 6. Recombination between the functional gene and the pseudogene (which contains several mutations) accounts for the increased frequency of this form of congenital adrenal hyperplasia. All forms of congenital adrenal hyperplasia are autosomal recessive. P450scc de®ciency has not yet been discovered to result from mutations in CYP11A1. Rather it results from mutations in StAR which reduce or block transport of cholesterol to the inner mitochondrial membrane (Lin et al., 1995). This defect can result in the absence of steroid hormone production which can be lethal without hormone replacement therapy because of the inability to produce mineralocorticoids necessary to control salt balance. Mutations in CYP11B1 (P450c11) or CYP11B2 (P450aldo) are not as serious because deoxycorticosterone and corticosterone have mineralocorticoid activity. The inability to produce steroid hormones also profoundly in¯uences phenotypic development. The absence of androgen production observed with either P450scc de®ciency or mutation in CYP17 (P450c17 de®ciency) leads to the absence of the male phenotype. Production of testosterone during fetal life by the male (XY) genotype is essential for development of male secondary sex characteristics (penis and scrotum). In the absence of androgen production the phenotype is female regardless of the genotype. Neither male nor female genotype go through puberty in these de®ciencies. CYP17 de®ciency also leads to hypertension because of overproduction of corticosterone. CYP21 de®ciency can reduce mineralocorticoid production leading to salt wasting and can lead to overproduction of androgens which enhance virilization in both males and females. Study of these genetic diseases have assisted in understanding the physiological roles of the steroidogenic P450s. 4.2. Bile acid synthesis Bile acid biosynthesis has long been known to take place in the liver and involves three di€erent forms of cytochrome P450 (Russell and Setchell, 1992). The ®rst and

40

J.A. Hasler et al. / Molecular Aspects of Medicine 20 (1999) 1±137

rate limiting step in this pathway takes place in endoplasmic reticulum where cholesterol 7a-hydroxylase (P450c7 product of CYP7A) converts cholesterol to 7a-hydroxycholesterol. A second microsomal P450, sterol 12a-hydroxylase in one of several enzymes involved in the production 5b-cholestane-3a, 7a, 12a-triol. This intermediate travels to the mitochondrion where sterol 27-hydroxylase (P450c27 product of CYP27) converts the triol to 5b-cholestane-3a, 12a, 27-tetrol. This intermediate is then converted to cholic acid (Russell and Setchell, 1992). An alternative hepatic pathway skips the sterol 12a-hydroxylase step producing chenodeoxycholic acid. The 27-hydroxylase is located in the inner mitochondrial membrane and requires ferredoxin and ferredoxin reductase for activity. CYP27 is the locus of one of the genetic diseases associated with bile acid biosynthesis, the rare autosomal recessive disease cerebrotendinous xanthomatosis (CTX) which leads to accumulation in the central nervous system of aberrant C27 bile alcohols (Russell and Setchell, 1992). While CYP7A is expressed exclusively in the liver, P450c27 is expressed in all nucleated cells. An alternative pathway for the biosynthesis of bile acids has been discovered in which P450c27 in extrahepatic cells catalyzes the ®rst step (Bjorkhem, 1992; Javitt, 1994). This alternative pathway leads to production of approximately 4% of the total bile acids and represents a pathway by which cholesterol can be removed from extrahepatic cells. The ®rst product of P450c27, 27-hydroxycholesterol, is transported through the circulation to the liver where it enters the primary (7a-hydroxylase) bile acid biosynthetic pathway. It has recently been found that human P450c27 can also catalyze the conversion of 27-hydroxycholesterol to 3b-hydroxy-5-cholestenoic acid which involves conversion of the 27 hydroxyl group to a 27-carboxylate group (Pikuleva et al., 1998). Both products of extrahepatic P450c27 can enter the hepatic bile acid pathway, the acid being the predominant product at low cholesterol levels. 27-Hydroxycholesterol is also thought to have additional cellular functions including regulation of cholesterol synthesis (Lund and Bjorkhem, 1995), cytotoxicity to di€erent cell types (Zhou et al., 1993) and platelet aggregation (Selley et al., 1996). P450c27 is the most widely distributed mitochondrial P450 in animal cells. 4.3. Vitamin D3 metabolism Three P450s are required for the activation and inactivation of Vitamin D3 (cholecalciferol) (Okuda et al., 1995). 7-Dehydrocholesterol, a late intermediate in the cholesterol biosynthetic pathway is photolyzed by UV light producing vitamin D3 . Vitamin D3 is converted to calcitriol (1,25-dihydroxycholecalciferol) by two sequential P450-dependent reactions. First is 25-hydroxylation which is catalyzed by P450c27 (Okuda, 1994). Recall that P450c27 is a mitochondrial P450 involved in bile acid biosynthesis where it hydroxylates the cholesterol side chain at C27. When cholecalciferol is presented to this enzyme as a substrate, hydroxylation occurs at C25. The product of this reaction is then converted to calcitriol by 1a-hydroxylation, also catalyzed by a mitochondrial P450, vitamin D3 1a-hydroxylase cytochrome P450 (P4501a product of CYPC27B). This enzyme has been known to exist in

J.A. Hasler et al. / Molecular Aspects of Medicine 20 (1999) 1±137

41

mitochondria for decades, but has remained an elusive P450. Because of its very low level of expression and presumed instability it has been dicult to impossible to purify. Only within the past two years has it been cloned and found to reside in the same gene family as P450c27 (Takeyama et al., 1997; Monkawa et al., 1997; Shinki et al., 1997, Fu et al., 1997). 1a-Hydroxylation generates the active form of vitamin D3 . A third mitochondrial P450, P450c24 (product of CYP24) inactivates calcitriol by hydroxylation on the side chain at position C24 which greatly reduces binding to the vitamin D receptor. P450c24 can also hydroxylate 25-hydroxycholicalciferol producing 24,25-dihydroxycholicalciferol. The kidney is the primary site of 1,25dihydroxycholecalciferol biosynthesis. As would be predicted, CTX also leads to abnormal vitamin D3 metabolism (Okuda et al., 1995). 4.4. Cholesterol biosynthesis The cholesterol biosynthetic pathway is very complex and is found in all nucleated cells. One step, 14a-demethylation of lanosterol is catalyzed by a microsomal cytochrome P450 enzyme, lanosterol 14a-demethylase cytochrome P450 (P45014DM product of CYP51). This is the most widely distributed form of cytochrome P450 in biology, being found in animals, fungi including yeast, plants and some bacteria (Noshiro et al., 1997; Yoshida et al., 1997). It has been an elusive mammalian P450, the human enzyme being cloned in 1996 (Stromstedt et al., 1996). P45014DM catalyzes a complex 3-step reaction requiring three moles of O2 and HADPH, ®rst leading to conversion of the methyl group to an alcohol, then to acid, before release as formic acid. Recently it has been found that the product of CYP51 accumulates in human follicular ¯uid and in bull testis, and has the ability to reinitiate activation of meiosis of mouse oocytes (Byskov et al., 1995). This suggests that in addition to being an intermediate in the cholesterol biosynthetic pathway, the product of CYP51 is a signalling sterol. The highest level of expression of human CYP51 is in postmeiotic (haploid) male germ cells (spermatids) (Stromstedt et al., 1998). Any function for this enzyme in spermatids beyond cholesterol biosynthesis is unknown. Since it is most highly expressed in postmeiotic germ cells it may not be important in meiosis in the male. Regulation of CYP51 gene expression in animal cells is via the sterol recognition element (SRE) pathway which regulates levels of all or most enzymes in the cholesterol biosynthetic pathway. It is also likely that CYP51 transcription can be mediated by cAMP during spermatogenesis (D. Rozman, unpublished). The very high level of conservation of CYP51 across phylogeny and its potential to serve multiple, essential biological roles make CYP51 a particularly interesting form of P450. 4.5. Conclusion Clearly P450 enzymes play crucial roles in biological processes in humans. In addition to those discussed herein, the P450-dependent metabolism of arachidonic acid as described by J. Capdevila et al. in Chapter 5 is critical. These two articles emphasize how critical P450 enzymes are in endogenous metabolism and indicate

42

J.A. Hasler et al. / Molecular Aspects of Medicine 20 (1999) 1±137

that human diseases arise from mutation of CYP genes encoding P450s required for endogenous metabolism. It can be expected that additional P450s involved in endogenous substrate metabolism will be found in humans. Some may be members of new gene families and others may be members of gene families generally thought to be involved in xenobiotic metabolism. In the overall study of P450s, endogenous substrate metabolism is a very important consideration. 4.6. Acknowledgments Studies of P450s involved in endogenous substrate metabolism are supported in the Waterman laboratory by USPHS grants DK28350 and GM37942. 5. Microsomal cytochrome P450 and eicosanoid metabolism 5.1. Introduction During the last decade, our understanding of the roles that lipids play in cell and organ biology has changed dramatically since, in addition to their accepted structural importance as the building blocks of cellular membranes, an extensive body of evidence has demonstrated that fatty acids, glycerolipids, glycerophospholipids, ceramides, etc., participate as mediators in a variety of transmembrane signaling cascades, as well as in cell di€erentiation, replication and apoptosis. The functional signi®cance of these mediators, further emphasized by their proposed roles in the pathophysiology of diseases such as in¯ammation, asthma, cancer, diabetes, and hypertension, has stimulated and intense research into the biochemistry, enzymology, and the regulation of lipid metabolism. The arachidonic acid (AA) cascade, composed of prostaglandin H2 synthase, lipoxygenases, and, more recently, microsomal P450, serves as a premier illustration of the role that lipid-derived mediators play in cell and organ function (Smith et al., 1991; Smith, 1992; DuBois et al., 1998; McGi€, 1991; Ford-Hutchinson et al., 1994; Oliw, 1994; Harder et al., 1995; Makita et al., 1996; McGi€ et al., 1996; Schwartzman et al., 1996; Rahman et al., 1997; Harder et al., 1997). After phospholipase A2 catalyzed release, AA is oxidized by the enzymes of the AA cascade. Metabolism by prostaglandin synthase generates an unstable cyclic endoperoxide (prostaglandin H2 ,) that rearranges enzymatically or chemically to prostaglandins (PGs), prostacyclin (PGI2 ) or thromboxane A2 (TXA2 ) (Smith et al., 1991). Metabolism by the lipoxidases generates several regioisomeric allylic hydroperoxides containing a cis, trans conjugated diene functionality. One of these, 5-hydroperoxyeicosatetraenoic acid (5-HPETE), serves as the precursor for the biosynthesis of leukotrienes (Ford-Hutchinson et al., 1994). The physiological and biomedical signi®cance of prostanoids and leukotrienes has been extensively documented (Smith, 1991; Smith 1992; Ford-Hutchinson et al., 1994; DuBois et al., 1998). Among these are their critical roles in pulmonary, vascular, and renal physiology as well as in the pathophysiology of in¯ammation, asthma, and, more recently, cancer

J.A. Hasler et al. / Molecular Aspects of Medicine 20 (1999) 1±137

43

(Smith et al., 1991; Smith 1992; Ford-Hutchinson et al., 1994; DuBois et al., 1998). On the other hand, the biological roles of many of the AA metabolites formed by the P450 system, the newest member of the AA metabolic cascade, are currently under extensive investigation (McGi€, 1991; Oliw, 1994; Harder et al., 1995; Makita et al., 1996; McGi€ et al., 1996; Schwartzman et al., 1996; Rahman et al., 1997; Harder et al., 1997). The studies of the role of microsomal P450 in the metabolism of AA were initiated in 1969 and continued, nearly 10 years latter, with the documentation of its spectral interactions with the P450 heme moiety and its e€ects on drug metabolism (DiAgustine and Fouts, 1969; Pessayre et al., 1979). A functional implication for P450 in AA metabolism was ®rst suggested in 1976 by the demonstration that the AA-induced aggregation of human platelets could be blocked by known P450 inhibitors (Cinti and Feinstein, 1976). However, the then growing pharmacological and toxicological importance of P450 focused attention of most investigators on its roles in drug and xenobiotic metabolism, It was not until 1981 that the role of microsomal P450 in the oxidative metabolism of arachidonic was unequivocally demonstrated when microsomal fractions and reconstituted P450 systems were shown to catalyze AA oxidation to products chromatographically distinct from prostanoids (Capdevila et al., 1981b; Capdevila et al., 1981a; Oliw and Oates, 1981; Morrison and Pascoe, 1981). Soon after, the structural characterization of most of the products generated from AA by incubates containing liver or kidney microsomal fractions was completed (Oliw and Oates, 1981; Morrison and Pascoe, 1981; Capdevila et al., 1982; Chacos et al., 1982; Oliw et al., 1982). It was evident from those original studies that the physiological importance of the AA substrate made those observations unique and likely to be functionally signi®cant. Furthermore, interest in these novel reactions of P450 was stimulated by: (a) the initial demonstration that some of its products displayed potent biological activities (Capdevila et al., 1983; Jacobson et al., 1984) and, (b) the documentation of its participation in the in vivo metabolism of endogenous AA pools (Capdevila et al., 1984a). These earlier studies established the metabolism of AA by P450 as a formal metabolic pathway, P450 as an endogenous member of the AA metabolic cascade and, more importantly, suggested functional roles for this enzyme in the bioactivation of the fatty acid. In recent years, the study of the biochemistry and biological signi®cance of these reactions has developed into an area of intense research. Although the physiological and/or pathophysiological implications of this pathway of AA metabolism remain to be fully understood, work from several laboratories is beginning to establish biochemical and functional correlations that can be interpreted as suggestive of a physiological function (McGi€, 1991; Oliw, 1994; Harder et al., 1995; Makita et al., 1996; McGi€ et al., 1996; Schwartzman et al., 1996; Rahman et al., 1997; Harder et al., 1997). In addition to the catalysis of AA oxygenation, microsomal P450 also participates in the metabolic transformation of oxygenated metabolites of AA (eicosanoids) by catalyzing both NADPH-independent and NADPH-dependent reactions. This distinction re¯ects the marked di€erences in types of oxygen chemistry involved in these reactions, i.e., the NADPH dependent activation of atmospheric oxygen or the NADPH-independent isomerization of arachidonic acid hydroperoxides. As with

44

J.A. Hasler et al. / Molecular Aspects of Medicine 20 (1999) 1±137

AA, the potential functional signi®cance of these reactions has stimulated considerable interest in their enzymology and regulation. For clarity, we will discuss ®rst those reactions associated with the metabolism of oxygenated eicosanoids by P450 and then concentrate on its role(s) in AA metabolism. 5.2. NADPH-independent metabolism of eicosanoids Cytochrome P450 is an active peroxidase that catalyzes the metabolism of a wide variety of organic hydroperoxides, including fatty acid hydroperoxides (White and Coon, 1980; Rahimtula and OÕBrien, 1974; Capdevila et al., 1984b). This peroxidase activity, initially described by O'Brien and collaborators (Rahimtula and OÕBrien, 1974), is associated with the ferric, Fe3‡ , state of microsomal P450, does not involve electron transfer from NADPH, and exhibits high catalytic rates. The mechanism of peroxide O±O bond cleavage, i.e., homolytic or heterolytic scission (White and Coon, 1980; Rahimtula and OÕBrien, 1974; Capdevila et al., 1984b), determines the catalytic outcome of these reactions and is highly dependent on: (a) the nature of the P450 isoform, and (b) the chemical properties of the organic hydroperoxide and the oxygen acceptor (White and Coon, 1980; Capdevila et al., 1984b). A homolytic pathway was proposed to account for the formation, from 15-hydroperoxyeicosatetraenoic acid (15-HPETE), of 11-and, 13-hydroxy-14,15-epoxyeicosatrienoic acids by rat liver microsomes (Weiss et al., 1987). Sequence homology analysis and spectral studies of human lung and platelet thromboxane synthase indicated the presence of structural features, including a heme-thiolate prosthetic group, typical of P450 type hemoproteins (Ohashi et al., 1992; Yokoyama et al., 1991). The heterolytic cleavage of the PGH2 endoperoxide and an oxygen atom transfer or oxenoid mechanism was proposed to account for the enzymatic formation of PGI2 and TXA2 from PGH2 (Hecker and Ullrich, 1989). The cDNAs coding for human thromboxane and bovine prostacyclin synthases were subsequently cloned, sequenced, and shown to code for cysteine-heme coordinated hemoproteins with 6 35% overall amino acid identity to members of the P450 3 and 7 gene families, respectively (Ohashi et al., 1992; Yokoyama et al., 1991; Hecker and Ullrich, 1989; Hara et al., 1994). The biomedical importance of these enzymes and their products to human vascular homeostasis is well documented (Smith et al., 1991; Smith, 1992; Ford-Hutchinson et al., 1994). In plants, a ¯axseed peroxidase was puri®ed, characterized, and its cDNA cloned and expressed (Song and Brash, 1991; Song et al., 1993). The enzyme catalyzed the heterolytic cleavage of 13-hydroperoxy linoleic acid to the corresponding allene oxide, a precursor in jasmonic acid biosynthesis (Song and Brash, 1991). Sequence analysis indicated a 6 25% overall identity to other P450s and the presence of a conserved cysteine residue involved in heme coordination (Song et al., 1993). The discovery and recent association of the peroxidase function of P450 to the biosynthesis of autacoids of animal or vegetal origin has opened novel and exciting research areas. The signi®cance of these P450 supported pathways to cell physiology and/or pathophysiology has only begun to be explored. The identi®cation of these enzymes, many of which are members of the arachidonate cascade with established

J.A. Hasler et al. / Molecular Aspects of Medicine 20 (1999) 1±137

45

physiological roles, as homologues of the P450 gene family should facilitate studies of their enzymology, molecular properties, and mechanisms and site of action. 5.3. NADPH-dependent metabolism of eicosanoids The NADPH-dependent metabolism of several eicosanoids by P450 is well established. The biological importance of these reactions resides in the fact that they: (a) increase eicosanoid structural diversity and, hence informational content, (b) may alter the pharmacological pro®le of the substrate, and (c) may participate in the regulation of steady state and/or stimulated levels of physiologically relevant molecules. While the consensus is that these reactions, for the most part, attenuate biological activity and initiate eicosanoid catabolism, recent studies have described the unique and potent biological properties of a few x-oxidized prostanoids (McGi€, 1991; McGi€ et al., 1996; Schwartzman et al., 1996). Nonetheless, in most cases, the sequence of steps leading to x/xÿ1 oxidized prostanoids from endogenous AA pools remains to be clari®ed. For the most part, the NADPH-dependent, P450 catalyzed, metabolism of prostanoids, leukotrienes, HETEs, and epoxyeicosatrienoic acids (EETs) result in the hydroxylation of these eicosanoids at the ultimate (C20 or x carbon) or penultimate carbon atoms (C19 or xÿ1 carbon). However, the epoxidation of infused PGI2 by a perfused kidney preparation (Wong et al., 1985), and the metabolism of 5,6- and 8,9-EET by prostaglandin H2 synthase are published (Oliw, 1984; Zhang et al., 1992). A detailed study of secondary metabolism of 12(R)-HETE and 14,15-EET by P450 has been reported (Jajjo et al., 1992; Capdevila et al., 1988). To date, however, none of the transformations described in this section, with the exception of x/xÿ1 hydroxylation, have been shown to occur in vivo and from endogenous precursors. Since the demonstration of C19 hydroxylated prostanoids in the human semen (Hamberg and Samuelsson, 1966), x/xÿ1 hydroxylation has become a recognized route for prostanoid metabolism. More recently, the hydroxylation at the C18 (xÿ3) carbon of PGE2 was reported (Oliw, 1989). Early studies demonstrated that prostanoid x and xÿ1 oxidation was NADPH-dependent, localized to the endoplasmic reticulum (Israelson et al., 1969), and catalyzed by microsomal P450 (Kupfer, 1982). Reconstitution studies using puri®ed enzymes or recombinant P450's showed that most of these reactions were catalyzed by members of the P450 4A gene subfamily. At present, approximately 10 4A isoforms have been cloned and/or isolated and puri®ed from rats (4A1, 4A2, 4A3 and 4A8), rabbits (4A4, 4A5, 4A6, 4A7) or humans (4A9 and 4A11) (Nelson et al., 1996). Many of these enzymes have been characterized enzymatically (Hardwick et al., 1987; Sharma et al., 1989a; Imaoka et al., 1989, 1990; Kimura et al., 1989a,b; Stromsteadt et al., 1990; Matsubara et al., 1987; Johnson et al., 1990; Roman et al., 1993; Yokotani et al., 1991, 1989; Sawamura et al., 1993; Kawashima et al., 1992; Palmer et al., 1993; Imaoka et al., 1993a; Wang, 1997). Table 3 summarizes the known substrate speci®city ad regioselectivity P450 4A isoforms involved in fatty acid and/or prostanoid x/xÿ1 hydroxylation (Hardwick et al., 1987; Sharma et al., 1989a; Imaoka et al., 1989, 1990; Kimura et al.,

46

J.A. Hasler et al. / Molecular Aspects of Medicine 20 (1999) 1±137

Table 3 Metabolism of fatty acids and prostanoids by P450 4A subfamily isoforms Isoform

Species

Enzymatic activities

4A1 4A2 4A3 4A8 4A4

Rat Rat Rat Rat Rabbit

4A5

Rabbit

4A6

Rabbit

4A7

Rabbit

4A9 4A11

Human Human

x-oxidation; lauric and arachidonic acid x/xÿ1 oxidation of lauric and arachidonic acid x/xÿ1 oxidation of lauric acid Unknown xÿoxidation of prostaglandin A, E, D and F2 a and palmitic and arachidonic acid x-oxidation of PGA1 , x/xÿ1 oxidation of lauric, palmitic and arachidonic acid x-oxidation of lauric, palmitic and arachidonic acid. Low PGA1 x-oxidation x-oxidation of PGA1 , lauric, palmitic, and arachidonic acids. Inactive towards PGE1 x-oxidation of lauric acid x-oxidation of lauric acid

1989a,b; Stromsteadt et al., 1990; Matsubara, 1987; Johnson et al., 1990; Roman et al., 1993; Yokotani et al., 1991, 1989; Sawamura et al., 1993; Kawashima et al., 1992; Palmer, 1993; Imaoka et al., 1993a; Wang et al., 1997). The x-oxidation of leukotriene B4 (LTB4 ) has been documented in whole animals, isolated cells, and subcellular fractions (Ford-Hutchinson, 1994; Jubiz et al., 1981; Powell, 1984; Shak and Goldstein, 1984; Romano et al., 1987). As with prostanoids, these reactions appear to serve mostly catabolic roles. Early biochemical studies suggested that the x-oxidation of LTB4 in leukocytes was catalyzed by a unique P450 isoform (Powell, 1984; Shak and Goldstein, 1984). Subsequently, a cDNA coding for a novel P450 isoform, Cyp P450 4F3, was cloned form human polymorphonuclear leukocytes and expressed (Kikuta et al., 1993). Recombinant P450 4F3 actively catalyzed the x-oxidation of LTB4 with a Km of 0.71 lM (Kikuta et al., 1993). Finally, the x-oxidation of 12(S)-HETE by polymorphonuclear leukocytes was demonstrated in 1984 (Wong et al., 1984; Marcus et al., 1984). Moreover, the last authors also showed that endogenous AA pools could be converted to 12,20dihydroxyeicosatetraenoic acid by a co-incubated mixture of human platelets and polymorphonuclear leukocytes (Marcus et al., 1984). Both 5- and 15-HETE are known to undergo x-oxidation by P450 (Flaherty and Nishihira, 1987; Okita et al., 1987). A P450 similar to that responsible for LTB4 metabolism, has been implicated in the x-oxidation of lipoxin A4 and B4 by human neutrophils and polymorphonuclear leukocytes (Boucher et al., 1991; Mizukami et al., 1993). 5.4. Metabolism of arachidonic acid: the ``P450 arachidonic acid monooxygenase pathway'' 5.4.1. Introduction The structural identi®cation of most of the P450-derived eicosanoids was soon followed by the chemical synthesis of the predominant metabolites isolated from incubates containing liver and kidney microsomal fractions (Manna et al., 1983;

J.A. Hasler et al. / Molecular Aspects of Medicine 20 (1999) 1±137

47

Falck et al., 1984; Moustakis et al., 1985). The access to sucient quantities of most P450-derived eicosanoids opened the door to extensive studies of the enzymology and biological signi®cance of these reactions. The early demonstration of the epoxyeicosatrienoic acids (EETs) as stimulants for the release of peptide hormones (Capdevila, 1983) and as inhibitors of Na‡ reabsorption in the distal nephron (Jacobson et al., 1984), focused interest on these novel P450 reactions and stimulated their subsequent molecular, biochemical and functional characterization. Soon after, the signi®cance of these observations was reinforced by the GC/MS documentation of chiral EETs pools in rat, rabbit and human organs, including kidney (Karara et al., 1989, 1990, 1992; Falck et al., 1987), and of EETs, DHETs and 20-hydroxyeicosatetraenoic acid (20-OH-AA) in human and rat urine (Capdevila et al., 1992; Schwartzman et al., 1991; Prakash et al., 1992). These results were important since they demonstrated endogenous biosynthesis of these eicosanoids, established P450 catalyzed AA monooxygenation as a formal metabolic pathway, P450 as a member of the endogenous AA cascade, and suggested functional roles for this enzyme system in cell and organ physiology. The studies of the renal signi®cance of this novel pathway of AA metabolism were initiated soon after its initial biochemical characterization and were stimulated, among other things, by the early demonstration of the potent transport and vasoactive properties of its metabolites (McGi€, 1991). Arachidonic acid is metabolized by liver microsomal P450s at rates that varied from 1 to 6 nmols of product/min/mg of microsomal protein and with apparent Km values for liver microsomes of 20±40 lM (Capdevila et al., 1981a; Falck et al., 1990; Capdevila et al., 1990). As with most of the enzymes of the arachidonic acid cascade, P450 will not oxidize, at signi®cant rates, the fatty acid when esteri®ed to phospholipids (Capdevila et al., 1990; Karara et al., 1992). During arachidonic acid metabolism, the P450 enzyme system catalyzes the NADPH-dependent, redox coupled activation of molecular oxygen an its delivery to the substrate ground state carbon skeleton. This feature, i.e., the NADPH-dependent, redox coupled activation of molecular oxygen, as opposed to the free radical-mediated activation of carbon atoms, distinguishes the P450 enzyme system from the other enzymes of the arachidonate cascade. Thus, while P450 functions as an active arachidonic acid monooxygenase, prostaglandin synthase and lipoxygenases are arachidonic acid dioxygenases. 5.4.2. Reactions catalyzed The analysis of the reaction products generated by microsomal fractions and/or puri®ed isoforms demonstrated that the P450 monooxygenase metabolizes AA by one or more of the following reactions: (a) Allylic and/or bis-Allylic Oxygenase generating, as ®nal products, six regioisomeric hydroxyeicosatetraenoic acids (HETEs) containing a cis,trans conjugated dienol (Oliw, 1994, 1993); (b) x/xÿ1 Hydroxylases which functionalyze sp3 carbons at or near the AA methyl terminus to 16-, 17-, 18-, 19-, or 20-hydroxyeicosatetraenoic acids (16-, 17-, 18-, 19-, and 20-OHAA, respectively) (Oliw 1994; Falck et al., 1990); and (c) Epoxygenase or ole®n epoxidation to 5,6-, 8,9-, 11,12-, or 14,15-EET (Oliw, 1994; Oliw and Oates, 1981;

48

J.A. Hasler et al. / Molecular Aspects of Medicine 20 (1999) 1±137

Chacos et al., 1982). As with most P450 activities, early studies demonstrated that the type of reaction catalyzed, as well as, the pro®le of metabolites generated was highly dependent on the animal and tissue source of microsomal enzymes, the nutritional and hormonal state of the animal, its sex, age and prior exposure to drugs, pesticides or to known inducers of microsomal P450 (Oliw, 1994; Makita et al., 1996; Capdevila et al., 1981a; Falck et al., 1990; Capdevila et al., 1990). For example, while rat and rabbit liver microsomal fractions catalyze preferentially arachidonic acid epoxidation (McGi€, 1991; Oliw, 1994; Oliw and Oates, 1981; Chacos et al., 1982), rat and rabbit kidney microsomes metabolize the fatty acid to generate mainly x and xÿ1 alcohols (McGi€, 1991; Oliw, 1994; Capdevila et al., 1981a, 1982, 1990, 1992; Oliw and Oates, 1981; Morrison and Pascoe, 1981; Oliw et al., 1982; Falck et al., 1990). Based on studies with either microsomal fractions or reconstituted system containing several solubilized and puri®ed P450 isoforms, it was concluded that P450 controls, in a protein speci®c fashion, the regioselectivity of arachidonic acid oxidation at two di€erent levels: (a) the type of reaction catalyzed, i.e., ole®n epoxidation (EETs), allylic oxidation (HETEs) or hydroxylation at the C16 C20 sp3 carbons (16-, 17-, 18-, 19-, and 20-OH-AA) and, (b) to a lesser extent, in selecting the regio-selectivity of oxygen insertion. Inasmuch as the role and/or relevance of P450 to in vivo HETE formation remains unsubstantiated, most research e€orts have concentrated on the epoxygenase and x/xÿ1 hydroxylase branches of the P450 AA monooxygenase pathway; the predominant P450 catalyzed reactions in most organ tissues. 5.4.2.1. AA x/xÿ1 hydroxylase reaction. The catalysis of fatty acid omega and omegaÿ1 oxidation (x/xÿ1 oxidation) is one of the best established P450 reactions (Kupfer, 1982). The regulation of these reactions by the a-subtype of ligand activated peroxisomal proliferator activated receptors (PPARa ) (Johnson et al., 1996) suggest that, in conjunction with peroxisomal b-oxidation reactions, fatty acid x/xÿ1 oxidation contributes to lipolytic pathways and participates in the control of body fatty acid homeostasis (Johnson et al., 1996). Fatty acid x/xÿ1 oxidation is regulated in vivo by a variety of factors including, animal age, diet, starvation, administration of fatty acids, hypolipidemic drugs, aspirin, steroids and diabetes (Johnson, 1996; Capdevila et al., 1992; Masters et al., 1989; Kupfer et al., 1988; Sharma et al., 1989b; Orellana et al., 1989). The P450-catalyzed hydroxylation of arachidonic acid at its x and xÿ1 carbons was ®rst documented, in 1981 when rabbit kidney cortex microsomes were shown to catalyze the NADPH-dependent formation of 19- and 20-hydroxyeicosatetraenoic acids (19-, and 20-OH-AA, respectively) (Morrison and Pascoe, 1981). Since then, these hydroxylation reactions has been demonstrated in several tissues, including human liver and kidney, as well as rat and rabbit liver and kidney (Oliw, 1994). More recently, the 16-, 17-, and 18-hydroxyeicosatetraenoic acids have been added to the list of products generated by the P450 arachidonic acid x/xÿ1 hydroxylase activity (Falck et al., 1990; Oliw, 1989, Laethem et al., 1993). While the oxygen chemistry and reaction mechanisms responsible for the x/xÿ1 hydroxylation of arachidonic acid of saturated fatty acids are probably similar, for arachidonic acid these reac-

J.A. Hasler et al. / Molecular Aspects of Medicine 20 (1999) 1±137

49

tions impose additional steric requirements on the P450 protein catalyst. Hydroxylation at the thermodynamically less reactive C16 through C20 and not at the chemically comparable C2 through C4 suggest a highly rigid and structured binding site for the arachidonic acid template. Thus, the binding site must be capable of positioning the acceptor carbon atoms not only in optimal proximity to the hemebound active oxygen but also with complete segregation of the reactive 5,6-, 8,9-, 11,12-, and 14,15-ole®ns and of the bis-allylic C7 , C10 and C13 methylene carbons. It is of interest that while several P450 4A isoforms are highly selective for the x-hydroxylation of fatty acids and prostanoids, a P450 isoform has yet to be identi®ed that is regioselective for xÿ1 hydroxylation of these molecules. Several rat, rabbit or human members of the P450 4A gene family have been puri®ed and/or cloned and expressed (Kupfer, 1982; Nelson et al., 1996; Hardwick et al., 1987; Sharma et al., 1989a; Imaoka et al., 1989, 1990, 1993a; Kimura et al., 1989a,b; Stromsteadt et al., 1990; Matsubara et al., 1987; Johnson et al., 1990; Roman et al., 1993; Yokotani et al., 1991; Yokotani et al., 1989; Sawamura et al., 1993; Kawashima et al., 1992; Palmer et al., 1993; Wang et al., 1997). While individual P450 4A isoforms show regioselectivity for either the x or the xÿ1 hydroxylation of lauric acid or prostanoids, to date, all the P450 4A isoforms characterized, metabolize arachidonic acid to either 20-OH-AA or to mixtures of 20-OH-AA and 19-OH-AA; i.e., none of these P450 4A isoforms show exclusive regioselectivity for the fatty acid C19 position. Interestingly, studies with inducers of microsomal P450, and the reconstitution of the AA hydroxylase with puri®ed enzymes showed that P450s 1A1, 1A2 and 2E1 may be responsible for the hydroxylations occurring at the C16 through C19 carbon atoms of AA (Falck et al., 1990; Laethem et al., 1993). Thus, while P450 1A1 and 1A2 are more or less selective for hydroxylations at C19 and C16 , respectively (Falck et al., 1990), puri®ed P450 2E1 metabolizes AA to 18(R)-and 19(S)-OH-AA along with other products (Laethem et al., 1993). Finally, the demonstration of 20-OH-AA excretion in human urine con®rmed the participation of the P450 in the x-hydroxylation of endogenous AA pools (Schwartzman et al., 1991; Prakash et al., 1992). The majority of the 20-OH-AA in rat urine is found conjugated to glucuronic acid, an established route for the excretion of hydroxylated compounds (Prakash et al., 1992). 5.4.2.2. The arachidonic acid epoxygenase reaction. The catalysis of AA epoxidation by P450 was inferred by the isolation of 11,12- and 14,15-dihydroxyeicosatrienoic acids (DHETs) from incubates containing kidney cortex microsomes, AA, and NADPH (Oliw, 1982). Soon after, rat liver microsomes were shown to catalyze the NADPH-dependent epoxidation of AA to 5,6-, 8,9-, 11,12-, and 14,15-EET (Chacos et al., 1982). To date, the catalysis of EET formation by puri®ed P450s, microsomal fractions, or isolated cell preparations has been demonstrated in numerous tissues, including liver, kidney, pituitary, brain, adrenal, endothelium, pancreas, and ovaries (McGi€, 1991; Oliw, 1994; Harder et al., 1995; Makita et al., 1996; McGi€ et al., 1996; Schwartzman et al., 1996). Two factors contributed to focus attention in the study of chemistry, biochemistry and physiological roles of this branch of the AA cascade: (a) the demonstration of the potent biological

50

J.A. Hasler et al. / Molecular Aspects of Medicine 20 (1999) 1±137

activities of its metabolites, the EETs, and (b) the establishment of EETs as endogenous constituents of organs such as liver, kidney, and lung and of human and rat plasma and urine (Makita et al., 1996; Capdevila et al., 1984a, 1992; Karara et al., 1989, 1990, 1992; Falck et al., 1987). In mammals, the epoxidation of polyunsaturated fatty acids to bis-allylic, cis-epoxides is unique to the P450 enzyme system and, at di€erence with the fatty acid x/xÿ1 oxygenase, more or less selective for AA. Finally, the fact that the P450-dependent epoxidation of AA generates only cis-epoxides suggest that epoxidation proceeds by a concerted pathway or that, alternatively, a rigid active site binding geometry restricts the freedom of C±C rotation for the transition state. The P450 isoform heterogeneity of the microsomal epoxygenases was ®rst indicated by studies showing that the regioselectivity of AA epoxidation was more or less organ speci®c, thus, for example, while brain microsomes formed 5,6-EET as their major epoxygenase product (Capdevila et al., 1983), the 11,12-EET is the predominant regioisomer formed by rat liver and kidney microsomal fractions (Chacos et al., 1982; Capdevila et al., 1992). The changes in EET chirality that resulted after animal treatment with selected P450 inducers further demonstrated the molecular heterogeneity of the AA epoxygenase (Capdevila et al., 1990). For example, Phenobarbital treatment increased, in a time-dependent fashion, the regio-and enantiofacial selectivity of the rat liver microsomal epoxygenase(s) (Capdevila et al., 1990). The Phenobarbital induced increases in stereoselectivity resulted in a remarkable inversion in absolute con®guration of the EETs produced by the microsomal enzymes (Capdevila et al., 1990). It was subsequently shown that AA epoxidation is highly enantioselective and that P450 controls, in an isoform speci®c fashion, the regio-and enantioselectivity of the epoxygenase reaction (Oliw, 1994; Falck et al., 1990; Capdevila et al., 1990). These results established that, in contrast to the cyclooxygenase and lipoxygenases enzymes, the regio-and enantioselectivity of the AA epoxygenase was variable and P450 isoform speci®c (Makita et al., 1996; Falck et al., 1990; Capdevila et al., 1990). Furthermore, amongst the enzymes of the arachidonate cascade, the P450 epoxygenase is unique in that its regio- and stereochemical selectivity is under regulatory control and can be experimentally altered, in vivo, by animal manipulation (Table 3) (Makita et al., 1996; Falck et al., 1990; Capdevila et al., 1990). Reconstitution of the P450 AA monooxygenase activity using puri®ed P450 isoforms and/or recombinant proteins demonstrated that members of the P450 2 gene family were the only isoforms that metabolized the fatty acid predominantly by enantioselective epoxidation (Makita et al., 1996; Capdevila et al., 1990, 1992; Falck et al., 1990; Laethem et al., 1993; Keeney et al., 1998). Isoforms of the 2B and 2C subfamily so far identi®ed as epoxygenases include rat 2B1, 2B2, 2B12, 2C11, 2C23, 2J3, 2J4 (Oliw, 1994; Falck et al., 1990; Laethem et al., 1993; Kenney et al., 1998; Zeldin et al., 1997) rabbit 2B4, 2C1 and 2C2 (Daikh et al., 1994; Knickle and Bend, 1994), and human 2C8, 2C9/10, and 2J2 (Zeldin et al., 1995; Wu et al., 1996). Importantly, P450s 2C23, 2C24 and 2C11, the three major 2C subfamily isoforms expressed in the rat kidney (Makita et al., 1996; Capdevila et al., 1992; Karara et al., 1994; Imaoka et al., 1993b; Makita et al., 1994), were cloned, expressed, identi®ed as active epoxygenases, and shown to

J.A. Hasler et al. / Molecular Aspects of Medicine 20 (1999) 1±137

51

account for most of the rat kidney epoxygenase activity (Karara et al., 1993; manuscript in preparation). Furthermore, P450s 2C8 and 2C23 are the predominant epoxygenases expressed in human and rat kidneys, respectively (Zeldin et al., 1995; Karara et al., 1993). While P450s 1A1, 1A2 and 2E1 are active AA x/xÿ1 oxygenases, they also produce low and variable amounts of EETs ( 6 20 of total products) (Falck et al., 1990; Capdevila et al., 1990; Laethem et al., 1993). A unique case is that of a P450 puri®ed from the livers of dioxin treated chick embryos (Rifkind et al., 1994). This protein has several structural features typical of proteins of the 1A gene subfamily, but metabolizes AA to EETs as the major reaction products (75% of total products) (Rifkind et al., 1994). The ability of a single 2C P450 protein to catalyze the enantioselective epoxidation of more than one fatty acid ole®n was clearly demonstrated using recombinant proteins (Zeldin et al., 1995; Karara et al., 1993; Imaoka et al., 1993b). For example, the ®rst recombinant epoxygenase characterized, P450 2C23, catalyzed the enantioselective epoxidation of AA to 8,9-, 11,12-, and 14,15-EET (27%, 54% and 19% of total products, respectively) (Karara et al., 1993). The recombinant protein generated 8(R),9(S)-, 11(R),12(S), and 14(S),15(R)-EETs with optical purities of 95%, 85% an 75%, respectively (Karara et al., 1993). The regio- and enantioselective formation of 11,12-, and 14,15-EET by recombinant P450 2C2 and a P450 2CAA, puri®ed from rabbit kidney cortex, was reported recently (Daikh et al., 1994). The structural analysis of the EETs generated by puri®ed 2B and 2C P450 epoxygenases showed that, among these proteins, none was selective for the epoxidation of a single fatty acid ole®n. Thus, although highly enantioselective, AA epoxidation shows a more limited regioselectivity (Capdevila et al., 1990; Karara et al., 1993). It is of signi®cance that the degrees of stereochemical selectivity displayed by AA epoxygenase isoforms are unusually high for P450 catalyzed oxidations of unbiased, noncyclic molecules such as AA. In view of its well known catalytic versatility, the in vitro catalysis of AA epoxidation by microsomal P450s was not completely unexpected. It was apparent then that the uniqueness and signi®cance of these reactions was going to be de®ned by whether or not the enzyme system participated in the in vivo metabolism of the fatty acid. Since asymmetric synthesis is an accepted requirement for the biosynthetic origin of most eicosanoids, this issue was resolved by the demonstration of the presence of chiral EETs in samples extracted from rat and human organs (Karara et al., 1989, 1990). A distinctive feature of the endogenous EET pools in rat liver and kidney was their presence esteri®ed to the sn-2 position of several cellular glycerophospholipids ( P 92% of the total liver EETs) (Karara et al., 1991). EET-glycerophospholipid formation required a multistep process, initiated by the P450 enantioselective epoxidation of AA, ATP-dependent activation to the corresponding EET-CoA derivatives, and EET enantiomer-selective lysolipid acylation (Karara et al., 1991). The observed in vivo EET esteri®cation process appears to be unique since most endogenously formed eicosanoids are either secreted, excreted, or undergo oxidative metabolism and excretion. There are, however, reports of esteri®cation by isolated cells of exogenously added HETEs and EETs (Brezinski and Serhan, 1990; Legrand et al., 1991).

52

J.A. Hasler et al. / Molecular Aspects of Medicine 20 (1999) 1±137

The biosynthesis of endogenous pools of phospholipids containing esteri®ed EET moieties in rat liver, kidney, brain and plasma and in human kidney and plasma (Makita et al., 1996; Karara et al., 1989, 1990, 1991, Karara et al., 1992; Falck et al., 1987), indicate new and potentially important functional roles for P450 in the generation of cellular glycerophospholipids pools containing oxidized fatty acid moieties. Furthermore these studies also show, in contrast to most eicosanoids, the potential for the cellular generation of performed bioactive EETs via hydrolytic reactions, thus obviating the need for AA oxidative zmetabolism. 5.4.3. Functional signi®cance of the AA monooxygenase metabolites The analysis of functional roles for the metabolites of the P450 AA monooxygenase has developed into an area of intense research and the list of biological activities attributed to these metabolites has grown considerably during the last few years. Among these, we emphasize: (a) Vasoactive properties: The EETs are in vitro systemic vasodilators with 5,6-EET as the most potent regioisomer while 8(S),9(R)EET, the endogenous enantiomer in rat kidney, is a stereoselective renal vasoconstrictor (McGi€, 1991; Oliw, 1994; Harder et al., 1995; Makita et al., 1996; McGi€ et al., 1996; Schwartzman et al., 1996; Rahman et al., 1997; Harder et al., 1997). On the other hand, 20-OH-AA, or products of its oxidative metabolism, are powerful vasoconstrictors while 19-OH-AA is a stereospeci®c renal vasodilator (McGi€, 1991; Oliw, 1994; Harder et al., 1995; Makita et al., 1996; McGi€ et al., 1996; Schwartzman et al., 1996; Rahman et al., 1997; Harder et al., 1997; Escalante et al.,1993; Ma et al.,1993). (b) Ion transport: The EETs, and 5,6-EET in particular, increase cytosolic Ca‡ concentrations in several cell systems (McGi€, 1991; Oliw, 1994; Harder et al., 1995; Makita et al., 1996; McGi€ et al., 1996; Schwartzman et al., 1996; Rahman et al., 1997; Harder et al., 1997), as well as the single channel open probability of Ca‡ activated K‡ channels (Hu and Kim, 1993; Zhou et al., 1996b). A role for the EETs in mediating the natriuretic and mitogenic responses to angiotensin II and epidermal growth factor (EGF) has been proposed (Harris et al., 1990; Burns et al., 1995; Madhun et al., 1991). Both natriuretic and diuretic e€ects have been described for 20-OH-AA (McGi€, 1991; Oliw, 1994; Harder et al., 1995; Makita et al., 1996; McGi€ et al., 1996; Schwartzman et al., 1996; Rahman et al., 1997; Harder et al., 1997). The stimulation and inhibition of renal Na‡ , K‡ /ATPase by 19- and 20-OH-AA, respectively, has been reported (McGi€, 1991; McGi€ et al., 1996). In rabbit mTALH cells 20-OH-AA and the corresponding 1,20-dicarboxylic acid blocked the Na‡ ,K‡ ,2Cl-cotransporter and a Na‡ /K‡ ATPase (McGi€, 1991; McGi€ et al., 1996). Earlier, mostly inhibitor based, studies de®ned the EETs as endothelium dependent vaso-relaxants (Hecker et al., 1994). Based on endothelial cell synthesis and release (Rosolowsky and Campbell, 1996), EET-mediated relaxation of preconstricted arteries (Hecker et al., 1994), and G-protein mediated EET opening of vascular smooth cell Ca‡‡ -activated K‡ channels, the EETs were identi®ed as endothelium-derived hyperpolarizing factors (EDHF) (Campbell et al., 1996; Li and Campbell, 1997). These exiting results, with their potential to provide a molecular understanding of EET vasoactivity, are contributing to integrate EET

J.A. Hasler et al. / Molecular Aspects of Medicine 20 (1999) 1±137

53

transport and vasoactive properties into a coherent mechanistic description, amenable to experimental analysis. Importantly, 20-OH-AA, recently identi®ed as an endogenous inhibitor of this type of Ca‡ activated K‡ channels (Al-Ping Zou et al., 1996) is also a powerful vasoconstrictor (Escalante et al., 1993; Ma et al., 1993) and behaves, therefore, as an EET functional antagonist. (c) Gene regulation: The potential for eicosanoid participation as ligands for orphan receptors is an area of intense current interest (Johnson et al., 1996). The mitogenic activities of 14,15-EET and 20-OH-AA have been reported (Harris et al., 1990; Burns et al., 1995; Fangming et al., 1995), and the P450 AA monooxygenase-dependent regulation of rat liver genes suggested (Tollet et al., 1995). Further, preliminary studies identi®ed 8(S),9(S)- and 14(R),15(R)-dihydroxyeicosatetraenoic acid, the EET hydration products, as ligands for the a and b subtypes of the peroxisomal proliferator activated receptor (PPAR). The early proposal by J.C. McGi€ and collaborators of a role for the kidney P450 AA monooxygenase in the pathophysiology of genetically controlled experimental hypertension (McGi€, 1991) provided new and unique opportunities for the study of genetic, biochemical and functional correlations indicative of physiological and/or pathophysiological signi®cance (McGi€, 1991; Oliw, 1994; Harder et al., 1995; Makita et al., 1996; McGi€ et al., 1996; Schwartzman et al., 1996; Rahman et al., 1997; Harder et al., 1997). Thus, based on: (a) biochemical and temporal correlates of renal AA monooxygenase gene expression and enzymatic activity with the development of high blood pressure in spontaneously hypertensive (SHR) rats (McGi€, 1991), (b) the prevention of hypertension in SHR rats by SnCl2 -mediated depletion of renal AA P450s (McGi€, 1991), (c) the normotensive e€ects of SnCl2 -mediated renal P450 depletion in hypertensive SHR animals (McGi€, 1991), and (d) the functional e€ects of 20-OH-AA and its oxygenated metabolites, kidney P450s were implicated in the development of hypertension in the rat SHR/WKY model of spontaneous hypertension and a pro-hypertensive role was identi®ed for products of the renal AA x/xÿ1 hydroxylase (McGi€, 1991; Harder et al., 1995; Makita et al., 1996; McGi€ et al., 1996; Schwartzman et al., 1996; Rahman et al., 1997; Harder et al., 1997). These results were recently con®rmed using hypertensive SHR animals and a more selective, mechanismbased, inhibitor of P450 (Ping et al., 1998). Similarly, a role for 20-OH-AA and P450 4A2 in salt sensitive hypertension was subsequently proposed based on: (a) inhibitor studies (Stec et al., 1997), (b) di€erences between Dahl salt sensitive (DS) and salt resistant rats (DR) in the activity and expression levels of their AA x/ xÿ1 hydroxylases (Ma et al., 1994b), and (c) normalization of Cl-transport in the TALH segment of DS rats by 20-OH-AA (Zhou et al., 1996a). Importantly, alleles at the P450 4A2 locus were found to co-segregate with hypertension in DS rats (Stec et al., 1996). On the other hand, an opposite, antihypertensive, role for the products of the AA epoxygenase was indicated by biochemical and functional correlates of epoxygenase activity, dietary salt intake and blood pressure (Makita et al., 1994, 1996). In Sprague-Dawley rats, excess dietary salt induces the kidney epoxygenase activity and markedly increases the urinary levels of its metabolites (Capdevila et al., 1992). Clotrimazole inhibition of the salt responsive epoxygenase

54

J.A. Hasler et al. / Molecular Aspects of Medicine 20 (1999) 1±137

leads to the development of a clotrimazole-dependent, salt sensitive hypertension (Makita et al., 1994). Furthermore, high salt diets induce the renal AA epoxygenase in normotensive Dahl salt resistant (DR) rats while, in contrast, under similar conditions, hypertensive Dahl salt sensitive (DS) animals fail to induced their salt responsive kidney AA epoxygenase (Makita et al., 1994). Finally, a polymorphism in the gene coding for a P450 2C epoxygenase has been documented in DS rats (Makita et al., 1996). However, co-segregation studies failed to demonstrate an association between this polymorphism and salt sensitivity in these animals. 5.5. Conclusion, unresolved issues, future perspectives As it has been emphasized repeatedly, the potential for physiological and/or pathophysiological function provides the needed justi®cation for the intensive studies of the role of P450 in arachidonic metabolism and bioactivation. The contributions of many laboratories has provided an extensive enumeration of biological activities associated with P450-derived eicosanoids (McGi€, 1991; Oliw, 1994; Harder et al., 1995; Makita et al., 1996; McGi€ et al., 1996; Schwartzman et al., 1996; Rahman et al., 1997; Harder et al., 1997). While exogenously added eicosanoids are useful tools for the analysis of biological function, this type of approach may result in limited and/or arti®cial descriptions of eicosanoid roles and biological signi®cance. Often, bioactive eicosanoids are but one component of what are usually complex, multistep signaling cascades that may include, among other things, coordinated, time dependent, changes in cytosolic ion, cyclic nucleotide and phosphoinositide concentrations, changes in the regulation of phosphorylation and de-phosphorylation cascades, phospholipase(s) activation, and changes in the turnover rate of selected glycerophospholipid pools. On the other hand, the last 5 years have been characterized by the application of molecular approaches to the studies of the biochemical and functional signi®cance of the P450-dependent AA epoxygenase(s) and x/xÿ1 oxygenase(s). As summarized, several P450 isoforms have been either cloned and/or their cDNAs expressed and their enzymatic activities, tissue and/or organ speci®c expression, regulation in salt sensitive and spontaneous hypertension characterized. Additionally, signal transduction pathways for several P450-derived eicosanoids are beginning to be dissected, and electrophysiology is providing important molecular insights into their ion transport e€ects and mode of action. Indeed, the progress achieved during the last few years situates us now in the threshold of an era in which, by the use of molecular genetic techniques, including gene disruption and /or transfection, we will be able to probe P450 isoform speci®c, gene-dependent phenotypes and thus, their functional roles. 5.6. Acknowledgments The authors of chapter 5 (J. Capdevila, V. Holla, C. Helvig and J. Falck) are grateful to the USPHSNIH NIDDK and NIGMS Institutes for their support of

J.A. Hasler et al. / Molecular Aspects of Medicine 20 (1999) 1±137

55

research carried out in their laboratories and cited here. The editorial assistance of Ms. Mary Couey is greatly appreciated. 6. E€ects of disease on expression and regulation of CYPs 6.1. Introduction It has been recognized that drug metabolism is impaired in severe liver disease for 40 years, but only in the last 10 years have studies started to specify the alterations among individual enzymes and to characterize the molecular basis for these changes. Likewise, when the author last reviewed drug metabolism in extrahepatic diseases (Farrell, 1987), comparatively little was known, other than extrapolation from animal studies, about which disease processes could alter the activity of hepatic drug metabolizing enzymes. Since then molecular approaches have allowed a more sophisticated and detailed analysis of alterations in drug metabolism in a variety of disease states, particularly in infections and with in¯ammatory modulators, and in some endocrine disorders. In this review, a major focus will be on CYP expression in liver disease but advances in these other areas will be considered. Connecting themes will be the extent to which clinically relevant changes in drug metabolism can be explained by altered constitutive expression of the genes that encode cytochrome P450 (P450) proteins, the CYP genes; this process has been referred to as dysregulation. Changes in the expression of P450 in extrahepatic tissues could also be relevant to such diverse processes as atherosclerosis, hypertension, cancer and hormonally dependent tumors, but these interesting topics are beyond the scope of the present work. 6.2. E€ects of liver disease 6.2.1. Changes in total hepatic P450 levels Many tissues express CYP genes, but only the liver exhibits activities of P450 enzymes high enough to play a signi®cant role in drug elimination. It might be expected, therefore, that liver disease, which can potentially reduce hepatic functional mass, could decrease the capacity for drug metabolism. Early studies of drug clearance in patients with liver disease tended to con®rm this, but not unambiguously so because in some conditions, and/or for some drugs there was no reduction in drug clearance (Nelson, 1964; Farrell et al., 1978; McLean and Morgan, 1991; Morgan and McLean, 1995). Further, while pathophysiologic determinants such as hepatic drug delivery (hepatic blood ¯ow, plasma protein binding), hepatic uptake, oxygen supply and bile secretion contribute to altered hepatic drug clearance in cirrhosis (reviewed in McLean and Morgan, 1991; Morgan and McLean, 1995), these variables do not explain all the discrepancies between di€erent types of drugs. Lowered hepatic levels of total P450 and related enzyme activities have been observed in severe liver disease for more than 25 yr (Schoene et al., 1972; Ahmad and Black, 1977; Farrell et al., 1979a; Hoensch et al., 1979; Boobis et al., 1980; Brodie

56

J.A. Hasler et al. / Molecular Aspects of Medicine 20 (1999) 1±137

et al., 1981). These changes contribute importantly to impairment of drug clearance and hepatic detoxication. P450s are clearly more susceptible to the e€ects of liver disease than are P450-reductase or phase II enzymes, such as glucuronyl transferases, sulphatases, N-acetyl transferases and amino acid transferases involved with drug and steroid conjugation reactions (Hoyumpa and Schenker, 1991). Drug metabolism catalyzed exclusively by the latter types of enzymes is preserved until the most advanced stages of liver failure. Indeed, some studies have found increased expression of glucuronyl transferases per unit cell in human cirrhosis (Debinski et al., 1995), suggesting that compensatory up-regulation can occur to o€set the reduced number of functional hepatocytes. In humans, signi®cant impairment of P450 expression is found only in patients with severe liver disease (Schoene et al., 1972; Ahmad and Black, 1977; Farrell et al., 1979a; Hoensch et al., 1979). Severe liver disease includes hepatitis with liver failure and decompensated cirrhosis. Conversely, hepatic total P450 levels and related enzyme activities are within the broad normal range in uncomplicated cases of cholestasis, mild-moderate hepatitis, steatosis (fatty liver) and in cases of clinically compensated cirrhosis (Ahmad and Black, 1977; Farrell et al., 1979a; Hoensch et al., 1979; Boobis et al., 1980; Brodie et al., 1981). Cirrhosis is the morphological end-state of chronic liver disease.It is characterized by densely connecting ®brosis and disruption of hepatic architecture by regeneration nodules. Changes in hepatic blood ¯ow result from distortion of the hepatic microcirculation (perisinusoidal ®brosis or capillarization) and eventually from portal hypertension, processes which lead to intrahepatic and extrahepatic ``shunting'' of blood away from hepatocytes. This impairs the clearance of rapidly metabolized (high clearance) compounds by decreasing drug delivery and thereby hepatic uptake. It may also create a di€usion barrier for the supply of oxygen, the cosubstrate for P450-dependent mixed function oxidases, thereby contributing to impaired metabolism in vivo (McLean and Morgan, 1991; Morgan et al., 1995; Hickey et al., 1995, 1996; Froomes et al., 1998). In general, the activity of P450-dependent drug oxidases, as measured directly in hepatic microsomes is also decreased in the cirrhotic human liver. In discussing the changes observed, it should be borne in mind that recent studies have employed explanted liver tissue of patients with end-stage liver disease coming to hepatic transplantation. They therefore represent a ``worst case'' or ``end-of-the-road'' scenario that may not apply to the majority of patients with liver disease observed in clinical practice. Another consideration is the extent to which hepatic CYPs in the diseased liver can respond to pharmacological induction or to other environmental factors that alter expression of P450 (Table 4). It has sometimes been assumed that a ``sick liver'' abrogates the protein synthetic response to transcriptional regulators. There is, however, little evidence to support this assumption, except in the extreme state of pre-terminal acute liver failure. To the contrary, several in vivo studies of drug clearance have demonstrated clearly that most patients with cirrhosis remain responsive to the stimulation of hepatic drug metabolism by such diverse agents as rifampicin (Levi et al., 1968), glutethimide (Farrell et al., 1979b), aldactone (Miguet

J.A. Hasler et al. / Molecular Aspects of Medicine 20 (1999) 1±137

57

Table 4 Factors that could in¯uence cytochrome P450 expression in liver disease Disease related

Genetic, environmental

Severity Type of pathophysiology, especially presence or absence of cholestasis Portosystemic shunting, and resultant hormonal changes Dedi€erentiation of hepatocytes Other disorder, e.g., diabetes mellitus Cytokines

Pharmacogenetics Enzyme induction by drugs, e.g. rifampicin, aldactone Cigarette smoking Nutrition

et al., 1980) and cigarette smoke (Joeres et al., 1988; Coverdale et al., 1995). It follows that the decreased expression of total P450 found in severe liver disease is, to a certain extent, enigmatic because the potential for up-regulation of the encoding CYP genes clearly remains. This raises the possibility that one explanation for decreased CYP expression might be failure of constitutive regulatory mechanisms to compensate for disease-related loss of hepatic enzyme mass. As discussed below, the speci®c changes in expression among individual CYPs is evidence in support of this proposal. 6.2.2. Changes in expression of individual CYPs Early studies using hepatic microsomal fractions showed unequal changes in oxidation rates for di€erent P450 substrates (Farrell et al., 1979a; Iqbal et al., 1990). These ®ndings have now been explained by di€erent e€ects of liver disease on individual P450 proteins or CYP genes (Table 5). As indicated by mRNA levels, the reductions are usually exerted at a pre-translational level. Even among patients with liver failure, expression of some hepatic CYPs remains una€ected and can be induced to supraphysiologic levels by pharmacologic agents. 6.2.2.1. CYP1A. The most reproducible changes of CYP expression in human cirrhosis are decreased CYP1A2 mRNA and P450 1A immunoreactive protein. Such changes have been reported in three separate studies (Guengerich and Turvey, 1991; Lown et al., 1992; George et al., 1995a,b), and corresponding reductions in P450 1Acatalyzed ethoxyresoru®n O-deethylase have been documented (Iqbal et al., 1990; George et al. 1995a). The profound fall in P450 1A probably explains the early observation that aryl hydrocarbon hydroxylase activity was much more reduced in Table 5 E€ects of di€erent types of liver disease on individual P450 proteinsa Type of cirrhosis Hepatocellular Cholestasis

P450 1A2 (%) b

29 18b

P450 2C (%)

P450 2E1 (%)

P450 3A (%)

57 34b

81 49b

25b 41

a Data are relative levels of individual P450 proteins expressed as a percentage of the mean value in controls (patients without signi®cant liver disease). b Values that were signi®cantly di€erent from control. Modi®ed from George et al. (1995a).

58

J.A. Hasler et al. / Molecular Aspects of Medicine 20 (1999) 1±137

liver from patients with severe liver disease than was P450 3A-dependent ethylmorphine N-demethylase (Farrell et al., 1979a). Patients with all types of cirrhosis appear to have equivalent reductions in CYP1A expression (Table 5). The importance of decreased expression of CYP1A is indicated by the profound changes in theophylline (Colli et al., 1988), ca€eine (Desmond et al., 1980) and antipyrine metabolism in liver disease (Branch et al., 1973; Andreasen et al., 1974; Farrell et al., 1978). These agents are all extensively metabolized by P450 1A enzymes, although other P450s (particularly 3A4, 2C8 and 2E1) catalyze additional reactions in some cases (Gu et al., 1992; Tassaneeyakul et al., 1994; Zhang and Kaminsky, 1995; Engel et al., 1996; Sharer and Wrighton, 1996). Decreased ca€eine and antipyrine metabolism appears to occur relatively early in progressive types of chronic liver disease, such as chronic viral hepatitis and primary biliary cirrhosis (Lautz et al., 1988; Williams and Farrell, 1989; Rosenblum et al. 1992; Coverdale et al., 1995). As a result, clearance tests based on such substrates have been employed to quantify liver function in attempts to monitor progress in chronic liver disease (reviewed in Farrell and Coverdale, 1998). It should be noted that P450 1A2 is induced to a considerable extent by cigarette smoking (Sesardic et al., 1988). Our observations of antipyrine clearance (Coverdale et al., 1995), and those of others on ca€eine clearance (Joeres et al., 1988) indicate that such induction is still present in cirrhosis. 6.2.2.2. CYP2C. These genes appear to be relatively less a€ected by liver disease than CYP1A. Thus, patients with cirrhosis eliminate tolbutamide, a P450 2C9 substrate, at a normal rate (Nelson, 1964). On the other hand, racemic mephenytoin can be used as a probe for CYP2C19, and the activity of this enzyme was rather sensitive to the e€ects of liver disease (Arns et al., 1997; Adedoyin et al., 1998). Using monospeci®c antibodies, Lown et al. (1992) found minor decreases in hepatic P450 2C9 protein levels in a small number of cases, but no change in 2C8. Guengerich and Turvey (1991) found no alteration in levels of 2C protein using a subfamily speci®c antibody, and with the same kind of probe, George et al. (1995a) found no change in ``hepatocellular'' liver diseases like chronic hepatitis. In contrast, there were signi®cant reductions of P450 2C protein levels in livers from patients in whom cirrhosis resulted from prolonged impairment of bile ¯ow (cholestasis). The di€erent results between studies could therefore be explained, at least in part, by the type and severity of liver disease. 6.2.2.3. CYP3A. Among these genes, CYP3A4 is quantitatively the most important P450 in human liver, accounting for 40% of total P450 and catalyzing the oxidation of at least 50% of currently used drugs (Guengerich et al., 1998). Earlier studies found only subtle or no changes in CYP3A expression and activity in cirrhosis. For instance, Guengerich and Turvey (1991), found a nonsigni®cant trend toward reduction in P450 3A protein in human cirrhosis, whereas Lown et al. (1992) observed a signi®cant reduction in the erythromycin breath test result (re¯ecting CYP3A activity in vivo), but no signi®cant alteration in hepatic 3A protein levels. On the other hand, pharmacokinetic studies have clearly demonstrated that the clearance of

J.A. Hasler et al. / Molecular Aspects of Medicine 20 (1999) 1±137

59

drugs metabolized by 3A proteins, such as lidocaine (Huet and Villeneuve, 1983; Meyer-Wyss et al., 1993) and nifedipine (Kleinbloesem et al., 1986) is impaired in cirrhosis. Findings from the authorÕs laboratory have helped resolve the apparent inconsistencies between in vitro and in vivo studies (George et al., 1995a). Thus, values for both 3A protein and 3A-catalyzed testosterone 6b-hydroxylase activity were signi®cantly less than controls in patients with some types of cirrhosis (hepatocellular disease). Conversely, levels of 3A mRNA and protein in cholestatic cirrhosis were more variable, and mean values were not sign®cantly less than controls. Interestingly, in vitro enzyme activities based on substrates extensively metabolized by CYP3A (ethylmorphine, aminopyrine) have shown the same preservation in primary biliary cirrhosis (a cholestatic liver disease) with signi®cant reductions in other liver diseases (Farrell et al., 1979a; Iqbal et al., 1990). 6.2.2.4. CYP2E1. Changes in expression of CYP2E1 protein in human liver disease are also variable. For instance, Lown et al. (1992) found no change in nine end-stage livers, whereas Guengerich and Turvey (1991) found a signi®cant decrease in mean values among 42 cirrhotic livers compared with controls. The ®ndings of George et al. (1995a,b) partly resolve these di€erences. Their results clearly show that the changes in CYP2E1 expression depend on the type of liver disease. Similar to 2C9, and in contradistinction to 3A, P450 2E1 protein and related enzyme activities were decreased in cholestatic types of cirrhosis but remained normal in hepatocellular liver diseases. It is noteworthy, therefore, that the study of Guengerich and Turvey (1991) mainly included cases of the latter types of liver disease. One possible reason for the sustained expression of CYP2E1 in hepatocellular diseases may be the in¯uence of alcohol and diabetes on expression of CYP2E1. It is clear that alcohol induces expression of CYP2E1, particularly by post-translational mechanisms (Hu et al., 1995; Lieber, 1997), while diabetes increases expression of CYP2E1 at both mRNA and protein levels (de Waziers et al., 1995). Indeed, it seems likely that the increased expression of CYP2E1 observed in alcoholic liver disease plays a role in pathogenesis of this disease (Nanji and Zakim, 1996; Lieber, 1997). The mechanism is attributable to generation of reduced oxygen species, which in turn contribute a state of intra-hepatic oxidative stress. Likewise, Weltman et al. (1996, 1998) have observed similar increases in expression of CYP2E1 in an animal dietary model of steatohepatitis, and in patients with the clinical entity of nonalcoholic steatohepatitis (NASH). The latter disease is particularly common in NIDDM, obesity and hyperlipidemia. In early studies, we observed decreased antipyrine metabolism in patients with NASH (Fiatarone et al., 1991), a ®nding which accords with decreased expression of CYPs other than 2E1 in the animal model (Weltman et al., 1996). Some in vivo studies also indicate that CYP2E1 activity is increased in patients with NASH (Leclerq et al., 1996). 6.2.2.5. Other CYPs. There are fewer data on other CYPs in human liver disease, although in vivo studies with enzyme-speci®c probes have yielded interesting data. Metabolism of debrisoquin was not altered in moderately severe liver disease,

60

J.A. Hasler et al. / Molecular Aspects of Medicine 20 (1999) 1±137

indicating that CYP2D6 may be refractory to the e€ects of liver disease (Adedoyin et al., 1998). The rat homologue, CYP2D1 seems correspondingly resistant to cirrhosis in the rat, as based on propranolol hydroxylation (Fenyves et al., 1993). Activity of P450 2A6, as indicated by coumarin 7-hydroxylation in vivo, also appears to be a€ected by liver disease (Sotaniemi et al., 1995), but in vitro studies are lacking. There are few published data on expression of the human fetal form CYP3A7, the substrate speci®city of which remains unclear. This gene is not usually expressed in the mature liver, but it has been suggested that expression may occur in some hepatocellular carcinomas (M.E. McManus and R. McKinnon, personal communication). Preliminary data from the authorÕs laboratory (J. George, C. Liddle ± unpublished results) indicate that some cirrhotic livers express CYP3A7 at the mRNA level, but there are no studies on expression at the protein level. Expression of CYP3A7 could indicate dedi€erentiation of hepatocellular function akin to the reversion to alpha-fetoprotein secretion, and as such could potentially be a premalignant marker, but this requires further study. 6.3. What is the mechanism for altered expression of hepatic CYPs in liver disease? 6.3.1. Changes in protein turnover Various explanations have been proposed for the decreased activity of hepatic drug metabolizing enzymes in patients with liver disease. These include impaired protein synthesis, enzyme inhibition or destruction, and decreased mass of functional hepatocytes. There are some data that infer a nutritional component (George et al., 1996), but none of these factors adequately explains all the above ®ndings. Further, there is no evidence that the cirrhotic rat liver is selectively unable to synthesise P450 proteins (Farrell and Zaluzny, 1984). In rats, bile duct ligation (BDL) induces severe cholestasis with liver cell injury. This is associated with generalised reductions in levels of all P450 proteins and P450reductase, a process that seems to be due to destruction mediated by the detergent action of bile acids because the changes that can be simulated in vitro by incubation of hepatic microsomal fractions with bile acids (Chen and Farrell, 1996). However, there is no other evidence for destruction of P450 enzymes in cirrhosis (Farrell and Zaluzny, 1985), and earlier studies failed to ®nd evidence of reversible enzyme inhibition. On the other hand, decreased mass of functional hepatocytes is pathologically plausible as an explanation for some of the changes in drug metabolizing enzymes observed in chronic liver disease, but it fails to explain why P450 enzymes are more a€ected than phase II enzymes or P450-reductase (Hoyumpa and Schenker, 1991; Debinski et al., 1995). It also fails to account for di€erences in expression levels of individual CYPs. 6.3.2. Changes in mRNA levels The most compelling evidence that nonspeci®c changes in protein synthesis or destruction do not account for altered expression of P450 in liver disease comes from recent studies of CYP expression in human cirrhotic liver explants (George et al.,

J.A. Hasler et al. / Molecular Aspects of Medicine 20 (1999) 1±137

61

Table 6 Correlations between CYP mRNAs and microsomal proteins in cirrhotic human livera

a

CYP

rs

P value

1A2 2C 2E1 3A

0.74 0.36 0.43 0.64