Chapter 6: Biotransformation of Xenobiotics

In 2003, my father (and Oliver's grandfather), Edward William Parkinson, died of Lou Gehrig disease, otherwise known as amyotrophic lateral sclerosis (ALS) in America and as motor neuron disease (MND) in Britain. He died beeping. The beeping came from a portable, electronic syringe that periodically injected my father with glycopyrrolate, an anticholinergic drug that blocks the production of saliva that many ALS patients have difficulty swallowing. Glycopyrrolate is a quaternary ammonium salt; hence, it is positively charged at physiological pH. As such, glycopyrrolate does not readily cross lipid bilayers, which is why it is injected intravenously or intramuscularly; it has few CNS effects (it does not readily cross the blood–brain barrier), and a relatively small volume of distribution at steady state (V_ss = 0.42 L/kg). Unchanged glycopyrrolate is rapidly eliminated in urine. The mean elimination half-life increases from 19 minutes in patients with normal kidney function to 47 minutes in patients with severe kidney impairment, indicating that renal disease impairs the elimination of glycopyrrolate. Although it is excreted in urine largely as unchanged drug, glycopyrrolate reinforces a number of principles about xenobiotic biotransformation, the most important of which is: xenobiotic biotransformation is the process—actually a series of enzyme-catalyzed processes—that alters the physiochemical properties of foreign chemicals (xenobiotics) from those that favor absorption across biological membranes (namely, lipophilicity) to those favoring elimination in urine or bile (namely, hydrophilicity). Without xenobiotic biotransformation, the numerous foreign chemicals to which we are exposed (which includes both man-made and natural chemicals such as drugs, industrial chemicals, pesticides, pollutants, pyrolysis products in cooked food, alkaloids, secondary plant metabolites, and toxins produced by molds, plants, etc) either unintentionally or, in the case of drugs, intentionally would—if they are sufficiently lipophilic to be absorbed from the gastrointestinal tract and other sites of exposure—eventually accumulate to toxic levels. Furthermore, absent xenobiotic biotransformation, many of the drugs in use today would have an unacceptably long duration of action. In contrast, drugs that are not lipophilic, such as glycopyrrolate, are not absorbed from the gastrointestinal tract unless their uptake is mediated by a transporter protein (hence, they are not orally active), and if they are administered parenterally, they are not obligated to undergo biotransformation (because they are already hydrophilic) to facilitate their elimination from the body.

The enzymes that catalyze xenobiotic biotransformation are often called drug-metabolizing enzymes. The acronym ADME stands for absorption, distribution, metabolism, and elimination. This acronym is widely used in the pharmaceutical industry to describe the 4 main processes governing drug disposition. The acronym is sometimes extended to include drug transport (AMDET) or drug toxicity (ADME-Tox). This chapter describes some fundamental principles of xenobiotic biotransformation, and describes the major enzyme systems involved in the biotransformation (or metabolism) of drugs and other xenobiotics. The examples given are biased toward drugs and human enzyme systems for 2 reasons. First, many of the fundamental principles of xenobiotic biotransformation stem from such studies. This is especially true of drugs with a narrow therapeutic index (where the toxic dose is not much greater than the therapeutic dose), which have revealed a large number of genetic and environmental factors that affect xenobiotic biotransformation and, hence, drug toxicity. Second, adverse drug events (ADEs) are one of the leading causes of death in the United States. Lazarou et al. (1998) estimated that in 1994 over 2 million hospitalized patients had serious ADEs (also known as adverse drug reactions or ADRs) and 106,000 had fatal outcomes, placing ADEs only behind heart disease, cancer, stroke, and pulmonary disease as a leading cause of death. Of the 548 new chemical entities approved by the Food and Drug Administration (FDA) between 1974 and 1999, ~10% (56 drugs) required black box warnings or were withdrawn from the market because of ADEs that included hepatotoxicity, cardiotoxicity (several as a result of drug–drug interactions), myelotoxicity, immunotoxicity, and warnings to pregnant women (Lasser et al., 2002; Thomas, 2002). Between 1998 and 2005 the number of serious ADEs reported voluntarily to the FDA through the Adverse Events Reporting System (AERS), known as MedWatch reports, increased 2.6-fold (from 34,966 in 1998 to 89,842 in 2005) and the number of ADE-related fatalities increased 2.7-fold (from 5519 in 1998 to 15,107 in 2005) (Moore et al., 2007). These increases in ADE incidence were 4 times greater than the increase in the total number of outpatient prescriptions written during the same period. In many cases, xenobiotic biotransformation is central to understanding these ADEs.

It is difficult to make categorical statements about xenobiotic biotransformation because there is an exception to every rule. Nevertheless, the following points, which might be considered principles or rules, apply in the majority of cases:

Point 1 Xenobiotic biotransformation or drug metabolism is the process of converting lipophilic (fat-soluble) chemicals, which are readily absorbed from the gastrointestinal tract and other sites, into hydrophilic (water-soluble) chemicals, which are readily excreted in urine or bile. There are exceptions even to this most basic rule. For example, acetylation and methylation are biotransformation reactions that can actually decrease the water solubility of certain xenobiotics. As a general rule, xenobiotics with a log D_7.4 >0 require biotransformation to facilitate their elimination (Williams et al., 2003b).

Point 2 The biotransformation of xenobiotics is catalyzed by various enzyme systems that can be divided into 4 categories based on the reaction they catalyze:

1. Hydrolysis (eg, carboxylesterase)

2. Reduction (eg, carbonyl reductase)

3. Oxidation (eg, cytochrome P450 [CYP])

4. Conjugation (eg, UDP-glucuronosyltransferase [UGT])

The mammalian enzymes involved in the hydrolysis, reduction, oxidation, and conjugation of xenobiotics are listed in Table 6-1, together with their principal subcellular location. The conjugation reactions include glucuronidation, sulfonation (often called sulfation), acetylation, methylation, conjugation with glutathione (GSH; mercapturic acid synthesis), and conjugation with amino acids (such as glycine, taurine, and glutamine). Examples of the major chemical groups that undergo biotransformation together with the enzymes that commonly mediate their biotransformation are given in Table 6-2 (Williams et al., 2003b; Testa and Krämer, 2008, 2010).

Xenobiotic biotransformation is generally catalyzed by enzymes, but there are exceptions. For example, hydrolysis of certain carboxylic and phosphoric acid esters, reduction of sulfoxides to sulfides (eg, rabeprazole), isomerization involving enol–keto tautomerization (eg, thalidomide), and the conjugation of certain xenobiotics with GSH can occur nonenzymatically at an appreciable rate. Certain reactions are catalyzed by gastric acid, such as the hydrolysis of esters and the conversion of indole-3-carbinol to a dimer that is a potent agonist of the aryl hydrocarbon receptor (AhR) and, consequently, an inducer of various enzymes including certain CYP enzymes (CYP1A1, 1A2, 1B1, and 2S1) (Grose and Bjeldanes, 1992).

Point 3 In general, individual xenobiotic-biotransforming enzymes are located in a single organelle. In Table 6-1, some enzymes are listed with 2 or more subcellular locations. However, in such cases, the enzyme name generally refers to 2 or more enzymes, each with its own distinct subcellular location. For example, the epoxide hydrolase located in microsomes is the same enzyme located in the plasma membrane but is a different enzyme from the epoxide hydrolase located in cytosol (ie, they are distinct gene products). From a practical perspective, it is noteworthy that during the homogenization of tissue and the preparation of subcellular fractions, a certain degree of cross-contamination of organelles occurs. For example, microsomes contain detectable levels of monoamine oxidase (MAO) and other outer mitochondrial enzymes, such as the mitochondrial amidoxime-reducing component (mARC) due to their contamination with the outer mitochondrial membrane. On the other hand, some xenobiotic-biotransforming enzymes are present in 2 or more subcellular locations. Some CYP enzymes, such as CYP2E1, are present in microsomes and the inner mitochondrial membrane, which has implications for the site of xenobiotic toxicity, as discussed in the section “CYP2E1.”

Point 4 In general, xenobiotic biotransformation is accomplished by a limited number of enzymes with broad substrate specificities. In humans, for example, 2 CYP enzymes—namely, CYP2D6 and CYP3A4—metabolize over half the orally effective drugs in current use (Gonzalez et al., 2011). The broad and sometimes overlapping substrate specificities of xenobiotic-biotransforming enzymes preclude the possibility of naming the individual enzymes after the reactions they catalyze (which is how most other enzymes are named). Many of the enzymes involved in xenobiotic biotransformation are arranged in families and subfamilies and named according to nomenclature systems based on the primary amino acid sequence of the individual enzymes. CYP2D6, for example, is in CYP family 2, subfamily D, and gene number 6. Some enzymes are given the same name across all mammalian species, whereas others are named in a species-specific manner. For example, some CYP enzymes, such as CYP1A1, CYP1A2, and CYP1B1, are so named in all mammalian species, whereas CYP enzymes in the CYP2, CYP3, and CYP4 families (with certain exceptions such as CYP2E1) are named in a species-specific manner. The convention of using italic and regular letters to distinguish between the gene and gene products (mRNA and protein), respectively, and the convention of using lower case letters to designate mouse genes and gene products will not be followed in this chapter.

The structure (ie, amino acid sequence) of a given xenobiotic-biotransforming enzyme may differ among individuals, which can give rise to differences in rates of drug metabolism. In general, a variant form of a xenobiotic-biotransforming enzyme (known as an allelic variant or allelozyme) has diminished enzymatic activity or level of expression compared with that of the wild-type enzyme, although this is not always the case (see the sections “Alcohol Dehydrogenase,” “Aldehyde Oxidase,” “Cholinesterases (AChE and BChE),” “Carboxylesterases,” “Epoxide Hydrolase,” “Flavin Monooxygenases,” “Glutathione Conjugation,” “Paraoxonases (Lactonases),” “Sulfonation,” “CYP1A1, 1A2, and 1B1,” “CYP1A1, 1A2, and 1B1,” “CYP2C8,” “CYP2C19,” and “CYP3A5”). The impact of amino acid substitution(s) on the catalytic activity of a xenobiotic-biotransforming enzyme may be substrate-dependent, such that an allelic variant may interact normally with some substrates (and inhibitors) but interact atypically with others.

The broad substrate specificity of xenobiotic-biotransforming enzymes makes them catalytically versatile but slow compared with most other enzymes (with the exception of hydrolytic reactions). The sequential oxidation, conjugation, and transport of a xenobiotic tend to proceed quicker at each subsequent step, which prevents the accumulation of intracellular metabolites. Were it not for the low catalytic turnover of CYP (one molecule of which may take several seconds or minutes to oxidize a single drug molecule), it would not be possible to achieve the once-a-day dosing characteristic of a large number of drugs.

Point 5 Hydrolysis, reduction, and oxidation expose or introduce a functional group (such as −OH, −NH₂, −SH, or −COOH) that can be converted to a water-soluble conjugate (see Table 6-2). For subsequent conjugation, the functional group introduced or exposed by hydrolysis, reduction, or oxidation must be nucleophilic (in the case of glucuronidation, sulfonation, methylation, acetylation, and conjugation with glycine or taurine) or electrophilic (in the case of glutathionylation).

The first 3 reactions (hydrolysis, reduction, and oxidation) are often called Phase 1 reactions, and the conjugation reactions are often called Phase 2 reactions. The classification of xenobiotic-biotransforming enzymes into Phase 1 and Phase 2 (and the extension of this system to classify xenobiotic transporters as Phase 3) has been criticized by Josephy et al. (2005), as being both misleading and contrary to the original principles of Williams (1959), the drug metabolism pioneer who coined the phrases to distinguish reactions that resulted in either a decrease or increase in xenobiotic toxicity (Phase 1) from reactions that resulted in only a decrease in toxicity (Phase 2). Notwithstanding the arguments against the Phase 1–Phase 2 classification, which were summarized in the previous version of this chapter (Parkinson and Ogilvie, 2008), the terms are widely used today not as R. T. William's originally intended but to distinguish biotransformation reactions involving oxidation, reduction, or hydrolysis (Phase 1) from those involving conjugation (Phase 2). The original idea that Phase 2 metabolism results in only detoxication is incorrect. Indeed, on a case-by-case basis, all xenobiotic-metabolizing enzymes are capable of increasing the toxicity of one or more xenobiotics, including the conjugating enzymes that R. T. Williams classified as Phase 2 (detoxifying) enzymes. For example, conjugation of carboxylic acid–containing drugs with glucuronic acid or coenzyme A (CoA) to form acyl glucuronides and acyl-CoA thioesters, respectively, is thought to be responsible for the hepatotoxicity observed with numerous nonsteroidal anti-inflammatory drugs (NSAIDs). In fact, in the United States, NSAIDs and other carboxylic acid–containing drugs represent the largest category of drugs whose clinical development was halted or whose regulatory approval was subsequently withdrawn because of liver toxicity in humans (Fung et al., 2001). Although Phase 1 biotransformation often precedes Phase 2 metabolism, some drugs, such as acetaminophen (Tylenol), are conjugated directly (by glucuronidation and sulfonation), whereas other drugs, such as gemfibrozil and diclofenac, undergo Phase 1 metabolism (by CYP [CYP2C8]) after Phase 2 metabolism (glucuronidation) (Ogilvie et al., 2006).

Point 6 Oxidation, reduction, hydrolysis, methylation, and acetylation generally cause a modest increase in the water solubility of a xenobiotic, whereas glucuronidation, sulfonation, glutathionylation, and amino acid conjugation generally cause a marked increase in hydrophilicity. The impact of biotransformation on the physiochemical properties and disposition of xenobiotics and their metabolites can be appreciated from modern approaches to drug development. For a drug to be absorbed by passive diffusion in the gut (so that it can be orally effective and distribute throughout the body) it must generally have a log P of 0 to 5 and a calculated total polar surface area (TPSA) less than 100 Ų (Smith and Dalvie, 2012). Water-soluble drugs (those with a negative log P) might be orally active if their absorption is mediated by an intestinal drug transporter. Extremely lipophilic drugs (log P > 5) tend to be too insoluble or they dissolve too slowly to be well absorbed from the intestine unless the dose is particularly low. Drugs must undergo desolvation (ie, they must dissociate from water molecules) to penetrate the lipid bilayer of cells. Drugs begin to lose their lipoidal permeability when their TPSA increases above ~75 Ų and lose it entirely at a TPSA of around 150 Ų. When drugs meet the aforementioned criteria (log P = 0−5, TPSA < 100 Ų), their metabolites formed by oxidation, reduction, hydrolysis, methylation, or acetylation also tend to meet these same criteria; hence, the metabolites formed by these reactions, like the parent drug, readily diffuse across biological membranes down a concentration gradient (unless their diffusion into tissues such as the brain is restricted by efflux transporters). Metabolites formed by conjugation reactions (other than methylation and acetylation) contain a negatively charged glucuronide, sulfate, or amino acid (glycine or taurine) moiety (or a zwitterionic GSH moiety) and consequently have a much lower log P and a much greater TPSA than the parent drug, so much so that they cannot readily diffuse passively across biological membranes. The elimination of such conjugates from the liver (where they are largely formed) requires active transport across the canalicular (apical) membrane so they can be eliminated in bile or across the sinusoidal (basolateral) membrane, which enables their transport into blood (where they are generally highly bound to albumin given their anionic nature [negative charge]) and then to the kidney for elimination in urine (by either glomerular filtration or possibly active transport [tubular secretion]). Drug conjugates are actively transported into bile primarily by MRP2 (ABCC2) on the canalicular membrane and actively transported into blood primarily by MRP1 (ABCC1; in the case of GSH conjugates) and MRP3 (ABCC3; in the case of glucuronide and sulfate conjugates) on the sinusoidal membrane, although on a case-by-case basis other transporters may be involved (Borst et al., 2006).

Point 7 Not all biotransformation reactions are catalyzed by the mammalian enzymes listed in Table 6-1. Some biotransformation reactions are catalyzed by enzymes in the gut microflora (largely anaerobic bacteria in the colon), whereas the biotransformation of still other xenobiotics is catalyzed by enzymes that participate in intermediary (endobiotic) metabolism. This principle is illustrated in Fig. 6-1 for 3 xenobiotics that are all converted to benzoic acid and then hippuric acid. The conversion of toluene to hippuric acid is catalyzed by some of the xenobiotic-biotransforming enzymes listed in Table 6-1: microsomal CYP converts toluene to benzoic acid in 3 oxidative steps (R-CH₃ → R-CH₂OH → RCHO → RCOOH), which introduces a functional group (namely, –COOH) that is conjugated with glycine to produce hippuric acid, which is excreted in urine. Cinnamic acid is also converted to benzoic acid but, in this case, the reaction is catalyzed by the mitochondrial enzymes involved in the β-oxidation of fatty acids. Quinic acid is also converted to benzoic acid, but this reductive, multistep reaction is catalyzed by gut microflora. Incidentally, the conversion of benzoic acid to hippuric acid is of historical interest because it is generally recognized as the first xenobiotic biotransformation reaction to be discovered (in dogs by Woehler in 1828, and in humans by Ure in 1841).

Some drugs are intentionally designed to be biotransformed by endobiotic-metabolizing enzymes. For example, the anti-HIV drug zidovudine (AZT) is converted to a triphosphate nucleoside by enzymes in the salvage pathway (nucleoside kinase, nucleoside monophosphate kinase [NMK], and nucleoside diphosphate kinase [NDK]). HIV reverse transcriptase inhibitors, such as zidovudine, are administered as nonphosphorylated analogs to facilitate their absorption and cellular uptake. Conversely, some drugs are administered as a phosphorylated prodrug (eg, fosamprenavir) to promote their water solubility and rate of dissolution. The luminal surface of the small intestine contains high levels of alkaline phosphatase, which hydrolyzes prodrugs such as fosamprenavir and thereby releases the active drug at the surface of the enterocyte, where it is readily absorbed. Generally speaking, kinase and alkaline phosphatase are not usually considered to be xenobiotic-biotransforming enzymes.

Point 8 Just as some xenobiotics are biotransformed by the so-called endobiotic-metabolizing enzymes (Point 7), certain endobiotics are biotransformed by the so-called xenobiotic-metabolizing enzymes. For example, the same CYP enzymes implicated in xenobiotic biotransformation also contribute to the hepatic catabolism of steroid hormones, and the same UGTs that conjugate xenobiotics also glucuronidate bilirubin, thyroid hormones, and steroid hormones. Interestingly, the CYP and UGT enzymes that play a major role in steroid metabolism (namely, CYP3A4 and UGT1A1) display positive homotropic cooperativity (ie, the enzymes have Hill coefficients >1) such that the rate of metabolism of steroid hormones by these enzymes is slow at low substrate concentrations but increases dramatically as the concentration of steroid hormone increases. Benzoic acid, a xenobiotic, is conjugated with glycine (as shown in Fig. 6-1), and so are bile acids (endobiotics). Certain leukotrienes are GSH conjugates. Conjugates of endogenous substrates, such as the glucuronide and sulfate conjugates of steroid and THs, tend to be transported out of the hepatocyte by MRP2, the same canalicular transporter that transports drug conjugates into bile. From the few examples in Points 7 and 8, it is apparent that, on a case-by-case basis, there is often no clear-cut distinction between endobiotic- and xenobiotic-biotransforming enzymes.

The human genome project has helped to establish that what were once thought to be 2 distinct enzymes, one involved in the metabolism of endobiotics and one involved in the metabolism of xenobiotics, are in fact one and the same enzyme. For example, the microsomal enzyme known as 11β-hydroxysteroid dehydrogenase is identical to the xenobiotic-metabolizing enzyme known as microsomal carbonyl reductase (gene symbol HSD11B1; aka SDR26C1). Likewise, hepoxilin A₃ hydrolase was thought to represent a fifth class of epoxide hydrolases, but was subsequently identified as the same enzyme as soluble epoxide hydrolase (sEH) (Cronin et al., 2011).

Point 9 Several xenobiotic-biotransforming enzymes are inducible, meaning their expression can be increased (upregulated) usually in response to exposure to high concentrations of xenobiotics. Induction is mediated by ligand-activated receptors (so-called xenosensors) that are activated by xenobiotics (ligands) to DNA-binding proteins that upregulate the transcription of various genes encoding xenobiotic-biotransforming enzymes, especially CYP enzymes, which are usually induced to the greatest extent (in terms of fold increase). The major xenosensors are aryl hydrocarbon receptor (AhR), which induces CYP1 enzymes, the constitutive androstane receptor (CAR), which induces CYP2B, 2C, and 3A enzymes, the pregnane X receptor (PXR), which also induces CYP2B, 2C, and 3A enzymes, and the peroxisome proliferator–activated receptor-α (PPARα), which induces CYP4 enzymes. CAR and PXR are closely related nuclear receptors, and they tend to be activated by the same ligands and bind to the same DNA-response elements (ie, the same discrete regions of DNA that control gene expression), for which reason there is said to be considerable cross-talk between these 2 xenosensors. Certain xenosensors are activated by endogenous ligands (eg, bilirubin, bile acids, and fatty acids activate CAR, PXR, and PPARα, respectively), and certain nuclear receptors, such as the vitamin D receptor (VDR), can mimic PXR and induce CYP3A4, which inactivates the active metabolite of vitamin D. These examples illustrate how xenosensors are not just involved in xenobiotic disposition but also play a role in endobiotic homeostasis. Two proteins called Kelch-like ECH-related protein (KEAP)-1 and nuclear factor E2 p45-related factor-2 (Nrf2) function as xenosensors inasmuch as they induce enzymes in response to oxidative stress, which is often associated with xenobiotic biotransformation. Binding of the transcription factor Nrf2 to the region of DNA known as the antioxidant response element (also known as the electrophilic response element [EpRE]) induces enzymes that detoxify electrophiles and metabolites that generate reactive oxygen species (ROS), including glutathione transferase (GST; GSTA1), microsomal epoxide hydrolase (mEH), aldo-keto reductase (AKRs; AKR7A, also known as aflatoxin aldehyde reductase), NAD(P)H-quinone oxidoreductase (NQO1, also known as DT-diaphorase), microsomal heme oxygenase, and glutamate–cysteine ligase (GCL), which catalyzes the rate-limiting step in GSH synthesis. Induction of GCL increases the rate of GSH synthesis under conditions of oxidative stress and a decrease in GSH concentration (Lee and Johnson, 2004).

Induction is a reversible, adaptive response to xenobiotic exposure. The induced enzymes (and transporters) usually accelerate the elimination of the xenobiotic that triggered the induction process, in which case the xenobiotic is said to be an autoinducer (one that induces its own metabolism). However, xenobiotics often induce enzymes that are not capable of metabolizing them, in which case the induction is said to be gratuitous. Induction is a pleiotropic response: activation of AhR, CAR, PXR, PPARα, and Nrf2 all results in alterations in the expression of numerous genes.

Suppression (downregulation) of drug-metabolizing enzymes is often association with inflammatory diseases (such as arthritis), cancer, infectious diseases (both bacterial and viral), vaccination, and treatment with certain proinflammatory biologics (therapeutic proteins). These disease processes activate nuclear factor kappa-B (NF-κB) (and other nuclear receptors), which suppresses the expression and induction of CYP and other xenobiotic-metabolizing enzymes because activated NF-κB suppresses all 4 xenosensors (AhR, CAR, PXR, and PPARα), as well as several other nuclear receptors). By reversing the disease process—such as lessening the inflammation associated with rheumatoid arthritis—some biologics (large drug molecules such as monoclonal antibodies and other types of therapeutic proteins) can reverse the suppression of drug-metabolizing enzymes and restore their activity to normal (predisease) levels (Morgan, 2009; Huang et al., 2010).

Point 10 The ability of certain xenobiotic-biotransforming enzymes to metabolize hormones and other endobiotics (Point 8) and the ability of certain xenobiotics to induce xenobiotic-biotransforming enzymes (Point 9) have implications for understanding an important mechanism by which certain xenobiotics can alter homeostasis or cause toxicity. Persistent exposure to xenobiotics that are enzyme inducers (especially CAR and PPARα activators) can increase the rate of steroid hormone oxidation by CYP and increase the rate of thyroid hormone glucuronidation and sulfonation that, in rodents, can lead to the development of Leydig cell tumors (due to elevated levels of LH and FSH) and thyroid follicular tumors (due to elevated levels of thyroid-stimulating hormone [TSH]), respectively (Grasso et al., 1991). Persistent exposure to enzyme inducers can also cause liver tumors, although the mechanism is not fully understood. Activation of certain xenosensors (CAR or PPARα) is critical to liver tumor development in mice, although upregulation of xenobiotic-biotransforming enzymes is less important than other xenosensor-dependent events, such as, in the case of the CAR activator phenobarbital, altered DNA methylation of numerous genes, upregulation of GADD45B (an antiapoptotic factor) and MDM2 (a negative regulator of the tumor suppressor p53), and the downregulation of gap junctional proteins (connexin 32), which diminishes cell–cell communication (Omiecinski et al., 2011b).

Phenobarbital, Wy-14643, methapyrilene, and Ponceau S are representatives of 4 classes of nongenotoxic rodent tumorigens (epigenetic tumor promoters) that cause hepatocellular hyperplasia and hypertrophy in association with proliferation of the endoplasmic reticulum, peroxisomes, mitochondria, and lysosomes, respectively (Grasso et al., 1991). In the first 2 cases, the hepatocellular changes depend on activation of CAR (in the case of phenobarbital) and PPARα (in the case of Wy-14643). Prolonged activation of these receptors in rodents results in the development of liver and/or thyroid tumors. However, this is thought to be a rodent-specific phenomenon because, in the case of phenobarbital and other enzyme-inducing antiepileptic drugs (EIAEDs), there is compelling epidemiological evidence that these chemicals do not cause liver tumors in humans even after >30 years of treatment (Parkinson et al., 2006).

The major UGT responsible for conjugating bilirubin is UGT1A1, which is inducible by CAR activators such as phenobarbital and scoparone (6,7-dimethylesculetin). At one time, the management of neonatal jaundice included treatment with phenobarbital to induce bilirubin conjugation, but this practice has been discontinued. However, the Chinese herbal Yin Zhi Huang (active ingredient scoparone) is still used to treat neonatal jaundice. Bilirubin is also a CAR agonist, and high levels of bilirubin induce UGT1A1 and bilirubin/bilirubin glucuronide transporters (MRP2 [ABCC2] and MRP3 [ABCC3]) to increase the rate of bilirubin elimination when bilirubin levels are elevated (eg, during hemolytic anemia).

Although prolonged activation of PXR in rats is associated with thyroid tumor promotion due to induction of triiodothyronine (T3) glucuronidation (Richardson and Klaassen, 2010), this effect is thought not to occur in humans. However, prolonged activation of PXR by rifampin in humans (which induces hepatic and intestinal CYP3A4) is associated with osteomalacia caused by an enhanced rate of conversion of the active metabolite of vitamin D to inactive metabolites by CYP3A4 (Xu et al., 2006; Zhou et al., 2006a).

Point 11 Xenobiotic biotransformation can alter the biological properties of a xenobiotic. It can make the xenobiotic less toxic (detoxication), but in some cases it can make it more toxic (activation). The oxidation of ethanol (alcohol) to acetaldehyde is an example of xenobiotic activation, and the subsequent oxidation of acetaldehyde to acetic acid is an example of detoxication. The biotransformation of drugs can result in (1) a loss of pharmacological activity (eg, the conversion of acetaminophen to acetaminophen glucuronide and the conversion of morphine to morphine-3-glucuronide), (2) no change in pharmacological activity (eg, the conversion of fluoxetine to its N-demethylated metabolite norfluoxetine), or (3) an increase in pharmacological activity (eg, the conversion of codeine to morphine, as well as the conversion of morphine to morphine-6-glucuronide). When the parent drug is pharmacologically active, diminished metabolism (due to a genetic deficiency or enzyme inhibition) can result in an exaggerated pharmacological response (Type A toxicity), such as the prolonged or increased hypotension caused by debrisoquine, alprenolol, bufuralol, metropolol, propranolol, and timolol in individuals lacking CYP2D6 (Turner et al., 2001; Ingelman-Sundberg, 2005). When a metabolite is the pharmacologically active component, diminished formation (due to a genetic deficiency or enzyme inhibition) can result in little or no pharmacological response (Type F toxicity), such as the diminished effects of codeine and tamoxifen in individuals lacking CYP2D6 and the diminished effects of clopidogrel in individuals lacking CYP2C19 (discussed in the sections “CYP2D6” and “CYP2C19”).

Point 12 In many cases, the toxicity of a xenobiotic is due to the parent compound (the compound that was absorbed), in which case xenobiotic biotransformation serves as a detoxication mechanism. This principle is underscored by the clinical observation that the incidence of ADEs is often higher in individuals with a poor metabolizer (PM) phenotype (discussed later in Point 24). However, xenobiotic-biotransforming enzymes can convert certain xenobiotics to reactive metabolites (eg, electrophiles and radicals), and this activation process plays an important role in chemical toxicity and chemical mutagenicity/carcinogenicity; the former often involves the covalent binding of reactive metabolites to critical cellular nucleophiles such as proteins, whereas the latter involves covalent binding to one or more purine or pyrimidine bases in DNA. Formation of reactive metabolites can also lead to irreversible CYP inhibition (a process known as metabolism-dependent inhibition [MDI] or time-dependent inhibition [TDI]), which is an important cause of drug–drug interactions. As illustrated in Fig. 6-2, carboxylic acid–containing drugs (especially the acetic and propionic acid–containing NSAIDs that cause liver toxicity) can be converted to reactive acyl glucuronides and/or acyl-CoA thioesters, whereas many neutral and basic drugs are converted by CYP to electrophilic metabolites or carbon-centered radicals that may cause liver toxicity (or damage to the cells harboring the activating CYP). Several nitrogen- and sulfur-containing drugs can be activated by peroxidases to heteroatom-centered radicals that can cause blood dyscrasias such as agranulocytosis/neutroprenia, aplastic anemia, and thrombocytopenia (Tang and Lu, 2010; Walsh and Miwa, 2011). In Fig. 6-2, the scheme depicting the formation and toxicity of reactive metabolites is an oversimplification. Furthermore, some drugs can be converted to 2 or more reactive metabolites. Tienilic acid, for example, contains a carboxylic acid group, which is converted to a reactive acyl glucuronide and acyl-CoA thioester, and a furan (a neutral group), which is converted by CYP to an epoxide and other reactive metabolites (such as an enedial).

CYP is particularly effective at converting proximate carcinogens to ultimate carcinogens by converting the former to electrophilic metabolites that bind to and mutate DNA, thereby leading to mutations and tumor initiation. The Ames bacterial mutagenicity assay, which is representative of several mutagenicity tests, is performed in the absence and presence of CYP (ie, liver microsomes) to evaluate the mutagenic potential of both the parent compound and any electrophilic metabolites, respectively. Polycyclic aromatic hydrocarbons (PAHs; combustion pollutants and components of tobacco smoke), aromatic amines (industrial chemicals), aflatoxin (a mycotoxin), cooked food pyrolysis products, and tobacco-specific nitrosamines are all examples of proximate carcinogens that require metabolic activation to form electrophilic, DNA-binding metabolites (ultimate carcinogens).

Certain anticancer drugs require activation by xenobiotic-metabolizing enzymes in order to exert their antineoplastic effects. For example, cyclophosphamide is activated by CYP (CYP2B6 and CYP3A4), and canfosfamide is activated by GST, namely, GSTP1-1, which is often overexpressed in tumors thereby conferring resistance against certain anticancer drugs (see the section “Glutathione Conjugation”) (Gonzalez and Tukey, 2006).

Liver microsomes are widely used in 3 in vitro systems to investigate the potential of drugs or drug candidates to be converted to reactive metabolites. First, liver microsomes from Aroclor 1254–treated (induced) rats are used in the Ames bacterial mutagenicity assay. A limitation of this assay is that it is well suited to identify reactive metabolites formed by CYP but not by other xenobiotic-biotransforming enzymes, examples of which are given in the section “Conjugation.” This limitation has been addressed through genetic engineering of Salmonella typhimurium so that test bacteria express one or more human xenobiotic-metabolizing enzymes (Iwata et al., 1998; Glatt and Meinl, 2004b). Second, human and rat liver microsomes are widely used to assess the formation of reactive metabolites that bind covalently to microsomal protein or GSH (or a related nucleophile such as N-acetylcysteine [NAC]). This system does a poor job of distinguishing hepatotoxic drugs from nonhepatotoxic drugs (Obach et al., 2008), but the distinction is improved when covalent binding is assessed in human hepatocytes and dose is taken into account (Bauman et al., 2009; Nakayama et al., 2009). Third, human liver microsomes are widely used to evaluate a drug candidate's potential to cause irreversible inhibition of the major drug-metabolizing CYP enzymes. False-negative results can occasionally occur when metabolism by a non-CYP enzyme is instrumental to CYP inactivation, such as the need to glucuronidate gemfibrozil in order to inactivate CYP2C8. In such cases of so-called system-dependent inhibition the use of human hepatocytes offers advantages over human liver microsomes (Parkinson et al., 2010).

Structural alerts (functional groups) for the bioactivation of xenobiotics to reactive metabolites that cause toxicity and/or CYP inactivation include the following (which is not an exhaustive list): (1) acetic and propionic acids (activated by UGT and/or acyl-CoA synthetase), (2) nitro aromatics (activated by reductases and/or CYP), and (3) a large number of functional groups that are activated by CYP and, in some cases, peroxidases: acetylenes, activated aromatics, 3-alkly indoles, anilines, cyclopropylamines, formamides, furans, hydrazines, pyrroles, sulfonylureas, thiazolidinones, thioamides, thiophenes, and thioureas (Walsh and Miwa, 2011).

Point 13 The toxicity and potential carcinogenicity of electrophilic metabolites produced by CYP and other xenobiotic-biotransforming enzymes is reduced and often altogether eliminated by their conjugation with GSH, which is often described as a noncritical cellular nucleophile. Conjugation with GSH can occur both enzymatically (by GST) and nonenzymatically. In the majority of cases, conjugation with GSH represents a detoxication reaction, one that protects critical cellular nucleophiles, such as protein and DNA, from covalent modification. However, there are cases (eg, methylbromide and dibromoethane) where conjugation with GSH actually produces a DNA-reactive (mutagenic) metabolite. Activation of haloalkanes and dihaloalkanes (which are used industrially as fumigants) is usually catalyzed by the Theta class of GSTs (Josephy, 2010). Depletion of GSH or genetic deficiency of a particular GST can predispose to certain drug toxicities. For example, the genetic loss of glutathione transferase-Mu (GSTM1) predisposes to the hepatotoxic effects of carbamazepine and valproic acid (Saruwatari et al., 2010), whereas the combined loss of GSTM1 and glutathione transferase-Theta (GSTT1) is a risk factor for troglitazone hepatotoxicity (Ikeda, 2011).

Point 14 The biotransformation of some xenobiotics results in the production of ROS, which can cause cell toxicity (including DNA damage) through oxidative stress and lipid peroxidation. GSH, GSTs, and glutathione peroxidases (GPXs) all limit the toxic effects of ROS just as they limit the toxicity of reactive metabolites formed directly from xenobiotics. Oxidative stress and the formation of electrophilic metabolites reduce GSH levels and thus result in the concurrent oxidation of KEAP-1, which then releases Nrf2, which in turn upregulates the expression of enzymes that detoxify electrophilic metabolites (eg, epoxides) and those metabolites that generate ROS (eg, quinones) (see Point 9). Certain xenobiotics and ischemic reperfusion injury cause an accumulation of unfolded proteins in the endoplasmic reticulum (where proteins are folded and posttranslationally modified), which gives rise to the unfolded protein response (UPR; also known as endoplasmic reticulum stress [ER stress]), which activates Nrf2 and Bax-inhibitor-1 (BI-1). The heme oxygenase induced by Nrf2 as well as BI-1 (which complexes with NADPH-cytochrome P450 reductase) protects cells from oxidative stress by inhibiting the production of ROS by CYP2E1 and other CYP enzymes (discussed in the section “CYP2E1”).

Point 15 The balance between activation and detoxication by xenobiotic-biotransforming enzymes is often a key determinant of chemical toxicity, and is often the basis for organ or species differences in toxicity. For example, aflatoxin is converted by liver microsomal CYP to a reactive epoxide that is thought to be responsible for the hepatotoxic and hepatocarcinogenic effect of this mycotoxin. The fact that this reaction occurs in the liver explains why aflatoxin causes liver toxicity and liver tumors. On this basis, mice would be expected to be more sensitive than rats to the hepatotoxic effects of aflatoxin because mice catalyze the epoxidation of aflatoxin faster than rats. However, through GSH conjugation, mice also detoxify aflatoxin epoxide faster than rats. Consequently, despite their slower rate of activation, rats are more susceptible than mice to the toxic effects of aflatoxin. Coumarin is hepatotoxic to rats, but not to humans. This species difference is attributable to differences in the metabolites formed by CYP. Rats convert coumarin to a reactive epoxide that rearranges to a reactive aldehyde, whereas humans convert coumarin to the relatively nontoxic metabolite 7-hydroxycoumarin (umbelliferone).

Point 16 Exposure to xenobiotics (especially drugs) is largely through oral ingestion, and the small intestine and liver are highly developed to limit systemic exposure to orally ingested xenobiotics, a process known as first-pass elimination (or presystemic elimination). The enterocytes at the tips of the small intestinal villi express the efflux transporters P-glycoprotein (ABCB1 or MDR1) and BCRP (ABCG2), which serves to limit xenobiotic absorption. The enterocytes express high levels of certain CYP and UGT enzymes, which biotransform a wide variety of xenobiotics. The liver expresses a number of uptake transporters that actively remove xenobiotics from the blood. They also express a number of efflux transporters that actively discharge xenobiotics or their metabolites (especially conjugates) into the bile canaliculus for biliary excretion, or that actively discharge xenobiotic metabolites (especially conjugates) across the sinusoidal membrane back into the blood for urinary excretion. The liver expresses the largest number and, with few exceptions, the highest concentrations of xenobiotic-biotransforming enzymes.

Although the liver contains higher concentrations of most xenobiotic-biotransforming enzymes, and because the number of hepatocytes in the liver exceeds the number of enterocytes in the small intestine, it might be assumed that, compared with the liver, the small intestine would make only a small contribution to first-pass metabolism, but this is not the case in part because in the small intestine xenobiotics are recycled by P-glycoprotein (ABCB1) and BCRP (ABCG2) on the lumenal surface of enterocytes, which delays absorption and increases the opportunity for xenobiotic biotransformation by intestinal enzymes such as CYP (especially CYP3A4), UGT, and sulfotransferase (SULT) (Benet, 2009). Furanocoumarins in grapefruit juice inhibit intestinal but not hepatic CYP3A4 and yet grapefruit juice increases systemic exposure to felodipine and other drugs that undergo first-pass metabolism by CYP3A4 (Paine et al., 2006). The impact of the CYP3A4 inhibitor ketoconazole on the disposition of the CYP3A4 substrate midazolam depends on whether midazolam is given orally or intravenously. When midazolam is given intravenously, such that its clearance is dependent only on hepatic metabolism, ketoconazole causes a 3- to 5-fold increase in the area-under-the-plasma-concentration-time curve (AUC). However, when midazolam is given orally, ketoconazole causes a 10- to 15-fold increase in AUC, the difference reflecting the significant role of intestinal metabolism to the presystemic clearance of midazolam. In rats, administration of the general CYP inhibitor 1-aminobenzotriazole (ABT) by the intravenous route (which inhibits only hepatic CYP enzymes) and the oral route (which inhibits both intestinal and hepatic CYP enzymes) can be used to assess the relative role of the intestine and liver to the first-pass metabolism of a drug or other xenobiotic (Strelevitz et al., 2006).

The small intestine and liver are exposed to high concentrations of xenobiotics, and they possess high levels of the enzymes that potentially convert xenobiotics to reactive or toxic metabolites. It is perhaps not surprising, therefore, that both tissues possess protective mechanisms to minimize the risk of xenobiotic toxicity and carcinogenicity. As already mentioned, both tissues have enzymes and transporters that facilitate the elimination of xenobiotics and their metabolites. In both tissues, several of the xenobiotic-biotransforming enzymes and transporters are inducible, enabling the liver and the small intestine to respond to high levels of xenobiotics by enhancing the rate of xenobiotic biotransformation and elimination. In the small intestine, the enterocytes at the villus tips undergo extensive turnover, such that the mature cells that are exposed to high levels of xenobiotics and/or reactive metabolites are quickly lost (exfoliated) and replaced in a matter of days. In liver, high levels of GSH (5-10 mM), a large proportion of diploid (binucleated) cells, and a high regenerative capacity all protect the liver from xenobiotic toxicity or help the liver to repair chemical-mediated toxicity. In addition, severely damaged hepatocytes can undergo apoptosis (cell-programmed death) to eliminate precancerous cells (ie, cells with extensive DNA damage).

Point 17 Some of the same mechanisms that protect the small intestine and liver from xenobiotic toxicity also protect certain organs such as the brain and reproductive organs. Xenobiotics that enter the systemic circulation are often excluded from the brain by the blood–brain barrier, which poses a physical barrier to xenobiotic transport (in the form of tight junctions) and a biochemical barrier notably in the form of the efflux transporters P-glycoprotein (ABCB1) and BCRP (ABCG2). Germ cells in the testis are protected in part by a blood–testis barrier, by high levels of GSH (and GST and GPX), and by high cell turnover (as in the case of enterocytes). The ovum, with virtually no cell turnover, possesses high levels of GSH (8 mM or more), which protects it from DNA-reactive electrophiles and from the oxidative stress that accompanies both xenobiotic exposure and fertilization. Efflux transporters and GSTs are often overexpressed in tumor cells as a result of chromosomal rearrangements that place the genes encoding these proteins under the control of a strong promoter. The overexpression of a transporter and/or GST can confer resistance to certain cancer chemotherapeutic agents (many of which are intended to be converted to reactive metabolites that damage DNA and kill tumor cells).

Point 18 In view of the important role of CYP in the metabolic activation of proximate carcinogens to ultimate carcinogens, it may seem paradoxical to list CYP induction among the defense mechanisms that protect organisms from the carcinogenic effects of xenobiotics. Activation by CYP is definitely required for certain xenobiotics to exert their carcinogenic effects, and induction of CYP is associated with an increase in the toxicity of certain xenobiotics. However, contrary to expectation, treatment of rodents with a CYP inducer prior to treatment with a known proximate carcinogen (such as aflatoxin, various nitrosamines, or PAHs) is generally associated with a decrease, not an increase, in tumor incidence (Parkinson and Hurwitz, 1991). The route of exposure to the carcinogen can affect the impact of enzyme induction; it protects against orally administered drugs but may increase the tumorigenicity of carcinogens applied directly to their site of action (Nebert et al., 2004), and prolonged treatment of rodents with an enzyme inducer after application of the initiating carcinogen can lead to an increase in tumor incidence because many enzyme inducers act as tumor promoters (see Point 10). Nevertheless, for the most part, enzyme induction provides protection against chemical toxicity and carcinogenesis (Nebert and Dalton, 2006; Shimada, 2006; Ma and Lu, 2007).

Point 19 Although the small intestine and liver contain the highest concentrations, xenobiotic-biotransforming enzymes are nevertheless widely distributed throughout the body. In terms of specific content (ie, the amount of enzyme on a per-milligram-protein basis), some of the highest concentrations of xenobiotic-metabolizing enzymes are to be found in nasal epithelium, a portal of entry for many volatile xenobiotics. The presence of high levels of xenobiotic-biotransforming enzymes in nasal epithelium is of interest because numerous xenobiotics cause nasal cytotoxicity and cancer in rodents (reviewed in Jeffrey et al., 2006). Xenobiotic-biotransforming enzymes in the lung, eye, and skin can be especially important for the metabolism of drugs delivered by inhalation, eye drop, or topical application, respectively. The kidney expresses several xenobiotic-biotransforming enzymes, in addition to numerous transporters that actively secrete xenobiotics (especially acidic metabolites) into urine. The kidney also metabolizes GSH conjugates formed in the liver and converts them to mercapturic acids (N-acetylcysteine conjugates).

Point 20 Species differences in xenobiotic-biotransforming enzymes are often the basis for species differences in both the qualitative and quantitative aspects of xenobiotic biotransformation and toxicity. As mentioned in Point 15, species difference in coumarin metabolism and toxicity (activation by epoxidation in rats, detoxication by aromatic hydroxylation in humans) reflects a species difference in hepatic microsomal CYP, whereas species differences in GST activity account for the difference between rats and mice in terms of their susceptibility to aflatoxin-induced liver toxicity. Species differences in the fetal expression of xenobiotic-biotransforming enzymes have raised questions about the suitability of laboratory animals to assess the risk of human teratogenicity because human fetal liver expresses CYP3A7, an enzyme capable of activating aflatoxin to reactive metabolites, whereas no such enzyme is expressed in the fetus of rodents (Li et al., 1997; Pang et al., 2012). Biotransformation, inhibition, and induction can occur in a species-specific manner. For example, furafylline and omeprazole are inhibitors and inducers of human CYP1A2, respectively, but they do not inhibit or induce CYP1A2 in rodents. Such species differences are the impetus for the development of chimeric mice (in which the mouse hepatocytes are replaced with human hepatocytes) (Strom et al., 2010) and so-called humanized mice, which involves substituting the murine genes that encode xenobiotic-metabolizing enzymes (and the receptors that regulate their expression) with their human counterparts (Gonzalez, 2003; Gonzalez and Yu, 2006).

In some species, entire pathways of xenobiotic biotransformation are low or absent from a particular species. For example, cats and dogs are considered poor glucuronidators and poor acetylators of xenobiotics, respectively. Because of their inability to glucuronidate drugs, cats are sensitive to acetaminophen hepatotoxicity (Baillie and Rettie, 2011). Dogs are poor acid secretors, which disfavors the gastric absorption of acidic drugs. Interestingly, although dogs can conjugate acidic drugs, they are poor metabolizers of acidic drugs in terms of oxidative biotransformation by CYP enzymes (Baillie and Rettie, 2011). Metabolism by gut microflora can occur in a species-dependent manner. For example, as shown in Fig. 6-1, quinic acid is converted to benzoic acid in humans and old-world monkeys but not in new-world monkeys, mice, rats, and most other laboratory animals (Adamson et al., 1970).

Point 21 In sexually mature rats and, to a lesser extent, mice there are marked gender differences in the expression of certain xenobiotic-biotransforming enzymes (both oxidative and conjugating enzymes) (Waxman and Holloway, 2009). In other species, including humans, gender differences either do not exist or generally represent less than a 2-fold difference, whereas the differences in rodents can be an order of magnitude or greater (Parkinson et al., 2004; Soldin et al., 2011). Against a large background of interindividual variation in each gender, it is often difficult to discern gender differences in xenobiotic biotransformation in humans, although, in general, women appear to have lower CYP1A2 and higher CYP3A4 activity than men (Parkinson et al., 2004). It is particularly interesting that, compared with men, women appear to have a higher incidence of idiosyncratic drug toxicity (Walgren et al., 2005), which has been the cause of a number of drug withdrawals and black box warnings (see Point 22).

Point 22 Idiosyncratic drug reactions (IDRs) are rare adverse events (generally <0.1%) that do not involve an exaggerated pharmacological response, do not occur in most patients at any dose, and typically do not occur immediately after exposure but do so after weeks or months of repeated administration (Walgren et al., 2005). IDRs are also known as Type B or hypersensitivity reactions and they can be divided into allergic and nonallergic IDRs. The former tend to develop relatively quickly (in days or weeks) and, after the drug is discontinued, patients respond robustly when rechallenged with the same or a closely related drug, whereas the latter tend to develop relatively slowly (with symptoms sometimes appearing after 6 months or more of drug treatment) and patients may or may not respond to rechallenge with the drug. Evidence for an immune component (ie, an acquired immune response) to allergic IDRs is often circumstantial or lacking, as is the evidence for the lack of an immune component to nonallergic IDRs, which possibly involve an innate immune response. Hepatotoxicity is a prevalent IDR. Drugs that were withdrawn from the US market or carry a black box warning for idiosyncratic hepatotoxicity are listed in Table 6-3. The drugs are structurally diverse and fall into numerous therapeutic classes.

Several factors predispose to IDRs including gender (women are more prone than men), age (the elderly are more prone), ethnicity, viral infection (a particularly strong predisposing factor), and other disease states (such as diabetes and obesity), possibly alcohol consumption, comedications that cause pharmacokinetic drug–drug interactions, genetic polymorphisms in xenobiotic-biotransforming enzymes and transporters (discussed in Point 23), and genetic polymorphisms that affect immune function especially allelic variants of human leukocyte antigens (HLA haplotypes), part of the major histocompatibility complex (MHC) (Alfirevic and Pirmohamed, 2010; Chalasani and Björnsson, 2010).

The best understood IDR is the skin hypersensitivity reaction to the anti-HIV drug abacavir, which is determined in a highly predictable manner by an individual's HLA subtype (Alfirevic and Pirmohamed, 2010). Abacavir binds to the antigen-presenting groove of MHC I molecule in patients carrying the B*5701 allele (in patients of all ethnic groups). In vitro studies have demonstrated that treatment of antigen-presenting cells positive for B*5701, but not for the related alleles B*5702 or B*5801, with abacavir can stimulate a CD8+ T-cell response. Topical exposure to abacavir (patch test) reveals activated CD8+ T cells in skin biopsies. Genetic testing for B*5701 is highly predictive; abacavir hypersensitivity is almost never seen in individuals without the B*5701 allele and it is seen in about half of B*5701-positive carriers. Genetic testing for B*5701 prior to abacavir treatment is required by the FDA and is cost effective (the cost of treating the severe skin lesions that occur in 5%–9% of abacavir-treated patients outweighs the costs of genetic testing). The strongest association (odds ratio >1000) between HLA alleles (HAL B*1502) and IDR is seen with carbamazepine-induced skin lesions (Stevens–Johnson syndrome [SJR] and toxic epidermal necrolysis [TEN]) in Han Chinese, Thai, Malay, and Indian populations. However, B*1502 is neither predictive of SJS/TEN in Japanese or Caucasian populations (in whom HLA-A*3101 is a moderate risk factor) (McCormack et al., 2011) nor predictive of other carbamazepine-induced IDRs such as maculopapular exanthema or drug reaction with systemic symptoms (DRESS) in Asian populations (Alfirevic and Pirmohamed, 2010).

Hypersensitivity to abacavir is unusual for 2 reasons. First, an individual's HLA haplotype (the expression of B*5701) is the dominant determining factor. Second, the parent drug, which binds to the antigen-binding grove of B*5701, rather than a (reactive) metabolite, appears to be the precipitating factor, although it is possible that a metabolite formed in peripheral blood mononuclear cells (PBMCs) may contribute to or be responsible for the IDR. In other IDRs, especially drug-induced liver injury (DILI), there is evidence that a reactive metabolite triggers the idiosyncratic response by forming a protein adduct (hapten) that triggers an acquired immune response (activation of killer T cells and/or antibody-producing B lymphocytes). There is strong evidence including the production of antibodies against drug-modified proteins that the idiosyncratic DILI caused by halothane and tienilic acid are due to the formation of reactive metabolites, which is why, during the process of drug development, considerable attention is devoted to avoiding drug candidates that are converted to reactive metabolites that bind covalently to protein (see Point 12) (Obach et al., 2008; Bauman et al., 2009; Nakayama et al., 2009; Tang and Lu, 2010; Ikeda, 2011; Park et al., 2011; Walsh and Miwa, 2011). However, formation of reactive metabolites (which occurs in all individuals) is insufficient to account for the rarity of idiosyncratic DILI. For many drugs that cause idiosyncratic DILI, HLA haplotype is a factor but not to the extent seen with abacavir (Alfirevic and Pirmohamed, 2010).

In several cases, the drugs that cause idiosyncratic liver injury (those listed in Table 6-3) have closely related structures that do not cause idiosyncratic DILI. For example, rosiglitazone and pioglitazone contain the same thiazolidinedione ring that in troglitazone is converted to a reactive metabolite, and yet only troglitazone has been withdrawn from the market because of idiosyncratic hepatotoxicity. This difference may be due to the much larger daily dose of troglitazone (300 mg) compared with rosiglitazone (4-8 mg) and pioglitazone (15-45 mg). Uetrecht (2001) has proposed the general rule that drugs administered at a daily dose of 10 mg or less do not cause idiosyncratic toxicity, and this rule is consistent with the dosing information shown in Table 6-3. Both dose (more than 50 mg/day) and extent of hepatic metabolism (greater than 50%) are now recognized as risk factors for idiosyncratic drug-induced liver injury (Lammert et al., 2008 and 2010; Chalasani and Björnsson, 2010). Other examples of closely related drugs where one does and one does not cause hepatotoxicity include tolcapone/entacapone, ticlopidine/clopidogrel, amineptine/tianeptine, tamoxifen/toremifene, ibufenac/ibuprofen, ebrotidine/famotidine, and niperotidine/ranitidine (Walgren et al., 2005). Labetalol has 2 chiral centers and is a racemic mixture of 4 enantiomers, one of which is dilevalol. Although labetrol carries a black box warming for hepatotoxicity, dilevalol was not approved by the FDA because it caused a high incidence of liver toxicity during clinical trials. In some cases the difference is thought to be due to differences in dose (which is invariably higher for the drug partner that causes hepatotoxicity), but in other cases the basis for the difference is not known. Furthermore, more than 20 drugs (many of which do not cause idiosyncratic toxicity) bind covalently to their pharmacological target, including aspirin, omeprazole (and other proton pump inhibitors), clopidogrel, and neratinib (Singh et al., 2011).

Just as the acquired immune system can be activated by drugs that cause Type B toxicity (idiosyncratic toxicity) so the innate immune system can be activated by drugs that cause Type C (off-target) toxicity. For example, acetaminophen (Tylenol), which is a leading cause of acute liver failure in humans (Kaplowitz, 2005), is activated to a reactive metabolite (N-acetylbenzo-p-quinoneimine) that binds covalently to mitochondrial proteins, which leads to cell death. The cell damage (necrosis) caused by acetaminophen causes the release of cellular constituents such as fragmented DNA (and other molecules that make up the so-called danger-associated molecular pattern [DAMP]) that bind to Toll-like receptors and other receptors that activate macrophages, which in turn recruit neutrophils and monocytes to the site of injury (Jaeschke et al., 2012). Whether activation of the innate immune system in this way is detrimental (by leading to a second round of hepatocellular damage) or beneficial (by limiting damage and promoting tissue repair) is controversial. However, it is noteworthy that, in the case of Type C toxicity, the innate immune system is activated by cellular components released from damaged cells (DAMP) and not by the drug or its metabolites (although in the example given it is recognized that the reactive metabolite of acetaminophen plays a critical role in causing the cell damage that subsequently activates the innate immune system). This contrasts with the direct role of the drug and/or its metabolites (including reactive metabolites) in Type B toxicity. In the case of idiosyncratic toxicity, the drug/metabolite plays a key role in activating the acquired immune system.

Point 23 The same dose of drug (or any other xenobiotic) administered to humans often results in large interindividual differences in exposure to the parent compound (typically measured as the plasma AUC). This variation can reflect genetically determined differences in the activity of xenobiotic-biotransforming enzymes or transporters (genetic polymorphisms) or environmental factors, such as drug–drug interactions.

The study of the causes, prevalence, and impact of heritable differences in xenobiotic-biotransforming enzymes is known as pharmacogenetics. It is the basis for the concept of individualized medicine, whereby the dosage of drugs with a narrow therapeutic index (such as anticancer drugs) is adjusted in accordance with an individual's genotype (ie, drug-metabolizing potential).

Genetic variation in a xenobiotic-metabolizing enzyme produces 4 discernible phenotypes: poor metabolizers (PMs), intermediate metabolizers (IMs), extensive metabolizers (EMs), and ultrarapid metabolizers (UMs), which arise from the expression of no functional alleles (−/−) for PMs, one nonfunctional allele and one partially functional allele or two partially active alleles (−/* or */*) for IMs, one or two functional alleles (+/−, +/*, or +/+) for EMs, and gene duplication ([+/+]n) for UMs. This traditional classification scheme has been revised on the basis of an activity score, which assigns to each allelic variant a functional activity value from one (for the wild-type or *1 allele) to zero (for any completely nonfunctional allele), as shown in Table 6-4 (Zineh et al., 2004).

For those enzymes that are polymorphically expressed, such as CYP2D6, the incidence of the PM (and UM) genotype can vary widely from one ethnic group to the next. For example, the CYP2D6 PM genotype is common among Caucasians but not Asians, whereas the converse is true of the CYP2C19 PM genotype.

When the clearance of a drug is largely determined by a polymorphically expressed enzyme, such as CYP2D6, exposure to the parent drug follows the rank order: PMs > IMs > EMs > UMs. Exposure to metabolites (especially the maximum concentration, C_max) follows the opposite rank order. The difference in exposure between PMs and EMs (ie, the fold increase in C_max or AUC associated with the PM phenotype vs the EM phenotype) depends on fractional metabolism (fm), a measure of how much of the drug's clearance is determined by CYP2D6 (or whatever enzyme is responsible for metabolizing the drug in question). The fold increase in AUC (AUC_PM/AUC_EM) is equal to 1/(1 − fm); hence, the impact of the polymorphism increases dramatically as fm approaches unity. If CYP2D6 accounted for half of a drug's clearance (fm = 0.5), exposure to that drug would increase 2-fold in CYP2D6 PMs. However, if CYP2D6 accounted for 90% of a drug's clearance (fm = 0.9), then exposure to that drug would increase 10-fold in CYP2D6 PMs. This is why PMs are often at increased risk of ADEs either due to an exaggerated pharmacological response to the drug or due to its toxic side effects. An exception is omeprazole and other proton pump inhibitors, which actually provide improved control over gastroesophageal reflux disease (GERD) in CYP2C19 PMs compared with EMs (Furuta et al., 2005).

There are a minority of cases where the UMs, not the PMs, are at increased risk, especially in cases where the polymorphic enzyme catalyzes the formation of a pharmacologically active or toxic metabolite. For example, CYP2D6 converts the prodrug codeine to the active metabolite morphine, and there are reports of morphine intoxication in CYP2D6 UMs, one case being the death of a baby being breastfed by a CYP2D6 UM mother who was prescribed codeine (Gasche et al., 2004; Koren et al., 2006).

Many cases have now been described where the PM phenotype requires dosage adjustment to prevent off-target (Type C) toxicity or an exaggerated pharmacological response (Type A toxicity). For example, CYP2D6 PMs are at increased risk for perhexiline hepatotoxicity and at increased risk for an exaggerated pharmacological response to debrisoquine and sparteine, 3 drugs that were not approved in the United States because of the high incidence of adverse events in the PMs. On the other hand, when drugs are converted to active metabolites by CYP2D6, PMs derive inadequate therapeutic effect (Type F toxicity). For example, CYP2D6 converts codeine to morphine; hence, codeine is a less effective analgesic in PMs. Similarly, CYP2D6 converts the breast cancer drug tamoxifen to endoxifen (which is 30- to 100-fold more potent than tamoxifen in suppressing estrogen-dependent cell proliferation); hence, CYP2D6 PMs are at increased risk for breast cancer recurrence following tamoxifen adjuvant therapy (Goetz et al., 2005). Other examples of genetic polymorphisms that affect drug disposition in humans include CYP2C8 (repaglinide, paclitaxel), CYP2C9 (warfarin), CYP2C19 (clopidogrel, omeprazole, lansoprazole, and pantoprazole), CYP3A5 (tacrolimus), UGT1A1 (irinotecan), UGT2B7 (morphine), thiopurine methyltransferase (TPMT) (6-mercaptopurine), N-acetyltransferase 2 (NAT2) (isoniazid), catechol-O-methyltransferase (COMT) (levodopa), dihydropyrimidine dehydrogenase (DPD) (5-fluorouracil), and the efflux transporter P-glycoprotein (digoxin) (Robert et al., 2005; Relling and Giacomini, 2006). Details of these genetic polymorphisms are given later in the chapter. Many of the drugs identified in parentheses have a narrow therapeutic index. Some victim drugs have a high therapeutic index; hence, they are therapeutically effective in PMs to UMs and cause roughly the same incidence of adverse effects regardless of genotype. However, in the case of victim drugs with a narrow therapeutic index, dosage must be increased in UMs to achieve a therapeutic effect and/or it must be decreased in PMs to prevent ADEs.

Genetic polymorphisms in xenobiotic-biotransforming enzymes have an impact on the incidence of certain environmental diseases. For example, ethnic differences in the incidence of polymorphisms in alcohol dehydrogenase (ADH) and aldehyde dehydrogenase (ALDH) impact the incidence of alcoholism (Li, 2000). Genetic polymorphisms in CYP2A6 impact the incidence of cigarette-smoking-induced lung cancer. Individuals lacking CYP2A6 are PMs of nicotine (so they tend to smoke less than CYP2A6 EMs) and are poor activators of tobacco-specific mutagens (so they form fewer DNA-reactive metabolites than do CYP2A6 EMs) (Kamataki et al., 2005).

Genetic polymorphisms can be the underlying cause of a disease. For example, the severe and mild hyperbilirubinemia associated with Crigler–Najjar syndrome and Gilbert syndrome, respectively, reflect a complete and partial loss of the UGT responsible for conjugating bilirubin, namely, UGT1A1. The hyperbilirubinemic Gunn rat is the rodent equivalent of Crigler–Najjar syndrome. Genetic polymorphisms in CYP1B1 are associated with primary congenital glaucoma.

Genetic polymorphisms have been described in laboratory animals. For example, laboratory-bred rabbits have about 50:50 chance of being a poor or rapid acetylator of certain drugs (such as isoniazid) because NAT2 is polymorphically expressed in rabbits. CYP1A2 is polymorphically expressed in dogs.

Point 24 Environmental factors can introduce as much variation in drug metabolism as can genetic factors, and this is especially true of drug–drug interactions. To take CYP3A4 as an example, PMs can be created pharmacologically with inhibitors (such as ketoconazole and erythromycin), whereas UMs can be created pharmacologically with inducers (such as rifampin and the herbal agent St. John's wort). Whereas the impact of genetic polymorphisms on drug disposition and safety is often identified during clinical trials, the impact of drug–drug interactions may not be identified until after the drug has been approved and is under postmarketing surveillance.

Drugs (and other xenobiotics) can be viewed from a victim and perpetrator perspective. A drug whose clearance is largely determined by a single route of elimination, such as metabolism by a single CYP enzyme, is considered a victim drug (also known as a sensitive or object drug). Such drugs have a high victim potential because a diminution or loss of that elimination pathway, either due to a genetic deficiency in the relevant CYP enzyme or due to its inhibition by another, concomitantly administered drug, will result in a large decrease in victim drug clearance and a correspondingly large increase in exposure (the magnitude of which will depend on fm, as indicated in Point 23). Perpetrators are those drugs (or other environmental factors) that inhibit or induce the enzyme that is otherwise responsible for clearing a victim drug. In other words, perpetrators are drugs that cause drug–drug interactions. Perpetrators are also known as precipitants. Genetic polymorphisms that result in the partial or complete loss of enzymatic activity can also be viewed as perpetrators inasmuch they cause a decrease in the clearance of—and an increase in exposure to—victim drugs. A combination of 2 perpetrators (such as 2 drugs that inhibit different enzymes or an inhibitory drug plus a genetic polymorphism) can have a dramatic effect on the disposition of a victim drug if they impair parallel pathways of clearance, a phenomenon called maximum exposure (Collins et al., 2006). This is illustrated by voriconazole, whose metabolic clearance is determined by CYP2C19 and CYP3A4 (which represent 2 parallel pathways of clearance). Based on plasma AUC, CYP2C19 PMs (individuals lacking CYP2C19) are exposed to roughly twice as much voriconazole compared with CYP2C19 EMs. Loss of CYP3A4 (and only CYP3A4) by administering ritonavir to CYP2C19 EMs also causes about a doubling in systemic exposure to voriconazole. However, when ritonavir is administered to CYP2C19 PMs (such that both CYP2C19 and CYP3A4 are inactive), exposure to voriconazole increases more than 25-fold. This example illustrates how 2 drugs (ritonavir and voriconazole) can interact dramatically in some individuals (CYP2C19 PMs) but not others (CYP2C19 EMs). Conversely, a perpetrator drug that inhibited only CYP2C19 would cause such inhibition only in CYP2C19 EMs (the PMs having no enzyme to inhibit). Many other examples of perpetrator–perpetrator–victim interactions (maximum exposure) are reviewed in Ogilvie and Parkinson (in press).

Terfenadine (Seldane), cisapride (Propulsid), astemizole (Hismanal), and cerivastatin (Baycol) are all victim drugs, so much so that they have all been withdrawn from the market (Fung et al., 2001). The first 3 are all victim drugs because they are extensively metabolized by CYP3A4. Inhibition of CYP3A4 by various antimycotic drugs, such as ketoconazole, and antibiotic drugs, such as erythromycin, decreases the clearance of terfenadine, cisapride, and astemizole and increases their plasma concentrations to levels that, in some individuals, cause ventricular arrhythmias (QT prolongation) that can result in fatal heart attacks. Cerivastatin is extensively metabolized by CYP2C8. Its metabolism is inhibited by gemfibrozil (actually by gemfibrozil glucuronide), and the combination of cerivastatin (Baycol) and gemfibrozil (Lopid) is associated with a high incidence of fatal, cerivastatin-induced rhabdomyolysis (Ogilvie et al., 2006).

Mibefradil (Posicor) is the only drug withdrawn from the US market largely because of its perpetrator potential. This calcium channel blocker not only causes extensive inhibition of CYP3A4 but also causes prolonged inhibition of the enzyme by virtue of being a metabolism-dependent inhibitor of CYP3A4. By inactivating CYP3A4 in an irreversible manner—such that restoration of normal CYP3A4 activity requires the synthesis of new enzyme—mibefradil inhibits CYP3A4 long after treatment with the drug is discontinued.

The in vitro technique of reaction phenotyping (also known as enzyme mapping) is the process of identifying which enzyme or enzymes are largely responsible for metabolizing a drug candidate (Williams et al., 2003b; Ogilvie et al., 2008). Reaction phenotyping allows an assessment of the victim potential of a drug candidate or other xenobiotic. Drug candidates can also be evaluated in vitro for their potential to inhibit or induce CYP enzymes, which allows an assessment of their perpetrator potential (Tucker et al., 2001; Bjornsson et al., 2003; http://www.fda.gov/Drugs/DevelopmentApprovalProcess/DevelopmentResources/DrugInteractionsLabeling/ucm080499.htm; Huang et al., 2008; EMA, 2012; FDA, 2012.

Herbal remedies can also interact with drugs. For example, St. John's wort contains hyperforin, which is a potent PXR agonist and, as such, is an inducer of CYP3A4 (along with several other xenobiotic-metabolizing enzymes) and P-glycoprotein (ABCB1). To prevent a loss of therapeutic efficacy, the use of St. John's wort is not recommended for patients on antirejection drugs (such as cyclosporine and tacrolimus), anti-HIV drugs (such as indinavir and nevirapine), anticoagulants (such as warfarin and phenprocoumon), or oral contraceptive steroids (Pal and Mitra, 2006). Food can affect drug disposition in a number of ways. It can affect absorption, which is why some drugs are taken with meals and others are taken between meals. Large quantities of cruciferous vegetables (eg, broccoli and Brussels sprouts) can induce hepatic CYP1A2 (Conney, 1982), whereas grapefruit juice can inhibit intestinal CYP3A4 (Paine et al., 2006).

Point 25 Although drug–drug interactions can cause an increase in the incidence of adverse events or, in the case of induction, a loss of therapeutic efficacy, not all drug–drug interactions are undesirable. For example, some anti-HIV drugs, such as ritonavir, inhibit CYP3A4 and thereby improve the pharmacokinetic profile of other anti-HIV drugs, such as saquinavir and lopinavir. Tumors often overexpress various transporters and GSTs, which limit the effectiveness of several anticancer drugs. Drugs that inhibit transporters and GST are being developed to enhance the efficacy of anticancer drugs.

Point 26 Although the amount of functional enzyme determines whether an individual is a PM, IM, EM, or UM, disease states and cofactor levels can impact the rate of xenobiotic metabolism. Not surprisingly, severe liver disease decreases the rate of metabolism of a large number of drugs whose clearance is determined by hepatic metabolism, as can kidney disease that decreases the hepatic expression of CYP2D6 (Rostami-Hodjegan et al., 1999). For certain xenobiotic-metabolizing enzymes, particularly the conjugating enzymes, cofactor availability can impact the rate of xenobiotic biotransformation. The liver usually has sufficient NADPH to saturate reactions catalyzed by CYP, but it does not have sufficient levels of UDP-glucuronic acid (UDPGA) or 3′-phosphoadenosine-5′-phosphosulfate (PAPS) to saturate the enzymes responsible for glucuronidating or sulfonating xenobiotics. Consequently, alterations in the levels of these cofactors can impact the rate or extent of glucuronidation and sulfonation. For example, fasting lowers hepatic levels of these cofactors, and fasting decreases the extent to which acetaminophen is conjugated. Consequently, fasting increases the extent to which acetaminophen is metabolized to a toxic metabolite by CYP (including CYP2E1, which is induced by fasting), for which reason fasting is suspected of being one of the risk factors for acetaminophen hepatotoxicity (Whitcomb and Block, 1994).

Point 27 Stereochemical aspects can play an important role in the interaction between a xenobiotic and its biotransforming enzyme (from both a substrate and an inhibitor perspective), and xenobiotic-biotransforming enzymes can play a key role in converting one stereoisomer to another, a process known as mutarotation or inversion of configuration.

A xenobiotic that contains a chiral center can exist in 2 mirror-image forms called stereoisomers or enantiomers. The biotransformation of some chiral xenobiotics occurs stereoselectively, meaning that one enantiomer (stereoisomer) is biotransformed faster than its antipode. For example, the antiepileptic drug Mesantoin^®, which is a racemic mixture of R- and S-mephenytoin, is biotransformed stereoselectively in humans, such that the S-enantiomer is rapidly hydroxylated (by CYP2C19) and eliminated faster than the R-enantiomer. The ability of some chiral xenobiotics to inhibit xenobiotic-biotransforming enzymes can also occur stereoselectively. For example, quinidine is a potent inhibitor of CYP2D6, whereas quinine, its antipode, has relatively little inhibitory effect on this enzyme. In some cases, achiral molecules (or achiral centers) are converted to a mixture of enantiomeric metabolites, and this conversion may proceed stereoselectively such that one enantiomer is formed preferentially over its antipode. Inversion of configuration is the process whereby one enantiomer is converted to its antipode via an achiral intermediate. This interconversion can occur nonenzymatically, as in the case of thalidomide, or it can be catalyzed by a xenobiotic-metabolizing enzyme, as in the case of lisofylline, or by an endobiotic-metabolizing enzyme, as in the case of simvastatin (Fig. 6-3).

In the case of thalidomide, inversion of configuration (as shown in Fig. 6-3) occurs spontaneously, although the process is facilitated by albumin (Eriksson et al., 1998). In the mid-to-late 1950s, mainly in Europe, thalidomide was prescribed to pregnant women in the first trimester to treat morning sickness. Unfortunately, whereas (R)-thalidomide is an effective sedative, the (S)-enantiomer is a teratogen that produces phocomelia (limb shortening) and other congenital defects in the offspring largely as a result of its ability to inhibit angiogenesis (vasculogenesis). The teratogenic effect of thalidomide cannot be circumvented by administering only the (R)-enantiomer because spontaneous or albumin-facilitated racemization quickly produces the teratogenic (S)-enantiomer. Thalidomide blocks the release of tumor necrosis factor-α (TNFα) and it has been approved for the treatment of erythema nodosum leprosum (an acute inflammatory reaction associated with lepromatous leprosy), and is useful in the treatment of several other inflammatory conditions and immune-mediated diseases. Because of its ability to inhibit angiogenesis (the process that supplies tumors with new blood vessels), thalidomide is also under investigation as an anticancer drug. It is ironic that a drug with antivascular side effects that was originally prescribed as a sedative is now in clinical trials for vascular diseases with sedation as a side effect (Franks et al., 2004).

Ketones can be reduced by carbonyl reductases to a mixture of enantiomeric secondary alcohols, and this can occur with a high degree of stereoselectivity. For example, pentoxifylline is reduced by carbonyl reductases in blood and liver to a mixture of secondary alcohols with the major metabolite having an S-configuration, as shown in Fig. 6-3. Interestingly, the minor metabolite, a secondary alcohol with the R-configuration, has pharmacological properties distinct from those of its S-antipode and its ketone precursor, pentoxifylline. This minor metabolite is known as lisofylline, which is under clinical investigation for the treatment of various diseases. Through the action of carbonyl reductase, lisofylline is oxidized to pentoxifylline and then reduced to its antipode (ie, R-alcohol → ketone → S-alcohol), the net result being an inversion of configuration (Lillibridge et al., 1996). The same type of interconversion explains why the administration of pure R-albuterol to human volunteers results in the formation of S-albuterol, just as the administration of pure S-albuterol leads to the formation of R-albuterol (Boulton and Fawcett, 1997).

In the case of simvastatin (which contains a β-hydroxylactone), the interconversion of 2 secondary alcohols (β-hydroxylactone ↔ α-hydroxy-lactone) involves hydrolysis of the lactone ring (by paraoxonase 3 [PON3] in humans) followed by formation of an acyl-CoA thioester with CoA, the first step in the β-oxidation of fatty acids, followed by dehydration to an achiral intermediate. Reversal of the steps by a combination of hydrolysis and lactonization (condensation) restores the lactone ring with the hydroxyl group in the original β-configuration or in the opposing α-configuration. In humans, both the initial hydrolysis of the lactone ring and the final lactonization (condensation) reaction are catalyzed by PON3 (Draganov and La Du, 2004). Conversion to an acyl-CoA thioester also plays a role in the chiral inversion of ketoprofen. In this case, only the R-enantiomer is converted to an acyl-CoA thioester, which isomerizes and undergoes hydrolysis to the S-enantiomer. Although both R- and S-ketoprofen are converted to acyl glucuronides, the former is hydrolyzed twice as quickly as the latter, which further contributes to the accumulation of the S-enantiomer (Regan et al., 2010).

Point 28 Mass spectrometry is widely used to characterize the structure of metabolites, and many instruments now come equipped with software to assist in this process, based on the fact that certain xenobiotic reactions are associated with discrete changes in mass (Holčapek et al., 2008). For example, the loss of 2 atomic mass units (amu) signifies dehydrogenation, whereas the loss of 14 amu usually signifies demethylation (−CH₂). Several reactions result in an increase in mass, including reduction (+2 amu = 2H), methylation (+14 amu = CH₂), oxidation (+16 amu = O), hydration (+18 amu = H₂O), acetylation (+42 amu = C₂H₂O), glucosidation (+162 = C₆H₁₀O₅), sulfonation (+80 amu = SO₃), glucuronidation (+176 amu = C₆H₈O₆), carbamoyl glucuronidation (+220 amu = C₇H₈O₈), and conjugation with GSH (+305 amu = C₁₀H₁₅N₃O₆S), glycine (+57 amu = C₂H₃NO), taurine (+107 amu = C₂H₅NO₂S), and glutamine (+107 amu = C₅H₇NO₃) (Holčapek et al., 2008). Conjugation of acidic drugs with CoA (to form acyl-CoA thioesters) increases mass by 749 amu, but these conjugates are not transported out of cells and, hence, are not detected in blood, bile, or urine.

Occasionally, routine changes in mass can arise from unexpected reactions. For example, ziprasidone is converted to 2 metabolites, each of which involves an increase of 16 amu, which normally indicates addition of oxygen (eg, hydroxylation, sulfoxidation, N-oxygenation). One of the metabolites is indeed formed by addition of oxygen to ziprasidone (sulfoxidation), as shown in Fig. 6-4 (Beedham et al., 2003). However, the other metabolite is formed by a combination of reduction (+2 amu) and methylation (+14). Therefore, nominal changes in mass can sometimes leave in doubt the biotransformation event that led to metabolite formation, which accounts for the popularity of accurate mass spectrometry, a technique that can distinguish, in the case of ziprasidone, for example, oxidation from a combination of reduction and methylation.

Mass spectrometry can typically provide information on which region of a molecule has undergone biotransformation, but in some cases it cannot distinguish between several closely related possibilities. For example, based on mass spectrometry alone, it might be possible to ascertain that a certain phenyl group has been hydroxylated. However, analysis by nuclear magnetic resonance (NMR) may be required to ascertain whether the hydroxylation occurred at the ortho, meta, or para position.

Point 29 As mentioned in Point 19, xenobiotic-biotransforming enzymes are widely distributed throughout the body. The advent of high-throughput assays based on quantitative real-time reverse-transcription polymerase chain reaction (QPCR) has facilitated comprehensive measurements of the levels of mRNA encoding numerous xenobiotic-biotransforming enzymes, transporters, and xenosensors in a wide range of human tissues (Nishimura et al., 2003, 2004; Furukawa et al., 2004; Nishimura and Naito, 2005, 2006). These data are very informative as far as mRNA expression is concerned, but it can be difficult to draw conclusions about the activity of a given enzyme in a given tissue based on mRNA levels alone. For instance, based on mRNA levels in human liver, the levels of the major xenobiotic-biotransforming CYP enzymes follow the rank order: CYP2E1 > 2A6 > 2C8 > 2C18 ≈ 1A2 ≈ 4A11 ≈ 2C9 ≈ 3A4 (Nishimura et al., 2003). This contrasts substantially from the levels of CYP protein (as determined by LC–MS/MS), which follow the rank order: CYP2C9 > 3A4 ≈ 2E1 ≈ 2A6 > 2C8 > 1A2 > 2D6 > 2B6 ≈ 2C19 > 3A5 ≈ 3A43 ≈ 3A7 (Kawakami et al., 2011; Ohtsuki et al., 2012). In the case of CYP2E1, hepatic mRNA levels are more than 17 times higher than the levels of CYP3A4 mRNA, but CYP2E1 protein levels in microsomes are comparable to those of CYP3A4. This discrepancy is due to the fact that, under normal conditions, most of the mRNA encoding CYP2E1 is sequestered in the cytoplasm and is not available for translation (Gonzalez, 2007). The mRNA for FMO2 is present at very high levels in human lung, but is not translated into functional enzyme due to the presence of a truncation mutation in Caucasians and Asians. However, about 33% of sub-Saharan Africans, 26% of African Americans, 7% of Puerto Ricans, and 2% of Mexicans have one normal allele and express a functional protein (Cashman and Zhang, 2006; Veeramah et al., 2008). Human CYP2A7 and CYP4B1 are full-length genes that encode enzymes incapable of incorporating heme and are therefore nonfunctional.

Mass spectrophotometric techniques for the quantitative analysis of proteins have been applied to drug-metabolizing enzymes and transporters (Kawakami et al., 2011; Sakamoto et al., 2011; Ohtsuki et al., 2012). The levels of CYP and UGT enzymes in human liver microsomes have been determined by LC–MS/MS and are described in the sections “Cytochrome P450” and “Glucuronidation and Formation of Acyl-CoA Thioesters.”

Hydrolysis

Mammals contain a variety of enzymes that hydrolyze xenobiotics containing such functional groups as a carboxylic acid ester (delapril and procaine), amide (procainamide), thioester (spironolactone), carbamate (irinotecan), phosphoric acid ester (paraoxon), acid anhydride (diisopropylfluorophosphate [DFP]), lactone (lovastatin), and thiolactone (erdosteine), most of which are shown in Figs. 6-5 and 6-6. The major hydrolytic enzymes are the carboxylesterases, cholinesterases, and paraoxonases (for which lactonase is a more encompassing name), but they are by no means the only hydrolytic enzymes involved in xenobiotic biotransformation. The first 2 classes of hydrolytic enzymes, the carboxylesterases and cholinesterases, are known as serine esterases because their catalytic site contains a nucleophilic serine residue that participates in the hydrolysis of various xenobiotic and endobiotic substrates and the stoichiometric (one-to-one) binding of organophosphorus (OP) compounds and other cholinergic neurotoxins.

Approximately 100 human gene products encode serine hydrolases that are classified as esterases, amidases, thioesterases, lipases, peptidases, or proteases (Evans and Cravatt, 2006; Ross and Crow, 2007; Testa and Krämer, 2008, 2010). The active-site serine residue of carboxylesterases, cholinesterases, and other serine hydrolases becomes phosphorylated (or phosphonylated) by OP compounds, such as those used as insecticides, herbicides, fungicides, nematicides, and plant growth regulators. Binding of OP compounds to carboxylesterases, cholinesterases, and other targets, some of which have been identified as receptors and enzymes involved in the hydrolysis of endobiotics (reviewed in Casida and Quistad, 2005), plays a key role in limiting the binding of OP compounds to acetylcholinesterase (AChE). Phosphorylation of AChE, which hydrolyzes acetylcholine and thereby terminates its neurotransmitter activity, is the principal mechanism of OP toxicity in mammals, insects, and nematodes, with 70% to 90% inhibition usually proving lethal. Reversal of this phosphorylation event with pyridinium oximes such as pralidoxime (2-PAM) and obidoxime (toxogonin) is one of the strategies to treat OP poisoning (Casida and Quistad, 2005).

Albumin, lipases, peptidases, proteases, ALDHs, and carbonic anhydrases have all been shown to have hydrolytic (esteratic) activity toward various xenobiotics. CYP can catalyze the cleavage of certain xenobiotics containing a carboxylic acid ester, phosphoric acid ester, or carbamate (see the section “Cytochrome P450” for examples).

In the presence of an alcohol, carboxylesterases and certain other hydrolytic enzymes can catalyze the transesterification of xenobiotics, which accounts for the conversion of cocaine (a methyl ester) to ethylcocaine (the corresponding ethyl ester) (Fig. 6-5). The same transesterification occurs with clopidogrel, which is converted from a methyl to an ethyl ester (Tang et al., 2006). Transesterification occurs when ethanol, not water, cleaves the catalytic transition state, that is, the esteratic bond between the active serine residue on the enzyme and the carbonyl group on the xenobiotic:

Enzyme−O−CO-R + CH₃CH₂OH → enzyme–OH + CH₃CH₂−O−CO-R.

In humans, the hydrolysis of xenobiotics (including many prodrugs) is largely catalyzed by microsomal carboxylesterases in liver (CES1 and CES2) and intestine (CES2), and by cholinesterases, paraoxonases, and albumin in blood (some of these enzymes are present in plasma; others are bound to erythrocytes in a species-dependent manner) (Li et al., 2005). Compared with many other mammalian species humans are unusual because they lack a plasma carboxylesterase (Li et al., 2005). However, human erythrocytes contain esterase D, a carboxylesterase used as a genetic marker for retinoblastoma (Wu et al., 2009). On a case-by-case basis, specific enzymes other than those mentioned above can be involved in xenobiotic hydrolysis. For example, the valine ester prodrugs of acyclovir (valacyclovir) and gangciclovir (valgangciclovir) are hydrolyzed by valacyclovirase (gene code BPHL), a serine- and α/β-fold-hydrolase that hydrolyzes other antiviral and anticancer nucleoside drugs such as zidovudine, floxuridine, and gemcitabine (Lai et al., 2008).

In general, esters are hydrolyzed more rapidly than amides, which can impact the duration and site of action of drugs. For example, procaine, a carboxylic acid ester, is rapidly hydrolyzed, which is why this drug is used mainly as a local anesthetic. In contrast, procainamide, the amide analog of procaine, is hydrolyzed much more slowly; hence, this drug reaches the systematic circulation, where it is useful in the treatment of cardiac arrhythmia. In general, enzymatic hydrolysis of amides occurs more slowly than esters, although electronic factors can influence the rate of hydrolysis. The presence of electron-withdrawing substituents weakens an amide bond, making it more susceptible to enzymatic hydrolysis.

The hydrolysis of xenobiotics by carboxylesterases and other hydrolytic enzymes is not always a detoxication process. Fig. 6-5 shows some examples in which carboxylesterases convert xenobiotics to toxic and tumorigenic metabolites.

In 1953, Aldridge classified hydrolytic enzymes on the basis of their interaction with OP compounds, classifying those that hydrolyze OP compounds as A-esterases, those that are inhibited by OP compounds as B-esterases, and those that do not interact with OP compounds as C-esterases. Although the terms are still used, the classification system of Aldridge can be somewhat confusing because it divides the paraoxonases into the A- and C-esterase classes: the human paraoxonase PON1 hydrolyzes OP compounds and so can be classified as an A-esterase, whereas PON2 and PON3 can be classified as C-esterases because they neither hydrolyze OP compounds nor, in most cases, are inhibited by them. Furthermore, carboxylesterases and cholinesterases, 2 distinct classes of hydrolytic enzymes, are both B-esterases according to Aldridge because both are inhibited by OP compounds.

Carboxylesterases

Carboxylesterases are predominantly microsomal enzymes (~60 kDa glycoproteins) that are present in liver, intestine, and a wide variety of tissues, including plasma in rats and mice but not humans. The 2 major human carboxylesterases involved in xenobiotic hydrolysis are CES1 and CES2, which differ in their tissue distribution and substrate specificity. Both enzymes are expressed in liver microsomes (although CES1 predominates) but only CES2 is expressed in intestinal microsomes (Nishimura and Naito, 2006; Ross and Crow, 2007). CES1 prefers to hydrolyze xenobiotics with a small alcoholic leaving group, whereas CES2 prefers to hydrolyze xenobiotics with a large alcoholic leaving group. In other words, methyl and ethyl esters tend to be hydrolyzed by CES1. This is illustrated in Fig. 6-5 for the hydrolysis of delapril by CES1 (which releases ethanol, a small alcohol) and the hydrolysis of procaine by CES2 (which releases a large alcohol). In the case of cocaine (the structure of which is shown in Fig. 6-5), the ethyl ester is hydrolyzed by CES1 (to release a small alcohol), whereas the benzoic ester is hydrolyzed by CES2 (to release a large alcohol). CES1 also catalyzes the transesterification of the methyl ester of cocaine, as shown in Fig. 6-5. CES1 is more active than CES2 at catalyzing the hydrolysis of oseltamivir, benazepril, cilazepril, quinapril, temocapril, imidapril, meperidine, delapril, and clopidogrel, whereas CES2 is more active than CES1 at hydrolyzing aspirin, heroin, cocaine benzoyl ester, 6-acetylmorphine, oxybutynin, and the anticancer drug irinotecan (also known as CPT-11) (Satoh and Hosokawa, 2006; Shi et al., 2006; Tang et al., 2006). Individual pyrethroids such as trans-permethrin represent an example of xenobiotics that are hydrolyzed by both CES1 and CES2 (Ross and Crow, 2007).

In addition to hydrolyzing xenobiotics, carboxylesterases hydrolyze numerous endogenous compounds, such as long- and short-chain acyl-glycerols (both monoacylglycerols and diacylglycerols), long-chain acyl-carnitine, long-chain acyl-CoA thioesters (eg, palmitoyl-CoA), retinyl ester, platelet-activating factor, and other esterified lipids. Carboxylesterases can also catalyze the synthesis of fatty acid ethyl esters, which represents a nonoxidative pathway of ethanol metabolism in adipose and certain other tissues. In the case of platelet-activating factor, carboxylesterases catalyze both the deacetylation of PAF and its subsequent esterification with fatty acids to form phosphatidylcholine (Satoh and Hosokawa, 1998).

In addition to CES1 and CES2, human liver microsomes contain another serine hydrolase known as AADAC, which stands for arylacetamide deacetylase. This enzyme catalyzes the hydrolysis of the phenacetin (a discontinued drug), 2-acetylaminofluorene (2-AAF; a carcinogen), and flutamide (an antiandrogen drug) (Watanabe et al., 2010). AADAC deacetylates phenacetin and 2-aminofluorene (2-AF) to aromatic amines that can be N-hydroxylated by CYP and then conjugated to form reactive metabolites. Examples of this type of metabolic activation are given later in the sections “Azo- and Nitro-Reduction” and “Conjugation”).

CES1 (the major liver form, which is also expressed in lung and other tissues) and CES2 (the major intestinal form, which is also expressed in kidney and brain) represent 2 of the 5 families of human carboxylesterases (Satoh and Hosokawa, 2006; Holmes et al., 2010). The other enzymes are CES3 (expressed in brain, liver, and colon), CES4A (previously called CES6 or CES8, which is expressed in brain, lung, and kidney), and CES5A (previously called CES7, which is expressed in brain, lung, and testis). A larger number of carboxylesterases have been identified in rats and mice (Holmes et al., 2010). Human CES1 is encoded by 2 genes (CSE1A1 and CES1A2) that differ only in the amino acid sequence of the encoded signal recognition peptide (SRP) that directs the enzyme to the endoplasmic reticulum (Satoh and Hosokawa, 2006).

Genetic polymorphisms that affect carboxylesterase activity or expression levels underscore the importance of CES1 and CES2 in drug metabolism. Genetic polymorphisms of CES1 affect the disposition of methylphenidate (Ritalin, a methyl ester) and the antiviral prodrug oseltamivir (an ethyl ester) (Zhu and Markowitz, 2009). A phenotype for CES1 that might be classified as high EM has been described. It arises from a single-nucleotide polymorphism (SNP) in the promoter region of CES1A2 (but not CES1A1) that increases the expression of CES1 and thereby increases the rate of hydrolysis of imidapril to its active metabolite imidaprilat, an angiotensin-converting enzyme (ACE) inhibitor, which increases its antihypertensive effect (Geshi et al., 2005). Genetic polymorphisms of CES2 affect the disposition of the anticancer drug irinotecan (CPT-11, a carbamate), which is converted by CES2 to the active metabolite SN-38, a topoisomerase inhibitor (Kubo et al., 2005). However, genetic polymorphisms of UGT1A have a greater impact on the disposition and toxicity of irinotecan, as detailed in the section “Glucuronidation and Formation of Acyl-CoA Thioesters.”

Certain carboxylesterases also have a physiological function in anchoring other proteins to the endoplasmic reticulum. For example, the lysosomal enzyme β-glucuronidase is also present in the endoplasmic reticulum, where it is anchored in the lumen by egasyn, a microsomal carboxylesterase designated Ces1e in mouse and rat (Holmes et al., 2010). Egasyn binds to β-glucuronidase at its active-site serine residue, which effectively abolishes the carboxylesterase activity of egasyn, although there is no corresponding loss of β-glucuronidase activity. Binding of OP compounds to egasyn causes the release of β-glucuronidase into plasma, which serves as the basis for a test for OP exposure (Fujikawa et al., 2005). The retention of β-glucuronidase in the lumen of the ER is thought to be physiologically significant. Glucuronidation by microsomal UGTs is a major pathway in the clearance of many of the endogenous aglycones (such as bilirubin) and xenobiotics (such as drugs). However, hydrolysis of glucuronides by β-glucuronidase complexed with egasyn in the lumen of the ER appears to be an important mechanism for recycling endogenous compounds, such as steroid hormones (Dwivedi et al., 1987). The acute-phase response protein, C-reactive protein, is similarly anchored in the endoplasmic reticulum by egasyn.

The mechanism of catalysis by carboxylesterases is analogous to the mechanism of catalysis by serine proteases. In the case of carboxylesterases, it involves charge relay among a catalytic triad comprising an acidic amino acid residue (glutamate [Glu₃₃₅]), a basic residue (histidine [His₄₄₈]), and a nucleophilic residue (serine [Ser₂₀₃]) (Yan et al., 1994; Satoh and Hosokawa, 1998). (These amino acid residues, numbered for a rat carboxylesterase, differ slightly in other species, but the overall location and function of these residues are the same in all mammalian carboxylesterases.) The mechanism of catalysis of carboxylesterases is shown in Fig. 6-7, and is discussed in more detail in the section “Epoxide Hydrolases.” OP compounds bind to the nucleophilic OH-group on the active-site serine residue to form a phosphorus–oxygen bond, which is not readily cleaved by water. Therefore, OP compounds bind stoichiometrically to carboxylesterases and inhibit their enzymatic activity, for which reason they are also classified as B-esterases (Aldridge, 1953). Surprisingly, the stoichiometric binding of OP compounds to carboxylesterases and cholinesterases is an important determinant of OP toxicity, as outlined in the section “Cholinesterases (AChE and BChE).”

As mentioned previously in this section, a plasma carboxylesterase, namely, Ces1c (previously known as Es1), is present in mouse plasma but not human plasma and accounts, at least in part, for the relative resistance of mice to the OP nerve agent soman. Duysen et al. (2011) demonstrated that plasma carboxylesterase knockout mice are considerably more susceptible to soman toxicity than wild-type mice.

Given the lack of plasma carboxylesterase in humans, the hydrolysis of xenobiotics in human blood is catalyzed by cholinesterases and paraoxonases (with a significant contribution from albumin on a case-by-case basis), as described in the following sections.

Cholinesterases (AChE and BChE)

Humans have 2 cholinesterases, namely, acetylcholinesterase (AChE; gene name ACHE) and butyrylcholinesterase (BChE, also known as pseudocholinesterase; gene name BCHE), which are related enzymes (about 54% identical). They are present in most tissues. The levels of BChE are higher than those of AChE except in brain and muscle, tissues where AChE terminates the action of the neurotransmitter acetylcholine. In human plasma, the levels of BChE are 100- to 1000-fold greater than those of AChE, although the latter enzyme is present in erythrocytes (Li et al., 2005). As the names imply, AChE and BChE have high activity toward acetylcholine and butyrylcholine, respectively. AChE is highly selective for acetylcholine and plays little or no significant role in the hydrolysis of xenobiotics, whereas BChE hydrolyzes numerous drugs (and other xenobiotics) including aspirin, bambuterol, chlorpropaine, cocaine, flestolol, heroin, irinotecan, isosorbide diaspirinate, methylprednisolone acetate, mivacurium, moxisylyte, n-octanoyl ghrelin, procaine, succinylcholine (suxamethonium), and tetracaine (Li et al., 2005). Eserine (physostigmine) is an inhibitor of both enzymes, whereas BW284C51 is a selective inhibitor of AChE, and iso-OMPA, bambuterol, tolserine, and bis-norcymserine are selective inhibitors of BChE (Liederer and Borchardt, 2006; Masson and Lockridge, 2010). Drugs that selectively inhibit brain AChE and BChE activity, such as rivastigmine (Exelon^®), have been used to treat Alzheimer disease. Other drugs that inhibit AChE and are used to treat Alzheimer disease include tacrine (Cognex^®), gelantamine (Reminyl^®), and donepezil (Aricept^®).

Both enzymes exist in 6 different forms with differing solubility: monomer (G1), dimer (G2), tetramer (G4), tailed tetramers (A4), double tetramers (A8), and triple tetramers (A12). G1, G2, and G4 contain 1, 2, and 4 subunits, each with a catalytic site. These various forms can each exist in 3 states: soluble (hydrophilic), immobilized (asymmetric), and amphiphilic globular (membrane-bound through attachment to the phospholipid bilayer) (Nigg and Knaak, 2000). All forms are expressed in muscle. In the case of AChE, the major form in brain is the tetramer G4 (anchored with a 20-kDa side chain containing fatty acids), but the major form in erythrocytes is the dimer G2 (anchored with a glycolipid-phosphatidylinositol side chain). In the case of BChE, the major form in plasma is the tetramer G4 (a glycoprotein with Mr 342 kDa). In both AChE and BChE, the esteratic site (containing the active-site serine residue) is adjacent to an anionic (negatively charged) site that interacts with the positively charged nitrogen on acetylcholine and butyrylcholine.

Genetic variants of AChE that eliminate its activity have not been described, which is not surprising given the key role that AChE plays in terminating neurotransmission by acetylcholine, although AChE knockout mice (AChE^−/−) are born alive and, despite developmental abnormalities, survive up to 21 days (Xie et al., 2000). Based on measurements of erythrocyte AChE activity, familial reductions of 30% have been reported, and a reduction of 50% has been linked to paroxysmal nocturnal hemoglobinuria (Nigg and Knaak, 2000).

More than 70 genetic variants of BChE have been described following the discovery of poor metabolizers (PMs) of succinylcholine (suxamethonium) and, later, mivacurium. Succinylcholine and mivacurium are muscle relaxants whose duration of action is determined by plasma BChE. Succinylcholine (1.5-2.0 mg/kg) is a rapidly acting muscle relaxant (the onset of paralysis takes 30-60 seconds) with a short duration of action (8-15 minutes) making it well suited for intubating patients. In some individuals, succinylcholine causes prolonged (60-120 minutes) paralysis (muscular relaxation and apnea), which led to the discovery of 2 BChE genetic polymorphisms (now known as BCHE*A and BCHE*K) (Cerf et al., 2002). The so-called A variant of BChE (Asp₇₀Gly) has markedly reduced enzymatic activity (less than 10% of the wild-type enzyme) but is relatively rare; about 1 in 300 Caucasians are homozygous for the A variant. The K variant (Ala₅₃₉Thr) is considerably more common (with one in 63 individuals being homozygous) but the K variant still retains approximately two thirds of its enzymatic activity. Consequently, the A variant causes a greater impairment of succinylcholine (and mivacurium) metabolism than does the K variant (La Du, 1992; Lockridge, 1992; Levano et al., 2008). Although the A variant has markedly diminished activity toward succinylcholine (due to a ~100-fold decrease in affinity [K_m]), it nevertheless has appreciable activity toward other substrates, such as acetylcholine and benzoylcholine. Wild-type BChE and the A variant are equally sensitive to the inhibitory effect of OP compounds, but the allelic variant is relatively resistant to the inhibitory effect of dibucaine, a local anesthetic, which forms the basis of a diagnostic test for its presence (frequently called a test for atypical pseudocholinesterase). The percent inhibition of hydrolysis of benzoylcholine by dibucaine (with both the substrate and inhibitor at 10 μM) is called the dibucaine number; it is 80% or more with wild-type BChE and about 40% with the A variant in most but not all cases (Cerf et al., 2002). The discovery of the A variant of BChE (the so-called atypical pseudocholinesterase) is of historical interest because it ushered in the new field of pharmacogenetics, a field pioneered by Dr Werner Kalow, after whom the K variant of BChE is named.

Carboxylesterases and cholinesterases in the blood and tissues play an important role in limiting the amount of cholinergic neurotoxins that reach AChE in the brain, marked inhibition (70%–90%) of which is lethal to mammals, insects, and nematodes. The cholinergic neurotoxins that are bound covalently to—or are hydrolyzed by—these enzymes include OP nerve agents (such as soman and sarin), OP pesticides (such as parathion, malathion, and chlorpyros, which are converted to oxons by CYP), carbamate pesticides (such as aldicarb, carbaryl, carbofuran), the naturally occurring OP compound anatoxin-a(S) from blue-green algae, physostigmine (eserine) from the Calabar bean, huperzine A from the club moss Huperzia serrata, solandine from green potatoes, and cocaine from the Erythroxylum coca plant. The covalent interaction between OP compounds and brain AChE is analogous to their binding to the active-site serine residue in all serine esterases (B-esterases). As previously mentioned, certain OP compounds are hydrolyzed by A-esterases (the paraoxonase PON1) but bind stoichiometrically and, for the most part, irreversibly to B-esterases (carboxylesterases and cholinesterases). Surprisingly, stoichiometric binding of OP compounds to carboxylesterase and cholinesterase (and perhaps to numerous other enzymes and receptors that have structural features common to serine esterases) plays an important role in limiting the toxicity of OP compounds. Numerous studies have shown an inverse relationship between serine esterase activity and susceptibility to the toxic effect of OP compounds. Factors that decrease serine esterase activity potentiate the toxic effects of OP compounds, whereas factors that increase serine esterase activity have a protective effect. For example, the susceptibility of animals to the toxicity of parathion, malathion, and diisopropylfluorophosphate (DFP) is inversely related to the level of plasma esterase activity (which reflects BChE activity and, in some species, carboxylesterase activity). Differences in the susceptibility of several mammalian species to OP toxicity can be abolished by pretreatment with selective serine esterase inhibitors such as cresylbenzodioxaphosphorin oxide, the active metabolite of tri-ortho-tolylphosphate (which is also known as tri-ortho-cresylphosphate [TOCP]). Knockout mice that lack plasma carboxylesterase (Ces1c, previously known as Es1) or AChE are more susceptible to the toxic effects of OP compounds, as are PON1 knockout mice in some but not all cases (discussed in the section “Paraoxonases (Lactonases)”) (Xie et al., 2000; Duysen et al., 2011). Somewhat surprisingly, BChE knockout mice are not more sensitive to OP toxicity than wild-type mice (Duysen et al., 2007), but this may not reflect the situation in humans because mouse plasma contains a carboxylesterase, whereas humans do not (see the section “Carboxylesterases”). In humans, BChE plays an important role in preventing OP compounds from reaching AChE in the brain. Carboxylesterases, cholinesterases, and paraoxonases are not the only enzymes involved in the detoxication of OP pesticides. Certain OP compounds are detoxified by CYP, flavin monooxygenases, and GSTs. However, paraoxonases, enzymes that catalyze the hydrolysis of certain OP compounds, appear to play a limited role in determining susceptibility to OP toxicity, as outlined in the following section.

Once bound to the active site of cholinesterase or carboxylesterase, OP compounds can undergo dealkylation reactions that further retard their release from the active-site serine residue of these enzymes. This process is called “aging.” In many cases the phosphorylated or phosphonylated enzyme can be reactivated by displacement of the OP adduct with nucleophilic compounds such as fluoride, hydroxamates, and oximates. The pyridinium oximes pralidoxime (2-PAM) and obidoxime (toxogonin) are used therapeutically as antidotes to OP poisoning. More potent oximes (HI-6 and MMB-4) are under development (Masson and Lockridge, 2010). Cocaine is hydrolyzed by CES1 and CES2 in liver and intestine, and by BChE in plasma. Genetic polymorphisms that reduce BChE activity toward succinylcholine (such as the A variant described earlier in this section) also reduce its activity toward cocaine (again by decreasing K_m), which exacerbates cocaine toxicity (Masson and Lockridge, 2010). Despite its important role in cocaine toxicity, BChE hydrolyzes cocaine slowly. Masson and Lockridge (2010) reviewed efforts based on molecular dynamics simulation to improve the hydrolytic function of BChE through site-directed mutagenesis. A BChE variant with 4 mutations hydrolyzes cocaine with a catalytic efficiency 1500- to 5000-fold greater than that of the wild-type enzyme, whereas a variant with 5 mutations is 6500 times more active. The latter man-made variant is of therapeutic interest for the treatment of cocaine overdose. Other variants of BChE are being developed with improved hydrolysis of OP compounds for the treatment of OP poisoning and protection against OP nerve agents.

Paraoxonases (Lactonases)

Paraoxonases catalyze the hydrolysis of a broad range of organophosphates, organophosphinites, aromatic carboxylic acid esters (such as phenylacetate), cyclic carbonates, lactones, and oxidized phospholipids. They are calcium-dependent enzymes containing a critical sulfhydryl (−SH) group; as such they are inhibited by EDTA, metal ions (Cu and Ba), and various mercurials such as phenylmercuric acetate (PMA), para-chloromercuribenzoate (PCMB), and the PCMB hydrolysis product, para-hydroxymercuribenzoate. (Note: Calcium must be added to measure paraoxonase activity in plasma prepared from EDTA-anticoagulated blood.) Based on the observation that A-esterases are inhibited by PCMB but not OP compounds, Augustinsson (1966) postulated that, in the case of paraoxonases, OP compounds bind to a nucleophilic SH-group on an active-site cysteine residue and form a phosphorus–sulfur bond, which is readily cleaved by water. A strong argument against this postulate is the fact that there is no loss of enzymatic activity when the only potential active-site cysteine residue in human paraoxonase (Cys₂₈₃) is substituted with serine or alanine (Sorenson et al., 1995). However, substitution with serine or alanine renders paraoxonase resistant to inhibition by PCMB, placing Cys₂₈₃ near but not in the active site of paraoxonase. Paraoxonase requires Ca²⁺, for both stability and catalytic activity, which raises the possibility that the hydrolysis of OP compounds by paraoxonase involves metal-catalyzed hydrolysis, analogous to that proposed for calcium-dependent phospholipase A2 or zinc-dependent phosphotriesterase activity (Sorenson et al., 1995). A structurally modified but catalytically active form of recombinant PON1 has been crystallized and shown by x-ray analysis to contain 2 calcium ions in a 6-fold β-propeller protein similar to that found in squid diisopropylfluorophosphatase (DFPase), a paraoxonase-like enzyme (Otto et al., 2009).

Humans express 3 paraoxonases designated PON1, PON2, and PON3. PON1 is present in liver microsomes and plasma, where it is associated exclusively with high-density lipoprotein (HDL). PON2 is not present in plasma but it is expressed in the inner mitochondrial membrane of vascular cells and many tissues (Devarajan et al., 2011). PON3 is expressed in liver and kidney microsomes and plasma. Only PON1 has appreciable arylesterase activity and the ability to hydrolyze the toxic oxon metabolites of OP insecticides such as parathion (paraoxon), diazinon (diazoxon), and chlorpyrifos (chlorpyrifos oxon) (Gupta and DuBois, 1998; Draganov and La Du, 2004). However, all 3 enzymes can catalyze the hydrolysis of various lactones, for which reason the name “lactonase” is more encompassing. A lactone derived from arachidonic acid (5-hydroxy-cicosate traeomic acid-1,5-lactone) is one of the few substrates hydrolyzed by all 3 paraoxonases (Gupta et al., 2009). Lactone hydrolysis of the statins lovastatin and simvastatin is catalyzed only by PON3. Reports of the same reaction being catalyzed by PON1 appear to be attributable to trace contamination with PON3 (Draganov and La Du, 2004). However, both PON1 and PON3 hydrolyze the lactone form of atorvastatin, which reverses the lactonization of atorvastatin, a reaction that involves formation of an acyl glucuronide of atorvastatin by UGT1A3 (Riedmaier et al., 2011). PON1 hydrolyzes thiolactones such as homocysteine thiolactone (an endobiotic substrate) and a thiolactone metabolite of clopidogrel (discussed later in the section “CYP2C19”), whereas PON3 hydrolyzes the lactone spironolactone (a diuretic drug) (Gupta et al., 2009). PON1 is the major plasma enzyme responsible for hydrolyzing the prodrug olmesartan medoxomil to its active metabolite (Ishizuka et al., 2012) and is one of the plasma enzymes responsible for hydrolyzing the structurally related prodrugs prulifloxacin and ceftobiprole medocaril to their active metabolite (Eichenbaum et al., 2012).

PON2, located on the inner mitochondrial membrane, appears to play no significant role in xenobiotic biotransformation, although it can hydrolyze the lactone dihydrocoumarin. In the inner mitochondrial membrane, PON2 is bound to coenzyme Q₁₀ and complexed with respiratory complex III where it protects the mitochondrion from oxidative damage from superoxide anion and its derivative ROS (hydrogen peroxide, peroxynitrate, and hydroxy radicals) (Devarajan et al., 2011).

PON1 has 2 prominent polymorphisms in the coding region, namely, Q₁₉₂R (Glu₁₉₂Arg) and L₅₅M (Leu₅₅Met), and several polymorphisms in the promoter region. Polymorphisms in the promoter region and the L₅₅M polymorphism in the coding region do not affect PON1 activity but they do affect expression levels (Gupta et al., 2009; Furlong et al., 2010). The commonest genetic polymorphism, Q₁₉₂R, affects PON1 activity in a substrate-dependent manner. The glutamine (Q₁₉₂) and arginine (R₁₉₂) allelozymes have the same hydrolytic activity toward para-nitrophenylacetate (a measure of arylesterase activity) and diazoxon (the oxon of diazinon) but R₁₉₂ is more active toward paraoxon (the oxon of parathion), chlorpyrifos oxon, the prodrugs olmesartan medoxomil and prulifloxacin, and the lactone pilocarpine, whereas Q₁₉₂ is more active toward soman, sarin, the thiolactone metabolite of clopidogrel (Bouman et al., 2011; discussed later in the section “CYP2C19”), and lipid peroxides, the significance of which is discussed later in this section (Gupta et al., 2009; Furlong et al., 2010; Eichenbaum et al., 2012; Ishizuka et al., 2012).

There is evidence to suggest that PON1 protects against atherosclerosis by hydrolyzing specific derivatives of oxidized cholesterol and/or phospholipids in atherosclerotic lesions and in oxidized low-density lipoprotein (LDL). For example, mice lacking PON1 (knockout mice or PON1 null mice) are predisposed to atherogenesis, whereas mice overexpressing PON1 are protected (Draganov and La Du, 2004; Gupta et al., 2009). Some studies show that individuals who are homozygous for the R₁₉₂ allele are at increased risk of atherosclerosis and ischemic stroke compared with individuals who are homozygous for the Q₁₉₂ allele, suggesting that the absence of the latter enzyme, the PON1 allelozyme with high activity toward lipid peroxides and offering more protection against LDL oxidation, is a risk factor for cardiovascular disease (Gupta et al., 2009; Dahabreh et al., 2010). However, the association is controversial. A complicating factor is that the R₁₉₂ allelozyme is expressed at higher levels than the Q₁₉₂ enzyme, which may partially offset its lower ability to hydrolyze lipid peroxides. Bayrak et al. (2011) proposed that, in terms of assessing the risk of atherosclerosis, an assessment of both plasma PON1 activity and genotype is more reliable than genotype alone. This seems appropriate in view of the finding that several environmental factors affect PON1 levels, which are upregulated by hypolipidemic drugs (fenofibrate and statins) and cardioprotective dietary components such as polyphenols (such as resveratrol), oleic acid, and olive oil, and downregulated by diabetes and a high-fat (proatherogenic) diet (Gupta et al., 2009).

It seems reasonable to assume that PON1 would play an important role in determining the susceptibility of humans to OP toxicity based on 3 considerations: first, PON1 hydrolyzes OP compounds (rather than simply bind them stoichiometrically like BChE). Second, the concentration of PON1/3 in human plasma is 10 times greater than that of BChE (50 mg/mL vs 5 mg/mL) (Li et al., 2005). Third, human plasma does not contain carboxylesterase (Li et al., 2005). PON1/3 may protect against OP toxicity in some but not all cases. PON1 knockout mice are no more susceptible than wild-type mice to the toxic effects of paraoxon (the active metabolite of parathion), and administration of human PON1 (either the R₁₉₂ or the Q₁₉₂ allelozyme) to PON1 knockout mice does not confer protection against paraoxon. However, administration of either the R₁₉₂ or Q₁₉₂ human allelozymes to PON1 knockout mice does protect against diazoxon toxicity, and administration of the R₁₉₂ allelozyme (but not the Q₁₉₂ allelozyme) protects against chlorpyrifos oxon (Cole et al., 2010). The allelozyme-dependent pattern of protection corresponds with the rate of hydrolysis of these OP oxons by R₁₉₂ and Q₁₉₂ (high and equal for diazoxon, high but greater with R₁₉₂ for chlorpyrifos oxon, and low but greater with R₁₉₂ for paraoxon). Although the concentration of BChE in human plasma is only one tenth that of PON1/3 (~50 nM vs 500 nM), its apparent second-order rate constant with OP compounds is very high (10⁷ to 10⁹ M⁻¹ min⁻¹) compared with the catalytic efficiency of PON1 (10⁵ M⁻¹ min⁻¹) (Masson and Lockridge, 2010).

Diisopropylfluorophosphatase (DFPase), a squid enzyme that catalyzes the release of fluoride from DFP (Fig. 6-5), is a hydrolytic enzyme related to the paraoxonases. It hydrolyzes the nerve gas agents sarin and soman (Liederer and Borchardt, 2006). Human paraoxonases do not hydrolyze DFP but are inhibited by this OP compound.

Prodrugs and Alkaline Phosphatase

Many prodrugs are designed to be hydrolyzed by hydrolytic enzymes (Liederer and Borchardt, 2006). Some prodrugs, such as propranolol ester, are hydrolyzed by both carboxylesterases and cholinesterases (both AChE and BChE), whereas others are preferentially or specifically hydrolyzed by carboxylesterases (capecitabine, irinotecan), BChE (bambuterol, methylprednisolone acetate), hPON1 (prulifloxacin, olmesartan medoxomil), or PON3 (lovastatin, simvastatin). On a case-by-case basis, specific enzymes other than those mentioned above can be involved in xenobiotic hydrolysis. For example, the valine ester prodrugs of acyclovir (valacyclovir) and gangciclovir (valgangciclovir) are hydrolyzed by valacyclovirase (gene code BPHL), a serine- and α/β-fold-hydrolase that hydrolyzes other antiviral and anticancer nucleoside drugs such as zidovudine, floxuridine, and gemcitabine (Lai et al., 2008). Some prodrugs are hydrolyzed with a high degree of stereoselectivity. For example, in the case of prodrugs of ibuprofen and flurbiprofen, the R-enantiomer is hydrolyzed about 50 times faster than the S-enantiomer. The acyl glucuronides of ibuprofen and other NSAIDs are also hydrolyzed stereoselectively (see the section “Glucuronidation and Formation of Acyl-CoA Thioesters”).

Some prodrugs, such as fosphenytoin (Cerebyx^®) and fosamprenavir (Lexiva^®), are designed to be hydrolyzed by alkaline phosphatase, high concentrations of which are present on the luminal surface of the enterocytes lining the wall of the small intestine. Hydrolysis of these prodrugs by alkaline phosphatase releases the active drug at the surface of the enterocytes, where it can be readily absorbed. Soluble epoxide hydrolase (sEH) is a bifunctional enzyme; its C-terminus contains an epoxide hydrolase domain, whereas its N-terminus contains a phosphatase domain, as described in the section “Epoxide Hydrolases.” Although it may play a role in the hydrolysis of endogenous phosphates, such as polyisoprenyl phosphates and lysophosphatidic acids (Oguro and Imaoka, 2012), sEH is an intracellular enzyme and, as such, can play no significant role in the hydrolysis of phosphate/phosphonate prodrugs.

As a result of their ability to hydrolyze prodrugs, hydrolytic enzymes may have clinical applications in the treatment of certain cancers. They might be used, for example, to activate prodrugs in vivo and thereby generate potent anticancer agents in highly selected target sites (eg, at the surface of tumor cells, or inside the tumor cells themselves). For example, carboxylesterases might be targeted to tumor sites with hybrid monoclonal antibodies (ie, bifunctional antibodies that recognize the carboxylesterase and the tumor cell), or the cDNA encoding a carboxylesterase might be targeted to the tumor cells via a viral vector. In the case of irinotecan, this therapeutic strategy would release the anticancer drug SN-38 in the vicinity of the tumor cells, which would reduce the systemic levels and side effects of this otherwise highly toxic drug (Senter et al., 1996). Some prodrugs, such as capecitabine, are activated by hydrolytic enzymes in the tumors themselves.

Peptidases

With the advent of recombinant DNA technology, numerous human peptides have been mass-produced for use as therapeutic agents, and several recombinant peptide hormones, growth factors, cytokines, soluble receptors, and humanized monoclonal antibodies currently are used clinically. To avoid acid precipitation and proteolytic degradation in the gastrointestinal tract, peptides are administered parenterally. Nevertheless, peptides are hydrolyzed in the blood and tissues by a variety of peptidases, including aminopeptidases and carboxypeptidases, which hydrolyze amino acids at the N- and C-terminus, respectively, and endopeptidases, which cleave peptides at specific internal sites (trypsin, for example, cleaves peptides on the C-terminal side of arginine or lysine residues) (Humphrey and Ringrose, 1986; Testa and Krämer, 2008, 2010). Peptidases cleave the amide linkage between adjacent amino acids; hence, they function as amidases. As in the case of carboxylesterases, the active site of peptidases contains either a serine or cysteine residue, which initiates a nucleophilic attack on the carbonyl moiety of the amide bond. As previously noted, the mechanism of catalysis by serine proteases, such as chymotrypsin, is similar to that by serine esterases (B-esterases).

β-Glucuronidase

β-Glucuronidase is present in liver lysosomes and microsomes (where it is bound to the lumen of the endoplasmic reticulum by egasyn [Ces1e], as mentioned in the section “Carboxylesterases”) and in gut microflora. The enzyme hydrolyzes xenobiotic glucuronides (in which the glucuronide is in the β-configuration). When a drug is glucuronidated directly and excreted in bile, hydrolysis by β-glucuronidase in the gut can release the aglycone (the parent drug) and result in a second phase of drug absorption, a process called enterohepatic circulation. In the case of the anticancer drug irinotecan, which is hydrolyzed to SN-11 (the pharmacologically active and toxic metabolite) and then glucuronidated, hydrolysis of SN-11-glucuronide by microflora β-glucuronidase is undesirable because it releases SN-11 in the colon and causes dose-limiting diarrhea. To reduce the risk of such an adverse event, an inhibitor of gut β-glucuronidase has been developed, which is described in more detail in the section “Glucuronidation and Formation of Acyl-CoA Thioesters.”

Epoxide Hydrolases

Epoxide hydrolases catalyze the trans-addition of water to alkene epoxides and arene oxides (oxiranes), which can form during the CYP-dependent oxidation of aliphatic alkenes and aromatic hydrocarbons, respectively. As shown in Fig. 6-8, the products of this hydrolysis are vicinal diols with a trans-configuration (ie, trans-1,2-dihydrodiols), a notable exception being the conversion of leukotriene A₄ (LTA₄) to leukotriene B₄ (LTB₄), in which case the 2 hydroxyl groups that result from epoxide hydrolysis appear on nonadjacent carbon atoms. Epoxide hydrolases play an important role in detoxifying electrophilic epoxides that might otherwise bind to proteins and nucleic acids and cause cellular toxicity and genetic mutations. In the case of PAHs, however, microsomal epoxide hydrolase plays a critical role in forming diol epoxides, the ultimate carcinogenic metabolites of benzo[a]pyrene (B[a]P), 7,12-dimethylbenz[a]anthracene (DMBA), and many other PAHs (discussed later in this section).

There are 4 distinct forms of epoxide hydrolase in mammals: microsomal epoxide hydrolase (mEH; gene name EPHX1), soluble epoxide hydrolase (sEH; gene name EPHX2), cholesterol-5,6-epoxide hydrolase (ChEH; gene yet to be characterized), and leukotriene A₄ hydrolase (LTA₄ hydrolase; gene name LTA4H) (Fretland and Omiecinski, 2000; Morisseau and Hammock, 2005). Hepoxilin A₃ hydrolase was thought to represent a fifth class of epoxide hydrolase but it was subsequently identified as being the same enzyme as sEH (Cronin et al., 2011). As their names imply, cholesterol-5,6-epoxide hydrolase (ChEH) and LTA₄hydrolase hydrolyze endogenous epoxides specifically, and have virtually no capacity to detoxify xenobiotic oxides. LTA₄ hydrolase is distinct from the other epoxide hydrolases because it is a bifunctional zinc metalloenzyme that has both epoxide hydrolase and peptidase activity, and because the 2 hydroxyl groups introduced during the conversion of LTA₄to LTB₄ are 8 carbon atoms apart. sEH is a bifunctional enzyme; its C-terminus contains an epoxide hydrolase domain, whereas its N-terminus contains a phosphatase domain. The latter domain is structurally related to members of haloacid dehalogenase (HAD) superfamily, which includes dehalogenases, phosphonatases, phosphomutases, phosphatases, and ATPases. The phosphatase domain of sEH is thought to play a role in the hydrolysis of endogenous phosphates, such as polyisoprenyl phosphates (which regulate cholesterol levels) and lysophosphatidic acids (Imig and Hammock, 2009; Oguro and Imaoka, 2012).

mEH hydrolyzes a wide variety of xenobiotics with an alkene epoxide or arene oxide. sEH hydrolyzes some xenobiotic epoxides and oxides, such as trans-stilbene oxide, but it also plays an important role in the hydrolysis of endogenous fatty acid epoxides, such as the epoxyeicosatrienoic acids (EETs) that are formed by epoxidation of arachidonic acid by CYP (particularly CYP2J2 and CYP2C9) and the leukotoxins that are formed by the epoxidation of linoleic acid by leukocytes (Fretland and Omiecinski, 2000; Morisseau and Hammock, 2005; Imig and Hammock, 2009). EETs are endothelin-derived hyperpolarizing factors (EDHFs) (ie, vasodilators) that possess anti-inflammatory properties and protect tissues from ischemic injury. Hydrolysis of EETs by sEH terminates their vasodilatory and anti-inflammatory effects. Accordingly, sEH is a potential therapeutic target for the treatment of various cardiovascular diseases such as hypertension and atherosclerosis (Imig and Hammock, 2009; Wang et al., 2010c). Several disubstituted ureas (including the anticancer drug sorafenib) have been identified as competitive inhibitors of sEH that are both potent (K_i values in the nanomolar range) and specific (they do not inhibit the phosphatase domain of sEH and they do not potently inhibit mEH). In addition to their vasodilatory and anti-inflammatory effects, EETs promote endothelial cell proliferation and migration; hence, they are angiogenic (ie, they stimulate blood vessel supply). A potential side effect of sEH inhibitors, therefore, may be potentiation of EET-mediated angiogenic resulting in accelerated tumorigenesis (Imig and Hammock, 2009; Wang et al., 2010c).

Although the levels vary from one tissue to the next, mEH has been found in the microsomal fraction (and in some cases the plasma membrane) of virtually all tissues, including the liver, testis, ovary, lung, kidney, skin, intestine, colon, spleen, thymus, brain, and heart. sEH is also widely distributed in tissues; high levels of sEH are present in the cytosol (and in some cases the lysosomes) of liver, kidney, brain, and vasculature and lower levels are present in lung, spleen, and testis. In general, mEH prefers monosubstituted epoxides and disubstituted epoxides with a cis configuration, such as cis-stilbene oxide, whereas sEH prefers tetrasubstituted and trisubstituted epoxides and disubstituted epoxides with a gem configuration (both substituents on the same carbon atom) or the trans configuration, such as trans-stilbene oxide, as shown in Fig. 6-9. In rodents, both sEH and mEH are inducible enzymes; sEH is under the control of PPARα, so it is induced following treatment of rats and mice with peroxisome proliferators, whereas mEH is under the control of Nrf2, so it is induced in response to oxidative stress or exposure to electrophiles and GSH depletors (see the section “Quinone Reduction—NQO1 and NQO2”). Treatment of mice with the CAR agonist phenobarbital induces mEH about 2- to 3-fold, whereas treatment with Nrf2 activators such as butylated hydroxytoluene (BHT), butylated hydroxyanisole (BHA), and ethoxyquin induces mEH by an order of magnitude or more. mEH is one of several proteins (so-called preneoplastic antigens) that are overexpressed in chemically induced foci and nodules that eventually develop into liver tumors.

Many epoxides and oxides are intermediary metabolites formed during the CYP-dependent oxidation of unsaturated aliphatic and aromatic xenobiotics. These electrophilic metabolites might otherwise bind to proteins and nucleic acids and cause cellular toxicity and genetic mutations. In general, sEH and mEH are found in the same tissues and cell types that contain CYP. For example, the distribution of epoxide hydrolase parallels that of CYP in liver, lung, and testis. In other words, both enzymes are located in the centrilobular region of the liver (zone 3), in Clara and type II cells in the lung, and in Leydig cells in the testis. The colocalization of epoxide hydrolase and CYP presumably ensures the rapid detoxication of alkene epoxides and arene oxides generated during the oxidative metabolism of xenobiotics.

Electrophilic epoxides and arene oxides are constantly produced during the CYP-dependent oxidation of unsaturated aliphatic and aromatic xenobiotics, and are potentially reactive to cellular macromolecules such as DNA and protein. Epoxide hydrolase can rapidly convert these potentially toxic metabolites to the corresponding dihydrodiols, which are less reactive and easier to excrete. Thus, epoxide hydrolases are widely considered as a group of detoxication enzymes. In some cases, however, further oxidation of a dihydrodiol can lead to the formation of diol epoxide derivatives that are no longer substrates for epoxide hydrolase because the oxirane ring is protected by bulky substituents that sterically hinder interaction with the enzyme. This point proved to be extremely important in elucidating the mechanism by which PAHs cause tumors in laboratory animals (Conney, 1982). Tumorigenic PAHs, such as B[a]P, are converted by CYP (particularly by CYP1B1 and CYP1A1) to a variety of arene oxides that bind covalently to DNA, making them highly mutagenic to bacteria. One of the major arene oxides formed from B[a]P, namely, the 4,5-oxide, is highly mutagenic to bacteria but weakly mutagenic to mammalian cells. This discrepancy reflects the rapid inactivation of B[a]P 4,5-oxide by epoxide hydrolase in mammalian cells. However, one of the arene oxides formed from B[a]P, namely, B[a]P 7,8-dihydrodiol-9,10-oxide, is not a substrate for epoxide hydrolase and is highly mutagenic to mammalian cells and considerably more potent than B[a]P as a lung tumorigen in mice.

B[a]P 7,8-dihydrodiol-9,10-oxide is known as a bay-region diol epoxide, and analogous bay-region diol epoxides are now recognized as tumorigenic metabolites of numerous PAHs. A feature common to all bay-region epoxides is their resistance to hydrolyation by mEH, which results from steric hindrance from the nearby dihydrodiol group. As shown in Fig. 6-10, B[a]P 7,8-dihydrodiol-9,10-oxide is formed in 3 steps: B[a]P is converted to the 7,8-oxide, which is converted to the 7,8-dihydrodiol, which is converted to the corresponding 9,10-epoxide (which can exist in 4 diastereomeric forms). The first and third steps are epoxidation reactions catalyzed by CYP (especially CYP1B1 and CYP1A1) or prostaglandin H synthase, but the second step is catalyzed by mEH. Consequently, even though mEH plays a major role in detoxifying several B[a]P, such as the 4,5-oxide, it nevertheless plays a role in converting B[a]P to its ultimate tumorigenic metabolite, B[a]P 7,8-dihydrodiol-9,10-oxide. Diol epoxides are also the carcinogenic metabolites of DMBA and many other PAHs (Shimada, 2006).

The importance of mEH in the conversion of PAHs to their ultimate carcinogenic metabolites, namely, diol epoxides, is illustrated by the observation that mEH knockout mice (mEH-null mice) are completely resistant to the tumorigenic effects of DMBA (Shimada, 2006). Genetic polymorphisms of human mEH also impact cigarette-smoking-induced cancer of the lung and upper aerodigestive tract (UADT) in a manner consistent with the protective effect conferred by a lack of mEH in DMBA-treated mice (Li et al., 2011b). More than 110 SNPs have been identified in the human EPHX1 gene, 2 of which have been studied in detail: One of them (Tyr₁₁₃His) decreases mEH activity (by about 40%) and the other (His₁₃₉Arg) increases mEH activity (by about 25%). The low-activity mEH (observed in individuals who are homozygous or heterozygous for the His₁₃₃ allelozyme) protects from tobacco-induced cancer of the lung and UADT, whereas the high-activity variant (observed in individuals who are homozygous or heterozygous for the Arg₁₃₉ allelozyme) predisposes to these cancers (Li et al., 2011b).

Not all epoxides are highly reactive and toxic to the cells that produce them. Some drugs actually contain an epoxide, such as scopolamine, tiotropium, and troleandomycin. Vitamin K epoxide is also a nontoxic epoxide, which is formed and consumed during the vitamin K–dependent γ-carboxylation of prothrombin and other clotting factors in the liver. Vitamin K epoxide is not hydrated by mEH but is reduced by vitamin K epoxide reductase. This enzyme is inhibited by warfarin and related coumarin anticoagulants, which interrupts the synthesis of several clotting factors. The major metabolite of carbamazepine is an epoxide, which is so stable that carbamazepine 10,11-epoxide is a major circulating metabolite in patients treated with this antiepileptic drug. (Carbamazepine is converted to a second epoxide, which is less stable and more cytotoxic, as shown in the section “Cytochrome P450.”) Circulating levels of carbamazepine 10,11-epoxide are elevated in individuals expressing the low-activity variant of mEH (the His₁₃₃ allelozyme) but, with one possible exception, this genetic polymorphism is not associated with an increase in the adverse effects of this or other anticonvulsant drugs (Daly, 1999). The exception is a case report of a man who had a defect in mEH expression and suffered acute and severe phenytoin toxicity (Morisseau and Hammock, 2005). Certain drugs, such as valpromide (the amide analog of valproic acid) and progabide (a γ-aminobutyric acid [GABA] agonist), cause clinically significant inhibition of mEH and may impair epoxide hydrolase activity more than genetic polymorphisms. These 2 drugs potentiate the neurotoxicity of carbamazepine by inhibiting mEH, leading to increased plasma levels of carbamazepine 10,11-epoxide and presumably its more toxic 2,3-epoxide (Kroetz et al., 1993). mEH can be inhibited in vitro by certain epoxides, such as 1,1,1-trichloropropene oxide and cyclohexene oxide, and stimulated by several alcohols, ketones, and imidazoles.

The microsomal and soluble forms of epoxide hydrolase show no evident sequence identity and, accordingly, are immunochemically distinct proteins (Beetham et al., 1995). Nevertheless, mEH and sEH catalyze reactions by the same mechanism, and similar amino acids are involved in catalysis, namely, a nucleophilic acid (Asp₂₂₆ in mEH and Asp₃₃₄ in sEH), a basic histidine (His₄₃₁ in mEH and His₅₂₃ in sEH), an orienting acid (Glu₄₀₄ in mEH and Asp₄₉₅ in sEH), and polarizing tyrosine residues (Tyr₂₉₉ and Tyr₃₇₄ in mEH and Tyr₃₈₂ and Tyr₄₆₅ in mEH) (Morisseau and Hammock, 2005). The mechanism of catalysis by epoxide hydrolase is similar to that of carboxylesterase, in that the catalytic site comprises 3 amino acid residues that form a catalytic triad, as shown in Fig. 6-7. The attack of the nucleophile Asp₂₂₆ on the carbon of the oxirane ring initiates enzymatic activity, leading to the formation of an α-hydroxyester-enzyme intermediate, with the negative charge developing on the oxygen atom stabilized by a putative oxyanion hole. The His₄₃₁ residue (which is activated by Glu₃₇₆ and Glu₄₀₄) activates a water molecule by abstracting a proton (H⁺). The activated (nucleophilic) water then attacks the Cγ atom of Asp₂₂₆, resulting in the hydrolysis of the ester bond in the acyl–enzyme intermediate, which restores the active enzyme and results in formation of a vicinal diol with a trans-configuration (Armstrong, 1999). The second step, namely, cleavage of the ester bond in the acyl–enzyme intermediate, resembles the cleavage of the ester or amide bond in substrates for serine esterases and proteases.

Although both epoxide hydrolase and carboxylesterase have a catalytic triad comprising a nucleophilic, basic, and acidic amino acid residue, there are striking differences in their catalytic machinery, which account for the fact that carboxylesterases primarily hydrolyze esters and amides, whereas epoxide hydrolases primarily hydrolyze epoxides and oxides. In the triad, both enzymes have histidine as the base and either glutamate or aspartate as the acid, but they differ in the type of amino acids for the nucleophile. Even during catalysis, there is a major difference. In carboxylesterases, the same carbonyl carbon atom of the substrate is attacked initially by the nucleophile Ser₂₀₃ to form an α-hydroxyester-enzyme ester that is subsequently attacked by the activated water to release the alcohol product. In contrast, 2 different atoms in epoxide hydrolase are targets of nucleophilic attacks. First the less hindered carbon atom of the oxirane ring is attacked by the nucleophile Asp₂₂₆ to form a covalently bound ester, and next this ester is hydrolyzed by an activated water molecule that attacks the Cγ atom of the Asp₂₂₆ residue, as illustrated in Fig. 6-7. Therefore, in carboxylesterase, the oxygen introduced to the product is derived from the activated water molecule. In contrast, in epoxide hydrolase, the oxygen introduced to the product is derived from the nucleophile Asp₂₂₆ (Fig. 6-7).

Carboxylesterases and epoxide hydrolases exhibit no primary sequence identity, but they share surprising similarities in the topology of the structure and sequential arrangement of the catalytic triad. Both are members of the α/β-hydrolase fold enzymes, a superfamily of proteins that includes lipases, esterases, and haloalkane dehydrogenases (Beetham et al., 1995). Functionally, proteins in this superfamily all catalyze hydrolytic reactions; structurally, they all contain a similar core segment that is composed of 8 β-sheets connected by α-helices. They all have a catalytic triad and the arrangement of the amino acid residues in the triad (ie, the order of the nucleophile, the acid, and the base in the primary sequence) is the mirror image of the arrangement in other hydrolytic enzymes such as trypsin. All 3 active-site residues are located on loops that are the best conserved structural features in the fold, which likely provides catalysis with certain flexibility to hydrolyze numerous structurally distinct substrates.

Reduction

Certain metals (eg, pentavalent arsenic) and xenobiotics containing an aldehyde, ketone, alkene, disulfide, sulfoxide, quinone, N-oxide, hydroxamic acid, amidoxime, isoxazole, isothiazole, azo, or nitro group are often reduced in vivo, although it is sometimes difficult to ascertain whether the reaction proceeds enzymatically or nonenzymatically by interaction with reducing agents (such as the reduced forms of glutathione, FAD, FMN, and NAD[P]). Some of these functional groups can be either reduced or oxidized. For example, aldehydes (R-CHO) can be reduced to an alcohol (R-CH₂OH) or oxidized to a carboxylic acid (R-COOH), whereas sulfoxides (R₁-SO-R₂) can be reduced to a sulfide (R₁-S-R₂) or oxidized to a sulfone (R₁-SO₂-R₂). Likewise, some enzymes, such as alcohol dehydrogenase (ADH), aldehyde oxidase (AO), and CYP, can catalyze both reductive and oxidative reactions depending on the substrate or conditions (eg, aerobic vs anaerobic). In the case of halogenated hydrocarbons, such as halothane, dehalogenation can proceed by an oxidative or reductive pathway, both of which are catalyzed by the same enzyme (namely, CYP). In some cases, such as azo-reduction, nitro-reduction, and the reduction of certain alkenes, the reaction is largely catalyzed by intestinal microflora.

Azo- and Nitro-Reduction

Prontosil and chloramphenicol are examples of drugs that undergo azo- and nitro-reduction, respectively, as shown in Fig. 6-11. Reduction of prontosil is of historical interest. Treatment of streptococcal and pneumococcal infections with prontosil marked the beginning of specific antibacterial chemotherapy. Subsequently, it was discovered that the active drug was not prontosil but its metabolite, sulfanilamide (para-aminobenzene sulfonamide), a product of azo-reduction. During azo-reduction, the nitrogen–nitrogen double bond is sequentially reduced and cleaved to produce two primary amines, a reaction requiring four reducing equivalents. Nitro-reduction requires six reducing equivalents, which are consumed in three sequential reactions, as shown in Fig. 6-11 for the conversion of nitrobenzene to aniline.

Azo- and nitro-reduction reactions are generally catalyzed by intestinal microflora. However, under certain conditions, such as low oxygen tension, the reactions can be catalyzed by liver microsomal CYP and NAD(P)H-quinone oxidoreductase (NQO1, a cytosolic flavoprotein that is also known as DT-diaphorase) and, in the case of nitroaromatics, by cytosolic AO. The anaerobic environment of the lower gastrointestinal tract is well suited for azo- and nitro-reduction, which is why intestinal microflora contributes significantly to these reactions. The reduction of quinic acid to benzoic acid is another example of a reductive reaction catalyzed by gut microflora, as shown in Fig. 6-1.

Nitro-reduction by intestinal microflora is thought to play an important role in the toxicity of several nitroaromatic compounds including 2,6-dinitrotoluene, which is hepatotumorigenic to male rats. The role of nitro-reduction in the metabolic activation of 2,6-dinitrotoluene is shown in Fig. 6-12 (Long and Rickert, 1982; Mirsalis and Butterworth, 1982). The biotransformation of 2,6-dinitrotoluene begins in the liver, where it is oxidized by CYP and conjugated with glucuronic acid. This glucuronide is excreted in bile and undergoes biotransformation by intestinal microflora. One or both of the nitro groups are reduced to amines by nitro-reductase, and the glucuronide is hydrolyzed by β-glucuronidase. The reduced/deconjugated metabolites are absorbed and transported to the liver, where the newly formed amine group is N-hydroxylated by CYP and conjugated with acetate or sulfonate. These conjugates form good leaving groups, which renders the nitrogen highly susceptible to nucleophilic attack from proteins and DNA; this ostensibly leads to mutations and the formation of liver tumors. Compared with females, male rates are more susceptible to hepatotumorigenicity of 2,6-dinitrotoluene due to their higher rate of bile secretion and therefore their higher rate of biliary excretion of 2,6-dinitrobenzylalcohol glucuronide. The complexity of the metabolic scheme shown in Fig. 6-12 underscores an important principle, namely, that the activation of some chemical tumorigens to DNA-reactive metabolites involves several different biotransforming enzymes and may take place in more than one tissue. Consequently, the ability of 2,6-dinitrotoluene to bind to DNA and cause mutations is not revealed in most of the short-term assays for assessing the genotoxic potential of chemical agents. These in vitro assays for genotoxicity do not make allowance for biotransformation by intestinal microflora or, in some cases, the conjugating enzymes.

Nitro-reduction by intestinal microflora also plays an important role in the biotransformation of musk xylene (1,3,5-trinitro-2-tbutyl-4,6-dimethylbenzene). Reduction of one or both of the nitro groups is required for musk xylene to induce (as well as markedly inhibit) liver microsomal CYP (namely, CYP2B) in rodents (Lehman-McKeemanm et al., 1999).

Carbonyl Reduction—AKRs and SDRs

A variety of xenobiotics contain a carbonyl function (R-CHO and R₁-CO-R₂) that undergoes reduction in vivo. The reduction of aldehydes to primary alcohols and of ketones to secondary alcohols is generally catalyzed in mammals by NAD(P)H-dependent reductases belonging to one of several superfamilies, the aldo keto reductases (AKRs), the short-chain dehydrogenases/reductases (SDRs), the medium-chain dehydrogenases/reductases (MDRs), ALDHs, and NAD(P)H-quinone oxido reductases (NQO) as shown in Table 6-5 (Jez and Penning, 2001; Oppermann et al., 2001; Matsunaga et al., 2006; Malatkova et al., 2010; Skarydova and Wsol, 2011). The AKRs are members of a superfamily of cytosolic enzymes that reduce both xenobiotic and endobiotic compounds, as their alternative names imply (Table 6-5). Dimeric dihydrodiol dehydrogenase and various members of the AKR superfamily, functioning as dihydrodiol dehydrogenases, can oxidize the trans-dihydrodiols of various polycyclic aromatic hydrocarbon oxiranes (formed by epoxide hydrolase) to the corresponding ortho-quinones, as shown previously in Fig. 6-10. The role of AKR as an oxidizing enzyme is discussed in the section “Dimeric Dihydrodiol Dehydrogenase.” As mentioned earlier (see Point 9 in the section “Introduction”), one of the AKRs, namely, AKR7A (also known as aflatoxin aldehyde reductase), is one of the many enzymes induced following activation of Nrf2 by oxidative stress, exposure to electrophiles, or depletion of GSH. Humans contain at least 71 SDR members, three of which, namely, cytosolic carbonyl reductases (CBR1 and CBR3) and a microsomal carbonyl reductase (HSD11B1), play a role in the reduction of a wide variety of carbonyl-containing xenobiotics (other species express more than two carbonyl reductases) (see http://www.sdr-enzymes.org). Erythrocytes also contain carbonyl reductase, which contributes significantly to the reduction of haloperidol, as shown in Fig. 6-13. From the alternative names given in Table 6-5, it is apparent that the both cytosolic and microsomal carbonyl reductases have been studied for their role in endobiotic metabolism, namely, the reduction of prostaglandin derivatives and 11β-hydroxysteroids, respectively (Kavanagh et al., 2008).

In certain cases, the reduction of aldehydes to alcohols can be catalyzed by ADH, as shown in Fig. 6-13 for the conversion of the sedative–hypnotic, chloral hydrate, to trichloroethanol. As shown in Table 6-5, ADHs belong to the medium chain dehydrognases/reductases (MDRs). They typically convert alcohols to aldehydes, for which reason they are discussed later in the section “Alcohol Dehydrogenase.” In the case of chloral hydrate, the reverse reaction is favored by the presence of the trichloromethyl group, which is a strong electron-withdrawing group.

The SDR carbonyl reductases are monomeric, NADPH-dependent enzymes present in erythrocytes and both the cytosolic and microsomal fractions of the liver, kidney, brain, and many other tissues. The major circulating metabolite of the antipsychotic drug, haloperidol, is a secondary alcohol formed by carbonyl reductases in the blood and liver, as shown in Fig. 6-13. Other xenobiotics that are reduced by carbonyl reductases include pentoxifylline (see Fig. 6-3), acetohexamide, daunorubicin, doxorubicin, loxoprofen, menadione, 4-nitroacetophenone, timiperone, and R-warfarin (Rosemond and Walsh, 2004). As shown in Fig. 6-3, the reduction of ketones to secondary alcohols by carbonyl reductases may proceed with a high degree of stereoselectivity, as in the case of pentoxifylline (Lillibridge et al., 1996).

Liver cytosol contains at least two carbonyl reductases (CBR1 and CBR3) and microsomes contain at least one other form of carbonyl reductase (HSD11B1), and these can differ in the degree to which they stereoselectively reduce ketones to secondary alcohols. For example, keto-reduction of pentoxifylline produces two enantiomeric secondary alcohols: one with the R-configuration (which is known as lisofylline) and one with the S-configuration, as shown in Fig. 6-3. Reduction of pentoxifylline by cytosolic carbonyl reductases results in the stereospecific formation of the optical antipode of lisofylline, whereas the same reaction catalyzed by microsomal carbonyl reductase produces both lisofylline and its optical antipode in a ratio of about 1 to 5 (Lillibridge et al., 1996).

The well-known microsomal carbonyl reductase is 11β-hydroxysteroid dehydrogenase (gene symbol HSD11B1, aka SDR26C1) (Skarydova and Wsol, 2011). However, there are several other microsomal carbonyl-reducing enzymes, such as the six human retinol dehydrogenases (eg, RDHs) and the six human 17β-hydroxysteroid dehydrogenases. Although these enzymes have not been well characterized for their ability to reduce carbonyl-containing xenobiotics, there is indirect evidence from the ratio of enantiomers of metabolites that multiple microsomal enzymes may play a role in the metabolism of some xenobiotics such as 7α-methyl-19-nortestotsterone (Skarydova and Wsol, 2011). There are at least two monomeric cytosolic carbonyl reductases (gene symbols CBR1 [aka SDR21C1] and CBR3 [aka SDR21C2]) that have broad, and somewhat overlapping, substrate specificities toward many endobiotics and xenobiotics such as menadione, isatin, 4-(methylnitrosamino)-1-(3-pyridyl)-1-butanone (NNK), daunorubicin, and doxorubicin (Malatkova et al., 2010). A human tetrameric mitochondrial carbonyl reductase, namely, CBR4 (aka SDR45C1), has also been identified with activity toward 9,10-phenanthrenequinone and 1,4-benzoquinone, and therefore acts as a mitochondrial quinone reductase (Malatkova et al., 2010).

In rat liver cytosol, the reduction of quinones is primarily catalyzed by NQO1 and NQO2 (see the section “Quinone Reduction—NQO1 and NQO2”), whereas in human liver cytosol, quinone reduction is catalyzed by both NQO and carbonyl reductases.

Various members of the AKR superfamily have been implicated in the reduction of such carbonyl-containing xenobiotics as the tobacco-specific nitrosamine NNK, acetohexamide, daunorubicin, naloxone, naltrexone, befunolol, ethacrynic acid, ketoprofen, ketotifen, haloperidol, loxoprofen, metyrapone, oxo-nortryptyline, and numerous aromatic and aliphatic aldehydes (Rosemond and Walsh, 2004). Putative AKR1A, 1B, 1C1 to 4, or 1D1-selective inhibitors, statil, flufenamic acid, phenolphthalein, and finasteride, respectively, can aid in the evaluation of which reductive enzyme(s) is(are) involved in the formation of specific metabolites. Many of the xenobiotics reduced by AKRs are also reduced by SDRs, and in most cases the relative contribution of individual carbonyl-reducing enzymes is not known. Genetic polymorphisms of human AKR1C3, AKR1C4, AKR7A2, and cytosolic carbonyl reductase (CBR1) have been associated with decreased metabolism of carbonyl-containing xenobiotics such as doxorubicine and daunorubicine in vitro (Gonzalez-Covarrubias et al., 2007; Bains et al., 2010). Allelic variants of human cytosolic AKR1C2 with reduced enzymatic activity toward the androgen 5α-dihydrotestosterone (DHT) have also been identified (Takahashi et al., 2009). There appear to be no reports of clinically significant drug–drug interactions involving the inhibition or induction of AKRs or SDRs (Rosemond and Walsh, 2004).

Disulfide Reduction

Some disulfides are reduced and cleaved to their sulfhydryl components, as shown in Fig. 6-14 for the alcohol deterrent, disulfiram (Antabuse). As shown in Fig. 6-14, disulfide reduction by glutathione is a 3-step process, the last of which is catalyzed by glutathione reductase. The first steps can be catalyzed by glutathione transferase, or they can occur nonenzymatically.

Sulfoxide and N-Oxide Reduction

Thioredoxin-dependent enzymes in liver and kidney cytosol have been reported to reduce sulfoxides, which themselves may be formed by CYP or flavin monooxygenases (Anders et al., 1981). Recycling through these counteracting enzyme systems, a process known as retro-reduction or futile cycling (Hinrichs et al., 2011), may prolong the half-life of certain xenobiotics. As shown in Fig. 6-15 sulindac is a sulfoxide that undergoes reduction to a sulfide, which is excreted in bile and reabsorbed from the intestine (Ratnayake et al., 1981). This enterohepatic cycling prolongs the duration of action of the drug such that this NSAID need only be taken twice daily. In human liver, glutaredoxin (GLRX) and thioredoxin may also be involved in reducing the mixed disulfide formed between xenobiotics and glutathione, as is the case in the formation of the pharmacologically active metabolite of the P2Y₁₂ inhibitor prasugrel (Hagihara et al., 2011).

Sulfoxide reduction may also occur nonenzymatically at an appreciable rate, as in the case of the proton pump inhibitor rabeprazole (Miura et al., 2006). Diethyldithiocarbamate methyl ester, a metabolite of disulfiram, is oxidized to a sulfine, which is reduced to the parent methyl ester by glutathione. In the latter reaction, two molecules of gluathione are oxidized with reduction of the sulfine oxygen to water (Madan et al., 1994) as shown below:

Just as sulfoxide reduction can reverse the effect of sulfoxidation, so the reduction of N-oxides can reverse the N-oxygenation of amines, which is catalyzed by flavin monooxygenases and CYP. Under reduced oxygen tension, reduction of the N-oxides of imipramine, tiaramide, indicine, and N,N-dimethylaniline can be catalyzed by mitochondrial and/or microsomal enzymes in the presence of NADH or NADPH (Sugiura and Kato, 1977). The NADPH-dependent reduction of N-oxides in liver microsomes appears to be catalyzed by CYP (Sugiura et al., 1976), although in some cases NADPH-cytochrome P450 reductase may play an important role.

As a class, N-oxides are not inherently toxic compounds. However, certain aromatic and aliphatic N-oxides have been exploited as bioreductive drugs (also known as DNA-affinic drugs) for the treatment of certain cancers and infectious diseases (Wardman et al., 1995). In these cases, N-oxides have been used as prodrugs that are converted to cytotoxic or DNA-binding drugs under hypoxic conditions. The fact that N-oxides of certain drugs are converted to toxic metabolites under hypoxic conditions is the basis for their selective toxicity to certain solid tumors (namely, those that are hypoxic and, hence, resistant to radiotherapy) and anaerobic bacteria. For example, tirapazamine (SR 4233) is a benzotriazine di-N-oxide that is preferentially toxic to hypoxic cells, such as those present in solid tumors, apparently due to its rapid activation by one-electron reduction of the N-oxide to an oxidizing nitroxide radical, as shown in Fig. 6-15 (Walton et al., 1992). This reaction is catalyzed by CYP and NADPH-cytochrome P450 reductase (Saunders et al., 2000). Two-electron reduction of the di-N-oxide, SR 4233, produces a mono-N-oxide, SR 4317, which undergoes a second N-oxide reduction to SR 4330. Like SR 4233, the antibacterial agent, quindoxin, is a di-N-oxide whose cytotoxicity is dependent on reductive activation, which is favored by anaerobic conditions. AQ4N is a di-N-oxide prodrug that is converted by N-oxide reduction to the potent topoisomerase II inhibitor 4QA (1,4-bis{[2-(dimethylamino)ethyl]amino}-5,8-dihydroxy-anthracene-9,10-dione) (Nishida et al., 2010). The reductive reaction is catalyzed by CYP2S1 and CYP2W1, two hypoxia-inducible CYP enzymes expressed in many solid tumors. The induction of CYP2S1 and CYP2W1 in hypoxic tumor cells provides a basis for their application in the selective bioreductive activation of antineoplastic drugs (Nishida et al., 2010).

Bioreductive alkylating agents, which include such drugs as mitomycins, anthracyclins, and aziridinylbenzoquinones, represent another class of anticancer agents that require activation by reduction. However, for this class of agents, bioactivation also involves a two-electron reduction reaction, which is largely catalyzed by NQO, which is described in the next section.

Quinone Reduction—NQO1 and NQO2

Quinones can be reduced to hydroquinones by two closely related, cytosolic flavoproteins, namely, NQO1 and NQO2. The former enzyme, NAD(P)H-quinone oxidoreductase-1, is also known as DT-diaphorase. The latter enzyme, NAD(P)H-quinone oxidoreductase-2, is also known as NRH-quinone oxidoreductase because it prefers the unusual electron donor dihydronicotinamide riboside (NRH) over NAD(P)H. Although they are closely related enzymes (both contain two 27-kDa subunits each with an FAD prosthetic group), NQO1 and NQO2 have different substrate specificities, and they can be distinguished on the basis of their differential inhibition by dicoumarol and quercetin (which are selective inhibitors of NQO1 and NQO2, respectively). NQO2 may have a physiological role in the metabolism of vitamin K hydroquinone (Chen et al., 2000).

An example of the type of reaction catalyzed by NQO is shown in Fig. 6-16. Formation of the hydroquinone involves a two-electron reduction of the quinone with stoichiometric oxidation of NAD[P]H without oxygen consumption. (The two-electron reduction of certain quinones can also be catalyzed by carbonyl reductase, especially in humans.) In contrast, NADPH-cytochrome P450 reductase, a microsomal flavoprotein, catalyzes the one-electron reduction of quinones to semiquinone radicals that, in addition to being reactive metabolites themselves, cause oxidative stress by reacting with oxygen to form reactive oxygen species, which leads to nonstoichiometric oxidation of NADPH and oxygen consumption, as shown in Fig. 6-16. The two-electron reduction of quinones is a nontoxic reaction—one that is not associated with semiquinone formation and oxidative stress—provided the resultant hydroquinone is sufficiently stable to undergo glucuronidation or sulfonation. However, there are quinone-containing xenobiotics that, despite undergoing two-electron reduction by NQO, produce semiquinone free radicals, oxidative stress, DNA damage, and cytotoxicity. Many of these xenobiotics are being developed as anticancer drugs because NQO1 is often overexpressed in tumor cells. The impact of the null allele NQO1*2 (discussed below in this section) on the sensitivity and resistance to antitumor quinones remains to be established (Siegel et al., 2011). The properties of the hydroquinone determine whether, during the metabolism of quinone-containing xenobiotics, NQO functions as a protective antioxidant or a pro-oxidant activator leading to the formation of reactive oxygen species and reactive semiquinone free radicals. The latter are thought to form not from the one-electron reduction of the quinone but from the two-electron reduction of the quinone (Q) to the hydroquinone (QH₂), which then undergoes one-electron oxidation or perhaps disproportionation to form the reactive semiquinone (QH):

QH₂ + Q ↔ 2QH.

Drugs or drug candidates that are activated by NQO to anticancer agents include the aziridinylbenzoquinone diaziquone, the anthraquinone mitoxantrone, the indolquinones mitomycin C and EO9 (an analog of mitomycin C that is more rapidly reduced by NQO1), and the anthracycline antibiotics daunorubicin and doxorubicin (Gutierrez, 2000). These so-called bioreductive alkylating agents are reduced by NQO1 to generate semiquinone free radicals and other reactive intermediates that undergo nucleophilic additions with DNA, resulting in single-strand DNA breaks. The reason such drugs are preferentially toxic to tumor cells is that tumor cells, especially those in solid tumors, are hypoxic, and hypoxia induces the synthesis of NQO1 (by a mechanism that involves the activator protein 1 [AP-1] and NF-κB response elements in the 5′-promoter region of the NQO1 gene). Therefore, tumor cells often express high levels of NQO1, which predisposes them to the toxic effects of quinone-reductive anticancer drugs such as mitomycin C. Interestingly, mitomycin C also upregulates the expression of NQO1, which may enable this anticancer drug to stimulate its own metabolic activation in tumor cells (Yao et al., 1997). Some cancer chemotherapeutic agents, such as the N-oxide SR 4233 (tirapazamine), are inactivated, not activated, by NQO, as shown in Fig. 6-15.

NQO can activate certain nitroaromatic compounds (R-NO₂) to the corresponding hydroxylamine (R-NHOH), which can be activated by acetylation or sulfonation (by pathways analogous to those shown in Fig. 6-12). Dinitropyrenes and the nitroaromatic compound CB 1954 are activated by NQO. The latter compound was under consideration as an anticancer agent. However, although it is activated by reduction by rat NQO, the nitroaromatic compound CB 1954 is not activated by human NQO.

Oxidative stress is an important component to the mechanism of toxicity of several xenobiotics that either contain a quinone or can be biotransformed to a quinone (Anders, 1985). The production of superoxide anion radicals and oxidative stress are responsible, at least in part, for the cardiotoxic effects of doxorubicin (adriamycin) and daunorubicin (daunomycin), the pulmonary toxicity of paraquat and nitrofurantoin, and the neurotoxic effects of 6-hydroxydopamine. Oxidative stress also plays an important role in the destruction of pancreatic β cells by alloxan and dialuric acid. Tissues, low in superoxide dismutase activity, such as the heart, are especially susceptible to the oxidative stress associated with the redox cycling of quinones. This accounts, at least in part, for the cardiotoxic effects of adriamycin and related anticancer agents, although other susceptibility factors have been proposed (Mordente et al., 2001).

As already mentioned in this section, it is now apparent that the structure of the hydroquinones produced by NQO determines whether the two-electron reduction of quinones results in xenobiotic detoxication or activation. Hydroquinones formed by two-electron reduction of unsubstituted or methyl-substituted 1,4-naphthoquinones (such as menadione) or the corresponding quinone epoxides are relatively stable to autoxidation, whereas the methoxyl, glutathionyl, and hydroxyl derivatives of these compounds undergo autoxidation with production of semiquinones and reactive oxygen species. The ability of glutathionyl derivatives to undergo redox cycling indicates that conjugation with glutathione does not prevent quinones from serving as substrates for NQO. The glutathione conjugates of quinones can also be reduced to hydroquinones by carbonyl reductases, which actually have a binding site for glutathione. In human carbonyl reductase, this binding site is Cys₂₂₇, which is involved in binding both substrate and glutathione (Tinguely and Wermuth, 1999). Although oxidative stress is an important mechanism by which quinones cause cellular damage (through the intermediacy of semiquinone radicals and the generation of reactive oxygen species), it should be noted that quinones are Michael acceptors, and cellular damage can occur through direct alkylation of critical cellular proteins and/or DNA (Bolton et al., 2000).

NQO1 is inducible up to 10-fold by two classes of inducers, which have been categorized as bifunctional and monofunctional inducers (Prochaska and Talalay, 1988). The bifunctional inducers include compounds such as β-naphthoflavone, benzo[a]pyrene, 3-methylcholanthrene, and 2,3,7,8-tetrachlorodibenzo-p-dioxin (TCDD or dioxin), which are AhR agonists that induce both oxidative enzymes (such as the CYP enzyme CYP1A1) and conjugating enzymes (such as GST and UGT). The monofunctional inducers are Nrf2 activators that tend to induce conjugating and other non-CYP enzymes (although in mice, monofunctional inducers can induce CYP2C55 and 2U1, as well as AO). The AhR agonists (the so-called bifunctional inducers) signal through the xenobiotic-response element (XRE), whereas Nrf2 activators (the so-called monofunctional inducers) signal through the antioxidant response element (ARE), which is also known as the electrophilic response element (EpRE). (Response elements are short sequences of DNA, often located in the 5′-promoter region of a gene, that bind the transcription factors that control gene expression.) The monofunctional inducers can be subdivided into two chemical classes: those that activate Nrf2 by causing oxidative stress through redox cycling (eg, the quinone, menadione, and the phenolic antioxidants tert-butylhydroquinone and 3,5-di-tert-butylcatechol) and those that activate Nrf2 by causing oxidative stress by depleting glutathione (eg, fumarates, maleates, acrylates, isothiocyanates, and other Michael acceptors that react with glutathione).

NQO1 is under the control of both AhR and Nrf2. The flavonoid β-naphthoflavone and the PAH B[a]P induce NQO1 (and other enzymes) by both mechanisms; the parent compound activates AhR, whereas electrophilic and/or redox active metabolites activate Nrf2. The situation with B[a]P is quite intriguing. This PAH binds directly to AhR, which binds to XRE and induces the synthesis of CYP1A1 and CYP1B1, which in turn convert B[a]P to electrophilic metabolites (such as arene oxides and diol epoxides) and redox active metabolites (such as catechols), as shown in Fig. 6-10. These electrophilic and redox active metabolites activate Nrf2 and induce various enzymes that protect against oxidative stress. However, the catechol metabolites of B[a]P are further converted by AKRs and/or dimeric dihydrodiol dehydrogenase to ortho-quinones (Fig. 6-10), and are thereby converted back into planar, hydrophobic compounds that are highly effective ligands for the AhR (Burczynski and Penning, 2000). This may be toxicologically important, because the AhR may translocate ortho-quinone metabolites of B[a]P into the nucleus, where they might damage DNA (Bolton et al., 2000).

Sulforaphane and various isothiocyanates are Nrf2 activators that are present in broccoli and are thought to be responsible for the anticarcinogenic effects of this cruciferous vegetable (Zhang et al., 1992). These Nrf2 activators induce GST (GSTA1), mEH, AKR (AKR7A, also known as aflatoxin aldehyde reductase), NQO1 (also known as DT-diaphorase), glutamate cysteine ligase, as well as genes involved in apoptosis. One isothiocyanate in particular, phenethyl isothiocyanate, has been found to activate Nrf2 and activate numerous genes in addition to those encoding xenobiotic-biotransforming enzymes and oxidant defense systems. Microarray studies carried out in wild-type and Nrf2 knockout mice treated with phenethyl isothiocyanate showed that the most highly inducible genes include the very low-density lipoprotein (VLDL) receptor, G-protein signaling modulator 2, early growth response 1, pancreatic lipase-related protein 2, histocompatibility 2 (K region), general transcription factor IIB, myoglobin, potassium voltage-gated channel Q2, and SLC39A10 (Hu et al., 2006). As with other xenosensors, activation of Nrf2 results in a pleiotypic response in which a large number of genes are activated (or repressed). As mentioned above (this section), hypoxia and the anticancer agent mitomycin C are also inducers of NQO1, which has implications for cancer chemotherapy.

NQO1 and NQO2 are polymorphically expressed enzymes, and several lines of evidence suggest that NQO1 and/or NQO2 plays a key role in protecting bone marrow from the hematotoxic effects of benzene or other environmental factors (Iskander and Jaiswal, 2005). In humans, a high percentage of individuals with myeloid and other types of leukemia are homozygous or heterozygous for a null mutant allele of NQO1. This polymorphism, NQO1*2, is a SNP (C₆₀₉T) that changes Pro₁₈₇ to Ser₁₈₇, which destabilizes the protein and targets it for rapid degradation by the ubiquitin proteasomal pathway; this polymorphism is also associated with increased risk of colorectal and esophageal cancers (Ross, 2005; Chen et al., 2012). Mice lacking NQO1 or NQO2 (knockout or null mice) have no developmental abnormalities but have increased granulocytes in the blood and myelogenous hyperplasia of the bone marrow (due to decreased apoptosis). Mice lacking NQO1 are substantially more susceptible than wild-type mice to benzene-induced hematotoxicity (Iskander and Jaiswal, 2005; Ross, 2005). The hematotoxicity of benzene is thought to involve its conversion to hydroquinone in the liver and its subsequent oxidation to benzoquinone by myeloperoxidase in the bone marrow (discussed later in the section “Peroxidase-Dependent Cooxidation”). NQO would be expected to play a role in detoxifying benzoquinone, as predicted, loss of NQO potentiates benzene hematotoxicity. However, loss of NQO also impairs apoptosis, which also represents a plausible explanation for the association between loss of NQO and increased susceptibility to benzene hematotoxicity. The latter mechanism (ie, impaired apoptosis) likely accounts for the observation that NQO1 and NQO2 null mice are more susceptible than wild-type mice to skin carcinogenesis by B[a]P and DMBA, an effect attributable to the diol epoxides, not the quinone metabolites, of these PAHs (Iskander and Jaiswal, 2005).

Dihydropyrimidine Dehydrogenase

In 1993, 15 Japanese patients died as a result of an interaction between two oral medications: sorivudine (a new antiviral drug for herpes zoster) and tegafur (a prodrug that is converted in the liver to the anticancer agent, 5-fluorouracil). The deaths occurred within 40 days of the Japanese government's approval of sorivudine for clinical use. The mechanism of the lethal interaction between sorivudine and tegafur is illustrated in Fig. 6-17, and involves inhibition of dihydropyrimidine dehydrogenase (DPD), an NADPH-requiring, homodimeric protein (Mr ~210 kDa) containing FMN/FAD and an iron–sulfur cluster in each subunit. The enzyme is located mainly in liver cytosol, where it catalyzes the reduction of 5-fluorouracil and related pyrimidines. Sorivudine is converted in part by gut flora to (E)-5-(2-bromovinyl)uracil (BVU), which lacks antiviral activity but which is converted by DPD to a metabolite that binds covalently to the enzyme. The irreversible inactivation (aka suicidal inactivation) of DPD by sorivudine causes a marked inhibition of 5-fluorouracil metabolism, which increases blood levels of 5-fluorouracil to toxic and, in some cases, lethal levels (Ogura et al., 1998; Kanamitsu et al., 2000).

Several genetic polymorphisms that result in a partial or complete loss of DPD activity, affecting ~8% of the population, have been described (van Kuilenburg et al., 2004; Robert et al., 2005). Severe 5-fluorouracil toxicity has also been documented in individuals who are heterozygous for loss-of-function allelic variants of DPD, and 5-fluorouracil lethality has been documented in rare individuals who are completely deficient in DPD (1 individual in about 10,000). 5-Fluorouracil is one of the most frequently prescribed anticancer drugs, for which reason assessing an individual's DPD genotype (by analyzing DNA for allelic variants) or phenotyping (by measuring DPD activity in peripheral blood mononuclear cells) is advocated prior to 5-fluorouracil or capecitabine (a 5-flurouracil prodrug) therapy so that the dosage of these anticancer drugs can be adjusted on an individual basis.

Dehalogenation

There are three major mechanisms for removing halogens (F, Cl, Br, and I) from aliphatic xenobiotics (Anders, 1985). The first, known as reductive dehalogenation, involves replacement of a halogen with hydrogen, as shown below:

In the second mechanism, known as oxidative dehalogenation, a halogen and hydrogen on the same carbon atom are replaced with oxygen. Depending on the structure of the haloalkane, oxidative dehalogenation leads to the formation of an acylhalide or aldehyde, as shown below:

A third mechanism of dehalogenation involves the elimination of two halogens on adjacent carbon atoms to form a carbon–carbon double bond, as shown below:

A variation on this third mechanism is dehydrohalogenation, in which a halogen and hydrogen on adjacent carbon atoms are eliminated to form a carbon–carbon double bond.

Both reductive and oxidative dehalogenations are catalyzed by CYP. (The ability of CYP to catalyze both reductive and oxidative reactions is explained later in the section “Cytochrome P450.”) Dehalogenation reactions leading to double bond formation are catalyzed by CYP and GST. These reactions play an important role in the biotransformation and metabolic activation of several halogenated alkanes, as the following examples illustrate.

The hepatotoxicity of carbon tetrachloride (CCl₄) and several related halogenated alkanes is dependent on their biotransformation by reductive dehalogenation (Plaa, 2000). The first step in reductive dehalogenation is a one-electron reduction catalyzed by CYP, which produces a potentially toxic, carbon-centered radical and inorganic halide. In the case of CCl₄, reductive dechlorination produces a trichloromethyl radical (^•CCl₃), which initiates lipid peroxidation and produces a variety of other metabolites, as shown in Fig. 6-18.

Halothane can also be converted by reductive dehalogenation to a carbon-centered radical, as shown in Fig. 6-19. The mechanism is identical to that described for carbon tetrachloride, although in the case of halothane the radical is generated through loss of bromine, which is a better leaving group than chlorine. Fig. 6-19 also shows that halothane can undergo oxidative dehalogenation, which involves oxygen insertion at the C–H bond to generate an unstable halohydrin (CF₃COHClBr) that decomposes to a reactive acylhalide (CF₃COCl), which can bind to cellular proteins (particularly to amine groups) or further decompose to trifluoroacetic acid (CF₃COOH).

Both the oxidative and reductive pathways of halothane metabolism generate reactive intermediates capable of binding to proteins and other cellular macromolecules. The relative importance of these two pathways to halothane-induced hepatotoxicity is species dependent. In rats, halothane-induced hepatotoxicity is promoted by those conditions favoring the reductive dehalogenation of halothane, such as moderate hypoxia (10%–14% oxygen) plus treatment with the CYP inducers, phenobarbital, and pregnenolone-16α-carbonitrile (PCN). In contrast to the situation in rats, halothane-induced hepatotoxicity in guinea pigs is largely the result of oxidative dehalogenation of halothane (Lunam et al., 1989). In guinea pigs, halothane hepatotoxicity is not enhanced by moderate hypoxia and is diminished by the use of deuterated halothane, which impedes the oxidative dehalogenation of halothane because the CYP-dependent insertion of oxygen into a carbon–deuterium bond is energetically less favorable (and therefore slower) than inserting oxygen into a carbon–hydrogen bond.

Halothane hepatitis in humans is a rare but severe form of liver necrosis associated with repeated exposure to this volatile anesthetic. In humans, as in guinea pigs, halothane hepatotoxicity results from the oxidative dehalogenation of halothane, as shown in Fig. 6-19. Serum samples from patients suffering from halothane hepatitis contain antibodies directed against neoantigens formed by the trifluoroacetylation of proteins. These antibodies have been used to identify which specific proteins in the endoplasmic reticulum are targets for trifluoroacetylation during the oxidative dehalogenation of halothane (Pohl et al., 1989).

The concept that halothane is activated by CYP to trifluoroacetylhalide, which binds covalently to proteins and elicits an immune response, has been extended to other volatile anesthetics, such as enflurane, methoxyflurane, and isoflurane. In other words, these halogenated aliphatic hydrocarbons, like halothane, may be converted to acylhalides that form immunogens by binding covalently to proteins. In addition to accounting for rare instances of enflurane hepatitis, this mechanism of hepatotoxicity can also account for reports of a cross-sensitization between enflurane and halothane, in which enflurane causes liver damage in patients previously exposed to halothane.

One of the metabolites generated from the reductive dehalogenation of halothane is 2-chloro-1,1-difluoroethylene (Fig. 6-19). The formation of this metabolite involves the loss of two halogens from adjacent carbon atoms with formation of a carbon–carbon double bond. This type of dehalogenation reaction can also be catalyzed by GSTs. GSH initiates the reaction with a nucleophilic attack either on the electrophilic carbon to which the halogen is attached (mechanism A) or on the halogen itself (mechanism B), as shown in Fig. 6-20 for the dehalogenation of 1,2-dihaloethane to ethylene. The insecticide DDT is detoxified by dehydrochlorination to DDE by a lyase (ie, DDT-dehydrochlorinase), as shown in Fig. 6-21. The activity of this GSH-dependent reaction correlates well with resistance to DDT in houseflies.

Dehydroxylation—mARC, Cytochrome b5,b5 Reductase, and Aldehyde Oxidase

Mitochondrial amidoxime reducing component (mARC) is a molybdenum-containing enzyme that, in the presence of NADH, cytochrome b₅, and NADH-cytochrome b₅ reductase, can catalyze the N-dehydroxylation of various amidoximes and related N-hydroxy compounds, as shown in Fig. 6-22 (Havemeyer et al., 2010). Reactions catalyzed by mARC/cytochrome b₅/NADH-cytochrome b₅ reductase include the N-dehydroxylation of amidoximes formed during the activation of the antiparasitic prodrug pafuramidine (DB-289), the N-dehydroxylation of N-hydroxy-sulfonamides, and the N-dehydroxylation of N-hydroxy-valdecoxib to its active principal, a cyclooxygenase-2 (COX-2) inhibitor (Saulter et al., 2005; Havemeyer et al., 2010). The mARC complex can also N-dehydroxylate the aryl hydroxylamine metabolites of carcinogenic arylamines such as 4-aminobiphenyl and 2-amino-1-methyl-6-phenyl-imidazol[4,5-b]pyridine (PhIP), N-hydroxy metabolites that are formed by CYP1A1, 1A2, 1B1, lactoperoxidase, and myeloperoxidase (see the sections “Cytochrome P450” and “Peroxidase-Dependent Cooxidation”) (Kurian et al., 2006). Because these carcinogenic aryl hydroxylamines can be further activated by glucuronidation, sulfonation, or acetylation in various tissues (see the sections “Glucuronidation and Formation of Acyl-CoA Thioesters,” “Sulfonation,” and “Acetylation”), reduction by mARC/cytochrome b₅/NADH-cytochrome b₅ reductase represents a competing detoxication pathway. Gut microflora can also catalyze dehydroxylation reactions as shown in Fig. 6-1 for quinic acid. Gut microflora can also reduce N- and S-oxides formed by FAD-containing monooxygenase (FMO) and/or CYP, as described for trimethylamine (TMA) N-oxide in the section “Flavin Monooxygenases.”

Aldehyde Oxidase—Reductive Reactions

Aldeyde oxidase (AO) is a cytosolic molybdoenzyme that catalyzes the oxidation of some xenobiotics and the reduction of others. The types of oxidative and reductive reactions catalyzed by AO are shown in Fig. 6-23. In contrast to the large number of drugs that are known to be (or suspected of being) oxidized by AO in vivo, only a few drugs are known to be (or suspected of being) reduced by AO in vivo, including nitrofurazone, zonisamide, and ziprasidone. The reductive metabolism of ziprasidone by AO is shown in Fig. 6-4. The features of AO and the oxidative reactions it catalyzes are discussed later in the section “Aldehyde Oxidase.”

Oxidation

Alcohol, Aldehyde, Ketone Oxidation–Reduction Systems

Alcohols, aldehydes, and ketones are oxidized by a number of enzymes, including alcohol dehydrogenase, aldehyde dehydrogenase, AKRs (such as those with dihydrodiol dehydrogenase activity), the molybdenum-containing enzymes (namely, AO and xanthine dehydrogenase [XD]/xanthine oxidase [XO]), and CYP. For example, simple alcohols (such as methanol and ethanol) are oxidized to aldehydes (namely, formaldehyde and acetaldehyde) by ADH. These aldehydes are further oxidized to carboxylic acids (formic acid and acetic acid) by ALDH, as shown in Fig. 6-24. Many of the aforementioned enzymes can also catalyze the reduction of xenobiotics, as discussed in the section “Reduction.”

Alcohol Dehydrogenase

ADHs belong to the MDRs, as shown in Table 6-5. ADHs are zinc-containing, cytosolic enzymes present in several tissues including liver (which has the highest levels), kidney, lung, and gastric mucosa (Agarwal, 1992; Ramchandani, 2004). These enzymes oxidize several types of alcohols including hydroxysteroids, retinol, ethanol, lipid peroxidation products, and other simple and complex (eg, ring-containing) alcohols. Human ADHs are dimeric proteins consisting of two ~40-kDa subunits designated α, β, γ, π, χ, σ (also previously known as μ), or ADH6 (the latter having no subunit designation). As shown in Table 6-5, there are 7 human ADHs, and these are categorized into 5 classes (I-V) based on patterns of tissue-specific expression, catalytic properties, and amino acid sequence. Class I comprises 3 hepatically expressed genes: ADH1A, ADH1B, and ADH1C, which were formerly known as ADH1, -2, and -3, respectively. The class I isozymes consist of homodimeric and heterodimeric forms of the 3 subunits (eg, αα, αβ, ββ, βγ, γγ, etc). Class II contains ADH4, which is composed of 2 pi subunits (ππ). Class III contains ADH5, which is composed of 2 chi subunits (χχ). Class IV contains ADH7, which is composed of 2 sigma subunits (σσ). Class V contains ADH6 (for which there is no subunit designation) (Brennan et al., 2004; Ramchandani, 2004). Therefore, there are over 20 human ADH isozymes and these differ in their substrate specificity and catalytic efficiency toward ethanol (Ramchandani, 2004). The human ADH genes have similar sequences (ie, 60%–70% identical coding regions), and all have 9 exons and 8 introns with the exception of ADH6, which lacks the last exon (Han et al., 2005). In addition, several hundred genetic variants have been described across the human ADH cluster which lies on chromosome 4q (Li et al., 2008). Alcohols can be oxidized to aldehydes by non-ADH enzymes in microsomes and peroxisomes, although these are quantitatively less important than ADH for ethanol oxidation (Lieber, 2004). The microsomal ethanol oxidizing system (formerly known as MEOS) is the CYP enzyme, CYP2E1. The corresponding peroxisomal enzyme is catalase. The oxidation of ethanol to acetaldehyde by these 3 enzyme systems is shown in Fig. 6-25.

Class I (ADH1A, 1B, 1C)

The class I ADH isozymes (ADH1A or α-ADH, ADH1B or β-ADH, and ADH1C or γ-ADH) are responsible for the oxidation of ethanol and other small, aliphatic alcohols, and they are strongly inhibited by pyrazole and its 4-alkyl derivatives (eg, 4-methylpyrazole). High levels of class I ADH isozymes are expressed in liver and adrenals, with lower levels in kidney, lung, blood vessels (in the case of ADH1B), gastric mucosa (in the case of ADH1C), and other tissues, but not brain. It is noteworthy that the liver expresses a very large amount of ADHs (approximately 3% of all soluble protein) and also expresses the widest variety of isozymes (Ramchandani, 2004).

The class I ADH isozymes differ in their capacity to oxidize ethanol. Polymorphisms have been well described for the class I ADH isozymes. Even the allelozymes, which differ in a single amino acid, differ markedly in their affinity (K_m) and/or capacity (V_max) for oxidizing ethanol to acetaldehyde. There are at least 3 allelic variants of ADH1B (ie, *1, *2, and *3), with a single amino acid change at position 48. The homodimer, β₂β₂, and heterodimers containing at least 1 β₂ subunit (ie, the ADH1B*2 allelozymes) are especially active in oxidizing ethanol at physiological pH. ADH1B*2 (formerly known as ADH2*2) is known as atypical ADH, and is responsible for the unusually rapid conversion of ethanol to acetaldehyde in up to 90% of the Pacific Rim Asian population (eg, Japanese, Chinese, Korean), whereas only ~10% of Caucasians express this allele. The ADH1B*3 is relatively common in individuals of African descent. The latter 2 alleles have greater activity toward ethanol than the ADH1B1*1 allele (Kimura and Higuchi, 2011). These population differences in ADH1B allelozyme expression contribute to ethnic differences in alcohol consumption and toxicity, as discussed in the section “Aldehyde Dehydrogenase.”

Unlike the allelic variants of ADH1B, the allelic variants of ADH1C do not differ markedly in their ability to oxidize ethanol. However, as in the case of the ADH1B allelozymes, the expression of the ADH1C allelozymes also varies from one ethnic group to the next. The 2 allelozymes of ADH1C, namely, ADH1C*1 (γ₁-ADH) and ADH1C*2 (γ₂-ADH), are, respectively, expressed ~50:50 in Caucasians and 90:10 in Pacific Rim Asians, with the *2 allele having somewhat higher activity toward ethanol than the *1 allele (Li, 2000; Kimura and Higuchi, 2011). An additional SNP in ADH1C has been found in up to 20% of some Native American populations (Ramchandani, 2004).

Class II (ADH4)

The class II enzyme ADH4 (π-ADH) is mainly expressed in liver (and to some extent in other gastrointestinal tissues), where it preferentially oxidizes larger alcohols (Ramchandani, 2004). ADH4 differs from the ADH1 isozymes in that it is less sensitive to pyrazole inhibition, but may play some role in ethanol oxidation, especially at high concentrations (Lockley et al., 2005; Edenberg, 2007; Kimura and Higuchi, 2011). Some studies support a role for polymorphisms of ADH4 in the susceptibility to alcoholism (Kimura and Higuchi, 2011).

Class III (ADH5)

The class III enzyme ADH5 (a homodimer of the χ-subunit) preferentially oxidizes long-chain alcohols (pentanol and larger), omega-hydroxy-fatty acids, and other alcohols (such as cinnamyl alcohol). Like ADH4, ADH5 is less sensitive to pyrazole inhibition than ADH1 enzymes. However, in contrast to ADH4, which is largely confined to the liver, ADH5 is ubiquitous, being present in virtually all tissues (including brain), where it catalyzes the rate-limiting step in detoxifying formaldehyde through oxidation of S-hydroxymethylglutathione (which is formed spontaneously from formaldehyde and GSH) to S-formylglutathione. In fact, ADH5 and the GSH-dependent formaldehyde dehydrogenase (also referred to in literature as FDH, ADH3, or S-nitrosoglutathione reductase [GSNOR]) are identical enzymes (Koivusalo et al., 1989; Edenberg, 2007; Just et al., 2011). This enzyme appears to be the ancestral form of ADH from which all other vertebrate ADHs have evolved, and so far functional polymorphisms have not been identified in ADH5 (Just et al., 2011).

Class IV (ADH7)

The class IV enzyme ADH7 (σ-ADH; also referred to as the μ subunit) is a low-affinity (high K_m), high-capacity (high V_max) enzyme, and is the most active of the medium-chain ADHs in oxidizing retinol (a member of the vitamin A family). It is the major ADH expressed in human stomach and other areas of the UADT (eg, stomach, esophagus, pharynx, gingiva, mouth, and tongue), as well as the eyes (Han et al., 2005). In contrast to the other ADHs, ADH7 is not expressed in adult human liver (Ramchandani, 2004). Among the human ADH forms, ADH7 has the highest activity toward ethanol (Han et al., 2005). Inasmuch as ADH7 is expressed in the upper gastrointestinal tract, where chronic alcohol consumption leads to cancer development, there is considerable interest in the role of ADH7 in the preabsorptive conversion of ethanol to acetaldehyde (a suspected upper GI tract carcinogen or cocarcinogen) and in its role in the metabolism of retinol (a vitamin required for epithelial cell growth and differentiation), which might be inhibited by alcohol consumption (Seitz and Oneta, 1998). The role for a protective effect of high-activity polymorphisms against alcoholism remains to be fully elucidated, with some studies showing a positive association (Han et al., 2005), and others no association (Duell et al., 2011). However, the A92G SNP (rs1573496; C → G) in ADH7 has been found to confer a reduced risk of squamous cell carcinoma of the head and neck in Caucasians (Wei et al., 2010).

Compared with hepatic ADH, gastric ADH has a lower affinity (higher K_m) but higher capacity (larger V_max) for oxidizing ethanol, the former being dominated by the class I ADH isozymes and the latter by the class IV enzyme ADH7. Although ethanol is largely biotransformed by hepatic ADH1, gastric ADH7 nevertheless can limit the systemic bioavailability of alcohol. This first-pass elimination of alcohol by gastric ADH7 can be significant depending on the manner in which the alcohol is consumed; large doses over a short time produce high ethanol concentrations in the stomach, which compensate for the low affinity (high K_m) of gastric ADH7. Young women have lower gastric ADH7 activity than do men, and gastric ADH7 activity tends to be lower in alcoholics (Frezza et al., 1990). Some alcoholic women have no detectable gastric ADH7, and blood levels of ethanol after oral consumption of alcohol are the same as those that are obtained after intravenous administration. Gastric ADH7 activity decreases during fasting, which is one reason alcohol is more intoxicating when consumed on an empty stomach. Several commonly used drugs (eg, cimetidine, ranitidine, aspirin) are noncompetitive inhibitors of gastric ADH7. Under certain circumstances these drugs increase the systemic availability of alcohol, although the effect is too small to have serious medical, social, or legal consequences (Levitt, 1993). About 30% of Asians appear to be genetically deficient in ADH7, the main gastric ADH. In addition to biotransforming ethanol and retinol, ADH7 also detoxifies the dietary carcinogen, nitrobenzaldehyde. It has been suggested that a lack of ADH7 in some Japanese subjects may impair their ability to detoxify nitrobenzaldehyde and may possibly be linked to the high rate of gastric cancer observed in the Japanese population (Seitz and Oneta, 1998).

Class V (ADH6)

The mRNA for class V ADH (namely, ADH6, for which there is no subunit designation) (Brennan et al., 2004; Ramchandani, 2004) has been found in fetal and adult liver (Edenberg, 2007). However, the protein has not yet been isolated from human tissue, so little is known about its in vivo function. ADH6 has been expressed in vitro and metabolizes ethanol with a K_m of approximately 28 mM and has higher affinity for benzyl alcohol (K_m 0.12 mM) and propanol (K_m 3.2 mM) (Zhi et al., 2000).

Aldehyde Dehydrogenase

ALDHs oxidize 4 major types of aldehydes: (1) saturated alkanals (eg, formaldehyde, acetaldehyde), (2) unsaturated alkenals (eg, acrolein), (3) aromatic aldehydes (eg, benzaldehyde), and (4) dicarbonyls (eg, glyoxal, malondialdehyde) to their corresponding carboxylic acids, generally with NAD⁺ as the cofactor. However, it should be noted that ALDH1L1 prefers NADP⁺ over NAD⁺, and ALDH3B1 can use either cofactor in a substrate-dependent manner. In addition, ALDH6A1 requires acetyl- or propionyl-CoA (Marchitti et al., 2007, 2008). Most of the enzymes also have esterase activity (Yoshida et al., 1998; Marchitti et al., 2008). Several ALDH enzymes are involved in the oxidation of xenobiotic aldehydes, such as those formed from ethanol, allyl alcohol, carbon tetrachloride, cyclophosphamide, and ifosfamide (Marchitti et al., 2007). Formaldehyde dehydrogenase, which specifically oxidizes formaldehyde that is complexed with GSH, is not a member of the ALDH family but is a class III ADH (ADH5) (Koivusalo et al., 1989; Edenberg, 2007; Just et al., 2011). At least 19 ALDH genes have been identified in humans, and a correspondingly large number of ALDH genes appear to be present in other mammalian species (Sládek, 2003; Vasiliou et al., 2004; Marchitti et al., 2007, 2008). The name, tissue distribution, subcellular location, and major substrate for the human ALDHs are summarized in Table 6-6. The ALDHs differ in their primary amino acid sequences. They may also differ in the quaternary structure. For example, ALDH3A1 is a dimer of two 85-kDa subunits, whereas ALDH1A1 and ALDH2 are homotetramers of 54-kDa subunits (Goedde and Agarwal, 1992; Marchitti et al., 2008).

As shown in Fig. 6-25, ALDH2 is a mitochondrial enzyme that, by virtue of its high affinity, is primarily responsible for oxidizing simple aldehydes, such as acetaldehyde (K_m for acetaldehyde <1 μM) and, by acting as a nitrate reductase, ALDH2 is the principal enzyme necessary for the activation of nitroglycerin (Marchitti et al., 2008). In fact, the genetic polymorphism ALDH2*2 is associated with decreased nitroglycerin efficacy in certain Chinese populations (Marchitti et al., 2008). Several genetic polymorphisms in human ALDH2 have been described, including the well-described ALDH2*2 point mutation (Glu₄₈₇ → Lys₄₈₇), which results in a loss of activity due to a greatly diminished affinity for NAD⁺ (Marchitti et al., 2008). The mutant ALDH2*2 allele is dominant, such that heterotetrameric ALDH2 proteins containing even a single ALDH2*2 subunit are inactive (Marchitti et al., 2008). A high percentage (40%–50%) of individuals of Asian descent are deficient in ALDH2 activity due to the presence of the ALDH2*2 allele. This same population also has a high incidence of the atypical form of ADH1B (ie, ADH1B*2), which means that they rapidly convert ethanol to acetaldehyde but only slowly convert acetaldehyde to acetic acid. (They also have a relatively high prevalence of a deficiency of ADH7 activity, which impairs gastric metabolism of ethanol.) As a result, many Asian subjects experience a flushing syndrome after consuming alcohol due to a rapid buildup of acetaldehyde, which triggers the dilation of facial blood vessels through the release of catecholamines. Some Native American populations also experience a flushing syndrome after consuming alcohol, possibly because they express ADH1B*2 (Ramchandani, 2004) or a different allelic variant of ALDH2 and/or because acetaldehyde oxidation in blood erythrocytes is impaired in these individuals, possibly due to the expression of a variant form of ALDH1A1. Both the functional genetic variants of ADH that rapidly convert ethanol to acetaldehyde (ie, ADH1B*2) and the genetic variants of ALDH that slowly detoxify acetaldehyde (ie, ALDH2*2) protect against heavy drinking and alcoholism. Inhibition of ALDH by disulfiram (Antabuse) causes an accumulation of acetaldehyde in alcoholics. The nauseating effect of acetaldehyde serves to deter continued ethanol consumption (Goedde and Agarwal, 1992). However, it is important to note that a predisposition toward alcoholism is not simply determined by factors that affect the pharmacokinetics of ethanol and its metabolites. Studies in humans and rodents implicate 5-HT1B, 2A, 3A, and 3B receptors, dopamine-related genes, GABA receptors, cholinergic muscarinic receptor 2, the endogenous opioid system, tryptophan hydroxylase, and neuropeptide Y as candidate targets of genetic susceptibility in the pharmacodynamic actions of ethanol (Li, 2000; Kimura and Higuchi, 2011). MAO may also be a risk factor for alcoholism, as discussed later (see the section “Amine Oxidases”).

Genetic deficiencies in other ALDHs impair the metabolism of other aldehydes, which is the underlying basis of certain diseases. For example, polymorphisms in certain ALDH genes may alter the risk for, or cause, the following: spina bifida (ALDH1A2), Sjögren–Larsson syndrome (ALDH3A2), paranoid schizophrenia (ALDH3B1), type II hyperprolinemia (ALDH4A1), γ-hydroxybutyric aciduria (ALDH5A1), methylmalonic aciduria (ALDH6A1), pyroxidine-dependent seizures (ALDH7A1), nonalcoholic steatohepatits (NASH) (ALDH9A1), and hyperammonemia and/or hypoprolinemia (ALDH18A1) (Marchitti et al., 2008).

The toxicological consequences of an inherited (ie, genetic) or acquired (eg, drug-induced) deficiency of ALDH illustrate that aldehydes are more cytotoxic than the corresponding alcohol. This is especially true of allyl alcohol (CH₂=CHCH₂OH), which is converted by ADH to the highly hepatotoxic aldehyde, acrolein (CH₂=CHCHO) (Marchitti et al., 2007, 2008).

The oxidation of ethanol by ADH and ALDH leads to the formation of acetic acid, which is rapidly oxidized to carbon dioxide and water. However, in certain cases, alcohols are converted to toxic carboxylic acids, as in the case of methanol and ethylene glycol, which are converted via aldehyde intermediates to formic acid and oxalic acid, respectively. Formic and oxalic acids are considerably more toxic than acetic acid. For this reason, methanol and ethylene glycol poisonings are commonly treated with ethanol, which competitively inhibits the oxidation of methanol and ethylene glycol by ADH and ALDH. The potent inhibitor of ADH, 4-methylpyrazole (fomepizole) (Lockley et al., 2005), is also used to treat methanol and ethylene glycol poisonings.

The reduction of aldehydes and ketones to primary and secondary alcohols by carbonyl reductases has already been discussed (see the section “Carbonyl Reduction—AKRs and SDRs”). In contrast to ADH and ALDH, carbonyl reductases typically use NADPH as the source of reducing equivalents. Aldehydes, especially aromatic aldehydes, can also be oxidized by AO and xanthine oxidoreductase (XOR), which are discussed in the section “Molybdenum Hydroxylases (Molybdoenzymes).”

Dimeric Dihydrodiol Dehydrogenase

As mentioned in the section “Carbonyl Reduction—AKRs and SDRs,” several members of the AKR superfamily are dihydrodiol dehydrogenases that oxidize the trans-dihydrodiols of various PAHs to the corresponding ortho-quinones, as shown in Fig. 6-10 (Penning, 1997; Burczynski and Penning, 2000). There is also a dihydrodiol dehydrogenase (gene symbol DHDH; completely unrelated to AKRs or SDRs) that encodes a dimeric enzyme that oxidizes trans-dihydrodiols of aromatic hydrocarbons to the corresponding catechols in a NADP⁺-dependent manner (Carbone et al., 2008). The overall reaction catalyzed by dimeric dihydrodiol dehydrogenase is a 2-electron oxidation of one of the hydroxyl groups of a trans-dihydrodiol to an intermediate ketol, which rapidly enolizes to the catechol (Carbone et al., 2008). It also activates naphthalene trans-dihydrodiol to 1,2-dihydroxynaphthalene, which rapidly auto-oxidizes to 1,2-naphthoquinone, a cytotoxic ortho-quinone that may be responsible for naphthalene-induced cataracts (Carbone et al., 2008). Dimeric dihydrodiol dehydrogenase is not active toward hydroxysteroids, but it does oxidize several endogenous sugars, including D-glucose and D-xylose, and may therefore play a role in the metabolism of dietary sugars with subsequent generation of NADPH. Dimeric dihydrodiol dehydrogenase can also catalyze reductions in the presence of NADPH, such as the carbonyl reduction of methylglyoxal and 3-deoxyglucosone, nitrobenzaldehydes, and camphoroquine (Carbone et al., 2008). Dimeric dihydrodiol dehydrogenase can be inhibited by L-ascorbic and isoascorbic acids and also by 4-hydroxyphenylketones and 2,6-dihydroxyanthraquinone, and is inactivated by magnesium chloride and some phosphate salts (Carbone et al., 2008). This enzyme was first characterized in rabbit liver, where it oxidizes benzene dihydrodiol to the genotoxic and immunotoxic catechol. It has also been isolated from pig, dog, and cynomolgus monkey (Macaca fascicularis). The purified enzyme is composed of identical subunits with molecular weights of 32 to 36 kDa, and amino acid sequence in humans indicates that each subunit is composed of 335 amino acids (Carbone et al., 2008). Dimeric dihydrodiol dehydrogenase is ubiquitously expressed in dogs and pigs, but only in the kidney of monkey, lens and intestine of rabbit, and apparently only the intestine of humans (Carbone et al., 2008).

It should be noted that PAHs can be oxidized by CYP to an arene oxide, an electrophilic and potentially toxic metabolite that is detoxified by its conversion to a trans-dihydrodiol by epoxide hydrolase. By oxidizing the trans-dihydrodiol to an ortho-quinone, AKRs generate yet another potentially toxic metabolite. These roles of AKRs (which can act as monomeric dihydrodiol dehydrogenases) in the metabolism of PAHs are well known but the activity of dimeric dihydrodiol dehydrogenase toward the trans-dihydrodiols of PAHs has not yet been reported (Carbone et al., 2008). In terms of their ability to cause oxidative stress and cellular toxicity, ortho-quinones can be considered equivalent to para-quinones, which were discussed earlier in the section “Quinone Reduction—NQO1 and NQO2.”

Molybdenum Hydroxylases (Molybdoenzymes)

There are 5 human molybdenum-containing enzymes (molybdoenzymes), namely, AO (gene symbol AOX1), XOR (also known as XO or XD, gene symbol XDH), sulfite oxidase (gene symbol SUOX), and the more recently discovered “mitochondrial amidoxime-reducing component” (mARC-1 and -2; gene symbols MOSC1 and MOSC2) (Havemeyer et al., 2011). AO and XOR are the 2 major molybdoenzymes that participate in the biotransformation of xenobiotics. The human mARC (consisting of mARC-1 and -2) is involved in the activation of amidoxime prodrugs and detoxication of some N-hydroxylated xenobiotics, as described in the section “Dehydroxylation—mARC, Cytochrome b₅, b₅ Reductase, and Aldehyde Oxidase.” Sulfite oxidase will not be described in detail here except that, as its name implies, the enzyme oxidizes sulfite, an irritating air pollutant, to sulfate, which is relatively innocuous, and is also required to metabolize the sulfur-containing amino acids cysteine and methionine in foods.

Genetic polymorphisms can affect all molybdoenzymes or individual enzymes. Homozygosity for nonfunctional forms of some of the genes involved in the synthesis of the active molybdenum cofactor (MoCo), namely, MOCS1, MOCS2, MOCOS, or GPHN, results in a complete lack of activity of all molybdoenzymes, and is clinically indistinguishable from isolated SUOX deficiency (Reiss and Hahnewald, 2010), a rare condition that gives rise to progressive neurological damage and death in early childhood (Reiss and Johnson, 2003; Feng et al., 2007). Isolated XOR deficiency results in the classical type I xanthinuria, which can lead to xanthine stones (calculi) because of the inability to oxidize xanthine and hypoxanthine to uric acid, but it is not life-threatening (Reiss and Hahnewald, 2010). Although no isolated forms of AO deficiency have been described, type II xanthinuria results from a combined deficiency of XOR and AO as evidenced by the inability to oxidize both xanthine and hypoxanthine as well as allopurinol to oxypurinol. Certain polymorphisms in AO do appear to cause a decrease or complete loss of activity toward certain substrates, as discussed in the section “Aldehyde Oxidase” (Hartmann et al., 2012). Not surprisingly, given its recent discovery, isolated mARC deficiency has not yet been described.

XOR and AO comprise the human members of the “XO family” of molybdoenzymes. Both enzymes are only active as homodimers, each with an ~145-kDa monomer consisting of an N-terminal region that binds 2 nonidentical iron–sulfur clusters of the [2Fe–2S] type, followed by an FAD-binding region and C-terminal domains that bind the molybdenum center (Wahl et al., 2010). In XOR, however, an additional NAD⁺ binding site is present. In XOR and AO, the molybdenum is ligated to an oxo ligand, a hydroxyl group, 2 dithiolene sulfurs, and a sulfido group (as shown in the upper left panel of Fig. 6-27). The molybdenum–sulfur bond is inserted through a posttranslational reaction catalyzed by MoCo sulfurase (gene symbol MOCOS), which substitutes one of the oxo groups of the MoCo with a sulfo double bond and is essential for the activity of these enzymes (Garattini et al., 2008). In contrast, in sulfite oxidase (the only fully accepted human member of the “sulfite oxidase family” of molybdoenzymes), the molybdenum is coordinated by the 2 sulfurs on the pterin, 2 oxo ligands, and a protein-derived cysteinyl sulfur (Wahl et al., 2010). Sulfite oxidase does not require posttranslational sulfuration of the MoCo to become active. The two 35 kDa human mARC enzymes may represent a separate family of molybdoenzymes in that they do not contain an FAD domain and may bind molybdenum in a manner different than either the XO or sulfite oxidase families. However, there is some evidence to suggest that the mARCs are in fact members of the human sulfite oxidase family of molybdoenzymes (Havemeyer et al., 2011). In addition, the mARCs share significant sequence similarity with the C-terminal domain of MOCOS and appear to require cytochrome b₅, cytochrome b₅ reductase, and NADH for catalytic activity (Wahl et al., 2010). Furthermore, whereas XOR, AO, and SUOX are localized in the cytosol, the 2 mARC proteins appear to be localized to the outer mitochondrial membrane (Havemeyer et al., 2011).

The catalytic cycle for XOR and AO involves an interaction between the molybdenum center with a reducing substrate, which results in the reduction of the MoCo, after which reducing equivalents are transferred intramolecularly to the flavin cofactors (which act as electron sinks), with reoxidation occurring via the flavin moiety by molecular oxygen (in the case of AO and the XO form of XOR) or NAD⁺ (in the case of the XD form of XOR). Electrons are transferred from the MoCo to the flavin cofactor through the intermediacy of the iron–sulfur centers. During substrate oxidation, AO and XOR are reduced and then reoxidized by molecular oxygen; hence, they function as true oxidases. The oxygen incorporated into the xenobiotic is derived from water rather than oxygen, which distinguishes these oxidases from oxygenases. The main functional difference between AO and XOR is the ability of the latter to use both NAD⁺ (in its XD form) and molecular oxygen (in its XO form), as the final acceptors of the reducing equivalents (Garattini and Terao, 2011). Reduction of oxygen by the XO form of XOR leads to the formation of hydrogen peroxide or superoxide anion (in a substrate-dependent manner) as follows:

RH + H₂O + O₂ → ROH + H₂O₂;

RH + H₂O + 2O₂ → ROH + 2O₂⁻ + 2H⁺.

Additional details of the catalytic cycle are described in the section “Xanthine Oxidoreductase.” XOR and AO catalyze the oxidation of carbon atoms with a low electron density such as the electron-deficient sp²-hybridized carbon atom, double bonded to nitrogen in aromatic azaheterocycles (R₁-CH=N-R₂) or oxygen in various aldehydes (R₁-CH=O). In aromatic azaheterocycles (such as purines, pyrimidines, pteridines, quinolines, iminium ions, and diazanaphthalenes), oxidation by AO (and in most cases by XOR) occurs at the carbon with the lowest electron density, which is typically the α-carbon (next to the nitrogen) but in some cases the γ-carbon (2 carbons removed) (Strolin-Benedetti, 2011). This contrasts with oxidation by CYP, which generally catalyzes the oxidation of carbon atoms with a high electron density. For this reason, xenobiotics that are good substrates for molybdoenzymes (nucleophilic oxidizing enzymes) tend to be poor substrates for CYP (an electrophilic oxidizing enzyme), and vice versa (Pryde et al., 2010). AO and, to a lesser extent, XOR hydroxylate the carbonyl carbon in certain aldehydes and convert them to carboxylic acids (R-CHO → R-COOH). Both molybdoenzymes hydroxylate the α-carbon in aromatic azaheterocycles (R₁-CH=N-R₂) to form α-hydroxyimines (R₁-COH=N-R₂) that rapidly tautomerize to the corresponding lactam (R₁-CO−NH-R₂) (α-aminoketone). Certain aromatic aldehydes, such as tamoxifen aldehyde and benzaldehyde, are good substrates for AO and XOR, whereas aliphatic aldehydes tend to be poor substrates. Consequently, AO and XOR contribute negligibly to the metabolism of acetaldehyde under normal conditions. Some reactions catalyzed by AO and XOR are shown in Fig. 6-26. It should be noted that although AO and XOR tend to oxidize similar substrates, there are some differences in substrate specificity (eg, vanillin is rapidly oxidized by AO, but there is little contribution of XOR) (Strolin-Benedetti, 2011). An interesting difference between XOR and AO is their regioselective oxidation of the purine prodrug 6-deoxyacyclovir, which is oxidized by XOR in the 6-membered ring to form the active drug acyclovir (discussed below) but is oxidized by AO in the 5-membered ring to form the pharmacologically inactive metabolite 6-deoxy-8-hydroxy-acyclovir (Testa and Krämer, 2008, 2010). In the case of nonaromatic azaheterocycles the preferred substrate is the protonated form (−CH=NH⁺−) because this form lowers the electron density at the α-carbon. For this same reason, N-alkylation of aromatic azaheterocycles to form a quaternary (positively charged) nitrogen next to the α-carbon improves oxidation by AO (more so than by XOR), as in the case of N-methylnicotinamide. Under certain conditions, both enzymes can also catalyze the reduction of xenobiotics (Testa and Krämer, 2008, 2010). Examples of reductive reactions catalyzed by AO are shown in Fig. 6-23.

Xanthine Oxidoreductase

XD and XO are 2 forms of the same cytosolic enzyme (XOR, gene symbol XDH) that differ in the electron acceptor used in the final step of catalysis. In the case of XD, the final electron acceptor is NAD⁺ (dehydrogenase activity), whereas in the case of XO the final electron acceptor is oxygen (oxidase activity). Although both XOR and AO are expressed in many tissues, the highest XOR expression in humans is found in the proximal intestine, liver, and lactating mammary glands (Strolin-Benedetti, 2011). XD is converted reversibly to XO by oxidation of sulfhydryl groups or irreversibly by proteolytic cleavage (Testa and Krämer, 2008, 2010). In humans, XOR is encoded by a single gene (ie, XDH). Under normal physiological conditions, XD is the predominant form of the enzyme found in vivo. However, during tissue processing, the dehydrogenase form tends to be converted to the oxidase form; hence, most in vitro studies are conducted with XO or a combination of XO and XD. The induction (upregulation) of XD and/or the conversion of XD to XO in vivo is thought to play an important role in ischemia–reperfusion injury, lipopolysaccharide (LPS)–mediated tissue injury, and alcohol-induced hepatotoxicity (Pacher et al., 2006). During ischemia, XO levels increase because hypoxia induces XOR gene transcription, and because XD is converted to XO. During reperfusion, XO contributes to oxidative stress and lipid peroxidation because the oxidase activity of XO involves the reduction of molecular oxygen, which can lead to the formation of ROS of the type shown in Fig. 6-16. Similarly, treatment with LPS, a bacterial endotoxin that triggers an acute inflammatory response, increases XO activity both by inducing XOR transcription and by converting XD to XO. The associated increase in oxidative stress has been implicated in LPS-induced cytotoxicity. Ethanol facilitates the conversion of XD to XO, and the conversion of ethanol to acetaldehyde provides a substrate and, hence, a source of electrons for the reduction of oxygen. Hereafter, the 2 forms of the enzyme will be referred to as XOR.

Typical reactions catalyzed by XOR are shown in Fig. 6-26. XOR contributes significantly to the first-pass elimination of several purine derivatives (eg, 6-mercaptopurine and 2,6-dithiopurine), and limits the therapeutic effects of these cancer chemotherapeutic agents. In contrast, certain prodrugs are activated by XOR. For example, the antiviral prodrugs 6-deoxyacyclovir and 2′-fluoroarabino-dideoxypurine, which are relatively well absorbed after oral dosing, are oxidized by XOR to their respective active forms, acyclovir and 2′-fluoroarabino-dideoxyinosine, which are otherwise poorly absorbed (see Fig. 6-26). Furthermore, XOR has been implicated in the bioactivation of mitomycin C and related antineoplastic drugs, although this bioactivation reaction is thought to be largely catalyzed by NQO1 (DT-diaphorase), as discussed previously in the section “Quinone Reduction—NQO1 and NQO2.” Other xenobiotic substrates of XOR include 6-thioxanthine, methotrexate, and the benzylic aldehyde metabolite of tolbutamide (Chladek et al., 1997; Testa and Krämer, 2008, 2010; Kudo et al., 2010).

XOR catalyzes an important physiological reaction, namely, the sequential oxidation of hypoxanthine to xanthine and uric acid, as shown in Fig. 6-26 (Rajagopalan, 1980). By competing with hypoxanthine and xanthine for oxidation by XOR, allopurinol inhibits the formation of uric acid, making allopurinol a useful drug in the treatment of gout (a complication of hyperuricemia). Allopurinol can also be used to evaluate the contribution of XOR to xenobiotic biotransformation in vivo. Like allopurinol, hydroxylated coumarin derivatives, such as umbelliferone (7-hydroxycoumarin) and esculetin (7,8-dihydroxycoumarin), are potent inhibitors of XOR. Allopurinol and other XOR inhibitors are being evaluated for the treatment of various types of ischemia–reperfusion and vascular injury that appear to be mediated, at least in part, by XOR (Pacher et al., 2006).

Monomethylated xanthines are preferentially oxidized to the corresponding uric acid derivatives by XOR. In contrast, dimethylated and trimethylated xanthines, such as theophylline (1,3-dimethylxanthine) and caffeine (1,3,7-trimethylxanthine), are oxidized to the corresponding uric acid derivatives primarily by CYP. Through 2 sequential N-demethylation reactions, CYP converts caffeine to 1-methylxanthine, which is converted by XOR to 1-methyluric acid. The urinary ratio of 1-methylxanthine to 1-methyluric acid provides an in vivo marker of XOR activity.

The mechanism of catalysis by molybdoenzymes depicted in Fig. 6-27 is based on studies of XOR by Okamoto et al. (2004), Amano et al. (2007), and Alfaro and Jones (2008). The molybdopterin active center of the enzyme consists of molybdenum (Mo) bound to the following groups: pterin cofactor, thioxo (=S), oxo (=O), and hydroxyl (–OH). For simplicity, this cofactor will be written as S=Mo–OH. The hydroxyl group (S=Mo–OH) is activated by a nearby glutamic acid residue in the enzyme to S=Mo–O⁻, which initiates a concerted nucleophilic attack on the substrate at the electron-deficient carbon atom double bonded to the nitrogen (H–CR₁=N-R₂) with hydride transfer from the α-carbon to the thioxo group to form the intermediate (HS–Mo–O–CR₁=N-R₂). The oxygen bound to the α-carbon in the substrate (eg, xanthine) is replaced by oxygen from water to complete formation of product (eg, uric acid). The resting enzyme is restored by the removal of 2 reducing equivalents, which may be transferred to NAD⁺ (in the case of XD) or oxygen (in the case of XO), as shown in Fig. 6-27. In the XO mode, XOR (and AO, which exists only in the XO mode) can transfer both reducing equivalents to a single molecule of oxygen to produce hydrogen peroxide or it can transfer the 2 electrons to 2 molecules of oxygen to produce superoxide anion. In addition to the concerted mechanism described above, a stepwise mechanism involving a tetrahedral intermediate is also possible (Alfaro and Jones, 2008).

Aldehyde Oxidase

The term AO is something of a misnomer, in that this enzyme does not solely act on aldehyde-containing substrates (see Table 6-7). AO is the second of 2 molybdoenzymes that play an important role in xenobiotic biotransformation, the other being XOR (discussed in the preceding section). Whereas XOR exists in 2 forms, a dehydrogenase form (XD) that relays electrons to NAD⁺ and an oxidase form (XO) that relays electrons to molecular oxygen, AO exists only in the oxidase form because it lacks an NAD⁺ binding site (Garattini and Terao, 2011). Another significant difference between these 2 molybdoenzymes is that high levels of XOR appear to be widely distributed throughout the body, whereas high levels of AO (or at least its mRNA) are found in the liver and adrenal gland, with somewhat less expression in the small and large intestines, ovary, prostate, proximal-, distal-, and collecting tubules of the kidney, the epithelia of trachea and bronchium, and the alveolar cells of the lung, with only detectable levels of transcript in the endocrine tissues, esophagus, pancreas, brain, and few other tissues, at least in humans (Pryde et al., 2010; Garattini and Terao, 2011). The human AOX1 gene is complex with 35 exons, and exists as an active homodimer. There is also considerable interindividual variability in the levels of AO in humans, which may be due to genetic (as discussed further below) as well as environmental factors such as age, disease state (eg, cancer, inflammatory conditions leading to altered levels of various cytokines), smoking, and drug use (Pryde et al., 2010). Apart from these differences, many of the features of XOR apply to AO, including subcellular location (cytosol), enzyme structure and cofactor composition, typical mechanism of catalysis, preference for oxidizing sp²-hybridized carbon atoms adjacent to the nitrogen atoms in nitrogen heterocycles, and its preference for oxidizing aromatic aldehydes over aliphatic aldehydes. Other AO substrate types include nitro- or nitroso-containing compounds as well as aldehyde or iminium ion intermediates formed by CYP, ADHs, or MAOs (Pryde et al., 2010). Furthermore, AO also transfers electrons to molecular oxygen, which can generate ROS (eg, superoxide anion, which dismutates to hydrogen peroxide) and lead to oxidative stress and lipid peroxidation (Garattini and Terao, 2011). However, in contrast to XOR, the physiological functions of AO remain largely unknown, although endogenous compounds such as indole-3-acetate, retinaldehyde, retinoic acid, nicotinamide, and pyridoxal are substrates for AO (Garattini and Terao, 2011).

Many SNPs in the human AOX1 gene have been described in the NCBI dbSNP database (http://www.ncbi.nlm.nih.gov/projects/SNP). In a population of 180 Italian individuals, relatively frequent nonsense, synonymous, or missense SNPs in AO have been described, with a total of 51 SNPs identified, with 10 individuals being homozygous for a SNP (Hartmann et al., 2012). Two of the SNPs resulted in a fast metabolizer phenotype (FMs; N₁₁₃₅S, frequency = 2.9%, and H₁₂₉₇R, frequency = 5.3%) and 1 in a PM phenotype (ie, R₉₂₁H, frequency = 0.3%). In addition, a relatively frequent nonsense mutation in exon 5 (Y₁₂₆stop, frequency = 0.026) was identified that results in a very short and nonfunctional 126-amino acid protein, but no subjects were homozygous for this loss-of-function mutation. An important point about these polymorphisms described so far is that the effects appear to be somewhat substrate-dependent based on the functional characterization of purified, heterologously expressed variants described above with the AO substrates benzaldehyde, phthalazine, phenathridine, and chloroquinazolinone (Hartmann et al., 2012).

Substrates for AO do not appear to correlate with any specific area of “chemical space” based on physicochemical properties, which have a wide variety of lipophilicity, polar surface area, polarity, and structure, in contrast to P450 substrates, for which there is a high correlation between higher log P and P450-mediated metabolism as measured by microsomal stability in the presence of NADPH (see Table 6-7 for example substrates) (Pryde et al., 2010). Therefore, although it has been assumed that, in general, xenobiotics that are good substrates for AO are poor substrates for CYP, and vice versa (Rettie and Fisher, 1999), there is some overlap, and structural motifs alone may be a better predictor of AO substrates (Pryde et al., 2010). Naphthalene (with no nitrogen atoms) is oxidized by CYP, but not by AO, whereas the opposite is true of pteridine (1,3,5,8-tetraazanaphthalene), which contains 4 nitrogen atoms. The intermediate structure, quinazolone (1,3-diazanaphthalene), is a substrate for both enzymes. This complementarity in substrate specificity reflects the opposing preference of the 2 enzymes for oxidizing carbon atoms; CYP prefers to oxidize carbon atoms with high electron density, whereas AO (and XOR) prefers to oxidize carbon atoms with low electron density. Because of their opposing mechanisms of action (electrophilic for CYP, nucleophilic for AO), the common practice in drug development of adding a fluorine atom to block a site of oxidation by CYP (to improve metabolic stability) can potentially increase the rate of metabolism by AO. In nearly all cases, oxidation of aromatic heterocycles by AO occurs at the α-carbon but occasionally, depending on ring structure, the γ-carbon (or even the β-carbon) has the lowest electron density and is the site of oxidation. Examples of γ-carbon (or β-carbon) oxidation by AO include N-methylnicotinamide, quinolone, cinnoline, and N-[(2′-dimethylamino)ethyl]acridine-4-carboxamide (DACA) (Schofield et al., 2000; Testa and Krämer, 2008, 2010). In terms of its role in drug metabolism, it is reasonable to conclude that human AO does play—and will continue to play—a significant role in the oxidation of a wide range of structurally diverse nitrogen heterocylic substrates (Pryde et al., 2010).

As shown in Figs. 6-23 and 6-26, AO can oxidize a number of substituted pyrroles, pyridines, pyrimidines, purines, pteridines, and iminium ions (see Table 6-7 for additional examples) by the mechanism described for XOR in the previous section (Alfaro and Jones, 2008). AO can oxidize aldehydes to their corresponding carboxylic acids, but the enzyme shows a marked preference for aromatic aldehydes (eg, benzaldehyde, tamoxifen aldehyde). Consequently, AO contributes negligibly to the oxidation of aliphatic aldehydes, such as acetaldehyde. Rodrigues (1994) found that, in a bank of human liver samples, AO activity toward N¹-methylnicotinamide varied more than 40-fold, whereas activity toward 6-methylpurine varied less than 3-fold. Although this suggests human liver cytosol contains 2 or more forms of AO, subsequent Southern blot analysis has provided evidence for only a single copy of the AO gene in humans (Terao et al., 1998), although genomic analysis of chromosome 2q has revealed the presence of 2 nearby pseudogenes (Garattini and Terao, 2011). Species differences in the number of functional AOX genes are discussed further below.

The substrate specificity of AO differs among mammalian species, often leading to erroneous conclusions about the potential metabolic stability of certain drugs in humans, as detailed below. AO activity among animal species can vary depending on the particular substrate, but in general it is high in monkeys and humans, low in rats, and essentially absent in dogs (because the latter only express AOX3L1 and AOX4) (Garattini and Terao, 2011). These differences have been ascribed to the size of the active site (Pryde et al., 2010). However, large differences in activity have also been found among individual strains of rats and mice (eg, high activity in Sea:SD rats and low in WKA/Sea rats; large differences between Sprague–Dawley and Wistar as well as between C129/C57 and CB57B1/6J mice) (Pryde et al., 2010). Gender differences between rats and mice due to hormonal regulation further complicate attempts to extrapolate stability of AO substrates from toxicologically relevant species to humans (eg, 2- to 4-fold higher activity in male mice than in female) (Pryde et al., 2010). Finally, whereas humans encode only a single functional AO (namely, AOX1; the others being transcribed pseudogenes, ie, AOX3P and 3L1P), mice and rats possess 4 functional AOX genes (namely, AOX1, 3, 4, and 3L1) (Garattini and Terao, 2011). Rhesus monkeys possess 2 functional AO (AOX3 and AOX3L1). In the mouse and monkey, AOX3L1 is expressed only in nasal mucosa (Garattini and Terao, 2011). A further complication is the observation of species differences in the relative roles of AO and XOR in xenobiotic biotransformation. For example, the 6-oxidation of antiviral deoxyguanine prodrugs is catalyzed exclusively in rats by XOR, but by AO in humans (Rettie and Fisher, 1999). Similarly, only human AO catalyzes the oxidation of 6-deoxypenciclovir to penciclovir (Strolin-Benedetti, 2011).

AO in human liver has proven to be rather unstable, which complicates an in vitro assessment of species differences in AO activity in frozen stocks of human liver cytosol (Rodrigues, 1994; Rettie and Fisher, 1999; Garattini and Terao, 2011). In addition, few human cell lines express functional AO (eg, HepG2 cells express the AOX1 transcript, but lack the functional enzyme) (Garattini and Terao, 2011). Fresh or cryopreserved human hepatocytes are therefore most likely to have relevant levels of AO activity.

AO is the second of 2 enzymes involved in the formation of cotinine, a major metabolite of nicotine excreted in the urine of cigarette smokers. The initial step in this reaction is the formation of a double bond (C=N) in the pyrrole ring, which produces nicotine Δ^1′,5′-iminium ion. Like nicotine, several other drugs are oxidized either sequentially or concomitantly by CYP and AO, including quinidine, azapetine, cyclophosphamide, carbazeran, and prolintane. Other drugs that are oxidized by AO include bromonidine, citalopram, proprionaldehyde, O⁶-benzylguanine, 6-mercaptopurine, metyrapone, quinine, pyrazinamide, methotrexate, vanillin, isovanillin, zaleplon, and famciclovir (an antiviral prodrug that is converted by AO to penciclovir) (for additional examples, see Table 6-7). Several pyrimidine derivatives are oxidized by AO, including 5-ethyl-2(1H)-pyrimidone, which is converted by AO to 5-ethinyluracil. Like sorivudine, 5-ethinyluracil is a metabolism-dependent (suicide) inactivator of DPD (see Fig. 6-17).

Raloxifene is an extraordinarily potent inhibitor of human AO (K_i values as low as 0.9 nM), whereas it is at least 3 orders of magnitude less potent in other species (Obach, 2004; Pryde et al., 2010). Perphenazine and menadione are also potent inhibitors of AO (IC₅₀ ~0.03 and ~0.2 μM) and are often used together with allopurinol to discriminate between AO- and XOR-catalyzed reactions. The ability of proadifen to inhibit AO is noteworthy because this methadone analog, commonly known as SKF 525A, is widely used as a CYP inhibitor. Hydralazine is an irreversible (time-dependent) inhibitor of AO. Nitroso-imidacloprid has been characterized as a mechanism-based inhibitor of rabbit AO with a K_I value of 1.3 mM and k_inact of 0.35/min (Dick et al., 2007). Several other inhibitors of AO have also been described, but raloxifene is probably the most useful for in vitro studies to evaluate the contribution of AO to the metabolism of a substrate (Pryde et al., 2010). For instance, if a substrate that undergoes oxidative metabolism is found (1) to have higher in vitro turnover in human hepatocytes than in human liver microsomes, (2) is oxidized in human liver S9 or cytosol in the absence of NADPH, and (3) contains a structure amenable to oxidation by aldehyde oxidase, then raloxifene can be used to confirm involvement of aldehyde oxidase in the metabolism of the substrate (Pryde et al., 2010).

Under certain conditions, AO and XOR can also catalyze the reduction of xenobiotics, including azo-reduction (eg, 4-dimethylaminoazobenzene), nitro-reduction (eg, 1-nitropyrene, imidacloprid and other nitroguanidines, nitromethylenes), N-oxide reduction (eg, S-(−)-nicotine-1′-N-oxide, imipramine N-oxide, and cyclobenzaprine N-oxide), nitrosamine reduction (eg, N-nitrosodiphenylamine), hydroxamic acid reduction (eg, N-hydroxy-2-acetylaminofluorene [NOH-AAF]), sulfoxide reduction (eg, sulindac; see Fig. 6-15, fenthion sulfoxide), quinone reduction (eg, diethylstilbestrol quinone), epoxide reduction (eg, B[a]P, 4,5-oxide), and heterocycle reduction (eg, ziprasidone and zonisamide) (Testa and Krämer, 2008, 2010; Pryde et al., 2010). Oximes (C=NOH) can also be reduced by AO to the corresponding ketimines (C=NH), which may react nonenzymatically with water to produce the corresponding ketone or aldehyde (C=O) and ammonia (Testa and Krämer, 2008, 2010). The reduction of ziprasidone by AO is shown in Fig. 6-4, and examples of other reductive reactions catalyzed by AO are shown in Fig. 6-23. Xenobiotic reduction by AO in vitro requires anaerobic conditions or the presence of a reducing substrate, such as N¹-methylnicotinamide, 2-hydroxypyrimidine, or benzaldehyde. These “cosubstrates” reduce the enzyme, which in turn catalyzes azo-reduction, nitro-reduction, etc, by relaying electrons to xenobiotics (rather than molecular oxygen). These unusual requirements make it difficult to assess the degree to which AO functions as a reductive enzyme in vivo.

Amine Oxidases

Human amine oxidases can be divided into 2 classes: (I) the familiar FAD-containing mitochondrial MAOs, MAO-A and -B (gene symbols MAOA and MAOB), as well as polyamine oxidase (PAO; gene symbol PAOX) and spermine oxidase (gene symbol SMOX); and (II) the copper-containing amine oxidases (CuAOs) that contain a tightly bound Cu^II and a quinone residue (typically 2,4,5-trihydroxyphenylalanine quinone [TPQ]) as the redox cofactor (Largeron, 2011). The latter class of human amine oxidases belongs to the larger class of so-called quinoproteins class, which are present in plants, animals, fungi, yeast, and bacteria. In humans the CuAOs include (1) the intracellular lysyl oxidase (gene symbol LOX), (2) diamine oxidase (DAO; also known as AOC1 [from “amine-oxidase, copper-containing] and DAO [gene symbol ABP1, from “amiloride binding protein”]), and (3) the 2 known so-called semicarbazide-sensitive copper-containing amine oxidases (or SSAOs), which are located in plasma and the plasma membranes of various tissues (gene symbols AOC2 for the retina-specific amine oxidase and AOC3 for the gene encoding what is probably the originally identified SSAO, which is the same as vascular adhesion protein [VAP1]) (Largeron, 2011).

Class I amine oxidases differ markedly from class II amine oxidases in their substrate specificity, but there is some overlap. For instance, dopamine and tyramine are good substrates for MAOs, SSAOs, and DAO, whereas benzylamine is a good substrate for both MAOs and the CuAOs (Largeron, 2011). Features of some of these enzymes will be discussed below, with examples of reactions catalyzed by MAO, DAO, and PAO shown in Fig. 6-28.

Class I Amine Oxidases: MAO-A and B, PAO, and SMOX

MAO (gene symbols MAOA and MAOB), spermine oxidase (gene symbol SMOX), and PAO (gene symbol PAOX) are involved in the oxidative deamination of primary, secondary, and tertiary amines (Benedetti, 2001; Agostinelli et al., 2004; Edmondson et al., 2004). Substrates for these enzymes include several naturally occurring amines, such as the monoamine serotonin (5-hydroxytryptamine), and monoacetylated derivatives of the polyamines spermine and spermidine. A number of xenobiotics are substrates for these enzymes, particularly MAOs. Oxidative deamination of a primary amine produces ammonia and an aldehyde, whereas oxidative deamination of a secondary amine produces a primary amine and an aldehyde. (The products of the former reaction—that is, an aldehyde and ammonia—are those produced during the reductive biotransformation of certain oximes by AO, as described in the section “Aldehyde Oxidase.”) The aldehydes formed by MAO are usually oxidized further by other enzymes to the corresponding carboxylic acids, although in some cases they are reduced to alcohols.

MAO is located throughout the brain, and is present at high levels in the liver, kidney, intestine, heart, blood platelets, and blood vessels (but absent from erythrocytes) in the outer membrane of mitochondria (although some activity has been found in other cellular compartments, possibly because human liver microsomes prepared from frozen tissue are contaminated with the outer mitochondrial membrane) (Pearce et al., 1996a; Testa and Krämer, 2008, 2010). Although MAOs are expressed in most tissues, there are some that express predominantly one form over the other. For example, MAO-A is the predominant form in human placenta, whereas MAO-B is the predominant form in human platelets, lymphocytes, and chromaffin cells (Strolin Benedetti et al., 2007). MAO substrates include milacemide (Fig. 6-28), a dealkylated metabolite of propranolol (Fig. 6-28), primaquine, haloperidol, citalopram, sertraline, doxylamine, 1-methyl-4-phenyl-1,2,5,6-tetrahydropyridine (MPTP), β-phenylethylamine, tyramine, catecholamines (eg, dopamine, norepinephrine, epinephrine), tryptophan derivatives (tryptamine, serotonin), and tryptophan analogs known as triptans, which include the antimigraine drugs almotriptan, sumatriptan, zolmitriptan, and rizatriptan (Strolin Benedetti et al., 2007; Testa and Krämer, 2008, 2010). In the case of the triptans, when a nitrogen exists in either a pyrrolidine or a piperidine ring, MAOs are not involved in their metabolism. Also, as a general rule, the introduction of a methyl group in the α-position relative to the nitrogen atom in what would otherwise be an MAO substrate renders it resistant to oxidative deamination by MAO (Strolin Benedetti et al., 2007).

MAO-A preferentially oxidizes serotonin (5-hydroxytryptamine), norepinephrine, and the dealkylated metabolite of propranolol, and is preferentially inhibited by clorgyline, whereas MAO-B preferentially oxidizes β-phenylethylamine, benzylamine, and citalopram, and is preferentially inhibited by l-deprenyl (selegiline) (Strolin Benedetti et al., 2007). Species differences in the substrate specificity of MAO have been documented. For example, dopamine is oxidized by MAO-B in humans, but by MAO-A in rats, and by both enzymes in several other mammalian species. The distribution of MAO in the brain shows little species variation, with the highest concentration of MAO-A in the locus coeruleus, and the highest concentration of MAO-B in the raphe nuclei. MAO-A is expressed predominantly in catecholaminergic neurons, whereas MAO-B is expressed largely in serotonergic and histaminergic neurons and glial cells. The distribution of MAO throughout the brain does not always parallel that of its substrates. For example, serotonin is preferentially oxidized by MAO-A, but MAO-A is not found in serotonergic neurons.

The mechanism of catalysis by MAO is illustrated as follows:

RCH₂NH₂ + FAD → RCH=NH + FADH₂;

RCH=NH + H₂O → RCHO + NH₃;

FADH₂ + O₂ → FAD + H₂O₂.

The substrate is oxidized by the enzyme, which itself is reduced (FAD → FADH₂). The oxygen incorporated into the substrate is derived from water, not molecular oxygen; hence, the enzyme functions as a true oxidase. The catalytic cycle is completed by reoxidation of the reduced enzyme (FADH₂ → FAD) by oxygen, which generates hydrogen peroxide (which may be a cause of oxidative stress). The initial step in the catalytic cycle is abstraction of hydrogen from the α-carbon adjacent to the nitrogen atom; hence, the oxidative deamination of xenobiotics by MAO is generally blocked by substitution of the α-carbon. For example, amphetamine and other phenylethylamine derivatives carrying a methyl group on the α-carbon atom are not oxidized well by MAO. (Amphetamines can undergo oxidative deamination, but the reaction is catalyzed by CYP.) The abstraction of hydrogen from the α-carbon adjacent to the nitrogen atom can occur stereospecifically; therefore, only one enantiomer of an α-substituted compound may be oxidized by MAO. For example, whereas MAO-B catalyzes the oxidative deamination of both R- and S-β-phenylethylamine, only the R-enantiomer is a substrate for MAO-A. The oxidative deamination of the dealkylated metabolite of propranolol is catalyzed stereoselectively by MAO-A, although in this case the preferred substrate is the S-enantiomer (which has the same absolute configuration as the R-enantiomer of β-phenylethylamine) (Benedetti and Dostert, 1994).

Clorgyline and l-deprenyl (selegiline) are metabolism-dependent inhibitors (ie, mechanism-based or suicide inactivators) of MAO-A and MAO-B, respectively. Both enzymes are irreversibly inhibited by phenelzine, a hydrazine that can be oxidized either by abstraction of hydrogen from the α-carbon atom, which leads to oxidative deamination with formation of benzaldehyde and benzoic acid, or by abstraction of hydrogen from the terminal nitrogen atom, which leads to formation of phenylethyldiazene and covalent modification of the enzyme, as shown in Fig. 6-28.

MAO has received considerable attention for its role in the activation of MPTP to a neurotoxin that causes symptoms characteristic of Parkinson disease in humans and monkeys but not rodents (Gerlach et al., 1991). In 1983, Parkinsonism was observed in young individuals who, in attempting to synthesize and use a narcotic drug related to meperidine (Demerol^®), instead synthesized and self-administered MPTP, which causes selective destruction of dopaminergic neurons in the substantia nigra. MPTP crosses the blood–brain barrier, where it is oxidized by MAO in the astrocytes (a type of glial cell) to 1-methyl-4-phenyl-2,3-dihydropyridine (MPDP⁺), which in turn autoxidizes to the neurotoxic metabolite, 1-methyl-4-phenylpyridine (MPP⁺), as shown in Fig. 6-29. Because it is transported by the dopamine transporter, MPP⁺ concentrates in dopaminergic neurons, where it impairs mitochondrial respiration. The neurotoxic effects of MPTP can be blocked with pargyline (an inhibitor of both MAO-A and MAO-B) and by l-deprenyl (a selective inhibitor of MAO-B) but not by clorgyline (a selective inhibitor of MAO-A). This suggests that the activation of MPTP to its neurotoxic metabolite is catalyzed predominantly by MAO-B. This interpretation is consistent with the finding that MAO-B knockout mice (ie, transgenic mice that lack MAO-B) do not sustain damage to the dopaminergic terminals of nigrostriatal neurons after MPTP treatment (Shih et al., 1999; Quinn et al., 2007).

Both genetic and environmental factors appear to play important roles in the etiology of Parkinson disease. Apart from MPTP, Parkinsongenic neurotoxins to which humans are exposed have not been identified unequivocally; hence, the environmental factors that cause Parkinson disease remain to be identified. It is interesting that the bipyridyl herbicide, paraquat, is similar in structure to the toxic metabolite of MPTP, as shown in Fig. 6-29. Some epidemiological studies have shown a positive correlation between herbicide exposure and the incidence of Parkinsonism in some but not all rural communities. Haloperidol can also be converted to a potentially neurotoxic pyridinium metabolite (Subramanyam et al., 1991).

MAO-B may be among the genetic factors that affect susceptibility to Parkinson disease. MAO-B activity in the human brain increases with aging, perhaps due to a proliferation of glial cells. It has been proposed that increased oxidation of dopamine by MAO-B in the elderly may lead to a loss of dopaminergic neurons in the substantia nigra, which underlies Parkinson disease. Such damage may be caused by the oxidative stress associated with the oxidative deamination of dopamine by MAO-B. In support of this proposal, patients with Parkinson disease have elevated MAO-B activity in the substantia nigra, and the MAO-B inhibitors l-deprenyl (selegiline), lazabemide, and rasagiline provide some symptomatic relief and can delay the progression of symptoms and need for levodopa (Sano et al., 1997; Löhle and Reichmann, 2011). Furthermore, there are allelic variants of MAO-B, some of which (such as alleles 1, A, B4, and G) appear to be associated with an increased risk of developing Parkinson disease, especially in women (Shih et al., 1999; Kang et al., 2006). No such association has been found between Parkinson disease and MAO-A gene polymorphisms (Williams-Gray et al., 2009). Cigarette smoking, which carries a number of health risks, nevertheless provides some protection against Parkinson disease (Gorell et al., 1999; Gu et al., 2010). Although the mechanism of protection remains to be determined, it is interesting to note that cigarette smokers are known to have decreased levels of MAO-B (and MAO-A) (Shih et al., 1999), the degree of which is proportional to cigarette usage (ie, it is dose related) (Whitfield et al., 2000).

MAO-A knockout mice have elevated brain levels of serotonin and a distinct behavioral syndrome, including enhanced aggression in adult males. The enhanced aggressive behavior exhibited by MAO-A knockout mice is consistent with the abnormal aggressive behavior in individuals who lack MAO-A activity due to a point mutation in the MAO-A gene (Shih et al., 1999). Other polymorphisms in the MAO-A gene appear to be risk factors for alcoholism among Euro-Americans and Han Chinese (Shih et al., 1999). MAO-B may also be a factor in alcoholism inasmuch as alcoholics (especially male Type 2 alcoholics) tend to have lower MAO activity in platelets, which only contain MAO-B. However, MAO-B activity is not lower in alcoholics when cigarette smoking status is taken into account, which suggests that MAO-B activity tends to be lower in alcoholics because smoking and alcohol dependence are strongly associated with each other (Whitfield et al., 2000).

Although not present in mitochondria, the PAO, PAOX (approved name: “polyamine oxidase [exo-N4-amino]”), and SMOX (approved name: “spermine oxidase”) resemble MAO in their cofactor requirement and basic mechanism of action. These enzymes use oxygen as an electron acceptor, which results in the production of hydrogen peroxide. However, spermine and spermidine are first acetylated by spermidine/spermine N¹-acetyltransferases (gene symbols SAT1 and 2), and subsequently oxidized by PAOX, which produces not only hydrogen peroxide but also 3-acetamidopropanal (Hakkinen et al., 2010). SMOX has very different substrate specificity, preferring nonacetylated polyamines (but not spermidine). The polyamines, spermine, spermidine, and the diamine precursor, putrescine (discussed below), are ubiquitous in mammalian cells. The intracellular levels of these amines are strictly controlled by several enzymes (including PAOX and SMOX) as well as various transporters (Hakkinen et al., 2010). Dysregulation of polyamine metabolism is associated with cancer and other diseases. Because of differences in polyamine regulation in various parasites, several xenobiotic polyamine synthesis inhibitors or substrates have been investigated as antiparasitic, chemopreventive, or chemotherapeutic drugs, including N-alkylated polyamine analogs such as diethylnorspermine, as well as some N-benzyl-substituted polyamines (Hakkinen et al., 2010).

The MAO inhibitor pargyline also inhibits PAOX. The anticonvulsant milacemide is one of the few xenobiotic substrates for PAOX (the structure of which is unrelated to polyamines), although it is also a substrate for MAO (Fig. 6-28) (Strolin Benedetti et al., 1992). By converting milacemide to glycine (via glycinamide), MAO plays an important role in the anticonvulsant therapy with milacemide (Benedetti and Dostert, 1994).

Class II Amine Oxidases (CuAOs): SSAOs, DAO, and LOX

The term SSAO is confusing because the term has been previously used to refer only to amine oxidases that are active only toward primary amines, although this has not been proven, and several publications that mention SSAOs actually describe DAO. There are at least 3 types of SSAOs (Largeron, 2011). The first (which probably includes the original SSAO) has the recommended name PrAOs, for “primary amine oxidases,” and includes gene symbols AOC2 and AOC3, the latter of which is probably the originally identified SSAO. The second SSAO is DAO (represented by AOC1; gene symbol ABP1, for “amiloride-binding protein”). A third type of SSAO is lysyl oxidase (LOX) (Largeron, 2011). All of these enzymes are CuAOs. In general, the SSAOs (ie, PrAOs) do oxidize primary monoamines and have little or no activity toward diamines. DAO (ABP1) oxidizes both diamines such as histamine and some primary amines. Neither DAO nor SSAOs have activity toward secondary and tertiary amines (Largeron, 2011). These PrAOs can be either soluble or membrane associated and are expressed in the liver, lung, lymph nodes, small intestine, and blood plasma (Testa and Krämer, 2008, 2010). The PrAO/SSAO known as AOC3 is a 180 kDa dimeric endothelial transmembrane glycoprotein that mediates leukocyte extravasation, for which reason it is widely known as the “vascular adhesion protein 1” [VAP1] (Kaitaniemi et al., 2009).

SSAOs/PrAOs catalyze fundamentally the same reaction catalyzed by MAO (Kaitaniemi et al., 2009):

RCH₂NH₂ + H₂O + O₂ → RCHO + H₂O₂ + NH₃.

The endogenous roles of the CuAOs include regulating levels of endogenous and exogenous levels of primary monoamines and polyamines by catalyzing their oxidative deamination (eg, methylamine, putrescine, and cadaverine [and specific lysine residues of extracellular matrix proteins in the case of LOX]), which produces an aldehyde, hydrogen peroxide, and ammonia (Yraola et al., 2009; Largeron, 2011). Because all of the latter species are potentially cytotoxic, these metabolic pathways can activate endobiotics and xenobiotics. The activity of plasma CuAOs is generally increased in various disease states, including Alzheimer disease, congestive heart failure, cirrhosis, Type 1 and 2 diabetes, atherosclerosis, and other inflammatory conditions, which leads to the overproduction of toxic species, especially aldehydes and hydrogen peroxide, and may further contribute to the disease process (Largeron, 2011).

Although CuAOs do not play a large role in xenobiotic oxidation, there is some evidence for the involvement of SSAOs in the oxidative deamination of a few xenobiotics, including allylamine, primaquine, mescaline, amlodipine, and amifostine (Testa and Krämer, 2008, 2010; Largeron, 2011). The calcium channel blocker amlodipine is first oxidized (mainly by CYP3A4) to dehydroamlodipine, with subsequent deamination to an aldehyde (consistent with the action of an SSAO) and finally oxidation to the corresponding carboxylic acid by an ALDH and/or AO (Largeron, 2011). Amifostine is a thiophosphate prodrug used as a cytoprotective adjuvant used in cancer chemotherapy and radiotherapy; it is rapidly converted to 2-[3-aminopropylamino] ethanethiol by alkaline phosphatase followed by oxidative deamination catalyzed by one or more CuAOs to an aldehyde that spontaneously decomposes to acrolein and cysteamine (Largeron, 2011). Benzylamine (and several analogs) may be used in vitro as a substrate for SSAOs (Yraola et al., 2009). MAO-B can also catalyze the deamination of benzylamine, but MAO-B acts by removal of the pro-R H-atom, whereas SSAOs act by removal of the pro-S H-atom (Testa and Krämer, 2008, 2010). There also appear to be some substrate selectivity between the individual SSAOs, AOC2 and 3, in that the preferred in vitro substrates of AOC2 (highly expressed in the retina) include 2-phenylethylamine, tryptamine, and tyramine, rather than the benzylamines and methylamine, which appear to be the preferred substrates for AOC3 (Kaitaniemi et al., 2009).

DAO is a cytosolic, copper-containing, pyridoxal phosphate-dependent enzyme that is highly expressed in lung, liver, kidney, intestine, and placenta, among other tissues, for which 2 crystal structures have been described (McGrath et al., 2010). It is located in the extracellular space (secreted form) or in the endoplasmic reticulum (Testa and Krämer, 2008, 2010). Its preferred substrates include histamine and simple alkyl diamines with a 4- or 5-carbon chain length such as putresine (1,4-diaminobutane) and cadaverine (1,5-diaminopentane) (McGrath et al., 2010). Diamines with carbon chains longer than 9 are not substrates for DAO, although they can be oxidized by MAO. DAO or a similar enzyme is present in cardiovascular tissue and appears to be responsible for the cardiotoxic effects of allylamine, which is converted by oxidative deamination to acrolein. Although histamine is a substrate for DAO (which catalyzes oxidative deamination of the primary amine to imidazole acetaldehyde by the secreted DAO) (McGrath et al., 2010), there is little DAO in brain (nor is there a receptor-mediated uptake system for histamine, in contrast to other neurotransmitters). For this reason, the major pathway of histamine metabolism in the brain is by methylation (see the section “Methylation”).

Aromatization

The conversion of MPTP to MPP⁺ (Fig. 6-29) is an example of a xenobiotic whose oxidation involves the introduction of multiple double bonds to achieve some semblance of aromaticity (in this case, formation of a pyridinium ion). Aromatization of xenobiotics is an unusual reaction, but some examples have been documented. A mitochondrial enzyme in guinea pig and rabbit liver can oxidize several cyclohexane derivatives to the corresponding aromatic hydrocarbon, as shown in Fig. 6-30 for the aromatization of cyclohexane carboxylic acid (hexahydrobenzoic acid) to benzoic acid. Mitochondria from rat liver are less active, and those from cat, mouse, dog, monkey, and human are inactive (Mitoma et al., 1958). The reaction requires magnesium, CoA, oxygen, and ATP. The first step is the formation of hexahydrobenzoyl-CoA, which is then dehydrogenated to the aromatic product. Glycine stimulates the reaction, probably by removing benzoic acid through conjugation to form hippuric acid. The conversion of androgens to estrogens involves aromatization of the A-ring of the steroid nucleus. This reaction is catalyzed by CYP19A1, one of the CYP enzymes involved in steroidogenesis. The major substrate of CYP19A1 is androstenedione, which is converted to estrone by 3 successive oxidation steps (Kuhl and Wiegratz, 2007). Gut microflora can catalyze aromatization reactions, as shown in Fig. 6-1 for the conversion of quinic acid to benzoic acid.

Evidence for the apparent aromatization of a few xenobiotics by other P450 enzyme has also been described. For instance, 19-nortestosterone derivatives can be converted to aromatic metabolites, namely, ethinylestradiol (from norethisterone and norethynodrel) and 7α-methylethinylestradiol (from tibolone) (Kuhl and Wiegratz, 2007). Formation of these A-ring aromatized steroids occurs in the liver, which does not express the steroidogenic enzyme CYP19A1 (aromatase). In addition, indoline (and some derivatives) is aromatized to indole in human liver microsomes, and mainly by recombinant human CYP3A4 (as well as CYP1A2, 2B6, 2C19, 2D6, and 2E1) by a dehydrogenation pathway (Sun et al., 2007). The aromatization mechanism involves two one-electron oxidations, which is different from that catalyzed by CYP19A1, which catalyzes 2 sequential carbon oxidations followed by cleavage of a carbon–carbon bond (Sun et al., 2007). The indoline-containing diuretic, indapamide, is also dehydrogenated to the aromatic indole by CYP3A4 (Sun et al., 2009). This type of dehydrogenation can lead to P450 inactivation, as exemplified by the mechanism-based inhibition of CYP3A4 by the 3-alkylindole-containing TNFα inhibitor, SPD-304, which forms an electrophilic 3-methylenindolenine (Sun and Yost, 2008).

Peroxidase-Dependent Cooxidation

The oxidative biotransformation of xenobiotics generally requires the reduced pyridine nucleotide cofactors, NADPH and NADH, although oxidation by AO and MAO is a notable exception. Another exception is xenobiotic biotransformation by peroxidases, which are heme-containing enzymes that couple the reduction of hydrogen peroxide (or a lipid hydroperoxide) to the one-electron oxidation of other substrates (O'Brien, 2000; Tafazoli and O'Brien, 2005). Several different peroxidases catalyze the biotransformation of xenobiotics (in addition to performing important physiological functions), and these enzymes occur in a variety of tissues and cell types. Peroxidases do not play an important role in the first-pass metabolism or clearance of drugs and most other xenobiotics because their contribution is usually negligible compared with CYP and other oxidative enzymes. However, peroxidases do play an important role in xenobiotic toxicity, especially the activation of drugs associated with idiosyncratic hepatotoxicity, blood dyscrasias (eg, agranulocytosis, neutropenia, aplastic anemia, and thrombocytopenia), and skin rashes, and the activation of xenobiotics (including the activation of proximate carcinogens to ultimate carcinogens) in skin, bladder, bone marrow, and various other extrahepatic tissues. As shown in Fig. 6-2, peroxidases form reactive intermediates by converting nitrogen- and sulfur-containing xenobiotics to heteroatom-centered radicals (Tang and Lu, 2010; Walsh and Miwa, 2011).

In humans, the family of peroxidases includes myeloperoxidase (gene symbol MPO), eosinophil peroxidase (EPO; gene symbol EPX), lactoperoxidase (gene symbol LPO), thyroid peroxidase (gene symbol TPO), catalase, 2 forms of prostaglandin H synthase (also known as PHS1 and 2 or cyclooxygenase-1 [COX-1] and COX-2; gene symbols PTGS1 and 2), as well as the selenium-containing glutathione peroxidases (gene symbols GPX1-8). Some basic features of these peroxidases are summarized in Table 6-8. These peroxidases all have physiological functions: MPO, EPO, and LPO are lysosomal enzymes present in neutrophils, eosinophils, and secretory cells of exocrine glands, respectively. During infection, MPO and EPO are released into phagocytic vacuoles (granules) and into the plasma, whereas LPO is released into saliva and tears, where they kill microorganisms and thereby provide protection against infectious agents such as bacteria and parasites. (Peroxidases are also present in milk: LPO is the predominant peroxidase in cow's milk, whereas MPO is the predominant peroxidase in human milk.) The hydrogen peroxide required by MPO and EPO is produced by a membrane-bound NADPH oxidase that is activated by the presence of infectious agents. Unlike MPO, EPO, and LPO, which are soluble enzymes, TPO is a membrane-bound peroxidase located on the apical membrane of thyroid follicular cells; it catalyzes the iodination of tyrosine residues and the oxidative coupling of di-iodinated and monoiodinated tyrosine residues in thyroglobulin to form T3- and thyroxine (T4)–bound thyroglobulin, from which thyroid hormones are released. An NAD(P)H-oxidase known as p138 TOX (gene symbol DUOX2), which is also localized on the apical plasma membrane, produces the hydrogen peroxide required by TPO to synthesize thyroid hormones. PHS1 and PHS2 are the peroxidase component of COX-1 and COX-2; they are enzymes that convert arachidonic acid (and closely related fatty acids) to a variety of eicosanoids (prostaglandins, leukotrienes, thromboxane, and prostacyclin). In contrast to the other peroxidases, PHS1 and PHS2 do not require a source of hydrogen peroxide (although they can use it); hence, their activity is not dependent on an H₂O₂-generating NAD(P)H-oxidase. Catalase is also a peroxidase. This peroxisomal enzyme catalyzes the disproportionation of hydrogen peroxide to water and oxygen (2H₂O₂ → 2H₂O + O₂). At low concentrations of hydrogen peroxide, catalase can catalyze the oxidation of ethanol (see Fig. 6-25) and various other small molecules. GPXs are a family of selenium-containing enzymes that also detoxify hydrogen peroxide (and lipid hydroperoxides) by reducing it to water, which is associated with the formation of oxidized GSH as follows:

2GSH + HOOH → GS-SG + 2H₂O.

In mammalian peroxidases (in contrast to plant peroxidases), the heme prosthetic group is covalently attached to the enzyme. Iron is bound to 4 pyrrole nitrogen atoms with the nitrogen on the imidazole ring of histidine serving as the usual fifth ligand. The sixth coordination position is vacant so that peroxidases can interact with hydrogen peroxide (or other hydroperoxides), just as the sixth coordinate position of hemoglobin and myoglobin (both of which have low peroxidase activity) is available to bind molecular oxygen (Tafazoli and O'Brien, 2005). The oxidation of a xenobiotic (X → XO or 2XH → 2X^• + H₂O) by peroxidase involves the conversion of hydrogen peroxide to water or the conversion of a hydroperoxide to the corresponding alcohol (ROOH → ROH), during which the peroxidase (Fe^III) is converted to spectrophotometrically distinct states known as compound I and compound II, shown as follows for the conversion of a xenobiotic phenol (XOH) to the corresponding phenoxyl radical (XO^•) (Tafazoli and O'Brien, 2005):

Peroxidase + ROOH → compound I + ROH;

Compound I + XOH → compound II + XO^• + H⁺;

Compound II + XOH → peroxidase + XO^• + H₂O.

In this example, the hydroperoxide is converted to the corresponding alcohol (ROOH → ROH), and the peroxyl oxygen atom is reduced to water (H₂O) by the one-electron oxidation of 2 molecules of xenobiotic (XOH) to produce 2 xenobiotic radicals (XO^•). In some cases, the oxygen from the hydroperoxide is incorporated into the xenobiotic itself (XH → XOH). Examples of each reaction are given later in this section. The conversion of ROOH to ROH involves the release of an oxygen atom that coordinates with the heme iron (initially in the ferric or Fe^III state). This iron-bound oxygen formally contains only 6 (instead of 8) valence electrons, making it a powerful oxidizing species. One electron is removed from the iron to produce Fe^IV=O, and a second electron is removed from the tetrapyrrole ring to produce a π-porphyrin cation radical (Por^•+), which corresponds to compound I (Fe^IV=O–Por^•+). The transfer of an electron from a xenobiotic (XOH → XO^• + H⁺) to the porphyrin cation radical produces compound II (Fe^IV=O–Por). In some cases, such as PHS, the electron donated to the initial Fe=O complex comes directly or indirectly from an amino acid rather than the tetrapyrrole ring, and this accounts in part for some of the differences among the various peroxidases (such as the ability of TPO to catalyze the iodination of tyrosine residues and the subsequent formation of thyroid hormones).

Although peroxidases are renowned for catalyzing one-electron oxidation reactions, they can catalyze the 2-electron oxidation of iodide (I⁻) to hypoiodous acid (HOI) and iodine which, in the case of TPO, is important for the synthesis of thyroid hormones:

Peroxidase compound I(O) + I⁻ + H⁺ → HOI;

Peroxidase compound I(O) + 2 × I⁻ + H⁺ → I₂ (iodine) + OH⁻.

MPO, EPO, and LPO can all oxidize the pseudohalide thiocyanate (SCN⁻) faster than iodide, which in turn is oxidized faster than bromide, which is oxidized much faster than chloride. In fact, MPO is the only peroxidase with appreciable ability to oxidize chloride to hypochlorous acid (HOCl). There is an inverse relationship between the rate of conversion of halides to hypohalous acids and their physiological plasma concentrations: chloride = 100 to 140 mM, bromide = 20 to 100 μM, and iodide = 0.1 to 0.5 μM. Consequently, the relatively low rate of chloride oxidation is offset by the high levels of chloride such that about half (20%–70%) of the hydrogen peroxide produced by activated neutrophils is converted to HOCl, which reacts with GSH, proteins, thiols, amines, unsaturated fatty acids, and cholesterol, all of which disrupt cell membranes and lead to cell lysis and death (of both infectious organisms and host cells). Formation of HOCl by MPO is important not only from a physiological perspective but also from a drug metabolism and toxicity perspective, as discussed later in this section. The preferred “halide” substrate for MPO is thiocyanate, which is present in plasma at concentrations ranging from 20 to 120 μM, which is sufficiently high that oxidation of thiocyanate consumes the other half of the hydrogen peroxide produced by activated neutrophils. Whereas MPO oxidizes chloride to HOCl, which is cytotoxic, it oxidizes thiocyanate to thiocyanogen (SCN₂), which rapidly hydrolyzes to hypothiocyanic acid (HOSCN). HOSCN is much less reactive and cytotoxic than HOCl, such that the oxidation of thiocyanate by peroxidases is a mechanism to remove hydrogen peroxide without forming HOCl. This represents an important mechanism of hydrogen peroxide detoxication in saliva, which contains low levels of catalase but high levels (1-5 mM) of thiocyanate, and in the stomach, where the levels of thiocyanate in parietal cells are 3 times greater than plasma levels, which allows gastric peroxidase to inactivate hydrogen peroxide that otherwise stimulates gastric acid secretion by stimulating histamine release from mast cells. Inactivation of gastric peroxidase by aspirin, dexamethasone, indomethacin, or methimazole can result in gastric ulceration due to impaired detoxication of hydrogen peroxide and impaired synthesis of cytoprotective prostaglandins (O'Brien, 2000).

MPO and EPO can be distinguished by (1) their cellular distribution (neutrophils/leukocytes vs eosinophils); (2) their ability to oxidize halides (MPO is better than EPO at converting chloride to HOCl, whereas EPO is better than MPO at converting bromide to hypobromous acid), and (3) their sensitivity to inhibitors (MPO is more sensitive to the inhibitory effect of cyanide over azide, whereas the opposite is true of EPO).

The level of EPO (or a related peroxidase) in uterine epithelial cells is regulated by estrogen; uterine peroxidase activity is inducible several hundred-fold by estrogenic steroids, including the synthetic estrogens diethylstilbestrol and tamoxifen. The level of TPO in thyroid follicular cells is regulated by TSH. The enzyme is inactivated by a variety of ethylenethiourea drugs, such as propylthiouracil and methimazole (which is used as an antithyroid drug in patients with Graves disease), as well as a number of naturally occurring flavonoid/resorcinol compounds that also have antithyroid effects. By inactivating TPO and impairing thyroid hormone synthesis, these chemicals trigger a large and prolonged increase in TSH that in rodents (but apparently not in humans) can result in thyroid follicular cell tumor formation.

MPO is an abundant enzyme in granulocytes, where it accounts for 5% of the dry weight. It can be used as a diagnostic marker to differentiate myeloid leukemia from lymphoid leukemia. MPO is one of the peroxidases implicated in the formation of reactive metabolites of drugs that cause idiosyncratic agranulocytosis (or other blood dyscrasias), including clozapine, aminopyrine, vesnarinone, propylthiouracil, dapsone, remoxipride, sulfonamides, procainamide, amodiaquine, and ticlopidine (Liu and Uetrecht, 2000; O'Brien, 2000; Tafazoli and O'Brien, 2005; Testa and Krämer, 2008, 2010). The nitro-reduced metabolite of tolcapone can be converted to a reactive ortho-quinoneimine by MPO and potentially lead to hepatotoxicity (Testa and Krämer, 2008, 2010). Many of these drugs are aromatic amines, and both MPO and PHS have also been implicated in the activation of several carcinogenic aromatic amines, such as benzidine, methylaminoazobenzene, and aminofluorene. Furthermore, inactivating polymorphisms in MPO (such as MPO-463A) appear to afford protection against both the activation of aromatic amines and PAH in tobacco smoke and the tumor-promoting effects of hydrogen peroxide formed by tobacco smoke–activated neutrophils (O'Brien, 2000).

In the presence of hydrogen peroxide and chloride (which are converted to HOCl by MPO), activated neutrophils and monocytes (including Kupffer cells) oxidize the aforementioned drugs to reactive intermediates (such as nitrogen-centered radicals, hydroxylamines, N-chloramines, and, in the case of aromatic amines with a hydroxyl group in the para-position, quinoneimines or the corresponding semiquinoneimine radicals). In the case of ticlopidine, MPO converts the thiophene ring of this antiplatelet drug to a thiophene-S-chloride, a reactive metabolite that rearranges to 2-chloroticlopidine (minor) and dehydro-ticlopidine (major), or reacts with GSH, as shown in Fig. 6-31. When catalyzed by activated neutrophils, ticlopidine oxidation is inhibited by low concentrations of azide and catalase. When catalyzed by purified myeloperoxidase, ticlopidine oxidation requires hydrogen peroxide and chloride, although all components of this purified system can be replaced with HOCl. It is not known whether drugs that cause agranulocytosis are activated in the bone marrow by neutrophils or their precursors that contain myeloperoxidase, or are activated in neutrophils in the general circulation. In the latter case, agranulocytosis would presumably involve an immune response triggered by neoantigens formed in neutrophils by the covalent modification of cellular component by one or more of the reactive metabolites formed by MPO. MPO is also present in Kupffer cells, which release the enzyme in the liver upon activation by hydrogen peroxide. It is possible to increase the toxicity of hydralazine (which is associated with the development of hepatitis and centrilobular necrosis) toward cultured hepatocytes by the inclusion of a hydrogen peroxide–generating system (Tafazoli and O'Brien, 2008).

MPO has also been implicated in the formation of pro-oxidant phenoxyl radicals of etoposide, a topoisomerase inhibitor that can cause myelogenous leukemia, and both MPO and PHS have been implicated in the formation of phenoxyl radicals of benzene, an industrial solvent linked to bone marrow suppression and leukemia. MPO and PHS cannot oxidize benzene itself. Liver CYP converts benzene to phenol, which in turn is oxidized to hydroquinone, which can be converted to DNA-reactive metabolites by MPO in bone marrow leukocytes and by PHS in bone marrow. The myelosuppressive effect of benzene can be blocked by the PHS inhibitor, indomethacin, which suggests an important role for PHS-dependent activation in the myelotoxicity of benzene. The formation of phenol and hydroquinone in the liver is also important for myelosuppression by benzene. However, such bone marrow suppression cannot be achieved simply by administering phenol or hydroquinone to mice, although it can be achieved by coadministering hydroquinone with phenol. Phenol stimulates the MPO- and PHS-dependent activation of hydroquinone. Therefore, bone marrow suppression by benzene involves the CYP-dependent oxidation of benzene to phenol and hydroquinone in the liver, followed by the phenol-enhanced, MPO- and PHS-catalyzed peroxidative oxidation of hydroquinone to reactive intermediates that bind to protein and DNA in the bone marrow (Fig. 6-32). It is noteworthy that the CYP enzyme responsible for hydroxylating benzene has been identified as CYP2E1 (see the section “Cytochrome P450”). Although CYP2E1 was first identified in liver, this same enzyme has been identified in bone marrow where it can presumably convert benzene to phenol and possibly hydroquinone. The importance of CYP2E1 in the metabolic activation of benzene was confirmed by the demonstration that CYP2E1 knockout mice are relatively resistant to the myelosuppressive effects of benzene (Gonzalez, 2003; Gonzalez and Yu, 2006). MPO has also been shown to convert phenol to diols. Benzene-diols are good substrates of EPO, LPO, and MPO and their oxidation by peroxidases leads to oxidative stress (Testa and Krämer, 2008, 2010).

Prostaglandin H synthase (formally known as prostaglandin-endoperoxide synthase 1 and 2, and also widely known as cyclooxygenase 1 and 2; gene symbols PTGS1 and 2 and collectively referred to below as “PHS”) is one of the most extensively studied peroxidases involved in xenobiotic biotransformation. As shown in Fig. 6-33, PHS is a dual-function enzyme composed of a cyclooxygenase, which converts arachidonic acid to PGG₂, a 15-hydroperoxide-endoperoxide (which involves the addition of 2 molecules of oxygen to each molecule of arachidonic acid), and a peroxidase that converts the 15-hydroperoxide to the corresponding 15-alcohol PGH₂, which can be accompanied by the oxidation of xenobiotics (a “cosubstrate”) and formation of ROS (Ramkissoon and Wells, 2011). PHS1 is localized in the endoplasmic reticulum and nuclear envelope and is constitutively expressed in a wide variety of tissues that synthesize and release eicosanoids (prostaglandins, leukotrienes, thromboxane, prostacyclin), which generally bind to G-coupled cell surface receptors and regulate cellular function, largely in a paracrine (local) fashion (Ramkissoon and Wells, 2011). In contrast, PHS2 is expressed in cells that respond to inflammatory cytokines, mitogens, tumor promoters, and AhR agonists (ie, agents such as TNFα, LPS, phorbol esters, TPA, TCDD) such as the macula densa of the kidney, the vas deferens, and certain regions of the brain (Ramkissoon and Wells, 2011). PGH2 produces prostaglandins that activate receptors on the nuclear membrane, and high levels of the enzyme are expressed in colorectal and other tumors, which appear to be important for tumor promotion and angiogenesis (the process of stimulating the blood vessel supply). Both enzymes play an important role in the activation of xenobiotics to toxic or tumorigenic metabolites, particularly in extrahepatic tissues that contain low levels of CYP. PHS2 (COX-2) also plays an important role in the subsequent response of tissues to cell damage and tumor initiation, and is a possible target for the treatment or prevention of certain types of cancer. Increasing evidence suggests that inhibition of COX-2 by NSAIDs may be part of the mechanism whereby these drugs protect against colon and other gastrointestinal cancers (Dubé et al., 2007; Wu et al., 2010). PHS has been evaluated for its contribution to the activation of phenytoin and other antiepileptic drugs, B[a]P, thalidomide, methamphetamine, and 3,4-methylenedioxymethamphetamine (Ramkissoon and Wells, 2011).

In certain cases, the oxidation of xenobiotics by peroxidases involves direct transfer of the peroxide oxygen to the xenobiotic, as shown in Fig. 6-33 for the conversion of substrate X to product XO. An example of this type of reaction is the PHS-catalyzed epoxidation of B[a]P 7,8-dihydrodiol to the corresponding 9,10-epoxide (see Fig. 6-10). Although PHS can catalyze the final step (ie, 9,10-epoxidation) in the formation of this tumorigenic metabolite of B[a]P, it cannot catalyze the initial step (ie, 7,8-epoxidation), which is catalyzed by CYP. Several lines of evidence suggest that both PHS1 and PHS2 play an important role in PAH-induced skin carcinogenesis. First, both PHS1 knockout and PHS2 knockout mice are resistant to PAH-induced skin cancer. (It is noteworthy that PHS1 knockout mice appear normal despite having only 1% of the prostaglandin levels of wild-type mice, whereas only 60% of PHS2 knockout mice survive until weaning, and the 40% that survive past weaning usually die within a year of kidney disease.) Both resveratrol (an inhibitor of PHS1 and CYP1A1) and SC-58125 (an inhibitor of PHS2) block PAH-induced skin cancer in mice.

PHS can also catalyze the 8,9-epoxidation of aflatoxin B₁, which is one of the most potent hepatotumorigens known. Epoxidation by CYP is thought to be primarily responsible for the hepatotumorigenic effects of aflatoxin B₁. However, aflatoxin B₁ also causes neoplasia of rat renal papilla. This tissue has very low levels of CYP, but contains relatively high levels of PHS, which is suspected, therefore, of mediating the nephrotumorigenic effects of aflatoxin (Fig. 6-34).

The direct transfer of the peroxide oxygen from a hydroperoxide to a xenobiotic is neither the only mechanism of xenobiotic oxidation by peroxidases nor the most common. Xenobiotics that can serve as electron donors, such as amines and phenols, can be oxidized to free radicals during the reduction of a hydroperoxide. In this case, the hydroperoxide is still converted to the corresponding alcohol, but the peroxide oxygen is reduced to water instead of being incorporated into the xenobiotic. For each molecule of hydroperoxide reduced (which is a 2-electron process), 2 molecules of xenobiotic can be oxidized (each by a one-electron process). Important classes of compounds that undergo one-electron oxidation reactions by peroxidase include aromatic amines, phenols, hydroquinones, and polycyclic hydrocarbons. Many of the metabolites produced are reactive electrophiles. For example, PAHs, phenols, and hydroquinones are oxidized to electrophilic quinones. Acetaminophen is similarly converted to a quinoneimine, namely, N-acetyl-benzoquinoneimine, a cytotoxic electrophile that binds to cellular proteins, as shown in Fig. 6-35. The formation of this toxic metabolite by CYP causes centrilobular necrosis of the liver. However, acetaminophen can also damage the kidney medulla, which contains low levels of CYP but relatively high levels of PHS; hence, PHS may play a significant role in the nephrotoxicity of acetaminophen. The 2-electron oxidation of acetaminophen to N-acetyl-benzoquinoneimine by PHS likely involves the formation of a one-electron oxidation product, namely, N-acetyl-benzosemiquinoneimine radical. Formation of this semiquinoneimine radical by PHS likely contributes to the nephrotoxicity of acetaminophen and related compounds, such as phenacetin and 4-aminophenol. Metabolites of the NSAID diclofenac (namely, 4′-hydroxydiclofenac and 5-hydroxydiclofenac) can also be converted to benzoquinoneimines by peroxidases (Testa and Krämer, 2008, 2010).

Like the kidney medulla, urinary bladder epithelium also contains low levels of CYP but relatively high levels of PHS. Just as PHS in kidney medulla can ostensibly activate aflatoxin and acetaminophen to nephrotoxic metabolites, so PHS in urinary bladder epithelium can potentially activate certain aromatic amines, such as benzidine, 4-aminobiphenyl, and 2-aminonaphthalene, to DNA-reactive metabolites that cause bladder cancer in certain species, including humans and dogs. PHS can convert aromatic amines to reactive radicals, which can undergo nitrogen–nitrogen or nitrogen–carbon coupling reactions, or they can undergo a second one-electron oxidation to reactive diimines. Binding of these reactive metabolites to DNA is presumed to be one of the underlying mechanisms by which several aromatic amines cause bladder cancer in humans and dogs. In some cases the one-electron oxidation of an amine leads to N-dealkylation. For example, PHS catalyzes the N-demethylation of aminopyrine, although in vivo this reaction is mainly catalyzed by CYP. In contrast to CYP, PHS does not appear to catalyze the N-hydroxylation of carcinogenic aromatic amines (an important step in their metabolic activation), although MPO and LPO have been shown to catalyze this reaction. In liver, activation of aromatic amines by N-hydroxylation is catalyzed predominantly by CYP1A2, whereas this same reaction in the bladder epithelium appears to be catalyzed by another enzyme, possibly a CYP enzyme other than CYP1A2, such as CYP2A13, CYP4B1, or CYP2S1 (Nakajima et al., 2006), or a peroxidase other than PHS.

Many of the aromatic amines known or suspected of causing bladder cancer in humans have been shown to cause bladder tumors in dogs. In rats, however, aromatic amines cause liver tumors by a process that is thought to involve N-hydroxylation by CYP, followed by conjugation with acetate or sulfonate, as shown in Fig. 6-12. This species difference has complicated an assessment of the role of PHS in aromatic amine–induced bladder cancer, because such experiments must be carried out in dogs. However, another class of compounds, the 5-nitrofurans, such as N-[4-(5-nitro-2-furyl)-2-thiazole]formamide (FANFT) and its deformylated analog 2-amino-4-(5-nitro-2-furyl)thiazole (ANFT), are substrates for PHS and are potent bladder tumorigens in rats. The tumorigenicity of FANFT is thought to involve deformylation to ANFT, which is oxidized to DNA-reactive metabolites by PHS. The ability of FANFT to cause bladder tumors in rats is blocked by the COX inhibitor, aspirin, which suggests that PHS plays an important role in the metabolic activation and tumorigenicity of this nitrofuran. Unexpectedly, combined treatment of rats with FANFT and aspirin causes forestomach tumors, which are not observed when either compound is administered alone.

Increased expression of PHS2 (COX-2) has been documented in a number of tumors, including human colorectal, gastric, esophageal, pulmonary, and pancreatic carcinomas (Gupta and DuBois, 1998; Molina et al., 1999; Dubé et al., 2007; Wu et al., 2010). Aspirin and other NSAIDs block the formation of colon cancer in experimental animals, and there is epidemiological evidence that long-term use of certain NSAIDs (aspirin and sulindac), but not acetaminophen (which does not inhibit COX-2), decreases the incidence of colorectal polyps and cancer in humans, and also decreases the number of deaths from esophageal, gastric, and rectal cancers. The incidence of intestinal neoplasms in Apc^Δ716 knockout mice is dramatically suppressed by crossing these transgenic animals with PHS2 (COX-2) knockout mice (Oshima et al., 1996). From these few examples it is apparent that PHS2 may play at least 2 distinct roles in tumor formation; it may convert certain xenobiotics to DNA-reactive metabolites (and thereby initiate tumor formation), and it may promote subsequent tumor growth, perhaps through formation of growth-promoting eicosanoids, such as PGE2, which can activate EGFR, ERKs, etc (Wu et al., 2010).

Many phenolic compounds can serve as reducing substrates for PHS peroxidase. The phenoxyl radicals produced by one-electron oxidation reactions can undergo a variety of reactions, including binding to critical nucleophiles, such as protein and DNA, reduction by antioxidants such as GSH, and self-coupling. The reactions of phenoxyl radicals are analogous to those of the nitrogen-centered free radicals produced during the one-electron oxidation of aromatic amines by PHS.

It was previously mentioned in this section that phenol can enhance the peroxidative metabolism of hydroquinone, which is an important component to benzene myelotoxicity. An analogous interaction has been described between the phenolic antioxidants, BHT, and BHA. In mice, the pulmonary toxicity of BHT, which is a relatively poor substrate for PHS, is enhanced by BHA, which is a relatively good substrate for PHS. The mechanism by which BHA enhances the pulmonary toxicity of BHT involves the peroxidase-dependent conversion of BHA to a phenoxyl radical that interacts with BHT, converting it to a phenoxyl radical (by one-electron oxidation) or a quinone methide (by 2-electron oxidation), as shown in Fig. 6-36. Formation of the toxic quinone methide of BHT can also be catalyzed by CYP, which is largely responsible for activating BHT in the absence of BHA. A similar outcome occurs with quercetin, which is oxidized to a quinonemethide in the presence of peroxidases (Testa and Krämer, 2008, 2010).

Several reducing substrates, such as phenylbutazone, retinoic acid, 3-methylindole, sulfite, and bisulfite, are oxidized by PHS to carbon- or sulfur-centered free radicals that can trap oxygen to form a peroxyl radical, as shown in Fig. 6-37 for phenylbutazone. The peroxyl radical can oxidize xenobiotics in a peroxidative manner. For example, the peroxyl radical of phenylbutazone can convert B[a]P 7,8-dihydrodiol to the corresponding 9,10-epoxide.

PHS is unique among peroxidases because it can both generate hydroperoxides and catalyze peroxidase-dependent reactions, as shown in Fig. 6-33. Xenobiotic biotransformation by PHS is controlled by the availability of arachidonic acid. The biotransformation of xenobiotics by other peroxidases is controlled by the availability of hydroperoxide substrates. Hydrogen peroxide is a normal product of cellular respiration, and lipid peroxides can form during lipid peroxidation. The level of these peroxides and their availability for peroxidase reactions depends on the efficiency of hydroperoxide scavenging by GPX and catalase.

Flavin Monooxygenases

Liver, kidney, intestine, brain, and lung (among other tissues) contain one or more FAD-containing monooxygenases (FMO) that oxidize soft nucleophilic nitrogen, sulfur, selenium, phosphorus, and rarely—iodine, carbon, and boron—atoms of xenobiotics (Ziegler, 1993; Lawton et al., 1994; Cashman, 1995, 1999, 2008; Rettie and Fisher, 1999; Cashman and Zhang, 2006; Mitchell, 2008; Testa and Krämer, 2008, 2010). Among these types of substrates, the best tend to be those with lone pairs of electrons available to form a polar, coordinate covalent (dative) bond (eg, hydroxylamines and piperidines) (Cashman, 2008; Testa and Krämer, 2008, 2010). The mammalian FMO gene family comprises 5 enzymes (designated FMO1-FMO5) that contain 532 to 558 amino acid residues each and are 48% to 60% identical in amino acid sequence within a species, whereas orthologous forms are 82% to 86% identical across species (Hines et al., 2002). The human enzymes range in apparent molecular weight from 52 to 64 kDa (Testa and Krämer, 2008, 2010). Up to 6 human FMO pseudogenes have also been observed, including one that shares 71% sequence identity with FMO3, namely, FMO6, for which 9 distinct transcripts were identified in liver, but not kidney. All of these variant FMO6 transcripts arise by alternative splicing and lead to truncated, and therefore nonfunctional, enzymes. The possibility remains that rare SNPs in the FMO6 pseudogene could lead to normal expression of FMO6 in some individuals, but this has not yet been documented (Hines et al., 2002). Each FMO enzyme contains a highly conserved glycine-rich region (residues 4-32) that binds 1 mol of FAD (noncovalently) near the active site, which is adjacent to a second highly conserved glycine-rich region (residues 186-213) that binds NADPH (or NADH). Other structural motifs have been reviewed in detail by Ziegler (2002).

Like CYP, the FMOs are microsomal enzymes that require O₂ and NADPH (or NADH in some cases) (Lai et al., 2011), and many of the reactions catalyzed by FMO also can be catalyzed by CYP. However, whereas CYP catalyzes 2 sequential one-electron oxidations, FMO catalyzes a single 2-electron oxidation (Cashman, 2008). In contrast to CYP enzymes, FMOs are not readily induced or inhibited, such that metabolism of a drug by FMO tends to decrease its victim potential (ie, its potential to be the object of drug–drug interactions) (Cashman, 2008). However, basal expression levels appear to vary as much as 20-fold between individuals, with the differences appearing to be caused by mutations in the promoter regions of at least FMO1 and 3 (Testa and Krämer, 2008, 2010). Consequently, the disposition of a drug primarily metabolized by FMO would likely show considerable interindividual variation and, for reasons outlined later in this section, infants and young children would likely be PMs.

Several in vitro techniques have been developed to distinguish reactions catalyzed by FMO from those catalyzed by CYP. In general, and in contrast to CYP, FMO is heat labile and can be inactivated in the absence of NADPH by warming microsomes to 50°C for one minute (Ogilvie et al., 2008). However, even the often-used procedure of preincubating human liver microsomes at 37°C prior to addition of NADPH can cause significant inactivation of FMO, possibly leading to an underestimation of the role of FMO in the metabolism of xenobiotics (Cashman, 2008). By comparison, CYP can be inactivated with nonionic detergent, such as 1% Emulgen 911, which has a minimal effect on FMO activity. The pH optimum for FMO-catalyzed reactions (pH 8-10) tends to be higher than that for most (but not all) CYP reactions (pH 7-8). Based on the latter approach, a higher rate of formation of N-oxides from tertiary amines in microsomal incubations at pH 10 relative to pH 7.4 provides strong evidence for the role of FMO in N-oxide formation (Cashman, 2008). Antibodies raised against purified CYP enzymes can be used not only to establish the role of CYP in a microsomal reaction but also to identify which particular CYP enzyme catalyzes the reaction. In contrast, antibodies raised against purified FMO do not inhibit the enzyme. The use of chemical inhibitors to ascertain the relative contribution of FMO and CYP to microsomal reactions is often complicated by a lack of specificity. For example, both cimetidine and SKF 525A, which are well-recognized CYP inhibitors, are substrates for FMO. Conversely, the FMO inhibitor methimazole is known to cause direct inhibition of CYP2B6 and CYP2C9, and MDI of CYP3A4 (Parkinson et al., 2011). Excellent FMO substrates, such as the strong nucleophile mercaptoimidazole, can potentially be used as competitive inhibitors of FMO-mediated N-oxidation of tertiary amines (Cashman, 2008). Zwitterions, positively multiple charged compounds, and diamines such as cadaverine are generally not FMO substrates (Krueger et al., 2006). The situation is further complicated by the observation that the various forms of FMO differ in their thermal stability and sensitivity to detergents and other chemical modulators (examples of which are described later in this section).

FMO catalyzes the oxidation of nucleophilic tertiary amines to N-oxides, secondary amines to hydroxylamines and nitrones, and primary amines to hydroxylamines and oximes. Amphetamine, benzydamine, chlorpromazine, clozapine, guanethidine, imipramine, methamphetamine, olanzapine, and tamoxifen are examples of nitrogen-containing drugs that are N-oxygenated by FMO (and by CYP in most cases). Only sp³-hybridized electron-rich (nucleophilic) nitrogen atoms—those with pK_a values 5 to 10—are substrates for N-oxygenation by FMO, hence the preference of FMO for tertiary and cyclic amines. Therefore, in general, FMO N-oxygenates tertiary aliphatic amines and aliphatic cyclic amines (CYP can also) but FMO does not usually N-oxygenate aromatic amines (pyridines and anilines such as clozapine) (Testa and Krämer, 2008, 2010). Heteroaromatics (with an sp²-hybridized nitrogen) and amides (which have low basicity [low pK_a]) are not N-oxygenated by FMO but they may be N-oxygenated by CYP. Thus, CYP can N-oxygenate both tertiary/cyclic amines and pyridines. In the case of anilines (arylamines), however, CYP (as will peroxidases, but rarely FMO) will N-hydroxylate them (Testa and Krämer, 2008, 2010). The N-oxygenation of primary and secondary aliphatic amines by FMO is more complex. For instance, the aliphatic secondary amine, methamphetamine (R-NH−CH₃), is N-oxygenated by FMO3 first to a hydroxylamine (R-NOH−CH₃) and then to a dihydroxylated intermediate (R-N⁺(OH)₂−CH₃) that loses water to form a nitrone (R-N⁺O−CH₃). The resulting nitrone can spontaneously hydrolyze to a ketone (phenylacetone in this case), such that this overall reaction represents an alternative deamination pathway (Testa and Krämer, 2008, 2010). In the case of the aliphatic primary amine, amphetamine (R-CH₂−NH₂), FMO again catalyzes 2 sequential N-hydroxylation reactions but in this case the N-dihydroxylated metabolite R-CH₂−N(OH)₂ loses water to form an oxime (R-CH=NOH). Primary aliphatic amines tend to be poor substrates for CYP; in fact, primary alkylamines, such as the N-demethylated metabolite of fluoxetine, are metabolically stable and potent inhibitors of CYP (discussed later in the section “Inhibition of Cytochrome P450”).

FMO also oxygenates several sulfur-containing xenobiotics (such as thiols, thioethers, thiones, and thiocarbamates) and phosphines to S- and P-oxides, respectively. Cimetidine and sulindac sulfide are examples of sulfur-containing drugs that are converted to sulfoxides by FMO. As in the case of N-oxygenation, S-oxygenation requires a soft nucleophilic heteroatom, such as that found in thioethers. Both FMO and CYP can convert thioethers to sulfoxides (R₁-S-R₂ → R₁-SO-R₂) but only CYP can convert sulfoxides to sulfones (R₁-SO-R₂ → R₁-SO₂-R₂). Electron-withdrawing groups next to the sulfur (as in the case of sulfoxides and many heteroaromatics) decrease the nucleophilicity of the sulfur atom and decrease or abolish its oxygenation by FMO, whereas electronic-donating groups have the opposite effect, which is why thioamides are excellent substrates for S-oxygenation by FMO. Fig. 6-15 shows how sulindac, a sulfoxide, is reduced to sulindac sulfide, only to be oxidized by FMO back to the parent drug in what is often called a futile cycle as discussed further below in this section. Hydrazines, iodides, selenides, and boron-containing compounds are also substrates for FMO. Examples of FMO-catalyzed reactions are shown in Fig. 6-38A and B. One of the more unusual reactions catalyzed by FMO1 is the oxidative defluorination of 4-fluoro-N-methylaniline to 4-hydroxy-N-methylaniline (ie, 4-N-methylaminophenol) through the intermediacy of a reactive quinoneimine (Driscoll et al., 2010). In the case of 4-fluoro-N-methylaniline, this oxidative defluorination reaction is facilitated by delocalization of the lone pair of electrons on the aniline nitrogen.

In general, the metabolites produced by FMO are the products of a chemical reaction between a xenobiotic and a peracid or peroxide, which is consistent with the mechanism of catalysis of FMO (discussed later in this section). The reactions catalyzed by FMO are generally detoxication reactions, although there are exceptions to this rule, which are described below (this section). Inasmuch as FMO attacks nucleophilic heteroatoms, it might be assumed that substrates for FMO could be predicted from their pK_a values (ie, from a measure of their basicity). Although there is some truth to this—for example, xenobiotics containing an sp³-hybridized nitrogen atom with a pK_a of 5 to 10 are generally good substrates for FMO—predictions of substrate specificity based on pK_a values alone are not very reliable presumably because steric effects influence access of substrates to the FMO active site, which is consistent with the reported lack of rabbit FMO2 activity toward imipramine and chlorpromazine (Rettie and Fisher, 1999; Krueger et al., 2006).

The mechanism of catalysis by FMO is depicted in Fig. 6-39. After the FAD moiety is reduced to FADH₂ by NADPH, the oxidized cofactor, NADP⁺, remains bound to the enzyme. FADH₂ then binds oxygen to produce a peroxide (ie, the C(4a)-hydroperoxyflavin of FAD). The FAD-hydroperoxide is relatively stable, probably because the active site of FMO comprises non-nucleophilic, lipophilic amino acid residues, and this is thought to be the enzyme's resting state (ie, the FAD-hydroperoxide is the form in which FMO exists prior to substrate binding). During the oxygenation of xenobiotics (depicted as X → XO in Fig. 6-39), the nucleophilic heteroatom (N or S) attacks the terminal oxygen of the C(4a)-hydroperoxyflavin resulting in oxygen transfer to the xenobiotic (to form an N-oxide or sulfoxide) and formation of C(4a)-hydroxyflavin. From the latter step, it is understandable why the metabolites produced by FMO are generally the products of a chemical reaction between a xenobiotic and a peroxide or peracid. The final step in the catalytic cycle involves dehydration of C(4a)-hydroxyflavin (which restores FAD to the ground state) and release of NADP⁺. This final step is important because it is rate-limiting, and it occurs after substrate oxygenation. Consequently, this step determines the upper limit of the rate of substrate oxidation. Therefore, all good substrates for FMO are converted to products at roughly the same maximum rate (ie, V_max is determined by the final step in the catalytic cycle). Binding of NADP⁺ to FMO during catalysis is important because it prevents the reduction of oxygen to H₂O₂. In the absence of bound NADP⁺, FMO would function as an NADPH-oxidase that would consume NADPH and cause oxidative stress through excessive production of H₂O₂. In some cases the reactive C(4a)-hydroperoxyflavin can function as an “enzymatic peroxide” to catalyze a Baeyer–Villiger oxidation, which involves the insertion of an oxygen atom into a carbon–carbon bond next to a carbonyl group (eg, aldehyde or ketone), to form an ester (eg, oxidation of salicylaldehyde by pig FMO1) (Lai et al., 2011). However, this reaction has not been described in humans with the notable exception of FMO5 (and only FMO5), which mediates the Baeyer–Villiger lactonization of ER-879819, the ketone metabolite of the poly(ADP-ribose) polymerase inhibitor, E7016 (Lai et al., 2011).

The oxygenation of substrates by FMO does not lead to inactivation of the enzyme, even though some of the products are strong electrophiles capable of binding covalently to critical and noncritical nucleophiles such as protein and GSH, respectively. The products of the oxygenation reactions catalyzed by FMO and/or the oxygenation of the same substrates by CYP can inactivate CYP. For example, the FMO-dependent S-oxygenation of spironolactone thiol (which is formed by the deacetylation of spironolactone by carboxylesterases, as shown in Fig. 6-5) leads to the formation of an electrophilic sulfenic acid (R-SH → R-SOH) that can undergo redox cycling, or be further converted to the reactive sulfinic acid (R-SO₂H) or sulfanyl radical (HS^•) that inactivates CYP and binds covalently to other proteins (Decker et al., 1991; Krueger et al., 2006; Testa and Krämer, 2008, 2010).

In humans, FMO plays a role in the biotransformation of several drugs (eg, albendazole, amphetamine, benzydamine, chlorpheniramine, cimetidine, clindamycin, clozapine, codeine, dasatinib, deprenyl, ethionamide, fenbendazole, guanethidine, itopride, methimazole, olanzapine, olopatadine, pargyline, ranitidine, sulindac sulfide, tamoxifen, tazarotene, thiacetazone, thioridazine, tozasertib, vandetanib, voriconazole, xanomeline, zimeldine, hydrazines such as procarbazine, and various dimethylaminoalkyl phenothiazine derivatives such as chlorpromazine and imipramine) and other xenobiotics (eg, aldicarb, cocaine, diallyl disulfide, N,N-dimethylaniline, disulfoton, fenthion, methyl-p-tolyl sulfide, methamphetamine, nicotine, phorate, thioacetamide, thiobenzamide, and tyramine), and in the activation of 3,3′-iminodipropionitrile to the neurotoxic N-hydroxy-3,3′-iminodipropionitrile (Ballard et al., 2007; Testa and Krämer, 2008, 2010; Wang et al., 2008; Francois et al., 2009; Weil et al., 2010; Strolin-Benedetti, 2011). As discussed later in this section, FMO also plays a role in the metabolism of some endogenous substrates such as TMA, cysteamine, cysteine conjugates, methionine, and lipoic acid (6,8-dithiooctanoic acid).

The major flavin monooxygenase in human liver microsomes, FMO3, is predominantly if not solely responsible for converting (S)-nicotine to (S)-nicotine N-1′-oxide (which is one of the reactions shown in Fig. 6-38A). The reaction proceeds stereospecifically; only the trans isomer is produced by FMO3, and this is the only isomer of (S)-nicotine N-1′-oxide excreted in the urine of cigarette smokers or individuals wearing a nicotine patch (Cashman and Zhang, 2006). Therefore, the urinary excretion of trans-(S)-nicotine N-1′-oxide can be used as an in vivo probe of FMO3 activity in humans. However, common polymorphisms in FMO1 (eg, FMO1*6) have been shown to be associated with nicotine dependence in European patients (Hinrichs et al., 2011). Furthermore, recombinant human FMO1 catalyzes the N-oxidation of (S)-nicotine approximately 4-fold more efficiently than recombinant human FMO3, and does so nonstereospecifically (ie, 55:45 cis:trans ratio of (S)-nicotine N-1′-oxide, similar to the situation with pig, rat, and rabbit FMO1), whereas FMO3 forms predominantly the trans-(S)-nicotine N-1′-oxide (as described above) (Hinrichs et al., 2011). FMO1 is extrahepatic, with high expression levels in the kidney, and some expression in the intestine and brain. If FMO1 contributes significantly to the metabolism of nicotine in the brains of smokers, it is possible that the N-oxide formed by FMO1 could be reduced back to nicotine in the brain by AO, a process known as retro-reduction or futile cycling (Hinrichs et al., 2011).

FMO3 is also the principal enzyme involved in the S-oxygenation of cimetidine, an H₂-antagonist that was once widely used in the treatment of gastric ulcers and other acid-related disorders (this reaction is shown in Fig. 6-38B). Cimetidine is stereoselectively sulfoxidated by FMO3 to an 84:16 mixture of (+) and (−) enantiomers, which closely matches the 75:25 enantiomeric composition of cimetidine S-oxide in human urine. Therefore, the urinary excretion of cimetidine S-oxide, like that of (S)-nicotine N-1′-oxide, is an in vivo indicator of FMO3 activity in humans.

Sulindac is a sulfoxide that exists in 2 stereochemical forms (as do most sulfoxides), and a racemic mixture of R- and S-sulindac is used therapeutically as a NSAID. As shown in Fig. 6-15, the sulfoxide group in sulindac is reduced to the corresponding sulfide (which is achiral), which is then oxidized back to sulindac by retro-reduction (futile cycling). In human liver, the sulfoxidation of sulindac sulfide is catalyzed by FMO3 with little or no contribution from CYP. At low substrate concentrations (30 μM), FMO3 converts sulindac sulfide to R- and S-sulindac in an 87:13 ratio (Hamman et al., 2000). Consequently, although sulindac is administered as a racemic mixture (ie, a 1:1 mixture of R- and S-enantiomers), the reduction of this drug to the corresponding sulfide and its preferential sulfoxidation by FMO3 to R-sulindac results in stereoselective enrichment of R-sulindac in plasma and urine.

This retro-reduction of sulfide S-oxides or amine N-oxides formed by FMO (or CYP) back to the parent may occur more often than is currently recognized, but it can be observed in vitro when the levels of an N- or S-oxide metabolite increase rapidly at first but then quickly reach a plateau as a balance between oxidation and reduction is reached (Cashman, 2008). There are at least 7 systems that can catalyze the retro-reduction of FMO metabolites: (1) AO, (2) quinone reductase, (3) hemoglobin itself or other heme-containing proteins such as CYP (which may or may not be enzymatically catalyzed), (4) XOR, (5) NADPH-cytochrome P450 reductase, (6) the mARC with NADH-cytochrome b₅/cytochrome b₅, and (7) gut microflora (Cashman, 2008; Testa and Krämer, 2008, 2010). For instance, amitriptyline N-oxide, imipramine N-oxide, cyclobenzaprine N-oxide, and brucine N-oxide are reduced back to the parent tertiary amine by quinone reductase, heme, and/or gut microflora (Cashman, 2008; Testa and Krämer, 2008, 2010). Nicotine N-oxide is retro-reduced to nicotine by AO (Cashman, 2008). Recombinant human CYP1A1, 2A6, 2D6, and 3A4 can also retro-reduce tamoxifen N-oxide to the tertiary amine (Cashman, 2008). The retro-reduction of TMA N-oxide to TMA by gut microflora is discussed later in this section (Cashman, 2008; Testa and Krämer, 2008, 2010).

In the case of sulindac sulfide, stereoselective sulfoxidation occurs not only with human FMO3 (the major drug-metabolizing FMO in human liver) but also with porcine FMO1 (the major form expressed in pig liver) and rabbit FMO2 (the major form expressed in rabbit lung) (Hamman et al., 2000). However, this conformity is the exception, rather than the rule. For example, in contrast to the stereoselective oxygenation of (S)-nicotine and cimetidine by human FMO3 (see above), FMO1 (which is the major FMO expressed in pig, rat, and rabbit liver) converts (S)-nicotine to a 1:1 mixture of cis- and trans-(S)-nicotine N-1′-oxide, and similarly converts cimetidine to a 1:1 mixture of (+) and (−) cimetidine S-oxide, respectively. Therefore, statements concerning the role of FMO in the disposition of xenobiotics in humans may not apply to other species, or vice versa.

Several sulfur-containing xenobiotics are oxygenated by FMO to electrophilic reactive intermediates. Such xenobiotics include various thiols, thioamides, 2-mercaptoimidazoles, thiocarbamates, and thiocarbamides. The electrophilic metabolites of these xenobiotics do not inactivate FMO, but they can covalently modify and inactivate neighboring proteins, including CYP. Some of these same xenobiotics are substrates for CYP, and their oxygenation to electrophilic metabolites leads to inactivation of CYP, a process known variously as MDI, mechanism-based inhibition, and suicide inactivation. 2-Mercaptoimidazoles undergo sequential S-oxygenation reactions by FMO, first to sulfenic acids and then to sulfinic acids (R-SH → R-SOH → R-SO₂H). These electrophilic metabolites, such as the sulfenic acid metabolite produced from spironolactone thiol (see above), bind to critical nucleophiles (such as proteins) or interact with GSH to form disulfides. The thiocarbamate functionality present in numerous agricultural chemicals is converted by FMO to S-oxides (sulfoxides), which can be further oxygenated to sulfones. These reactions involve S-oxygenation adjacent to a ketone, which produces strong electrophilic acylating agents, which may be responsible for the toxicity of many thiocarbamate herbicides and fungicides. The hepatotoxicity of thiobenzamide is dependent on S-oxidation by FMO and/or CYP. As shown in Fig. 6-38B, the S-oxidation of thiobenzamide produces an S-oxide, which can rearrange to an oxathiirane (a 3-membered ring of carbon, sulfur, and oxygen) on photolysis or thermolysis. However, such oxathiiranes are readily reduced back to the S-oxide. In vivo, the S,S-dioxide is more likely to form, which readily tautomerizes to iminosulfinic acid, binds covalently to protein (which leads to hepatocellular necrosis), or rearranges to benzamide, a reaction known as oxidative group transfer. In general, thiols with a negative charge on the S-atom (eg, dithioacids) tend to be good substrates of FMOs, whereas less nucleophilic S-atoms tend to be oxidized by CYP (Testa and Krämer, 2008, 2010).

Endogenous FMO substrates include cysteamine, which is oxidized to the disulfide, cystamine, and TMA, which is converted to TMA N-oxide (Fig. 6-38). By converting cysteamine to cystamine, FMO may serve to produce a low-molecular weight disulfide exchange agent, which may participate in the formation of disulfide bridges during peptide synthesis or the renaturation of proteins. By converting TMA to TMA N-oxide, FMO3 (but neither FMO1 nor 5) converts a malodorous and volatile dietary product derived from the reduction of dietary TMA N-oxide by gut bacteria (eg, from choline, lecithin, carnitine, and especially fish, which contain large amounts of TMA N-oxide) to an inoffensive metabolite. TMA smells of rotting fish (because bacteria in dead fish rapidly convert the high levels of TMA N-oxide to TMA), and people who are genetically deficient in FMO3 suffer from trimethylaminuria or fish odor syndrome, which is caused by the excretion of TMA in urine, sweat, and breath (Ayesh and Smith, 1992; Phillips and Shephard, 2008). Although the lack of functional FMO3 (and hence trimethylaminuria) occurs only rarely (eg, ~1% of British Caucasians), it is now known to be caused by one of at least 30 mutations that decrease or abolish FMO3 activity (Cashman et al., 2000; Cashman and Zhang, 2006; Phillips and Shephard, 2008). As might be expected, trimethylaminuria is associated with an impairment of nicotine N-oxidation and other pathways of drug biotransformation that are primarily catalyzed by FMO3 (Rettie and Fisher, 1999). For instance, benzydamine N-oxygenation and oxidation of sulindac sulfide has been correlated with FMO3 genotype in trimethylaminurics (Cashman and Zhang, 2006; Testa and Krämer, 2008, 2010; Francois et al., 2009). Some polymorphisms decrease while others increase the activity of FMO3 toward certain substrates, while others affect the promoter region and alter the amount of enzyme expressed (Phillips and Shephard, 2008).

Humans and other mammals express 5 different flavin monooxygenases (FMO1, FMO2, FMO3, FMO4, and FMO5) in a species- and tissue-specific manner, as shown in Table 6-9, based on mRNA levels, which may not reflect protein levels (adapted from Cashman, 1995; Cashman and Zhang, 2006; Zhang and Cashman, 2006). The major hepatic FMOs expressed in humans are FMO3 and FMO5, whereas FMO1 is the major FMO expressed in rat, rabbit, and pig liver. In humans FMO1 is expressed in fetal liver and is very rapidly downregulated after birth (although not in kidney) while hepatic FMO3 is gradually upregulated over months to years, in contrast to other mammals (Shephard et al., 2007; Phillips and Shephard, 2008; Testa and Krämer, 2008, 2010). This delay between the hepatic silencing of FMO1 and expression of FMO3 means that most infants probably have very little drug-metabolizing FMO present in the liver for several months (Shimizu et al., 2011). In addition, given the fact that apparently only FMO3 catalyzes the N-oxidation of TMA produced from gut bacteria, this begs the question of why all infants do not have fish odor syndrome. In a cohort of 6 patients with genetically confirmed trimethylaminuria, the average age of presentation was 6 (±3.6) years (Chalmers et al., 2006). As a genetic deficiency of FMO3, trimethylaminuria is obviously present from birth. However, the phenotype only becomes apparent when an affected child is weaned, and foods with high amounts of choline (eg, eggs, liver, etc) or TMA N-oxide (eg, seafood) are introduced (Chalmers et al., 2006). Diagnostic delay also presumably leads to the higher than expected average age of presentation. Transient trimethylaminuria can however occur in neonates and infants if their diet sporadically contains higher than usual levels of choline, etc (Testa and Krämer, 2008, 2010).

Previously, FMO3 has been considered to be the dominant FMO in human liver, and although FMO5 expression is higher, it has activity toward fewer xenobiotic substrates, and likely plays only a minor role in xenobiotic metabolism, with the exception of E7016, the poly(ADP-ribose) polymerase inhibitor noted above (Cashman and Zhang, 2006; Lai et al., 2011). FMO5 is also the most highly expressed FMO in the small intestine, in terms of mRNA. In humans, high levels of FMO1 are expressed in the kidney, and low levels of FMO2 are expressed in the lungs of Caucasians and Asians.

Lung microsomes from most mammals, particularly rabbit, mouse, monkey, and other nonhuman primates, contain high levels of functional FMO2, and of the FMO transcripts detected in human lung, the FMO2 transcript is found at the highest levels by far (Zhang and Cashman, 2006). The uncharacteristically low level of active FMO2 in human lung is caused by a mutation (a C → T transition at codon 472 in exon 9) in the major human FMO2*2A allele found in essentially all Caucasian and Asian populations, which results in a premature stop codon and the synthesis of a nonfunctional, truncated protein (one lacking the last 64 amino acid residues from the C-terminus) (Dolphin et al., 1998; Veeramah et al., 2008; Francois et al., 2009). In contrast, 26% of African Americans, 7% of Puerto Ricans, and 2% of Mexicans have one normal allele and express a functional protein (Cashman and Zhang, 2006). In sub-Saharan Africa, approximately 33% of individuals possess at least one functional FMO2*1 allele, and in some subpopulations the incidence approaches 50% (Veeramah et al., 2008). In pulmonary microsomes from individuals who express one or more functional FMO2 alleles, the protein level is equal to or greater than that of CYP enzymes. Functional FMO2 has high activity toward thioureas and thioamides, which implies that individuals who express a functional FMO2 may be more susceptible to the toxic effects of sulfenic or sulfinic acids formed from S-oxygenation of phenylthiourea, α-naphthylthiourea, and ethylenethiourea, whereas Caucasians and Asians may be more susceptible to thioether-containing organophosphate pesticides such as phorate and disulfoton, for which the parent compounds are more toxic than products of the FMO2-catalyzed reaction (Cashman and Zhang, 2006; Veeramah et al., 2008). FMO2 has also been shown to catalyze S-oxygenation of the second-line antitubercular drugs ethionamide and thiacetazone, leading to the formation of reactive metabolites such as the sulfenic and sulfinic acids, carbodiimide derivatives, and S-oxides (Francois et al., 2009). Because these second-line antitubercular agents are prodrugs that must be converted by the mycobacteria in FMO-like reactions to the reactive products, it is possible that administration to individuals with at least one FMO2*1 allele could decrease their efficacy and increase pulmonary toxicity.

FMO4 is expressed at low levels in the brain of several mammalian species, where it might terminate the action of several centrally active drugs and other xenobiotics. It is unstable, however, which makes its characterization somewhat difficult (Cashman and Zhang, 2006).

The various forms of FMO are distinct gene products with different physical properties and substrate specificities. For example, FMO2 N-oxygenates n-octylamine, whereas such long aliphatic primary amines are not substrates for FMO1, although they stimulate its activity toward other substrates (in some cases causing a change in stereospecificity). Conversely, short-chain tertiary amines, such as chlorpromazine, orphenadrine, and imipramine, are substrates for FMO1 but not FMO2. FMO2 exhibits no activity toward phenothiazine derivatives with only a 3-carbon side chain or 1,3-diphenylthiourea (Krueger et al., 2006). FMO3 is highly selective in the N-oxygenation of TMA, whereas FMO5 is selective for the N-oxygenation of short-chain aliphatic primary amines such as N-octylamine, but has little activity toward other typical FMO substrates, with the exception of E7016, the poly(ADP-ribose) polymerase inhibitor noted previously (Krueger et al., 2006; Lai et al., 2011). The substrate specificity of FMO4 also is somewhat restricted. Certain substrates are oxygenated stereospecifically by one FMO enzyme but not another. For example, FMO2 and FMO3 convert (S)-nicotine exclusively to trans-(S)-nicotine N-1′-oxide, whereas the N-oxides of (S)-nicotine produced by FMO1 are a 1:1 mixture of cis and trans isomers. FMO2 is heat stable under conditions that completely inactivate FMO1, and FMO2 is resistant to anionic detergents that inactivate FMO1. Low concentrations of bile acids, such as cholate, stimulate FMO activity in rat and mouse liver microsomes but inhibit FMO activity in rabbit and pig liver.

The FMO enzymes expressed in liver microsomes are not under the same regulatory control as CYP enzymes. In rats, the expression of FMO1 is suppressed rather than induced by treatment with phenobarbital or 3-methylcholanthrene (although some studies point to a modest [~3-fold] induction of rat FMO1 by 3-methylcholanthrene). Indole-3-carbinol, which induces the same CYP enzymes as 3-methylcholanthrene, causes a marked decrease in FMO activity in rat liver and intestine. A similar decrease in FMO3 activity occurs in human volunteers following the consumption of large amounts of Brussels sprouts, which contain high levels of indole-3-carbinol and related indoles. The decrease in FMO3 activity may result from direct inhibition of FMO3 by indole-3-carbinol and its derivatives rather than from an actual decrease in enzyme levels (Cashman, 1999). Increased levels of nitric oxide (NO), as occurs during acute inflammation (eg, sepsis or endotoxemia), which activates nitric oxide synthase-2 (NOS-2), can inhibit FMO (as well as CYP) (Mitchell, 2008). However, the AhR agonist, TCDD, was found to induce FMO2 and FMO3 mRNA levels in mice by 30- and 80-fold, respectively (Tijet et al., 2006).

The levels of FMO3 and, to a lesser extent, the levels of FMO1 in mouse liver microsomes are sexually differentiated (female > male) due to suppression of expression by testosterone. The opposite is true of FMO1 levels in rat liver microsomes, the expression of which is positively regulated by testosterone and negatively regulated by estradiol. In pregnant rabbits, lung FMO2 is positively regulated by progesterone and/or corticosteroids.

Species differences in the relative expression of FMO and CYP appear to determine differences in the toxicity of the pyrrolizidine alkaloids, senecionine, retrorsine, and monocrotaline. These compounds are detoxified by FMO, which catalyzes the formation of tertiary amine N-oxides, but are activated by CYP, which oxidizes these alkaloids to pyrroles that generate toxic electrophiles through the loss of substituents on the pyrrolizidine nucleus (details of which appear in the section “Cytochrome P450”). Rats have a high pyrrole-forming CYP activity and a low N-oxide-forming FMO activity, whereas the opposite is true of guinea pigs. This likely explains why pyrrolizidine alkaloids are highly toxic to rats but not to guinea pigs. Many of the reactions catalyzed by FMO are also catalyzed by CYP, but differences in the oxidation of pyrrolizidine alkaloids by FMO and CYP illustrate that this is not always the case. Species differences in the levels of FMO can also complicate the findings of preclinical toxicology and metabolism studies because most toxicologically useful small animal species express FMO1 as the dominant hepatic FMO in contrast to human, in which FMO3 is the dominant form (Cashman, 2008). However, one animal model that may be more similar to the human hepatic expression of FMO is the female mouse, which has relatively high levels of FMO3 and 5 (Cashman, 2008).

Cytochrome P450

Of all the xenobiotic-biotransforming enzymes, the CYP enzyme system ranks first in terms of catalytic versatility and the sheer number of xenobiotics it detoxifies or activates to reactive intermediates. The highest levels of CYP enzymes involved in xenobiotic biotransformation are found in liver endoplasmic reticulum (microsomes), but CYP enzymes are present in virtually all tissues. Some of the so-called microsomal CYP enzymes are also located on the inner membrane of mitochondria, the importance of which is discussed in the section “CYP2E1.” CYP enzymes play a very important role in determining the intensity and duration of action of drugs, and they also play a key role in the detoxication of xenobiotics. CYP enzymes in liver and extrahepatic tissues play important roles in the activation of xenobiotics to toxic and/or tumorigenic metabolites. The catalytic versatility of CYP enzymes is apparent from Table 6-2, which shows some of the many chemical groups that can be metabolized by CYP.

Microsomal and mitochondrial CYP enzymes play key roles in the biosynthesis or catabolism of steroid hormones, bile acids, fat-soluble vitamins such as vitamins A and D, fatty acids, and eicosanoids such as prostaglandins, thromboxane, prostacyclin, and leukotrienes, which underscores the catalytic versatility of CYP.

The human CYP superfamily contains 55 functional genes and 60 pseudogenes (Zhou et al., 2009a; Guengerich et al., 2010; http://drnelson.uthsc.edu/CytochromeP450.html). Many sources place the number of functional genes at 57; however, the 2 additional genes, namely, CYP2A7 (originally named IIA4) and CYP4B1, encode enzymes incapable of incorporating heme, rendering them catalytically inactive (note, however, that functional CYP4B1 is expressed in other mammalian species) (Yamano et al., 1990; Ding et al., 1995; Baer and Rettie, 2006). As shown in Table 6-10, the 55 human CYP enzymes can be broadly categorized on the basis of their role in (1) xenobiotic biotransformation, (2) fatty acid/eicosanoid hydroxylation/epoxidation, (3) steroidogenesis, (4) bile acid synthesis, vitamin D activation/inactivation, (5) retinoic acid metabolism, and (6) unknown function (a diminishing group of so-called orphan enzymes). In several cases, there is no clear functional distinction in terms of endobiotic and xenobiotic metabolism because, as noted in Table 6-10, there are many examples of CYP enzymes playing an important role in the metabolism of both an endobiotic and a drug or other xenobiotic. CYP2J2 and, in animals, CYP4B1 are examples of enzymes that ride the xenobiotic–endobiotic fence. In terms of endobiotic metabolism, CYP enzymes play a role in both catabolism and anabolism (several different CYP enzymes play a role in steroid hormone and bile acid synthesis). For example, CYP enzymes both activate vitamin D₃ to 1α,25-dihydroxyvitamin D₃ (1,25-(OH)₂-D₃) (CYP2R1, CYP2J2, CYP3A4, CYP27A1, and CYP27B1) and inactivate the active metabolite (CYP24A1 and CYP3A4). Arachidonic acid is epoxidated by CYP2C8, CYP2C9, and especially CYP2J2 to vasodilatory epoxyeicosatrienoic acids (EETs) but is converted by ω-hydroxylation to the vasoconstrictor 20-hydroxyeicosatetraenoic acid (20-HETE) by various CYP4A and CYP4F enzymes. The major xenobiotic-biotransforming CYP enzymes in human liver microsomes belong to families 1, 2, and 3, which are shown in Table 6-11 along with their homologs in the nonclinical species widely used in drug safety testing.

All CYP enzymes are heme-containing proteins. In some cases, such as many members of the CYP4 family, the heme moiety is covalently attached to the protein but in most other cases it is attached noncovalently (Ortiz de Montellano, 2008). The heme iron in CYP is usually in the ferric (Fe^III) state. When reduced to the ferrous (Fe^II) state, CYP can bind ligands such as O₂ and carbon monoxide (CO). The complex between ferrous CYP and CO absorbs light maximally at 450 nm, from which CYP derives its name (Pigment 450). The absorbance maximum of the CO complex differs slightly among different CYP enzymes and ranges from 447 to 452 nm. All other hemoproteins that bind CO absorb light maximally at ~420 nm. The unusual absorbance maximum of CYP is due to an unusual fifth ligand to the heme (a cysteine thiolate). The amino acid sequence around the cysteine residue that forms the thiolate bond with the heme moiety is highly conserved in all CYP enzymes. When this thiolate bond is disrupted, CYP is converted to a catalytically inactive form called cytochrome P420. By competing with oxygen, CO inhibits CYP. The inhibitory effect of CO can be reversed by irradiation with light at 450 nm, which photodissociates the CYP–CO complex. These properties of CYP are of historical importance (Omura, 2011). The observation that treatment of rats with certain chemicals, such as 3-methylcholanthrene, causes a shift in the peak absorbance of CYP (from 450 to 448 nm) provided some of the earliest evidence for the existence of multiple forms of CYP in liver microsomes. The conversion of CYP to cytochrome P420 by detergents and phospholipases helped to establish the hemoprotein nature of CYP. The inhibition of CYP by CO and the reversal of this inhibition by photodissociation of the CYP–CO complex established CYP as the microsomal and mitochondrial enzyme involved in drug biotransformation and steroid biosynthesis (Omura, 2011).

The basic reaction catalyzed by CYP enzymes is monooxygenation in which one atom of oxygen is incorporated into a substrate, designated RH, and the other is reduced to water with reducing equivalents derived from NADPH, as follows:

Although CYP functions as a monooxygenase, the products are not limited to alcohols and phenols due to rearrangement reactions (Guengerich, 1991, 2001b, 2007; Isin and Guengerich, 2007). During catalysis, CYP binds directly to the substrate and molecular oxygen, but it does not interact directly with NADPH or NADH. The mechanism by which CYP receives electrons from NAD(P)H depends on the subcellular localization of CYP. In the endoplasmic reticulum, which is where most of the CYP enzymes involved in xenobiotic biotransformation are localized, electrons are relayed from NADPH to CYP via a flavoprotein called NADPH-cytochrome P450 reductase (also known as an oxidoreductase; gene symbol POR). Within this flavoprotein, electrons are transferred from NADPH to CYP via FMN and FAD. In mitochondria, which house many of the CYP enzymes involved in steroid hormone biosynthesis and vitamin D metabolism, electrons are transferred from NAD(P)H to CYP via 2 proteins: an iron–sulfur protein called ferredoxin (gene symbol FDX1) and an FMN-containing flavoprotein called ferredoxin reductase (gene symbol FDXR). These proteins are also known as adrenodoxin and adrenodoxin reductase. In bacteria such as Pseudomonas putida, which express P450cam (CYP101A1), electron flow is similar to that in mitochondria (NADH → flavoprotein → putidaredoxin → P450).

There are some notable exceptions to the general rule that CYP requires a second enzyme (ie, a flavoprotein) for catalytic activity. One exception applies to 2 CYP enzymes involved in the conversion of arachidonic acid to eicosanoids, namely, thromboxane A synthase (CYP5A1, gene symbol TBXAS1) and prostaglandin I₂ synthase, which is also known as prostacyclin synthase (CYP8A1, gene symbol PTGIS). These 2 CYP enzymes convert the endoperoxide, PGH₂, to thromboxane (TXA₂) and prostacyclin (PGl₂) in platelets and the endothelial lining of blood vessels, respectively. In both cases, CYP functions as an isomerase and catalyzes a rearrangement of the oxygen atoms introduced into arachidonic acid by cyclooxygenase (see Fig. 6-33). The plant CYP, allene oxide synthase (CYP74A1), and certain invertebrate CYP enzymes also catalyze the rearrangement of oxidized chemicals (Guengerich, 2001b).

The second exception is represented by CYP102 (P450BM3) from the bacterium Bacillus megaterium, which produces a CYP enzyme linked directly to a flavoprotein to form a single, self-sufficient fusion protein. Some bacterial CYP enzymes are thermophilic (such as CYP119, CYP174A1, and CYP231A2), which has attracted attention for their potential industrial applications (Nishida and Ortiz de Montellano, 2005). The thermophilic P450 enzyme from Sulfolobus acidocaldarius, namely, CYP119, was used to confirm (and settle a long-standing debate) that the final step in the catalytic cycle of CYP is formation of compound I (por^•+Fe^IV=O), as in the case of peroxidases, which is discussed later in this section (Rittle and Green, 2010). Most mammalian CYP enzymes are not synthesized as a single enzyme containing both the hemoprotein and flavoprotein moieties, but this arrangement is found in the nitric oxide (NO) synthases. In addition to its atypical structure, the P450 enzyme expressed in B. megaterium, CYP102, is unusual for another reason: it is inducible by phenobarbital, which provided insight into the mechanism of CYP induction.

Phospholipids and cytochrome b₅ also play an important role in CYP reactions (McLaughlin et al., 2010). CYP and NADPH-cytochrome P450 reductase are embedded in the phospholipid bilayer of the endoplasmic reticulum, which facilitates their interaction. With the notable exception of CYP2W1 (discussed later in this section), CYP and NADPH-cytochrome P450 reductase face the cytoplasmic side of the endoplasmic reticulum. When the C-terminal region that anchors NADPH-cytochrome P450 reductase in the membrane is cleaved with trypsin, the truncated flavoprotein can no longer support CYP reactions, although it is still capable of reducing cytochrome c and other soluble electron acceptors. The ability of phospholipids to facilitate the interaction between NADPH-cytochrome P450 reductase and CYP does not appear to depend on the nature of the polar head group (serine, choline, inositol, ethanolamine), although certain CYP enzymes (those in the CYP3A subfamily) have a requirement for phospholipids containing unsaturated fatty acids. In vitro experiments established that cytochrome b₅ can stimulate various CYP reactions by either increasing V_max or decreasing K_m, which was initially interpreted as evidence that cytochrome b₅ can donate the second of 2 electrons required by CYP. However, the same stimulation occurs with heme-depleted cytochrome b₅ (Yamazaki et al., 1996, 2001). The stimulatory effect of cytochrome b₅ is now attributed to its effect on CYP conformation and/or its ability to facilitate the interaction between CYP and NADPH-cytochrome P450 reductase. Experiments in conditional knockout mice establish that cytochrome b₅ has an important stimulatory effect on CYP in vivo (McLaughlin et al., 2010). Liver microsomes contain numerous forms of CYP but contain a single form of NADPH-cytochrome P450 reductase (POR) and cytochrome b₅ (CYB5A).

For each molecule of NADPH-cytochrome P450 reductase in rat liver microsomes, there are 5 to 10 molecules of cytochrome b₅ and 10 to 20 molecules of CYP. In human liver microsomes the ratio of CYP to NADPH-cytochrome P450 reductase is slightly lower (closer to 5:1). NADPH-cytochrome P450 reductase will reduce electron acceptors other than CYP, which enables this enzyme to be measured based on its ability to reduce cytochrome c (which is why NADPH-cytochrome P450 reductase is often called NADPH-cytochrome c reductase). NADPH-cytochrome P450 reductase can transfer electrons much faster than CYP can use them, which more than likely accounts for the low ratio of NADPH-cytochrome P450 reductase to CYP in liver microsomes. Low levels of NADPH-cytochrome P450 reductase may also be a safeguard to protect cells from the often deleterious one-electron reduction reactions catalyzed by this flavoprotein (see Fig. 6-16). NADPH-cytochrome P450 reductase also supports heme oxygenase, a Nrf2-inducible enzyme, the significance of which is discussed in the section “CYP2E1.”

Whereas microsomal cytochrome b₅ is encoded by CYB5A, CYB5B encodes the cytochrome b₅ found in the outer mitochondrial membrane where it supports the molybdoenzyme mARC in the dehydroxylation of certain amidoximes and related compounds (see the section “Dehydroxylation—mARC, Cytochrome b₅, b₅ Reductase, and Aldehyde Oxidase” and Fig. 6-22).

The catalytic cycle of CYP involves 8 steps (A → H), as shown in Fig. 6-40 for the oxidation of a substrate (RH) to its hydroxylated metabolite (ROH) (Dawson, 1988; Guengerich, 2007; Rittle and Green, 2010; Johnston et al., 2011; Hrycay and Bandiera, 2012). In this scheme, iron is shown bound to its fifth ligand, a heme thiolate provided by a highly conserved cysteine (Cys) residue. The first steps of the cycle (A → G) involve the activation of oxygen to compound I, and the final steps involve substrate oxidation by compound I (G → H) followed by release of the metabolite (ROH) to restore the enzyme to its resting (ferric) state (H → A). Following the binding of substrate (RH) to CYP (A → B), the heme iron is reduced from the ferric (Fe^III) to the ferrous (Fe^II) state by the introduction of a single electron from NADPH-cytochrome P450 reductase (B → C). In many cases the reduction of CYP is facilitated by substrate binding because binding of the substrate in the vicinity of the heme moiety converts the heme ferric iron from a low-spin to a high-spin state, although some enzymes, such as CYP1A2 and CYP2E1, are naturally in the high-spin state and can be reduced in the absence of substrate. It is at stage C, when the iron is in the ferrous state, that CYP can bind oxygen and CO. It is at this stage of the cycle that, under reduced oxygen tension, CYP can reduce certain substrates (see “Other Reactions” at the bottom of the catalytic cycle shown in Fig. 6-40). In the third step (C → D) oxygen binds to the ferrous iron, which transfers an electron to oxygen to form ferrisuperoxo anion (ie, ferric-bound superoxide anion), designated Cys−Fe^IIIO₂⁻. At this stage the cycle can be interrupted (uncoupled) to release superoxide anion and restore the enzyme to its resting (ferric) state (see “Other Reactions” in Fig. 6-40). In the fourth step (D → E), a second electron is introduced from NADPH-cytochrome P450 reductase, which is delocalized over the thiolate bond, to form the supernucleophilic ferriperoxo intermediate ⁻Cys−Fe^IIIO₂⁻. Uncoupling of the cycle at this stage releases hydrogen peroxide and restores the enzyme to its resting (ferric) state (see “Other Reactions” in Fig. 6-40). In the fifth step (E → F), addition of a proton (H⁺) converts the supernucleophilic ferriperoxo intermediate to its corresponding hydroperoxide, the ferrihydroperoxy intermediate ⁻Cys−Fe^IIIOOH. In the sixth step (F → G), addition of a second proton and release of water converts the ferrihydroperoxy intermediate to compound I, an iron^IV-oxo porphyrin radical cation species (an oxidizing species previously described in the section “Peroxidase-Dependent Cooxidation”). The formation of compound I (por^•+Fe^IV=O) by protonation of the ferrihydroperoxy intermediate involves the heterolytic cleavage of oxygen with the 2-electron oxygen atom going to water, a reaction facilitated by the strong electron-donating effects of the heme thiolate anion. Heterolytic (2-electron) cleavage of oxygen to produce compound I places the iron in the perferryl (Fe^V) oxidation state, which is a considerably stronger oxidant than that formed by homolytic cleavage of oxygen, which produces the less reactive Por Fe^IV–OH with iron in oxidation state IV. (Note: It is somewhat confusing that compound I is written as por^•+Fe^IV=O because, without taking the porphyrin ring into account, the formula gives the erroneous impression that the iron is in the Fe^IV [ferryl] state, whereas it is actually in the Fe^V [perferryl] state.) In the seventh step (G → H), the highly electrophilic oxygen from compound I is transferred to the substrate (RH) to produce metabolite (ROH). In the final step (H → A), the metabolite is released, which restores the enzyme to its initial resting (ferric) state.

Compound I is the electrophilic oxo species responsible for the vast majority of CYP-catalyzed reactions (Hrycay and Bandiera, 2012). It can be formed as described above with NADPH and oxygen, or it can be formed by oxygen transfer from various organic hydroperoxides and peracids (X-OOH, such as cumene hydroperoxide, tert-butylhydroperoxide, and meta-chloroperbenzoic acid) in a one-step reaction called the “peroxide shunt,” as shown in “Other Reactions” in Fig. 6-40. Formation of CYP compound I from peroxy compounds (P450 + X-OOH → P450 compound I + X-OH) is identical to the formation of compound I in peroxidases (see the section “Peroxidase-Dependent Cooxidation”). However, by virtue of its unusual fifth ligand (the heme thiolate from cysteine) and, perhaps more importantly, its active site topology, CYP can catalyze a far greater array of oxidative reactions (or catalyze them much faster) than those typically seen with peroxidases (even chloroperoxidase, which is unusual among peroxidases for containing a Cys thiolate as the fifth ligand to the heme like that in CYP). CYP reactions supported by organic peroxides and peracids are not affected by CO, an inhibitor of reactions supported by NADPH/O₂.

Although P450 compound I generated by the peroxide shunt is as catalytically versatile as that generated by NADPH/O₂, differences in the ratio of metabolites formed from a single substrate, differences in the relative rates of oxidation of 2 or more substrates, and differences in kinetic isotope effects (the influence of deuterium substitution at the site of oxidation) have led to the “multiple oxidants” hypothesis. The concept is supported by the finding that site-directed mutagenesis of a highly conserved threonine residue in bacterial and mammalian CYP enzymes can differentially affect reaction rates. For example, site-directed mutagenesis (Thr268Ala) of the bacterial enzyme CYP102A1 (P450BM3) increases the rate of sulfoxidation relative to N-demethylation of a single substrate, namely, dimethyl-(4-methylsulfanylphenyl)amine, whereas another mutation (Phe87Ala) has the opposite effect. Although the concept that CYP forms 2 or more oxidizing species is well accepted, the basis for the multiplicity is a source of considerable debate (Chandrasena et al., 2004; Jin et al., 2004; Newcomb and Chandrasena, 2005; Sheng et al., 2009; Hrycay and Bandiera, 2012). The 2 major competing theories are the 2-state and the 2-oxidant hypotheses, both of which have supporting evidence. The 2-state hypothesis posits that compound I is the only oxidizing species but it exists in 2 states, a low-spin state (that favors N-demethylation) and a high-spin state (that favors sulfoxidation) (Newcomb and Chandrasena, 2005; Hrycay and Bandiera, 2012). The 2-oxidant hypothesis posits that compound I is the major electrophilic oxidizing species but that its precursor, the ferrihydroperoxo intermediate (⁻Cys−Fe^IIIOOH), can function as a relatively strong nucleophilic but weak electrophilic oxidizing species (Chandrasena et al., 2004; Jin et al., 2004; Sheng et al., 2009). This is supported by the observation that hydrogen peroxide (which does not form compound I but instead forms the preceding ferrihydroperoxo intermediate or its equivalent, namely, ferric iron–bound hydrogen peroxide) can support certain reactions that are not supported by organic hydroperoxides and peracids. The 2 mechanisms of generating multiple oxidizing species are not mutually exclusive and both may be needed to explain all the catalytic properties of CYP. There are also competing theories concerning the mechanism of substrate oxidation by P450 compound I (Hrycay and Bandiera, 2012). Hereafter, the mechanism of carbon hydroxylation is discussed in terms of just one possibility, namely, hydrogen atom transfer (HAT) from the site of substrate hydroxylation by compound I to form a carbon radical followed by oxygen rebound in a nonconcerted, stepwise manner even though experiments with so-called ultrarapid radical clocks support a nonradical mechanism (Newcomb and Chandrasena, 2005; Hrycay and Bandiera, 2012). In reactions involving heteroatoms (N and S) the initial step involves (among other possibilities) single electron transfer (SET) leading to S- and N-oxygenation or HAT leading to S- and N-dealkylation, as discussed later in this section (Li et al., 2009a; Roberts and Jones, 2010).

CYP catalyzes several types of oxidation reactions, including:

1. Hydroxylation of an aliphatic or aromatic carbon

2. Epoxidation of a double bond

3. Heteroatom (S-, N-, and I-) oxygenation and N-hydroxylation

4. Heteroatom (O-, S-, and N-) dealkylation

5. Oxidative group transfer

6. Cleavage of esters and carbamates

7. Dehydrogenation

In the first 3 cases, oxygen from P450 compound I is incorporated into the substrate, which otherwise remains intact. In the fourth case, oxygenation of the substrate is followed by a rearrangement reaction leading to cleavage of an amine (N-dealkylation) or an ether (O- and S-dealkylation). Oxygen from P450 compound I is incorporated into the alkyl-leaving group, producing an aldehyde or ketone. In the fifth case, oxygenation of the substrate is followed by a rearrangement reaction leading to loss of a heteroatom (oxidative group transfer). The sixth case, the cleavage of esters and carbamates, resembles heteroatom dealkylation in that the functional group is cleaved with incorporation of oxygen from P450 compound I into the leaving group, producing an aldehyde. In the seventh case, 2 hydrogens are abstracted from the substrate with the formation of a double bond (C=C, C=O, or C=N), with the reduction of oxygen from P450 compound I to water. It should be noted that this long list of reactions does not encompass all the reactions catalyzed by CYP (Guengerich, 2001b, 2007). CYP can catalyze reductive reactions (such as azo-reduction, nitro-reduction, N-oxide reduction, sulfoxide reduction, and reductive dehalogenation), ring expansion or ring formation, dearylation, dearomatization, isomerization (such as the conversion of PGH₂ to thromboxane and prostacyclin), and oxidative dehalogenation (as described previously for FMO; see Fig. 6-38). During the synthesis of steroid hormones, CYP catalyzes the cleavage of carbon–carbon bonds, which occurs during the conversion of cholesterol to pregnenolone by cholesterol side-chain cleavage enzyme (CYP11A1, which is also known as P450_scc) and the aromatization of a substituted cyclohexane, which occurs during the conversion of androgens to estrogens by aromatase (CYP19A1, also known as CYP19 and P450_arom).

Examples of aliphatic and aromatic hydroxylation reactions catalyzed by CYP are shown in Figs. 6-41 and 6-42, respectively. The hydroxylation of aromatic hydrocarbons may proceed via an oxirane intermediate (ie, an arene oxide) that isomerizes to the corresponding phenol. Alternatively, aromatic hydroxylation can proceed by a mechanism known as direct insertion. The ortho- and para-hydroxylation of chlorobenzene proceed via 2,3- and 3,4-epoxidation, whereas meta-hydroxylation proceeds by direct insertion, as shown in Fig. 6-43. When aromatic hydroxylation involves direct insertion, hydrogen abstraction (ie, cleavage of the C–H bond) is the rate-limiting step, so that substitution of hydrogen with deuterium or tritium considerably slows the hydroxylation reaction. This isotope effect is less marked when aromatic hydroxylation proceeds via an arene oxide intermediate. Arene oxides are electrophilic and, therefore, potentially toxic metabolites that are detoxified by such enzymes as epoxide hydrolase (see Figs. 6-8 to 6-10) and GST (see the section “Glutathione Conjugation”). Depending on the ring substituents, the rearrangement of arene oxides to the corresponding phenol can lead to an intramolecular migration of a substituent (such as hydrogen or a halogen) from one carbon to the next. This intramolecular migration occurs at the site of oxidation and is known as the NIH shift, so named for its discovery at the National Institutes of Health.

Aliphatic hydroxylation involves insertion of oxygen into a C–H bond. The initial step involves HAT to form a carbon radical followed by oxygen rebound, shown as follows:

In the case of simple, straight-chain hydrocarbons, such as n-hexane, aliphatic hydroxylation occurs at both the terminal methyl groups and the internal methylene groups. In the case of fatty acids (both saturated and unsaturated) and their derivatives (ie, retinoic acid and eicosanoids such as prostaglandins and leukotrienes), aliphatic hydroxylation occurs at the ω-carbon (terminal methyl group) and the ω-1 carbon (penultimate carbon), as shown for lauric acid in Fig. 6-41. For thermodynamic reasons, most CYP enzymes preferentially catalyze the ω-1 hydroxylation of fatty acids and their derivatives, but one group of CYP enzymes (those encoded by the CYP4 gene family) preferentially catalyzes the less energetically favorable ω-hydroxylation of fatty acids, which can be further oxidized to dicarboxylic acids and undergo chain shortening by β-oxidation (Baer and Rettie, 2006; Johnston et al., 2011).

Like CYP, the molybdoenzymes AO and XO can also catalyze the carbon oxidation of xenobiotics as outlined in the section “Molybdenum Hydroxylases (Molybdoenzymes).” As an electrophilic oxidizing enzyme, CYP generally prefers to catalyze the oxidation of carbon atoms with a high electron density, whereas the nucleophilic oxidizing enzymes AO and XO preferentially catalyze the oxidation of carbon atoms with a low electron density (such as the sp² carbon atom double bonded to a nitrogen atom in various nitrogen heterocycles). For this reason, xenobiotics that are good substrates for CYP enzymes tend to be poor substrates for AO, and vice versa (Pryde et al., 2010).

Xenobiotics containing a carbon–carbon double bond (ie, alkenes) can be epoxidated (ie, converted to an oxirane) in an analogous manner to the oxidation of aromatic compounds to arene oxides. Just as arene oxides can isomerize to phenols, so aliphatic epoxides can isomerize to the corresponding ene-ol, the formation of which may involve an intramolecular migration (NIH shift) of a substituent at the site of oxidation (examples of intramolecular shifts accompanying epoxidation are given in the section “Activation of Xenobiotics by Cytochrome P450”). Like arene oxides, aliphatic epoxides are also potentially toxic metabolites that are inactivated by other xenobiotic-metabolizing enzymes such as epoxide hydrolase and GST. Alkynes can be epoxidated by CYP to ketocarbenes (which can be further oxidized to carboxylic acids). The conversion of an alkyne to a ketocarbene via epoxidation or other possible oxidation intermediates is shown in the following scheme:

Oxidation of some aliphatic alkenes and alkynes produces metabolites that are sufficiently reactive to bind covalently to the heme moiety of CYP, a process known as suicide inactivation or MDI (discussed later in the section “Inhibition of Cytochrome P450”). As previously discussed in the section “Epoxide Hydrolases,” not all epoxides are highly reactive electrophiles. Although the 3,4-epoxidation of coumarin produces a hepatotoxic metabolite, the 10,11-epoxidation of carbamazepine produces a stable, relatively nontoxic metabolite (Fig. 6-43). EETs are endogenous epoxides with vasodilatory, anti-inflammatory, and angiogenic properties (Imig and Hammock, 2009; Wang et al., 2010c). They are formed by CYP2C8, CYP2C9, and CYP2J2 (discussed later in this section) and inactivated by sEH (see the section “Epoxide Hydrolases”).

In the presence of NADPH and O₂, liver microsomes catalyze the oxygenation of several sulfur-containing xenobiotics, including chlorpromazine, cimetidine, lansoprazole, pantoprazole, and omeprazole. Sulfur-containing xenobiotics can potentially undergo 2 consecutive sulfoxidation reactions: one that converts the sulfide (S) to the sulfoxide (SO), which occurs during the sulfoxidation of chlorpromazine and cimetidine, and one that converts the sulfoxide (SO) to the sulfone (SO₂), which occurs during the sulfoxidation of omeprazole and lansoprazole, as shown in Fig. 6-44. Albendazole is converted first to a sulfoxide and then to a sulfone. Both CYP and FMO can sulfoxidate sulfides to sulfoxides (S → SO) but only CYP can covert sulfoxides to sulfones (SO → SO₂) (see the section “Flavin Monooxygenases”) (Testa and Krämer, 2008, 2010). Accordingly, the sulfoxidation of the proton pump inhibitors omeprazole, lansoprazole, and pantoprazole to sulfones is catalyzed by CYP (CYP3A4) and not by FMO. Examples of sulfoxidation reactions catalyzed by FMO and/or CYP are shown in Figs. 6-38B and 6-44. In the presence of NADPH and O₂, liver microsomes catalyze the oxygenation of several nitrogen-containing xenobiotics, including chlorpromazine, doxylamine, oflaxacin, morphine, nicotine, MPTP, methapyrilene, methaqualone, metronidazole, pargyline, pyridine, senecionine, strychnine, TMA, trimipramine, and verapamil, all of which are converted to stable N-oxides. Whereas S-oxygenation might be catalyzed by both CYP and FMO, N-oxygenation is more likely to be catalyzed by just one of these enzymes. FMO prefers to N-oxygenate aliphatic amines (particularly tertiary and cyclic amines) with a highly nucleophilic/basic nitrogen atom (pK_a 5–10) but, in contrast to CYP, FMO cannot N-oxygenate aromatic amines, for which reason the N-oxygenation of pyridines and (iso)quinolones is usually catalyzed only by CYP, as is the case for pyridine-containing xenobiotics such as the tobacco-specific nitrosamine NNK and the antihistamine temelastine, and the isoquinoline-containing muscle relaxant 6,7-dimethoxy-4-(4′-chlorobenzyl)isoquinoline. Methods to distinguish the role of CYP versus FMO in microsomal N- and S-oxygenation reactions are described in the section “Flavin Monooxygenases.”

Although direct oxygen transfer is possible (Li et al., 2009a), the initial step in heteroatom oxygenation by CYP could involve SET from the heteroatom (N, S, or I) to P450 compound I, shown as follows for sulfoxidation:

SET from N, O, or S to P450 compound I may also be the initial step in heteroatom dealkylation, but in this case abstraction of the electron from the heteroatom is quickly followed by abstraction of a proton (H⁺) from the α-carbon atom (the carbon atom attached to the heteroatom) to form a carbon radical. Oxygen rebound leads to hydroxylation of the α-carbon, which then rearranges to form the corresponding aldehyde or ketone with cleavage of the α-carbon from the heteroatom, as shown in the following scheme for the N-dealkylation of an N-alkylamine:

Alternatively, it seems more likely that the initial step in heteroatom dealkylation may be HAT from the α-carbon atom to compound I to produce a carbon radical that undergoes oxygen rebound and rearrangement as shown in the following scheme (Li et al., 2009a; Roberts and Jones, 2010):

In general, CYP catalyzes the N-dealkylation, not the N-oxygenation, of amines. N-Oxygenation by CYP can occur if the nitrogen is next to an electron-donating group (making the nitrogen electron rich) or if α-protons are either absent (eg, aromatic amines) or inaccessible (eg, quinidine). In the case of aromatic amines (anilines; aryl-NH₂), N-oxidation by CYP usually results in the formation of hydroxylamines (aryl-NHOH). Some hydroxylamines are further oxidized to the corresponding nitroso metabolite, as observed with sulfamethoxazole (aryl-NH₂ → aryl-NOH → aryl-NO). Primary aromatic amides (aryl-CONH₂) are not substrates for N-oxidation, but secondary aromatic amides (aryl-NHCOR) are often N-hydroxylated by CYP (often by CYP1A2) to produce N-hydroxyamines (also known as hydroxamic acids). N-Hydroxylamines and hydroxyamides are of toxicological interest because under acidic conditions they can dissociate to form reactive nitrenium ions. Phenacetin, an aromatic amide (aryl-NH−CO−CH₃), causes kidney toxicity because it is converted by N-hydroxylation to a hydroxamic acid (aryl-NOH−CO−CH₃) that in the low pH of urine undergoes acid-catalyzed conversion to a reactive nitrenium ion (aryl-NOH−CO−CH₃ + H⁺ → aryl-N⁺−CO−CH₃ + H₂O) (Testa and Krämer, 2008, 2010). N-Hydroxylation of aromatic amines with subsequent O-acetylation or O-sulfonation is 1 mechanism by which tumorigenic aromatic amines, such as 2-AAF and 4-aminobiphenyl, are converted to electrophilic reactive intermediates that bind covalently to DNA (discussed later in the sections “Glucuronidation and Formation of Acyl-CoA Thioesters” and “Acetylation”).

Numerous xenobiotics are N-, O-, or S-dealkylated by CYP (but not by FMO), and some examples of these heteroatom dealkylation reactions are shown in Fig. 6-45. The dealkylation of xenobiotics containing an N-, O-, or S-methyl group results in the formation of formaldehyde, which can easily be measured by a simple colorimetric assay to monitor the demethylation of substrates in vitro. The expiration of ¹³C- or ¹⁴C-labeled carbon dioxide following the demethylation of drugs containing a ¹³C- or ¹⁴C-labeled methyl group has been used to probe CYP activity in vivo (Watkins, 1994). The activity of the human CYP enzymes involved in the N-demethylation of aminopyrine, erythromycin, and caffeine can be assessed by this technique. Although caffeine has 3 N-methyl groups, all of which can be removed by CYP, the major pathway in humans involves N₃-demethylation of caffeine to paraxanthine (see Fig. 6-45). Fig. 6-45 also presents an example of a xenobiotic (octamethylcyclotetrasiloxane, a component of cosmetics and deodorants) that undergoes silicone demethylation by CYP (Varaprath et al., 1999).

Whereas the metabolism of N-methyl amines by CYP generally results in N-demethylation, with the hydroxylation of the methyl group leading to the release of formaldehyde, the metabolism of N-methyl amides and carbamates by CYP can result in the formation of a stable methyl-hydroxylated metabolite, one that does not release formaldehyde (which would otherwise complete an N-demethylation reaction) possibly because of the hydroxymethyl metabolite formation of a 6-membered configuration that is stabilized by hydrogen bonding, illustrated as follows:

Zolpidem and camazepam are amide- and carbamate-containing drugs, respectively, that undergo such N-methyl-hydroxylation reactions.

Seto and Guengerich have shown that the N-demethylation and N-deethylation of N-ethyl-N-methylaniline not only proceed at different rates (with N-demethylation proceeding up to 20 times faster than N-deethylation) but also proceed by different mechanisms (Seto and Guengerich, 1993; Guengerich, 2001a). The initial step in the N-deethylation reaction involves HAT from the α-carbon atom (ie, the carbon atom attached to the nitrogen), whereas the initial step in the N-demethylation reaction involves a much faster reaction, namely, a SET from the nitrogen atom, which is transferred to P450 compound I. Although the N-demethylation of N,N-substituted amines proceeds by the relatively rapid process of SET, the N-demethylation of N,N-substituted amides, where the adjacent carbonyl causes electrons to be withdrawn from the nitrogen atom, proceeds by the relatively slow process of HAT, for which reason the latter reactions, in contrast to the former, show a large intrinsic isotope effect when the hydrogen atoms are replaced with deuterium (because it requires more energy to break a C–D bond than a C–H bond).

In addition to N-dealkylation, some primary amines can also undergo oxidative deamination by CYP, which is an example of oxidative group transfer. The mechanism of oxidative deamination is similar to that of N-dealkylation: the α-carbon adjacent to the primary amine is hydroxylated, which produces an unstable intermediate that rearranges to eliminate ammonia with the formation of an aldehyde or ketone. The conversion of amphetamine to phenylacetone is an example of CYP-catalyzed oxidative deamination, as shown in Fig. 6-46. However, primary aliphatic amines tend to be poor substrates for CYP; in fact, primary alkylamines, such as the N-demethylated metabolite of fluoxetine, are metabolically stable and potent inhibitors of CYP inhibitors (discussed later in the section “Inhibition of Cytochrome P450”). Oxidative deamination is also catalyzed by MAO and FMO (Testa and Krämer, 2008, 2010). In the example given above, however, the substrate, amphetamine, contains an α-methyl group that renders it a poor substrate for MAO (as described in the section “Amine Oxidases”).

In addition to oxidative deamination, CYP catalyzes 2 other types of oxidative group transfer, namely, oxidative desulfuration and oxidative dehalogenation. In all cases the heteroatom (N, S, or halogen) is replaced with oxygen. As shown in Fig. 6-47, oxidative desulfuration converts parathion, which has little insecticidal activity, to paraoxon, which is a potent insecticide. The same reaction converts thiopental to pentobarbital. Diethyldithiocarbamate methyl ester, a metabolite of disulfiram, also undergoes oxidative desulfuration. The initial reaction involves S-oxidation by CYP or FMO to a sulfine (R₁R₂C=S → R₁R₂C=S⁺−O⁻). In the presence of GSH and GST, this sulfine either is converted back to the parent compound (R₁R₂C=S⁺−O⁻ + 2GSH → R₁R₂C=S + H₂O) or undergoes desulfuration (R₁R₂C=S + 2GSH → R₁R₂C=O + GSSG + H₂S) (Madan et al., 1994). CYP catalyzes both reductive and oxidative dehalogenation reactions (Guengerich, 1991). During oxidative dehalogenation, a halogen and hydrogen from the same carbon atom are replaced with oxygen (R₁R₂CHX → R₁R₂CO) to produce an aldehyde or acylhalide, as shown in Fig. 6-19 for the conversion of halothane (CF₃CHClBr) to trifluoroacetylchloride (CF₃COCl). Oxidative dehalogenation does not involve a direct attack on the carbon–halogen bond, but it involves the formation of an unstable halohydrin by oxidation of the carbon atom bearing the halogen substituent. The carbon–halogen bond is broken during the rearrangement of the unstable halohydrin. When the carbon atom contains a single halogen, the resulting product is an aldehyde, which can be further oxidized to a carboxylic acid or reduced to a primary alcohol. When the carbon atom contains 2 halogens, the dihalohydrin intermediate rearranges to an acylhalide, which can be converted to the corresponding carboxylic acid (see Fig. 6-19). Aldehydes and, in particular, acylhalides are reactive compounds that can bind covalently to protein and other critical cellular molecules. The immune hepatitis caused by repeated exposure of humans to halothane and related volatile anesthetics is dependent on oxidative dehalogenation by CYP, with neoantigens produced by the trifluoroacetylation of proteins, as shown in Fig. 6-19. Another example of oxidative dehalogenation by CYP is provided by studies of dasatinib metabolism by Li et al. (2009b), who, in order to prevent the formation of a reactive quinoneimine metabolite that bound to GSH, used fluoro and chloro substitution in an attempt to block para-hydroxylation of the substituted aniline in dasatinib, which is a metabolism-dependent inhibitor of CYP3A4. The halogen substitutions did not block the formation of GSH adducts of the para-hydroxylated metabolite. The mechanism of oxidative dehalogenation of the substituted aniline in dasatinib (followed by glutathionylation) was presumed to arise from an epoxide intermediate that underwent dehalogenation during its conjugation with GSH. Another example of oxidative defluorination is shown in Fig. 6-38 for the oxidative defluorination of 4-fluoro-N-methylaniline by FMO (Driscoll et al., 2010).

CYP can catalyze the reductive dehalogenation of halogenated alkanes (see Figs. 6-18 and 6-19) and the reduction of certain azo- and nitro-containing xenobiotics, although these latter reactions are largely catalyzed by gut microflora (see Figs. 6-11 and 6-12). The ability of CYP to reduce xenobiotics can be understood from the catalytic cycle shown in Fig. 6-40. Binding of a substrate to CYP is followed by a one-electron reduction by NADPH-cytochrome P450 reductase. Under aerobic conditions, reduction of the heme iron to the ferrous state permits binding of oxygen. Anaerobic conditions, in contrast, interrupt the cycle at this point, which allows CYP to reduce those substrates capable of accepting an electron. Therefore, CYP can catalyze reduction reactions, such as azo-reduction, nitro-reduction, N-oxide reduction, sulfoxide reduction, and reductive dehalogenation, particularly under conditions of low oxygen tension. In effect, the substrate rather than molecular oxygen accepts electrons and is reduced. In fact, oxygen acts as an inhibitor of these reactions because it competes with the substrate for the reducing equivalents. The toxicity of many halogenated alkanes is dependent on their biotransformation by reductive dehalogenation. The first step in reductive dehalogenation is a one-electron reduction catalyzed by CYP, which produces a potentially toxic, carbon-centered radical and inorganic halide. The conversion of CCl₄to a trichloromethyl radical and other toxic metabolites is shown in Fig. 6-18. The activation of the prodrug AQ4N (ie, 1,4-bis{[2-(dimethylamino-N-oxide)ethyl]amino}-5,8-dihydroxyanthracene-9,10-dione) by N-oxide reduction to form 4QA (ie, 1,4-bis{[2-(dimethylamino)ethyl]amino}-5,8-dihydroxyanthracene-9,10-dione), a potent topoisomerase II inhibitor, by the hypoxia-inducible enzymes CYP2S1 and CYP2W1 represents a potentially new cancer chemotherapy, as discussed later in this section.

The oxidative desulfuration of parathion (see Fig. 6-46) involves the production of an intermediate that rearranges to paraoxon (as shown in Fig. 6-47). This same intermediate can decompose to 4-nitrophenol and diethylphosphorothioic acid, which are the same products formed by the hydrolysis of parathion (Fig. 6-47). In addition to facilitating the hydrolysis of phosphoric acid esters, CYP also catalyzes the cleavage of certain carboxylic acid esters and carbamates, as shown in Fig. 6-47. Carboxylic acid esters typically are cleaved by carboxylesterases and cholinesterases (see the section “Hydrolysis”), which results in the formation of an acid and an alcohol (R₁COOCH₂R₂ + H₂O → R₁COOH + R₂CH₂OH). In contrast, CYP converts carboxylic acid esters to an acid plus aldehyde (R₁COOCH₂R₂ + [O] → R₁COOH + R₂CHO), as shown in Fig. 6-47. The deacylation of loratadine, a carbamate, is the major route of biotransformation of this nonsedating antihistamine. The reaction is catalyzed predominantly by CYP (namely, CYP3A4 with a minor contribution from CYP2D6), with little contribution from carboxylesterases (Yumibe et al., 1996).

CYP can also catalyze the dehydrogenation of a number of compounds, including acetaminophen, nifedipine, and related dihydropyridine calcium channel blockers, sparteine, nicotine, and testosterone, as shown in Fig. 6-48. Dehydrogenation by CYP converts acetaminophen to its hepatotoxic metabolite, N-acetylbenzo-p-quinoneimine (NAPQI), as shown in Fig. 6-35. The formation of a double bond during the conversion of digitoxin (dt₃) to 15′-dehydro-dt₃ leads to cleavage of the terminal sugar residue to produce digitoxigenin bisdigitoxoside (dt₂) (Fig. 6-48), which can similarly be converted to 9′-dehydro-dt₂, which undergoes digitoxosyl cleavage to digitoxigenin monodigitoxoside (dt₁). In contrast to digitoxin, the latter metabolite is an excellent substrate for glucuronidation. In rats, the CYP enzymes responsible for converting digitoxin to dt₁ (namely, the CYP3A enzymes) and the UGT responsible for glucuronidating dt₁ are inducible by dexamethasone, PCN, and spironolactone, all of which protect rats from the toxic effects of digitoxin. The dehydrogenation of nicotine produces nicotine Δ^1′,5′-iminium ion, which is oxidized by cytosolic AO to cotinine, a major metabolite of nicotine excreted in the urine of cigarette smokers (see Fig. 6-48). Although nicotine can be N-oxygenated by FMO, dehydrogenation by CYP2A6 is responsible for 80% of the clearance of nicotine from cigarette smoking (Hukkanen et al., 2005).

Testosterone is dehydrogenated by CYP to 2 metabolites: 6-dehydrotestosterone, which involves formation of a carbon–carbon double bond, and androstenedione, which involves formation of a carbon–oxygen double bond, as shown in Fig. 6-48. The conversion of testosterone to androstenedione is one of several cases where CYP converts a primary or secondary alcohol to an aldehyde or ketone, respectively. The reaction can proceed by formation of a gem-diol (2 hydroxyl groups on the same carbon atom), with subsequent dehydration to a keto group, as shown in Fig. 6-25 for the conversion of ethanol to acetaldehyde. However, gem-diols are not obligatory intermediates in the oxidation of alcohols by CYP, and in fact the conversion of testosterone to androstenedione by CYP2B1 (the major phenobarbital-inducible CYP enzyme in rats) does not involve the intermediacy of a gem-diol but proceeds by direct dehydrogenation (Fig. 6-48). In contrast, a gem-diol is involved in the formation of androstenedione from epi-testosterone (which is identical to testosterone except the hydroxyl group at C₁₇ is in the α-configuration, not the β-configuration) (Wood et al., 1988). The fact that formation of androstenedione from epi-testosterone involves formation of a gem-diol, whereas its formation from testosterone does not, makes it difficult to generalize the mechanism by which CYP converts alcohols to aldehydes and ketones.

Liver microsomes from all mammalian species contain numerous CYP enzymes (Table 6-11), each with the potential to catalyze the various types of reactions shown in Figs. 6-41 to 6-48. In other words, all of the CYP enzymes expressed in liver microsomes have the potential to catalyze xenobiotic hydroxylation, epoxidation, dealkylation, oxygenation, dehydrogenation, and so forth. The broad and often overlapping substrate specificity of liver microsomal CYP enzymes precludes the possibility of naming these enzymes for the reactions they catalyze, which are now categorized into families and subfamilies and named individually on the basis of their amino acid sequence. As shown in Tables 6-10 and 6-11, the CYP enzymes involved in xenobiotic biotransformation belong mainly to the CYP1, 2, and 3 gene families, although on a case-by-case basis CYP enzymes in other gene families play a key role in xenobiotic biotransformation. The CYP enzymes involved in endobiotic metabolism generally have the same name in all mammalian species. Some of the xenobiotic-biotransforming CYP enzymes have the same name in all mammalian species, whereas others are named in a species-specific manner. For example, all mammalian species contain 2 CYP enzymes belonging to the CYP1A subfamily, and in all cases these are known as CYP1A1 and CYP1A2 because the function and regulation of these enzymes are highly conserved among mammalian species. The same is true of CYP1B1, 2E1, 2R1, 2S1, 2U1, and 2W1, all of which are highly conserved homologs that can be given the same name across mammalian species. In most other cases, functional or evolutionary relationships are not immediately apparent; hence, the CYP enzymes are named in a species-specific manner, and the names are assigned in chronological order regardless of the species of origin. For example, human liver microsomes express CYP2D6, but this is the only functional member of the CYP2D subfamily found in human liver. CYP2D7 and 2D8 are human pseudogenes. The other members of this subfamily (ie, CYP2D1-CYP2D5 and CYP2D9 onward) are the names given to CYP2D enzymes in other species.

Without exception, the levels and activity of each CYP enzyme have been shown to vary from one individual to the next, due to environmental and/or genetic factors. Decreased CYP enzymatic activity can result from (1) a genetic mutation that either blocks the synthesis of a CYP enzyme or leads to the synthesis of a catalytically compromised, inactive, or unstable enzyme, which gives rise to the PM and IM genotypes; (2) exposure to an environmental factor (such as an infectious disease or an inflammatory process) that suppresses CYP enzyme expression, or (3) exposure to a xenobiotic that inhibits or inactivates a preexisting CYP enzyme. By inhibiting CYP, one drug can impair the biotransformation of another, which may lead to an exaggerated pharmacological or toxicological response to the second drug. In this regard, inhibition of CYP by a drug (and suppression of CYP by infection, vaccination, or inflammation) essentially mimics the effects of a genetic deficiency in CYP enzyme expression (ie, these environmental factors mimic the IM or PM genotype depending on the degree to which they decrease CYP activity). Genetic deficiencies in CYP expression, CYP inhibition, and, to a lesser extent, CYP suppression all contribute significantly to interindividual variability in drug metabolism and toxicity, and inhibition of CYP activity is a major cause of drug–drug interactions. Examples of the impact of these genetic and environmental factors on drug metabolism and toxicity are given later in this section in the overviews of individual CYP enzymes and in the section “Inhibition of Cytochrome P450.”

Increased CYP enzymatic activity can result from (1) gene duplication leading to overexpression of a CYP enzyme, which gives rise to the UM genotype; (2) gene mutations in the coding or promoter region that increase expression, activity, or stability of CYP; (3) exposure to drugs and other xenobiotics that induce the synthesis or retard the degradation of CYP; or (4) exposure to drugs and other xenobiotics that stimulate the activity of a preexisting enzyme (a process known as homotropic or heterotropic activation depending on whether the drug stimulates its own metabolism or the metabolism of other drugs, respectively). Activation of CYP has been documented in vitro and in some in situ situations, such as the pronounced (up to 25-fold) activation of R-warfarin 10-hydroxylation by quinidine in rabbit liver microsomes and perfused rabbit liver (Chen et al., 2004a). However, in general, activation does not appear to be a major cause of drug–drug interactions. Although duplication of functional CYP2D6 genes has been documented (and shown to be relevant to drug metabolism and safety [see the section “CYP2D6”]), induction of CYP by xenobiotics is the most common mechanism by which CYP enzyme activity is increased to a pharmacologically relevant extent. By inducing CYP, one drug can stimulate the metabolism of a second drug and thereby decrease or ameliorate its therapeutic effect. Enzyme induction is a particular concern when it compromises the therapeutic effectiveness of drugs that have a narrow therapeutic index and are being used to treat a life-threatening illness, such as anti-HIV drugs, antirejection drugs (such as cyclosporine and tacrolimus), and oral anticoagulants (such as warfarin), or when it is used with drugs that exhibit a quantal (all-or-nothing) dose–response relationship, such as oral contraceptive steroids (which either block or do not block ovulation and thereby provide or do not provide protection against pregnancy).

The environmental factors known to affect P450 levels include medications (eg, prescription drugs as well as herbal remedies), foods (eg, cruciferous vegetables, charcoal-broiled beef), social habits (eg, alcohol consumption, cigarette smoking), and disease status (diabetes, infection, inflammation, vaccination, liver and kidney disease, and both hyperthyroidism and hypothyroidism). When environmental factors influence CYP enzyme levels, considerable variation may be observed during repeated measures of xenobiotic biotransformation (eg, drug metabolism) in the same individual. Such variation is not observed when alterations in CYP activity are determined genetically.

Due to their broad substrate specificity, it is possible that 2 or more CYP enzymes can contribute to the metabolism of a single compound. Three human CYP enzymes, CYP1A2, CYP2E1, and CYP3A4, can convert the commonly used analgesic, acetaminophen, to its hepatotoxic metabolite, NAPQI (Figs. 6-34 and 6-48). It is also possible for a single CYP enzyme to catalyze 2 or more metabolic pathways for the same drug. For example, CYP2D6 catalyzes both the O-demethylation and 5-hydroxylation (aromatic ring hydroxylation) of methoxyphenamine, and CYP3A4 catalyzes the 3-hydroxylation and N-oxygenation of quinidine, the 1′- and 4-hydroxylation of midazolam, the tert-butyl-hydroxylation and N-dealkylation of terfenadine, and several pathways of testosterone oxidation, including 1β-, 2β-, 6β-, and 15β-hydroxylation and dehydrogenation to 6-dehydrotestosterone (Figs. 6-41 and 6-48).

The pharmacological or toxic effects of certain drugs are exaggerated in a significant percentage of the population due to a heritable deficiency in a CYP enzyme (Meyer, 1994; Tucker, 1994; Zhou et al., 2009b). The observation that individuals who are genetically deficient in a particular CYP enzyme are PMs of one or more drugs illustrates a very important principle, namely, that the rate of elimination of drugs can be largely determined by a single CYP enzyme. This observation seems to contradict the fact that CYP enzymes have broad and overlapping substrate specificities. The resolution to this apparent paradox lies in the fact that although more than one human CYP enzyme can catalyze the biotransformation of a xenobiotic, they may do so with markedly different affinities. Consequently, xenobiotic biotransformation in vivo, where only low substrate concentrations are usually achieved, is often determined by the CYP enzyme with the highest affinity (lowest apparent K_m) for the xenobiotic. For example, the 5-hydroxylation of lansoprazole, which represents the key route of elimination of this proton pump inhibitor, is catalyzed by both CYP2C19 and CYP3A4. However, these reactions are catalyzed by CYP3A4 with such low affinity that the 5-hydroxylation (and hence the clearance) of lansoprazole in vivo is largely determined by CYP2C19 (Pearce et al., 1996b). When several CYP enzymes catalyze the same reaction, their relative contribution to xenobiotic biotransformation is determined by the kinetic parameter, V_max/K_m, which is a measure of in vitro intrinsic clearance at low substrate concentrations (<10% of K_m).

A drug whose clearance is largely determined by a single CYP enzyme (or any single route of elimination) is said to have high victim or object potential, meaning its rate of clearance will be decreased by genetic polymorphisms and inhibitory drugs that result in a loss of that particular CYP activity and increased by drugs and other xenobiotics that induce that particular CYP enzyme. For example, a drug that is largely cleared by CYP2D6 will be slowly metabolized in CYP2D6 PMs and rapidly metabolized in CYP2D6 UMs, whereas a drug whose clearance is largely determined by CYP3A4 will be slowly metabolized in the presence of ketoconazole (and other CYP3A4 inhibitors) and rapidly metabolized in the presence of rifampin or St. John's wort (and other CYP3A4 inducers). The extent to which genetic polymorphisms and CYP inhibitors/inducers impact the disposition of a drug (or any other xenobiotic) is determined by fm, the fraction of clearance attributable to the affected enzyme. This principle applies to all xenobiotic-biotransforming enzymes and is explained in detail in Point 24 in the section “Introduction.” Point 24 also covers the principle of drug–drug interactions in terms of perpetrator–victim interactions (in which the disposition of one drug, the victim, is impacted by a single perpetrator such as an inhibitory drug or a genetic polymorphism) and perpetrator–perpetrator–victim interactions, otherwise known as “maximum exposure,” which involves the dramatic (synergistic) impact of 2 perpetrators blocking 2 parallel pathways of clearance of the victim drug (Collins et al., 2006; Ogilvie and Parkinson, in press).

Accordingly, for drugs under development, considerable attention is paid to identifying which CYP enzyme or enzymes are involved in eliminating the drug, a process known as reaction phenotyping or enzyme mapping. Four in vitro approaches have been developed for reaction phenotyping (Ogilvie et al., 2008). Each has its advantages and disadvantages, and a combination of approaches is usually required to identify which human CYP enzyme is responsible for metabolizing a xenobiotic. The 4 approaches to reaction phenotyping are as follows:

1. Correlation analysis, which involves measuring the rate of xenobiotic metabolism by several samples of human liver microsomes and correlating reaction rates with the variation in the level or activity of the individual CYP enzymes in the same microsomal samples. This approach is successful because the levels of the CYP enzymes in human liver microsomes vary enormously from sample to sample as shown in Table 6-12.

2. Chemical inhibition, which involves an evaluation of the effects of known CYP enzyme inhibitors on the metabolism of a xenobiotic by human liver microsomes. Chemical inhibitors of CYP, which are discussed later, must be used cautiously because most of them can inhibit more than one CYP enzyme. Some chemical inhibitors are metabolism-dependent inhibitors that require biotransformation to a metabolite that inactivates or noncompetitively inhibits CYP.

3. Antibody inhibition, which involves an evaluation of the effects of inhibitory antibodies against selected CYP enzymes on the biotransformation of a xenobiotic by human liver microsomes. Due to the ability of antibodies to inhibit selectively and noncompetitively, this method alone can potentially establish which human CYP enzyme is responsible for biotransforming a xenobiotic. Unfortunately, the utility of this method is limited by the availability of specific inhibitory antibodies.

4. Biotransformation by purified or recombinant (cDNA-expressed) human CYP enzymes, which can establish whether a particular CYP enzyme can or cannot biotransform a xenobiotic, but it does not address whether that CYP enzyme contributes substantially to reactions catalyzed by human liver microsomes. The information obtained with purified or recombinant human CYP enzymes can be improved by taking into account large differences in the extent to which the individual CYP enzymes are expressed in human liver microsomes. The specific content of the major xenobiotic-biotransforming enzymes in human liver microsomes has been determined by mass spectrometry and the values are summarized in Table 6-12. The extrapolation of metabolic rates obtained with recombinant enzymes to those expected to occur in liver microsomes can be based on specific content, relative activity factors (RAF), or a combination of both called intersystem extrapolation factor (ISEF) (Proctor et al., 2004; Ogilvie et al., 2008).