26510318-Bio-Transformation-of-Xenobiotics

biotransformation. In many instances, however, P450 induction does not necessarily enhance the biotransformation of the inducer, in which case the induction is said to be gratuitous. Drugs are also known to induce enzymes that play no role in their biotransformation. For example, omeprazole induces CYP1A2, even though the disposition of this acid-suppressing drug is largely determined by CYP2C19. Some of the most effective inducers of cytochrome P450 are polyhalogenated aromatic hydrocarbons, such as polychlorinated derivatives of dibenzo-p-dioxin (PCDDs), dibenzofurans (PCDFs), azobenzenes and azoxybenzenes, biphenyl (PCBs), and naphthalene. In general, highly chlorinated compounds are resistant to biotransformation and cause a prolonged induction of cytochrome P450 and other enzymes. Due to the increased demand for heme, persistent induction of cytochrome P450 can lead to porphyria, a disorder characterized by excessive accumulation of intermediates in the heme biosynthetic pathway. In 1956, widespread consumption of wheat contaminated with the fungicide hexachlorobenzene caused an epidemic of porphyria cutanea tarda in Turkey. Another outbreak occurred in 1964 among workers at a factory in the United States manufacturing 2,4,5-trichlorophe- noxyacetic acid (the active ingredient in several herbicides and in the defoliant Agent Orange). The outbreak of porphyria cutanea tarda was caused not by the herbicide itself but by a contaminant, 2,3,7,8-tetrachlorodibenzo-p-dioxin, also known as dioxin and TCDD. Drugs that cause P450 induction have not been shown to cause porphyria cutanea tarda under normal circumstances, but phenobarbital, phenytoin, and alcohol are recognized as precipitating factors because they cause episodes of porphyria cutanea tarda in individuals with an inherited deficiency in the hemebiosynthetic enzyme uroporphyrinogen decarboxylase.

The mechanism of P450 induction has been studied extensively in rats and other laboratory animals (Gonzalez, 1989; Okey, 1990; Ryan and Levin, 1990; Porter and Coon, 1991). Currently, five classes of P450 enzyme inducers are recognized, which are represented by 3-methylcholanthrene, phenobarbital, preg- nenolone-16 -carbonitrile (PCN), clofibric acid, and isoniazid. The first four inducers cause a marked ( 10-fold) increase in the rate of transcription of one or more specific P450 enzymes. For each of these four inducers, the trans-acting factor that mediates the transcriptional activation of the P450 genes has been identified, as shown in Table 6-7. For the most part, these trans-acting factors are ligand-activated receptors that must dimerize with another protein in order to form a DNA-binding protein that can bind to discrete regions of DNA (cis-acting factors or response elements) and thereby activate gene transcription (Honkakoski and Negishi, 2000; Waxman, 1999).

Treatment of rats with 3-methylcholanthrene causes a marked ( 20-fold) induction of CYP1A1 and CYP1A2. Liver microsomes

Table 6-7

Receptors Mediating the Induction of P450 Enzymes

from untreated rats contain low levels of CYP1A2 and virtually undetectable levels of CYP1A1. In addition to polycyclic aromatic hydrocarbons, such as 3-methylcholanthrene and benzo[a]pyrene, inducers of the CYP1A enzymes include flavones (e.g., -naph- thoflavone), polyhalogenated aromatic hydrocarbons (e.g., TCDD, 3,3 ,4,4 ,5,5 -hexachlorobiphenyl), acid condensation products of indole-3-carbinol, and certain drugs and food additives (e.g., chlorpromazine, phenothiazine, clotrimazole, ketoconazole, miconazole, isosafrole). Induction of CYP1A1 involves transcriptional activation of the CYP1A1 gene, which, together with message stabilization, results in an increase in the levels of mRNA and newly synthesized protein.

In the absence of an inducer, transcription of the CYP1A1 gene is suppressed by a repressor protein, which accounts for the low constitutive levels of CYP1A1 in most species. (Guinea pig and rhesus monkey appear to express CYP1A1 constitutively and hence are exceptions to this general rule). Induction of CYP1A1 involves both derepression and activation of transcription by the Ah receptor. Although this cytosolic receptor binds several aromatic hy- drocarbons, such as 3-methylcholanthrene and benzo[a]pyrene, the ligand with the highest binding affinity is TCDD, which is why the Ah receptor is also known as the dioxin receptor (Whitlock, 1993). The Ah receptor is normally complexed in a 1:2 ratio with heatshock protein (hsp90), which dissociate upon binding of ligand to the Ah receptor, enabling the receptor to be phosphorylated by tyrosine kinase. The activated Ah receptor then enters the nucleus and forms a heterodimer complex with the Ah-receptor-nuclear translocator Arnt. Inside the nucleus, the Ah receptor-Arnt complex binds to regulatory sequences [known as dioxin-responsive elements (DRE) or xenobiotic responsive elements (XRE)] and enhances the transcription of the CYP1A1 gene and other genes with an XRE or XRE-like sequence in their upstream enhancer region [namely CYP1A2, DT-diaphorase, glutathione S-transferase, UDPglucuronosyltransferase (UGT1A6 and UGT1A7), and aldehyde dehydrogenase]. The XRE is only a small segment of DNA (the consensus sequence is 5 -TXGCGTG-3 , where X is normally T or A), which can be located more than a thousand bases from the initiation site for transcription. The enhancer region of the CYP1A1 gene contains multiple XREs, which accounts for the marked ( 100-fold) increase in CYP1A1 mRNA and protein levels following exposure to ligands for the Ah receptor.

Arnt was initially thought to be a cytosolic protein that simply facilitates the translocation of the ligand-bound Ah receptor into the nucleus. It is now recognized as an important component of the receptor complex that binds to DNA and activates transcription of genes under the control of the Ah receptor. For heterodimer formation, the Ah receptor must be bound to ligand and possibly phosphorylated, and Arnt must be phosphorylated, ap-

parently by protein kinase C. The Ah receptor is often compared with the steroid/thyroid/retinoid family of receptors, which also bind ligands in the cytoplasm and are translocated to the nucleus where they bind to DNA and enhance gene transcription. However, the Ah receptor is a novel ligand-activated transcription factor, very distinct from these other receptors. Whereas the steroid/thyroid/ retinoid receptors have “zinc-finger” DNA-binding domains and form homodimers, the Ah receptor forms a heterodimer with Arnt; both of these contain a basic helix-loop-helix (bHLH) domain near their N-terminus. The basic region binds DNA and the helix- loop-helix is involved in protein–protein interactions. The XRE recognized by the Ah receptor-Arnt complex contains a sequence of four base pairs (5 -GCGT-3 ) that is part of the recognition motif for other bHLH proteins (Whitlock, 1993).

In mice, the Ah receptor is encoded by a single gene, but there are four allelic variants: a low-affinity form known as Ahd (an104-kDa protein expressed in strains DBA/2, AKR, and 129) and three high-affinity forms known as Ahb 1 (an 95-kDa protein expressed in C57 mice), Ahb 2 (an 104-kDa protein expressed in BALB/c, C3H, and A mice), and Ahb 3 (an 105-kDa protein expressed in MOLF/Ei mice) (Poland et al., 1994). Mice that express the low-affinity form of the receptor (Ahd) require higher doses ( 10 times) of TCDD to induce CYP1A1. Even though they respond to high doses of TCDD, Ahd mice are called nonresponsive because it is not possible to administer sufficient amounts of polycyclic aromatic hydrocarbons to cause induction of CYP1A1. Although several genetic alterations give rise to the four allelic variants of the Ah receptor, the low-affinity binding of ligands to the Ahd receptor from nonresponsive mice is attributable to a single amino acid substitution (Ala375 Val375).

In vivo, the Ahd and Ahb genotypes can be distinguished by phenotypic differences in the effects of 3-methylcholanthrene treatment on the duration of action of the muscle relaxant zoxazolamine. Treatment of nonresponsive (Ahd) mice with 3-methyl- cholanthrene results in no change in zoxazolamine-induced paralysis time. In contrast, such treatment of responsive (Ahb) mice causes an induction of CYP1A1, which accelerates the 6-hydrox- ylation of zoxazolamine and reduces paralysis time from about 1 h to several minutes.

Some of the genes regulated by the Ah receptor contain other responsive elements and their expression is controlled by other transcription factors (Nebert, 1994). To a limited extent, the induction of CYP1A1 by polycyclic aromatic hydrocarbons can also be mediated by another cytosolic receptor known as the 4Sbinding protein, which has been identified as the enzyme glycine N-methyltransferase. The first intron of the CYP1A1 gene contains a glucocorticoid-responsive element (GRE); hence the induction of CYP1A1 by TCDD can be augmented by glucocorticoids. DTdiaphorase is inducible up to tenfold by two classes of agents: chemicals like 3-methylcholanthrene and TCDD that bind to the Ah receptor and chemicals that cause oxidative stress, such as menadione, tert-butylhydroquinone, and 3,5-di-tert-butylcatechol, which produce reactive oxygen species through redox cycling reactions. These latter effects are mediated by the antioxidant responsive elements (ARE); therefore these enzymes are inducible by so-called monofunctional agents that do not induce CYP1A1 (see the section on DT-diaphorase above). The flavonoid -naph- thoflavone, the polycyclic aromatic hydrocarbon benzo[a]pyrene, and the polyhalogenated aromatic hydrocarbon TCDD all induce DT-diaphorase by both mechanisms; the parent compound binds

to the Ah receptor and is responsible for inducing CYP1A1 as well as DT-diaphorase via the XRE, whereas electrophilic and/or redox active metabolites of -naphthoflavone, benzo[a]pyrene, and TCDD are responsible for inducing glutathione S-transferase as well as DT-diaphorase via the ARE (Radjendirane and Jaiswal, 1999). The ARE core sequence for enzyme induction (5 -GTGA- CAAAGC-3 ) is similar to the AP-1 DNA-binding site (5 - TGACTCA-3 ), which is regulated by redox status (see the earlier section on quinone reduction).

The enhancer region of CYP1A2 also contains an XRE (or XRE-like sequences), so inducers of CYP1A1 are also inducers of CYP1A2. However, the induction of CYP1A2 differs from that of CYP1A1 in several respects: it occurs at lower doses of inducer; it often involves stabilization of mRNA or enzyme from degradation, and it requires liver-specific factors. The first of these differences (dose–response) may explain why low-level exposure of humans to CYP1A inducers results in an increase in hepatic levels of CYP1A2 but not CYP1A1. The second of these differences (namely stabilization of mRNA and/or enzyme) apparently explains why compounds that form stable complexes with CYP1A2, such as isosafrole, can induce CYP1A2 even in nonresponsive (Ahd) mice. The third difference (the requirement for hepatic factors) explains why CYP1A2 is not expressed or inducible in extrahepatic tissues, hepatoma-derived cell lines, or isolated hepatocytes cultured under conditions that do not restore liver-specific gene expression. CYP1B1 is also regulated by the Ah receptor. In contrast to CYP1A2, CYP1B1 is primarily expressed in extrahepatic tissues, which is also true of human CYP1A1.

Liver microsomes from untreated rats contain low levels of CYP2B2 and extremely low or undetectable levels of CYP2B1, which are structurally related enzymes (97 percent identical) with very similar substrate specificities. Treatment of rats with phenobarbital causes a marked ( 20-fold) induction of cytochromes CYP2B1 and CYP2B2. In addition to barbiturates, such as phenobarbital and glutethimide, inducers of the CYP2B enzymes include drugs (e.g., phenytoin, loratadine, doxylamine, griseofulvin, chlorpromazine, phenothiazine, clotrimazole, ketoconazole, miconazole), pesticides (e.g., DDT, chlordane, dieldrin), food additives (e.g., butylated hydroxytoluene and butylated hydroxyanisole), personal care ingredients [e.g., octamethylcyclotetrasiloxane (or D4)] and certain polyhalogenated aromatic hydrocarbons [e.g., 2,2 ,4,4 ,5,5 -hexachlorobiphenyl and 1,4- bis[2]-(3,5-chloropyridyloxy)benzene or TCPOBOP]. Treatment of rats with phenobarbital also results in a twoto fourfold increase in the levels of CYP2A1, CYP2C6, and CYP3A2, as well as a 50 to 75 percent decrease in the levels of CYP2C11, which is present only in adult male rats. In addition, treatment of rats with phenobarbital generally causes an increase (50 to 300 percent) in the concentration of cytochrome b5, NADPH-cytochrome P450 (c) reductase, epoxide hydrolase, aldehyde dehydrogenase, glutathione S-transferase and UDP-glucuronosyltransferase. Indeed treatment of rodents with phenobarbital and related inducers causes hepatocellular hyperplasia and/or hypertrophy, which is accompanied by a proliferation of the endoplasmic reticulum.

The mechanism of induction of CYP2B1 by phenobarbital has not been elucidated in as much detail as the induction of CYP1A1 by TCDD. As in the case of CYP1A1, induction of CYP2B1 involves transcriptional activation of the CYP2B1 gene, which, together with message stabilization, results in an increase in the levels of mRNA and newly synthesized protein. As shown in Table 6-7,

CYP2B1 induction by phenobarbital appears to be mediated by CAR , which must dimerize with another nuclear receptor, the retinoid X receptor (RXR), which is activated by 9-cis-retinoic acid. CAR is a constitutively activated receptor, meaning it is active as a DNA-binding protein in the absence of a bound ligand. However, androstanol and androstenol have been identified as high affinity CAR ligands, although these steroids are not CYP2B1 inducers. It would appear that androstanol and androstenol (or related steroids) bind to CAR and block its inherent ability to function as a DNA-binding protein. Accordingly, induction by phenobarbital appears to involve the displacement of these androstanes from CAR . Once derepressed, CAR can dimerize with RXR, enter the nucleus and bind to the phenobarbital-responsive element (PBRE) preceding the CYP2B1 gene and numerous other phenobarbital-responsive genes. The fact that, according to this model, phenobarbital-type inducers need only displace androstanes from CAR , which presumably does not require highly specific binding, may help to explain why CYP2B1 inducers lack any discernible structure-activity relationship.

The bacterium B. megaterium contains two cytochrome P450 enzymes, known as P450BM-1 and P450BM-3, that are inducible by phenobarbital and other compounds that induce rat CYP2B1 (Liang et al., 1995). The 5 -enhancer regions of these bacterial genes contain a 15-base-pair DNA sequence that, when deleted or mutated, results in P450BM-1 or P450BM-2 expression. A similar 15-base-pair DNA sequence, which is known as the Barbie box (after barbiturate), is present in the 5 -enhancer region of CYP2B1, CYP2B2, and numerous other mammalian genes that are transcriptionally activated by phenobarbital. All Barbie boxes contain a four base pair sequence (5 -AAAG-3 ), making this the likely site of DNA-protein interactions. In the absence of inducer, a repressor protein, Bm3R1, binds to the Barbie box and impedes or prevents transcription of the structural gene. Binding of the inducer to this repressor causes its dissociation from the Barbie box, which results in increased transcription. However, in light of the discoveries surrounding CAR , it is now clear that the key feature of the bacterial phenobarbital induction mechanism, namely removal of the repressor protein Bm3R1 from the Barbie box, is not the mechanism of phenobarbital induction in mammals.

Treatment of rodents with PCN causes an induction of CYP3A1 and CYP3A2—two independently regulated CYP3A enzymes with very similar structures (87 percent similar) and substrate specificities. In contrast to CYP3A1, CYP3A2 is present in liver microsomes from untreated rats, although the levels of this enzyme decline markedly after puberty in female rats. Consequently, CYP3A2 is a male-specific protein in mature rats, and it is inducible in mature male but not mature female rats (whereas CYP3A1 is inducible in mature male and female rats). In addition to PCN, inducers of CYP3A enzymes include steroids (e.g., dexamethasone and spironolactone), macrolide antibiotics (e.g., troleandomycin and erythromycin estolate), and azole antifungals (e.g., clotrimazole, ketoconazole, and miconazole). Although it primarily induces the CYP2B enzymes (see above), phenobarbital is also an inducer of CYP3A2.

The induction of CYP3A1, like that of CYP1A1 and CYP2B1, involves transcriptional activation of the structural gene, which, together with message stabilization, results in an increase in the levels of CYP3A1 mRNA and newly synthesized protein. The mechanism of induction of CYP3A1 involves the binding of PCN and related inducers to an orphan receptor known as the pregnane-X

receptor (PXR). The ligand-activated PXR dimerizes with RXR (which is activated by 9-cis-retinoic acid), and the heterodimer is translocated to the nucleus where it binds to response elements that activate the transcription of CYP3A1 and other PCN-inducible genes. The ligand-binding properties of PXR vary among mammalian species. PCN is a ligand for rodent PXR, whereas rifampin is not. The converse is true of human PXR, which explains why PCN, but not rifampin, is an effective CYP3A inducer in rodents, whereas rifampin, but not PCN, is an effective inducer of human CYP3A4.

In the case of macrolide antibiotics, such as troleandomycin and erythromycin, induction of CYP3A1 involves both transcriptional activation and stabilization of the newly synthesized enzyme against protein degradation. This latter effect involves biotransformation of the macrolide antibiotic to a metabolite that binds tightly to the heme moiety. The induction of CYP3A1 by macrolide antibiotics is often masked by their ability to function as metabolismdependent inhibitors. The enzyme-inducing effects of clotrimazole, ketoconazole, and miconazole are similarly masked by the ability of these azole antimycotics to bind to and inhibit cytochrome P450 enzymes (including CYP3A1). Substrates and ligands stabilize CYP3A1 by inhibiting its cAMP-dependent-phosphorylation on Ser393, which otherwise denatures the protein and targets it for degradation in the endoplasmic reticulum.

Treatment of male rats with clofibric acid causes a marked induction (up to 40-fold) of CYP4A1, CYP4A2, and CYP4A3, which are three independently regulated CYP4A enzymes with similar substrate specificities (Sundseth and Waxman, 1991). CYP4A1 is expressed in the liver, whereas CYP4A3 is expressed in the liver and kidney. CYP4A2 is expressed in the liver and kidney of male rats, but it is neither expressed nor inducible in female rats. In addition to clofibric acid, inducers of CYP4A enzymes include perfluorodecanoic acid, phthalate ester plasticizers, 2,4-dichlorophe- noxyacetic acid (2,4-D), ciprofibrate and other hypolipidemic drugs, aspirin and other NSAIDs, nicotinic acid, dehydroepiandrosterone sulfate, and leukotriene receptor antagonists (MK-0571 and RG 7512). A feature common to all these CYP4A enzyme inducers is their ability to cause proliferation of hepatic peroxisomes.

As in the case of CYP1A1, CYP2B1, and CYP3A1, the induction of CYP4A enzymes by clofibric acid involves transcriptional activation of the structural gene, which results in an increase in the levels of mRNA and newly synthesized protein. The transcription factor that activates the CYP4A genes is the peroxisome proliferator-activated receptor (PPAR ), a member of the steroid/ thyroid/retinoid superfamily of nuclear receptors that regulates transcription of the genes for fatty acyl-CoA oxidase, bifunctional enzyme (enoyl-CoA hydratase/3-hydroxyacyl-CoA dehydrogenase), and fatty acid binding protein (Muerhoff et al., 1992; Demoz et al., 1994). Binding of a peroxisome proliferator to PPAR results in the formation of a heterodimer with RXR, which is activated by 9-cis-retinoic acid. The ligand-bound heterodimer of PPAR and RXR binds to a regulatory DNA sequence known as the peroxisome proliferator response element. Three PPREs have been identified in the 5 -enhancer region of the rabbit CYP4A6 gene. Each of these elements contains a sequence known as DR1, an imperfect repeat of the nuclear receptor binding consensus sequence separated by one nucleotide (PuGGTCA N PuGGTCA). Peroxisome proliferators can bind to PPAR stereoselectively; therefore, the enantiomers of certain drugs differ in their ability to

induce CYP4A and cause a proliferation of peroxisomes. Androsterone sulfate and arachidonic acid are physiologic ligands for PPAR .

Treatment of rats with isoniazid causes a twoto fivefold induction of CYP2E1. In sexually mature rats, the levels of liver microsomal CYP2E1 are slightly greater in female than in male rats. In addition to isoniazid, inducers of CYP2E1 include ethanol, acetone, pyrazole, pyridine, ketoconazole, fasting, and uncontrolled diabetes. A common feature of CYP2E1 inducers is their ability to inhibit or be biotransformed by CYP2E1 and/or their ability to increase serum ketone bodies. Like CYP3A1, CYP2E1 is induced both by transcriptional activation of the gene and stabilization of the protein against degradation (Koop and Tierney, 1990). Induction of CYP2E1 can also involve mRNA stabilization and/or increased efficiency of mRNA translation. The mechanism of CYP2E1 induction varies even among closely related inducers. For example, although diabetes and fasting both increase the levels of CYP2E1 mRNA by 10-fold, the increase with diabetes results from mRNA stabilization, whereas the increase with fasting results from increased gene transcription. Acetone and other substrates stabilize CYP2E1 by blocking its cAMP-dependent-phosphorylation on Ser129, which otherwise causes the denaturation and degradation of this enzyme. The enzyme-inducing effects of ethanol, acetone, pyrazole, pyridine, and isoniazid are often masked in vivo by the binding of these substrates to CYP2E1.

In mature rats, the levels of certain P450 enzymes are sexually differentiated; that is, they are higher in either male or female rats. Male-specific enzymes include CYP2A2, CYP2C11, CYP2C13, CYP3A2, and CYP4A2. The only known female-spe- cific P450 enzyme is CYP2C12, although the levels of several other P450 enzymes are greater in female than male rats, including CYP2A1, CYP2C7, and CYP2E1. These gender-related differences in P450 enzyme expression are due in large part to sex differences in the pattern of secretion of growth hormone, which is pulsatile in male rats and more or less continuous in females (Waxman et al., 1991). Treatment of mature male rats with various xenobiotics perturbs the pattern of growth hormone secretion and causes a partial “feminization” of P450 enzyme expression, which includes decreased expression of CYP2C11. Sex differences in the expression of P450 enzymes occur to a limited extent in mice, but no marked sex differences in P450 expression have been observed in dogs, monkeys, or humans.

Enzymatic assays have been developed to monitor the induction of the aforementioned P450 enzymes. A series of 7-alkoxyre- sorufin analogs has proven very useful for monitoring the induction of rat and mouse CYP1A and CYP2B enzymes. CYP1A enzymes preferentially catalyze the O-dealkylation of 7-methoxyresorufin and 7-ethoxyresorufin, whereas CYP2B enzymes preferentially catalyze the O-dealkylation of 7-pentoxyre- sorufin and 7-benzyloxyresorufin. The effects of treating rats with phenobarbital on the levels of liver CYP2A1, CYP2B1/2, CYP2C11, and CYP3A1/2 can be monitored by changes in specific pathways of testosterone oxidation. For all practical purposes, the rates of testosterone 2 -, 7 -, and 16 -hydroxylation accurately reflect the levels of CYP2C11, CYP2A1, and CYP2B1/2, respectively. The 2 -, 6 -, and 15 -hydroxylation of testosterone collectively reflect the levels of CYP3A1 and/or CYP3A2. Induction of CYP2E1 can be monitored by increases in 4-nitrophenol hydroxylase, aniline 4-hydroxylase, and chlorzoxazone 6-hydrox- ylase activity, although none of these reactions is specifically cat-

alyzed by CYP2E1. The 12-hydroxylation of lauric acid appears to be catalyzed specifically by CYP4A enzymes. Indeed the -hydroxylation of fatty acids and their derivatives (such as eicosanoids) appears to be a physiological function of these enzymes. CYP4A enzymes also catalyze the 11-hydroxylation of lauric acid, but this reaction is also catalyzed by other P450 enzymes, including the CYP2B and CYP2E enzymes. Induction of CYP4A enzymes by clofibric acid increases the ratio 12to 11-hydroxy- lauric acid, whereas induction of CYP2B enzymes by phenobar-

bital or CYP2E1 by isoniazid has the opposite effect.

In addition to measuring certain enzyme activities, changes in the levels of specific P450 enzymes can also be monitored by immunochemical techniques, such as Western immunoblotting. When P450 induction involves increased gene transcription and/or mRNA stabilization, the increase in mRNA levels can be measured by Northern blotting. These techniques are particularly useful for detecting P450 induction by chemicals that bind tightly to the active site of cytochrome P450 enzymes and thus mask their detection by enzymatic assays. Such chemicals include macrolide antibiotics (e.g., erythromycin and troleandomycin), methylenedioxycontaining compounds (e.g., safrole and isosafrole), azole antimycotics (e.g., clotrimazole, ketoconazole, and miconazole) and musk xylene (Lehman-McKeeman et al., 1999).

Numerous phenobarbital-type inducers and peroxisome proliferators are epigenetic tumorigens (Grasso et al., 1991). Rodents treated chronically with these chemicals develop liver and/or thyroid tumors. The liver tumors seem to be a consequence of hepatocellular hyperplasia/hypertrophy and the sustained proliferation of either the endoplasmic reticulum (in the case of CYP2B inducers) or peroxisomes (in the case of CYP4A inducers). The thyroid tumors are the result of UDP-glucuronosyltransferase induction, which accelerates the glucuronidation of thyroid hormones, leading to a compensatory increase in thyroid-stimulating hormone (TSH). Sustained stimulation of the thyroid gland by TSH leads to the development of thyroid follicular tumors. These epigenetic mechanisms of chemical-induced tumor formation do not appear to operate in humans. Prolonged treatment ( 35 years) with anticonvulsants, such as phenobarbital or phenytoin, does not increase the incidence of liver or thyroid tumor formation in humans. Prolonged elevation of TSH in humans does not lead to tumor formation but causes goiter, a reversible enlargement of the thyroid gland associated with iodide deficiency and treatment with drugs that block thyroid hormone synthesis. Chemicals that cause peroxisome proliferation in rodents do not do so in humans and other primates, possibly because of low levels of PPAR in primate liver and/or the presence of other transcription factors (such as LXR and the thyroid hormone receptor) that bind to or near the PPRE consensus site. Gene knockout mice lacking PPAR are refractory to peroxisome proliferation and peroxisome proliferator–induced changes in gene expression (including induction of CYP4A and - oxidation enzymes). Furthermore, PPAR -null mice are resistant to hepatocarcinogenesis when fed a diet containing WY-14,643, a potent peroxisome proliferator and nongenotoxic carcinogen (Gonzalez et al., 1998).

PHASE II ENZYME REACTIONS

Phase II biotransformation reactions include glucuronidation, sulfonation (more commonly called sulfation), acetylation, methylation, conjugation with glutathione (mercapturic acid synthesis), and

conjugation with amino acids (such as glycine, taurine, and glutamic acid) (Paulson et al., 1986). The cofactors for these reactions, which are shown in Fig. 6-45, react with functional groups that are either present on the xenobiotic or are introduced/exposed during phase I biotransformation. With the exception of methylation and acetylation, phase II biotransformation reactions result in a large increase in xenobiotic hydrophilicity, so they greatly promote the excretion of foreign chemicals. Glucuronidation, sulfation, acetylation, and methylation involve reactions with activated or “high-energy” cofactors, whereas conjugation with amino acids or glutathione involves reactions with activated xenobiotics. Most phase II biotransforming enzymes are mainly located in the cytosol; a notable exception is the UDP-glucuronosyltransferases, which are microsomal enzymes (Table 6-1). Phase II reactions generally proceed much faster than phase I reactions, such as those catalyzed by cytochrome P450. Therefore, the rate of elimination of xenobiotics whose excretion depends on biotransformation by cytochrome P450 followed by phase II conjugation is generally determined by the first reaction.

Glucuronidation

Glucuronidation is a major pathway of xenobiotic biotransformation in mammalian species except for members of the cat family (lions, lynxes, civets, and domestic cats) (Miners and Mackenzie, 1992; Mackenzie et al., 1992; Burchell and Coughtrie, 1992; Burchell, 1999; Tukey and Strassburg, 2000). Glucuronidation requires the cofactor uridine diphosphate-glucuronic acid (UDPglucuronic acid), and the reaction is catalyzed by UDPglucuronosyltransferases (UGTs), which are located in the endoplasmic reticulum of liver and other tissues, such as the kidney, intestine, skin, brain, spleen, and nasal mucosa (Fig. 6-46). Examples of xenobiotics that are glucuronidated are shown in Fig. 6-47. The site of glucuronidation is generally an electron-rich nucleophilic heteroatom (O, N, or S). Therefore, substrates for glucuronidation contain such functional groups as aliphatic alcohols and phenols (which form O-glucuronide ethers), carboxylic acids (which form O-glucuronide esters), primary and secondary aromatic and aliphatic amines (which form N-glucuronides), and free

Figure 6-46. Synthesis of UDP-glucuronic acid and inversion of configuration (A B) during glucuronidation of a phenolic xenobiotic (designated RO ).

Note that these microsomal enzymes face the lumen of the endoplasmic reticulum.

sulfhydryl groups (which form S-glucuronides). In humans and monkeys, more than thirty tertiary amines, including tripelennamine, cyclobenzaprine and imipramine, are substrates for N-glu- curonidation, which leads to formation of positively charged quaternary glucuronides (Hawes, 1998). Certain xenobiotics—such as phenylbutazone, sulfinpyrazone and feprazone—contain carbon atoms that are sufficiently nucleophilic to form C-glucuronides. Coumarin and certain other carbonyl-containing compounds are glucuronidated to form arylenol-glucuronides. In addition to numerous xenobiotics, substrates for glucuronidation include several endogenous compounds, such as bilirubin, steroid hormones, and thyroid hormones. A listing of over 350 UDP-glucuronosyltrans- ferase substrates is available at www.AnnualReview.org (Tukey and Strassburg, 2000).

Glucuronide conjugates of xenobiotics and endogenous compounds are polar, water-soluble conjugates that are eliminated from the body in urine or bile. Whether glucuronides are excreted from the body in bile or urine depends on the size of the aglycone (parent compound or phase I metabolite). In rat, glucuronides are preferentially excreted in urine if the molecular weight of the aglycone is less than 250, whereas glucuronides of larger molecules (aglycones with molecular weight 350) are preferentially excreted in bile. Molecular weight cutoffs for the preferred route of excretion vary among mammalian species. The carboxylic acid moiety of glucuronic acid, which is ionized at physiologic pH, promotes excretion because (1) it increases the aqueous solubility of the xenobiotic and (2) it is recognized by the biliary and renal organic anion transport systems, which enables glucuronides to be secreted into urine and bile. The cofactor for glucuronidation is synthesized from glucose-1-phosphate, and the linkage between glucuronic acid and UDP has an -configuration, as shown in Fig. 6-46. This configuration protects the cofactor from hydrolysis by

-glucuronidase. However, glucuronides of xenobiotics have a-configuration. This inversion of configuration occurs because glucuronides are formed by nucleophilic attack by an electron-rich atom (usually O, N, or S) on UDP-glucuronic acid, and this attack occurs on the opposite side of the linkage between glucuronic acid and UDP, as shown in Fig. 6-46. In contrast to the UDP-glucuronic acid cofactor, xenobiotics conjugated with glucuronic acid are substrates for -glucuronidase. Although present in the lysosomes of some mammalian tissues, considerable -glucuronidase activity is present in the intestinal microflora. The intestinal enzyme can release the aglycone, which can be reabsorbed and enter a cycle called enterohepatic circulation, which delays the elimination of xenobiotics. Nitrogen-glucuronides are more slowly hydrolyzed by -glucuronidase than O- or S-glucuronides, whereas O-glucuronides tend to be more stable to acid-catalyzed hydrolysis than N- or S-glucuronides. The potential for glucuronides to be hydrolyzed in the presence of acid or base complicates the analysis of conjugates in urine or feces.

The C-terminus of all UDP-glucuronosyltransferases contains a membrane-spanning domain that anchors the enzyme in the endoplasmic reticulum. The enzyme faces the lumen of the endoplasmic reticulum, where it is ideally placed to conjugate lipophilic xenobiotics and their metabolites generated by cytochrome P450 and other microsomal phase I enzymes. The lumenal orientation of UDP-glucuronosyltransferases poses a problem because UDPglucuronic acid is a water-soluble cofactor synthesized in the cytoplasm. A transporter has been postulated to shuttle this cofactor into the lumen of the endoplasmic reticulum, and it may also shuttle UDP (the byproduct of glucuronidation) back into the cytoplasm for synthesis of UDP-glucuronic acid, as shown in Fig. 6-46. In vitro, the glucuronidation of xenobiotics by liver microsomes can be stimulated by detergents, which disrupt the lipid bilayer of the endoplasmic reticulum and allow UDP-glucuronosyltransferases free access to UDP-glucuronic acid. High concentrations of detergent can inhibit UDP-glucuronosyltransferases, presumably by disrupting their interaction with phospholipids, which are important for catalytic activity.

Radominska-Pandya et al. (1999) have proposed that UDPglucuronosyltransferases form homoand heterodimers in the endoplasmic reticulum, which are stabilized by substrate binding (which gains access to the active site by diffusion through the lipid bilayer). According to this model, UDP-glucuronic acid gains access to the active site via a proteinaceous channel formed between the two monomers. After conjugation, the product glucuronide and UDP are expelled into the cytosol by the same channel used for entry of the cofactor, after which the dimer dissociates to allow pairing of the monomers with other UDP-glucuronosyltransferases. This model obviates the need for a transporter to shuttle UDPglucuronic acid from the cytosol to the lumen of endoplasmic reticulum. It is not known whether such dimerization, if it occurs, alters the substrate specificity of the individual UDP-glucurono- syltransferases, which would have implications for studies designed to determine the substrate specificity of recombinant enzymes, which are invariably expressed individually.

Cofactor availability can limit the rate of glucuronidation of drugs that are administered in high doses and are conjugated extensively, such as aspirin and acetaminophen. In experimental animals, the glucuronidation of xenobiotics can be impaired in vivo by factors that reduce or deplete UDP-glucuronic acid levels, such as diethyl ether, borneol, and galactosamine. The lowering of UDPglucuronic acid levels by fasting, such as might occur during a se-

The arrow indicates the site of glucuronidation.

vere toothache, is thought to predispose individuals to the hepatotoxic effects of acetaminophen, although even then hepatotoxicity only occurs with higher-than-recommended doses of this analgesic (Whitcomb and Block, 1994).

The existence of multiple forms of UDP-glucuronosyltrans- ferase was first suggested by the observation that in rats developmental changes in glucuronidation rates were substrate-dependent, and the glucuronidation of xenobiotics could be differentially affected by treatment of rats with chemicals known to induce cytochrome P450. Based on their ontogeny and inducibility, the UDP-glucuronosyltransferase activities in rat liver microsomes were categorized into four groups. The activity of enzyme(s) in the

first group peaks 1 to 5 days before birth, and it is inducible by 3-methylcholanthrene and other CYP1A enzyme inducers. Substrates for the group 1 enzyme(s) tend to be planar chemicals, such as 1-naphthol, 4-nitrophenol, and 4-methylumbelliferone. The activity of enzyme(s) in the second group peaks 5 days after birth and is inducible by phenobarbital and other CYP2B enzyme inducers. Substrates for the group 2 enzyme(s) tend to be bulky chemicals, such as chloramphenicol, morphine, 4-hydroxybiphenyl, and monoterpenoid alcohols. The activity of enzyme(s) in the third group peaks around the time of puberty ( 1 month) and is inducible by PCN and other CYP3A enzyme inducers. Substrates for the group 3 enzyme(s) include digitoxigenin monodigitoxoside

(dt1), a metabolite of digitoxin formed by CYP3A (see Fig. 6-43), and possibly bilirubin. The activity of enzyme(s) in the fourth group also peak around the time of puberty ( 1 month) and is inducible by clofibrate and other CYP4A enzyme inducers. Substrates for the group 4 enzyme(s) include bilirubin but not dt1, which distinguishes group 3 from group 4 UDP-glucuronosyltransferases.

Although this classification system still has some practical value, it has become evident that the four groups of UDPglucuronosyltransferases do not simply represent four independently regulated enzymes with different substrate specificities. This realization stems from various studies, including those conducted with Gunn rats, which are hyperbilirubinemic due to a genetic defect in bilirubin conjugation. The glucuronidation defect in Gunn rats is substrate-dependent in a manner that does not match the categorization of UDP-glucuronosyltransferases into the four aforementioned groups. For example, in Gunn rats the glucuronidation of the group 2 substrates, morphine and chloramphenicol, is not impaired, whereas the glucuronidation of 1-naphthol, dt1 and bilirubin (group 1, 3, and 4 substrates) is low or undetectable. The induction of UDP-glucuronosyltransferase activity by 3-methyl- cholanthrene, PCN, and clofibric acid is impaired in Gunn rats, whereas the induction by phenobarbital is normal. (Although phenobarbital does not induce the conjugation of bilirubin in Gunn rats, it does so in normal Wistar rats.) Only when the UDP-glu- curonosyltransferases were cloned did it become apparent why the genetic defect in Gunn rats affects three of the four groups of UDPglucuronosyltransferases that are otherwise independently regulated as a function of age and xenobiotic treatment (Owens and Ritter, 1992). It is now apparent that the UDP-glucuronosyltrans- ferases expressed in rat liver microsomes belong to two gene families, UGT1 and UGT2. The former gene family contains at least seven enzymes, all of which belong to the same subfamily designated UGT1A. The individual members of the rat UGT1A subfamily are UGT1A1, 1A2, 1A3, 1A5, 1A6, 1A7, and 1A8. UGT1A4 is not a member of the rat UGT1A subfamily, although additional members may yet be identified. The second UGT gene family in rats is divided into two subfamilies, UGT2A and UGT2B; the former contains a single member (UGT2A1) whereas the sec-

ond contains at least six members (UGT2B1, 2B2, 2B3, 2B6, 2B8, and 2B12). Members of gene family 2 are all distinct gene products (i.e., UGT2A1 and the six UGT2B enzymes are encoded by seven separate genes). In contrast, members of family 1 are formed from a single gene with multiple copies of the first exon, each of which can be connected in cassette fashion with a common set of exons (exons 2 to 5). This arrangement is illustrated in Fig. 6-48 for the human UGT1 gene locus. In rats, the fourth copy of exon 1 is a pseudogene, hence, there is no UGT1A4 in rats. In humans, the second copy of exon 1 is a pseudogene, hence, there is no UGT1A2 in humans. (The human UGT1A gene locus is discussed later in this section.) The rat UGT1A gene locus is known to contain eight versions of the first exon, which produce seven functional UGT1A enzymes. Additional members of the rat UGT1A subfamily are thought to exist, including one that glucuronidates dt1, a metabolite of digitoxin (see Fig. 6-43).

A simplified view of the UGT1A gene locus is that the multiple UGT1A enzymes are constructed by linking different substrate binding sites (encoded by multiple copies of exon 1) to a constant portion of the enzyme (encoded by exons 2 to 5). This constant region is involved in cofactor binding and membrane insertion. This method of generating multiple forms of an enzyme from a single gene locus is economical, but it is also the genetic equivalent of putting all of one’s eggs in the same basket. Whereas a mutation in any one of the UGT2 enzymes affects a single enzyme, a mutation in the constant region of the UGT1 gene affects all enzymes encoded by this locus. In the Gunn rat, a mutation at codon 415 introduces a premature stop signal, so that all forms of UDP-glucuronosyltransferase encoded by the UGT1 locus are truncated and functionally inactive. The UDP-glucuronosyltransferases known to be encoded by the rat UGT1 locus include the 3-methyl- cholanthrene-inducible enzyme that conjugates planar molecules like 1-naphthol (UGT1A6 and UGT1A7), the phenobarbitaland clofibric acid-inducible enzyme that conjugates bilirubin (UGT1A1 and, to a lesser extent, UGT1A4), and the PCN-inducible enzyme that conjugates dt1 (which will be named when the first exon for this enzyme is cloned and localized within the UGT1A gene locus). All of these UGT1 enzymes are defective in Gunn rats.

The second family of rat UDP-glucuronosyltransferases, which share less than 50 percent of amino acid sequence identity with the first family, are divided into two subfamilies (UGT2A and UGT2B), and its members are distinct gene products. The single member of the 2A subfamily, UGT2A1, is expressed specifically in olfactory epithelium where it conjugates a wide variety of substrates. The six known members of the rat UGT2B subfamily (UGT2B1, 2, 3, 6, 8, and 12) are expressed in liver and various extrahepatic tissues. Members of the UGT2B subfamily are named in the order they are cloned, regardless of the species of origin (much like the nomenclature system for most of the P450 enzymes). UGT2B enzymes have been cloned from rat (forms 1, 2, 3, 6, 8, and 12), humans (forms 4, 7, 10, 11, 15, and 17), mouse (form 5), and rabbits (forms 13, 14, and 16). A rabbit UDP-glucuronosyl- transferase has been classified as UGT2C1; however, UGT2C genes in other mammalians have not been identified. In rats, at least one UGT2B enzyme (UGT2B1) is inducible by phenobarbital. The gene encoding this enzyme is not defective in Gunn rats; therefore, treatment of Gunn rats with phenobarbital induces the glucuronidation of substrates for UGT2B1. However, UGT2B1 does not conjugate bilirubin (a reaction mainly catalyzed by UGT1A1 and/or UGT1A4), which is why phenobarbital cannot induce the conjugation of bilirubin in Gunn rats. UGT2B1 is the main enzyme responsible for catalyzing the 3-O-glucuronidation of morphine, which is markedly increased by treatment of rats with phenobarbital. Whereas as Gunn rats are genetically defective in all UGT1A enzymes, LA rats are selectively defective in UGT2B2, which allowed this enzyme to be identified as the principal enzyme responsible for glucuronidating androsterone and triiodothyronine (T3) in rats (Burchell, 1999). The multiple forms of human UDPglucuronosyltransferase are also products of either a single UGT1A gene locus (see Fig. 6-48) or multiple UGT2 genes. The human UGT1A locus contains 12 potential copies of the first exon, although only nine transcripts have been identified (UGT1A1, 1A3, 1A4, 1A5, 1A6, 1A7, 1A8, 1A9, and 1A10), all of which are transcribed in vivo into functional enzymes with the possible exception of UGT1A5 (Tukey and Strassburg, 2000). The UGT2 genes expressed in humans include UGT2A1 (which is expressed only in olfactory tissue in an analogous manner to the corresponding rat enzyme), and UGT2B4, 2B7, 2B10, 2B11, 2B15, and 2B17. The tissue distribution and substrate specificity of the human UGT1 and UGT2 enzymes have been reviewed by Tukey and Strassburg (2000). Suffice it to say that these enzymes are expressed in a wide variety of tissues, and some enzymes—including UGT1A7, 1A8, 1A10, 2A1, and 2B17—are expressed only in extrahepatic tissues, which has implications for the common practice of using human liver microsomes to investigate the role of glucuronidation in drug metabolism. Numerous UGT1 and UGT2 enzymes are expressed throughout the gastrointestinal tract, where they contribute significantly to the first-pass elimination of numerous xenobiotics. Several UGT2B enzymes are expressed in steroid-sensitive tissues, such as prostate and mammary gland, where they presumably terminate the effects of steroid hormones.

Probe drugs have been identified for some but not all of the human UDP-glucuronosyltransferases, including UGT1A1 (bilirubin), UGT1A4 (imipramine), UGT1A6 (1-naphthol and possibly acetaminophen), UGT1A8 (propofol), and UGT2B7 (morphine) (Burchell, 1997). The glucuronidation of morphine by UGT2B7 involves conjugation of the phenolic 3-hydroxyl and the alcoholic 6-hydroxyl group in a 7:1 ratio. The 6-O-glucuronide is 600 times

more potent an analgesic than the parent drug, whereas the 3-O-glucuronide is devoid of analgesic activity. UGT2B7 is present in the brain, where it might facilitate the analgesic effect of morphine through formation of the 6-O-glucuronide, which presumably does not readily cross the blood-brain barrier and may be retained in the brain longer than morphine (Tukey and Strassburg, 2000).

In humans, Crigler-Najjar syndrome and Gilbert’s disease are congenital defects in bilirubin conjugation analogous to that seen in Gunn rats. The major bilirubin-conjugating enzyme in humans is UGT1A1. Genetic polymorphisms in exons 2-5, which affect all enzymes encoded by the UGT1A locus, and polymorphisms in exon 1, which specifically affect UGT1A1, have been identified in patients with Crigler-Najjar syndrome and Gilbert’s disease. More than thirty genetic polymorphisms are associated with these diseases (Tukey and Strassburg, 2000). Some polymorphisms are associated with type I Crigler-Najjar syndrome, a severe form of the disease characterized by a complete loss of bilirubin-conjugating activity and marked hyperbilirubinemia, whereas others are associated with the less severe type II Crigler-Najjar syndrome or Gilbert’s disease. The milder forms of hyperbilirubinemia respond to phenobarbital, which stimulates bilirubin conjugation presumably by inducing UGT1A1. Type I Crigler-Najjar syndrome is associated with impaired glucuronidation of propofol, 17 - ethinylestradiol and various phenolic substrates for UGT1A enzymes.

There is some evidence for genetic polymorphisms in some of the human UGT2 enzymes. For example, oxazepam is glucuronidated by UGT2B7, which preferentially glucuronidates S-oxazepam over its R-enantiomer. Ten percent of the population appear to be poor glucuronidators of S-oxazepam, which may or may not reflect genetic polymorphisms of the UGT2B7 gene. Such polymorphisms appear to be the underlying cause of alterations in hyodeoxycholate glucuronidation in gastric mucosa (Tukey and Strassburg, 2000). Human UGT1A6 glucuronidates acetaminophen, and the glucuronidation of acetaminophen in humans is enhanced by cigarette smoking and dietary cabbage and brussels sprouts, which suggests that human UGT1A6 is inducible by polycyclic aromatic hydrocarbons and derivatives of indole 3-carbinol (Bock et al., 1994). However, direct evidence for induction of human UGT1A6 is lacking, even though rat UGT1A6 is highly inducible by polycyclic aromatic hydrocarbons and other CYP1A inducers. (It should be noted that the UGT1A enzymes in rat and human are named according to the location of the first exon relative to the shared exons within the UGT1A gene locus. Consequently, the identically named UGT1A enzymes in rats and humans need not necessarily glucuronidate the same substrates nor be under similar regulatory control.) Nevertheless, ligands for the Ah receptor, such as those present in cigarette smoke, induce CYP1A2, which would be expected to enhance the hepatotoxicity of acetaminophen. Increased acetaminophen glucuronidation may explain why cigarette smoking does not enhance the hepatotoxicity of acetaminophen. Conversely, decreased glucuronidation may explain why some individuals with Gilbert’s syndrome are predisposed to the hepatotoxic effects of acetaminophen (De Morais et al., 1992). Low rates of glucuronidation also predispose humans to the adverse gastrointestinal effects of irinotecan, a derivative of camptothecin (Gupta et al., 1994). Low rates of glucuronidation predispose newborns to jaundice and to the toxic effects of chloramphenicol; the latter was once used prophylactically to prevent

opportunistic infections in newborns until it was found to cause severe cyanosis and even death (gray baby syndrome).

Glucuronidation generally detoxifies xenobiotics and potentially toxic endobiotics, such as bilirubin, for which reason glucuronidation is generally considered a beneficial process. However, steroid hormones glucuronidated on the D-ring (but not the A-ring) cause cholestasis, and induction of UDP-glucuronosyltransferase activity has been implicated as an epigenetic mechanism of thyroid tumor formation in rodents (Curran and DeGroot, 1991; McClain, 1989). Inducers of UDP-glucuronosyltransferases cause a decrease in serum thyroid hormone levels, which triggers a compensatory increase in thyroid-stimulating hormone (TSH). During sustained exposure to the enzyme-inducing agent, prolonged stimulation of the thyroid gland by TSH ( 6 months) results in the development of thyroid follicular cell neoplasia. Glucuronidation followed by biliary excretion is a major pathway of thyroxine biotransformation in rodents whereas deiodination is the major pathway (up to 85 percent) of thyroxine metabolism in humans. In contrast to the situation in rodents, prolonged stimulation of the thyroid gland by TSH in humans will result in malignant tumors only in exceptional circumstances and possibly only in conjunction with some thyroid abnormality. Therefore, chemicals that cause thyroid tumors in rats or mice by inducing UDP-glucurono- syltransferase activity are unlikely to cause such tumors in humans. In support of this conclusion, extensive epidemiologic data in epileptic patients suggest that phenobarbital and other anticon-

vulsants do not function as thyroid (or liver) tumor promoters in humans.

In some cases, glucuronidation represents an important event in the toxicity of xenobiotics. For example, the aromatic amines that cause bladder cancer, such as 2-aminonaphthalene and 4-amino- biphenyl, undergo N-hydroxylation in the liver followed by N-glucuronidation of the resultant N-hydroxyaromatic amine. The N-glucuronides, which accumulate in the urine of the bladder, are unstable in acidic pH and thus are hydrolyzed to the corresponding unstable, tumorigenic N-hydroxyaromatic amine, as shown in Fig. 6-49. A similar mechanism may be involved in colon tumor formation by aromatic amines, although in this case hydrolysis of the N-glucuronide is probably catalyzed by -glucuronidase in intestinal microflora. Some acylglucuronides are reactive intermediates that bind covalently to protein by mechanisms that may or may not result in cleavage of the glucuronic acid moiety, as shown in Fig. 6-49. Several drugs—including the NSAIDs diclofenac, diflunisal, etodolac, ketoprofen, suprofen, and tolmetin—contain a carboxylic acid moiety that is glucuronidated to form a reactive acylglucuronide. Neoantigens formed by binding of acylglucuronides to protein might be the cause of rare cases of NSAIDinduced immune hepatitis. Binding of acylglucuronides to protein can involve isomerization reactions that lead to retention of a rearranged glucuronide moiety (Fig. 6-49). Formation of a common neoantigen (i.e., one that contains a rearranged glucuronic acid moiety) might explain the allergic cross-reactivities (cross-sensiti-