Genetic and environmental factors associated with variation of human xenobiotic glucuronidation and sulfation.

Glucuronidation and sulfation are phase 2 metabolic reactions catalyzed by large families of different isoenzymes in man. The textbook view that glucuronidation and sulfation lead to the production of harmless conjugates for simple excretion is not valid. Biologically active and toxic sulfates and glucuronides are produced and leed to adverse drug reactions, including immune hypersensitivity. Considerable variation in xenobiotic conjugation is observed as a result of altered expression of UDP-glucuronosyltransferases (UGTs) and sulfotransferases (STs). Recent cloning and expression of human cDNA encoding UGTs and STs has facilitated characterization of isoform substrate specificity, which has been further validated using specific antibodies and human tissue fractions. The availability of cloned/expressed human enzymes and specific antibodies has enabled the investigation of xenobiotic induction and metabolic disruption leeding to adverse responses. Genetic polymorphisms of glucuronidation and sulfation are known to exist although the characterization and assessment of the importance of these variations are hampered by appropriate ethical studies in men with suitable safe model compounds. Genetic analysis has allowed molecular identification of defects in well-known hyperbilirubinemias. However, full characterization of the specific functional roles of human UGTs and STs requires rigorous kinetic and molecular analyses of the role of each enzyme in vivo through the use of specific antibodies and inhibitors. This will leed to the better prediction of variation of xenobiotic glucuronidation and sulfation in man.


Introduction
Glucuronidation and sulfation of pharmacologically active xenobiotic compounds and endogenous substances is a major phase 2 detoxication system in man and has profound effects on the disposition, metabolism, and excretion of many drugs (1,2).
Glucuronidation is a major detoxication pathway in all vertebrates. Glucuronide formation is catalyzed by a family of UDPglucuronosyltransferases (UGTs) using thousands of endobiotic and xenobiotic compounds as substrates (3). The xenobiotic substrate range of an individual UGT isoform may be dictated by the evolved structure of an individual UGT to accept endobiotic substrates such as bilirubin or a steroid (4). A thorough understanding of the evolved endobiotic UGT substrate range is essential. Although an extensive list of drugs is glucuronidated by humans (3), it remains difficult to determine and predict UGT specificity. However, the fallibility of the evolved detoxication systems has been revealed by the implication of drug glucuronides in adverse drug reactions that resulted in hypersensitivity of immune response (5).
The cytosolic sulfotransferases (STs), derived from the ST gene superfamily, catalyze the sulfation of a host ofxenobiotics This paper was prepared as background for the Workshop on Susceptibility to Environmental Hazards convened by the Scientific Group on Methodologies for the Safety Evaluation of Chemicals (SGOMSEC) held [17][18][19][20][21][22] March 1996 in Espoo, Finland. Manuscript received at EHP 5 November 1996; accepted 18 November 1996. 644620. E-mail: bburchell@ninewells.dundee.ac.uk Abbreviations used: CN, Crigler-Najjar syndrome; GS, Gilbert's syndrome; PAPS, 3'-phosphoadenosine 5'-phosphosulfate; ST, sulfotransferase; UGT, UDP-glucuronosyltransferase; YAC, yeast-affiliated chromosome. and endogenous compounds, including steroids, bile acids, thyroid hormones, and neurotransmitters (2). Sulfotransferases of various kinds may be found in species as diverse as bacteria and man. Although in mammals sulfation, in common with most phase 2 reactions, is generally considered to serve a detoxication function, it also plays a role in, for example, steroid biosynthesis (6). For certain compounds (e.g. aromatic hydroxylamines and hydroxamic acids, benzylic alcohols of polycyclic aromatic hydrocarbons) sulfation is the terminal step in their bioactivation to mutagens (7). Many xenobiotics are substrates for both UGTs and STs, and the different subcellular location and kinetic properties of the enzymes and the availability of cosubstrate influence the relative contribution of each system. The availability of cloned human UGTs and STs has allowed for significant progress (3), although our excitement needs to be tempered by our lack of knowledge of the contribution of each isoform to xenobiotic conjugation in vivo. Moreover, the use of "rate" data obtained using recombinant cell lines may lead to prediction of in vivo pharmacokinetics, although there are a number of limitations that have to be circumvented (8).
This minireview will assess recent work on the interindividual variation of glucuronidation and sulfation in man caused by genetic differences in expression of UGTs or STs. Further, we shall briefly examine the bioactivation of xenobiotics following conjugation with sulfate and glucuronic acid as a mechanism of potential toxicity in man.

The UGT Family of Genes and Enzymes
More than 14 human liver UGT cDNAs have been cloned and classified into two subfamilies based on sequence analysis. The UGT1 subfamily of enzymes glucuronidate xenobiotic phenols and bilirubin, while UGT2 enzymes glucuronidate steroids and bile acids. An early observation was that UGT1 cDNA clones shared an identical C-terminal coding sequence whereas the N-terminal 246 amino acids show a striking lack of identity (24 vs 49%) (9).
Southern blot analysis indicated that the region encoding the conserved 3' half of four separate human UGT1 cDNAs (the common domain) was a single copy in the human genome, suggesting a role for alternative splicing in the synthesis of different isoforms. In support of this, the common domain and the isoform-specific 5' half of the four UGT1 cDNAs colocalized to chromosome 2 at 2q.37 (10). Owens et al. (11) described the existence of a gene complex by the isolation of overlapping cosmid clones containing six alternative substratedetermining first exons upstream of the four exons that make up the common domain. The human liver cDNA clone HP4 isolated by Wooster et al. (12) had the same common domain sequence as other UGT1 cDNAs but contained a novel substrate-determining exon that was termed UGT1*02 and was not among those already genomically cloned. This suggested that the UGTI gene locus was larger than had previously been described, presumably extending further upstream. Human genomic Southern blotting indicated the presence of multiple sequences homologous to the 5' portion of UGTI*02 (13). Further work has shown that the human UGTI gene is a single-copy gene that consists of 4 common exons and more than 13 variable exons that span more than 200 kb of the human genome (C Brierley and B Burchell, unpublished data).
In contrast, comparison of members of the UGT2 gene family, the steroid metabolizing isoforms, indicates that amino acid differences between different isoforms of this family occur throughout the length of the protein. Three UGT2B genes were mapped to chromosome 4 using somatic cell hybrid cell lines and polymerase chain reaction. To determine whether they were clustered we screened a yeast-affiliated chromosome (YAC) library and isolated a five-YAC contig. One of these YACs was found to contain at least three UGT2B genes in 195 kb. Fluorescence in situ hybridization analysis indicated that the UGT2B gene cluster was localized at 4ql3 (14).

Use of Cloned and Expressed
Human Liver UGTs to Determine Substrate Specificity Human UGT isoforms have been identified by gene sequencing and cDNA cloning (9), but only a few of these are known to be expressed in vivo. Indeed, one sequence is known to contain stop codons that would prevent functional enzyme expression (9). Consequently, only a few human UGTs have been substantially characterized.
Six cloned human hepatic UGT cDNAs were stably expressed in tissue culture cell lines. More than 100 drug xenobiotics and endobiotics were used as substrates for glucuronidation catalyzed by the cloned human transferases to determine the chemical structure accepted as substrates (9).
A cloned human bilirubin UGT (UGT 1*1) accepted a wide diversity of compounds such as phenols, anthraquinones, flavones, and steroids. Seven substrates were glucuronidated at rates comparable to or higher than bilirubin. Octyl gallate was glucuronidated at the highest rate among all substrates tested. The UGT1*1 isoform exhibited stereospecificity towards simple nonplanar phenols and estrogens. The increasing number of carbon atoms in the alkyl side chain of gallate analogues improved the specific activity from no glucuronide formed with gallic acid to 0.72 and 1.90 mmol/min/mg protein in propyl and octyl gallate, respectively (15). Seventeen steroids were tested and only four that had both 17pand 3,-hydroxy substituents were glucuronidated (15).
Recently, UGT1*4 has been shown to catalyze the N-glucuronidation of drugs and xenobiotics (9). Subfamily two contains at least five UGTs catalyzing steroid or bile acid glucuronidation UGT-2B4 and UGT-2B7 (steroid/bile acid UGTs) also catalyzed the glucuronidation of some xenobiotics (9). Levels of UGT activity were sufficient to allow determination of kinetic parameters for the enzyme reaction. Further, metabolism and toxicity of drugs could be studied by their addition to recombinant cell lines in culture, and extraction of the media allowed analysis of glucuronide formation. Suggested probe substrates for individual UGT isoforms are listed in Table 1.
Use of Antibodies to Validate Functional Specificity of Human UGTs in Human Liver Microsomes Some problems in the use of expressed human drug-metabolizing enzymes in the analysis of drug metabolism and drug-drug interactions have been discussed in a recent commentary (8). The value of these in vitro systems is their relevance to human drug metabolism in vivo.
Recently, attempts have been made to assess the contribution of a specific UGT to the glucuronidation of an endobiotic or xenobiotic in a particular tissue by using inhibitory antibodies. Monospecific polyclonal antibodies raised against the Nterminal portion (14-150 residues) of human liver UGT2B4 protein expressed in E. coli were used to immunoinhibit and immunoprecipitate this transferase from human liver and kidney microsomes (16). These experiments demonstrated that UGT2B4 activity was responsible for more than 90% of the hyodeoxycholate 6-0glucuronidation activity in human liver microsomes, but did not contribute significantly toward the glucuronidation of estriol, 4-hydroxyesterone, 1 -naphthol, or hyodeoxycholic acid (16). Similarly, antibodies were raised against the N-terminal half of UGT1*6 expressed in E. coli, and immunoinhibition analysis of human liver microsomes demonstrated that this isoenzyme represented between 20 and 50% of the total microsomal 1-naphthol glucuronidation (17). UGT activities towards hyodeoxycholic acid, 4-hydroxybiphenyl, 4-t-butylphenol, and bilirubin were not inhibited by these specific anti-UGT1*6 antibodies. Biologically Active and Potentially Toxic Glucuronides Glucuronidation has been described as a safe detoxication process and glucuronides were never considered biologically active intermediates. However, in recent years the potential toxicity and biological activity of certain glucuronides have been well recognized.
There are several notable examples reported in which such metabolites have been found to be pharmacologically active. (-)-Morphine is glucuronidated in a stereoselective manner to (-)-morphine-3-glucuronide and (-)-morphine-6-glucuronide in the liver (18). Detailed pharmacologic characterization of the glucuronides has established that (-)-morphine-6-glucuronide is 650 times more potent as an analgesic than the parent drug, whereas morphine-3glucuronide is a potent antagonist of morphine and has no analgesic activity (18). This discovery has led to (-)-morphine-6glucuronide being commercially marketed.
The Role of Acyl Glucuronidation in Drug Immune Hypersensitivity Acyl glucuronides are formed when conjugation with glucuronic acid occurs via a carboxyl group, resulting in an estertype linkage. Ester-type glucuronides are much more unstable than ether-linked glucuronides and can easily undergo nucleophilic substitution. The chemical properties of acyl-linked glucuronides are extensively reviewed (19).
One important reaction that acyl glucuronides undergo is acyl migration, a process whereby the aglycone moves from the 1-hydroxyl group of the glucuronic acid sugar to the 2-, 3-,or 4-hydroxyl groups. This rearrangement of the glucuronide leads to ,B-glucuronidase-resistant isomers and is completely reversible with one exception: the Ci-glucuronide does not appear to reform from the C2-isomer (19). The extent of acyl migration may only become detectable when the excretion of conjugates is impaired and their plasma concentrations are raised (19).
The rate of acyl migration differs from compound to compound and their stability is also highly variable (19). At physiological or slightly alkaline pH, acyl migration and hydrolysis of acyl glucuronides is extensive (19).
A number of acyl glucuronides have been shown to bind irreversibly to proteins in vitro and in vivo (20). Evidence exists for two principal mechanisms of this reversible (covalent) binding; however, it is not known which of these is principally responsible (21).

Known Polymorphisms of Glucuronidation in Man
Polymorphic drug glucuronidation in man has not been extensively studied due to problems in the selection of suitable drugs for screening. The identification and measurement of metabolites, including the interaction between pathways of glucuronidation and sulfation, complicate interpretation of the data. The known polymorphisms are listed in Table 2.
Codeine glucuronidation was decreased in Han Chinese when compared to a Swedish population (22). Paracetamol glucuronidation in small Chinese and Caucasian groups has been compared but no interethnic differences were observed (23). In a more recent study of paracetamol glucuronidation, the distribution in a random population appeared to be unimodel although skewed (24). Glucuronidation was also clearly affected by gender, oral contraceptives, steroids, and smoking, further indicating the problems in determination of genetic polymorphism (24).
Liu et al. (25) reported that the glucuronide excretion of the hypolipidemic drug clofibrate in a healthy Caucasian population followed a normal distribution, whereas that of fenofibrate appeared to be distributed into the distinct normal groups. However, a follow-up familial study has shown lack of a genetic polymorphism in the glucuronidation of fenofibrate (26).
Another study, of a French population, examined variation of dextrorphan glucuronidation. Again, a normal distribution was observed (27). Analysis of the polymorphic variation of diflunisal phenol and acyl glucuronide showed a unimodal population distribution, especially when females using oral contraceptives were excluded (28).
Recently, oxazepam administered as a racemic mixture was shown to be preferentially excreted as the (S)-glucuronide and a low SIR glucuronide ratio was used to assess poor glucuronidation of oxazepam (29). A group of 10% of the whole population was determined to be poor glucuronidators of (S)-oxazepam which suggested a genetic relationship to the UGT2B7 isoform. However, oxazepam may not be solely glucuronidated by UGT2B7 in vivo and this relationship requires additional investigation. Nonetheless, this is the most interesting example of a polymorphism of drug glucuronidation to date.
Menthol glucuronidation is considerably reduced in many patients with Crigler-Najjar syndrome, such that the menthol excretion test is often used as a confirmatory diagnosis of the disorder and as a study of the genetic inheritance of the disease (30,31). In a study of a family in which two CN children were born from the same marriage, only 5 of 16 members of the family showed a normal menthol glucuronide output. This menthol test demonstrated likely heterozygotes in the family members who were not revealed by serum bilirubin analyses (30). In other studies the test failed to discriminate between icteric patients and anicteric family members suspected to be heterozygotes. Indeed, jaundiced patients may show normal menthol glucuronide excretion (32). Bloomer et al. (33) also detected normal menthol conjugation in the parents of a Crigler-Najjar child. These accumulated data suggest that menthol glucuronidation is independently variable within the population.
Menthol is conjugated with glucuronic acid prior to excretion in bile and urine (31) and is therefore an interesting, relatively harmless test compound for study of polymorphism of drug glucuronidation in man. However, glucuronidation of menthol was only reduced to about 20% of controls in some of the Crigler-Najjar type 1 patients. This suggests that menthol may be a substrate for more than one UGT isoenzyme, which may complicate studies of menthol glucuronidation in the normal population.
The known defects of glucuronidation in humans are best illustrated within the group of a hereditary hyperbilirubinemias, (5) in which molecular genetic studies have revealed the association between genetic defects and loss of function. Crigler-Najjar Syndrome Crigler-Najjar syndrome (CN) is a familial form of severe unconjugated hyperbilirubinemia caused by a dysfunction in bilirubin glucuronidation. The molecular basis of these syndromes has recently been characterized by enzymological, immunochemical, and molecular genetic analysis. In vitro analysis of CN type 1 liver samples has revealed that as well as showing complete absence of bilirubin UGT activity, some of these patients also poorly glucuronidate phenols, 5-hydroxytryptamine, and the drugs ethinylestradiol and propofol (3).
Recently, as a result of the elucidation of the UGTI gene complex, several groups have determined some of genetic lesions that cause CN (5). The mutations that result in CN syndrome have been found in the exons 1,2,3,4, and 5 encoding the constant region of all UGT1 proteins, which explains the decreased activity towards other aglycones as well as bilirubin (5). More than 20 genetic lesions have been demonstrated to be associated with the severe hyperbilirubinemia of CN (5).

Gilberts Syndrome
Gilbert's syndrome (GS), a familial hyperbilirubinemia determined to be present in up to 5% of the population in 1980, is characterized by a mild unconjugated hyperbilirubinemia (34). Decreased formation of bilirubin diglucuronide and increased levels of bilirubin monoglucuronide were found in bile in parallel with decreased hepatic UGT activity (35). This disease provides an opportunity to study variation in drug glucuronidation due to the prevalence of the familial disorder within the population. There is no obvious indication of impaired drug oxidation, acetylation, or sulfation (36). Decreased clearance of several drugs, such as tolbutamide, rifamycin, josamycin, and paracetamol, has been observed (37), although decreased clearance was not apparently associated with a decreased rate of glucuronidation measured in overnight urine samples (36).
De Morais et al. (38) have reported that paracetamol glucuronide formation, measured in six GS patients by clearance from plasma within 2 hr, was 31% lower than in normal controls. The timing of measurements may be critical in determination of these significant differences.
Recent work has shown that the mildly affected members of families in which CN type 2 occurs are heterozygous for mutations in the UGT subfamily 1 (UGTI genes) may be classified as GS (39). However, the incidence of CN-2 in the population is very rare and the frequency of alleles causing CN-2 would not be sufficient to explain the population incidence of GS.
A study by Bosma et al. suggested a correlation between homozygosity for a 2bp insertion in the TATA box upstream of UGTI*1 exon 1 and GS; no mutations were found in the coding sequence of the UGTI *1 gene (40). We demonstrated that the primary genetic factor contributing to the variation in the serum and total bilirubin concentration in an eastern Scottish population is the sequence variation reported by Bosma et al. (40); a direct correlation is only revealed by a controlled study of a drug-free, alcohol-free nonsmoking population (41). We showed that the 7/7 genotype is associated with GS and occurs in 10 to 13% ofthe population (41).
Drugs, alcohol, and smoking induce human bilirubin UGT (41) and thereby interfere with the phenotype in the general population, creating latent GS patients.
It is now recognized that there are better xenobiotic substrates of UGT1*1 (the major bilirubin-metabolizing form) than bilirubin itself, examples including octyl gallate and emodin (15), which obviously have the potential to cause jaundice by competitive inhibition of UGT1*1, especially in GS patients in whom hepatic activity is reduced to 35% of normal levels.
Cimetidine has been shown to specifically increase the urinary excretion of naproxen acyl glucuronide, in contrast to other naproxen glucuronide metabolites (48).
Dexamethasone and phenobarbital cause up to a 2-fold induction of digitoxigenin monodigitoxoside UGT in human liver (49). Ethanol treatment of a GS patient led to a 3-fold decrease in serum bilirubin associated with a corresponding 2.5-fold increase in hepatic bilirubin UGT activity (50), and bilirubin UGT mRNA, encoded by the UGTJ*1 gene, was reported to be selectively induced in human livers from patients treated with phenytoin and phenobarbital (51). Indeed, widespread consumption of therapeutic and social drugs, e.g., ethanol, and exposure to environmental chemicals induce UGTs and may mask genetic variation, for example in GS patients (above).

Thyroid Hyperplasia
Plasma thyroxine (T4) concentrations are monitored by the pituitary gland; if thyroxine levels decrease, the pituitary secretes thyroid-stimulating hormone (TSH) that increases the production of thyroxine by the thyroid follicular cells. TSH synthesis is under negative feedback regulation by thyroxine; hence, when plasma thyroxine concentrations return to normal TSH secretion stops. However, continuous depression of plasma thyroxine results in sustained increases of plasma TSH, leading to follicular cell proliferation, hyperplasia, and ultimately neoplasia (52). Plasma thyroxine levels may be decreased by either direct inhibition of its synthesis or by increased metabolism and excretion. A major route of thyroxine metabolism is conjugatiort with glucuronic acid in the liver, followed by biliary excretion.
Rats given four microsomal enzyme inducers had elevated UGT activity toward T3 and T4 of up to 60%, and the circulating T3 and T4 concentrations fell to 70 to 75% of control levels. Furthermore this reduction in hormone levels was not mediated by the thyroid (53). This effect results in continuous TSH synthesis and thyroid hyperplasia. This hyperplastic mechanism is one that is potentially operative in many species induding humans. Glucuronidation of thyroxine is catalyzed by bilirubin UGT and other transferases in human liver (54). Therefore chemical induction of bilirubin UGT by alcohol and other drugs could increase risk of hypothyroidism and subsequent thyroid hyperplasia.

Xenobiotic Inhibition of Glucuronidation
In intact cell preparations (isolated cells, perfusions, in vivo) glucuronidation may Environmental Health Perspectives * Vol 105, Supplement 4. June 1997 be inhibited by either a decrease of UDPGA availability or by inhibition decrease of UGT activity. No highly efficient, selective inhibitors of glucuronidation in vivo or in intact cells are available at present, although Mulder et al. (55) present a fairly recent review.
The studies of inhibition of UGTs are difficult and confused by the microsomal membrane location of transferases. Many compounds that have been considered inhibitors of glucuronidation cause membrane disruption or reduce cellular UDPGA concentrations (56).
Specific analysis is perhaps more easily achieved by measurement of endogenous compound glucuronidation such as the inhibition of bilirubin UGT activity. Gentamycin was shown to be a weak inhibitor of bilirubin UGT in vitro, but also inhibited salicylamide glucuronidation in vivo (57). Bilirubin has been shown to specifically inhibit 3a-hydroxyestradiol glucuronide formation, but not 170hydroxyestradiol glucuronide formation by human liver microsomes, indicating a stereospecific role of bilirubin UGT in formation of this endogenous glucuronide (15). Novobiocin, which caused unconjugated hyperbilirubinemia in animals and man, was shown to exert noncompetitive inhibition of rat microsomal UDP-glucuronosyltransferase in digitonin-activated preparations with either bilirubin or UDPglucuronic acid. Human microsomal bilirubin UGT activity was also inhibited by novobiocin, whereas 1-naphthol UDPglucuronosyltransferase was unaffected (58). Further studies with purified rat liver bilirubin UGT suggested that novobiocin competitively inhibited bilirubin binding to the enzyme, but the drug did not appear to be a substrate for the purified enzyme.
D-ring glucuronides of estriol, testosterone, and dihydrotestosterone possess pharmacological activity and mediate cholestasis (59). In contrast, A-ring conjugates of these steroids are inactive (60). Human bilirubin UGT is the major enzyme responsible for 3-estradiol-3-glucuronide formation in human liver (15). Therefore, the bile ductular cholestasis described in CN patients might be due to the genetic defect in the human bilirubin UGT, which in turn results in excess formation of the cholestatic E2-17f3-glucuronide rather than the noncholestatic E2-3-glucuronide (61).
Androsterone glucuronide is the predominant C-19 steroid glucuronide in plasma (62). The marked rise in testosterone during puberty was strongly correlated with increases of androsterone glucuronide and androstane-3a, 17,-diol glucuronide. Both steroid glucuronides are plasma biochemical markers of adrenal hyperandrogenism in hirsuitism in women (63) and in virilizing congenital adrenal hyperplasia (64). Direct inhibition of UGTs by competitive xenobiotics could significantly affect the production of steroid glucuronides and, in turn, the physiological function responsive tissues such as ovary.
Tertiary amine drugs, chlorpromazine, amitriptyline, imipramine, promethazine, and cyproheptadine were potent inhibitors of the glucuronidation of testosterone, androsterone, and estriol. Structural features required for inhibition were a rigid tricyclic ring and either a dimethylaminopropyl or a methylpiperidine side chain (65). Ketoprofen and other substrates of human UGT2B7 are competitive inhibitors of (S)-oxazepam glucuronidation in human liver microsomes, but not effective inhibitors of (R)-oxazepam glucuronidation, indicating a stereoselective, competitive inhibition of oxazepam glucuronidation, which may be catalyzed by two different UGTs (66).

Organization and Characterization of the Human Sulfotransferase Family
In man, five ST gene products have been identified and characterized and, as is the case for most mammalian STs, these can be classified into two distinct subfamilies (based on their amino acid sequence identities and substrate specificities): phenol ST (PST) and hydroxysteroid ST (HST) ( Table 3) (67,68). PSTs, particularly M-PST and P-PST, sulfate numerous xenobiotics, thyroid hormones, and biogenic amines; the family also includes the estrogen ST (EST) that sulfates both endogenous and xenobiotic estrogens. A (to date) single but somewhat promiscuous human HST isoform exists, which sulfates many diverse steroids, bile salts, cholesterol, and xenobiotic alcohols. It is likely, however, that other human STs exist, since additional STs are present in other mammalian species, for example, a member of the rat PST family (STI CI) that shows remarkable specificity for aromatic hydroxylamines and hydroxamic acids (69). A new nomenclature system for the classification of all STs is currently being devised, which will be generally in line with those adopted for the cytochromes P450 and the UGTs. cDNAs encoding all the known human STs have been isolated, sequenced, and expressed either in bacterial or mammalian cell-based heterologous expression systems. These are proving valuable in defining the boundaries of substrate specificity of the various forms. In general, the kinetic and catalytic properties of these recombinant, expressed STs are as expected. There appears to be some microheterogeneity at the amino acid sequence level with certain STs, in particular with the P-PST. cDNAs encoding this isoform have been isolated in a number of laboratories, and at least three allelic variants differing by up to five amino acids seem to be present within the population. The effects of these variants on the substrate specificity and kinetic parameters of P-PST remain to be determined. A P-loop motif (GXXGXXK) located towards the C-terminus is conserved in all sulfotransferases, and this has been demonstrated to be involved in binding the cosubstrate, 3'-phosphoadenosine 5'-phosphosulfate (PAPS) (70). Site-directed mutagenesis experiments are beginning to identify amino acids important for catalysis (71).
Each ST appears to be transcribed from a single gene, although there is some evidence for alternate splicing from 5' noncoding "exons" that may play a role in tissue-specific expression. STs are expressed in many tissues, with marked selectivity for the different forms. P-PST appears almost ubiquitous and the liver is a major site for P-PST, HST, and EST expression, although M-PST is a minor form in this tissue. This is in contrast to the small intestine, where M-PST appears to be the major form. HST plays a  (94) fundamental biological role during development, as it is responsible for the production of dehydroepiandrosterone sulfate by the fetal adrenal gland during the second half of pregnancy (72), which in turn is the major substrate for placental estrogen biosynthesis. Both M-and P-PST are expressed in the brain, although the function remains unclear. Platelets, which contain P-PST and M-PST but not EST or HST proteins, are one of the most widely studied tissues, although their relevance to other tissues has remained a matter of some debate. The availability of good quality, isoform-specific antibodies is essential for the thorough characterization and analysis of any enzyme family, and this has proven particularly difficult for the STs. Numerous antibodies have been raised against, in particular, rat and human STs, but the high degree of sequence identity among the three known human PSTs (P-PST, ST1A2, and M-PST) means that antibodies distinguishing between them have so far eluded the field. It was demonstrated that synthetic peptides could be used to induce antibodies against human HST and members of the PST family (73), and advancement of this work has now yielded the first antibodies specific for M-PST and for P-PST (Rubin et al., unpublished data).

Biologically Active and Toxic Sulfates
It has been known (but perhaps not widely) for many years that sulfation does not always result in a reduction in biological activity, from Millers' pioneering work on the role of sulfation in the bioactivation of chemical procarcinogens (7). For the antihypertensive and hypertrichotic drug minoxidil, bioactivation by sulfation is beneficial, since the sulfate ester is the pharmacologically active species (74). Elegant experiments demonstrated that the ST responsible for this reaction is expressed in the hair follicles, providing in situ activation of topically applied minoxidil (75), which has found clinical application in the treatment of baldness.
A classical pathway of bioactivation of aromatic amines such as 2-aminofluorene involves N-hydroxylation by cytochrome P450 (forming an aromatic hyc!roxylamine) followed by reaction with PST to form a highly unstable N-0-sulfate that rapidly and spontaneously decomposes, resulting in the formation of a highly reactive arylnitrenium ion that is able to adduct to DNA and proteins.
The requirement for sulfation in the bioactivation of promutagens can be demonstrated in vitro using the Ames test (76). There is also a significant amount of good (albeit circumstantial) evidence for the key role of sulfation in the bioactivation of certain chemical procarcinogens in vivo. The brachymorphic mouse, which has a dramatically reduced capacity for PAPS synthesis, is remarkably resistant to chemical-induced cancer when fed compounds (such as aromatic amines) that require sulfation for activation to mutagens (7). Male rats are much more susceptible to hepatocarcinogenesis following exposure to aromatic amines than female rats, and male rats have up to 10-fold higher levels of expression of the PST(s), which sulfate aromatic hydroxylamines; and the carcinogenicity of aromatic amines is much reduced when coadministered with potent inhibitors of PSTs (7). Human liver PST(s) are primarily responsible for the sulfation of aromatic hydroxylamines and hydroxamic acids (77), whereas benzylic alcohols of polycyclic aromatic hydrocarbons (which are also conditional on sulfation for bioactivation) are preferentially sulfated by HST (78).
The upsurge of interest in sulfation and the STs involved in these bioactivation pathways is principally a result of two factors. First, the demonstration that the hydroxylated metabolites of heterocyclic amines formed during the cooking of meat and fish, which are widely believed to play an important role in colon cancer, are substrates for PST(s) (79,80). Second, there appears to be polymorphic expression of P-PST within the human population. An individual's genetic complement of STs may therefore be an important risk factor in susceptibility to cancer of the colon and possibly other tissues. Testing this hypothesis is therefore an important health priority.

Known Polymorphisms of Sulfation in Man
Platelet P-PST and M-PST activity varies widely within the population (81). Much of the work on sulfation pharmacogenetics has been done using the platelets because they are a readily accessible tissue and because the level of P-PST in platelets appears to correlate with that in other tissues from the same individual, in particular the cerebral cortex, small intestinal mucosa and liver (81). Platelets, of course, are unlikely to play a role in drug metabolism; the simple measurement of enzyme activity may not provide the whole picture, since there are several closely related PSTs known to exist in humans with undoubted overlapping substrate specificity. The discovery of isoenzyme-selective substrates is therefore critical although potentially very difficult when P-PST (ST1A3) and ST1A2 share 96% amino acid identity (67). The difficulties attached to producing molecular probes to distinguish the various forms at the DNA, mRNA, or protein level are also considerable.
Classical enzyme activity measurements in platelets from a large number of individuals and within families have suggested a significant heritability for both P-and M-PST, and polymorphic expression of the P-PST form in this tissue (81). The correlation between P-PST enzyme activity and enzyme protein levels in platelets has also been shown (81). Similarly, statistically significant correlations between the extent of paracetamol (acetaminophen) sulfation in vivo and the levels of platelet M-PST and P-PST enzyme activity (paracetamol is a substrate for both enzymes) have been demonstrated (82), although this drug is also subject to other metabolic routes.
There is also population variability in the expression of human HST, with an apparent bimodal distribution of enzyme activity in a series of 94 liver samples, although the variation was only 4.6-fold (83). Recent sequence analysis of the gene coding for HST indicated the presence of restriction fragment length polymorphisms and nucleotide mutations within the population, and such analyses will facilitate the study of molecular mechanisms of ST regulation in man (84).

Regulation of Sulfotransferases by Xenobiotics
Sulfotransferases are considered to be refractory to induction by "classical" xenobiotic inducers. Recent analysis of the effect of peroxisome proliferators on rat liver ST expression demonstrated a suppression of the major rat PSTs ASTIV (or STIAl) and STlCl (85), and the suppression of rat liver HST expression after treatment with 3-methylcholanthrene has also been shown (86).

Xenobiotic Inhibition of Sulfation and Human Risk Inhibition in Intact Cells and in Vivo
Sulfation can be inhibited either by direct inhibition of the enzymes or by interfering Environmental Health Perspectives * Vol 105, Supplement 4-June 1997 with the biosynthesis of the cosubstrate PAPS through, for example, reduction in inorganic sulfate availability. 2,6-Dichloro-4-nitrophenol (DCNP) and pentachlorophenol (PCP) are potent inhibitors of PST activity (in humans, specifically of the P-PST isoform[s]). Inhibition of sulfation in vivo, in perfused organs and intact cells, and in vitro can be achieved with these chemicals and with many other compounds (87). nhibiton ofST Activity in Vt"r by Drugs and Dietry Compounds The important role of sulfation in modulating the biological activity of key endobiotics and the fact that all human STs number both xenobiotics and endobiotics among their substrates mean that interference with this function by inhibition may have important consequences for the body's physiology and biochemistry.
The compounds DCNP and PCP, particularly potent and selective inhibitors of human P-PST in vitro, were used in the pre-recombinant DNA technology era as tools for classification of the different human STs. Studies on the inhibition of human liver steroid sulfation by a wide range of drugs demonstrated that a number of commonly used pharmaceuticals were potent inhibitors of the sulfation of DHEA by HST and the sulfation of estrone by EST (88). Clomiphene, danazol, spironolactone, cyproterone, and chlorpromazine were all able to inhibit HST activity with an IC50 value of less than 10 pM, and EST activity was inhibited by cyclizine, ibuprofen, chlorpheniramine, dimenhydrinate, and tamoxifen, again with IC50 values < 10 pM. These observations have implications for adverse drug reactions; but their effects in vivo, either in humans or animals, remains to be determined. Another obvious source of xenobiotic inhibitors is the diet, and a number of natural and synthetic dietary chemicals were shown to potently inhibit the sulfation of 17a-ethinylestradiol and dopamine (89). These included vanillin (a naturally occurring flavoring), octyl gallate (an antioxidant), and tartrazine (a synthetic colorant). A number of flavonoids, such as quercetin, are also potent inhibitors of ST activity (90); and red wine, which contains very high levels of polyphenolic compounds, is a very potent and selective inhibitor of human P-PST with a 2000-fold dilution resulting in 50% inhibition of this enzyme activity (91). The importance of these dietary inhibitors may lie in their ability to inhibit the bioactivation of dietary procarcinogens by STs (particularly P-PST) and thus act as natural chemoprotectants.

Concluding Remarks
The "undergraduate textbook" concept of glucuronidation and sulfation as purely detoxication mechanisms is clearly outdated. As we understand more about these complex and fascinating enzyme systems, it becomes increasingly clear that there is a fine balance between their beneficial and harmful properties. The combination of an individual's genetically determined complement of these, and other, xenobiotic metabolizing enzymes and the environmental pressures upon them are obviously major factors in determining which way the balance tips. Not until we have learned much more of the structure/function relationships, regulation and pharmacogenetics of the UGTs and STs, however, will we be able to fully exploit their potential as risk assessment tools.