Abstract
Glucuronidation reaction is catalyzed by mammalian uridine diphosphoglucuronosyl transferases by using uridine diphosphoglucuronic acid as a cosubstrate. Conjugation of glucuronic acid to nucleophilic functional groups in chemical entities results in formation of glucuronides. As anticipated, a number of nucleophilic functional groups such as hydroxyl, phenolic, acyl, primary secondary and tertiary amino, etc. in a diverse set of chemical compounds are known to form the corresponding glucuronides. Glucuronides have been reported to be formed at carbon atoms, selenium atoms, and upon N-carbamoylation of primary and secondary amino groups. Glucuronides are also believed to be the end products of metabolism. However, there are examples where glucuronidation results in further oxidative or conjugative biotransformation reactions. The objective of this review is to highlight unusual glucuronide conjugates. Diglucuronide conjugates reported in the literature fall under two distinct categories. Use of prefixes such as “bis” versus “di” has been previously proposed for separating the two types of diglucuronides. In spite of this, literature reports for diconjugative glucuronide metabolites reflect interchangeable use of “bisglucuronides” and “diglucuronides.” Furthermore, the application of such prefixes does not adhere to recommendations of International Union of Pure and Applied Chemistry nomenclature for substituent groups. Therefore, an effort is made in this review to document the historic reports for diglucuronides into two distinct types for sake of clarity and to allow differentiation between the two types of diconjugative metabolites. Overall, this commentary centers on unusual glucuronide metabolites that result from conjugation at uncommon functional groups, glucuronides undergoing ensuing oxidative or conjugative metabolic transformations. Structural and mechanistic aspects are also discussed.
Introduction
Conjugation of glucuronic acid moiety to a variety of functional groups in molecules is termed as glucuronidation or glucuronosylation. Glucuronidation of endobiotics and xenobiotics uses uridine diphosphoglucuronic acid (UDP-GlcUA) as an activated cosubstrate and occurs at alcoholic, phenolic, primary, secondary, and tertiary amino groups in addition to carboxylic acids. Some acidic carbon atoms, thiols, and amides have also been known to be glucuronidated. Mechanism of glucuronidation is shown in Fig. 1. Glucuronides, the end products of this reaction, are polar, ionized at physiological pH, and higher in molecular weight compared with their parent drugs, thus readily removable from the body. Monoglucuronides are generally considered to be final metabolites, although there is recent evidence with some of the nonsteroidal anti-inflammatory agents that glucuronides may be substrates for subsequent metabolism. Enzymes that catalyze glucuronidation reaction are referred to as uridine diphosphoglucuronosyl transferases (UGTs). Various attributes of UGTs such as nomenclature, structure, function, polymorphisms, in vitro and in vivo models, regulation, and tissue distribution have been published in many reviews and expert opinions that serve as an excellent source of the collective information (Burchell et al., 1998; King et al., 2000; Tukey and Strassburg, 2000; Fisher et al., 2001; Guillemette, 2003; Mackenzie et al., 2003, 2005; Argikar et al., 2008).
UGTs are microsomal enzymes and are located such that their active site faces the lumen of the endoplasmic reticulum. UGT superfamily of enzymes is subdivided into four distinct families: UGT1, UGT2, UGT3, and UGT8 (Mackenzie et al., 2005). UGT families are further grouped into prominent subfamilies: UGT1A (chromosome 2q37), UGT2A (chromosome 4q13), and UGT2B (chromosome 4q13). The UGT1A family in humans comprises several enzymes that share a conserved C terminus containing the UDP-GlcUA binding pocket, encoded by four conserved exons (3′ exons). Twelve variable region exons (cassette exons) encoding the N-terminus region that contains a substrate binding site are spliced to form constant region exons that make the individual enzymes (Remmel et al., 2009). Possibly due to gene duplication events that led to formation of different variable region exons in the UGT1A family (exon 1A1 to exon 1A13P), the UGT1A isozymes exhibit overlapping substrate activities. In general, UGT1A1 exhibits specificity toward bilirubin, bile acids, and phenols. UGT1A4 catalyzes glucuronidation of aliphatic and heterocyclic amines, whereas UGT1A6 catalyzes glucuronidation of phenolic substrates, specifically exhibiting preference toward planar phenols. In contrast, UGT1A8, UGT1A9, and UGT1A10 catalyze relatively bulky phenols.
Unlike, the UGT1A family, the UGT2B enzymes are formed from respective solitary genes. UGT1A and UGT2B enzymes are differentially expressed in liver, kidney, and intestine. UGT2A family is expressed exclusively in olfactory tissues. UGT3A (UGT3A1, UGT3A2, chromosome 5p13.2) and UGT8A (UGT8A1, chromosome 4q26) families have also been identified via genome sequencing, to be present in humans. UGT3A1 has been reported to form N-acetylglucosamidines (Mackenzie et al., 2008). In contrast, UGT3A2 has been shown to form xylosides and glucosides (Mackenzie et al., 2011). Interestingly, UGT8 has been shown to play a role in synthesis of glycosphingolipids and cerebrosides by galactosidating ceramide (Mackenzie et al., 2008).
The purpose of this review is to draw attention to unusual glucuronides; for example, glucuronide conjugates that are formed at functional groups other than those commonly reported, such as hydroxyl, phenolic, amino, and carboxylic acids, etc. The first segment of this review refers to uncommon glucuronides based on the functional groups at which they are conjugated. This initial section reviews the literature for carbon-, thio-, seleno (Se)-, and carbamoyl-linked glucuronides. Glucuronides are generally thought to be end products or terminal reactions. The latter part of this review focuses on examples of deviations from this notion and emphasizes glucuronides undergoing subsequent metabolism, either oxidative or conjugative. Thus, this review highlights unusual glucuronides and makes an attempt to understand the formation of such glucuronides from perspectives of chemical reactivity of parent drug, stability of intermediates if involved, UGT isoforms responsible, and selectivity across animal species.
Usual Glucuronides
A comprehensive list of substrates, inducers, and inhibitors of human UGT enzymes has been published Remmel et al. (2009). Most reports for glucuronides represent conjugation at nucleophilic groups such as phenol, aliphatic alcohol, and carboxylic acid, primary, secondary, and tertiary amino groups. Such functional groups are nucleophilic in nature and therefore are easily suited for conjugation via nucleophilic substitution reaction. UDP-GlcUA is used as a cosubstrate in the reaction, and UDP is released as a by-product in formation of a xenobiotic or an endobiotic glucuronide. Postulated mechanism of glucuronidation is shown in detail in Fig. 1. Over the years, there have been a few reports of conjugation reactions occurring at groups that may not be perceived as obvious nucleophiles. Such reports that lead to formation of unusual glucuronides are documented below.
Carbon-Linked Glucuronides
A carbon atom is not considered to be a nucleophilic atom. However, xenobiotics containing strongly acidic carbons or enolic acids are known to form C-linked glucuronides (Fig. 2). The earliest reports of C-glucuronides were for Δ6-tetrahydrocannabinol, sulfinpyrazone, and phenylbutazone. A C-linked glucuronide of Δ6-tetrahydrocannabinol, a psychotomimetic and analgesic agent, was reported first in vitro, and then in vivo, in mice at C-4 position (Yagen et al., 1977; Levy et al., 1978). A surprising aspect in the formation of this metabolite is that Δ6-tetrahydrocannabinol bears a free hydroxyl group, and an O-glucuronide conjugate at the same group could only be identified in trace amounts (Harvey and Paton, 1980). The uricosuric agent sulfinpyrazone, containing an acidic carbon in the pyrazolidinedione group, was shown to form a C-glucuronide conjugate. This metabolite was reported to be a major metabolite in humans (Dieterle et al., 1980). A secondary C-glucuronide metabolite of sulfinpyrazone, i.e., a metabolite of its sulfone metabolite, has also been reported (Dieterle et al., 1975). Studies have shown that recombinant human UGT1A9 catalyzed the formation of sulfinpyrazone-C-glucuronide in vitro (Kerdpin et al., 2006). Likewise, anti-inflammatory and antipyretic agent phenylbutazone was reported to be metabolized to a C-linked glucuronide conjugate on pyrazolidinedione ring, also by UGT1A9 (Nishiyama et al., 2006). Approximately 40% of total urinary radioactivity after a single human dose was attributed to glucuronides at C-4 of phenylbutazone in this study. Among the two glucuronides that were identified, direct C-4 glucuronide conjugate of phenylbutazone was reported to be greater than a secondary metabolite, a C-4 glucuronide of γ-hydroxy-phenylbutazone (Dieterle et al., 1976). As expected, suxibuzone, a prodrug of phenylbutazone where C-4 position was chemically blocked, did not form the corresponding C-linked glucuronide (Yasuda et al., 1982). Feprazone, another such drug containing a substituted pyrazolidinedione ring, was also reported to form a C-glucuronide as a major metabolite in human (Yamaguchi et al., 1979). It is interesting to note that a metabolite of feprazone, p-hydroxy feprazone, formed C-glucuronide to a lesser extent compared with O-glucuronide (Berry and Parke, 1988). A C-glucuronide on ethynyl moiety of sedative-hypnotic drug ethchlorvynol has been reported as a major urinary metabolite in rabbits (Abolin et al., 1980). Disposition of C-glucuronides has not been well studied. Recent studies with synthetically prepared C-glucuronides of mifepristone showed that the disposition properties of the synthetic analogs were similar to their corresponding O-linked analogs (Sorensen et al., 2003). To date, an accepted test for identification of C-glucuronides has been stability against β-glucuronidase-mediated hydrolysis, in addition to NMR and mass spectroscopic methodologies. This β-glucuronidase-mediated hydrolysis test is a crucial aspect in differentiation of C-linked glucuronides because other usual glucuronides (except α-glucuronides, which formed upon acyl migration) undergo complete hydrolysis when treated with β-glucuronidase enzyme (Argikar et al., 2006). In fact, once formed, the C-C bond appears to be unsusceptible to enzyme hydrolysis or reconversion to parent. Although it may be speculated that C-glucuronides may be formed at acidic carbon atoms, the mechanism of formation of C-glucuronides has not yet been fully understood. It may be fair to estimate that enzyme specificity and chemical properties of a substrate play a role in such a reaction. Extensive reaction phenotyping and perception of structure activity relationship with respect to this reaction is necessary to promote our understanding.
Sulfur-Linked Glucuronides
The earliest report for detection of an S-linked glucuronide dates back to early 1950s. An S-glucuronide was reported as a secondary metabolite of benzothiazole-2-sulfonamide (Clapp, 1956). S-Linked glucuronides reported until now are shown in Fig. 3. Benzothiazole-2-sulfonamide was found to be first metabolized to its corresponding mercapturic acid, which subsequently formed the S-glucuronic acid conjugate. Mercapturates, either administered as parent drugs or formed from sequential metabolism of an administered xenobiotic, have been shown to be metabolized to their corresponding S-glucuronides. Pyrithione-S-glucuronide has been reported as a major urinary metabolite in rat, rabbit, pig, and monkey after administration of pyrithione alone or as a reduced dipyrithione magnesium sulfate complex (Mitoma et al., 1983). Disulfiram (i.e., tetraethylthiuram disulfide), a drug used to manage chronic alcoholism, was reported to be metabolized to an S-glucuronide in humans (approximately 0.75% of the administered dose) (Kaslander, 1963). This particular metabolite was isolated as an S-linked glucuronide of the reduced monomer in human urine, subsequent to 3-day administration of disulfiram. As may be imagined, thiophenol was reported to form an S-glucuronide in mouse liver slices (Dutton and Illing, 1969). Para-nitrothiophenol was also found to be metabolized to an S-linked glucuronide in guinea pig liver microsomes (Smith et al., 1992). Butylate, a thiocarbamate herbicide also known as Sutan, was shown to be subsequently metabolized to an S-linked-glucuronide, as identified in male and female rat urine (Peffer et al., 1991). The metabolic pathway was hypothesized to be driven by S-oxidation, followed by subsequent GSH-driven nucleophilic displacement and GSH adduct formation. This GSH conjugate was further hypothesized to be metabolized to a mercapturic acid only to be conjugated as a glucuronide.
An S-glucuronide was identified as a downstream metabolite of 1,2,4-tricholorobenzene in rat bile. 1,2,4-Trichlorobenzene underwent bioactivation through epoxide formation, followed by formation of a GSH adduct. This GSH adduct was thought to be further cleaved by gut microflora to form the corresponding mercapturate and, subsequently, an S-glucuronide (Bakke et al., 1992). The hepatoprotective agent malotilate is metabolized to a dithiol metabolite. This dithiol metabolite has been reported to undergo conjugation with glucuronic acid to form an S-glucuronide in rats (Nakaoka et al., 1990; Nakaoka, 1990). Tanaproget is a nonselective progesterone receptor agonist. S-Glucuronide of tanaproget at iminothiol/oxazine thione moiety was reported in rat and human liver microsomes (Keating et al., 2006). AR-C133611XX, a CXCR2 receptor antagonist, has been shown to be metabolized to an S-linked conjugate in dog hepatocytes as a major metabolic pathway. Structure elucidation by NMR revealed the S-glucuronide to be present on triazole thione moiety in the molecule (Martin et al., 2003). Finally, phenotyping information on formation of S-glucuronides is scarce. HMR1098, a KATP antagonist, was reported to form an S-glucuronide in vitro in human liver and kidney microsomes and in vivo in rat and dog bile. Human UGT isoforms UGT1A1 and UGT1A9 were found to catalyzed this reaction in vitro, with UGT1A1 exhibiting a higher intrinsic clearance than its corresponding kidney-expressed isoform (Ethell et al., 2003). S-Linked glucuronides are illustrated in Fig. 3.
Selenium-Linked Glucuronides
Selenium is an essential and trace ion. Disposition of selenium occurs either through conjugation with amino acids (Wrobel et al., 2003) or sugars (Suzuki et al., 2005), followed by elimination in urine. Ebselen (PZ 51; 2-phenyl-1,2-benzoisoselenazol-3(2H)-one), a selenium containing organic compound with antioxidant and anti-inflammatory properties, was metabolized to a selenoglucuronide. This unusual metabolite was identified in bile in an isolated perfused rat liver model (Muller et al., 1988). Likewise, preliminary evidence obtained from high-performance liquid chromatography-UV and glucuronidase-based analysis implied that a selenoglucuronide was possibly a secondary metabolite of diphenyl diselenide in mouse urine (Adams et al., 1989). Compounds reported to form Se-linked glucuronides are depicted in Fig. 4.
Carbamoyl Glucuronides
Carbamoyl glucuronidation is a relatively infrequent disposition process. Chemically amino acids are well known to form carbamic acids spontaneously upon reaction with carbon dioxide (Morrow et al., 1974). Tocainide (mexiletine) provided the earliest evidence of N-carbamoylation followed by glucuronidation of a xenobiotic. Tocainide is an antiarrhythmic drug that forms an N-carbamoyl glucuronide at the primary amino group (Elvin et al., 1980; Ronfeld et al., 1982). Because most carbamic acids identified to date have been unstable, it is unknown whether N-carbamoyl glucuronide formation is a sequential reaction of substrate with CO2 followed by glucuronidation or a concerted process that occurs because of rapid equilibrium between the parent drug and its carbamate. Most likely, the reaction varies from substrate to substrate depending on the stability of corresponding carbamates. Compounds undergoing N-carbamoyl glucuronidation are shown in Fig. 5. Mechanistic studies with 13C-labeled CO2 suggest that formation of N-carbamoyl glucuronide of compound 1 was driven by exogenous CO2 in equilibrium with CO2 dissolved in the buffer in rat liver microsomes (Gunduz et al., 2010). Carbamic acids can convert back to parent moieties upon hydrolysis (Tremaine et al., 1989) or even during sample preparation. Rimantadine N-carbamoyl glucuronides have been identified by derivatization followed by gas chromatography-mass spectrometry analysis (Arnaout et al., 1989) or even by solution phase by NMR analysis, as demonstrated for derivatives of piperazine (Kirsch et al., 2000) and desmethyl org 1377 and desmethyl org 5222 (Delbressine et al., 1990). Trans-esterification has also been applied for trapping unstable carbamic acids of benzazepine (Straub et al., 1988) and a compound containing a primary amino group, referred to as compound I (Leung et al., 2001). Shaffer et al. (2005) recently identified an N-carbamoyl glucuronide of a GABA agonist in rats and monkeys. It was reported that N-carbamoylation proceeded through a stable carbamic acid intermediate that could be isolated from plasma. Such other stable carbamic acid was reported by Hayakawa et al. (2003) for garenoxacin in rat and monkey bile and in dog feces and for metoclopramide in our laboratory (Argikar et al., 2010).
N-Carbamoyl glucuronidation has been reported for xenobiotics containing primary and secondary amino functionalities such as carvedilol (Schaefer, 1992), rimantadine (Arnaout et al., 1989), sertraline (Obach et al., 2005), mofegiline (Trapnell et al., 1998), sitagliptin (Vincent et al., 2007), and amosulalol (Nakaoka et al., 1990). N-Carbamoyl glucuronides have been reported in in vitro and in vivo biotransformation studies across a variety of species such as mouse, rat, dog, rabbit, monkey, guinea pig, as well as human. N-Carbamoyl glucuronide of carvedilol was identified in rat and dog liver microsomal incubations (Schaefer, 1992), and that of BVT.2938 was identified in rat, monkey, and human liver microsomal incubations (Tjernberg et al., 1998). N-Carbamoyl glucuronide conjugate of mofegiline was identified in vivo in dog urine and in vitro in dog liver microsomal incubations (Trapnell et al., 1998), and amosulalol-N-carbamoyl glucuronide was isolated from mouse bile (Nakaoka et al., 1990). N-Carbamoyl glucuronide conjugate of varenicline was identified in plasma and urine of rat, monkey, and human, but not mouse (Obach et al., 2006). Garenoxacin-N-carbamoyl glucuronide was detected in multiple species: rat and monkey bile and dog urine and feces (Hayakawa et al., 2003). Tocainide-N-carbamoyl glucuronide has been reported to be present in urine of guinea pig, dog, cat, and rabbit (Gipple et al., 1982). An N-carbamoyl glucuronide of desethyl ropinrole (SK&F-104557) has been reported in urine of monkey and man, but not mouse and rat (Ramji et al., 1999). N-Carbamoyl glucuronides of dipeptidyl peptidase IV (DPPIV) inhibitory stereoisomers ILT-threo and ILT-allo were reported in rat, dog, and monkey in vitro, but they were found to be of significance in vivo only in dogs (Beconi et al., 2003). We recently identified an N-carbamoyl glucuronide of another DDPIV inhibitor, compound 1, in vitro, in mouse, rat, dog, monkey, hamster, and human liver microsomes (Gunduz et al., 2010). Thus, there does not appear to be species selectivity for formation of N-carbamoyl glucuronidation. An increasing number of N-carbamoyl glucuronides have been reported in recent times. One can imagine that this may be due to wide spread availability of soft ionization mass spectrometry techniques and advanced data-dependent acquisition techniques such as data-dependent scanning methodology (Bushee and Argikar, 2011). In addition, this may be the result of an increasing number of biotransformation studies that are being conducted on new chemical entities with primary or secondary amino functionalities; groups that are typically used in drug discovery to enhance solubilization properties of new chemical entities.
Drug metabolism is referred to as a process of transforming xenobiotic molecules into more polar entities, thus facilitating their elimination. Historically, metabolic processes have been referred to as “phase I” and “phase II,” a classification proposed by Williams (1947). This was based on the observation that oxidative reactions were more commonly studied, were also thought to be detoxification steps due to lack of knowledge of subsequent bioactivation steps, and were almost always preceded by conjugation. In-depth analyses of biotransformation processes reveal that drug metabolism examples are best classified as oxidative, conjugative, or hydrolytic, rather than temporal terms such as phase I and phase II (David Josephy et al., 2005). It is even more apparent that conjugation reactions be called as such, and not called phase II reactions, when one considers the plethora of examples of endogenous and exogenous molecules that undergo direct conjugation reactions (such as glucuronidation, sulfation, and acetylation) (Remmel et al., 2007) in addition to examples of glucuronides that undergo subsequent metabolism, as discussed in this ensuing portion of the review. The following sections disclose examples of conjugative metabolites, namely glucuronides that undergo subsequent metabolic transformations.
Diglucuronide Conjugates
A comprehensive list of chemical entities that metabolized to their corresponding diglucuronides is shown in Table 1. To date, two different types of diglucuronides have been identified. Both types are formed by conjugation of glucuronic acid on the same aglycone, i.e., the parent xenobiotic or endobiotic molecule. In the first type, glucuronidation occurs at two different functional groups on the same molecule, whereas in the other, glucuronidation occurs on the first glucuronide (glycone) moiety. Murai et al. (2005) used the prefix “bis” to describe compounds with two glucuronic acid groups on different functional groups of a molecule. “Di” was used to describe compounds where the two glucuronic acid groups were covalently linked to each other. However, literature reports for diconjugative glucuronide metabolites reflect interchangeable use of bisglucuronides and diglucuronides. Furthermore, from an organic chemistry perspective, prefixes such as di and “tri” are recommended to be used for simple substituents, whereas prefixes such as bis and “tris” are recommended to be used for complex substituents according to International Union of Pure and Applied Chemistry (IUPAC; http://www.chem.qmul.ac.uk/iupac/misc/numb.html) nomenclature. Therefore, an easier, clearer yet chemically correct way is documented in this review for the clarity of description and to allow differentiation between the two types of diglucuronides. The diglucuronides are documented as follows: 1) “discrete diglucuronides” for conjugates formed after glucuronidation at two different functional groups on the same molecule, and 2) “linked diglucuronides” for conjugates formed whenever the second glucuronidation occurs on the first glucuronide's sugar moiety (glycone moiety). The very first diglucuronide to be reported was that of bilirubin in 1957 (Billing et al., 1957; Schmid, 1957). However, it was not until a couple of decades later that the structure of this diglucuronide was unequivocally resolved after isolation from human and dog bile (Gordon et al., 1976, 1977). The mechanism of formation of the bilirubin diglucuronide as a result of glucuronidation at two different carboxylic acid groups and specificity of reaction conditions were subsequently resolved in liver microsomal preparations (Jansen et al., 1977; Gordon and Goresky, 1980; Gordon et al., 1983). Thus, the bilirubin diglucuronide is a discrete diglucuronide. Figure 6 illustrates structures and reported sites of diglucuronidation for bilirubin and many other subsequently reported compounds. Likewise, dihydroxydibutyl ether was reported to be metabolized to its diglucuronide conjugate and excreted in rat bile and urine (Corbic et al., 1982). Benzo[a]pyrene, a known carcinogen, was shown to be bioactivated to benzo[a]pyrine3,6-quinone, which was subsequently metabolized to benzo[a]pyrine3,6-hydroquinone. This tertiary metabolite was metabolized to the corresponding benzo[a]pyrene-3,6-diglucuronide in a stepwise manner (Lind, 1985). This particular diglucuronide, identified in vitro in rat liver microsomes, was later found to be predominantly eliminated in bile (Bevan and Sadler, 1992). Multiple diglucuronides of quercetin and methyl quercetin have been reported in rat tissues after administration of 14C-labeled quercetin glucoside (Graf et al., 2005). It is interesting to note that another study showed the presence of such multiple diglucuronides of quercetin in human urine after consumption of cooked onions (Hong and Mitchell, 2004). However, neither of these studies was able to point out the exact attachment of the glucuronic acid moieties on the molecules due to 1) the presence of multiple hydroxy groups on flavonoid molecules and 2) the fact that only tandem mass spectrometry data, but no NMR after isolation and collection of metabolite peaks, were reported. Just like many other poly-phenols undergoing diglucuronidation (summarized in Table 1), resveratrol was also shown to be metabolized to two diglucuronides in human plasma, but, interestingly enough, one conjugation occurred at a phenolic hydroxy, whereas the other occurred at a C atom alpha to a phenolic hydroxy group (Burkon and Somoza, 2008).
Uncannily, diglucuronides reported for testosterone and estrone are unlike any other and are linked diglucuronides, as mentioned in the present review. It is thought that 2′-hydroxyl of the glucuronic acid moiety is the nucleophile that leads to formation of linked diglucuronides, similar to glycosides identified for a variety of plants. During the formation of such glucuronides, 2′-hydroxy group acts as a nucleophile, i.e., X− as represented in Fig. 1, and R group represents a monoglucuronide. The glycone portion of such linked diglucuronides is similar to a group of disaccharides called nonreducing glycosides or glycosyl glycosides, also known as disaccharides without a free hemi-acetal group. The classification of carbohydrates is highlighted in IUPAC-International Union of Biochemistry and Molecular Biology, Joint Nomenclature Committee (Section 2-Carb-36.2, http://www.chem.qmul.ac.uk/iupac/2carb/36.html#361; and Section 2-Carb-22, http://www.chem.qmul.ac.uk/iupac/2carb/22.html) nomenclature recommendations.
Testosterone and estrone form testosterone-17-O-diglucuronide and estrone-3-O-diglucuronide, respectively, as a result of repeated glucuronidation indirectly at the same hydroxyl linkage. The structures of these glucuronides have been confirmed by NMR, and apparently it was indeed the 2′-hydroxyl of the glucuronic acid moiety that acted as a nucleophile to form the linked diglucuronide end product (Murai et al., 2005). Estradiol and estriol showed two similar metabolites each at both 17-hydroxy and 3-hydroxy positions (Murai et al., 2005). These in vitro studies were carried out in rat, dog, monkey, and human liver microsomes fortified with UDP-GlcUA, and the metabolites were species-specific. Androsterone diglucuronide at 3-O-position was identified in dog liver microsomes (Murai et al., 2005). Similar repeat diglucuronides or linked diglucuronides of epiandrosterone and dehydroepiandrosterone were identified in dog urine (Murai et al., 2008). In general, UGT phenotyping has not been performed for many of these reported diglucuronidation reactions that result in either discrete or linked diglucuronides. Thus far, reports only point toward the involvement of UGT1A1 and UGT1A9 for benzo[a]pyrene-3,6-quinol and UGT1A6 and UGT1A7 for chrysene-3,6-diphenol (Bock et al., 1992, 1999). Only UGT1A8 was found to carry out the diglucuronidation of dihydrotestosterone, where the second glucuronide was attached at the C2′ position of the first glycone moiety (Murai et al., 2006). Moreover, there is an interesting hypothesis that UGT enzyme dimers form monoglucuronides, whereas tetramers or oligomers form diglucuronides. This was supported by studies using radiation inactivation analysis, a technique to determine functional target sizes or enzymatic molecular weights in complex membranes such as microsomes. The limited evidence is based on the studies carried out for a handful of substrates: benzo[a]pyrene-3,6-quinol, 3,6-dihydroxychrysene, p-nitrophenol, phenolpthalein, estrone, testosterone, and bilirubin (Peters et al., 1984; Gschaidmeier and Bock, 1994). Therefore, whether the same interpretations are true for many other UGT substrates is debatable.
Diflunisal acyl glucuronide was reported to undergo subsequent glucuronidation at the phenolic group to form diflunisal-diglucuronide; however, this occurred only upon acyl migration (King and Dickinson, 1991). Most notably, the 2-O-linked and 3-O-linked positional isomers of the diflunisal acyl glucuronide underwent subsequent glucuronidation at the free phenolic hydroxy, but the 1-O-linked and 4-O-linked isomers did not. Furthermore, the phenolic glucuronide did not undergo a second glucuronidation at the free acyl group to form the same diglucuronide conjugate. It is noteworthy that only selected regioisomers of migrated diflunisal acyl glucuronide formed the corresponding diglucuronide (King and Dickinson, 1991). Changes in intramolecular hydrogen bonding, steric, electronic, and merely even conformatory changes may lead to differential metabolic profiles. Chemically synthesized alkylated analogs of bilirubin have also been reported to form diglucuronides, predominantly eliminated in bile in Sprague-Dawley rats. However, two synthetic analogs of bilirubin that contain sterically hindering geminal-dimethyl groups formed only the corresponding monoglucuronides but not the diglucuronides (McDonagh and Lightner, 1994). Therefore, it is quite possible that chemical modifications at the molecular level may alter the tendency of closely related structural analogs to form diglucuronides. We recently demonstrated alternation of glucuronidation kinetics for steroid hormones by variant UGT1A4 isozymes (Zhou et al., 2010). The effect of enzyme polymorphisms on the formation kinetics of such diglucuronides is yet to be elucidated. Finally, it is hypothesized that glucuronidation and subsequent diglucuronidation results in inactive metabolites for steroid hormones. Disposition properties of linked diglucuronides of steroid hormones are not well studied. It is possible that the diglucuronidation step may serve as another process by which the circulating concentrations of active steroids may be maintained by the body, and, therefore, it may be hypothesized that such a process may lead to modulation of steroid activity and homeostatic control exerted by the body.
Glucuronide-Sulfate Diconjugates
Glucuronosyl metabolites undergoing sulfation or sulfonyl metabolites undergoing glucuronidation lends an interesting dynamic to biotransformation pathways. Each of the precursor molecules (either a sulfate or a glucuronide metabolite) is hydrophilic, polar, ionized at physiological pH, and, in theory, suitable for elimination; thus, further conjugation of such metabolites (either sulfates or glucuronides) is considered unlikely. However, such metabolic reactions have been documented. A comprehensive list of molecules that are reported to form glucurono-sulfate diconjugates is documented in Table 1. A novel glucuronide-sulfate diconjugate of glycyrrhetinic acid was identified by Jing et al. (2008). Glucuronide-sulfate diconjugates have been identified for some other compounds, the earliest being a urinary diconjugate of mestranol, a 3-methyl ether of ethinylestradiol. The glucuronide-sulfate metabolite was formed after demethylation of mestranol in guinea pigs (Abdel-aziz and Williams, 1974). Naproxen, a nonsteroidal anti-inflammatory agent, undergoes oxidative dealkylation to form 6-O-desmethyl naproxen. An acyl glucuronide-sulfate conjugate of this primary metabolite of naproxen was identified as a major biliary metabolite in male Sprague-Dawley rats (Jaggi et al., 2002). Efavirenz, a non-nucleoside reverse-transcriptase inhibitor of HIV-1, has also been shown to form a glucuronide-sulfate diconjugate in rat urine and bile (Mutlib et al., 1999). It is noteworthy that in this instance, a sulfate metabolite but not a glucuronide metabolite was also identified, hinting that sulfation may have preceded glucuronidation in formation of efavirenz-8-O-sulfate-14-O-glucuronide. Glucuronide-sulfate diconjugates of quercetin and methyl quercetin have been reported in rat tissues after administration of 14C-labeled quercetin glucoside (Graf et al., 2005). Another study showed the presence of a sulfoglucuronide of quercetin in human urine after consumption of cooked onions (Hong and Mitchell, 2004). However, neither of these studies was able to point out the exact attachment of the glucuronic acid or sulfate moieties on the molecules due to presence of multiple hydroxy groups on flavonoid molecules and/or due to the fact that only tandem mass spectrometry data, but no NMR after isolation and collection of metabolite peaks, were reported. A phenolic-sulfate and acyl glucuronide diconjugate of antioxidant ferulic acid was identified in urine and plasma of male Wistar rats (Zhao et al., 2003). Naveglitazar, a similar phenolic-sulfate and acyl glucuronide diconjugate of a dual peroxisome proliferator-activated receptor agonist, was identified in bile of male Fischer-344 rats (Yi et al., 2007). Bisphenol-A-glucuronide-sulfate diconjugate was reported as major metabolite in hepatocytes of male Fischer-344 rats. In this instance, bisphenol-A-glucuronide was a major metabolite in mouse and human hepatocytes (Pritchett et al., 2002). Likewise, a sulfate-glucuronide diconjugate of resorcinol was identified in urine of Fischer-344 rats after oral administration of resorcinol (Kim and Matthews, 1987). In this case, males showed a tendency to excrete more of the diconjugate compared with the females. The same is true for in vitro metabolism of the proestrogenic compound methoxychlor. In studies conducted in liver slices from mouse, rat, quail, and rainbow trout, di-demethylated methoxychlor-4-O-sulfate-4′-O-glucuronide was identified as a rat-specific metabolite (Ohyama et al., 2004). Additional comparative studies with di-demethylated methoxychlor in male and female rat liver slices revealed that di-demethylated methoxychlor-4-O-sulfate-4′-O-glucuronide was higher in male rat liver slices compared those of females (Ohyama et al., 2005). At this time, it is unclear which of the two conjugation processes precedes the other. Thus far, data obtained from the studies has been equivocal. Either initial sulfonation and subsequent glucuronidation or initial glucuronidation followed by sulfonation are equally plausible for the diconjugated metabolites observed so far. Even if extensive studies were carried out to decipher which of the two steps preceded the other, one may envision a scenario in which the sequence of metabolism may vary from compound to compound. It is interesting to note that from the evidence so far, these sulfate-glucuronide diconjugates appear to be reported in rats and guinea pigs compared with any other species. Whether this is due to a UGT or sulfonyl transferase isozyme that is specifically expressed in rats or guinea pigs, and even higher in male rats compared with female rats, remains to be ascertained. However, these observations may be skewed by the fact that most preclinical biotransformation studies are conducted in a rat model, and that more in vivo biotransformation studies with xenobiotics may be conducted using rodents more than a higher preclinical species such as dog or monkey. Finally, reversibility is another aspect of consideration. As shown with naveglitzar (Yi et al., 2007), the formation of the sulfoglucuronide was reversible. Detection of these glucuronides in biological matrices may be dependent on the balance between rates of formation and reversible conversion back to the monoconjugative metabolite or the parent drug. The impact of freeze-thaw cycles, enzymatic or chemical hydrolysis tissue matrices, bioanalytical extraction, and extraction solvents on the stability of sulfoglucuronide diconjugates is yet to be studied. Sulfoglucuronides reported so far are shown in Fig. 7.
Glucuronides Undergoing Subsequent Oxidative Metabolism
A general assumption is that if oxidation and glucuronidation at two different positions in a parent molecule are found to result in a metabolite, the oxidation step always precedes the conjugation step. Unless the corresponding glucuronide is isolated, collected, and purified, and incubation with oxidative cofactors is carried out in the presence of cytochromes P450, one may never know whether the converse it true. Gemfibrozil metabolism is one such example. Gemfibrozil is metabolized to gemfiborzil-1-O-glucuornide as a major metabolic pathway (Okerholm et al., 1976). Oxidation of this glucuronide, but not gemfibrozil itself, at a benzylic position has been reported as a bioactivation pathway, resulting in potent irreversible inhibition of CYP2C8 (Ogilvie et al., 2006; Baer et al., 2009). Likewise, diclofenac glucuronide is shown to be metabolized oxidatively by CYP2C8 (Kumar et al., 2002). It is interesting to note that CYP2C8 has also been shown to metabolize estradiol-17β-glucuronide to its 2-hydroxy analog (Delaforge et al., 2005). When oxidative metabolism of muraglitazar acyl glucuronide and peliglitazar acyl glucuronide was investigated in human liver microsomes fortified with NADPH, multiple oxidative metabolites of the glucuronide reference standards were observed (Zhang et al., 2011). Rates of metabolism in this study were comparable for both the acyl glucuronides. Whether CYP2C8 is involved in the formation of the same metabolites is unclear. One may gain insight into the metabolic role of CYP2C8 from its crystal structure and relevant active site mutation studies. Experimental evidence suggests that arginine 241 residue may indirectly affect anionic substrate hydroxylation due to alterations in packing the helixes G and B' that form the active site (Melet et al., 2004). Glucuronides undergoing subsequent oxidative metabolism are depicted in Fig. 8. Evidence for oxidative metabolites subsequent to glucuronidation comes from in vitro studies carried out for diclofenac, gemfibrozil, and estradiol. However, for muraglitazar, such metabolites were also observed in vivo in rat, monkey, and human bile, and in some cases rat plasma. It is interesting to note that four of the five glucuronides shown above are acyl glucuronides. This is possibly due to increasing awareness and, therefore, the increasing number of studies on disposition of acyl glucuronides due to their bioactivation potential either by trans-acylation or glycation mechanisms.
Summary
The glucuronide metabolites reviewed in this article are fascinating from an absorption, distribution, metabolism, and excretion perspective. The effect of queer physicochemical properties of these unusual glucuronides on absorption, distribution, metabolism, and excretion characteristics is not well understood. For example, C-linked glucuronides are resistant to β-glucuronidase-mediated hydrolysis, and diglucuronides are bulkier molecules and are more polar compared with their monoglucuronide analogs. Formation, stability, and distribution kinetics of such unique metabolites are exciting and yet unknown aspects in drug metabolism and pharmacokinetics. Thus far, the data pose more questions than answers. For example, if a sulfonyl metabolite undergoes subsequent glucuronidation, then it must reach the inner lumen of endoplasmic reticulum of a hepatocyte. What transport mechanisms are involved in this process? How do substrate specificity and enzyme selectivity as well as enzyme polymorphisms influence the transformation of monoglucuronides to subsequent oxidative and diconjugative metabolites? Does one encounter differences in rate and extent of formation of such metabolites across preclinical species and humans or even across genders within a given species? Furthermore, the pharmacological and toxicological impacts of such metabolites are attractive prospects. There are not very many reports available in the literature where such unusual glucuronides have been isolated, purified, and characterized for activity against the pharmacological target of the corresponding aglycone, or for that matter against other known targets. It is unlikely, although not impossible, that a conjugative metabolite may be pharmacologically active, as exemplified by morphine-6-glucuronide (Joel et al., 1985). In an effort to quantitate the circulating levels of lignans and isoflavonoids in human plasma, Adlercreutz et al. (1993) separated diglucuronides, disulfates, and glucurono-sulfo-diconjugates into an inactive fraction by ion exchange chromatography. Lastly, known drug interactions involving UGTs, such as UGT2B7, are reviewed by Remmel et al. (2008). Almost all reported drug-interaction studies have involved formation of monoglucuronides, either as primary or secondary drug metabolites. It is conceivable that formation and subsequent stability of diconjugates add another layer of complexity to this somewhat already multifaceted field of drug interactions with other drugs, metabolizing enzymes, and transporters. Overall, our current understanding of the unusual glucuronide metabolites reviewed herein challenges the traditional concept of biotransformation reactions and elimination pathways.
Authorship Contributions
Wrote or contributed to the writing of the manuscript: Argikar.
Footnotes
Article, publication date, and citation information can be found at http://dmd.aspetjournals.org.
ABBREVIATIONS:
- UDP-GlcUA
- uridine diphosphoglucuronic acid
- UGT
- uridine diphosphoglucuronosyl transferase
- Se
- seleno
- DPPIV
- dipeptidyl peptidase IV.
- Received February 16, 2012.
- Accepted April 19, 2012.
- Copyright © 2012 by The American Society for Pharmacology and Experimental Therapeutics
References
Upendra A. Argikar is a Research Investigator in the Metabolism and Pharmacokinetics Department at Novartis Institutes for Biomedical Research, Inc. (Cambridge, MA). He received a B.S in Pharmacy (2002) from the University of Mumbai and a Ph.D. in Medicinal Chemistry (2006) from the University of Minnesota. His doctoral work involved the study of effects of age, induction, regulation, and polymorphisms on conjugative metabolism of antiepileptic drugs. He joined Novartis in 2006, and his research focused on metabolite identification, structure elucidation, and understanding biotransformation pathways of new chemical entities and clinical candidates. For the past several years, he has continued to be interested in conjugative metabolism by uridine glucuronosyl transferase enzymes, a topic on which he has coauthored book chapters and research papers.