Basic Reaction and Substrate Specificity
Depending on the state of the XOR protein and the availability of electron acceptors, the oxidation of hypoxanthine to xanthine and of xanthine to uric acid occurs by one of two reactions (Fig. 111-2).4 Under physiological conditions the enzyme exists as xanthine dehydrogenase [EC 18.104.22.168] and uses NAD+ as the electron acceptor; the reaction products are oxidized substrate and NADH.5 Both in vivo and in vitro, the dehydrogenase can be converted into xanthine oxidase [EC 22.214.171.124], which utilizes molecular oxygen as the electron acceptor and yields, in addition to oxidized substrate, superoxide (O2 −) and hydrogen peroxide (H2O2).5
Basic reactions catalyzed by the dehydrogenase and oxidase forms of xanthine oxidoreductase (XOR).
Of the naturally occurring purine compounds, XOR oxidizes not only hypoxanthine and xanthine but also adenine (see Chap. 108). Many synthetic purine derivatives, some of which are used as drugs, bind with very low affinity to XOR and thus lack practical significance as substrates.6 The exceptions are 6-mercaptopurine, which is the active compound derived from azathioprine, and allopurinol, which is effectively converted by the enzyme to oxypurinol, a tight-binding and virtually irreversible inhibitor of XOR.7 Purine ribonucleosides are extremely poor substrates, 2′-deoxyribonucleosides only slightly better. In addition to purine derivatives, XOR oxidizes a variety of pyrimidines, aldehydes, and pterins. These compounds have been studied mainly by using enzyme preparations from animal sources,8 but the substrate specificity of human XOR is generally similar to that of the enzyme from animals.6
NAD+ is the physiological electron acceptor for xanthine dehydrogenase. Although this enzyme form is also able to utilize molecular oxygen,4,9 the reaction has little importance under physiological conditions, because it is effectively inhibited by NAD+ 10 and the Km for O2 is fivefold higher than that of the oxidase.11 The oxidase form is unable to use NAD+ as the cosubstrate. Nitrosocysteine may also act as an electron acceptor, being itself decomposed to cysteine and NO in the process.12 Another mechanism of NO formation is reduction of nitrite by xanthine oxidase, in which NADH is required as the electron donor.13 Both enzyme forms are able to use a number of synthetic electron acceptors, including methylene blue, ferricyanide, and phenazine metho- or ethosulfate.
cDNA and Protein Structure
Crystallographic data on XOR itself are not available, but the active-site structure seems to be quite similar to that of a crystallized molybdoenzyme, aldehyde oxidoreductase from Desulfovibrio gigas.14 Conventional protein studies and cloning of the cDNA15– 17 (GenBank D11456) have provided information about the structure of human XOR and allow basic conclusions concerning functional properties.
The polypeptide chain of human XOR consists of 1333 amino acids, compared with 1331 in rat,18 1335 in mouse,19 and 1358 in chicken20 liver enzymes. All the mammalian enzymes are highly homologous at the amino acid level. For example, human and rat enzymes are 90 percent and mouse and rat enzymes 94 percent homologous, whereas human and Drosophila enzymes are only 52 percent homologous.
The purified protein is a homodimer, consisting of identical subunits of ∼150 kDa as estimated by SDS-PAGE.6,21 The amino acid composition deduced from the cDNA sequence corresponds to a molecular mass of 146 kDa. The protein is not glycosylated.
Each subunit contains a molybdenum center, an FAD center, and two Fe-S centers.4 Limited proteolysis of the polypeptide chain of the rat liver enzyme by trypsin, which results in conversion of dehydrogenase to active oxidase, cuts the polypeptide chain at two sites (Lys184 and Lys551). This initially produces three fragments, of 20 kDa, 40 kDa, and 85 kDa, which are not dissociated under nonreducing conditions.18,22 A similar pattern of XOR proteolytic fragments can be seen by treatment of purified xanthine dehydrogenase in vitro with a number of proteolytic enzymes, (e.g., trypsin, chymotrypsin, subtilisin, or papain5,18,22,23) or spontaneously in human liver homogenates, with initial proteolysis close to the N-terminus followed by cleavage at a site further downstream.21 However, it is likely that in most cases the degradation of the 150-kDa polypeptide represents a purification artifact.6,24 Sequence comparisons with proteins containing similar functional groups have indicated that the N-terminal 20-kDa fragment contains the two Fe-S centers, the middle 40-kDa fragment contains the FAD center with the associated NAD+ binding site, and the C-terminal 85-kDa fragment contains the molybdenum center.4,16 Electron microscopy suggests that the three proteolytic fragments correspond to distinct globular structures25 that may represent functional domains (Fig. 111-3). Although the size of the N-terminal fragment, 20 kDa, corresponds to the size of the fragment cleaved during irreversible conversion in vitro and in vivo, the relationship between nicking of the polypeptide chain at this point and modification of the NAD+ binding site in the middle fragment is unclear.
The three structural and functional domains of xanthine oxidoreductase.
The molybdenum is present in the form of a molybdenum cofactor, similar in structure to that in several other molybdoenzymes (see Chap. 128), and is covalently linked to two essential cysteines. The N-terminal part of the XOR molecule is rich in cystine residues, which are essential for binding the Fe atoms in the two nonidentical Fe-S centers.
Kinetics and Reaction Mechanism
Detailed kinetic studies of human XOR have not been published. However, the virtually identical primary structures and active-site conformations of the mammalian enzymes studied allow extrapolations from animal experiments.
In a purified preparation from human liver, in which 90 percent of the enzyme was in the oxidase form, the estimated Km values were 9 μM for hypoxanthine, 7 μM for xanthine, and 5 μM for adenine.6 Approximately similar values for hypoxanthine and xanthine have been found for XOR from other sources, including an oxidase preparation from bovine milk. When the bovine milk enzyme was purified in the dehydrogenase form, the Km for xanthine was clearly lower, at <1 μM. The Km for NAD+ of the latter preparation was 7 μM, and the Km of the oxidase form for oxygen was 53 μM.11 This means that at normal atmospheric pressure the oxidase may not be fully saturated with its cosubstrate, which may partly account for the findings of higher activity of the oxidase form with artificial electron acceptors than with O2. When reduced rat liver xanthine oxidase reacts with molecular oxygen, hydrogen peroxide and superoxide are produced at a molar ratio of approximately 4:1, similar to the ratio found for the human liver enzyme.6 Xanthine dehydrogenase also reacts with O2 and produces reactive oxygen metabolites, but its Km is fivefold that of the oxidase (260 μM), and the reaction is strongly inhibited by NAD+.4,26 The dehydrogenase form is also able to oxidize NADH, with a Km of 1.2 μM, by oxygen and yield superoxide as a by-product. This reaction occurs at the FAD site and thus is not inhibited by allopurinol or oxypurinol.27 On the other hand, reduction of nitrite occurs at the molybdenum site, the electron donor NADH being bound at the FAD site.13 Despite some controversial reports, inhibition of XOR by excess purine substrate or product does not seem physiologically relevant.
A large number of inhibitors of XOR have been developed, mainly to treat hyperuricemia. Allopurinol, the prime example, is readily oxidized by XOR, with a Km of 2 μM.7 The product, oxypurinol, has a high affinity for the reduced molybdenum site, resulting in essentially irreversible inhibition, with an overall Ki of 85 nM.7 Diphenyliodonium acts at the flavin site, inhibiting dehydrogenase and NADH oxidizing activity.
Studies on the mechanism of the enzymatic reaction have indicated that the purine substrate is bound to the molybdenum center and donates two electrons, reducing Mo(VI) to Mo(IV). The oxygen incorporated into the oxidized substrate is derived not from the essential oxo group bound to the Mo center but from a water ligand at the molybdenum site.28,29 After the initial reduction of the molybdenum, there is a rapid intramolecular transfer of electrons via the Fe-S clusters to the flavin center, the latter then being reoxidized by either NAD+ or O2. In both cases, first a bivalent and then a univalent reduction of oxygen produces H2O2 and superoxide4.
Enzyme Forms and their Interconversion
Depending on the experimental conditions, purification of mammalian xanthine oxidoreductases yields one of two interconvertible enzyme forms.5 By rapid purification, avoidance of proteolysis, and use of dithiothreitol, the enzyme from rat liver26 or bovine milk11 remains in the dehydrogenase form, which is its physiological state in intact cells. Xanthine dehydrogenase can be converted into oxidase by sulfhydryl (SH) group oxidation (Fig. 111-4). Titration of bovine milk dehydrogenase with the disulfide-forming reagent 4,4′-dithiopyridine suggests that complete conversion is associated with the generation of four new disulfide bridges.11 This alters the conformation of the enzyme protein and results in loss of the NAD+ binding site. The process can be reversed by the SH-reducing agents dithiothreitol or β-mercaptoethanol. In the reversible conversion of rat liver xanthine dehydrogenase to oxidase, Cys535 and Cys992 are implicated.30
Molecular forms of xanthine oxidoreductase (XOR). XORD = dehydrogenase, XORO(R) = reversible oxidase, XORO(IR) = irreversible oxidase.
Conditions favoring reversible oxidase formation include glutathione depletion,11,31,32 ischemia without or with reperfusion, and presence of oxidants such as hydrogen peroxide and peroxynitrite.33 Preparation of tissue extracts for enzyme assay or purification also leads to conversion, if thiol reagents are not included in the homogenization medium. If proper precautions are taken, the proportion of the enzyme in the oxidase form in tissues from a number of animal species, including humans, varies between 7 and 35 percent. The possibility that even this residual oxidase activity is an artifact cannot be excluded.
Irreversible conversion of dehydrogenase into oxidase occurs through proteolytic cleavage of a fragment of ∼20 kDa from each subunit of the dimeric enzyme,34 followed by proteolysis further downstream, yielding additional fragments of 40 and 85 kDa.24 This conversion may occur during extraction and purification of the enzyme as well as in cells and tissues under various pathophysiological conditions, in which irreversible conversion is preceded by an increase in reversible oxidase.35,36 The protease responsible for the conversion in vivo has not been identified,37– 39 but a mitochondrial protease has been implicated.24
Although proteolysis is the basis for the irreversible dehydrogenase-to-oxidase conversion, fragments are dissociated only under reducing conditions, otherwise remaining associated with the enzyme molecule.18 In the in vivo irreversible conversion, that is, without exogenous proteases, the fragment is cleaved from the N-terminus of the dehydrogenase molecule, between Gln183 and Lys184,24 and subsequently further downstream, possibly at Gln550 (Fig 111-5). Although XOR, as well as other molybdoenzymes, has been relatively well conserved in evolution, both the reversible and the irreversible conversion of dehydrogenase to oxidase occur only in mammalian species. The reason may be that Cys992, which is involved in the reversible conversion, is conserved only in mammalian XORs.30
Primary proteolytic sites of xanthine oxidoreductase. Tr = tryptic cleavage sites, Pr = cleavage site of mitochondrial protease, ? = putative cleavage site of mitochondrial protease.
Cyanide treatment converts xanthine dehydrogenase into an inactive form by removing an essential sulfur atom from the molybdenum center (Fig. 111-4). This desulfo-enzyme can be partly reactivated in vitro in the presence of rhodanese plus thiosulfate and a sulfhydryl reagent40 or of mercaptopyruvate sulfotransferase plus mercaptopyruvate. Since substantial amounts of desulfo-XOR have been found in homogenates of rat liver41 and purified preparations from cow's milk, the desulfo-sulfo interconversion may have physiological significance. In samples of human liver and intestine, specific activity is 2.7–3.0 nmol/min per milligram XOR protein, whereas in human milk, specific activity is only 20 percent of that.21 This indicates that milk contains the enzyme mostly in inactive form. In another preparation from human milk, less than 2 percent of the enzyme was active, and a fivefold increase in activity was obtained by a resulfuration procedure.42 XOR protein may also be inactivated by the loss of molybdenum. This may be due to a mutation (see Chap. 128) or to experimental tungsten supplementation, which results in reversible replacement of the molybdenum atom by tungsten.43 Enzyme forms from bovine milk have been characterized that lack either just the molybdenum atom or the molybdopterin cofactor.44 The demolybdo-enzyme is not known to have physiological significance.
The Xanthine Oxidoreductase Gene
Xanthine oxidoreductase is coded for by a single gene, located on human chromosome 2p2245– 47 and in a homologous region on mouse chromosome 21.48 The gene has been cloned and sequenced from mouse48 and human,49 and data are also available on the 5′-untranslated region and the 5′-end of the rat gene.50– 51 The completely characterized mammalian genes show a complex but highly similar structure, consisting of 36 exons and 35 introns, with well-conserved splice sites, the main difference being in the length of the second intron. The genes are quite large, spanning at least 60 kb. Although the cDNA and protein sequences of mammalian and insect (Drosophila melanogaster,52 D. pseudoobscura,53 and Calliphora vicina 54) xanthine dehydrogenases are more than 50 percent homologous, the insect genes consist of only four or five exons, with well-conserved splice sites compared with those in the mammalian genes.
The 5′-flanking region of the human XOR gene contains a number of consensus sequences of potential regulatory interest. These include binding motifs for C/EBP, NF-κB, AP-1, AP-2, ETS-1, ATF, GATA, and homeobox transcription factors, as well as responsive elements for IL-1, IL-6, tumor necrosis factor α (TNF-α), interferon-γ, glucocorticoids, and thyroid hormones.49 Which of these actually participate in the regulation of enzyme activity in vivo remains to be established. It should be pointed out, however, that despite the remarkable similarity in the structure of the respective genes themselves, the 5′-untranslated regions in human, rat, and mouse show a number of differences. These may account for the known differences in organ-specific expression and regulation of XOR activity between species.
Many different assay methods for XOR activity have been described, usually to obtain an estimate of quantities of both the oxidase and the dehydrogenase forms. The significant issues concerning the assay method include preparation of samples from biologic sources, choice of substrate (natural or artificial), incubation conditions, and measurement of product. Since the effect of these variables is tissue specific, the assay should be optimized for each application.
Xanthine oxidoreductase is cytosolic and thus can be measured in high-speed supernatants of homogenized or sonicated tissues or cells. To prevent reversible conversion of dehydrogenase to oxidase, dithiothreitol should be included in the sample preparation buffer. Protease inhibitors (e.g., phenylmethylsulfonylfluoride and leupeptin) have been recommended to prevent irreversible conversion and EDTA to avoid heavy metal inhibition. Removal of low-molecular-weight inhibitors (e.g., endogenous purines) by dialysis or gel filtration may be desirable, and necessary if activity is to be measured in human plasma with a high uric acid content. The optimal pH varies between 5.6, with pterin as substrate, and 8.4, with xanthine as substrate, but the actual pH chosen will depend on whether maximal sensitivity or physiological relevance is the prime consideration.
The classic method for assaying xanthine oxidoreductase activity is spectrophotometric, using xanthine as the substrate and measuring uric acid formation at 295 nm (ε = 1.1 × 10−4 M−1 cm−1) in the absence (for oxidase) and presence (for dehydrogenase plus oxidase) of NAD+. To avoid possible NADH inhibition, pyruvate and LDH can be added. Hypoxanthine, although more soluble, is not as suitable a substrate because of its two-stage oxidation and potential underestimation of enzyme activity by measurement of uric acid alone.
Since XOR activity in most human tissues is low, the sensitivity of the assay becomes an issue. By using [14C]-xanthine as the substrate and separating the radioactive uric acid by HPLC for measurement, sensitivity can be substantially increased.55
A fluorometric enzyme assay has been developed using pterin as substrate and measuring the product isoxanthopterin.56 Its sensitivity is about two orders of magnitude greater than that of the spectrophotometric method. In this assay, methylene blue must be used as the electron acceptor, but, at least in human liver and rat tissues, it gives results similar to those obtained with NAD+. However, in quantitative terms, the activity levels are not comparable with those obtained with the natural substrate. Although pterin appears to be a poor substrate for aldehyde oxidase, the overlapping specificities of the molybdoenzymes make the use of physiological substrates preferable.
Cell and Organ Distribution
Marked species differences in activity of XOR, as well as in expression of its mRNA, in various organs and cells have been documented.3 In human tissues, conflicting data have been presented, reflecting (at least in part) differences in assay methodology and specificities of antibodies or hybridization probes. Although animal data are relevant in interpreting disease models using animals, we will mainly review human studies on enzyme localization.
In human autopsy samples, substantial activity is present only in liver and small intestine.21 In kidney and lung, low or undetectable activity is found, whereas in other organs (e.g., brain, myocardium, skeletal muscle), activity measurements are consistently negative. These findings do not rule out expression of enzyme activity in a small subpopulation of cells in these organs. Levels of immunoreactive XOR protein correlate well with enzyme activity, which suggests that substantial amounts of inactive enzyme forms are not present.21 Enzyme activity is present in human milk, although at about two orders of magnitude lower than that in cow's milk, but most of the enzyme protein in milk from either source is inactive, presumably in the desulfo form. Normal human plasma does not contain either enzyme activity or protein.
Northern hybridization or ribonuclease protection assay of mRNA extracted from human organs detects expression of XOR only in intestine and liver.57,58 In autopsy samples from human newborn infants, quantitative RT-PCR analysis indicated the presence of mRNA in heart, brain, lung, kidney, liver, and intestine, in increasing order, but the relative amount in liver and intestine was several orders of magnitude larger than that in heart.58
The cellular localization of human XOR has been studied using immunohistochemistry.59,60 We found most of the intestinal enzyme in duodenal and jejunal epithelium, with little or no protein detectable in large intestine and rectum.60 Capillary endothelium in the gut showed slight antibody staining, but the enzyme is clearly not localized exclusively in these cells, as suggested in bovine studies.61 In human liver, hepatocytes are most strongly stained, with some antigen present in the endothelium and less in Kupffer cells. In hepatocytes the enzyme appears to be located in the cytoplasm, analogous to the case with rat liver, in which the cytoplasmic location was confirmed by immunoelectron microscopy.62 In other organs, immunologically detectable xanthine oxidoreductase has been shown in capillary endothelium and vascular smooth muscle cells from the heart and skeletal muscle,63 as well as in retina, which also showed staining of the cones.64 Resting mammary epithelium contains detectable enzyme protein,65 which becomes much more abundant during lactation.60 Malignant cells from human breast tumors are negative for the protein.65
Cell lines of human origin generally express very low levels of active XOR under normal culture conditions. An exception seems to be endothelial cells from aorta,66 whereas umbilical vein endothelial cells have either low or unmeasurable activity.67 An explanation may be loss of enzyme activity upon adaptation of the cells to culture, which has been demonstrated in rat pulmonary epithelial cells.68
Regulation of Xanthine Oxidoreductase Expression
Animal and cell culture experiments have indicated that the activity of XOR is regulated at least at the transcriptional and posttranslational levels by a number of effectors. These responses appear to be not only species but also cell type specific.
Changes in oxygen tension have an inverse effect on enzyme activity in many cells, hypoxia causing induction67,69– 72 and hyperoxia suppression of activity.67,73 The effect of hypoxia seems to be both transcriptional and posttranslational67,69,70 Hypoxia in cell culture does not convert xanthine dehydrogenase into oxidase.67,71,74 In whole lung from rats exposed to hypoxia, XOR activity and mRNA levels are doubled.75
Hyperoxia causes rapid loss of enzyme activity, partly through inactivation by reactive oxygen metabolites,76,77 and partly through a decrease in mRNA levels.67,78
Nitric oxide may participate in the regulation of XOR; on the other hand, the product of the enzyme, superoxide, reacts with NO to form peroxynitrite. Enzyme activity in bovine endothelial cells appears to be rapidly and reversibly inhibited by NO itself and by NO donors, e.g., nitrosoglutathione.72 Elevation of NO levels by L-arginine supplementation suppresses enzyme activity in normoxic and hypoxic rat endothelial cells as well as in rat lung79 but has no effect on mRNA levels. On the other hand, inhibitors of NO synthesis induce enzyme activity in normoxic and especially hypoxic cells.79 However, some of the effects of NO may be only apparent, accounted for by oxidation of uric acid by peroxynitrite, while true inhibition of the enzyme under physiological conditions, in the presence of purine substrates, may be minimal.80
The effects of inflammatory mediators and cytokines on XOR activity are well documented in experimental animals but again vary depending on the cell and species. For example, TNF-α and interferon-γ, as well as interleukins 1 and 6, induce enzyme activity in bovine renal epithelial cells,81 but of these mediators only interferon-γ is active in rat pulmonary endothelial cells.82 The mechanism of the increase seems to be transcriptional activation, since mRNA levels are elevated.81,82 Lipopolysaccharide induces enzyme activity and mRNA production in most mouse tissues except those of the gastrointestinal tract and heart19,83 and doubles enzyme activity and mRNA quantities in rat lung.75 Strong induction can also be seen in the alveolar epithelium of the lactating mouse mammary gland. In rat macrophages, interferon-γ has a paradoxical effect, since XOR mRNA is increased but enzyme activity disappears. The explanation seems to be inactivation of the enzyme by nitric oxide, the production of which is rapidly induced by the cytokine in macrophages.84 Inflammatory mediators may also convert xanthine dehydrogenase to oxidase without altering total enzyme activity.85
Examples of hormonal regulation of XOR include glucocorticoid induction of activity in bovine renal81 and mouse breast83 epithelial cells; in the latter, prolactin has a synergistic effect.
The molecular basis for regulation of human XOR activity has not been elucidated. However, with the recent cloning of the 5′-untranslated region, it will be possible to analyze which of the several consensus recognition sites for transcriptional regulators have physiological significance.