Binding of Xanthine Oxidase to Glycosaminoglycans Limits Inhibition by Oxypurinol*

Although the binding of xanthine oxidase (XO) to glycosaminoglycans (GAGs) results in significant alterations in its catalytic properties, the consequence of XO/GAG immobilization on interactions with clinically relevant inhibitors is unknown. Thus, the inhibition kinetics of oxypurinol for XO was determined using saturating concentrations of xanthine. When XO was bound to a prototypical GAG, heparin-Sepharose 6B (HS6B-XO), the rate of inactivation for uric acid formation from xanthine was less than that for XO in solution (kinact = 0.24 versus 0.39 min–1). Additionally, the overall inhibition constant (Ki) of oxypurinol for HS6B-XO was 2–5-fold greater than for free XO (451 versus 85 nm). Univalent electron flux (\batchmode \documentclass[fleqn,10pt,legalpaper]{article} \usepackage{amssymb} \usepackage{amsfonts} \usepackage{amsmath} \pagestyle{empty} \begin{document} \(\mathrm{O}_{2}^{{\bar{{\cdot}}}}\) \end{document} formation) was diminished by the binding of XO to heparin from 28.5% for free XO to 18.7% for GAG-immobilized XO. Similar to the results obtained with HS6B-XO, the binding of XO to bovine aortic endothelial cells rendered the enzyme resistant to inhibition by oxypurinol, achieving ∼50% inhibition. These results reveal that GAG immobilization of XO in both HS6B and cell models substantially limits oxypurinol inhibition of XO, an event that has important relevance for the use of pyrazolo inhibitors of XO in clinical situations where XO and its products may play a pathogenic role.

The interactions of reactive oxygen species (ROS) 1 with biomolecules that result in alterations in cell function or overt cellular damage have been proposed as contributing to the pathogenic mechanisms of various disease processes (1)(2)(3). Xanthine oxidoreductase is a molybdenum-pterin protein that serves as the rate-limiting enzyme catalyzing the oxidation of hypoxanthine to xanthine and, finally, to urate. Upon sulfhydryl oxidation or limited proteolysis, the dehydrogenase form of xanthine oxidoreductase (xanthine dehydrogenase) is converted to an oxidase (xanthine oxidase or XO), which utilizes O 2 as the terminal e Ϫ acceptor, yielding superoxide (O 2 . ) and hydrogen peroxide (H 2 O 2 ) rather than NADH. Under inflammatory conditions, XO serves as a significant source of O 2 . and H 2 O 2 in the vasculature (4 -8). During such states, it has been demonstrated that xanthine dehydrogenase is released into the circulation, is rapidly (Ͻ1 min) converted to XO, and binds with high affinity (K d ϭ 6 nM) to positively charged glycosaminoglycans (GAGs) on the surface of vascular endothelial cells (9,10). In this location, XO can generate ROS that, in turn, can modulate the bioavailability of nitric oxide (⅐NO) and, thus, vascular cell signaling (8). Xanthine oxidase displays an affinity for heparin sulfate-containing GAGs on endothelial cells; intravenous administration of heparin results in increases in plasma XO activity, suggesting that XO is bound to the vascular endothelium in both humans and animal models of disease (9,11,12). Sequestration of proteins by GAGs does the following: 1) increases their local concentration by up to an order of magnitude; 2) diminishes their rotational and translational mobility; and 3) limits the diffusion of substrates and products, thus creating a microenvironment that differs significantly from the experimental conditions typically used to determine the kinetics of substrate and inhibitor association (13). To more closely model clinically relevant conditions where the enzyme electrostatically associates with endothelial cell GAGs (4,11,12), XO was bound to heparin-Sepharose 6B (XO-HS6B). The binding of XO to heparin-Sepharose affects the kinetics of enzyme activation to result in an increased K m for xanthine, alters the properties of both product-and substrate-dependent enzyme inhibition with increases in K i for xanthine, uric acid and guanine, and also limits superoxide dismutase scavenging of XOderived O 2 . (14). Together, these data suggest that elevations in circulating XO will result in subsequent endothelial association of the enzyme with a prolonged production of ROS that may be relatively resistant to catalytic scavengers. Allopurinol, a pyrazolo-based suicide inhibitor of XO, has been used for several decades in the treatment of hyperuricemia. Recent reports demonstrate that allopurinol administration is also beneficial in both clinical situations and animal models of vascular disease (8,(15)(16)(17). Because endothelial associated XO plays a critical role in oxidant-mediated vascular dysfunction and because GAG association clearly alters the kinetics of enzyme activation, the inhibition kinetics of oxypurinol (the more water soluble and active metabolite of allopurinol) toward XO immobilized by heparin and vascular endothelial cell GAGs was investigated and compared with oxypurinol inhibition kinetics of XO in solution.
* This work was supported by National Institutes of Health Grants R01-HL06611 (to M. M. T. and B. A. F.) and (1T32GM6349003) (to E. E. K.), Fogarty International Center, National Institutes of Health Grants RO3TW05682) (to R. R.) and RO3TW001493 (to H. R.), and funding from the Guggenheim Foundation (to R. R.) and the Cardiome Pharma Corporation. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
§ These authors contributed equally to the manuscript.

EXPERIMENTAL PROCEDURES
Materials-Heparin-Sepharose CL-6B and PD10 G25 Sephadex columns were from Amersham Biosciences. Xanthine, native cytochrome c, and oxypurinol were from Sigma. Medium 199 and serum were from Invitrogen.
Xanthine Oxidase Purification and Binding to Heparin-Sepharose 6B-Xanthine oxidase was purified from fresh bovine cream and stored in ammonium sulfate at 4°C (18). Xanthine oxidase was bound to HS6B as described previously (14). Briefly, XO (2 mg/ml) was added to a fixed amount of gel (0.05 g, dry weight), and the mixture was stirred gently in 5 mM potassium P i , pH 7.4 (2-ml final volume) at 25°C for 30 min. The suspension was centrifuged at 10,000 ϫ g for 5 min and washed, and the pellet was resuspended in 5 mM potassium P i , pH 7.4.
Enzyme Analysis-Enzyme activity was determined as the rate of uric acid formation at 292 nm in 50 mM potassium P i , pH 7.4 (⑀ ϭ 11 mM Ϫ1 cm Ϫ1 ) using a Shimadzu UV 2401PC spectrophotometer. Reaction mixtures consisted of 2 milliunits of HS6B-XO and varying concentrations of xanthine in 50 mM potassium P i , pH 7.4. Reactions were initiated by the addition of xanthine and maintained under continuously stirred conditions at 25°C. Univalent flux was measured as the proportion of superoxide ( For analysis of enzyme reactivation kinetics, the reduced oxypurinol-XO complex was prepared by incubating 1 ml of XO-HS6B (2.8 mg/ml in 5 mM potassium P i containing 1.5 mg/ml bovine serum albumin, pH 7.4) at 25°C with an excess of dithionite and 100 M oxypurinol for 20 min under N 2 , centrifuging, washing, and resuspending (20). To determine reactivation constants, purified oxypurinol-reduced XO complex in 5 mM potassium P i , pH 7.4, with 1.5 mg/ml bovine serum albumin was incubated at 37°C in the absence or presence of 10 and 200 M xanthine. At various time points, aliquots were removed, and XO activity was determined as described above. First-order reactivation constants (k react ) were calculated from the slope of plots of ln (20).
Cellular Studies-Bovine aortic endothelial cells were isolated as described previously (21). Primary cell culture, routine passage, and experimental manipulations were all conducted in the absence of proteases. Cells were propagated by subculturing at a 1:4 ratio in medium 199 containing 5% fetal calf serum, 5% iron-supplemented and defined calf serum, and 10 M thymidine in 75-cm 2 flasks. Upon reaching confluence, the medium was replaced with fresh medium (medium 199(Ϫ) formulated without xanthine and hypoxanthine) containing XO (5 milliunits/ml) for 1 h. The XO-containing media were removed, cells were washed, fresh medium 199(Ϫ) was added that contained designated concentrations of oxypurinol, and reactions were initiated by the addition of 50 M xanthine for 1 h. Xanthine oxidase activity was determined by measuring uric acid formation via high pressure liquid chromatography with diode array detection as described previously (9). Uricase activity, as determined by the addition of known concentrations of uric acid in medium with or without cells (37°C) and then monitoring the loss of uric acid over time, was undetectable.
Statistics-Data were analyzed for statistically significant differences using a one way analysis of variance followed by a Tukey's range test for the multiple pairwise comparisons.

Inhibition of Uric Acid and Superoxide Formation-Oxypuri-
nolinhibitedbothfreeandHS6B-XOinatime-andconcentrationdependent fashion; however, the extent of XO inhibition by oxypurinol was less for the HS6B-immobilized enzyme at any given xanthine concentration (Fig. 1). Complete inhibition of the HS6B-immobilized enzyme was not attainable at clinically relevant concentrations of inhibitor (Ͻ90 M; see Table III); rates of uric acid production at 75 M oxypurinol did not differ from those at 20 M (not shown).
To  (19), was determined. Free XO displayed a greater proportion of univalent flux than did HS6B-XO in the absence of inhibitor (Table I). For both free and bound XO, the univalent electron flux percentage was not altered by oxypurinol in spite of decreasing uric acid formation rates, indicating that the rate of univalent reduction of molecular oxygen was proportionally decreased as well.
Inactivation and Reactivation Kinetics-The apparent inactivation constant (kЈ inact ) of oxypurinol for XO was obtained from the rate of uric acid formation as described previously (20). The kЈ inact versus kЈ inact /oxypurinol concentration was plotted for both free and HS6B-XO (Fig. 2). The true first-order inactivation constant (k inact ) and the dissociation constant (K d ) were obtained from the intercept of the y axis and the negative slope, respectively. The k inact for free XO was greater than HS6B-XO at 10 M xanthine and saturating XO concentrations, indicating that HS6B-XO was inhibited more slowly by oxypurinol than by free XO. The calculated K d was ϳ2-fold greater than that for free XO both at low and saturating substrate concentrations (Table II).
Oxypurinol forms an inhibitory complex with XO in the fully reduced Mo(IV) state (22,23,24). Hypoxanthine, xanthine, and allopurinol all provide this reductive function toward molybdenum. Dithionite can also reduce XO, permitting full oxypurinol association, XO inhibition, and determination of rates of spontaneous and xanthine-dependent reactivation ( Fig. 3 and Table  II). Spontaneous reactivation of HS6B-XO (0.0016 min Ϫ1 ) was 10-fold slower than the reported value for free XO (ϳ 0.018 min Ϫ1 ) (20). The addition of xanthine increased HS6B-XO reactivation in a concentration-dependent manner with k react values of 0.0048 and 0.0288 min Ϫ1 at 10 and 200 M xanthine, respectively. From these data, an overall inhibition constant (K i ) was determined for HS6B-XO at low and saturating xanthine concentrations, with a K i 2-5-fold greater than that obtained for free XO (Table II).
Inhibition of Cell-associated XO-To add cell biological context to the observation that GAG immobilization of XO results in resistance to oxypurinol inhibition, the extent of oxypurinol inhibition of bovine aortic endothelial cell XO was determined following exposure of cells to XO (Fig. 4). Comparable with GAG-bound XO, oxypurinol-dependent inhibition of uric acid formation in XO-treated bovine aortic endothelial cells was significantly less than for free XO. Increasing oxypurinol concentration up to 400 M, far exceeding typical therapeutic concentrations (Ͻ 90 M), showed no additional XO inhibition (see Table III).

DISCUSSION
Although XO exhibits a net negative charge at physiological pH, cationic amino acid motifs on the surface of the protein result in its high affinity for negatively charged GAGs (K d ϭ 6 nM) (9,14,25,26). This association alters its kinetic properties, increasing the K m and K i for substrates and products when compared with XO in solution because of a decrease in binding affinity (14). In healthy humans, plasma xanthine plus hypoxanthine levels are typically 2 M, suggesting that immobilized XO would manifest reduced enzymatic activity (K m ϭ 21.2 M for xanthine). However, during a variety of pathologic conditions, plasma xanthine plus hypoxanthine concentrations can rise to 20 -100 M, a range sufficient to reach and exceed half-maximal activity for the immobilized enzyme (Table III). Such increases in substrate concentration, coupled with a potential immobilization-induced increase in XO stability, indi-   2. Comparison of the inactivation kinetics of HS6B-XO versus free XO. The pseudo first-order inactivation constant (kЈ inact ) was plotted with respect to the ratio of kЈ inact to oxypurinol concentration, and the dissociation constant (K d ) and the inactivation first-order rate constant (k inact ) for free (f) (k inact ϭ 0.4 min Ϫ1 ; K d ϭ 0.82 M) and bound XO (q) (k inact ϭ 0.26 min Ϫ1 ; K d ϭ 0.45 M) were determined. The kЈ inact was obtained using 50 M xanthine for the free XO (2 milliunits/ ml) and 200 M xanthine for the bound enzyme (2 milliunits/ml) in 50 mM potassium P i , pH 7.4, at 25°C. K d represents the dissociation constant of oxypurinol from the initially formed oxypurinal-reduced complex, and kЈ inact represents the first-order constant for the conversion into the stable complex (kЈ inact ϭ k inact Ϫ K d ).  cate that XO bound to vascular cell GAGs could serve as an important source of ROS. In support of this finding, treatment with allopurinol improved vascular function both in animal models and in clinical vascular diseases such as diabetes and heart failure, where endothelial-associated XO is implicated as a significant source of ROS (8,15,16). Although there is an abundance of data regarding the impact of pyrazolo XO inhibitors such as allopurinol or oxypurinol on hyperuricemia, information on the effect of the production of uric acid and ROS when GAGs immobilize XO is limited. In humans, allopurinol (plasma half-life is 1-3 h) is rapidly converted to its active metabolite, oxypurinol, which persists with a plasma half-life of 12-30 h (27). Thus, examination of the inhibitory properties of oxypurinol toward immobilized XO is of biological interest and clinical relevance. Substantially more oxypurinol was required to inhibit bound XO (K i ϭ 230 nM for free XO versus 405 nM for GAG-bound XO). Following the binding of XO to GAGs, it has been determined that immobilization, not diffusional limitations, alters the affinity of XO for xanthine at the molybdenum site (14). Because the reduction of XO at the molybdenum site is a prerequisite for oxypurinol binding, the higher K m of immobilized XO for the required reductant (xanthine) results in lower rates of inhibition for oxypurinol toward the bound enzyme and may also explain the slower rates of HS6B-XO reactivation.
When tested in a cultured endothelial cell model, exogenously added XO binds to GAGs and displays significant resistance to inhibition by oxypurinol (Fig. 4). As with HS6B-XO, complete inhibition of cell-associated XO activity does not occur at clinically relevant oxypurinol concentrations (Table III). These similarities in cell-associated XO responses to oxypurinol affirm the usefulness of using HS6B-XO as a model for studying the kinetic parameters of this proinflammatory and pathogenic enzyme when associated with the vasculature. However, a modest proportion of cell-associated XO becomes internalized over time (9); therefore, it is also possible that the inability of oxypurinol to achieve complete inhibition of cell-associated XO may be due to limited access of the inhibitor to this internalized protein as well as to alterations caused by GAG binding of XO to the cell surface. Future studies may be directed to evaluate the extent to which cell internalized XO accounts for limitations to oxypurinol inhibition of cell-associated XO.
During diverse pathological states such as type I diabetes, sickle cell anemia, and heart failure, circulating XO levels are substantially elevated (6,7,15) (Table III). Negatively charged GAGs comprising the glycocalyx of vascular endothelial cells in turn may sequester circulating XO and significantly amplify its local concentration (28,29). This amplification occurs at an anatomical site critical for vascular cell ⅐NOand redox-dependent signaling reactions that modulate vessel relaxation as well as vessel wall-inflammatory cell interactions and intimal cell proliferation. In conditions such as ischemia where ATP catabolism is increased, the concentrations of purine substrates are also elevated (Table III). This confluence of inactivation-resistant XO and increased levels of substrate in the vascular compartment sets the stage for enhanced formation of O 2 . and H 2 O 2 . Such increases in rates of ROS production limit the bioavailability of ⅐NO as well as augment the formation of ⅐NO-derived species such as ONOO Ϫ that can potentiate oxidative inflammatory injury (8). Recent studies affirm such a role for XO in vascular disease by demonstrating increases in NO-dependent vasodilatation upon treatment with allopurinol (8,30,31). In the aggregate, these data demonstrate that GAG association with XO results in an enzyme with an increased half-life that is resistant to oxidative inactivation. Importantly, these results begin to provide insight into the interactions of XO and XO-derived ROS with the vascular surface. Additionally, these observations reveal that in a biological milieu, XO inhibitors such as oxypurinol do not fully inhibit the production of XO-derived ROS or uric acid and indicate the need for development of alternative approaches to achieve inhibition of cellassociated XO.