The Resolution of Active and Inactive Xanthine Oxidase by Affinity Chromatography*

The resolution of fully functional xanthine oxidase from nonfunctional enzyme by the technique of affinity chromatography is described. Both forms of the enzyme possess the full complement of the oxidation-reduction components molybdenum, FAD, iron, and acid-labile sulfur. Nonfunctional xanthine oxidase does not possess the cyanolyzable persulfide group and can be partially activated by incubation with sodium sulfide. The presence of the active center persulfide is essential for catalytic activity with xanthine as substrate but is not required for DPNH-ferricyanide reductase activity. It is also required for the properties of


SUMMARY
The resolution of fully functional xanthine oxidase from nonfunctional enzyme by the technique of affinity chromatography is described. Both forms of the enzyme possess the full complement of the oxidation-reduction components molybdenum, FAD, iron, and acid-labile sulfur. Nonfunctional xanthine oxidase does not possess the cyanolyzable persulfide group and can be partially activated by incubation with sodium sulfide. The presence of the active center persulfide is essential for catalytic activity with xanthine as substrate but is not required for DPNH-ferricyanide reductase activity. It is also required for the following properties of xanthine oxidase: (a) rapid bleaching of the oxidation-reduction chromophores of the enzyme by xanthine or by sodium borohydride, (b) development of the "rapid" molybdenum electron paramagnetic resonance signals by xanthine or NaBH4, (c) rapid development of the electron paramagnetic resonance signals of the reduced iron sulfur chromophores by xanthine or NaBH4, (d) interaction with the inhibitors arsenite and cyanide, (e) development of the molybdenum electron paramagnetic resonance signals associated with inactivation of the enzyme by formaldehyde.
A reaction mechanism for xanthine oxidase involving the persulfide linkage is proposed.
The presence of an inacstive form in santhine osidase l)repartitions was first suggested by Morel1 (I) to account for the biphasic course of enzyme reduction by substrate.
Work by Hart et al. (2) indicat.ed the esistence of nonfunctional xanthine oxidase containing the full complement of the oxidation-reduction groups molybdenum, F.211, iron, arid S= in addition to the inactive form lacking n~olyt~denutr~. Additional evidence for the presence of this inac%ive form has come from studies on the mechanism of inactivation by iodoncetamide (3) and on the interaction of xanthine oxidaxo with various pyrazolo (3,5). These studies, and those of M&art011 et al. (6), indicate that * This work was supported by Grants G1L111106 and GM12176 from the United States Public Health Service.
active xanthine oxidase is rapidly reduced by substrate while inactive enzyme is ody slowly reduced. It would be of obvious advantage to have a method for seljaration of functional from nonfunctional enzyme so as to delineate their respective chemical, physical, and catalytir properties.
A comparative study of the two forms would be expected to give information concerning the factor responsible for loss of functionality.
Previous data show a stoichiometric complement of oxidatiorl-reduction groups for enzyme preparations containing 30% of the inactive form (7), indicating the presence of another component essential for activity.
Recent work in this laboratory (8) has shown that inactivation of xanthine oxidase by cyanide is due to cyanolysis of a persulfide group required for catalysis. The possibility was considered that lability of this persulfide might be responsible for nonfunctional enzyme. The present work provides further evidence for this hypothesis. In addition, the isolation of >95c/; functional enzyme is described.
The resolution of functional from nonfunctional cnzyme was achieved by taking advantage of the strong complesforming of active enzyme with the pyrazolo(3,4-d)pyrimidines (4, 5,9). This property was utilized with a derivatized Sepharose gel, resolution thus being achieved by affinity column chromatography.

General Procedures
Milk santhine osidase was isolated as descAed previously (7). Xanthine-oxygen reductase artivity was measured spectro-photometric~ally at 295 run at 25" and expressed as AFR2j" values (10). Enzyme concentrations were estimated spectrophotometrically with an E450 = 37,800 cni+ M-' per molecule of enzyme-bound FAD (7). Molybdenum was determined qualititatively with the dithiol method of Ringley (11) ami iroll as the o-phenanthroline chelate after trichloroacetic acid denaturation of the enzyme (12). Labile sulfitlc was determilled by a modification of the method of Fogo and I'opowsky (13). Most of the solvent was removed in ~acuo, and the resulting black syrup was poured onto crushed ice, stirred well, and extracted with diethyl ether (4 X 400 ml). The ethereal phase was washed once with water, dried over sodium sulfate, and filtered.
The slightly yellow solution was reduced in volume to about 200 ml. The resulting precipitate was dissolved in 150 ml of absolute ethanol, and 3-aminopropanol (12 ml, 0.16 mole) was added. The reaction mixture was warmed on the steam bath until the remaining ether had evaporated, and the ethanolic solution was then refluxed for 13 hours. The ethanol was removed in vacua, producing a thick yellow slurry, which was diluted with 25 ml of water and stored at 0" for 2 days. The resulting solid was collected, dissolved in 100 ml of hot water, treated with charcoal, filtered, cooled, and the resulting off-white precipitate (11.48 g, 40%) collected, m.1'. 213-215".
After stirring for 20 hours at room temperature, the solution was diluted with 100 ml of ethyl acetate, then washed with 5% sodium bicarbonate (2 X 100 ml) and water (100 ml). The organic phase was dried over magnesium sulfate and the solvent removed in vacua, producing an oil. This was chromatographed on silica gel and eluted with chloroform-methanol, 20 : 1. The resulting product was crystallized from ethyl acetate-hexane, (

Preparation of Activated Sepharose
A solution of cyanogen bromide (5.0 g, 0.047 mole) in 50 ml of water was added to a rapidly stirred suspension of 50 ml of Sepharose 6R in 50 ml of water.
The pH of the solution was maintained at 11 by addition of 4 N sodium hgdroside. After 3 min no further change was observed in the pH, and the reaction mixture was diluted with ice, filtered on a coarse sintered glass filter, and washed with cold sodium bicarbonate buffer, pH 9.0 (3 X 200 ml). A solution of compound III (60 mg, 0.20 mmole) in 50 ml of cold buffer was added at once, with stirring, and the suspension was transferred to a beaker and stirred at 4" overnight.
The activated Sepharose was collected and washed with water.
About 30 mg of Compound III were attached to the Sepharose in this manner as estimated from the ultraviolet absorption spectrum of the washings. The activated Sepharose was stored as a gel at 4".

Isolation of Low and Nigh Activity Xanthine Oxidase by Afinity
Chromatography-The 4-substituted pyrazolo(3,4-d)pyrimidine attached to the Sepharose was converted to its 6-hydroxy derivative by incubation at room temperature with xanthine oxidase in 0.1 M PPi, pH 8.5. The conversion was assumed complete when a small sample of gel would no longer reduce an anaerobic sample of enzyme.
Anaerobiosis was established in the derivatized Sepharose column (1 x 15 cm) by equilibration under nitrogen with 0.01 M Na2S204 in 0.1 M PPi, pH 8.5. Approximately 0.6 pmole (expressed as E .FAD) of salicylate-free xanthine oxidase in a volume of 2 ml was reduced with a few crystals of dithionite, applied to and slowly washed into the column.
(Preliminary results indicated interference by salicylate with the complexforming properties of the column.) The flow was stopped and the charged column was equilibrated overnight at room temperature to allow sufficient time for complex formation with active enzyme. Low activity xanthine oxidase was then eluted with the dithionite-containing buffer (Fig. 1). The column was then washed aerobically with cold 0.1 M PPi, pF1 8.5, containing 1 INM salicylate and 0.2 mM EDTA.
A great enhancement of color intensity of the reddish brown band of bourld xanthine osidase became apparent as the iron and flavin chromophores reosidized.
Flow was stopped and the column allowed to stand in the cold room for 3 to 4 days to permit the slow reoxidation of reduced enzyme-bound molybdenum and the accompanying release of xanthine oxidase frorn the gel (4, 5). Elution of the column with the same buffer results in a reddish brown band with very high activity.
This activity is stable for several weeks in the presence of salicylate but is significantly less stable in its absence.
Fractions with similar specific activities were pooled and concentrated by ammonium sulfate precipitation (0.35 g per ml). This step had no effect on the specific activity.
Spectral Properties-& shown in Fig. 2, the absorption spectrum of low AFR2j" xanthine oxidase is quite similar to that of the high activity enzyme in the visible spectral region.
Differences are apparent in the near-ultraviolet region with a maximum in the difference spectrum at 320 nm. This difference spectrum is quite similar to that produced by cyanide inactivation (8,16). The visible absorption spectra indicate identical iron and flavin compositions for both forms. Analysis showed 1 mole of molybdenum and 4 moles of iron and acid-labile sulfur per mole of Fa4D for both low and high XFR enzyme fractions. borohydride samples.
The instrument settings were the same as those in Fig. 4. atonl as the common site for inhibition of enzymic activity. Previous work in this laboratory (8) showed that the liberation of SCN-as well as the extinction change at 320 nm produced by cyanide treatment was proportional to the AFRz5' of different enzyme preparations.
This proportionality has also been found upon cyanide treatment of low and high AFR2j" enzymes isolated from the affinity column (Fig. 3). These data indicate that the catalytically essential persulfide group (8) is not present in inactive enzyme.
The extent of extinction change at 380 nm upon arsenite treatment is also proportional to xanthine oxidase activity (Fig.  4). In agreement with the results of Coughlan et al. (16), arsenite treatment of cyanide-treated enzyme produced no spectral changes. Similarly, arsenite incubation of nonfunctional enzyme resulted in insignificant spectral changes. These data, coupled with the fact that arsenite interferes with cyanide inactivation of the enzyme, indicate the persulfide group to be one of the ligands in the arsenite-xanthine osidase complex. Emyme-Addition of substrate to an anaerobic solution of xanthine oxidase, produces an initial rapid phase of bleaching followed by a slow, secondary phase (7). If the secondary bleaching is due to reducation of inactive enzyme by reduced active enzyme (I), the degree of initial bleaching by substrate should be greater with enzyme of increasing specific activity.
In agreement with previous results (4-6), Fig. 5 shows a linear dependence of the extent of substrate bleaching with increasing specific activity. High AFRZbO enzyme (204 to 205) gave less than 5% secondary bleaching when incubated with an excess of xanthine under anaerobic conditions, supporting the previous conclusion that fully functional enzyme should have an .4FRz5" value of about 210.
The ext,ent of bleaching of xanthine oxidase absorbance by dithionite was identical for enzyme solutions of different specific activit,ies.
However, when sodium borohydride was used as reductant, the extent of reduction was dependent on the specific activity.
As shown in Fig

I
ing was identical with that produced by xanthine, thereby indicating that borohydride can reduce only functional enzyme. Borohydride did not reduce cyanide-treated xanthine oxidase and did not affect the AFRz5' of active enzyme.
In a companion series of experiments, the extent of non-hence iron reduction by either xanthine or borohydride was determined by electron paramagnetic resonance spectroscopy. The intensities of the g = 2.11 and 2.02 signals (produced by either reductant) relative to the signal obtained with dithionite increased linearly with increasing AFR@ values (Fig. 6).
Recently, McGartoll et al. (6) have reported the linear increase in the intensities of the rapid molybdenum electron paramagnet'ic resonance signals and the decrease in intensities of the slom molybdenum signals with increasing enzymir activity. The data in Fig. 7 substantiate their observations and emphasize again the similar behavior of xanthine and borohydride as reductants.
Lack of Requirement of PersulJide Group in Reactions with NASH-Deflavo xanthine oxidase is devoid of NADH-Fe(CN)G' reductase activity (19), thereby indicating the FAD as the acceptor of reducing equivalents from NADH.
Therefore if the lesion causing inactivation of native enzyme was associated with the molybdenum site, one would expect the same diaphorase activity for both high and low ,4FR 250 forms of xanthine oxidase. This was found to be the case; no difference in NADH-ferricyanide reductase activity was found, no matter the AFR250 value. In agreement with earlier reports (16,19), cyanide treatment had no effect on diaphorase activity.
Interaction of Pyrazolo(S,4-d)pyrimidines with Differing AFR Enzymes-Previous work has shown that alloxanthine forms a strong complex with reduced xanthine oxidase (4, 5). The fractional molar ratio of alloxanthine to active site needed for complete inhibition indicated that only functional enzyme could bind alloxanthine.
The mode of inhibition was shown to be a tight complex between alloxanthine and MoIV since two oxidizing equivalents were required to reconstitute catalytic activity.    (Fig. 8) indicates a maximal theoretical AFR25" value of 210 when extrapolated to a molar ratio of 1.0. The binding of various pyrazolo(3,4-Qyrimidines to reduced xanthine oxidase can be detected as a rapid increase in absorption in the visible region (4, 5). A particularly distinct difference spectrum with a maximum at 500 nm occurs rapidly when reduced xanthine oxidase is incubated with 4, B-dimerraptopyraeolo(3,4-d)pyrimidine.
As shown in Fig. 9, the extinction of this difference maximum is linearly related to xanthine oxidase activity.
Evidence for a slow rate of binding of the 4,6-dimercapto compound to inactive enzyme is also presented in Fig. 9. The difference extinction maximum for the inactive enzymepyrazolo(3,4-d)pyrimidine complex is located at 540 nm. The identical difference spectrum of the cyanide-inactivated enzyme (as well as the slow rate of its formation) with that of inactive  I  I  I  I  I  I  I  I  I  IO  20  30  TIh%  MIN5P  60  70  60  90 FIQ.
10. Activation of low AFRz5' xanthine oxidase by Na& Xanthine oxidase, AFRz6" = 3, was incubated at 45' with 0.01 M Na8 in 0.1 M PPi, pH 8.5, for the time indicated. xanthine oxidase indicates a very similar type of binding.
The persulfide group thus appears to influence the rate of binding of pyrazolo (3,4-d)pyrimidines to the molybdenum site as well as influencing the spectral properties of their complexes.
Reactivation of Low AFR25" Xanthine Oxidase-In view of the many similar properties of low AFRZ5" xanthine oxidase with those of the cyanide-treated enzyme, one would expect reactivation with Na2S as previously shown for cyanide-inactivated xanthine oxidase (8). As shown in Fig. 10 incubation of AFR@ = 3 enzyme with 0.01 M Na*S resulted in an increase in specific activity to AFRzSo = 55. It should be noted that the conditions for activation are rather critical; the rate of activation is markedly dependent on temperature, and becomes significant only at temperatures where denaturation also begins to occur.
Dithionite Tiirafion of Native and Cyanide-treated Xanthine Oxidase-In view of the-indicated catalytic requirement for the persulfide group and to probe its possible role in catalysis, dithionite titrations were performed on the cyanide-treated enzyme (with no persulfide group) and on native enzyme of approximately 80% functionality with the method of Foust et al. (20). The results are shown in Fig. 11. Because of the lack of a sharp spectral end point for full reduction at 450 run, end points were estimated from the appearance of dithionite absorption at 315 nm, a wavelength which is very close to the isosbestic point of oxidized and reduced xanthine osidase.
The results of duplicate titrations for both forms of enzyme were identical within experimental error.
The extrapolated dithionite absorption indicates an end point at 4 moles of dithionite per mole of enzyme flavin, in agreement with previous data (7). These data indicate that WJ electron equivalent< are taken up by the persulfide group on complete reduction.

DISCUSSIOiT
The resolution of a santhine oxidase preparation into low and high activity fractions by affinity chromatography provides conclusive evidence for the existence of nonfunctional enzyme found between the active and inactive enzyme is the presence of in preparations containing stoichiometric amounts of molyb-the persulfide in functional enzyme. This is indicated by the denum, FAD, iron, and acid-labile sulfur. As indicated in Fig. linear correspondence of SCN-production and Acaeo with the 1, the activity increased across each eluted band. The small AFRt5' value of the enzyme upon cyanide treatment. The amount of activity in the first band is due presumably to the near-ultraviolet spectral properties of the inactive enzyme and presence of a small amount of uncomplexed functional enzyme.
the identical spectral properties of the complexes formed with The finding of less than theoretical specific activities in the sub-4,6-dimercaptopyrazolo(3,4-d)pyrimidine and reduced nonsequent bands is due probably to some spontaneous inactivation functional enzyme or cyanide-inactivated enzyme also provide during chromatography and also to some complex forming of further evidence for their similarity.
The finding that both ininactive enzyme to the gel. Spectral changes (Fig. 9) indicate active and cyanide-treated enzymes can be partially activated that reduced inactive enzyme slowly forms a complex with by incubation with Na2S provides conclusive evidence that the 4,6-dimercaptopyrazolo (3,4-d)pyrimidine.
Based on the asloss of a persulfide group is responsible for nonfunctionality in sumption that there are two active sites per enzyme molecule enzyme preparations containing the full complement of oxida-(for which no contrary evidence exists), one can visualize three tion-reduction components. forms of xanthine oxidase molecules: a species with both sites There are several pieces of evidence which indicate that the nonfunctional, a species with one functional and one nonfunc-persulfide is close to or associated with the enzyme-bound molybtional site, and a species with both sites functional.
The insensitivity of diaphorase activity to the AFRZSo molecules with 50% functionality could presumably be in both value indicates that FAD reduction by NADH is unimpaired inactive and active fractions and thus may contribute to the by destruction of the persulfide. The red shift of the 4,6-dimerdifficulty of getting a complete separation of functional from captopyrazolo(3,4-d)pyrimidine enzyme-bound molybdenum nonfunctional enzyme. complex ( Fig. 9) and its slow rate of formation in the absence of The only difference in chetiical composition which has been the persulfide is stronger evidence for the proximity Additional support from election paramagnetic resonance studies is the correlation of intensity of the slow molybdenum signal with enzymes of decreasing AFR*@ (Fig. 7). The cyanide-treated enzyme gives the strongest "slow" molybdenum signal. The extrapolated intensities indicate that in fully functional enzyme preparations, no "slow" molybdenum signal should be observed, in full agreement with the results of McGnrtoll et al. (6). Electron paramagnetic resonance studies of the anaerobic reduction of the enzyme upon cyanide treatment (8) indicat,e molybdenum to be the initial site of reduction.
The linear correlation of xanthine-oxygen reductase activity with degree of substrate bleaching (Fig. 5), intensity of iron electron paramagnetic resonance signals (Fig. 6) and the intensity of rapid molybdenum electron paramagnetic resonance signals (Fig. 7) are in agreement with the results of McGartoll et al. (6). The results indicate that only functional enzyme can be rapidly reduced by substrate while dithionite reduces both functional and nonfunctional enzyme. Of particular significance is the observation that sodium borohydride reduces only functional enzyme; the rapid reduction of 450.nm absorbance, intensity of iron electron paramagnetic resonance signals and intensit,y of "rapid" molybdenum signals parallel the results with santhine as a reductant.
This finding has mechanistic significance in that the borohydride ion reduces by hydride ion transfer.
The observation that borohydride, aldehydes, and purines reduce only functional enzyme indicates that the persulfide group is required to facilitate enzyme reduction (with the molybdenum being the initial acceptor of reducing equivalents). These data imply that reduction of xanthine oxidase by aldehydes or by purines might also occur by a hydride ion transfer. Dithionite titrations (Fi,. v 11) indicate that the persulfide group does not take up any reducing equivalents; hence its function is probably to facilitate the hydride ion transfer reaction.
In an attempt to explain these results, as well as to provide a model for future experimentation, we would like to propose the mechanism for substrate reduction of enzyme and hydroxylation as outlined in Fig. 12.
The tautomeric form of xanthine shown is that proposed by Bergmann et al. (21) as the structure acted on by xanthine oxidase. With its 6+ charge on the C-8 position, this tautomer would be in a suitable structure for attack by the persulfide. The reaction intermediate of reduced enzyme and the persulfidebound substrate would then be subject to displacement by hydroside ion, resulting in the regeneration of the persulfide and liberation of the hydroxylated product. This mechanism has the attractive feature of separating the hydroxylation reaction from the reoxidation of the reduced enzyme, and would, in fact, permit several molecules of substrate to be hydroxylated under anaerobic conditions at the expense of still further reduction of the enzyme. Indeed, in a previous publication, the anaerobic formation of 33 molecules of uric acid per functional active site, has been reported (5). The proposed mechanism is also consistent with the findings of Pick and Bray (22) and Bray et al. (23), who have demonstrated that the "rapid" molybdenum electron paramagnetic resonance signal shows interaction with a proton derived from the C-8 position of xanthine.
A mechanism involving hydride ion transfer from substrate to enzyme has also been suggested previously by Rajagopalan and Handler (24).
The proposed mechanism is attractive in that it can also account for the oxidation of aldehydes by addition of the persulfide to the electron-deficient carbonyl carbon. The observations that xanthine oxidase oxidizes preferentially unhydrated aldehydes (25) and that thiols add to unhydrated aldehydes rather than to the hydrated species (26), are also consistent with the proposed mechanism.
Unfortunately, the available knowledge of persulfide chemistry is limited and, to our knowledge, the reactions of persulfides with aldehydes and nitrogen heterocyclic compounds have not been studied.
Thiols are known to add to the carbonyl carbon of aldehydes and have recently been shown to catalyze the deuterium-hydrogen exchange at the C-5 position of uracil (27). More studies are required, however, before the chemical properties of persulfides can be compared to those of the sulfides.