The Conversion from the Dehydrogenase Type to the Oxidase Type of Rat Liver Xanthine Dehydrogenase by Modification of Cysteine Residues with Fluorodinitrobenzene*

When rat liver xanthine dehydrogenase was incubated with fluorodinitrobenzene (FDNB) at pH 8.5, the total enzyme activity decreased gradually to a limited value of initial activity with modification of two lysine residues in a similar way to the modification of bovine milk xanthine oxidase with FDNB (Nishino, T., Tsushima, K., Hille, R. and Massey, V. (1982)J. Biol. Chem. 257, 7348–7353). After modification with FDNB, the two peptides containing dinitrophenyl-lysine were isolated from the molybdopterin domain after proteolytic digestion and were identified as Lys754 and Lys771 by sequencing the peptides. During the modification of these lysine residues, xanthine dehydrogenase was found to be converted to an oxidase form in the early stage of incubation. Incorporation of the3H-dinitrophenyl group into enzyme cysteine residues was 0.96 mol per enzyme FAD for 68% conversion to the oxidase form. The modified enzyme was reconverted to the dehydrogenase form by incubation with dithiothreitol with concomitant release of3H-dinitrophenyl compounds. After modification with3H-FDNB followed by carboxymethylation under denaturating conditions, the enzyme was digested with proteases. Three3H-dinitrophenyl-labeled peptides were isolated and sequenced. The modified residues were identified to be Cys535, Cys992 and Cys1324. These residues are conserved among the all known mammalian enzymes, but Cys992 and Cys1324 are not conserved in the chicken enzyme. Cys1324 of the rat enzyme was found not to be involved in the conversion from the dehydrogenase to the oxidase by limited proteolysis experiments, but Cys535 and Cys992 which seemed to be modified alternatively with FDNB appear to be involved in the conversion.

exists as the NAD-dependent dehydrogenase type (XDH) in freshly prepared samples, i.e. it exhibits low xanthine-O 2 reductase activity but high xanthine-NAD reductase activity even in the presence of O 2 (3,4). But during extraction or purification procedures, the enzyme can be easily converted to an O 2 -dependent oxidase type (XO), i.e. it exhibits low xanthine-NAD activity but high xanthine-O 2 reductase activity. This conversion occurs reversibly by oxidation of sulfhydryl groups or irreversibly through proteolysis (3)(4)(5)(6)(7)(8)(9)(10). As H 2 O 2 and O 2 Ϫ are formed as products when molecular oxygen is used as an electron acceptor, the conversion from XDH to XO has been proposed as the basis of the mechanism of recirculation injury (11).
Although the enzyme is easily converted to XO, the enzyme can be purified as the dehydrogenase (XDH) form by rapid purification (6) or in the presence of thiol reagents such as mercaptoethanol or DTT (7,8), or as an XO form which can be converted to XDH by incubation with thiols (9). The differences in structural, spectroscopic and kinetics properties between XDH and XO obtained by analyses of the purified enzymes were reviewed recently (12,13). Upon modification of the protein molecule either by proteolysis or disulfide formation, significant conformational changes occur, particularly around the flavin (8,14,15). The enzyme from rat liver can be converted to XO by limited proteolysis, which results in nicking at two positions which were identified by sequencing the NH 2 termini of three isolated peptides (10). However, the identification of the cysteine residues whose oxidation is responsible for XDH to XO was not easy because of the involvement of many cysteine residues being modified to form disulfide bridges during the conversion by ordinary sulfhydryl modifiers. Waud and Rajagopalan (16) observed that as many as 14 cysteine residues in rat liver enzyme were titrated with DTNB during conversion from XDH to XO. More recently rat liver and bovine milk enzymes can be converted to XO by 4,4Ј-dithiodipyridine (4-PDS) as a modifier forming 4 disulfide bridges, i.e. eight cysteine residues were associated with this modification (8,17). It is likely that parts of cysteine residues are close enough to readily form disulfide bridges within the enzyme because cysteine disulfide bridges were formed by reaction with 4-PDS rather than single cysteine residues being modified (8,17). Although it has been reported that a single cysteine residue is involved in the conversion (18), the method employed in the report is not appropriate for the determination of the number of residues in chemical modification studies, as discussed by Rakitzis (19).
In this work, the cysteine residues responsible for XDH to XO interconversion were determined by chemical modification using fluorodinitrobenzene (FDNB). Chemical modification of bovine milk XO by FDNB was reported previously to show modification of two lysine residues with a 6-fold decrease in xanthine:O 2 oxidoreductase activity (20). XO of rat liver, re-* This work was supported by a Grant-in-Aid (08249104) for Science Research on Priority Area "Molecular Biometallics" and a Grant-in-Aid (09480167) for Science Research from the Ministry of Education, Science, Sports and Culture of Japan. 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.

EXPERIMENTAL PROCEDURES
Material and Methods-Rat liver xanthine dehydrogenase was purified from livers of 8 -10-week old rats, by the methods previously reported (21). The ratio of absorbance at 280 nm to that at 450 nm of the purified enzyme was around 5.2. The enzyme activity was measured according to the method reported previously (9). XDH activity was determined following absorbance change at 340 nm, and XO activity was at 295 nm without NAD ϩ . The activity-to-flavin ratio at 25°C (AFR 25 ) was obtained by dividing the change in absorbance/minute at 295 nm in the presence of NAD under aerobic conditions by the absorbance at 450 nm of the enzyme used in the assay. The D/O ratio as defined by Waud and Rajagopalan (6) was determined as the ratio of the absorbance change at 295 nm under aerobic conditions in the presence of NAD to that in the absence of NAD. The enzyme concentration was determined spectrophotometrically using a value of 35,800 M Ϫ1 cm Ϫ1 for the molar absorbance/enzyme-bound FAD (6). The freshly purified enzyme was converted to the XDH form, with a D/O ratio of 9 -10, by incubation with 5 mM DTT. The completely reversible XO form was prepared by treatment with 4-PDS according to the methods described previously (9,17). 3 H-FDNB was obtained from New England Nuclear Co. and was diluted with unlabeled FDNB to a specific activity of 259,000 dpm/nmol. Molar extinctions at 360 nm of 17,700 M Ϫ1 cm Ϫ1 for DNP-lysine and of 17,500 M Ϫ1 cm Ϫ1 for average DNP-amino acids were used (22). A molar extinction of 16,000 M Ϫ1 cm Ϫ1 at 340 nm was used for DNP-cysteine (22) before and after DTT treatment.
Chemical Modification of Rat Liver XDH or XO with FDNB, and Purification and Sequence Analysis of Peptides Containing Labeled DNP-Amino Acids-XDH was prepared as described above and the first reaction was carried out with unlabeled 700 M FDNB in 50 mM potassium phosphate buffer, pH 7.8, for 10 min at 25°C to react with nonspecific residues. The reaction was then quenched by cooling the solution quickly to 0°C on ice followed by immediate passing through a Sephadex G-25 column at 4°C. The second reaction was carried out with 700 M 3 H-FDNB in 0.1 M pyrophosphate buffer pH 8.5 for strictly 10 min to convert XDH to XO. The labeled XDH was carboxymethylated under anaerobic conditions without prior DTT reduction and was then digested with a mixture of V8 protease Staphylococcus aureus (Boehringer Mannheim) and TPCK-treated trypsin (Sigma) for 3 h at 37°C in 0.1 M ammonium bicarbonate buffer, pH 7.8. The peptides were purified by using three different C 18 reverse phase HPLC columns from Capcellpak (Shiseido), Microsorb (Rainin), and Bondapak (Waters). Peptide detection was performed photometrically at 360 nm for specific absorbance of DNP-amino acids or at 226 nm.
XO, which was prepared either by spontaneous oxidation or by 4-PDS treatment (17), was incubated with various concentrations of FDNB in 0.1 M pyrophosphate buffer, pH 8.5, at 25°C. XO modified with FDNB was partially digested with trypsin and was separated into three fragments under denatured conditions by the method described previously (10).
The finally purified labeled peptides were subjected to amino acid sequence analysis using an Applied-Biosystem gas-phase amino acid sequencer, model 477A, equipped with PTH analyzer 120A. Incorporation of radioactivity into the amino acid residue was confirmed by determination of radioactivity of the aliquots of each cycles of PTHamino acids.
Partial Digestion of Native XDH with Endoproteinase Glu-C from Staphylococus aureus V 8 (Protease V 8 )-The partial digestion of native XDH with Protease V 8 was carried out in 50 mM ammonium bicarbonate buffer, pH 7.8, containing 0.4 mM EDTA with or without 10 mM DTT at 25°C with the ratio of substrate to enzyme of 25 to 1 (weight/weight). Aliquots were withdrawn at 2, 4, or 20 h of incubation, and activities were determined. To measure the intrinsic XDH activity, after proteolysis the enzyme was incubated with 10 mM DTT overnight to reduce disulfide bonds that may have been formed by auto-oxidation during partial digestion. The samples withdrawn at 4 and 20 h were subjected to ultrafiltration using Centricon 10 microconcentrators, and a part of the filtrate was applied to a C 18 reverse phase column equilibrated with 20% acetonitrile containing 0.1% trifluroacetic acid at a flow rate of 0.5 ml/min, which was monitored at 226 nm. The amino acid sequence of every peak was determined.

Xanthine Dehydrogenase Can Be Converted to an Oxidase by
Modification with FDNB-When native XDH having a D/O ratio between 9 and 10 was incubated with FDNB at pH 8.5, the total enzyme activity was decreased gradually to a limiting value of the initial activity as shown in Fig. 1. This type of inactivation by prolonged incubation of enzyme with FDNB was also observed with rat liver XO as well as bovine milk XO (20) and is described below. During short time incubation with FDNB, however, XDH was found to be converted to XO, with 80% conversion within 10 min at pH 8.5. As the enzyme was not converted to XO without FDNB under these conditions and the conversion rate was much faster than the decrease of total activity, it was suggested that this conversion might be due to rapid modification of cysteine residues independently of modification of the residues responsible for irreversible inactivation of both enzyme forms by prolonged incubation with FDNB. When aliquots of enzyme, which had been converted to various degrees of XO activity, were incubated with 10 mM DTT overnight, it was found that while total activity declined steadily with time of treatment with FDNB, 80 -90% of the residual activity was present as XDH after DTT treatment, indicating that the modification is reversible.
Although rapid conversion from XDH to XO suggests that the conversion is not due to spontaneous formation of a disulfide bridge but to modification of cysteine residues, it is still possible that the FDNB-modified cysteine residue might be attacked by another cysteine residue to result on the formation of disulfide bridge. Such cases are known to occur in the modification of rat liver (17) or bovine milk XDH (8) with 4-PDS. To check this possibility, the reconversion of the FDNB-modified as well as the 4-PDS-modified XDH was performed by incubation with DTT. The XO enzyme modified by 4-PDS was rapidly reconverted to XDH with 5 mM DTT and the conversion was almost complete within 30 min (data not shown). On the other hand, it took several hours for the XO enzyme modified by FDNB to reconvert to 80% of XDH with 10 mM DTT (data not shown).
Incorporation of 3   M FDNB at 25°C. The total activity, which is shown as percentage of the original activity remaining at any given time, was determined photometrically by following the absorbance change at 295 nm with NAD (total activity; q) or the oxidase activity, which is shown as percentage of the total activity remaining, was determined without NAD (type-O activity; E). As a control, the enzyme was also incubated without FDNB, and XO activity was followed (Ⅺ). After DTT treatment of each FDNB modified enzyme, XO activity expressed as percentage of the total activity remaining was determined at 295 nm without NAD (Ç).
was not rapid at pH 7.8 (data not shown), to identify the responsible cysteine residues for the D-O conversion, subsequent reactions were performed in two steps to minimize nonspecific modifications. In the first step reaction, the XDH (86.4% of dehydrogenase activity, and 13.6% of oxidase activity) was incubated with 700 M unlabeled FDNB at 25°C for 10 min at pH 7.8 in potassium phosphate buffer followed by gel filtration using Sephadex G-25 to remove excess unlabeled FDNB, as described under "Experimental Procedures." After the first step reaction, the AFR value decreased from 100 to 86, suggesting some incorporation of DNP group into lysine residues. However, most of the enzyme was retained as XDH (75% dehydrogenase activity, 25% oxidase activity). During this modification, unlabeled nonspecific DNP-Cys and DNP-Lys residues were estimated to be 2.3 and 0.3 mol/mol of enzyme FAD, respectively, using absorbance at 360 nm and 340 nm before and after DTT treatment. In the second step reaction, the sample was reacted with 3 H-FDNB at 25°C for 10 min at pH 8.5 in pyrophosphate buffer. After the second reaction, the enzyme showed only 7.5% of dehydrogenase activity, indicating that XDH was almost completely converted to XO during the second step reaction. The spectra of the DNP incorporated enzymes after first and second reaction are shown in Fig. 2. Incorporation of total 3 H-DNP compound into the enzyme was 2.25 mol/mol of enzyme FAD, of which 0.92 mol was released upon DTT treatment in concomitant re-conversion from XO to XDH, and was therefore considered to be incorporated into cysteine residue(s). The AFR value decreased from 86 to 52 during the second reaction. This is due to the further modification of lysine residues, consistent with the existence of some DNP compound (estimated to be 1.33 mol/mol of enzyme) remaining with the enzyme after DTT treatment. Incorporation of 0.92 mol of label/mol of enzyme FAD into cysteine residue(s) during 67% XDH to XO conversion corresponds to 1.36 mol of DNP/mol of enzyme FAD for 100% conversion.
Purification, Sequencing of 3 H-DNP Peptides and Sequence Comparison-6 nmol of enzyme prepared as above was carboxymethylated without DTT reduction to avoid possible transfer of the DNP group to other sulfhydryl residues. After dialysis, the sample was digested for 3 h at 37°C in 0.1 M ammonium bicarbonate buffer pH 7.8 with a mixture of V 8 protease and trypsin. The recovery of radioactivity in this step was 86%. All digested samples were purified by using the three kinds of reverse phase HPLC column as described under "Experimental Procedures." The recovery of radioactivity in each step of HPLC chromatography was about 90%, and the overall recovery of radioactivity that was releasable with DTT treatment was 65%. In the final steps of HPLC chromatography, the peptides were separated into three main peaks where 90% of the total radioactivity was associated. All radioactive purified peptides were subjected to gas phase protein sequencing and the obtained sequences were as follows: 1) Asp-Met-3 H-DNP-Cys-Gly-Lys, 2) Asn-3 H-DNP-Cys-Trp-Lys, and 3) Asn-3 H-DNP-Cys-Lys.
As was described in the previous section, 1.36 mol of DNP/ mol of enzyme FAD was incorporated if 100% D-O conversion was assumed to occur. But the radioactivity incorporation into these peptides was approximately equal among these three peptides; the ratio of radioactivities of the three peptides was 1.1/1.4/1, corresponding to an estimated amount of modification of each cysteine of 0.43, 0.54, and 0.39 mol/FAD, respectively. The explanation for these numbers of modification will be considered under "Discussion." From comparison with the complete amino acid sequence of the rat enzyme deduced previously from cDNA cloning (10), these peptides were identified as containing Cys 535 , Cys 992 , and Cys 1324 . As the mammalian enzyme is known to be converted from XDH to XO and whereas XDHs from other sources such as chicken and probably Drosophila are not, it is noteworthy to see the sequences alignment around these cysteine residues from various sources (23-34) (Fig. 3, panel a-c). It is found that residues corresponding to both Cys 535 and Cys 992 are conserved in all mammalian XDHs, whereas the residue corresponding to Cys 992 is replaced by arginine in chicken XDH. The regions around Cys 535 are relatively not well conserved among the enzymes from various species, although others are well conserved. In chicken XDH, the corresponding residue to Cys 535 is conserved but is followed by glutamic acid, whereas it is followed by glycine in mammalian XDHs. The residue is not conserved in XDHs from other sources than vertebrae, where aspartic residues exist in the region near the residue corresponding to Cys 535 of the rat enzyme. It is intriguing to note that all of these residues of aldehyde oxidase are replaced by tyrosine residues, with exception of aldehyde oxidoreductase from Desulfovibrio gigas, where peptide sequences corresponding to those of the modified residues in rat XDH are missing.
Evidence for That Cys 1324 Was Not Responsible for XDH/XO Conversion-During the course of FDNB modification and digestion experiments, it was noticed that protease V 8 showed very different patterns of partial digestion of XDH, compared with that of trypsin. Tryptic digestion resulted in proteolysis at two positions resulting in the formation of 20-, 40-, and 85-kDa fragments, with concomitant conversion from XDH to XO (10). If this tryptic digestion was carried out in the presence of 5 mM DTT under the same conditions, however, XDH retaining 82% initial activity without conversion into XO was obtained with formation of only 20-and 125-kDa fragments as analyzed by SDS-PAGE (Fig. 4). This indicates that proteolysis at the interconnecting segment between 20 and 125 kDa was not responsible for the irreversible conversion, but the segment between 40 and 80 kDa was.
By protease V8 treatment, most of the enzyme remained with apparently the same size as the native one, and the enzyme was partly digested by prolonged incubation at 25°C resulting in only a small amount of formation of 85-and 62-kDa fragments, which were detected by SDS-PAGE (Fig. 4). The activity (initial FIG. 2. The absorption spectra of DNP incorporated rat liver XDH. The spectra were recorded after reaction of rat liver XDH with FDNB followed by gel filtration with Sephadex G-25 equilibrated with potassium phosphate buffer pH 7.8. See detailed in text under "Results." 1, native XDH before modification; 2, the FDNB modified enzyme after the first reaction at pH 7.8; 3, the FDNB modified enzyme after the second reaction at pH 8.5; 4 (Ç-Ç), the difference spectrum between 2 and 1; 5 (q-q), the difference spectrum between 3 and 2.
D/O ϭ 9) remained mostly (60 -70%) as XDH after V 8 digestion with or without 5 to 10 mM DTT, with apparently no effect of DTT itself on V 8 digestion. The lower XDH activity compared with the control enzyme (80 -90%), which was incubated without V8 protease, might be due to minor nicking at the segment between 40-and 85-kDa domains. During digestion, aliquots were withdrawn after 4 and 20 h and applied to Centricon 10 ultrafiltrators to separate the small fragments of digested peptides that could be missed in SDS-PAGE analysis. From each filtrate, almost the same amount of a single peptide was obtained, judged from the elution pattern of HPLC (data not shown), indicating that this segment was easily hydrolyzed by V 8 protease. The peptide at each incubation time point was sequenced and was found to be the same, composed of the nine amino acid residues of the C-terminal peptide of XDH, containing Cys 1324 . (Asn 1323 -Cys-Lys-Ser-Try-Ser-Val-Arg-Ile 1331 ). Amino acids analysis indicated that the amount of peptides was nearly stoichiometric to the original enzyme, and the amino acid composition was coincident to the C-terminal nineamino acid peptide (data not shown). The sample of 60 -70% of XDH, which had lost its C-terminal peptide of nine amino acids, was again converted to XO when it was kept at 4°C for several weeks without DTT under aerobic conditions. These results indicate that the cysteine residue of the C-terminal peptide is not involved in the conversion of XDH to XO.
The Modification of XO with FDNB-As in the case of bovine milk XO, the oxidase type of rat liver xanthine dehydrogenase, which was prepared by the methods described under "Experimental Procedures," was inactivated to a limited value by incubation with various concentrations of FDNB. As shown in the inset of Fig. 5, the enzyme activity decreased at a rate dependent on the FDNB concentration down to a level around 30% of the initial activity. The fact that this remaining activity is due to the modified enzyme, and is not due to 30% of remaining native enzyme, was confirmed by the lack of any further decrease when more FDNB was added to the reaction mixture after the limiting activity level was reached (data not shown). The incorporation of FDNB into the enzyme was approximately 3.5 mol/enzyme FAD with 70% loss of activity calculated using an average molar extinction of 17,500 M Ϫ1 cm Ϫ1 of DNP amino acids. When inactivated enzyme was treated with DTT, some amount of DNP-compounds was released without reactivation of total catalytic activity; inactivated enzyme was still associated with 2-2.4 mol of DNP/mol of enzyme FAD. These results suggested that the FDNB modification responsible for the loss of activity is not likely to be due to modification of cysteine, histidine, or tyrosine residues, but due to modification of lysine residue(s) (36). These inactivation modes and changes of absorption spectrum of rat enzyme by modification with FDNB were quite similar to those of inactivated DNP-milk XO reported previously (20). In the bovine milk XO, it was shown by rapid reaction kinetics that the reduction rate of DNP-enzyme by xanthine was slower than that of native enzyme, indicating that the xanthine binding site was modified by FDNB (20). Furthermore, two lysine residues near the xanthine binding  (panels a-c) and lysine (panel d) residues of rat XDH with the sequences of other XDHs or aldehyde oxidases from various sources. Asterisks indicate the residues modified with FDNB. Black, the residue conserved in most of the enzymes; gray, the residue highly conserved in the enzymes. The sequences are the following: XDH r, rat liver XDH (10); XDH h, human liver XDH (23); XDH m, mouse liver XDH (24); XDH b, bovine milk XDH (25); XDH c, chicken liver XDH (26); XDH dm, Drosophila melanogaster XDH (27,28); XDH dp, Drosophila pseudoobscura XDH (29); XDH cv, Calliphora vicina XDH (30); XDH an, Aspergillus nidulans XDH (31); AOX b, bovine aldehyde oxidase (32); AOX h, human aldehyde oxidase (33); AOR dg, aldehyde oxidoreductase from Desulfovibrio gigas (34).  1 and 7).
site were suggested to be modified during incubation.
To identify the modified residues in the rat enzyme, isolation of peptide fragments modified with FDNB were performed after digestion of the modified enzyme by trypsin. Purified rat liver XO was incubated with 700 M of FDNB in 0.1 M pyrophosphate buffer pH 8.5 containing 0.4 mM EDTA at 25°C for 30 min. The reaction was stopped by cooling on ice and immediately was subjected to gel filtration on a Sephadex G-25 column to remove excess FDNB. The eluted DNP-enzyme still retained 36% residual activity and was incubated with 10 mM DTT overnight. After DTT treatment, the number of DNPresidues associated with the protein decreased from 3.4 to 2.5 mol/mol of enzyme FAD and the 36% residual activity was unchanged. The DNP-enzyme was digested by trypsin at 30°C in 50 mM potassium phosphate buffer, pH 7.8, overnight to produce nicked DNP-enzyme (2.12 mol of DNP/mol of E-FAD). The nicked DNP-XDH was subjected to gel filtration after denaturation using HPLC column and found to be dissociated into three fragments as described previously with native rat liver enzyme (10). The elution pattern of gel filtration using HPLC (10) is shown in Fig. 6 and recoveries of the peptides are summarized in Table I. Peak 1, which is considered to be an incompletely digested fragment, has only 10% retained DNP compounds as judged by the 360 nm absorbance. The peptide fragment of 85 kDa contained 76% of the total absorbance at 360 nm of all fragments. As the 85-kDa fragment is the molybdopterin-containing domain (13), this result is consistent with the results obtained from the experiments with bovine milk XO, that FDNB reacts with the residues near to the xanthine binding site. The spectrum of the 85-kDa fragment is typical of a DNP-peptide with absorption maximum at 360 nm. The 85-kDa fragment was further digested with trypsin and V 8 protease and DNP-peptides were purified using Capcellpak and Bondapak of C 18 reverse-phase HPLC columns and were sequenced. The two DNP-peptides were isolated and the sequences were as follows: 1) Thr 746 -Asn-Cys-Thr-Ile-Ala-Val-Pro-DNP-Lys 754 -Gly-Glu 756 ,2) Leu 762 -Phe-Val-Ser-Thr-Gln-Asn-Thr-Met-DNP-Lys 771 -Thr 772 .
DNP-Lys was identified as authentic PTH-DNP-Lys by protein sequencing. The sequences of these DNP-peptides belong to that of the largest fragment of 85 kDa (10), and the residues are also conserved in the sequence of the bovine milk enzyme. These residues are well conserved in all known mammalian XOs (Fig. 3, panel d). As the mode of inhibition of the bovine milk enzyme was very similar to that of rat enzyme, the same lysine residues in milk enzyme could well be modified by FDNB.

DISCUSSION
In this study we modified cysteine residues using FDNB, a less specific thiol-modifying reagent. XDH pretreated with unlabeled FDNB at lower pH was modified with 3 H-DTNB at pH 8.5, with modification of 1.36 mol cys 3 H-DNP/FAD accompanying conversion of XDH to XO. It is unlikely that FDNB modification was due to disulfide bridge formation as in the case of 4-PDS (8,17) but was due to a single cysteine modification since the rate of reactivation of modified enzyme by DTT was significantly slower than the XO formed by reaction with 4-PDS. The lack of disulfide formation in this modification may be due to lower reactivity of dinitrophenyl-cysteine than dithiopyridyl-cysteine toward other cysteine residues that may exist near the modified cysteine residue. Radioactivity was incorporated into the enzyme upon modification with concomitant conversion from XDH to XO, and furthermore, the incorporated 3 H-DNP was slowly released by DTT treatment with reconversion to XDH. Although only 1.36 mol DNP-residues were incorporated and distributed between three cysteine residues, some After incubation of XO with FDNB(30% residual activity), the sample was digested by trypsin at 30°C in 50 mM potassium phosphate buffer at pH 7.8 overnight and then carboxymethylated as described under "Experimental Procedures." The nicked DNP-enzyme was applied on the combined columns of TSK-SW 3000 and 4000 in series. The spectrum of the DNP-peptide was obtained after elution from HPLC columns.

TABLE I
The recovery of the fragment containing modified amino acid residues by FDNB in 3 fragments of XO The molecular weights of the eluted peptide samples shown in Fig. 6 were estimated by SDS-PAGE using the standard proteins. The recovery of peptides from gel filtration was calculated by absorbance at 280 nm, and the recovery of DNP-residues content in each of the peptides was estimated by absorbance at 360 nm. Non-involvement of Cys 1324 for the conversion has been also confirmed by site-directed mutagenesis experiments using Baculovirus/insect cell system where Cys 1324 was replaced by serine. 2 The explanation for the fact that only one mol/FAD incorporation into two different cysteine residues might be that one of the two cysteine residues, which may form a disulfide bridge during reversible conversion of XDH to XO, is alternatively modified so that dinitrophenylation of either of the residues may disturb the further modification of the other cysteine residue by steric hindrance. Thus, it is likely that Cys 535 and/or Cys 992 are responsible for the conversion from XDH to XO. We have attempted to address this question by replacement of these residues with serine, but expression of the mutant enzymes has so far been unsuccessful. The mammalian enzyme is known to be converted from XDH to XO (12,13). Although it is not well characterized in XDHs from other sources, it is known that at least XDHs from chicken and probably Drosophila are not converted to XO. It is intriguing that the residues corresponding to both Cys 535 and Cys 992 are all conserved in mammalian XDHs (10,(23)(24)(25), whereas the residue corresponding to Cys 992 is replaced by arginine in chicken XDH (26). In the chicken XDH, the residue corresponding to Cys 535 is followed by glutamic acid (26) whereas it is followed by glycine in mammalian XDHs (10,(23)(24)(25). The lack of spontaneous formation of disulfide bridges in chicken XDH might be explained by such a replacement. The differences in surrounding residues might also explain that the residue corresponding to Cys 535 in chicken XDH cannot be easily modified by sulfhydryl modifying reagents. Most probably the conformational change might occur particularly around the FAD moiety in the XDH to XO conversion by modification of these residues (8,13,14). The residue corresponding to Cys 535 is not conserved in XDHs from sources other than vertebrae, suggesting that these enzymes might not be converted into XO forms by sulfhydryl oxidation.
The Cys 535 residue is in the NAD (37) and FAD-binding domain (13), whereas Cys 992 and Cys 1324 are in the molybdopterin domain (10). These residues could be located at the interface of these domains close enough to form a disulfide bridge resulting in the conformational changes. The only available three dimensional structure of enzymes of the xanthine oxidase family is at the moment the structure of aldehyde oxidoreductase from Desulfovibrio gigas (38). As this enzyme lacks an FAD domain but possesses molybdopterin and iron-sulfur center domains (34), the residue corresponding to Cys 535 is missing. The residue corresponding to Cys 992 in this enzyme is also missing (34), but the residues corresponding to the neighboring residues of Cys 992 such as Thr 610 , His 611 or Lys 612 are actually located at the surface of the molybdopterin domain (38). These residues might be supposed to be at the interface to the FAD domain of XDH. In other aldehyde oxidases the discussed cysteine residues are replaced by tyrosine residues. All of these enzyme are known to exist solely as an oxidase form.
As with bovine milk XO, rat liver XDH was also inactivated to a limited value by rather prolonged incubation with various concentrations of FDNB. These inactivation modes and changes of absorption spectrum of the rat enzyme by modification with FDNB were quite similar to those of inactivated DNP-milk XO reported previously (20). In the bovine milk XO it was shown by rapid reaction kinetics that the reductive half reaction of the DNP-enzyme by xanthine was slower than that of native enzyme, indicating that the residues near the xanthine binding site were modified (20). In the present studies the modified residues of the rat XO form were identified by isolation of peptide fragments modified with FDNB. The residues were actually in the 85-kDa domain which contains the molybdopterin cofactor (10). As the mode of inhibition of rat XDH was very similar to that of bovine milk XO, the same residues would presumably be modified in the milk enzyme. These residues are conserved among all of the known mammalian XOs (10,(23)(24)(25). Although they are not conserved in aldehyde oxidoreductase from Desulfovibrio gigas (34), the presumed location of these residues in the three-dimensional structure suggests that they are directly associated neither with the molybdopterin cofactor nor with the possible active site pocket (38), suggesting that the inhibition is due to a conformational change around the xanthine binding site. Incomplete inactivation with FDNB is consistent with this conclusion.