Structural and Conformational Analysis of the Oxidase to Dehydrogenase Conversion of Xanthine Oxidoreductase*

Xanthine oxidoreductase (XOR) is a 300-kDa ho-modimer that can exist as an NAD (cid:1) -dependent dehydrogenase (XD) or as an O 2 -dependent oxidase (XO) depend- ing on the oxidation state of its cysteine thiols. Both XD and XO undergo limited cleavage by chymotrypsin and trypsin. Trypsin selectively cleaved both enzyme forms at Lys 184 , while chymotrypsin cleaved XD primarily at Met 181 but cleaved XO at Met 181 and at Phe 560 . Chymotrypsin, but not trypsin, cleavage also prevented the reductive conversion of XO to XD; thus the region surrounding Phe 560 appears to be important in the interconversion of the two forms. Size exclusion chromatography showed that disulfide bond formation reduced the hydrodynamic volume of the enzyme, and two-di-mensional gel electrophoresis of chymotrypsin-digested XO showed significant, disulfide bond-mediated, conformational heterogeneity in the N-terminal third of the enzyme but no evidence of disulfide bonds between the N-terminal and C-terminal regions or between XOR subunits. These results indicate that intrasubunit disulfide bond formation leads to a global conformational change in XOR that results in the exposure of the region surrounding Phe 560 . Conformational changes within this region in turn appear to play a critical role in the interconversion between the XD

Xanthine oxidoreductase (XOR) 1 plays an important function in vertebrate metabolism by catalyzing the oxidation of xanthine to uric acid, the rate-limiting step in purine degradation (1). Native XOR is a 300,000-dalton dimer composed of identical and catalytically independent subunits. Each subunit consist of 1333-1358 amino acids, depending on the species, and contains binding sites for molybdopterin, iron, and flavin co-factors (2, 3). The genes and full-length coding regions for XOR from mammals, chickens, insects, and fungi have been described (4 -10). Analysis of these cDNAs has shown that the amino acid sequence of XOR is highly conserved across the phylogenetic spectrum. The enzymes from rats, mice, humans, and cattle exhibit ϳ90% identity, while the chicken and Drosophila enzymes exhibit, respectively, 70 and 52% identity with mammalian enzymes (3).
In mammals, XOR can exist in two enzymatic forms: a dehydrogenase (XD; EC 1.1.3.204) which utilizes NAD ϩ as an electron acceptor and an oxidase (XO; EC 1.1.3.22) which utilizes O 2 as an electron acceptor (11,12). Both enzymatic forms are the product of the same gene and are identical in size, subunit composition, and co-factor requirements (2, 3). To date, XD appears to be the predominant enzyme form in freshly prepared mammalian tissues (11)(12)(13); however, it is often isolated in the XO form during purification (2). It is unclear if XD and XO serve distinct biological roles. High levels of the XO have been associated with tissue injury and certain diseases (14,15) and are thought to contribute to oxidative damage of cells through the generation of cytotoxic oxygen metabolites (H 2 O 2 and O 2 . ) (14). The XD form, on the other hand, may be an important component in the defense against oxygen radical damage through its role in the synthesis of uric acid, a potent antioxidant (16). The cellular mechanism(s) that regulate the relative tissue levels of the XD and XO forms of XOR are poorly defined. XD can be converted to XO in a reversible process by heating or oxidation of cysteine thiols to form disulfide bonds (11,17). The oxidative conversion of XD to XO has been shown to be associated with the oxidation of selected cysteines (17), loss of NAD ϩ binding affinity (3), alterations in redox and kinetic properties (18,19), and conformational changes at the flavin-binding site (20,21). However, specific protein domains involved in these changes have not been identified, and the underlying molecular basis of the redox-mediated XD to XO conversion is still uncertain. In addition, it is unclear to what extent conformational changes within the flavin-binding site reflect larger changes in the global conformation of XOR and whether such changes are necessary for the conversion of XD to XO. XD can also be irreversibly converted to XO by proteolysis (6,11,17). Recent crystallographic studies have shown that proteolysis of XD leads to structural alterations near the flavin cofactor domain (22). However, the crystallographic structure of intact XO generated by cysteine oxidation has not been solved, and thus it is unclear how similar the structure of disulfide-bonded XO is to proteolyzed XO. In the present study we have used peptide mapping, chromatography, and two-dimensional gel electrophoresis to identify differences in the solution conformations of XD and XO and to investigate the role conformation changes play in the XD to XO conversion.

EXPERIMENTAL PROCEDURES
Materials-XOR (2.5-3.1 IU/mg) isolated from bovine milk in the oxidase form (XO) according to Waud et al. (23) was a generous gift from Dr. Joseph McCord (Webb-Waring Antioxidant Research Institute, University of Colorado Health Sciences Center). Sequencing grade trypsin and chymotrypsin were purchased from Boehringer-Ingleheim Inc. BS3 (bis(sulfosuccinimidyl) suberate) was purchased from Pierce. Polyvinylidene difluoride membranes (Hyperbond) were purchased from Beckman Instruments Inc. Centricon ultrafiltration membranes were purchased from Amicon Inc. Other chemicals and reagents were obtained from either Sigma or Fischer Scientific.
Preparation of XD and Determination of XD and XO Isoforms-Purified XO was converted to XD by incubation with 5-10 mM DTT at room temperature for 1-2 h (17). The enzyme was then cooled to 4°C and subjected to ultrafiltration at 4°C using Centricon 10,000-Da cutoff membranes to remove DTT. Enzymatic activity specifically associated with either XD (xanthine:NAD ϩ oxidoreductase activity) or XO (xanthine:O 2 oxidoreductase activity) was calculated as the allopurinolsensitive rates of aerobic formation of uric acid from xanthine in the presence (XD) or absence (XO) of 0.6 mM NAD ϩ , and the ratio of XD to XO activities (D/O) was calculated according to Waud and Rajagopolan (12) as described previously (24). Total XOR activity (both XD and XO activities) was routinely assayed by the reduction of dichloroindophenol using xanthine as a substrate (xanthine/dichloroindophenol assay) as described previously (24). Total protein was quantified using the bicinchoninic acid method (25).
Protease Digestion-XOR (5-20 g) in the XD or XO form was incubated at room temperature (25°C) with 1/20 (w/w) chymotrypsin or trypsin in 50 mM Tris, 150 mM NaCl, pH 8, for 90 min. The reaction was stopped by the addition of phenylmethylsulfonyl fluoride. The samples were then assayed for enzyme activity and/or processed for SDS-PAGE analysis as described below.
Electrophoretic Analyses in One and Two Dimensions-Electrophoresis was performed using a mini-Protean II gel apparatus (Bio-Rad) with 0.75-mm spacers. For one-dimensional electrophoresis samples were diluted with an equal volumes of 2ϫ concentrated SDS-PAGE sample buffer (0.125 M Tris, 4% SDS, 20% glycerol, 10% 2-mercaptoethanol, pH 6.5), heated for 3 min at 95°C, and separated by electrophoresis in 10% acrylamide gels containing 0.1% SDS at room temperature. For twodimensional electrophoresis samples were diluted with equal volumes of 2ϫ concentrated SDS-PAGE sample buffer without 2-mercaptoethanol, heated for 3 min at 95°C, and subjected to electrophoresis in 10% acrylamide gels as described above. When the tracking dye reached the bottom of the gel, electrophoresis was stopped and the protein-containing lanes were excised and prepared for electrophoresis in the second dimension by incubation in 3 ml of SDS-PAGE sample buffer containing 5% 2-mercaptoethanol for 1 h at 60°C. After reduction the lanes were layered on top of the second dimension gel (10% acrylamide, 0.1% SDS) perpendicular to the direction of electrophoresis. Gaps were sealed with melted agarose (1%) containing 2% SDS, 50 mM DTT. After electrophoresis protein bands were detected by silver staining (26).
Chemical Cross-linking Studies-XO was cross-linked by reaction with a 50-fold molar excess of BS3 in phosphate-buffered saline, pH 7.5, according to the manufacturer's instructions.
Protein Sequence Determination-N-terminal amino acid sequence analyses were performed by automated Edman degradation of proteins electroblotted to polyvinylidene difluoride (27) using an Applied Biosystems Inc. model 477A automated protein sequencing instrument equipped with an in-line model 120 HPLC.
Size Exclusion-HPLC Analysis-HLPC-size exclusion chromatography (HPLC-SEC) analysis was performed on a Beckman System Gold Instrument using a 7.8 ϫ 600-mm SEC-3000 column (Phenomenex Inc.) equilibrated in phosphate-buffered saline, pH 7.5. Chromatography was performed at room temperature at a flow rate of 1 ml/min. The elution of protein and enzyme activity were monitored at 220 nm using an online UV detector and (where indicated by collecting fractions) using an electronically activated fraction collector (Gilson, FC 203B) that was synchronously integrated to sample injection. Determination of XD and XO retention times was performed using a 25-l injection loop and sample protein concentrations of 1 mg/ml. XO was converted to XD by incubation with 10 mM DTT at room temperature for at least 1 h prior to analysis. Mean retention times of each enzyme form were determined from a minimum of five HPLC-SEC analyses. The SEC properties of cross-linked XO were determined using a 100-l injection loop. Eluted protein was collected into 0.3-ml fractions. XO activity was determined using the xanthine/dichloroindophenol assay. The electrophoretic characteristics of the cross-linked protein were determined by SDS-PAGE analysis and silver staining as described above.

XD and XO Have Different Proteolytic Digestion Patterns-
Differences in the environment surrounding the flavin co-factor binding regions of XD and XO have been reported previously (20,21). To determine whether these differences reflect broader structural changes we investigated the trypsin and chymotrypsin digestion patterns of both catalytic forms (Fig. 1). Chymotrypsin cleavage of XO (D/O ϭ 0.1; 90% XO form) generated three major cleavage products with estimated molecular weights on SDS-PAGE of 85,000 (band 1), 64,000 (band 2), and 44,000 fragments (band 3). In contrast, chymotrypsin cleavage of XD (D/O ϭ 7.1; 88% XD form) yielded a single major product with an estimated molecular weight of 125,000 (band 5) and minor amounts of a 44,000 fragment (band 6). Trypsin cleavage of either isozyme of XOR generated a single 125-kDa fragment (Fig. 1B, bands 7 and 8). In agreement with earlier studies (6,11,17) neither chymotrypsin nor trypsin digestion affected total XOR activity or altered the apparent molecular weights of the isozymes (data not shown).
The identities of the cleavage sites were determined by Nterminal sequencing and comparing the experimentally deter- mined sequences to the deduced sequence of bovine XD cDNA (7). Table I shows the sequences of the first 7 amino acids of the major XD and XO cleavage products, the position of these sequences within the enzyme, and the corresponding cleavage site. The N-terminal sequences of the 125-kDa fragments (bands 7 and 8) generated by trypsin digestion of XD or XO correspond to Lys 185 -Leu 191 and are consistent with cleavage of both enzyme forms at Lys 184 . The sequence of the 125-kDa fragment (band 5) generated by chymotrypsin digestion of XD corresponds to Asn 182 -Thr 188 and is consistent with cleavage at Met 181 . Based on the deduced amino acid sequence of bovine XD the calculated molecular mass values for peptide fragments from Asn 182 to Val 1331 (the C terminus of bovine XD) and Lys 185 -Val 1331 are 127,225 and 126,854 kDa, respectively (28). These values are in good agreement with SDS-PAGE size estimates (ϳ125 kDa) for the major fragments generated by trypsin digestion of XD or XO and chymotrypsin digestion of XD, and indicate that these fragments contain the C terminus of XOR.
The N-terminal sequence of the 85-kDa fragment (band 1) generated by chymotrypsin digestion of XO corresponds to Gln 561 -Gln 567 and is consistent with cleavage at Phe 560 . The sequence of the 64-kDa fragment (band 2) corresponds to the N terminus of purified XOR (Thr 2 -Phe 7 ). The sizes of the 64-and 85-kDa fragments by SDS-PAGE agree reasonably well with the predicted molecular masses of the N-terminal (Thr 2 -Phe 560 ; 61,678 daltons) and C-terminal (Gln 561 -Val 1331 ; 84,829 daltons) peptide fragments generated by cleavage at Phe 560 . These results suggest that the 64-and 85-kDa fragments are primary products of chymotrypsin digestion of XO. The sequence of the first 7 amino acids of the 44-kDa fragment (band 3) corresponds to Asn 182 -Thr 188 and is consistent with cleavage at Met 181 . The size of this fragment agrees with that predicted for a peptide fragment from Asn 182 to Phe 560 (42,413 daltons) and suggests that the 44-kDa fragment is generated by cleavage of the 64-kDa fragment at Met 181 . Hydropathy analysis (28) of the amino acid sequences surrounding the cleavage sites ( Fig. 2) shows that Met 181 and Lys 184 lie in a hydrophilic region, while Phe 560 is in a relatively hydrophobic region of XOR. These results demonstrate that native XD and XO possess distinct solution conformations characterized, in part, by exposure of a hydrophobic region containing Phe 560 in the XO form.
XD and XO Possess Different Hydrodynamic Properties-To determine whether the conformational differences in XD and XO represented global changes in their conformations we investigated their hydrodynamic properties by size exclusion chromatography. Fig. 3 shows the elution properties of XD (D/O ϭ 6.2) and XO (D/O Ͻ 0.1) during HPLC-SEC at pH 7.5. Both forms of the enzyme eluted as single peaks (the small amount of absorbance at later retention times in the XD sample is due to DTT used to prepare XD) and appeared to be homogenous in composition. However, the average retention time of XO (17.52 Ϯ 0.11 min; n ϭ 6) was longer than that of XD (17.13 Ϯ 0.07 min; n ϭ 5). These results indicate that disulfide bond reduction leads to a significant change in the frictional coefficient of XOR and is consistent with an increase in the hydrodynamic volume of the enzyme. The nature of the disulfide bonds within XO was investigated by two-dimensional peptide mapping of chymotrypsin-digested XO under nonreduced and reduced conditions.  Organization of Disulfide Bonds in XO-A representative two-dimensional peptide map of chymotrypsin-digested XO is shown in Fig. 4A. The digested enzyme was denatured in SDS in the absence of reducing agents and subjected to SDS-PAGE under nonreducing (first dimension) and reducing (second dimension) conditions. In addition to residual intact XO at 145 kDa, prominent bands at 85, 64, and 44 kDa were observed to lie on the diagonal indicated by arrowheads. Additional bands were also observed to the right of the diagonal at 64 and 44 kDa. The lack of significant staining to the left of the diagonal indicates that there are few, if any, disulfide bridges between undigested XO subunits or between the 85-, 64-, and 44-kDa fragments generated by chymotrypsin digestion. The presence of groups of bands spreading horizontally to the right of the 64and 44-kDa bands indicates the existence of electrophoretic heterogeneity within these fragments. Since the first dimension was carried out in SDS under nonreducing conditions and fragments within each group have the same size, this observation suggests that this heterogeneity is due to disulfide bondmediated conformational differences within these fragments. To verify that the electrophoretic heterogeneity within the 64and 44-kDa fragments is the result of disulfide bonding, XO was first digested with chymotrypsin and then incubated with 5 mM DTT prior to two-dimensional peptide mapping. Fig. 4B shows that under these conditions the 85-, 64-, and 44-kDa fragments appear as tight bands on the diagonal (arrowheads), and there is no evidence of electrophoretic heterogeneity associated with these fragments.
Cross-linking Prevents Thiol-mediated Conversion of XO to XD-To determine whether a conformational switch is required for the conversion of XO to XD we investigate the effects of cross-linking on the conversion process. Table II shows the effects of cross-linking XO with the amine-specific reagent BS3 on the ability of DTT to convert XO to XD as indicated by an increase in the D/O ratio. The initial D/O values of cross-linked and non-cross-linked (control) enzyme were Ͻ0.1. After incubating with 10 mM DTT for 2 h the D/O of control XO increased to nearly 4, whereas the D/O value of the cross-linked enzyme was still less than 0.1. BS3 did not react with cysteine thiols or interfere with the ability of DTT to reduce XO cysteine thiols (data not shown). Moreover it did not interfere with the XD activity of enzyme that was reduced prior to adding crosslinking reagent (Table II). Thus, the inability of DTT to convert BS3 cross-linked XO to XD does not appear to be due to a specific interference of BS3 with XD enzymatic activity or its reaction with cysteine thiols.
The nature of the BS3 cross-linked enzyme was characterized by HPLC-SEC and SDS-PAGE. The majority of the crosslinked enzyme eluted with the same retention time as native  (Fig. 5). SDS-PAGE analysis of cross-linked XO (Fig. 5, inset) showed that it was composed of bands between 180 and 200 kDa as well as bands near the top of the gel and at ϳ150 kDa. The results demonstrate that cross-linking primarily occurs within, and/or between, individual XO subunits and that there is relatively little cross-linking between enzyme molecules or formation of large cross-linked enzyme aggregates. Therefore, cross-linking also does not appear to influence the assembly state of the enzyme. Together these results suggest that cross-linking with BS3 prevents a conformational change that is necessary for the reductive conversion of XO to XD.
Cleavage of XO at Phe 560 Prevents Conversion to XD-Since exposure of the region containing Phe 560 is a distinct feature of the XO conformation we tested the possibility that alterations to this region by proteolytic cleavage would interfere with conversion to XD by thiol reduction. Fig. 6 shows the effects of trypsin and chymotrypsin digestion on the ability of DTT to convert XO to XD (Fig. 6A) and on the conversion of XD to XO (Fig. 6B). XO digested with trypsin for up to 2 h could still be converted to XD by incubation with 5 mM DTT for 1 h; however, digestion with chymotrypsin for as little as 30 min completely prevented conversion. In contrast, neither chymotrypsin nor trypsin digestion led to significant conversion of XD to XO. Since chymotrypsin and trypsin cleave XO and XD at proximal sites (Met 181 and Lys 184 , respectively) within their N-terminal regions it is unlikely that cleavage within this region is responsible for the failure of DTT to convert chymotrypsin-cleaved XO to XD. Conversely, the selective cleavage at XO at Phe 560 by chymotrypsin suggests that structural alterations within this region are responsible for the inability of XO to be converted to XD by DTT reduction. DISCUSSION It is well established that mammalian XOR can be reversibly converted from an NAD ϩ -dependent dehydrogenase to an O 2dependent oxidase by oxidation of cysteine thiols to disulfide bonds (11,12). The present study demonstrates that the conversion of XD to XO involves a global conformational change that is coupled to a reduction in the apparent hydrodynamic size of XOR. The conformational change was mapped to a specific region surrounding Phe 560 and was shown to be a necessary step in the conversion process.
Although the specific activities of our preparations indicate that 30 -50% of the purified enzyme is catalytically inactive (2), evidence from the literature and from the present studies suggests that active and inactive XOR molecules possess similar structures and undergo equivalent conformational changes. First, catalytically inactive forms are present in many highly purified and physically homogenous XOR preparations (2). The inactive forms appear to result from specific alterations of the The inset shows the SDS-PAGE and silver stain analysis of selected fractions across the peak. The numbers adjacent to the XO activity profile indicate the fractions assayed by SDS-PAGE. The elution of protein was monitored at 220 nm and the elution XOR activity was monitored using the xanthine/dichloroindophenol assay ("Experimental Procedures") and is shown as the relative rate of xanthine oxidation/min per 20-l aliquot. molybdopterin cofactor (2) and not from alterations in the structural properties of the protein. Second, the hydrodynamic and electrophoretic properties of our purified XOR preparations demonstrate that the enzyme is intact and is composed of a relatively homogenous population of molecules. Size exclusion chromatography of the XD and XO isoforms also indicates that there is a uniform response to thiol reduction of XO. Third, the precise proteolytic patterns observed for both the XD and XO isoforms indicate that within a given isoform population the overall conformations are relatively homogenous. The limited and highly selective cleavage of XD and XO by trypsin and chymotrypsin is consistent with previous proteolysis studies of rat and bovine XOR (6, 29) and with subtilisin cleavage of chicken XD (8) and indicates that the native conformations of both isoforms are highly ordered and that homologous forms of avian and mammalian XOR have similar structural domains. Trypsin selectively cleaved the bovine XD and XO isoforms at Lys 184 . The absence of significant amounts of additional trypsin cleavage products indicates this region is highly accessible in the native conformations of both enzyme forms and that the degree of local segmental motion within this region is significantly greater than that of other regions of the enzyme (30,31). Chymotrypsin selectively cleaved XD and XO at an adjacent site in this region (Met 181 ). The close proximity of the Met 181 and Lys 184 sites combined with earlier findings that trypsin cleaves rat liver XD at Lys 185 (6) and subtilisin cleaves chicken liver XD at a homologous site (Lys 248 ) suggests that flexible structural properties are a general feature of this region in both mammalian and avian XOR molecules. Comparison of the amino acid sequences of mammalian and avian enzymes shows that Lys 184 and Met 181 are located in a highly conserved, hydrophilic region of the enzyme at the boundary between the iron-sulfur and flavin domains (8,10,22).
In contrast to the high susceptibility of Met 181 and Lys 184 to proteolytic attack in both XD and XO isoforms, Phe 560 is cleaved by chymotrypsin only when it is in the XO form. Hydropathy analysis shows that Phe 560 is in a relatively hydrophobic region of the enzyme that would be expected to exhibit reduced solvent exposure under physiological conditions. The observation that this site becomes accessible to chymotrypsin only after XD is converted to XO therefore suggests that disulfide bond formation produces significant structural changes within this region of XOR. Based on sequence alignment comparisons of XOR family members (32) and crystallographic studies of bovine XD (22), the Phe 560 site appears to be in an unstructured region near the boundary of the flavin and molybdopterin domains. Thus, disulfide bond formation appears to produce conformational changes that delineate the extent of the flavin-binding region of XOR. Furthermore, size exclusion analysis indicates that disulfide bond formation reduces the hydrodynamic size of XOR. These results extend previous observations (20,21) of conformational changes within the flavinbinding site following thiol oxidation-dependent conversion of XD to XO by demonstrating that the conversion involves specific structural alterations and global changes in the hydrodynamic properties of XOR.
The pattern of chymotrypsin digestion of bovine milk XD is different from that previously reported for rat liver XD (11). In this earlier study, chymotrypsin cleavage of XD generated significant amounts of 64-and 44-kDa fragments. However, the D/O value of enzyme used in these studies was 1.3; thus their preparation appeared to contain significant amounts of the XO form. In addition, the digestion was carried out at 37°C, which potentially leads to local unfolding (30) and has been reported to result in the conversion of XD to XO (17). Therefore, the basis for the difference in chymotrypsin cleavage pattern may reside in the relative XD/XO composition of the sample and/or temperature-dependent conformational changes, rather than actual differences in the conformations of the rat and bovine enzymes. This conclusion is supported by the similarity of the chymotrypsin cleavage pattern of chicken XDH at 25°C (33) to that reported here for bovine XDH. Since avian XOR exists only in the XDH form (8) and contains a region that is highly homologous to that surrounding Phe 560 of bovine XOR (8), it is likely that the inaccessibility of the Phe 560 region to chymotrypsin is a general feature of the XDH isoform.
Initial studies of the XD to XO conversion suggested that the cysteine residues affecting the dehydrogenase activity of XOR were located in the cysteine-rich N-terminal region (6,17). However, recent studies of rat liver XOR have led to the proposal that Cys 535 and Cys 992 are involved in the conversion through the formation of disulfide bonds between these residues (34). Our two-dimensional electrophoresis results provide evidence of significant disulfide bond formation within the N-terminal region of XOR (the 44-and 64-kDa fragments) but not between residues located in the C-terminal fragment (the 85-kDa band, which contains Cys 992 ) and N-terminal fragment (the 64-kDa band, which contains Cys 535 ). Since these residues are conserved in all mammalian enzymes studied to date, our data suggest the Cys 535 and Cys 992 form disulfide bonds with as yet unidentified Cys residues in the 64-and 85-kDa regions, respectively.
The basis of the disulfide bond heterogeneity in the N-terminal region of XO is unknown. However, milk XO appears to be formed by the action of a membrane-bound sulfhydryl oxidase during the secretion process (13,35). If the conformation of XD leads to clustering of cysteine residues it is possible that this enzyme might catalyze formation of disulfide bonds within groups of closely associated cysteine residues resulting from the generation of XO molecules with different disulfide bond combinations and consequently different conformations and electrophoretic mobilities. The formation and rearrangement of disulfide bond isomers by folding catalysts such as proteindisulfide isomerase are known to occur during the folding of nascent proteins and are part of the process of generating proper native conformations (36). A similar process may occur in the conversion of XD to XO during milk secretion and lead to the formation of disulfide bond isomers of XO.
Two lines of evidence from the present study demonstrate that the conversion of XO to XD is dependent on specific structural changes associated with disulfide bond reduction. The first is that cross-linking XO prevents the reductive conversion of XO to XD. Cross-linking did not interfere with XOR activity, and the majority of the cross-linking was intramolecular and occurred without major changes in the conformation of XO. Thus it appears that the inability of DTT to convert crosslinked XO to XD is not due to alterations in substrate utilization or artifactual intermolecular interactions between crosslinked XOR dimers. The second is that selective cleavage of XO at Phe 560 prevents conversion to XD by DTT. Because trypsin and chymotrypsin cleave XO at nearby sites in the N-terminal region but only chymotrypsin cleavage prevents conversion to XD, it appears that structural changes within the region around Phe 560 are critical for the conversion process. As yet it is unclear if cleavage at Phe 560 alone or a combination of cleavage at Phe 560 and Met 181 is required to prevent reductive conversion of XO to XD. However, time course studies indicate that Phe 560 cleavage occurs first and that cleavage at this site correlates with the inhibition of the XO to XD conversion. 2 Support for the importance of the region surrounding Phe 560 comes from observations that modification of Cys 535 of rat liver XOR converts XD to XO (34) and from crystallographic data that show that this region is disordered in intact bovine XD and proteolyzed bovine XO and constitutes a linking segment between the flavin and molybdopterin domains (22). Our results do not support earlier conclusions, based on proteolytic digestion of rat XOR, that alterations within the cysteine-rich Nterminal region by cleavage at Lys 185 are responsible for the XD to XO conversion (6, 17). As indicated above, technical differences, especially the use of elevated temperatures for proteolytic digestion experiments, may be the basis for this discrepancy. In fact, more recent studies of rat liver XOR have shown that cleavage at Lys 185 does not induce the D to O conversion of rat XOR (34). Together with evidence from recent investigations of the role of cysteine residues in the XD to XO conversion (34), the results from our study indicate that the XD-XO interconversion requires a conformational switch in the region surrounding Phe 560 . Alterations that affect the structure of this region, such as proteolysis, decouple this switch and interfere with the interconversion process.
Although the conversion of XD to XO represents a pathologically important event due to increased production of cytotoxic reactive oxygen species by the XO form (37), the biological importance of the conformational change associated with the conversion is less established. A potential physiological function of this conformational change is in milk lipid secretion. It has been known for nearly a century that XOR is highly enriched in cow's milk (2). Numerous studies have documented that the primary form of the milk enzyme is XO (11,13,38), and it has been demonstrated that XOR is a major protein constituent of the membrane that surrounds lipid globules in milk (39). Immunofluorescence studies have shown that XOR is selectively associated with the apical plasma membrane in lactating cattle (40), and histochemical evidence suggests that the membrane-associated form is XO (41). We recently demonstrated that the primary form of XOR in mouse mammary tissue is XD, that rapid conversion of the XD form to the XO form occurs during milk secretion, and that membranes surrounding milk lipid globules contain an enzyme capable of converting XD to XO (13). Thus, it is possible that the conformational changes associated with the XD to XO conversion, in particular the exposure of the hydrophobic region around Phe 560 , are important in the association of XOR with the apical membrane in mammary epithelial cells and may play a role in milk lipid secretion. Electron microscopic studies have shown that the proteins on the inner surface of milk fat globule membranes are structurally organized into highly ordered hexagonal arrays (42). It will be interesting to determine whether changes in the conformation of XOR contribute to the organization of these structures.