Oxidative dissociation of human alpha 2-macroglobulin tetramers into dysfunctional dimers.

Human alpha 2-macroglobulin is a broad-spectrum, homotetrameric antiproteinase that can maximally bind up to two proteinase molecules in a ternary complex. Proteinases cleave the inhibitor within a peptide stretch termed the bait region and induce the emergence of internal thiol esters whose nucleophilic scission precede a major conformational change which entraps enzymes within molecular cages. In a previous study, leukocyte-generated hypohalous acids and N-haloamines were identified as the first examples of physiologically relevant inactivators of the antiproteolytic activity of alpha 2-macroglobulin (Reddy, V. Y., Pizzo, S. V., and Weiss, S. J. (1989) J. Biol. Chem. 264, 13801-13809), but the mechanisms whereby the oxidants damaged the inhibitor remained undefined. We now demonstrate that N-chloramines (RNCl) destroy the antiproteolytic activity of alpha 2-macroglobulin in an unusual biphasic process that results in the formation of inactive alpha 2-macroglobulin half-molecules. In the first phase, 8 eq of RNCl reacted with each alpha 2-macroglobulin subunit to generate a partially oxidized antiproteinase containing 8 methionyl sulfoxide residues/monomer. Structure-function analyses demonstrated that the oxidized inhibitor retained its homotetrameric structure as well as its ability to entrap proteinases. In marked contrast, the oxidation of an additional 6 methionyl residues and a single tryptophanyl residue fractured the alpha 2 M homotetramer across its non-covalent axis into two pairs of disulfide-linked dimers. Despite the fact that the oxidized dimers displayed normal bait regions whose cleavage by proteinases initiated thiol ester scission, all antiproteolytic activity was lost. Furthermore, the oxidized dimers were unable to undergo the critical conformational changes normally associated with bait region cleavage or thiol ester scission. Together, these results demonstrate that chlorinated oxidants destroy the antiproteolytic activity of alpha 2-macroglobulin by attacking a subset of methionyl and tryptophanyl residues whose oxidation mediates the dissociation of the native homotetramer into conformationally locked dimers.

Human az-macroglobulin is a broad-spectrum, homotetrameric antiproteinase that can maximally bind up to two proteinase molecules in a ternary complex. Proteinases cleave the inhibitor within a peptide stretch termed the bait region and induce the emergence of internal thiol esters whose nucleophilic scission precede a major conformational change which entraps enzymes within molecular cages. In a previous study, leukocytegenerated hypohalous acids and N-haloamines were identified as the first examples of physiologically relevant inactivators of the antiproteolytic activity of azmacroglobulin (Reddy, V . Y., Pizzo, S. V . , and Weiss, S. J. (1989) J. BioZ. Chem. 264, 13801-13809), but the mechanisms whereby the oxidants damaged the inhibitor remained undefined. We now demonstrate that N-chloramines (RNCl) destroy the antiproteolytic activity of azmacroglobulin in an unusual biphasic process that results in the formation of inactive az-macroglobulin half-molecules. In the first phase, 8 eq of RNCl reacted with each az-macroglobulin subunit to generate a partially oxidized antiproteinase containing 8 methionyl sulfoxide residuedmonomer. Structure-function analyses demonstrated that the oxidized inhibitor retained its homotetrameric structure as well as its ability to entrap proteinases. In marked contrast, the oxidation of an additional 6 methionyl residues and a single tryptophanyl residue fractured the azM homotetramer across its non-covalent axis into two pairs of disulfide-linked dimers. Despite the fact that the oxidized dimers displayed normal bait regions whose cleavage by proteinases initiated thiol ester scission, all antiproteolytic activity was lost. Furthermore, the oxidized dimers were unable to undergo the critical conformational changes normally associated with bait region cleavage or thiol ester scission. Together, these results demonstrate that chlorinated oxidants destroy the antiproteolytic activity of az-macroglobulin by attacking a subset of methionyl and tryptophanyl residues whose oxidation mediates the dissociation of the native homotetramer into conformationally locked dimers. * This research was supported by Grants AI-23870 and HL-28024 from the National Institutes of Health and Council for Tobacco Research Grant 2786. 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. The chlorinated oxidants and proteolytic enzymes released from triggered human neutrophils can compromise the function of virtually every known plasma proteinase inhibitor by (i) oxidizing critical cysteinyl or methionyl residues, or (ii) hydrolyzing peptide bonds located at or near reactive site domains (e.g. [1][2][3][4][5][6][7][8]. Until recently, the only plasma proteinase inhibitor believed to be resistant to oxidative or proteolytic inactivation has been the multifunctional antiproteinase, a,-macroglobulin (azM)' (9)(10)(11). a2M is a large molecular mass glycoprotein (-720 kDa) comprised of four identical subunits which are disulfide-linked into dimers that then associate into tetramers through strong, non-covalent interactions (9,10). Each tetramer contains two proteinase-binding sites and can, under optimal conditions, bind a maximum of two proteinase molecules (9,10). Although the organization of the a2M subunits which allows for the formation of proteinase-binding sites remains controversial (i.e. a binding site may be fashioned from two monomers that are either non-covalently or covalently linked), azM dimers generated by a variety of techniques retain antiproteolytic activity (12)(13)(14)(15)(16)(17)(18).
Unlike all other proteinase inhibitors, a2M (as well as the closely related human a-macroglobulin, pregnancy zone protein) inhibits proteinases from all four catalytic classes by trapping the targeted enzyme in a molecular cage from which its dissociation is prevented (9,101. According to current models of a2M function, an attacking proteinase initiates its own entrapment by cleaving one of the more than 10 peptide bonds that are arranged in a short sequence near the center of the a,M subunit (9,10). Cleavage anywhere within this sequence (termed the bait region) allows the proteinase to regenerate its active site, but results in a perturbation of azM structure that initiates a conformational change causing p-cysteinyl-y-glutamyl thiol esters to emerge from hydrophobic pockets (9, 10). The thiolesters then react with either water or a nucleophilic group on the proteinase, which in the latter case, leads to the covalent linkage of the a2M subunit to the enzyme (9,10). Although the formation of covalent cross-links between a2M and proteinases is not a necessary prerequisite for trapping, the scission of the thiol ester precedes or signals a second, more dramatic change in conformation which terminates in the formation of a non-dissociable enzyme-inhibitor complex (9,10).
Despite the complexity of the a2M trapping mechanism, the fact that the native inhibitor does not contain free cysteinyl groups, that methionyl residues have not been demonstrated essential to its antiproteolytic function, and that human pro-Oxidative Dissociation of a2M Tetramers teinases capable of catalyzing an inactivating cleavage2 have never been identified, it is not surprising that a2M has long been considered the "fail-safe" plasma antiproteinase at inflammatory sites. However, recent studies have demonstrated that triggered leukocytes can irreversibly destroy a2M function by attacking the inhibitor with the halogenated oxidants, hypochlorous acid (HOC1) or hypobromous acid (HOBr) (11). Although halogenated oxidants represent the only known example of a physiologic a2M inactivator, the molecular mechanisms underlying the destruction of the antiproteinase remain unknown. Efforts to either identify the critical moieties oxidized in the a2M tetramer or define the functional consequences of their modification have been hampered in part, by the extreme chemical reactivity and low selectivity of the hypohalous acids (11). Interestingly, however, while comparing the ability of HOCl and derivative N-chloramines (RNCl) to inactivate a2M under cell-free conditions, we noted that RNCl act as highly selective oxidants which would potentially permit structure-function analyses of the oxidized antiproteinase (11). Accordingly, we now demonstrate that a2M is inactivated following the co-oxidation of a subset of 6 methionyl residues and a single tryptophanyl residuela2M monomer. Coincident with these oxidative modifications, the native tetramer is sheared across its non-covalent axis into covalently linked dimers. Despite the fact that the oxidized dimers display normal bait regions which transduce the conformational signals necessary for thiol ester scission, the modified a2M molecules are unable to covalently bind or trap attacking proteinases. Together, these results not only define the mechanism by which physiologically relevant oxidants destroy a2M activity, but also serve to identify a subset of methionyl and tryptophanyl residues that appear to play a critical role in controlling a2M structure and function.
a2M Antiproteolytic Actiuity-At indicated time points, the reaction of RNCl with aZM was terminated by adding methionine (10 mM) to quench residual oxidants (11) and an aliquot of the native or oxidized aZM reaction mixture incubated with a 2-fold molar excess of porcine pancreatic elastase (Elastin Products; effective concentration determined by active site titration) for 15 min at 25 "C (22). Functional a,M was then assessed on the basis of the ability of the antiproteinase to either (a) protect the amidolytic activity of porcine pancreatic elastase activity from an excess of a,-proteinase inhibitor or, (b) inhibit porcine pancreatic elastase-dependent proteolysis of insoluble elastin (11). In the amidase assay, aliquots of azM that had been treated with pancreatic elastase were subsequently incubated alone or with a 2-fold molar excess of a,proteinase inhibitor (Calbiochem) for 15 min at 25 "C. Amidase activity was assayed with methoxysuccinyltrialanyl p-nitroanilide (1 mM; Sigma) and unbound elastase defined as the difference in moles of substrate hydrolyzed between aliquots incubated alone and with a,-proteinase inhibitor (11). In the elastinolysis assay, azM samples that had been incubated with pancreatic elastase were subsequently mixed with 200 pg of 3H-labeled insoluble elastin (3100 counts/ midpg; reductively labeled with PHINaBH, as described in Ref. 23) for 24 h at 37 "C (11). At the end of the incubation period, an aliquot of the An inactivating cleavage is defined as a hydrolytic event in which the proteinase destroys azM function without being trapped. reaction mixture was removed, centrifuged at 10,000 x g (10 min), and solubilized radioactivity determined by p-scintillation counting.
The determination of k,.,, for the interaction of pancreatic elastase with native or oxidized azM was determined by competition assay with a,-proteinase inhibitor as described by Virca and Travis (24). In brief, aZM (0.14 w) was mixed with cy1-proteinase inhibitor and pancreatic elastase at a molar ratio of 1:lO:l (a,M/a,-proteinase inhibitor/ pancreatic elastase) in Dulbecco's buffer, pH 7.4, at 25 "C. Aliquots were assayed at timed intervals for both amidolytic activity (as the cyZ" elastase complex) and non-complexed a,-proteinase inhibitor, and the kass, calculated as described (24).
Polyacrylamide Gel Electrophoresis-Samples of native or oxidized a,M were examined by electrophoresis directly or following incubation with either (i) a 2-fold molar excess of pancreatic elastase or bovine trypsin (Sigma) for 15 min at 25 "C, (ii) 200 m M methylamine (2 h at 25 "C), or (iii) 200 m M methylamine followed by pancreatic elastase. In selected experiments, native or oxidized azM that had been incubated with pancreatic elastase or trypsin were treated with 12.5 p~ [3Hldiisopropylfluorophosphate (4 CUmmol; DuPont-New England Nuclear) to label antiproteinase-bound serine proteinases. Non-denaturing PAGE and SDS-PAGE were performed in 5% gels as previously described (11,25). Samples prepared under non-reducing conditions contained 1.5 m iodoacetate while samples containing active proteinases were incubated with 1 m M phenylmethanesulfonyl fluoride prior to denaturation. Where indicated, samples were electrophoresed on 6 1 5 % gradient gels (Jule Inc.) after equilibration into a 25 m Tris-HC1, 192 n m glycine buffer, pH 8.2, for 30 min. Gels were fmed with trichloroacetic acid and stained with Coomassie Brilliant Blue as described (11).
Size Exclusion Chromatography-Dialyzed native or oxidized a2M was subjected to chromatography on a prepacked Superose-6 column (Pharmacia LKB Biotechnology) as described (26). The chromatography buffer was phosphate-buffered saline, and the flow rate was 0.4 mumin. The elution volumes (V,) of blue dextran (2000 kDa) and NaN3 were 6.4 and 19.7 ml, respectively.
Electron Microscopy-Preparations of native or oxidized azM (-70 nM) were adsorbed to thin carbon films for 1 min at 22 "C, negatively stained with 2.0% uranyl formate, air-dried at 25 "C and examined by transmission electron microscopy as described (27).
Amino Acid Analysis of Native and Oxidized a,M-Amino acid analysis was carried out as described previously with hydrolysis performed at 155 "C for 45 min (28). Methionine sulfoxide was determined as methionine after acid hydrolysis of CNBr-treated samples (29). Alkaline hydrolysis and determination of tryptophan content was performed as described earlier (30). The concentration of aromatic amino acids was also determined spectrophotometrically by multicomponent analysis following the denaturation of the a2M samples in 6 M guanidine before or after reduction and carboxymethylation (31). Given that the number of tyrosine residues in native or oxidized aZM was constant (see "Results"), the number of tryptophan residues was determined using the formula: where a is the concentration of tryptophan, b is the concentration of tyrosine, and c is a constant which fixes the number of tryptophan residues in native azM at 11 as determined by sequence analysis (32). The constant c was determined from the deconvolution of native azM to be 50.85. Bityrosine content was monitored fluorometrically at 325 nm excitation and 410-420 nm emission (33). Determination of Carbonyl Content-Carbonyl content of native or oxidized aZM was determined spectrophotometrically following the generation of 2,4-dinitrophenylhydrazone derivatives (29,34). The derivatized azM was freed of excess 2,4-dinitrophenylhydrazine by either (a) chromatography over a Sephadex PD-10 column (Pharmacia LKB Biotechnology, Inc.), or (6) HPLC gel filtration using a Zorbax GF-250 column (Mac-Mod Analytical, Chadds Ford, PA). The carbonyl content was then determined using the formula: where a is the molecular weight of the a,M subunit, b is the absorbance value of the hydrazone at 370 nm, and c is the absorbance value at 276 nm (proportional to protein concentration). The molar absorption coefficient of the hydrazone is 22,000 M-, cm-l (34). When samples were analyzed by HPLC, the areas of the 370 and 276 nm peak were used for b and c, respectively.
Methylaminolysis and Sulfhydryl Titration of a,M-The integrity of the p-cysteinyl-y-glutamyl thiol ester of native or oxidized a2M was assessed by monitoring the (i) formation of a substituted alkylamide with ["TICH,NH,, or (ii) appearance of sulfhydryl groups (9, 10). For CH:*NH2 incorporation studies, native or oxidized a2M (0.17 p~) was incubated with 25 mM [14CICHnNH2 (3.66 mCi/mmol; DuPont-New England Nuclear) for 18 h a t 25 "C (35). Following native PAGE, the gels were fixed in 30% methanol and 10% acetic acid for 15 min and processed for fluorography.
The appearance of sulfhydryls was monitored by quantitating the reaction of the liberated thiol with 5,5'-dithiobis(2-nitrobenzoic acid) assuming a molar absorption coeficient of 13,600 M -~ cm-I a t 412 nm for the p-nitrothiophenolate anion (36). An aliquot of either native or oxidized azM (0.67 p~) was added to a cuvette containing 2 mM 5,5'dithiobis(2-nitrobenzoic acid) in Dulbecco's buffer, pH 7.4, a t 25 "C. The reaction was then initiated by the addition of either methylamine (200 mM), pancreatic elastase (0.67 p~) , or bovine trypsin (0.67 p~) . The second-order rate constant was determined from the pseudo-first-order rate constant as described (36).
NH,-terminal Sequence Analysis-Native or oxidized azM (100 pmol) was incubated with an equimolar quantity of pancreatic elastase, human neutrophil elastase (Calbiochem), or human fibroblast collagenase (purified and activated as described in Ref. 37) for 5 min a t 25 "C. The reaction mixtures containing serine proteinases or metalloproteinases were then treated with phenylmethanesulfonyl fluoride (1 mM) or ophenanthroline (0.1 mM), dialyzed overnight against water, lyophilized, and sequenced as described (37). (25 VM) to native azM led to the rapid consumption of 18.9 eq of oxidant/azM subunit (n = 2) within 15 min resulting in a complete loss of antiproteolytic activity (Fig. lA). In contrast, an equimolar quantity of NHzCl reacted with aZM a t a slower rate allowing recognition of a distinct biphasic pattern (Fig. lA). During the early rapid phase of oxidation (i.e. the first 15 rnin), 7.8 2 0.8 eq (n = 8, mean 1S.D.)" of NHzCI reacted with each azM subunit (this partially oxidized form of azM is henceforth referred to as azMox-l). aZMox-l appeared identical to native azM as assessed by pore limit gel electrophoresis under native conditions ( Fig. 2 A ) , and retained full antiproteolytic activity Following the generation of the azMox.l species, the continued reaction of NHzCI with the antiproteinase resulted in the consumption of an additional -8 eq oxidant by 2 h (Fig. lA; this form of azM is herein referred to as azM,x.Z). In direct contrast to the apparently native structure of azMox-l, azMox.z structure was dramatically altered as assessed by native pore-limit gel electrophoresis ( Fig. 2A ). Under denaturing conditions, aZMox-2 migrated closely with native azM dimers (Fig. 2B 1  tive azM dimers generated by cadmium or mild acid ( Fig. 3; Refs. 14, 17,18). Significantly, the oxidized dimers were completely unable to protect pancreatic elastase from inhibition by al-proteinase inhibitor (Fig. lA) and the loss in activity correlated closely with the formation of (data not shown). The loss in activity was not restricted to pancreatic elastase since similar results were obtained when azMox.z was incubated with trypsin or human neutrophil elastase (data not

B
shown). a2Mox-2 was also unable to exert antiproteolytic activity when porcine pancreatic elastase was incubated with insoluble elastin in the presence of the oxidized inhibitor (98.7 2 1.2% inactivated, n = 3). Interestingly, the sensitivity of the antiproteinase to oxidative inactivation was limited to native a2M since the addition of NH2Cl to preformed azM-pancreatic elastase complexes failed to either reverse proteinase inhibition or dissociate the tetramer into dimers despite the fact that comparable amounts of oxidant were consumed (18.1 2 0.6 eq NH2Cl consumeda2M subunit, n = 3). Finally, although NHzCl is a relatively lipophilic oxidant with stronger chlorinating potential than N-chloramines derived from primary amines (381, hydrophilic RNCl (e.g. p-alanine chloramine or Tris chloramine) similarly generated a2Mox-1 and a2Mox.z following the reaction of -8 and -16 eq of the chlorinated oxidants/azM subunit, respectively (Fig. 1B). As expected, all changes in a2M structure or function were completely prevented when the antiproteinase was incubated with chlorinated oxidants in the presence of an excess of methionine (data not shown).
Amino Acid Modifications of azMox., and azMOx.-Because sulfur-containing amino acids rank among the most sensitive targets to oxidative modification by chlorinated amines (11, initial efforts focused on determining the status of methionyl residues and thiol esters (native a2M contains no free cysteinyl groups) in samples of oxidized a2M. In comparison to native a2M (which contained 1.1 2 1.3 methionine sulfoxides/a2M subunit, n = 41, the sulfoxide content of azMox-l increased to 8.8 = 1.1 residues ( n = 4) reflecting the net oxidation of -8 methionine residues. Although the integrity of the thiol esters cannot be assessed directly by amino acid analysis, small molecular size nucleophiles are able to react with the ester to generate a substituted alkylamide at Glxgs2 and a titratable S H group (9,10). Thus, native a2M and a2Mox-l were incubated with [14Clmethylamine and incorporation qualitatively monitored by SDS-PAGE and fluorography while the simultaneous appearance of thiol moieties was quantitated with 5,5'-dithio-bis(2-nitrobenzoic acid). As shown in Fig. 4, [14C]methylamine was incorporated comparably into both forms of a2M, while quantitative analysis demonstrated that the native a2M and a2Mox.l exposed 1.00 2 0.01 thioYazM subunit ( n = 5) and 0.98 2 0.01 thi~l/a~M,,.~ subunit (n = 3), respectively. Finally, the full amino acid analysis of the native azM and a2Mox-l were compared in an attempt to identify any additional oxidative modifications, but no other significant changes could be detected (Table I). Given the fact that (i) 8 eq oxidant reacted with each azM subunit to generate a2Mox.l and, (ii) that 1 mol of NH2Cl reacts with 1 mol of methionyl residues to generate 1 mol of methionyl sulfoxide residues, these data demonstrate that aZMox-l is generated following the oxidation of -8 of the 25 methionyl residues/a2M subunit.
Following the formation of azMox-2 dimers, the methionyl sulfoxide content increased by an additional 6 residues/subunit (i.e. to 15.0 2 1.4 for a net increase of 6.3 2 1.4 sulfoxides/ a2Mox-2 subunit relative to a2Mox-l, n = 4) while thiol ester integrity remained intact as assessed by either [14Clmethylamine incorporation (Fig. 4) or thiol exposure (i.e. a2Mox-2 released 0.96 2 0.01 thioYsubunit, n = 3). To determine whether the oxidation of the 6 additional methionyl residues alone correlated with the inactivation process, azMox-l was incubated with NH2Cl, the reaction terminated at various times with methionine, residual antiproteolytic activity quantitated, and the partially inactivated inhibitor analyzed for methionyl content. As shown in Fig. 5A, the loss of the additional 6 methionyl residues correlated closely with the loss of azM activity ( r = 0.99). Furthermore, when a2M was incubated with stimulated human neutrophils, the methionine sulfoxide content of the oxidized a2M similarly correlated with the loss of antiproteinase activity (Fig. 5A).
Although the oxidation of the additional methionyl residues correlated with the inactivation process, the consumption of 8 eq of RNCl during the generation of a2Mox-2 leaves unaccounted the moieties targeted by the 2 remaining eq of oxidant.
Amino acid analysis failed to identify reproducible changes in oxidized azM relative to native aZM (Table I), and bityrosine could not be detected by fluorometric analysis (data not shown). Although variabilities in the tryptophan yield following alkaline hydrolysis complicated interpretation, regression analysis of plots of a2M activity versus tryptophan content yielded a slope that was significantly different from zero, suggesting a loss of -1-2 tryptophan residuesla2M subunit ( r = 0.55; data not shown). In order to obtain a more precise determination of the tryptophan content, variably inactivated samples of azM were submitted to multicomponent analysis of the UV second derivative spectrum (Fig. 5B). Whereas aZMox-l exhibited no loss in tryptophan content, increasing levels of oxidative inactivation resulted in a linear loss of a single tryptophan residue/ azM subunit with no attendant changes in other aromatic moieties detected. Similar results were obtained when azM was inactivated by stimulated neutrophils (Fig. 5B). When the oxidized azM samples were reduced and carboxymethylated prior to spectral analysis in order to minimize contributions from protein secondary structure, an identical loss of a single tryptophanyl residuelsubunit was detected (data not shown). Finally, because 2 mol of RNCI can oxidize 1 mol of tryptophan to the oxindole which, in turn, should be detectable on the basis of its ability to react with 2,4-dinitrophenylhydrazine (39-41), samples of oxidized azM were examined quantitatively for the formation of the protein-hydrazone adduct. As shown in Fig. 6, the oxidation of the single tryptophanyl residue was associated with the corresponding formation of a -0.56 carbonyls/azM subunits. Taken together, these results indicate that the RNCImediated destruction of azM antiproteolytic activity occurs via the oxidation of 6 methionyl residues and a single tryptophanyl residue/azM subunit.
Functional Characterization of the Oxidized azM Dimers-azM dimers generated by a variety of non-oxidative processes express significant antiproteolytic activity by retaining intact bait regions and thiol esters, and undergoing conformational changes that permit covalent linkage and dimer-dimer reassociation (10,(12)(13)(14)(15)(16)(17)(18). Because modifications of any of these events could upset the antiproteolytic potential of azM dimers, bait region hydrolysis, thiol ester scission, and proteinase entrapment by oxidized azM were next assessed. Following the addition of a 2-fold molar excess of porcine pancreatic elastase to native aZM, aZMox-l, or aZMox-2, bait region cleavage was examined by SDS-PAGE under reducing conditions. As shown in Fig. 7A, the subunits in the native azM tetramer (molecular mass of -185 kDa) were cleaved to yield the expected 85-95 kDa doublet (termed the I V , and I V b fragments; Refs. 9, 10).
Although identical results were obtained with azMox-l, the I V b fragment generated during O L~M~~-~ cleavage migrated aberrantly (Fig. 7 A ). In addition, a variable portion of the oxidized inhibitor could not be reduced to the monomer, apparently reflecting the generation of intersubunit cross-links (Fig. 7A). When U~M,,.~ was alternatively incubated with either human neutrophil elastase or human fibroblast collagenase, the generated I V , fragments again migrated in an identical fashion (data not shown). Because these results were consistent with the possibility that the primary cleavage site had occurred in an anomalous domain, samples of native or oxidized azM that had been incubated with either porcine pancreatic elastase, human neutrophil elastase, or human fibroblast collagenase were submitted for NHz-terminal sequence analysis. Surprisingly, bait region cleavage sites in native azM, O L~M~, -~, a n d (Y~M,,.~ matched perfectly with all three proteinases (Fig. 7B 1. Thus, differences in apparent molecular masses observed following reduced SDS-PAGE most likely reflect an altered mobility of the oxidized fragments in the electrophoretic field. ppically, bait region cleavage signals a conformational  onine, 10% of the a2M was inactivated and the tryptophan content increased to 10.9 residuedsubunit ( n = 2). change that increases the reactivity of the thiol ester such that it exists in a short-lived nascent state and subsequently undergoes hydrolysis or covalently binds to the attacking proteinase (9, 10). To first determine whether chlorinated oxidants modified the reactivity of the thiol ester, native or oxidized a2M were incubated with 200 mM methylamine a t pH 7.4 and the rate of thiol exposure monitored. As shown in Fig. 8 A , only marginal differences in the rates of thiol ester cleavage were observed. In each case, semilogarithmic plots of thiol ester cleavage as a function of time were linear (data not shown), indicating that nucleophilic attack of the uncharged amine on the thiol ester is a bimolecular reaction occumng under pseudo-first-order conditions (42,431. Accordingly, the second-order rate constants of the inhibitor-methylamine interactions were calculated as 6.  native a2M, aZMox-l, and  azM,x.2, respectively (42, 43). When thiol exposure was monitored in proteinase-treated samples of native or oxidized a2M, gross differences were similarly not detected (Fig. 8B ). Together, these data indicate that the bait region-thiol ester signaling proceeds in a near intact fashion in oxidatively modified a2M.

Oxidative Dissociation of ( Y~M Tetramers
Following scission of thiol esters in a2M tetramers or dimers, attacking proteinases are bound either by a non-covalent trap- The arrows indicate the locations of the cleavage sites determined, except for the cleavage site of native a,M by fibroblast collagenase which was obtained from the literature (67). The cleavage site identified for neutrophil elastase differs from that reported previously (68) and is (22). HNE. human neutroohil elastase: PPE. porcine pancreatic elas-more similar to that reported for the rat a-macroglobulin analogue, alIR ping mechanism or through stable E-lysyl-(proteinase)-y-glutase; HFC,' human fibroblast collagenase.

Oxidative
Conformational Flexihility of a2M,,x.,The inability of a2Mox"L to bind or trap attacking proteinases following bait region cleavage and thiol ester scission suggests that the oxidized dimer cannot undergo the major conformational changes necessary for the generation of the molecular cage. Earlier studies indicate that methylamine-treated tetramers or dimers undergo a conformational change that partially shields the bait region from proteolytic cleavage (12, 44). Hence, native a2M, a2Mox.l, and a2M,x.2 were first incubated with 200 mM methylamine for 2 h a t p H 8.0 and then reacted with pancreatic elastase for either 10 min or 1 h. As shown in Fig. lOA, methylamine induced a conformational change in a2M and a2M,x.1 as denoted by the expected mobility shift observed following native gel electrophoresis. However, no shift was detected with aZMox.2 (Fig.  1OA ). Furthermore, whereas only limited bait region hydrolysis occurred following the reaction of elastase with methylaminetreated a2M or a2M,,x.l (Fig. 10B), the bait region of a2Mox.2 remained equally susceptible to proteolytic attack before and after methylamine treatment (compare Fig. 1OB with Fig. 7A 1. Identical results were obtained when the proteinase/a2M,,, ratio was decreased to 0.1, or when trypsin was used in place of elastase as the attacking proteinase (data not shown). Thus, is incapable of undergoing the conformational changes normally associated with thiol ester scission.  5 ) or trypsin (lane 6 ) could not be detected. tr,M, is defined as the "fast" form of u,M due to its increased mobility under nondenaturing conditions (Refs. 9-11).  7 ) were incubated with an equimolar quantity of pancreatic elastase for 10 min (lanes2,5, and 8, respectively) or 60 min (lanes 3.6 , and 9, respectively) and analyzed by reduced SDS-PAGE a s described under "Experimental Procedures."

DISCUSSION
Despite the fact that leukocyte-derived halogenated oxidants remain the only known examples of physiologically relevant species capable of compromising aZM antiproteolytic activity (111, the molecular mechanisms underlying inactivation have not been previously defined. In the present study, RNCl were employed as relatively selective oxidants that inactivated aZM in a distinct, biphasic manner. While azMox-l displayed normal functional properties following the oxidation of 8 methionyl residuedsubunit, the loss of these "endogenous antioxidants" rendered the antiproteinase sensitive to subsequent RNC1-mediated dissociation and inactivation. Indeed, in the second phase of oxidative attack, the loss of 6 additional methionyl residues correlated closely with the observed changes in aZM structure and the attendant loss of function. However, attempts to link azM dissociation or inactivation to methionine oxidation alone were complicated by stoichiometric analyses which demonstrated that an additional 2 eq of RNCl had reacted with the antiproteinase. Earlier studies suggested that RNCl are highly selective oxidants that display an apparent absolute specificity for cysteinyl and methionyl residues at alkaline pH (45). Although thiol esters conceivably present a n additional target for oxidative attack, all four thiol esters/tetramer remained intact in fully inactivated azM. Interestingly, however, amino acid analyses of alkaline hydrolysates of Q~M , , .~ indicated the loss of 1-2 of the 11 trytophanyl residues detected per subunit. When aZMox-2 was subsequently examined by the more sensitive technique of second-derivative spectroscopy (31), the loss of a single tryptophanyl residuehbunit was also found to closely correlate with the inactivation process. Under acidic conditions (i.e. pH <6.5), RNCl have been reported capable of oxidizing accessible tryptophanyl residues to oxindoles with an apparent stoichiometry of -2. 0-251 (39, 40, 46, 47). Nonetheless, as originally proposed by Schecter et al. (451, tryptophanyl residues may be localized in microenvironments where their sensitivity to oxidative attack at alkaline pH is enhanced. Indeed, consistent with the predicted formation of oxindoles, aZMox.2 also contained -0.6 carbonyls/subunit. Although this value does not account for all of the oxidized tryptophan, a substoichiometric yield of carbonyls was noted during the metal-catalyzed oxidation of glutamine synthetase (48).
In addition to the oxidative events described above, a variable proportion of the azMox-z dimers appeared to be crosslinked by p-mercaptoethanol-resistant bonds. Current evidence suggests that protein multimers may be generated by dityrosine cross-links or Schiff base formation (49-52). While spectral analyses failed to detect dityrosine formation following NHzC1-mediated inactivation, we have not specifically ruled out the formation of Schiff bases. However, it should be stressed that other forms of cross-links may be formed. For example, Drozdz et al. (51) reported that a HzOz-myeloperoxidase-C1system could multimerize lysozyme even after available amino groups were acetylated via the putative formation of oxyindole cross-links. Until the mechanism of cross-link formation in azM is determined, fractional losses of unidentified moieties remain a consideration. Nonetheless, the relative selectivity of RNC1mediated oxidations coupled with the stoichiometric yield of oxidized amino acids relative to NHzCl consumed, lead us to conclude that the loss of 6 methionyl residues and a single tryptophanyl residue4 account for all of the major oxidative "hits" responsible for azM inactivation.
Following the oxidative transformation of aZMox.l into aZMox-2, the homotetramer was dissociated along its non-cova-Although tryptophan oxidation products could have generated absorbing species that masked the loss of additional tryptophan residues RNCl with oxidized azM is most consistent with the oxidation of a single by spectrophotometric analysis, the reaction of only 2 additional eq of residue (39, 40, 46, 47). lent axis into dimers that failed to entrap cognate proteinases. At first glance, it is tempting to attribute this loss of functional activity to the dissociation of oxidized azM. However, azM dimers generated artificially by treating human aZM with either urea, acid, mild reductants, or divalent cations can retain considerable functional activity (12-18, 26, 53, 54). Indeed, cadmium-induced dimers, which appear morphologically similar to the oxidized dimers generated in our study, are able to both bind and shield proteinases in a near normal fashion (13, 14,171. Furthermore, several naturally occurring dimeric azMhomologues, including human pregnancy zone protein, express antiproteolytic activity (9, 10, 55-57). Thus, loss of a2Mox-2 function appears most consistent with the inability of the oxidized dimers to undergo the major conformational changes requisite for efficient proteinase entrapment. Earlier studies have characterized chemically modified forms of azM that display losses in antiproteolytic function secondary to changes in conformational mobility (34,58, 59), but oxidative attack remains the only physiologically relevant example of this form of damage. In this regard, the relative roles that methionine and tryptophan oxidation play in perturbing azM structure and function are unknown. However, oxidation of critical methionyl residues can dissociate protein dimers presumably by altering conformation as a consequence of transforming hydrophobic methionyl residues into hydrophilic sulfoxides (60-62). The possibility that methionine oxidation directly generates aZM dimers which unmask a critical tryptophanyl residue is especially intriguing and efforts to identify the positions of the oxidized methionyl and tryptophanyl residues are underway.
To date, an increasing number of proteinase inhibitors have been shown to be sensitive to oxidative attack (e.g. al-proteinase inhibitor, az-antiplasmin, plasminogen activator inhibitor-1, and secretory leukoproteinase inhibitor), leading us to postulate that neutrophils use chlorinated oxidants to protect secreted proteolytic enzymes from premature inactivation in the extracellular milieu (1). Despite the fact that azM inactivation required a significant number of oxidative hits (i.e. 15 residues oxidizedhubunit or 60 residuedtetramer), it should be noted that only -1% of the 5804 amino acid residues found in each tetramer were affected. Indeed, al-proteinase inhibitor function, long considered one of the most sensitive targets to attack by chlorinated oxidants, is depressed by neutrophils following the oxidation of 4 methionyl residues (1, 20). Although only l of the methionyl residues is believed to be critical to loss of function (i.e. Met358 in the PI position), the oxidation of 4 of the 394 residues reinforces the contention that an attack of -1% of the amino acids remains physiologically relevant. These data are not meant to suggest, however, that all proteinase inhibitors are sensitive to chlorinated oxidants. Indeed, oxidative conditions sufficient to damage azM have little effect on al-antichymotrypsin5 or plasminogen activator-inhibitor-2 (63). Nonetheless, recent studies demonstrating the presence of inactive azM as well as increases in the methionyl sulfoxide and carbonyl content of proteins recovered from inflammatory sites in vivo lend support to the physiologic relevance of these oxidative events in a pathophysiologic setting (1, 64-66).