Methionine Sulfoxide and Proteolytic Cleavage Contribute to the Inactivation of Cathepsin G by Hypochlorous Acid

Using myeloperoxidase and hydrogen peroxide, activated neutrophils produce high local concentrations of hypochlorous acid (HOCl). They also secrete cathepsin G, a serine protease implicated in cytokine release, receptor activation, and degradation of tissue proteins. Isolated cathepsin G was inactivated by HOCl but not by hydrogen peroxide in vitro. We found that activated neutrophils lost cathepsin G activity by a pathway requiring myeloperoxidase, suggesting that oxidants generated by myeloperoxidase might regulate cathepsin G activity in vivo. Tandem mass spectrometric analysis of oxidized cathepsin G revealed that loss of a peptide containing Asp108, which lies in the active site, associated quantitatively with loss of enzymatic activity. Catalytic domain peptides containing Asp108 were lost from the oxidized protein in concert with the conversion of Met110 to the sulfoxide. Release of this peptide was blocked by pretreating cathepsin G with phenylmethylsulfonyl fluoride, strongly implying that oxidation introduced proteolytic cleavage sites into cathepsin G. Model system studies demonstrated that methionine oxidation can direct the regiospecific proteolysis of peptides by cathepsin G. Thus, oxidation of Met110 may contribute to cathepsin G inactivation by at least two distinct mechanisms. One involves direct oxidation of the thioether residue adjacent to the aspartic acid in the catalytic domain. The other involves the generation of new sites that are susceptible to proteolysis by cathepsin G. These observations raise the possibility that oxidants derived from neutrophils restrain pericellular proteolysis by inactivating cathepsin G. They also suggest that methionine oxidation could render cathepsin G susceptible to autolytic cleavage. Myeloperoxidase may thus play a previously unsuspected role in regulating tissue injury by serine proteases during inflammation.

Neutrophils play a key role in host defense by migrating to sites of infection, where they phagocytose invading microorganisms (1). After a neutrophil encloses a microbe, the resulting phagosome fuses with granules containing microbicidal and digestive enzymes to form a phagolysosome. The azurophilic granules of neutrophils are rich in serine proteases, including cathepsin G and neutrophil elastase, which play critical roles in killing bacteria (2)(3)(4).
Although such proteases are important for tissue homeostasis and host defense, an imbalance between proteases and their inhibitors is implicated in tissue damage during inflammation (5). It is therefore likely that plasma-derived inhibitors of cathepsin G, such as ␣ 1 -antichymotrypsin and ␣ 1 -antitrypsin, help limit proteolysis (6). However, traditional enzyme kinetics cannot fully explain the regulation of proteolysis by neutrophils. Thus, even in the presence of plasma-derived protease inhibitors, neutrophils produce evanescent "quantum bursts" of pericellular proteolytic activity (7). Cathepsin G could thus promote local proteolysis of a range of protein and peptide substrates, including cytokines, neutrophil chemoattractants, clotting factors, extracellular matrix, and G protein-coupled protease-activated receptors (8 -10).
The substrate-binding site of cathepsin G lies in a cleft between two six-stranded ␤-barrel domains (11,12). It contains the catalytic triad Asp 108 -His 64 -Ser 201 , which forms a charge relay system that cleaves the peptide bond of proteins. The active site of a protease contains a series of subsites (S) 1 that interact with particular residues in peptide substrates (P). The site for amidolysis is defined as the peptide bond between amino acids P 1 and P 1 Ј in the substrate peptide P n . . . P 2 , P 1 , P 1 Ј, P 2 Ј . . . P n Ј (13). The primary specificity pocket, S 1 , which accommodates the side chain of the P 1 amino acid of a peptide substrate, provides the critical interaction site for serine proteases (12). Cathepsin G has the unusual ability to recognize either large, hydrophobic amino acids (Phe, Leu, and Met) or basic amino acids (Lys and Arg) in the P 1 site of target peptides and proteins (12, 14 -16), indicating that it has both chymotrypsin-and trypsin-like specificity.
Phagocytosis of pathogens by neutrophils triggers a burst of cyanide-insensitive oxygen consumption by the phagocyte NADPH oxidase (17,18). This membrane-associated electron transport system generates superoxide, which dismutates to hydrogen peroxide (H 2 O 2 ). Another major component of azurophilic granules, myeloperoxidase, can use H 2 O 2 and chloride ions to generate hypochlorous acid (HOCl), a potent cytotoxic oxidant (Reaction 1) (19,20), Myeloperoxidase is the only known enzyme that produces HOCl in humans at plasma concentrations of halide ion (21).
Recent studies with mice deficient in myeloperoxidase or cathepsin G strongly support the proposal that these enzymes are necessary for killing fungi and Gram-negative bacteria (21,22). However, reactive intermediates produced by myeloperoxidase have also been implicated in tissue damage at sites of inflammation (23). Indeed, oxidation products identical to those generated by the enzyme in vitro have been found in proteins, lipids, and nucleic acids in inflamed human tissue (24 -27), indicating that myeloperoxidase and HOCl might be pathogenic. Moreover, phagocytes store myeloperoxidase and cathepsin G in the same secretory compartment, and degranulation of these components is likely to create high local concentrations of both enzymes near the cell surface (28), where NADPH oxidase is located. Indeed, secreted cathepsin G binds tightly to the cell surface of neutrophils (2)(3)(4). Oxidative modulation of the inflammatory response may be a general regulatory mechanism, because HOCl can inactivate a wide range of products that are secreted by neutrophils (29 -31). However, the molecular details of protein oxidation by reactive intermediates remain poorly understood.
In the current studies, we demonstrate that HOCl, a specific product of myeloperoxidase, potently inactivates cathepsin G by a pathway that involves oxidation of a specific methionine residue and autolytic cleavage. This finding raises the possibility that myeloperoxidase might restrain the activity of cathepsin G near the surface of neutrophils. If so, it might help limit, rather than promote, inadvertent proteolysis of host proteins and tissue.

Materials
Cathepsin G (from human sputum, salt-free, lyophilized) was obtained from Elastin Products Co. (Owensville, MO). NaOCl, H 2 O 2 , CF 3 COOH, and HPLC grade CH 3 CN were obtained from Fisher Scientific. Unless otherwise indicated, all other materials were purchased from Sigma.

Methods
Oxidation Reactions-Reactions were carried out at 37°C for 30 min in buffer A (10 mM phosphate-buffered saline, containing 138 mM NaCl, and 2.7 mM KCl, pH 7.4) with 1 M or the indicated concentrations of cathepsin G or 400 M TIQNDIMLLQLSR (Biopeptide, San Diego, CA). Reactions were initiated by adding oxidant (HOCl or H 2 O 2 ) and terminated by adding a 10-fold mole excess (relative to oxidant) of L-methionine. Concentrations of HOCl and H 2 O 2 were determined spectrophotometrically (⑀ 292 ϭ 350 M Ϫ1 cm Ϫ1 and ⑀ 240 ϭ 39.4 M Ϫ1 cm Ϫ1 , respectively (32,33)).
Human Neutrophils-Human neutrophils were isolated from EDTAanti-coagulated blood by buoyant density centrifugation using Polymorph-Prep (Robbins Scientific, Sunnyvale, CA) (34). Neutrophils were washed twice at 4°C by centrifugation in buffer B (Hanks' balanced salt solution, pH 7.4; prepared without calcium chloride, magnesium chloride, magnesium sulfate, sodium bicarbonate, and phenol red (Invitrogen); 100 M diethylenetriaminepentaacetic acid was included to inhibit metal-catalyzed reactions (35)). Neutrophils were resuspended in buffer B at 37°C and immediately used for experiments. Cathepsin G (1 M) was exposed for 1 h to neutrophil (1 ϫ 10 6 cells/ml) in buffer B at 37°C. Neutrophils were activated with 200 nM phorbol 12-myristate 13-acetate. Reactions were terminated by pelletting the cells by centrifugation at 4°C. Supernatants were concentrated under vacuum and digested with trypsin for mass spectrometric analysis.
Cathepsin G Enzymatic Activity-Following the addition of L-methionine to scavenge oxidants, the activity of cathepsin G (70 nM) was assayed in buffer C (0.1 M Tris-HCl, 0.5 M NaCl, pH 7.4) using 510 M N-succinyl-AAPF-p-nitroanilide (suc-AAPF-NA) as substrate (36). Control experiments demonstrated that L-methionine did not affect the activity of cathepsin G. Suc-AAPF-NA (6.4 mM) was initially solubilized in Me 2 SO. Samples were incubated in individual wells of a 96-well microtiter plate at 37°C after adding suc-AAPF-NA, and the initial rate of change in absorbance was monitored at 410 nm (SpectraMax 190, Molecular Devices, Sunnyvale, CA).
To determine cathepsin G activity in neutrophils, quiescent and activated neutrophils were sonicated and then incubated with 510 M suc-AAPF-NA for 60 min at 37°C in buffer C (pH 7.4). Each reaction mixture was clarified by centrifugation, and the initial rate of change in absorbance of the supernatant was determined at 410 nm (37). Western blotting was performed on proteins separated by SDS-PAGE using an antibody specific for cathepsin G (37). For proteolysis studies, oxidized or native TIQNDIMLLQLSR (400 M) was incubated at 37°C with cathepsin G (2 M) in buffer A. To inactivate cathepsin G, enzyme was incubated with 3.5 mM phenylmethylsulfonyl fluoride (PMSF; stock solution 35 mM in ethanol) in buffer A at 0°C for 30 min, and then extensively dialyzed against buffer A (containing 10% ethanol) at 4°C.
Mice Deficient in Myeloperoxidase-All procedures involving animals were approved by the Washington University Animal Studies Committee. Myeloperoxidase-deficient mice were generated by targeted mutagenesis (38). Mice were maintained in a specific pathogen-free barrier facility with a 12-h light/dark cycle and provided with water and food ad libitum. Wild type and mutant mice were littermates derived from interbreeding of heterozygous mice, were in the C57/BL6 genetic background, and were sex-and age-(8 -10 weeks) matched. Neutrophils were isolated from bone marrow and the peritoneum as described previously (37).
HPLC-Synthetic peptides and cathepsin G proteolytic products were separated on a reverse-phase column (Vydac C18 MS column, 25 ϫ 2.1 mm i.d., Grace Vydac, Hesperia, CA) at a flow rate of 0.3 ml/min using a Beckman HPLC system (Fullerton, CA). Peptides were detected by monitoring absorbance at 214 nm. Peptides were eluted using solvent A (0.06% CF 3 COOH in H 2 O) and solvent B (0.06% CF 3 COOH in 80% CH 3 CN and 20% H 2 O) with a linear gradient of 10% to 60% solvent B over 60 min.
LC-ESI-MS-Cathepsin G was incubated overnight at 37°C with sequencing grade modified trypsin (Promega, Madison, WI) at a ratio of 25:1 (w/w) cathepsin G/trypsin in buffer D (1 M urea, 50 mM NH 4 HCO 3 ). Before trypsin digestion, cathepsin G was reduced with dithiothreitol and alkylated with iodoacetamide. Digestion was halted by acidifying with CF 3 COOH to a final pH of 2-3. LC-ESI mass spectrometric analyses (39) were performed in the positive ion mode with a Finnigan Mat LCQ ion trap instrument (San Jose, CA) coupled to a Waters 2690 HPLC system (Milford, MA). Peptides were separated on a reversephase column (Vydac C18 MS column, 25 ϫ 2.1 mm i.d.) at a flow rate of 0.2 ml/min using solvent C (0.2% HCOOH in H 2 O) and solvent D (0.2% HCOOH in 80% CH 3 CN and 20% H 2 O). Peptides were eluted using a linear gradient of the following: 10% to 40% solvent D over 50 min, and then 40% to 70% solvent D over 10 min for tryptic digests; 10% solvent D for 5 min, and then 10% to 50% solvent D over 55 min for cathepsin G proteolysis of native or oxidized cathepsin G; and 10% to 50% solvent D over 60 min for synthetic peptides. The electrospray needle was held at 4500 V. The sheath gas, nitrogen, was set at 80 units. The collision gas was helium. The temperature of the heated capillary was 220°C.

Cathepsin G Is Inactivated by HOCl but Not by H 2 O 2 -To
assess whether reactive intermediates generated by phagocytes can directly influence the proteolytic activity of cathepsin G, we exposed the enzyme to HOCl, H 2 O 2 , or the myeloperoxidase-H 2 O 2 -Cl Ϫ system for 30 min at 37°C at neutral pH in a physiological buffer containing plasma concentrations of chloride, sodium, and phosphate. After adding methionine to scavenge any residual oxidant, we determined whether the enzyme could degrade suc-AAPF-NA, a synthetic peptide substrate (33). Cathepsin G exposed to increasing concentrations of HOCl progressively lost its proteolytic activity, which was inhibited by ϳ50% at a 10:1 mole ratio of HOCl to protein (Figs. 1, A and B, and 2A). Higher concentrations of HOCl completely inhibited proteolysis. In striking contrast, the same range of H 2 O 2 concentrations did not affect the proteolytic activity cathepsin G (Figs. 1A and 2A). Cathepsin G exposed to the complete myeloperoxidase-H 2 O 2 -Cl Ϫ system also lost activity (Fig. 1A). Inactivation required each component of the reaction mixture: myeloperoxidase, H 2 O 2 , and Cl Ϫ (data not shown). The IC 50 for HOCl and peroxide in the complete myeloperoxidase system were ϳ10:1 and 20:1 (mol: mol), respectively (Fig. 1A). Inactivation of cathepsin G by the myeloperoxidase system was blocked by catalase, a scavenger of H 2 O 2 , and azide or 3-aminotriazole, two inhibitors of heme enzymes (data not shown). Methionine, a potent scavenger of HOCl, also blocked the reaction. These results demonstrate that inactivation of cathepsin G by myeloperoxidase requires active enzyme, Cl Ϫ , and H 2 O 2 .
Oxidative Inactivation of Cathepsin G Is Quantitatively Associated with Loss of a Tryptic Digest Peptide Containing Asp 108 of the Catalytic Triad-To determine whether oxidation of specific amino acid residues and/or enhanced proteolysis might contribute to the oxidative inactivation of cathepsin G, we digested unmodified and oxidized enzyme with trypsin and identified the resulting peptides through LC-ESI-MS. We used high concentrations of cathepsin G (17 M) for these experi-ments to facilitate MS analysis of the protein. It is important to note, however, that inactivation of cathepsin G depended only on the mole ratio of oxidant to protein (Fig. 1, B and C). Thus, when 200 nM cathepsin G was exposed to 3 M HOCl (a 15:1 mole ratio of oxidant to protein), the activity of the proteinase was inhibited by ϳ85%. These observations indicate that physiologically plausible concentrations of HOCl, but not H 2 O 2 , can convert cathepsin G to a catalytically inactive form.
The trypsin digest of native cathepsin G contained 18 peptides that covered 70% of the sequence of the native protein.
Due to poor retention on the HPLC column, we were unable to identify small peptides and single amino acid residues. When cathepsin G was exposed to HOCl (15:1, oxidant/protein, mol/ mol) and digested with trypsin, LC-ESI-MS analysis revealed eight new peaks of material (Table I). MS/MS analysis indicated that the eight peptides were derived from residues 35-48, 104 -116, and 132-149 of cathepsin G. Five of the peptides (derived from all three regions of the protein) exhibited an increase of 16 atomic mass units, suggesting that each one had gained 1 oxygen atom (Table I). MS/MS analysis demonstrated that methionine was oxygenated in four of these peptides (Metϩ16) and that tryptophan was oxygenated in the fifth (Tryϩ16). The sixth peptide had gained 32 atomic mass units, and its methionine had been converted to a sulfone, while the seventh and eighth peptides (derived from residues 132-149) had lost 2 or 4 atomic mass units, respectively. We recently showed that oxidative cross-linking of adjacent tryptophan and glycine residues creates these modifications (40,41). In contrast, we observed only a very low level of one oxidized peptide (residues 104 -116) in native cathepsin G or cathepsin G that was exposed to H 2 O 2 . These observations indicate that HOCl, but not H 2 O 2 , extensively oxidizes methionine residues in cathepsin G. It should be noted that HOCl reacts much more rapidly than H 2 O 2 with thioethers (42-44). M) was incubated with HOCl at the indicated mole ratio. C, the indicated concentration of cathepsin G was incubated with HOCl (15:1, mol/mol). All reactions were carried out for 30 min at 37°C in buffer A (10 mM phosphate-buffered saline, pH 7.4). Reactions were initiated by adding oxidant (HOCl or H 2 O 2 ) and terminated by adding a 10-fold mole excess (relative to oxidant) of L-methionine. Control experiments demonstrated that L-methionine did not affect enzyme activity. The activity of cathepsin G was assessed using suc-AAPF-NA as the substrate. Results represent the means and standard deviations of triplicate determinations. Similar results were observed in three independent experiments.

FIG. 2. Proteolytic activity (A) and abundance of peptide TIQNDIMLLQLSR (B) of cathepsin G exposed to HOCl or H 2 O 2 .
Cathepsin G (17 M) was incubated with HOCl or H 2 O 2 at the indicated mole ratio and its enzymatic activity was assessed as described in the legend to Fig. 1. A, enzymatic activity of cathepsin G. B, TIQNDIM-LLQLSR in a tryptic digest of reduced, alkylated cathepsin G was quantified by LC-ESI-MS. Results represent the means of duplicate determinations from three independent experiments.
To investigate the relationship between inactivation of cathepsin G by HOCl and the appearance of oxidation products in the enzyme, we focused on TIQNDIMLLQLSR, one of the tryptic digest peptides containing the Asp 108 of the catalytic triad (Table I). Cathepsin G exposed to increasing concentrations of oxidant was digested with trypsin and analyzed by LC-ESI-MS. At low concentrations of HOCl (5-20:1, mol/mol, oxidant/protein), loss of the native peptide TIQNDIMLLQLSR associated quantitatively with loss of enzymatic activity (Fig. 2). In striking contrast, cathepsin G was not inactivated by the same concentrations of H 2 O 2 ( Fig. 2A). Moreover, there was no loss of TIQNDIMLLQLSR in cathepsin G exposed to H 2 O 2 (Fig. 2B). These results implicate oxidative modification of TIQNDIM-LLQLSR in the loss of enzymatic activity of cathepsin G exposed to low mole ratios of HOCl.
HOCl Oxidizes Specific Methionine Residues to Generate Unique Proteolytic Cleavage Sites in Cathepsin G-Because loss of the peptide TIQNDIMLLQLSR (residues 104 -116) associated strongly with loss of enzymatic activity, we used reconstructed ion chromatograms (Fig. 3) and MS/MS analyses ( Fig. 4) to characterize this region of cathepsin G after oxidation with HOCl and digestion with trypsin. Peptides TIQNDI(Mϩ16)LLQLSR, TIQNDI(Mϩ16)LL, and NDI(Mϩ16)LL were observed in cathepsin G that had been oxidized with HOCl but not in native enzyme or cathepsin G exposed to H 2 O 2 (Figs. 3 and 4). All three peptides contained methionine sulfoxide (Metϩ16). Importantly, two resulted from cleavage at a P 1 site (Leu 112 ) recognized by cathepsin G but not by trypsin (which recognizes Arg and Lys). The third, NDI(Mϩ16)LL, was cleaved at Gln-Asn at the N terminus of the peptide as well as at Leu 112 in the P 1 site. Because native cathepsin G has not been reported to recognize Gln in the P 1 site, HOCl appears to generate new proteolytic cleavage sites by oxidizing specific methionine residues in cathepsin G. Because cathepsin G can cleave proteins and peptides at P 1 sites that contain leucine residues and because we observed only TIQNDI(Mϩ16)LL and NDI(Mϩ16)LL in the oxidized protease, oxidation of Met 110 could generate a site in cathepsin G that the enzyme can cleave. Importantly, all three peptides also contained Asp 108 , which lies in the catalytic triad of the enzyme, suggesting that proteolytic cleavage might help inactivate the protease.
Loss of TIQNDIMLLQLSR in cathepsin G exposed to low concentrations of HOCl occurred in parallel with the appearance of TIQNDI(Mϩ16)LLQLSR, TIQNDI(Mϩ16)LL, and NDI(Mϩ16)LL (Fig. 5). It is noteworthy that the relative abundances of the oxidized peptides derived from TIQNDIM-LLQLSR decreased when the mole ratio of HOCl to cathepsin G was Ͼ20 (Fig. 5). At high mole ratios of oxidant, other processes, such as random fragmentation of the peptide backbone by HOCl, may contribute to inactivation of cathepsin G and loss of precursor peptide (42). Collectively, these observations indi-cate that when cathepsin G is exposed to low concentrations of HOCl, oxidation of the region containing Asp 108 of the catalytic triad likely accounts for loss of enzymatic activity. The conversion of methionine 110 to its sulfoxide is likely to be an important event in this process. Moreover, we observed non-tryptic cleavage sites in cathepsin G exposed to HOCl, suggesting that oxidation generates unique autolytic or proteolytic cleavage sites in the proteinase that contribute to oxidative inactivation of the enzyme.
Cathepsin G Exposed to HOCl Releases an Oxidized Catalytic Domain Peptide by a Pathway Sensitive to Inhibition by PMSF-To determine whether oxidation promotes proteolysis of intact cathepsin G, we exposed the enzyme to HOCl (15:1, oxidant/protein, mol/mol) and analyzed the reaction mixture with LC-ESI-MS (Fig. 6). Because we did not digest the protein with trypsin, any peptides detected in the reaction mixture presumably were generated by cathepsin G itself. Under these conditions, we readily detected an ion of m/z 734.2, the anticipated m/z of the peptide NDI(Mϩ16)LL, which is derived from the catalytic site of the enzyme (Fig. 6B). MS/MS analysis confirmed the identity of the peptide. Formation of this peptide was directly proportional to the concentration of HOCl when the mole ratio of oxidant to protein was increased from 0 to 10 TABLE I Detection of modified peptides in cathepsin G exposed to HOCl Cathepsin G (17 M) was exposed to HOCl (255 M) for 30 min at 37°C in buffer A (10 mM phosphate-buffered saline, pH 7.4). Reactions were initiated by adding HOCl and terminated by adding L-methionine (2.6 mM). Oxidized cathepsin G was reduced, alkylated, and digested with trypsin, and the tryptic digest was analyzed with LC-ESI-MS and MS/MS.  3. LC-ESI-MS analysis of cathepsin G exposed to HOCl. Cathepsin G was incubated for 30 min at 37°C in buffer A alone or buffer A supplemented with HOCl (15:1, mol/mol, oxidant/protein). Reactions were initiated by adding oxidant and terminated by adding a 10-fold mole excess of L-methionine. Cathepsin G was then reduced, alkylated, and digested with trypsin, and the tryptic digest was analyzed with LC-ESI-MS. Reconstructed ion chromatograms of tryptic products of (A) native cathepsin G and (B) HOCl-exposed cathepsin G. ( Fig. 6D). At higher mole ratios of oxidant, the amount of peptide released from oxidized cathepsin G decreased. Treating the enzyme with PMSF, a phosphonate derivative that irreversibly sulfonylates the serine residue at the active site of serine proteases, prior to oxidation blocked the release of the oxidized peptide (Fig. 6C). Production of oxidized peptide was directly proportional to the concentration of cathepsin G in the reaction mixture (data not shown). Moreover, the product yield of the oxidized peptide was dependent on the mole ratio of HOCl to cathepsin G but was independent of the cathepsin G concentration. These results imply that proteolysis was occurring by a unimolecular reaction pathway. Thus, oxidized cathepsin G may proteolytically cleave itself, releasing NDI(Mϩ16)LL, which contains Asp 108 of the catalytic triad. The correlations between the loss of precursor peptide, the appearance of oxidized peptides, and the loss of enzymatic activity raise the possibility that proteolytic cleavage contributes to the inactivation of cathepsin G that has been exposed to HOCl. These observations further suggest that conversion of the thioether side chain of Met 110 to sulfoxide generates unique autolytic cleavage sites in cathepsin G that contribute to oxidative inactivation of the enzyme.
High concentrations of HOCl have been proposed to oxidatively fragment proteins by a reaction pathway that may involve formation of N-chloramides of the peptide bond, yielding a random mixture of peptides (45). To explore whether this mechanism participates in the oxidative inactivation of cathepsin G, we incubated the enzyme alone or with varying concentrations of HOCl for 5 min at 37°C in a physiological buffer at neutral pH. After we exposed cathepsin G, either native or pretreated with PMSF, to a high concentration of HOCl (100:1, mol/mol, oxidant/protein), we observed no intact protein on SDS-PAGE (data not shown). This observation suggests an oxidative mechanism that randomly cleaves cathepsin G into a mixture of lower molecular weight proteins that would be difficult to visualize on SDS-PAGE (45). The ⑀ amino group of lysine residues exposed to high concentrations of HOCl has been proposed to undergo deamination (46), which may also contribute to loss of staining of oxidized proteins and peptides by Coomassie Blue. It is important to note that a much higher concentration of HOCl (Ͼ50:1) was required for the apparent random fragmentation of cathepsin G than for enzyme inactivation. Thus, this pathway is not likely to contribute directly to cathepsin G inactivation at low mole ratios of oxidant to enzyme (Ͻ20:1).
Detection of Proteolytically Released Peptide NDI(Mϩ16)LL in Cathepsin G Exposed to Activated Neutrophils-To determine whether oxidants generated by neutrophils might help regulate the activity of cathepsin G, we incubated human neutrophils for 1 h at 37°C in a physiological buffer (pH 7.4) supplemented with cathepsin G (1 M). The autolytic, oxidized peptide NDI(Mϩ16)LL was readily detected by LC-ESI-MS in the supernatant of cells activated with phorbol 12-myristate 13-acetate (Fig. 7A). The identity of the peptide was confirmed using MS/MS analysis (Fig. 7B). A low level of the peptide was detected with quiescent cells, but production of the peptide increased markedly when the neutrophils were activated with phorbol 12-myristate 13-acetate (Fig. 7C). Generation of the peptide was inhibited completely by methionine (a scavenger of HOCl), catalase (a scavenger of H 2 O 2 ), or azide (a heme poison). Moreover, we detected the same oxidized peptide when cathepsin G was exposed to the myeloperoxidase-H 2 O 2 -Cl Ϫ system. These observations strongly suggest that HOCl generated by the myeloperoxidase system of activated neutrophils oxidizes cathepsin G, promoting cathepsin G proteolysis and release of the oxidized peptide NDI(Mϩ16)LL.

Activated Neutrophils Isolated from Myeloperoxidase-deficient Mice Have Higher Cathepsin G Activity Than Those from
Wild-type Mice-To further determine whether oxidants generated by myeloperoxidase might help regulate cathepsin G in vivo, we compared cathepsin G activity in two groups of neutrophils. The first group was harvested from bone marrow of wild-type and myeloperoxidase-deficient mice; these cells were expected to be relatively quiescent. The second group consisted of neutrophils recruited into the peritoneum of wild-type and myeloperoxidase-deficient mice with glycogen and then activated by challenging each group of animals with intraperitoneal Klebsiella pneumoniae. It is important to note that this latter approach should result in phagocytosis of bacteria together with activation of NADPH oxidase and secretion of myeloperoxidase and cathepsin G into the phagolysosome. After harvesting the neutrophils, we monitored the activity of FIG. 5. Quantification of peptide TIQNDIMLLQLSR and its oxidized products in a tryptic digest of cathepsin G exposed to HOCl. Cathepsin G was incubated with HOCl at the indicated mole ratio for 30 min at 37°C in buffer A. Reactions were initiated by adding oxidant and terminated by adding L-methionine. The enzymatic activity of native or oxidized cathepsin G was assessed using suc-AAPF-NA as the substrate (Fig. 2A). In parallel studies, cathepsin G was reduced, alkylated, and digested with trypsin, and the tryptic digest was ana- Detection of autolytically released peptide NDI(M؉16)LL in cathepsin G exposed to HOCl. Native (B) or PMSF-treated (C) cathepsin G was exposed to HOCl (15:1, mol/mol, oxidant/protein) for 60 min at 37°C in buffer A. In parallel studies, cathepsin G alone was incubated in buffer A (A). The reaction mixture was then analyzed directly by LC-ESI-MS with selected reaction monitoring of ions of m/z 734.2. Note that under these conditions only peptides derived from the proteolytic activity of cathepsin G itself will be detected. D, HOCl dependence for relative abundance of peptide NDI(Mϩ16)LL in cathepsin G exposed to HOCl. cathepsin G in supernatants of sonicated cells, using the specific peptide substrate suc-AAPF-NA (36).
The resting neutrophils derived from bone marrow of wildtype and myeloperoxidase-deficient cells had almost identical cathepsin G activity (1.40 Ϯ 0.05 versus 1.47 Ϯ 0.07 units/mg protein; n ϭ 3) (Fig. 8). In contrast, cathepsin G activity was 50% lower in neutrophils activated in vivo and then isolated from wild-type mice as in those activated in myeloperoxidasedeficient mice (0.65 Ϯ 0.04 versus1.29 Ϯ 0.07 units/mg protein; n ϭ 3; p Ͻ 0.01, Student's t test). Western blotting with a specific antibody demonstrated that the levels of immunoreactive cathepsin G were comparable in resting or activated wildtype and myeloperoxidase-deficient neutrophils (data not shown). These observations suggest that myeloperoxidase modulates cathepsin G activity in activated neutrophils, perhaps by oxidatively inactivating the protease. One important oxidant is likely to be HOCl. However, myeloperoxidase generates both chlorinating and nitrating intermediates in vivo (47,48), and it is possible that a variety of oxidants contribute to the inactivation of cathepsin G in our animal studies.
Oxidized Methionine Residues Direct the Regiospecific Cleavage of Synthetic Peptides That Mimic the Catalytic Domain of Cathepsin G-Detection of methionine sulfoxide in the peptide released from cathepsin G after HOCl exposure raised the possibility that oxidation generates new proteolytic sites in the enzyme. To explore this hypothesis, we exposed the synthetic peptide TIQNDIMLLQLSR to HOCl (1:1, mol/mol) for 30 min at 37°C in buffer A and then determined its susceptibility to cleavage by cathepsin G. HPLC analysis revealed a single major peak of new material. LC-ESI-MS and MS/MS analyses identified this product as TIQNDI(Mϩ16)LLQLSR. In contrast, no oxidation products were observed after H 2 O 2 treatment, even with a 20-fold mole excess of H 2 O 2 . Thus HOCl, but not H 2 O 2 , can convert the methionine in the synthetic peptide to a sulfoxide under our experimental conditions.
Oxidation of TIQNDIMLLQLSR depended on both HOCl concentration and the length of incubation (data not shown). Resting neutrophils were isolated from the bone marrow of myeloperoxidase-deficient and wild type mice (37). To prepare activated cells, neutrophils were recruited into the peritoneum of mice using glycogen. After 4 h, the mice were infected intraperitoneally with K. pneumoniae (4 ϫ 10 8 CFUs), and the cells were isolated 1 h latter (37). Cathepsin G activity in cell lysates were assessed using suc-AAPF-NA as the substrate. Results represent the means Ϯ S.D. values of three independent experiments.
The reaction of the peptide with HOCl reached completion in Ͻ1 min, indicating that exposing TIQNDIMLLQLSR to low concentrations of HOCl rapidly produced a high yield of TIQNDI(Mϩ16)LLQLSR. In contrast, the yield of Metϩ16 in TIQNDIMLLQLSR decreased when the concentration of HOCl was high, as did the total amount of intact and oxidized peptides detectable in the reaction mixture. This decreased yield was perhaps due to HOCl-dependent stochastic cleavage of peptide bonds (45). We observed similar results with peptides derived from TIQNDIMLLQLSR of cathepsin G that had been exposed to high HOCl concentrations (Fig. 5).
To investigate the proteolysis of TIQNDIMLLQLSR by cathepsin G, we incubated the peptide with enzyme at 37°C in a physiological buffer at neutral pH (Fig. 9). HPLC analysis revealed three major products (Fig. 9B), which MS/MS analysis identified as TIQNDIM, TIQNDIML, and TIQNDIMLL. These results indicate that cathepsin G cleaves peptides and proteins at Leu-Gln, Leu-Leu, and Met-Leu, as has been shown previously (15).
To determine whether oxidation can generate additional cleavage sites in the peptide, we exposed TIQNDIMLLQLSR to HOCl (1:1, mol/mol), quenched the reaction with methionine, and exposed the mixture of native and oxidized peptides to cathepsin G. HPLC analysis of the reaction mixture revealed that TIQNDIM, TIQNDIML, and TIQNDIMLL were derived from the native peptide. In addition, we observed two major peaks and one minor peak of new material, which MS/MS analysis identified as TIQNDI(Mϩ16)L, NDI(Mϩ16)LL, and TIQNDI(Mϩ16)LL (Fig. 9C). In contrast, no new peaks of material were observed when the peptide was exposed to a much higher concentration of H 2 O 2 (20:1, mol/mol). These results show that TIQNDI(Mϩ16)LLQLSR contains two of the cleavage sites that cathepsin G recognizes in the native peptide: Leu-Gln and Leu-Leu. However, the oxidized peptide was resistant to cleavage at (Metϩ16)-Leu. Importantly, it had a new cleavage site, at Gln-Asn (Figs. 9C and 10). Thus, when HOCl oxidizes the synthetic peptide TIQNDIMLLQLSR, it generates the same unique proteolysis site as when it oxidizes cathepsin G.
We used reverse-phase HPLC to quantify the influence of time on product yields when native and oxidized TIQNDIM-LLQLSR were incubated with cathepsin G (Fig. 9). The oxidized peptide was degraded more rapidly than the native peptide, suggesting that conversion of Met to Metϩ16 increases the susceptibility of the peptide to proteolysis (Fig. 9A). With native TIQNDIMLLQLSR, the major proteolytic product after short incubation periods was TIQNDIMLL (Fig. 9B). This peak rapidly declined as TIQNDIM appeared. The level of TIQNDIML remained relatively constant during the incubation. These observations suggest that TIQNDIMLLQLSR is cleaved by cathepsin G at Leu-Gln to yield TIQNDIMLL, which in turn is cleaved at Leu-Leu and then Met-Leu to produce TIQNDIML and then TIQNDIM (Fig. 11).
The progress curve for proteolysis of oxidized TIQNDI(Mϩ16)LLQLSR by cathepsin G was different (Fig. 9C) from that for the native peptide. As incubation time increased, the yield of all three cleaved oxidized peptides initially increased. The product yields were TIQNDI(Mϩ16)LL Ͼ TIQNDI(Mϩ16)L Ͼ NDI(Mϩ16)LL. With prolonged incubation, the concentration of TIQNDI(Mϩ16)LL gradually declined as the concentrations of NDI(Mϩ16)LL and TIQNDI(Mϩ16)L gradually increased. These observations suggest that cathepsin G can readily cleave the oxidized peptide at either Leu-Gln or Leu-Leu (Fig. 11). The oxidized peptide, but not TIQNDIMLLQLSR, is also cleaved at Gln-Asn (Figs. 10 and 11). These observations indicate that cathepsin G can cleave only the oxidized peptide at Gln-Asp and that different residues in native and oxidized TIQNDIMLLQLSRs are susceptible to proteolytic cleavage by cathepsin G. DISCUSSION Cathepsin G plays a critical role in host defense, cytokine release, and receptor activation, but dysregulation of its proteolytic activity has been implicated in tissue injury (2-4, 8 -10). Our observations indicate that one potential mechanism for limiting its activity involves HOCl, a specific product of myeloperoxidase. We found that isolated cathepsin G could be inactivated by HOCl but not H 2 O 2 . Moreover, when we isolated neutrophils from the peritonea of mice infected with K. pneumonia, cathepsin G activity was lower in wild-type neutrophils than in myeloperoxidase-deficient neutrophils, suggesting that oxidants derived from myeloperoxidase can regulate the activity of the proteinase in vivo. Tandem MS analyses of tryptic digests of cathepsin G exposed to HOCl demonstrated that enzyme inactivation associated strongly with loss of the peptide containing Met 110 . Importantly, the crystal structure of cathepsin G demonstrates that Met 110 resides in close proximity to Asp 108 , a key component of the enzyme's catalytic triad (11,12). Moreover, loss of the peptide containing Met 110 occurred in concert with the appearance of peptides containing (Metϩ16) 110 . Thus, loss of enzymatic activity apparently involved conversion of methionine to its sulfoxide in a single localized region of cathepsin G. Our observations may be physiologically relevant, because we found that activated myeloperoxidase-deficient neutrophils had more cathepsin G activity than wild-type neutrophils during inflammation.
Oxidation of Met 110 may contribute to cathepsin G inactivation by at least two distinct mechanisms. The first involves the introduction of unique proteolytic sites into the enzyme by alterations in local (and perhaps distant) structure. We found that low concentrations of HOCl promoted the degradation of cathepsin G, as assessed by the appearance of the peptide NDI(Mϩ16)LL containing (Metϩ16) 110 . Importantly, this oxidized peptide also contained catalytic Asp 108 (Fig. 12). Pretreatment with PMSF, a phosphonate that potently inhibits serine proteases, inhibited the appearance of this oxidized peptide. Moreover, we detected NDI(Mϩ16)LL when cathepsin G was exposed to activated neutrophils, suggesting that HOCl generated by myeloperoxidase could play a physiolog-ical role in oxidizing the proteinase. These findings suggest that oxidation of methionine residues generates proteolytic cleavage sites that permit cathepsin G to be inactivated. The production of NDI(Mϩ16)LL was linearly dependent on the concentration of cathepsin G, suggesting that oxidized cathepsin G autolytically digested itself, releasing the oxidized peptide.
Oxidation of Met 110 in concert with disruption of the charge relay system formed by Asp 108 -His 64 -Ser 201 may provide a second mechanism for inactivating cathepsin G. The crystal structure of cathepsin G reveals that Met 110 is in close proximity to catalytic Asp 108 (11,12). In turn, the latter residue is juxtaposed with His 64 , which serves as the general base that activates the hydroxyl group of Ser 201 (Fig. 12). This raises the possibility that introduction of a sulfoxide into the tightly packed local environment of Asp 108 , His 64 , and Ser 201 disrupts the charge relay system that is critical for proteolysis by cathepsin G. It is worth noting that oxidation of a single methionine residue also inactivates subtilisin, a bacterial serine protease (49,50). This family of proteinases uses the Asp-His-Ser catalytic triad, although both the primary sequence and the FIG. 12. Met 110 and the catalytic triad of cathepsin G. Met 110 (red) lies next to Asp 108 in the crystal structure of cathepsin G (11). The relative positions of the residues in the catalytic triad (Asp 108 -His 64 -Ser 201 , cyan) must be maintained for the enzyme to retain proteolytic activity. Thus, converting the methionine residue to a sulfoxide may disturb the tightly packed local environment of Asp 108 , His 64 , and Ser 201 , facilitating proteolysis of NDI(Mϩ16)LL (green) and disrupting the enzyme's charge relay system. FIG. 11. Proposed reaction pathways for cleavage of native and oxidized TIQNDIMLLQLSR by cathepsin G. Model system studies (Figs. 9 and 10) indicate that the oxidized peptide, but not the native peptide, is cleaved at Gln-Asn by cathepsin G. The same cleavage site is apparent in cathepsin G exposed to HOCl. These observations indicate that different residues in native and oxidized TIQNDIMLLQLSR are susceptible to cleavage by cathepsin G. They also suggest that oxidation of methionine residues generates proteolytic cleavage sites that permit cathepsin G to be inactivated. three-dimensional structure of subtilisin are completely different from that of cathepsin G (11,49). Thus, oxidation of a specific methionine residue may disrupt the tightly packed environment of the charge relay system in both subtilisin and cathepsin G.
Model system studies indicate that, of all the amino acid side chains, the thiol and thioether groups of cysteine and methionine react most rapidly with HOCl (K 2 ϭ 3 ϫ 10 7 M Ϫ1 s Ϫ1 and 4 ϫ 10 7 M Ϫ1 s Ϫ1 , respectively (42,44)). Therefore, these two residues might be the first to be targeted when HOCl oxidizes proteins. Consistent with this proposal, our study of a synthetic peptide, TIQNDIMLLQLSR, which mimics the tryptic digest peptide from cathepsin G that contains Asp 108 and Met 110 , demonstrated that HOCl rapidly converts the methionine residue to the sulfoxide in near quantitative yield. In contrast, thiols react slowly with H 2 O 2 (K 2 Ͻ 5 ϫ 10 1 M Ϫ1 s Ϫ1 (43)), and peroxide did not oxidize TIQNDIMLLQLSR under our experimental conditions. When this peptide was oxidized with HOCl and then incubated with cathepsin G, it was converted to NDI(Mϩ16)LL, the same peptide that appears after HOCl oxidizes intact cathepsin G. Thus, alteration of the same residue in the peptide and the oxidized protein may confer susceptibility to cleavage by cathepsin G. Collectively, these observations indicate that oxidation of methionine residues can direct the regiospecific cleavage of peptides and proteins by cathepsin G. These findings may be of broader significance, because we recently showed that neutrophil elastase, another serine protease that is abundant in neutrophils, also degrades itself as it is oxidatively inactivated by HOCl (51).
Based on these observations, we propose the following model for the oxidative regulation of the proteolytic activity of cathepsin G. When the enzyme is exposed to low concentrations of HOCl, Met 110 is oxidized to methionine sulfoxide, which disrupts the interactions of the catalytic triad and generates unique autolytic cleavage sites in the intact protein at Gln 106 -Asn 107 and Leu 112 -Gln 113 . Proteolytic cleavage of this region releases NDI(Mϩ16)LL, which contains Asp 108 . Other methionine residues may also be oxidized, and these oxidations may also promote autolysis and loss of proteolytic activity. When cathepsin G is exposed to high concentrations of HOCl, however, other oxidative processes, including random fragmentation of the peptide backbone, degrade the protein (45,46).
Many lines of evidence indicate that activated phagocytic cells inflict oxidative tissue injury in humans (1,5,20,23). However, the results of clinical trials of antioxidants in the prevention of human disease have generally been disappointing (52). This observation raises the possibility that oxidants such as HOCl are also involved in suppressing inflammation. Our demonstration that HOCl restrains the proteolytic activity of cathepsin G may have important implications for understanding the role of reactive intermediates in limiting tissue damage. Indeed, we have recently shown that HOCl inactivates matrilysin (MMP-7) by oxidatively cross-linking adjacent tryptophan and glycine residues in the catalytic domain of the enzyme (39,40). This inactivation mechanism is distinct from the well studied mechanisms involving tissue inhibitors of metalloproteinases. Our findings suggest that local, pericellular production of HOCl by phagocytes is a physiological mechanism for governing proteinase activity during inflammation. The failure of antioxidants to prevent human inflammatory diseases in clinical trials may in part reflect a beneficial regulatory effect of oxidants in inflamed tissue.
In conclusion, our studies establish a potential role for HOCl as a regulator of the proteolytic activity of cathepsin G, a serine protease implicated in host defense, cytokine release, and tis-sue injury. The underlying mechanism involves the oxidation of a specific methionine residue, which in turn may disrupt the catalytic charge relay system and introduce proteolytic cleavage sites into the enzyme. This complex interplay between the oxidative and proteolytic systems of neutrophils raises the possibility that oxidants protect hosts from protease-mediated tissue degradation.