Characterization of Two Azurophil Granule Proteases with Active-site Homology to Neutrophil Elastase*

Much of the tissue damage associated with emphysema and other inflammatory diseases has been attributed to the proteolytic activity of neutrophil elastase, a major component of the azurophil granule. Recently, two additional azurophil granule proteins with NH2-terminal sequence homology to elastase were isolated (Gabay, J. E., Scott, R. W., Campanelli, D., Griffith, J., Wilde, C., Marra, M. N., Seeger, M., and Nathan, C. F. (1989) Proc. Natl. Acad. Sci. U.S.A. 86, 5610-5614) and designated azurophil granule protein 7 (AGP7) and azurocidin. Azurocidin and AGP7 represent significant protein components of the azurophil granule, together comprising approximately 15% of the acid-extractable protein as judged by reverse-phase high performance liquid chromatography analysis. AGP7 migrates on sodium dodecyl sulfate-polyacrylamide gel electrophoresis as four distinct glycoforms of molecular mass 28-34 kDa, whereas azurocidin exhibits three predominant bands with molecular mass of 28-30 kDa. Treatment of intact azurophil granules with [3H]diisopropyl fluorophosphate resulted in labeling of elastase, cathepsin G, and AGP7, whereas azurocidin was not labeled. Tryptic mapping of 3H-labeled AGP7 allowed us to identify and sequence the active-site polypeptide that has 70% identity to elastase over 20 residues. The active site peptide of azurocidin was also identified by sequence analysis of tryptic fragments and showed 65% identity to the active site of elastase. Surprisingly, the catalytic serine of azurocidin is replaced by glycine, explaining its inability to label with [3H]diisopropyl fluorophosphate. Thus, we have identified two azurophil proteins closely related to neutrophil elastase, one of which has apparently lost its proteolytic activity due to mutation of the catalytic serine.

Much of the tissue damage associated with emphysema and other inflammatory diseases has been attributed to the proteolytic activity of neutrophil elastase, a major component of the azurophil granule. Recently, two additional azurophil granule proteins with NH2terminal sequence homology to elastase were isolated (Gabay, J. E., Scott, R. W., Campanelli, D., Griffith, J., Wilde, C., Marra, M. N., Seeger, M., and Nathan, C. F. (1989) Proc. Natl. Acad. Sci. U. S. A. 86, 5610-5614) and designated azurophil granule protein 7 (AGP7) and azurocidin. Azurocidin and AGP7 represent significant protein components of the azurophil granule, together comprising approximately 15% of the acid-extractable protein as judged by reversephase high performance liquid chromatography analysis. AGP7 migrates on sodium dodecyl sulfate-polyacrylamide gel electrophoresis as four distinct glycoforms of molecular mass 28-34 kDa, whereas azurocidin exhibits three predominant bands with molecular mass of 28-30 kDa. Treatment of intact azurophil granules with [3H]diisopropyl fluorophosphate resulted in labeling of elastase, cathepsin G, and AGP7, whereas azurocidin was not labeled. Tryptic mapping of 'H-labeled AGP7 allowed us to identify and sequence the active-site polypeptide that has 70% identity to elastase over 20 residues. The active site peptide of azurocidin was also identified by sequence analysis of tryptic fragments and showed 65% identity to the active site of elastase. Surprisingly, the catalytic serine of azurocidin is replaced by glycine, explaining its inability to label with [3H]diisopropyl fluorophosphate. Thus, we have identified two azurophil proteins closely related to neutrophil elastase, one of which has apparently lost its proteolytic activity due to mutation of the catalytic serine.
Neutrophils bind to target organisms via Fc and complement receptors and internalize them in phagosomes. The phagosomes then fuse with primary (azurophil) granules to form phagolysosomes in which granule components destroy the ingested organism (1, 2). The proteolytic activity within primary granules has usually been ascribed to two well characterized serine proteases, cathepsin G and neutrophil elastase. In addition to their physiologic role in the phagolysosome, these granule proteases have important extracellular effects, particularly at sites of inflammation (3). For example, the destruction of lung tissue in emphysema results from an * The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "aduertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. $ To whom correspondence should be addressed: Invitron Corporation, 301 Penobscot Dr., Redwood City, CA 94063.
imbalance between the level of neutrophil elastase and that of al-proteinase inhibitor, its major physiological inhibitor (4, 5). The elastaselotl-proteinase inhibitor balance may also be involved in other inflammatory disease states in which neutrophil accumulation is accompanied by tissue damage such as adult respiratory distress syndrome (6, 7).
The contribution of other proteases to neutrophil-mediated inflammatory states is unclear. Recently a granule protease designated proteinase 3 was purified and described by Kao et al. (8). Like elastase, proteinase 3 degrades elastin in vitro and causes extensive tissue damage when administered to hamsters by tracheal instillation. In addition, Gabay and coworkers (9) recently isolated two novel neutrophil proteins, azurocidin and azurophil granule protein 7 (AGP7),' with sequence homology to elastase but with unknown proteolytic function. Here we report structural and functional characterization of these two proteins, which may be important in evaluating the proteolytic component of neutrophil function.

EXPERIMENTAL PROCEDURES
Materials-Diisopropyl fluorophosphate (DFP) was purchased from Sigma, [3H]DFP from Amersham Corp., polyvinylidene difluoride from Millipore, 4-vinylpyridine from Aldrich, and TPCK-treated trypsin from Worthington. Isolation and Subcellular Fractionation of Granulocytes-Granulocytes were isolated from buffy coats as described previously (10). Isolated granulocytes (2 X 10' cells/ml in phosphate-buffered saline, pH 7.4) were treated with 5 mM DFP for 15 min at 4 "C. The cells were washed, resuspended in lysis buffer (10 mM Pipes, pH 6.8, 100 mM KCl, 3 mM NaCl, 3.5 mM MgCl*), and disrupted by nitrogen cavitation (Parr Instrument Co., Moline, IL). Granules were isolated from the postnuclear supernatant by Percoll density gradient centrifugation as described previously (11). The azurophil granule fraction was collected, and Percoll was removed from the granules by centrifugation (180,000 X g, 2 h). When isolated granules were labeled with [3H]DFP, treatment of whole cells with the inhibitor was omitted. Isolated granules from 2.3 x 10' cells were treated with 20 &i [3H] DFP (180 mCi/mmol) for 1 h at room temperature before freezing at -70 "C for later extraction.

Preparation of Granule
Extracts-Isolated granules were lysed by five freeze/thaw cycles on dry ice/ethanol and after addition of an equal volume of 100 mM glycine, pH 2.0, were extracted with vigorous agitation for 40 min at room temperature. The acid extract was centrifuged at 30,000 x g for 20 min and at 200,000 X g for 30 min to obtain a soluble fraction for chromatography.

RESULTS
Acid extraction of intact granules and reverse-phase HPLC of extracted proteins revealed 10 major protein peaks (Fig. 1). Analysis of the peaks by NHz-terminal sequencing and sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) revealed a relatively simple complement of proteins, similar to that described previously for an azurophil granule membrane fraction (9) but with notable differences. One new peak (peak 4a) was identified and found to contain primarily eosinophil-derived neurotoxin. Peak 5 consisted of lysozyme with little major basic protein detected. The predominant species in each peak is derived from neutrophils with the exception of peaks 3 and 4a. Only these peaks increased in size when granule preparations from eosinophil-rich granulocytes were analyzed (data not shown).
The predominant species comprising peak 6 (azurocidin) and peak 7 (AGP7) each exhibited NH?-terminal sequence homology to neutrophil elastase, cathepsin G, and other members of the trypsin superfamily (9). Table I  son of the first 20 residues of azurocidin and AGP7 to other granule-associated leukocyte proteases. AGP7 is identical to cathepsin G and the cytotoxic T cell proteases (granzymes) across a stretch of 8 residues from Pro9 to Alal and shows 70% homology to granzyme B. Azurocidin is less similar (35%) to the granzymes than AGP7, and of the granule proteases, it is most clearly related to elastase. Analysis of peak 4 (cathepsin G), peak 6 (azurocidin), peak 7 (AGP7), and peak 8 (elastase) by SDS-PAGE is shown is Fig. 2. AGP7 migrated as four discrete bands in the M, range 28,000-34,000. The relative size and staining intensity were similar to those seen for glycoforms of elastase. For azurocidin, the three bands around M, 30,000 were somewhat diffuse, indicating a more heterogenous range of glycoforms.
Both azurocidin and AGP7 represent major components of the azurophil granule extract, comprising 11.0 f 0.4% and 4.3 + 0.5%, respectively of extracted protein as determined by computer integration of reverse-phase HPLC peaks (triplicate analyses). Although yields of granules and extracted proteins were somewhat variable among preparations, recoveries of azurocidin and AGP7 were typically around 200 and 100 ng/ lo6 cells, respectively, compared with approximately 150 ng/ lo6 cells for elastase. Our detection of these proteins as major components may be attributable to extraction and purification under acidic conditions. We have observed a pronounced tendency of both azurocidin and AGP7 to precipitate when an acid extract is adjusted toward neutral pH or when the salt concentration of the extract is raised (data not shown).
To determine whether AGP7 and azurocidin are active as packaged in granules, we treated isolated azurophil granules with [3H]DFP, a membrane-permeant reagent that covalently labels the active-site serine of serine proteases. Proteins were extracted from labeled granules and fractionated by size exclusion and reverse-phase HPLC. Analysis of reverse-phase column fractions identified cathepsin G, AGP7, and elastase but not azurocidin as labeled peaks (Fig. 3). The extent of labeling of AGP7 was comparable to that of elastase and cathepsin G.
Specific labeling of AGP7 with [3H]DFP was demonstrated directly by SDS-PAGE of reverse-phase-purified AGP7 followed by electroblotting onto polyvinylidene difluoride membranes. Identical blotted lanes were prepared; one was sliced and counted for the presence of [3H]DFP, and the other was Coomassie stained and subsequently utilized for sequence analysis. The major peak of radioactivity was associated with an approximately 30-kDa band that yielded the NH* terminal sequence expected for AGP7 (Fig. 4).
To determine the active site sequence of AGP7, [3H]DFPlabeled material was reduced, alkylated, and subjected to digestion with trypsin. A single 3H-labeled tryptic peptide with the sequence shown in Table II was isolated by reversephase HPLC. Comparison of this sequence with the activesite region of neutrophil elastase revealed extensive homology, including the expected alignment of the catalytic serine at position 173 (Table III).
Although peak 6 was not labeled by [3H]DFP, we were able to identify a putative active site peptide by sequence analysis of the major tryptic fragments. This peptide has strong homology with the active site of elastase and AGP7 so that alignment of the fragment was unequivocal (Table III). The residue that aligns with the catalytic Ser'73 of elastase is Gly' of this peptide. The absence of the active-site serine in azurocidin likely accounts for the failure of peak 6 to label with [3H]DFP and identifies azurocidin as a proteolytically inactive member of the trypsin superfamily.
The insolubility of AGP7 near neutral pH noted above complicated efforts to determine substrate specificity. Because serine proteases are generally resistant to denaturing treatments, we tested reverse-phase-purified AGP7 and elastase for amidolytic and esterolytic activity against a panel of chromogenic substrates (Table IV). Only the p-nitrophenyl ester of N-benzyloxycarbonylvaline was cleaved by AGP7, and no amidolytic activity was identified. Elastase retained activity against its characteristic substrates. Azurocidin and AGP7 are primary granule proteins that show extensive NH*-terminal sequence similarity to proteases of the trypsin superfamily. Azurocidin appears to be most closely related to elastase, whereas AGP7 is 70% identical to granzyme B, a serine protease found in the granules of cytotoxic T lymphocytes.
Labeling of AGP7 with [3H]DFP demonstrates that it is an  N-Succiny:Ala-Ala-Val-Ala-pNAs N-Succiny:Ala-Ala-Pro-Leu-pNA N-Succiny:Ala-Ala-Pro-Phe-pNA N-Succinyl- active serine protease. The extent of AGP7 labeling with this reagent was comparable to that of elastase and cathepsin G. We were able to take advantage of the covalent modification of AGP7 with DFP to identify a single tryptic fragment containing the active-site serine. The sequence of the labeled peptide was identical to the active-site sequence of human neutrophil elastase at 10 of the first 13 residues of the peptide with two of the three mismatches being conservative substitutions.
Azurocidin, on the other hand, did not incorporate [3H] DFP. Further structural analysis of the molecule revealed sequence strongly similar to the active site of neutrophil elastase. In azurocidin, however, the catalytic serine is replaced by glycine. Thus azurocidin, like the LY subunit of nerve growth factor (12), is a member of the trypsin superfamily in which the catalytic serine is absent. It was shown previously that azurocidin is antimicrobial in vitro (9). The data presented here demonstrating the inability of azurocidin to function as a protease support the idea that azurocidin acts primarily as an antimicrobial protein in uivo. It should be noted, however, that cathepsin G is a potent antimicrobial protein (2, 3, 9) but retains its proteolytic activity. Both azurocidin and AGP7 appear to be among the more abundant proteins of azurophil granules. We recovered up to 400 ng of azurocidin and 200 ng of AGP7/106 neutrophils after reverse-phase HPLC. The total elastase content of neutrophils is estimated to be 1,500 ng/106 cells (3) compared with our recoveries that range up to 300 ng/106 cells, Considering expected losses during granule isolation, extraction, and chromatography, the data are consistent with granules containing azurocidin and AGP7 in amounts comparable to that of elastase. That these molecules have not been described previously may result from poor solubility of azurocidin and AGP7 at neutral pH under nondenaturing conditions. The propensity of AGP7 to precipitate when the pH is raised or when the NaCl concentration is increased has hampered purification attempts under nondenaturing conditions for substrate and inhibitor studies. The apparent insolubility of both proteins also raises the possibility that they may be associated with the granule membrane.
Kao and co-workers (8) have reported the existence of a neutrophil protein, proteinase 3, that cleaves cY-naphthyl acetate, a nonspecific esterase substrate, and is distinct from neutrophil elastase and cathepsin G. Purified proteinase 3 degrades elastin with an efficiency equivalent to that of neutrophil elastase (8). Proteinase 3, like AGP7, was found in multiple isoforms in the molecular weight range around 30,000 and was extracted from granules under strongly acidic conditions. Initial purification attempts using procedures described by these workers resulted in complete loss of AGP7. The relationship of proteinase 3 to AGP7 is under investigation.
Azurocidin shows antimicrobial activity in vitro and probably plays a role in bacterial killing within the phagolysozome. AGP7 is only slightly microbicidal, however, suggesting that this protein has a different function. One interesting possibility is that AGP7 contributes to the proteolytic degradation at inflammatory sites that has previously been attributed solely to elastase and cathepsin G.