Biosynthesis and processing of proteinase 3 in U937 cells. Processing pathways are distinct from those of cathepsin G.

Proteinase 3 is a human polymorphonuclear leukocyte serine proteinase that degrades elastin in vitro and causes emphysema when administered by intratracheal insufflation into hamsters. Proteinase 3, stored in the azurophilic granules, is expressed in progenitor cells of myeloid origin. In the present study, the biosynthesis, processing, and intracellular transport of the enzyme was investigated in the human myelomonocytic cell line U937. Proteinase 3 is initially identified as a 35-kDa precursor and converted into the 29-kDa mature form within 3 h. By using a combination of techniques including amino-terminal sequencing, we identified the 35-kDa form as a zymogen containing an activation dipeptide but lacking the amino-terminal 25 residues, presumably the result of cleavage by a signal peptidase. Tunicamycin treatment and alkalinization of acidic cell compartments with NHCl did not prevent the processing of the proteinase 3 zymogen into the mature form, suggesting that the enzyme is targeted to the cytoplasmic granules by a mechanism other than the mannose 6-phosphate receptor. Brefeldin A inhibited the zymogen processing, suggesting that the dipeptide cleavage occurred in a post-Golgi organelle. The enzyme responsible for the removal of the dipeptide is a cysteine proteinase since E-64d, a class-specific inhibitor, prevented processing. However, treatment of cells with a dipeptidyl peptidase I inhibitor, Gly-Phe-diazomethyl ketone and with the lysosomotropic agents, NHCl and chloroquine, did not prevent dipeptide cleavage, indicating that the processing enzyme for proteinase 3 is not dipeptidyl peptidase I. In contrast, Gly-Phe-diazomethyl ketone inhibited cleavage of the dipeptide from cathepsin G. This indicates that processing of proteinase 3 is distinct from that of cathepsin G. Proteinase 3 is also processed at the COOH-terminal extension. Cleavage takes place next to Arg-222, suggesting that a trypsin-like proteinase is involved in the COOH-terminal processing.

Proteinase 3 (PR-3, 1 EC 3.4.21.76), is a third neutral serine proteinase in azurophilic granules of human polymorphonuclear leukocytes (PMNL) (1,2), distinct from elastase (HLE, EC 3.4.21.37) and cathepsin G (Cat G, EC 3.4.21.20). PR-3 degrades several structural proteins in vitro, including elastin, suggesting that it plays a role in several PMNL-mediated physiologic and pathologic events (1,3). Physiologically, the proteolytic activity of PR-3 may facilitate the movement of neutrophils from the vasculature through basement membranes at sites of inflammation (4) or may assist in the digestion of phagocytosed microorganisms (5). Pathologically, the elastolytic property of PR-3 suggests a role in the development of emphysema. This possibility is supported by the demonstration of emphysematous lesions in the lungs of hamsters following intratracheal insufflation of PR-3 (1). The prospect that PR-3 also is a mediator of airway injury is strengthened by its resistance to inhibition by secretory leukocyte protease inhibitor, the primary serine proteinase inhibitor in human upper respiratory tract (3,6).
PR-3, in addition to its proteolytic activity against extracellular matrix proteins, has a variety of other potentially important actions. PR-3 is identical to myeloblastin, which has been ascribed a central role in the control of growth and differentiation of leukemic cells (7). Recently, PR-3 was found to degrade the 28-kDa mammalian heat shock protein (8), previously linked to differentiation of normal and neoplastic cells and to cleave the nuclear factor B subunit p65 (9). PR-3 also has microbicidal activity that is independent of its serine proteinase activity (10). Perhaps most importantly, PR-3 is the antigen recognized by cytoplasmic-staining anti-neutrophil cytoplasmic autoantibodies in patients with Wegener's granulomatosis (11)(12)(13), a disease characterized by a prominent neutrophilic vasculitis. PR-3 could contribute to the pathogenesis of this disease either by inactivating complement pathway inhibitor (C1) (14) or by its presence on the surfaces of PMNL and human endothelial cells, turning these cells into targets for activation by cytoplasm-staining anti-neutrophil cytoplasmic autoantibodies (15,16). Because of these diverse roles, an indepth knowledge of factors that influence the expression of PR-3 are of substantial importance. In the present investigation, we report on the biosynthesis, processing, and intracellular transport of this enzyme. sequencing were from Applied Biosystems (Foster City, CA). EN-LIGHTNING and EN 3 HANCE spray were from DuPont NEN. All other chemicals not specifically mentioned were high quality grade from Sigma.
Identification of COOH Termini of the PMNL Serine Proteinases-PR-3, HLE, and Cat G were purified as described previously (1) and then further purified by HPLC to ensure that the preparations were free from peptides produced by autolysis. The highly purified proteinases (30 g) were suspended in 30 l of 4 M guanidine hydrochloride in 80% acetic acid, and then 30 l of iodosobenzoic acid (10 mg/ml dissolved in the same solvent) was added (10-fold excess over the protein). Iodosobenzoic acid cleaves peptide bonds on the carboxyl side of Trp (17,18), which is strategically located near the COOH termini of PR-3, HLE, and Cat G (see Fig. 1) The reaction mixture was incubated for 24 h in the dark at room temperature. The reaction was terminated by diluting with water followed by lyophilization. The peptides in the digest were fractionated by HPLC and sequenced. The sequence data was confirmed by mass spectrum analysis.
Antisera-Rabbit polyclonal antisera were raised to purified PR-3 and HLE as described previously (1). Rabbit polyclonal antisera also were raised to the synthetic peptides representing the partial NH 2terminal prepro region (Nab) and the COOH-terminal region (Cab) of PR-3 deduced from the cDNA and from sequencing peptides of the mature protein (see Ref. 7 and "Results"). For Nab, the peptide consisted of 13 residues, cA Ϫ12 LLLSGAARAAE Ϫ1 , which correspond to the putative 10 amino acids of the pre region and two amino acids of the pro region. For Cab, the peptide consisted of 11 residues, cggR 221 RVEAKGR 228 , corresponding to the putative eight amino acids of the COOH-terminal peptide extension of PR-3, and two additional glycines that were incorporated into this peptide to increase its size. An extra cysteine was added to the NH 2 -terminal side of each peptide to facilitate conjugation with the carrier protein. Similarly, peptides cS 220 EDNPCPHPRD 230 and cS 224 FKLLDQMETPL 235 were synthesized corresponding to regions of COOH-terminal peptide extensions of HLE and Cat G, respectively. Using an immunogen conjugation kit obtained from Pierce, the peptides were coupled to keyhole-limpet hemocyanin (KLH) through the NH 2 -terminal cysteine residues. Rabbits were immunized with keyhole-limpet hemocyanin-coupled peptides (100 -200 g) emulsified in Freund's complete adjuvant and then boosted with the same amount of antigen in Freund's incomplete adjuvant at 2-week intervals until a titer of Ͼ4000 was obtained. The presence of antipeptide antibodies was assessed by dot-blot analysis using the peptides linked to ovalbumin as the antigens.
Metabolic Labeling and Immunoprecipitation-Cells seeded at a den-sity of 2 ϫ 10 6 /ml were incubated in the selective amino acid-deficient medium supplemented with 10% (v/v) dialyzed heat-inactivated fetal calf serum for 1 h prior to the addition of 100 Ci/ml [ 35 S]methionine or 10-fold concentrated [ 3 H]isoleucine. After a 30 -60-min pulse, the cells were separated from the conditioned medium by centrifugation. For chase experiments, labeled cells were resuspended in complete RPMI and incubated for the indicated time periods. At the completion of the pulse or chase period, cells were transferred to ice and recovered by centrifugation at 10,000 ϫ g for 15 s. Cell pellets lysed with 1 ml of chilled radioimmunoprecipitation assay buffer, pH 8, containing 10 mM Tris/Tris-HCl, 140 mM NaCl, 1% (v/v) Nonidet P-40, 0.5% (w/v) sodium deoxycholate, 0.1% (w/v) SDS, and a proteinase inhibitor mixture (1 mM phenanthroline, 50 M 3,4-dichloroisocoumarin, 10 M E-64) (20).
Cell lysates and conditioned media were each mixed with 50 l of a 10% suspension of protein A-Sepharose CL-4B beads in radioimmunoprecipitation assay buffer and incubated for 1 h at 4°C with gentle agitation. Supernatants were recovered by centrifugation, and then 50 l of protein A beads, preincubated with 20 l of the appropriate antiserum, was added and incubated overnight at 4°C with gentle mixing. Protein A-bound immune complexes were then washed, suspended in 40 l of SDS sample buffer, boiled for 5 min, and analyzed by SDS-polyacrylamide gel electrophoresis (SDS-PAGE) and fluorography or radiosequencing.
SDS-PAGE and Fluorography-SDS-PAGE was performed (21) in 1.5-mm thick 12.5% acrylamide minigels. Gels were stained with Coomassie Blue R-250. For fluorography destained gels were treated with ENLIGHTNING for 30 min according to the manufacturer's directions. After drying, the gels were exposed to Kodak X-Omat XAR2 film at Ϫ80°C.
Radiosequence Analysis-For radiosequence analysis, immunoprecipitates were subjected to SDS-PAGE as above and then electroblotted onto a proBlott membrane using a Bio-Rad mini Trans-Blot system (22,23). Transfer time was 1 h in a buffer, pH 8.3, containing 25 mM Tris, 192 mM glycine, and 10% (v/v) methanol at 200 mA. The membrane was then stained for 1 min in 0.1% Coomassie Blue R-250 in 40% (v/v) methanol and 1% (v/v) acetic acid and destained in 50% (v/v) methanol. The dried membrane surface was sprayed evenly with three light coats of EN 3 HANCE spray. The treated membrane was wrapped in plastic and exposed to Kodak film as above at Ϫ80°C. Using the developed film as a template, appropriate bands were excised from the membrane and sequenced by Edman degradation on an automated pulse-liquid sequenator (ABI 477A, Applied Biosystem). The sequenced samples were collected and counted for 10 min in 10 ml of Optifluor scintillation fluid. The radioactivity from each cycle was plotted against the cycle number and aligned with the sequence of the appropriate proteinase.
Dipeptidyl Peptidase I (DPP-I, EC 3.4.14.1) Assay-DPP-I activity in U937 cells was assayed using Gly-Phe-␤-naphthylamide, as described by McGuire et al. (24). Briefly, the reaction mixture contained 50 mM acetate/acetic acid buffer, pH 5.5, 1 mM EDTA, 1 mM dithiothreitol, 30 mM NaCl, and 0.1 mM Gly-Phe-␤-naphthylamide in a final volume of 1 FIG. 1. PMNL serine proteinases primary structure, location of peptide sequences used for peptide antisera production, and proposed COOH-terminal cleavage sites. The amino acid sequences are deduced from the cDNA and genomic studies reported for PR-3 (7, 34, 56 -58) HLE (26,59), and Cat G (60). All of the residues are numbered starting with the first amino acid residue of the active proteinase. The residues in the prepro region are assigned negative numbers relative to the first amino acid residue. The peptide sequences used for raising antisera are shown in boldface. Using conventional protein sequencing methodology, the sequence obtained for the COOH-terminal tryptic peptide of purified PR-3 (3) and that reported for purified HLE (61) are shown in italics. The sequence obtained for the peptide resulting from the cleavage of each proteinase by idosobenzoic acid is underlined. The arrow next to Arg indicates the probable cleavage site of the COOH-terminal extension peptide.
ml. The reaction was stopped after 30 min of incubation at 37°C by adding 1 ml of 50 mM glycine-NaOH buffer, pH 10.5. The fluorescence of the liberated ␤-naphthylamine was measured in a spectrophotofluorometer set at excitation and emission wavelengths of 335 and 405 nm, respectively. The fluorescence of known concentrations of ␤-naphthylamine was used to calculate the enzyme activity.

Biosynthesis of PR-3-
In order to study the biosynthesis and processing of PR-3, pulse-chase experiments were carried out in U937 cells using [ 35 S]methionine followed by immunoprecipitation. An autofluorogram of a representative experiment is shown in Fig. 2. These studies demonstrated that after 30 min of pulse labeling (0 min chase), PR-3 is identified as a 35-kDa protein that is subsequently processed into lower molecular weight forms. By the end of a 3-h chase, the 35-kDa biosynthetic form is almost completely converted into a 29-kDa form, a size comparable with that of mature PR-3 purified from the human PMNL (1).
Characterization of Biosynthetic PR-3 Precursors-The higher molecular mass (35 kDa) of the precursor protein compared with the apparent molecular mass (29 kDa) of the later biosynthetic form or that of purified mature PR-3 suggested that the larger form may include a preprosequence on the NH 2 -terminal side and/or an extension peptide on the COOHterminal end. This possibility was explored in several ways. A comparison of the amino acid composition of the predicted sequence of PR-3 (the expected residues at position 1-229) and the determined composition of mature PR-3 (25) indicated that PR-3 has a COOH-terminal extension that is removed during processing. In order to gain insight into the processing of the COOH-terminal extension peptide, we initially determined the COOH-terminal amino acid of PR-3 using iodosobenzoic acid to generate peptides. The amino acid sequence of the smallest peptide started with Ile and ended with Arg at positions 6 and 7. Mass spectrum analysis of this peptide confirmed that the amount of arginine recovered at cycle 7 was not a residual Arg from cycle 6 and thus demonstrated that the last two residues of this peptide were Arg. This peptide matches residues 216 -222 of the deduced sequence of PR-3 cDNA (see Fig. 1), indicating that the COOH-terminal amino acid of PR-3 is Arg 222 . The results establish that PR-3 has a seven-residue COOHterminal extension that is removed during processing to the mature enzyme. In parallel experiments carried out with HLE and Cat G, we found that these proteinases also have COOHterminal arginine residues, Arg 219 and Arg 223 , respectively indicating that COOH-terminal extension peptides of HLE (SED-NPCPHPRDPDPASRTH) (26) and Cat G (SFKLLDQMETPL) (27) were removed during the processing (28). Next, U937 cells were pulse labeled for 30 min with radioactive methionine, and Nab or Cab were used for the immunoprecipitation studies. The antibodies did not precipitate the 35-kDa form, suggesting that the signal peptide and COOH-terminal extension peptide had been removed. The same results were obtained with HLE and Cat G Cab. Similar results were obtained when the cells were pulse labeled for periods as short as 5 min, suggesting that the signal peptide and the COOH-terminal extension were removed co-translationally. Alternatively, the peptide antibodies may have been unable to precipitate the precursor protein.
To further investigate the nature of the precursor processing, we used radiosequencing to identify the amino terminus of the precursor forms of PR-3. For these studies, cells were labeled with [ 3 H]isoleucine since isoleucine is at the amino terminus of mature PR-3 (see Fig. 1). The results are shown in Fig. 3. The 35-kDa form of PR-3 contained an isoleucine at position 3. Thus, the earliest identifiable biosynthetic form (35 kDa) was pro-PR-3, which consists of a dipeptide preceding the aminoterminal isoleucine of mature PR-3 (see Fig. 1). These results confirmed that the 25-amino acid signal peptide was removed co-translationally. In the later biosynthetic form of PR-3 (29 kDa), the prosegment (Ala Ϫ2 -Glu Ϫ1 ) was removed resulting in isoleucine at position one.
The Role of Glycosylation in the Processing of PR-3-Since removal of the dipeptide from the precursor protein did not account for the apparent molecular weight differences between the early 35-kDa (pro) and the late 29-kDa (mature) forms of PR-3, we next investigated the contribution of glycosylation at consensus sites (Asn-X-(Ser/Thr)) at Asn residues 102 and 147. We determined the effects of tunicamycin, which blocks the Asn-linked core glycosylation by inhibiting the addition of Nacetylglucosamine to dolichol phosphate, the first step in the formation of core oligosaccharides. Cells were preincubated with various concentrations of tunicamycin prior to pulse la- Biosynthetic forms of PR-3 were immunoprecipitated from U937 cells using PR-3 antiserum following 1 h of pulse and 4 h of chase. The immunoprecipitates were subjected to SDS-PAGE and blotted to proBlott membrane. The protein bands of interest were cut out from the membrane and sequenced using Edman degradation followed by scintillation counting of radioactivity of the product released during each cycle. The radioactivity (counts/min) released from each degradation cycle was plotted against cycle number depicted on the abscissa. Single-letter amino acid code of the pro-PR-3 or mature PR-3 sequence is shown below the cycle numbers to match the position of the labeled amino acid. beling with [ 35 S]methionine in the continued presence of the inhibitor. The results demonstrate that increasing concentrations of tunicamycin resulted in the earliest biosynthetic form of PR-3 having a molecular mass of 29 kDa (Fig. 4). Tunicamycin did not lead to secretion of PR-3 biosynthetic forms into the conditioned medium (data not shown). The results indicate that conversion of the 35-kDa form to the 29-kDa form is due largely to "trimming" of Asn-linked oligosaccharides.
Next we conducted chase experiments with [ 3 H]isoleucine in the presence of tunicamycin to explore the role of glycosylation in PR-3 trafficking and processing. The radiosequence analysis demonstrated that inhibition of glycosylation did not prevent the isoleucine shift from the third to the first position during the chase period (data not shown). Thus, glycosylation is not essential for cleavage of the pro dipeptide and by inference is not critical for transport of the protein through intracellular compartments.
Effect of Brefeldin A (BFA) on Processing of PR-3-To investigate the subcellular location of the dipeptide removal that converts the zymogen to an active form, we examined the effect of BFA on processing of PR-3. BFA, a fungal product, is known to inhibit the transport of proteins out of the Golgi apparatus by inducing resorption of the Golgi into the endoplasmic reticulum. BFA-treated cells were labeled with [ 3 H]isoleucine, chased for 4 h in the presence of the drug, and then subjected to immunoprecipitation, electrophoresis, blotting, and sequencing. The radiosequence results of PR-3 immunoprecipitated from BFA-treated cells revealed an isoleucine in the third position. This suggests that excision of the activation dipeptide from PR-3 occurs in a post-Golgi compartment. Similarly, BFA blocked removal of the activation dipeptide from HLE and Cat G (Fig. 5).

A Cysteine Proteinase(s) Is Involved in Cleavage of the Dipeptide from PR-3-Recently,
McGuire et al. (29) reported that the granules of U937 cells (analogous to azurophilic granules of PMNL) contain a cysteine exopeptidase, DPP-I, that removes the dipeptide from the Cat G zymogen to convert it to the active mature form of the enzyme. To examine the possibility that the enzyme responsible for removing the dipeptide from PR-3 might belong to the same catalytic class as DPP-I, we tested the effects of various cysteine proteinase inhibitors. Of the inhibitors tested, N-ethylmaleimide ( E-64d inhibited more than 96% within 1 h of treatment. This difference may be related to the slower entry of the former two inhibitors into the cells via pinocytosis (31,32). We examined whether inhibition of cysteine proteinases with E-64d prevented PR-3 processing. We found that not only pro-PR-3 processing but also processing of pro-HLE and pro-Cat G was abrogated by E-64d (Fig. 6). These results indicate that, like Cat G, a cysteine proteinase is involved in processing of PR-3 and HLE.
Since E-64d is known to inhibit calcium-dependent cysteine proteinases (calpains) within cells (32), we examined the effects of EGTA (1 and 5 mM), a general calcium-dependent proteinase inhibitor, and other metalloproteinase inhibitors, EDTA (1 and 5 mM) and o-phenanthroline (100 M), on the processing of PR-3. These inhibitors did not block the processing ruling out the possibility that calpain-type proteinases are involved. 1. In striking contrast, and consistent with the results obtained by McGuire et al. (29), pro-Cat G was not processed in the presence of GF-CHN 2 as evidenced by the presence of radioactive isoleucine at positions 3 and 4. Similar to PR-3, the processing of pro-HLE was not inhibited by GF-CHN 2 (data not shown). These results demonstrate that the processing of PR-3 (and HLE) is distinct from that of Cat G.

DPP-I Is Not Involved in the Processing
Influence of Lysosomotropic Inhibitors on the Processing of PR-3-DPP-I is an enzyme that is active in the acidic milieu of lysosome-like granules. Thus, to further explore the role of this enzyme in PR-3 processing, we investigated the effect of raising the pH of the granules with NH 4 Cl and chloroquine. Cells were incubated with 5 mM NH 4 Cl or 25 M chloroquine during the pulse and chase periods followed by immunoprecipitation and radiosequencing. The results shown in Fig. 7 demonstrate that NH 4 Cl does not block the proenzyme processing of PR-3, HLE, or Cat G. Distinct from the findings with NH 4 Cl, chloroquine prevented the processing of pro-Cat G, but not of pro-PR-3 or pro-HLE. These results indicate that the processing enzyme for PR-3 and HLE is insensitive to raising intragranular/lysosomal pH, providing further evidence that DPP-I is not the enzyme responsible for the dipeptide removal from pro-PR-3. The results with chloroquine also provide further evidence that the post-translational processing of PR-3 (and HLE) is different from that of Cat G. DISCUSSION We have previously reported the identification (1), biochemical characterization (3), and gene structure (34, 35) of PR-3, an elastolytic PMNL serine proteinase. The present investigation focused on the biosynthesis, processing, and intracellular transport of the enzyme. The biosynthesis studies showed that PR-3 is initially detectable as a 35-kDa form and over 3 h is converted to a 29-kDa protein, a size identical to the mature enzyme. The mass of the larger precursor was comparable with the 33-kDa size deduced from the cDNA, for PR-3 comprised 256 amino acids plus the two carbohydrate chains at the Asnlinked glycosylation sites. However, our data, which include 1) the failure to immunoprecipitate the precursor with an antibody directed to the NH 2 -terminal prepro region and 2) the radiosequence analysis that identified Ile in the third position in the 35-kDa precursor, demonstrate that removal of the 25amino acid signal peptide occurs early in the biosynthesis of the enzyme probably in a co-translational fashion. The amino-terminal cleavage occurs between Ala Ϫ3 and Ala Ϫ2 and fits the rules predicted for signal peptidases according to von Heijne (36). The signal peptide cleavage results in the earliest identi-fiable precursor being a zymogen containing a dipeptide (Ala Ϫ2 -Glu Ϫ1 ) before the NH 2 -terminal Ile of the active enzyme. This, however, leaves a disparity between the observed molecular mass of the earliest precursor (35 kDa) and the predicted molecular mass (30.6 kDa) of the glycosylated protein minus the signal peptide. This difference may be explained by the reduced mobility of the protein in SDS-PAGE due to its cationic nature and by "trimming" of the Asn-linked oligosaccharides (37).
Following trimming of the oligosaccharides, the maturation of pro-PR-3 to the biologically active proteinase proceeds via excision of the dipeptide (Ala Ϫ2 -Glu Ϫ1 ) at the acidic residue. This dipeptide is homologous to other proteinases of hematopoietic origin including HLE and Cat G (28,29,38), human mast cell chymase (39), mouse mast cell chymase (40), and mouse cytotoxic lymphocyte granzyme B (41). In the case of Cat G and suggested for the other hematopoietic cell-derived serine proteinases, the cysteine exopeptidase DPP-I has been reported to be the processing enzyme involved in the excision of the dipeptide (29). This prompted us to investigate its role in the processing of PR-3 (and HLE). When DPP-I was inhibited by GF-CHN 2 , the processing of pro-Cat G was prevented, but pro-PR-3 and pro-HLE were processed at their acidic residues. In addition, alkalinization of the secretory granules with NH 4 Cl, which should prevent the dipeptide removal by DPP-I, since the enzyme requires an acidic milieu for activity, had no effect on the processing of pro-PR-3, pro-HLE, or pro-Cat G. In contrast to the results of the present investigation, Lindmark et al. (42) reported that NH 4 Cl inhibited the processing of both HLE and Cat G in U937 cells. However, the site of inhibition in the processing pathway was not determined, and prevention of the dipeptide removal was not demonstrated.
Chloroquine, which has a similar effect on the pH of the granules as NH 4 Cl, surprisingly showed differential effects on the processing of the PMNL serine proteinases. The processing of pro-Cat G was inhibited, but that of pro-PR-3 and pro-HLE was not. This raises the possibility that the observed effect of both the DPP-I inhibitor and chloroquine on the processing of pro-Cat G may be related to the blockage of the transport of the enzyme through the subcellular organelles rather than inhibition of DPP-I and that the transport of Cat G may be different from that of PR-3 and HLE. While it is clear from our data that DPP-I is not involved in the processing of PR-3 or HLE, inhibition of the dipeptide excision by E-64d establishes that a cysteine proteinase, perhaps related to DPP-I, is involved in FIG. 6. Inhibition of cysteine proteinases with E-64d in U937 cells prevents cleavage of activation dipeptide from PR-3 zymogen. Cells were preincubated for 1 h with 0.24 mM E-64d prior to labeling with [ 3 H]isoleucine for 1 h and chased for 4 h in the continued presence of the inhibitor. At the end of chase period, cells were immunoprecipitated with proteinase-specific antisera, and the purified labeled proteins were analyzed as described in legend to Fig. 3. Note that the inhibitor blocks the cleavage of the activation dipeptides from PR-3, HLE, and Cat G. this processing step. Ongoing work in our laboratory is aimed at identifying the processing enzyme.
In the present investigation, analysis of the COOH terminus established arginyl residues as the final two amino acids of that mature PR-3 (Arg 221 -Arg 222 ). Thus PR-3 is shorter by seven amino acid residues on the COOH-terminal side than would be predicted from the cDNA. The COOH termini for HLE and Cat G are monoarginyl residues (Arg 219 and Arg 223 , respectively, Fig. 1). This implicates a basic amino acid directed endoproteinase in the processing of the COOH termini of these enzymes. Recently several mammalian subtilisin-like convertases have been identified that cleave at dibasic and monobasic sites of proproteins (43)(44)(45)(46). Among these enzymes, furin would be a candidate enzyme for processing of PMNL proteinases. However, furin is localized in the Golgi compartment (47). While studies by Salvesen and Enghild (28) on the processing of Cat G are consistent with removal of its COOH-terminal extension peptide in the Golgi, our results suggest that the COOH-terminal extension of PR-3 is likely removed co-translationally, indicating that its processing enzyme is probably localized to the endoplasmic reticulum. In addition, it has been shown that furin requires a precursor protein comprising the consensus cleavage site Arg-Xaa-(Lys/Arg)-Arg or Arg-Xaa-Xaa-Arg (47)(48)(49)(50), indicating importance for an Arg at P 4 2 posi-tion. None of the PMNL proteinases contain Arg at the P 4 position. Thus basic amino acid directed endoproteinases other than furin likely are involved in processing the PMNL proteinases. In preliminary studies, N ␣ -p-tosyl-L-lysine chloromethyl ketone and phenylmethanesulfonyl fluoride did not inhibit the processing of the COOH terminus, suggesting that the responsible enzyme, although trypsin-like, is likely not a serine proteinase.
The azurophilic granules of PMNL are considered analogous to lysosomes (51) of nonmyeloid cells. In the present investigation, we compared the targeting of PR-3 to granules with sorting of lysosomal enzymes. Following synthesis, lysosomal enzymes in nonmyeloid cells are transported from endoplasmic reticulum to Golgi, during which time high mannose side chains are modified and mannose 6-phosphate residues are added as a recognition marker for receptor targeting to the lysosome. In the acidic intralysosomal environment, the lysosomal enzyme dissociates from the receptor and the proenzyme is proteolytically processed. We found that PR-3 zymogen was not processed to the mature form when we treated cells with BFA, an agent that specifically blocks protein transport distal to the Golgi, suggesting that the PR-3 precursor is transported beyond the Golgi compartment before processing of the propetide. The cleavage of the propeptide from the PR-3 precursor in the presence of tunicamycin suggests that the transport of PR-3 to the proper cell compartment is not mannose 6-phos- 2 The nomenclature introduced by Schechter and Berger (55) is used to describe the position of amino acid in a peptide substrate. P 1 , P 2 , etc. refer to amino acid residues of the substrate (or inhibitor) in the aminoterminal direction from the scissile bond, and P 1 Ј, P 2 Ј etc. refer to amino acid residues in the carboxyl-terminal direction from the scissile bond.
The corresponding subsites of the enzyme refer to S 1 , S 2 , etc. and S 1 Ј, S 2 Ј, etc., respectively. At the end of chase period, cells were immunoprecipitated with proteinase-specific antisera, and the purified labeled proteins were analyzed as described in legend to Fig. 3. Note that NH 4 Cl did not inhibit processing of any of the proteinases. In contrast, chloroquine inhibited processing of Cat G, and partially inhibited the processing of HLE but had no effect on PR-3. phate receptor-mediated. In addition, agents that increase the pH of acid compartments did not prevent the processing of PR-3 precursor into the mature form, further evidence that the mannose 6-phosphate receptor is not required for the transport to the storage granules and processing of PR-3. These findings are consistent with results from previous investigations of various azurophilic granule proteins such as myeloperoxidase (52) and defensins (53), indicating that the sorting mechanisms are distinct from those of typical lysosomal enzymes (54).
In summary, the processing of PR-3 in U937 cells requires three proteolytic cleavages, two on the amino-terminal side and one on the carboxyl-terminal side of 256-amino acid prepro-PR-3. The initial amino-terminal cleavage results in the removal of a 25-amino acid endoplasmic reticulum-targeting signal sequence. In additional cleavages, the dipeptide (Ala Ϫ2 -Glu Ϫ1 ) propiece from the amino-terminal side preceding Ile 1 and seven-amino acid-long peptide COOH-terminal to Arg 222 are removed leaving the 222-amino acid PR-3 found in PMNL. Our results show that the processing of PR-3 in U937 cells is distinct from that of Cat G.