Molecular Cloning and Expression of Human Alveolar Macrophage Cathepsin S, an Elastinolytic Cysteine Protease*

Human alveolar macrophages (HAM) express an elastase activity of acidic pH optimum inhibitable by cysteine protease inhibitors. Recent studies indicate that the only known eukaryotic elastinolytic cysteine protease, cathepsin L, cannot completely account for this activity. In order to search for additional cysteine proteases with elastinolytic activity, low degeneracy oligonucleotide primers based on regions of strong homology among the known cysteine proteases were used to screen reverse-transcribed HAM RNA for cysteine proteases by the polymerase chain reaction. Among the cDNA sequences generated was a 493-base pair product highly homologous to bovine cathepsin 5. Screening of a HAM cDNA eukaryotic expression library with this cDNA yielded a 1.7-kilobase full-length cDNA highly homologous to bovine cathepsin S (-85% iden- tical). This cDNA predicts a 331-amino acid prepro-cathepsin. Expression of this cDNA in COS cells re- vealed the active enzyme to be a single chain 28-kDa protease, as judged by active site labeling with a novel iodinated analogue of N-(~-3-trans-carboxyoxirane-2-carbonyl)-~-leucylamido-(4-guanido)butane The recombinant enzyme to of this activity retained pH 7.0. Labeling of HAM with the active site probe revealed these cells express a 28-kDa cysteine protease, and Northern blot analysis revealed the presence of a -1.7-kilobase cathepsin S mRNA. These data establish that human macrophages express at least two cysteine proteases with elastinolytic activity. The relatively broad pH range of human cathepsin S activity suggests this enzyme may contribute to the contact-dependent elastase activity of live human alveolar macrophages.

Human alveolar macrophages (HAM) express an elastase activity of acidic pH optimum inhibitable by cysteine protease inhibitors. Recent studies indicate that the only known eukaryotic elastinolytic cysteine protease, cathepsin L, cannot completely account for this activity. In order to search for additional cysteine proteases with elastinolytic activity, low degeneracy oligonucleotide primers based on regions of strong homology among the known cysteine proteases were used to screen reverse-transcribed HAM RNA for cysteine proteases by the polymerase chain reaction. Among the cDNA sequences generated was a 493-base pair product highly homologous to bovine cathepsin 5. Screening of a HAM cDNA eukaryotic expression library with this cDNA yielded a 1.7-kilobase full-length cDNA highly homologous to bovine cathepsin S (-85% identical). This cDNA predicts a 331-amino acid preprocathepsin. Expression of this cDNA in COS cells revealed the active enzyme to be a single chain 28-kDa protease, as judged by active site labeling with a novel iodinated analogue of N-(~-3-trans-carboxyoxirane-2carbonyl)-~-leucylamido-(4-guanido)butane (E-64). The recombinant enzyme was found to be elastinolytic toward 3H-labeled elastin (bovine ligamentum nuchae) at pH 5.5 but with 26% of this activity retained at pH 7.0. Labeling of HAM with the active site probe revealed these cells express a 28-kDa cysteine protease, and Northern blot analysis revealed the presence of a -1.7-kilobase cathepsin S mRNA. These data establish that human macrophages express at least two cysteine proteases with elastinolytic activity. The relatively broad pH range of human cathepsin S activity suggests this enzyme may contribute to the contact-dependent elastase activity of live human alveolar macrophages.
Macrophages are thought to be involved in connective tissue remodeling associated with chronic inflammation, injury, and healing (1). Part of their role in remodeling is mediated by proteolytic degradation of extracellular matrix proteins. Proteolysis of extracellular matrix elements by macrophages is a cooperative process involving proteases of serine, metallo, and cysteine classes (2). Human alveolar macrophages, for example, degrade elastin by a contact-dependent *This work was supported by National Health Service Grants 35653 and 44712, and by a grant from the Council for Tobacco Research. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. mechanism involving the neutral proteases plasminogen activator and a metalloprotease, as well as acidic enzymes of the cysteine protease class (3,4). Plasminogen activator contributes to extracellular matrix metabolism possibly by exposing matrix elastin to contact zones with macrophages where elastolysis actually occurs and/or by activation of latent metalloproteases (5). We have previously reported, on the basis of inhibitor profiles, evidence that cysteine proteases are a major enzyme class involved in the contact-dependent elastolysis mediated by human lung macrophages (6).
To date only one eukaryotic cysteine protease, cathepsin L, has been demonstrated to be elastinolytic (7). Human alveolar macrophages express cathepsin L (8). However, recent studies indicate that cathepsin L alone cannot explain the elastinolytic activity of human alveolar macrophages. Macrophages obtained from cigarette smokers have much higher intracellular acidic elastase activity than those of nonsmokers, and yet the amounts of active cathepsin L are equivalent between nonsmoker and smoker cells (9). These data suggest that a second cathepsin, distinct from cathepsin L, could contribute to macrophage elastolysis. To identify novel cathepsins, we designed low degeneracy oligonucleotide primers based on regions of strong amino acid homology among the known cathepsins and screened reverse-transcribed human alveolar macrophage RNA for expressed cathepsins. Using this strategy we identified a previously uncharacterized human cathepsin, cathepsin S, and obtained the full cDNA by subsequent analysis of a macrophage cDNA library. Expression of this cDNA in COS cells revealed an enzyme of 28 kDa with elastase activity, confirming the expression of a second acidic elastase by these cells.
Primer 1 was based on the conserved active site region; primer 2 was designed from another conserved region near the 3' end of all known eukaryotic cysteine proteases (10-14). In order to minimize the degeneracy of the pool of oligonucleotides coding for the selected amino acid sequences (>lOOO-fold for each sequence), the codon usage of known human and bovine cathepsin cDNAs for these sequences was employed (10-14). Human alveolar macrophage total RNA was prepared as previously described (15). One pg of RNA was heated to 65 "C for 3 min and then reverse transcribed in 50 pl of reaction solution: 5 p1 of 5 X RT buffer (GIBCO), 2.5 pl of 10 mM dNTPs (Pharmacia LKB Biotechnology Inc.), 2 p1 of antisense primer 2 (0.4 pg/pl), 1 pl of RNAsin (GIBCO), 1.5 p1 of bovine serum albumin (GIBCO), and 600 units of Moloney murine leukemia virus (GIBCO). This mixture was kept at room temperature for 10 min, 42 "C for 1 h, and 95 "C for 5 min. PCRs were done by using 5 pl of reverse transcription mixture and 3 pl of primer 1 (0.6 pg/pl), 3 pl of primer 2 (0.4 pg/pl), 2 pl of 10 mM dNTPs, 5 pl of dimethyl sulfoxide (Sigma), 10 pl of 10 X PCR buffer (Promega), 2.5 units of Taq DNA polymerase (Promega), and HzO to 100 pl. The PCR temperature cycle was 94 "C X 1 min, 55 "C X 2 min, 72 "C X 2.5 min, repeated for 30 cycles. Fifty pl of the PCR reaction constituents were separated on a 2% agarose gel, and a region of bands of -500 bp was isolated from the gels by electroelution onto a DEAE membrane (Schleicher & Schuell). No attempt was made to obtain a "pure" cDNA band in order to maximize the recovery of multiple cathepsin cDNAs.
The recovered DNA was subcloned using the pCRlOOO TA cloning system (Invitrogen). Miniprep DNA was prepared from 55 white colonies and digested with pCRlOOO polylinker restriction enzymes. Twenty-eight of these were found to have cDNA inserts. Those 28 clones were selected for DNA sequencing with Sequenase T-7 DNA polymerase according to the supplier's recommendations (U. S. Biochemical Corp.). Two of these clones have 493-bp insert sequences that are 86% identical to bovine cathepsin S (10).
Human Alveolar Macrophuge cDNA Library Screening-A human macrophage eukaryotic expression library (pcDNA I) was commercially prepared (Invitrogen) as previously described (15). The 493-bp PCR product was labeled with [LU-~'P]~ATP by random hexamer extension (Multiprime, Amershan Corp.) and used for colony hybridization screening according to standard methods (16). Ten positive candidates were selected from the secondary screening and preliminary DNA sequencing performed. Three of the candidate clones were found to have sequences that exactly matched the 493-bp PCR product internal to the degenerate primers. Only one was in the correct orientation for expression. The coding region of this clone was subsequently fully sequenced in both strands using overlapping primers. When digested with pcDNA I polylinker restriction enzymes, BamHI and XbaI (Boehringer Mannheim), this clone had a -1.7-kb insert.
Protein Expression in COS Cells-COS-7 cells (ATCC) were subcultured overnight in 90-mm plastic tissue culture dishes in Dulbecco's modified Eagle's medium (GIBCO) containing 10% fetal bovine serum (GIBCO). Transfection was done by a DEAE dextran/chloroquine method (17). Preliminary experiments indicated that 30 pg of miniprep DNA yielded maximal expression, which was similar at 48and 72-h post-transfection. After 2-3 days of post-transfection culture, the cells were washed with phosphate-buffered saline (pH 7.4), scraped, suspended in phosphate-buffered saline, and counted. One ml of lysis buffer (40 mM sodium acetate, 1 mM EDTA, 1% Triton X-100, pH 5.5) was added per 10 million cells and incubated at 37 "C for 1 h. Cell lysates were clarified by centrifugation at 450 X g and kept on ice until assayed or labeled as described below.
Cysteine Proteinase Active Site Labeling-In order to screen macrophages for additional cysteine proteases a compound potentially reactive with the active site of all cysteine proteases and suitable for iodination was designed. An analogue of E-64, N-(~-3-trans-carboxyoxirane-2-carbonyl)-~-leucyl-2-(p-hydroxyphenyl)ethylamide (termed in this report JPM-565), was chosen. The starting material for synthesis of JPM-565 was diethyl-L-trans-epoxysuccinic acid, obtained in three steps from diethyl-L-tartrate according to the procedure of Mori and Iwasawa (18). Saponification of this material to the mono-acid was accomplished in 93% yield. Boc-Leu-2-(p-hydroxypheny1)-ethylamide was prepared in 97% yield from Bocleucine and tyramine via the mixed anhydride method. After deprotection (4 N HCl/dioxane), Boc-Leu-2-(p-hydroxyhenyl)-ethylamide was in turn coupled to the epoxy acid by the mixed anhydride method in 74% yield. Hydrolysis of the resulting ethyl ester gave 79% yield of JPM-565. This analogue of E-64 containing a phenol moiety was used as a cysteine proteinase active site probe. Preliminary experiments established that the acid elastase activity of alveolar macrophage lysates is completely abolished by 1 p~ E-64 or JPM-565. Iodination was done as follows. A glass tube was coated with IODO-GEN (Pierce Chemical Co.) as directed by the manufacturer. To this was added 10 pl of phosphate buffer (50 mM sodium phosphate, pH 5.5), 10 pl of Nalz5I (0.5 mci), and 25 pl of a 1 mM solution of JPM-565 in 50% ethanol, 50% phosphate buffer. After 10 min on ice, 455 pl of cold phosphate buffer was added. The solution was then applied to a Sep-Pack cartridge (Millipore) that had been washed with 5 ml of acetonitrile followed by 10 ml of phosphate buffer. The cartridge was extensively washed with Hz0 to remove free NalZ5I, and then the iodinated probe was eluted with 4 ml of acetonitrile. For these experiments, aliquots of eluant were frozen directly and stored until use. Concentration of active probe was determined by active site titration against purified cathepsin B (Calbiochem) (aliquots containing 10 p~ cathepsin B in 0.05% Triton X-100,20 mM sodium acetate, 1 mM EDTA, 3 mM DTT were incubated with varying amounts of probe for 1 h at 37 "C, followed by SDS-PAGE and autoradiography) and found to be 1 p~; this represents 16% recovery of initial JPM-565. Transfected COS cells were prepared for labeling by lysing at 10' cells/ml in 1% Triton X-100,40 mM sodium acetate, 1 mM EDTA, pH 5.5, for 1 h at 37 "C. For labeling, 50 p1 of lysate was made 3 mM DTT and 20 nM lZ5I-labeled JPM-565, kept at 37 "C for 1 h, then mixed 1:l with 2 X reduced sample buffer and subjected to SDS-PAGE (12% gel) and autoradiography. Human alveolar macrophages obtained by bronchoalveolar lavage (15) were lysed and labeled in a similar manner.
EZustase Assay~"[~H]Elastin (bovine ligamentum nuchae, Elastin Products, St. Louis, MO) was prepared by reductive borohydration as previously described (19). The specific activity of the preparation used in this study was 1300 cpm/pg protein. Assay plates contained -200 pg of elastin per 16-mm well (Falcon). All final assay mixtures were made 3 mM DTT and assayed in duplicate. Lysates of COS transfectants and controls (250 pl) were mixed with 250 p1 of assay buffer (0.05% Triton X-100,20 mM sodium acetate, 1 mM EDTA, pH 5.5) and incubated on plates for 24 h at 37 "C. To determine the effect of pH on elastase activity of expressed enzyme, 50-pl aliquots of lysate were mixed with 450 pl of 50 mM buffers of various pH (sodium acetate, pH 4.5, 5, 5.5; potassium phosphate, pH 6, 6.5, 7; Tris, pH 7.5,8) containing 0.05% Triton X-100 and 1 mM EDTA and assayed as above. Inhibitor studies were done by assaying 67 p1 of lysate in 433 pl of assay buffer in the presence of various inhibitors, as above. The samples were removed, centrifuged at 14,000 X g, and assayed for solubilized tritium by scintillation counting. Data are expressed as micrograms of elastin solubilized by 24 h, after correcting for background activity measured using buffer blanks.
Northern Blot Analysis-Human alveolar macrophage RNA, 20 pg/lane, was electrophoresed on a 1.2% agarose gel containing 1% formaldehyde, transferred to a nylon membrane (Genescreen Plus, Du Pont-New England Nuclear) by capillary blotting and UV-irradiated, as previously described (15). A 493-bp cathepsin S cDNA was generated by PCR and labeled with 32P as described above. A cathepsin L cDNA, nucleotide positions 692-1191 (12), was also generated by PCR from reverse-transcribed macrophage RNA and labeled. Blots were prehybridized and hybridized with the 32P-labeled PCR products, washed in 0.5 X SSC, 0.1% SDS at room temperature for 1 h and 0.1 X SSC, 0.1% SDS at 42 "C for 3 h, then autoradiographed at -80 "C.

SDS-Polyacrylamide Gel Electrophoresis and Autoradiography-
The gel system was essentially as described by Laemmli, modified as previously described (15, 20). Gels for autoradiography were stained with Coomassie Blue G-250, destained, dried, and exposed to Kodak X-OMAT film at -80 "C.

RESULTS
Human Alveolar Macrophage Cathepsin S cDNA-Based on two highly conserved regions within previously sequenced cysteine proteases, low degeneracy primers were designed and used to amplify reverse-transcribed macrophage RNA by PCR. Electrophoresis of the PCR products revealed multiple bands of -500 bp. After subcloning these DNAs as a mixture into pCRlOOO and subsequent DNA sequencing of clones containing inserts, cDNA sequences were obtained that were similar to those of the known human macrophage cysteine proteases, i.e. cathepsins B, H, and L (data not shown). In addition a 493-bp cDNA product showing 86% identity to the Human Macrophage Cathepsin S corresponding region of bovine spleen cathepsin S was observed (10). This cDNA was used to screen a human alveolar macrophage cDNA library. The entire coding region of one full-length clone has been extensively sequenced using vector and internal primers (Fig. 1). This cDNA predicts a 331amino acid preprocathepsin. Compared with available partial bovine cathepsin S sequences, it shows 85 and 87% identity at the amino acid and cDNA levels, respectively (10).
Northern Blot Analysis of Human Alveolar Macrophage Cathepsin S mRNA- Fig. 2 illustrates the reaction of the 32Plabeled cathepsin S cDNA (493-bp PCR product) with human alveolar macrophage RNA. Cathepsin S mRNA was detected as a single band migrating at -1.7 kb, consistent with the size of human cathepsin S predicted by cDNA cloning and that reported for bovine cathepsin S mRNA (10). The specificity of the cathepsin S cDNA hybridization was verified by reprobing the filter with a 32P-labeled cathepsin L cDNA (not shown). This revealed a single, distinct band slightly lower than cathepsin S, consistent with the reported size of cathep-sin L mRNA, -1.5 kb (12), and indicated that the cathepsin S cDNA was not detecting other cathepsin mRNAs under the hybridization conditions employed. Among the known cathepsins, cathepsin L has the closest sequence homology both within the region probed (57% identical) and overall (49%) to cathepsin S. COS Cell Expression of Human Macrophage Cathepsin S-Miniprep DNA (30 pg) from clones containing 1.7-kb inserts was transfected into subconfluent COS cells. Expression of enzyme was assessed by active site labeling of COS cell lysates (pH 5.5) with lZ51-labeled JPM 565. Labeled lysates were then analyzed by SDS gel electrophoresis and autography of dried gels. As illustrated in Fig. 3, untransfected COS cells or COS cells undergoing sham transfection with DNA clones containing either no insert or cathepsin S cDNA inserted in the reverse orientation express a cysteine protease of 33 kDa. Although no attempt was made to characterize this enzyme, the size of the labeled band is identical to that expressed by human alveolar macrophages, which has been isolated and

BCS:
V a l U e t Leu His Asn Leu Glu His Ser net Gly Wet His Ser Tyr Asp Leu Gly U e t Aln 81s Leu Gly Aap 82   Human alveolar macrophage total RNA (20 pg) was separated on an agarose/formaldehyde gel, blotted to a nylon membrane, and probed using the 32P-labeled cathepsin S PCR fragment. Cathepsin S mRNA is detected as a band of -1.7 kilobases. Positions of the 18 and 28 S ribosomal RNAs used to estimate the mRNA size were determined by UV shadowing.

FIG. 3. Coexpression of cathepsin S and elastase activity in COS cells.
COS cells were transfected with cathepsin S plasmid, or controls, as described under "Experimental Procedures." After 3 days of culture post-transfection, cells were lysed in 1% Triton X-100, 40 mM sodium acetate, 1 mM EDTA, pH 5.5, labeled with the cysteine proteinase active site probe 1251-labeled JPM-565, and subjected to SDS-PAGE under reducing conditions and autoradiography. Human alveolar macrophages (obtained from a cigarette smoker) were lysed and labeled in a similar manner for comparison. Additional COS cell lysate was assayed for elastase activity against insoluble, tritiated elastin as described in the text. Lune I , COS cells in routine culture; lune 2, COS cells subjected to transfection protocol with no plasmid added; lanes 3 and 4, transfection with cathepsin S plasmid, reverse orientation; lane 5, transfection with cathepsin S plasmid, correct orientation.  Effect of pH and Protease Inhibitors on Recombinant Cathepsin S Elastase Activity-Data in Table I show the effect of various known inhibitors of cysteine proteases on the elastinolytic activity of human cathepsin S expressed in COS cells. The effect of these inhibitors on the elastinolytic activity of alveolar macrophage lysates is shown for comparison. All tested inhibitors of cysteine proteases blocked enzyme activity to some degree. Interestingly, the relative potency of the various diazomethylketone inhibitors is the same as that previously reported for these inhibitors against purified human cathepsin L (21). Also of note, the relative potency of these inhibitors on the acidic elastase activity of alveolar macrophage lysates is the same as for cathepsin S, although the percentage of inhibition is lower for the macrophage lysates than the COS cell transfectants.
Previously reported data indicate that bovine cathepsin S has activity over a relatively broad range of pH (22). T o determine the pH dependence of the elastinolytic activity of recombinant human cathepsin S, aliquots of COS cell transfectant lysates were mixed with various pH buffers, and residual elastase activity was assessed in a 24-h assay. Although activity decreased with increasing pH of the buffers, elastinolytic activity -25% of maximum was still measurable at pH 7 and was completely inhibited by 1 p~ E-64.

DISCUSSION
Molecular cloning of human cathepsin S has established its expression by human alveolar macrophages and defined functional features of the enzyme not previously appreciated.
Macrophage cathepsin S cDNA predicts a 331-amino acid preprocathepsin. Expression of this cDNA (Fig. 3) indicates the active enzyme is 28 kDa. This contrasts with the reported size of 23 kDa for active bovine cathepsin S (22). Based on these size considerations, inspection of Fig. 1 indicates active human cathepsin S has an additional 35-40 amino acids amino-terminal to the start residue of bovine cathepsin S. Of note, this amino-terminal sequence contains the only acceptor site for N-linked glycosylation (residue 104) in the entire coding region. Such sites are thought to be necessary for posttranslation mannosylation important to channeling of lysosomal enzymes into lysosomes (23). Since the reported sequences for active bovine cathepsin S do not show an N-glycosyl acceptor site, it seems likely that incompletely processed forms of bovine cathepsin S with sequence homology to the human enzyme exist and promote lysosomal channeling. Currently no sequence information for precursor forms of bovine cathepsin S are available to test this prediction.
Demonstration of the elastinolytic activity of human cathepsin S suggests this enzyme is closely related to cathepsin L, the other known cathepsin with elastinolytic activity. Indeed human cathepsin S shows closest sequence homology to cathepsin L (49%) and less homology to cathepsins H (31%) and B (23%). This functional and structural similarity is also reflected by the inhibitor profile of recombinant cathepsin S (Table I). Cbz-Phe-Phe-CHN, is a good inhibitor of cathepsin L but only weakly inhibits cathepsin B (24). Cbz-Phe-Phe-CHN, was observed to be an effective inhibitor of cathepsin S. Of note, in prior studies of elastin degradation by human cathepsin L, it was observed that no detectable elastin degradation occurred above a pH of 6.0 (8). In contrast, using similar buffer conditions to that of the prior work, we found cathepsin S to retain 25% of maximal elastinolytic activity at We have previously reported that lysates (pH 5.5) of human alveolar macrophages have elastase activity (9), and data shown in Fig. 3 confirm that these lysates contain a 28-kDa cysteine protease. Moreover the lysate elastase activity and the 28-kDa enzyme co-purify,' suggesting this 28-kDa enzyme is cathepsin S. However, data in Table I show similarities and differences between the inhibitor profile of recombinant cathepsin S and macrophage lysate elastase activities. Activities in both the COS cell lysates containing cathepsin S and in the macrophage lysates are completely inhibited by the classspecific inhibitor, E-64, and by cystatin C. In contrast, only 20% or less of the macrophage lysate activity was inhibitable by either Cbz-Tyr-Ala-CHN2 or Cbz-Tyr-Tyr(o-butyl)-CHN,, whereas over 75% of the cathepsin S activity was blocked by either of these inhibitors. We suspect that these differences reflect larger amounts of active cathepsin B in macrophage lysates (Fig. 3) as compared with the COS cell lysates (Fig. 3, lane 5 ) competing for binding with these inhibitors. It is also possible that there are additional cysteine proteases in the macrophage lysates. However, cathepsin S appears to be the major elastase active in these macrophage lysates.
Demonstration of the elastinolytic activity of cathepsin S establishes that human alvedlar macrophages express at least three elastinolytic proteases: a 92-kDa gelatinase (25), cathepsin L, and cathepsin S. These cells thus have the potential to mediate focal elastin degradation in tissues, and such degradation has been documented in vitro. The close physical pH 7-8. association between macrophages and sites of lung injury in cigarette smokers suggests this potential may also be expressed in vivo (26,27). Future studies are needed to explore the extent to which cigarette smoking amplifies cathepsin S activity in the microenvironment of lung macrophages over that of nonsmoker cells and to define the relative contribution of cathepsin S to the overall elastinolytic activity of these cells. The finding that human cathepsin S has appreciable elastinolytic activity at both acidic and neutral pH suggests that, at least among the cathepsins, this enzyme may be uniquely tailored to function at surface contact sites of cells and elastin where the interfacial pH could be expected to vary between acidic and neutral.