Jun 29, 1987 - We thank Fred Woessner, Jr. (University of Miami Medical. School) forthe ... A. M., Trump, B. F. & Harris, C. C. (1985) Science 227,. 1174-1179.
Proc. Nati. Acad. Sci. USA
Vol. 84, pp. 6725-6729, October 1987 Biochemistry
Human skin fibroblast stromelysin: Structure, glycosylation, substrate specificity, and differential expression in normal and tumorigenic cells (extracellular matrix metalloproteases/protein purification/cDNA cloning)
SCOrr M. WILHELM, IVAN E. COLLIER, ANNEMARIE KRONBERGER, ARTHUR Z. EISEN, BARRY L. MARMER, GREGORY A. GRANT, EUGENE A. BAUER, AND GREGORY I. GOLDBERG Division of Dermatology, Department of Medicine, Washington University School of Medicine, St. Louis, MO 63110
Communicated by David M. Kipnis, June 29, 1987
We have purified and determined the comABSTRACT plete primary structure of human stromelysin, a secreted metalloprotease with a wide range of substrate specificities. Human stromelysin is synthesized in a preproenzyme form with a calculated size of 53,977 Da and a 17-amino acid long signal peptide. Prostromelysin is secreted in two forms, with apparent molecular masses on NaDodSO4/PAGE of 60 and 57 kDa. The minor 60-kDa polypeptide is a glycosylated form of the major 57-kDa protein containing N-linked complex oligosaccharides. Zymogen activation by trypsin results in the removal of 84 amino acids from the amino terminus of the enzyme generating a 45-kDa active enzyme species. Human stromelysin is capable of degrading proteoglycan, fibronectin, lamiiin, and type IV collagen but not interstitial type I collagen. The enzyme is not capable of activating purified human fibroblast procollagenase. Analysis of its primary structure shows that stromelysin is in all likelihood the human analog of rat transin, which is an oncogene transformation-induced protease. The pattern of enzyme expression in normal and tumorigenic cells revealed that human skin fibroblasts in vitro secrete stromelysin constitutively (1-2 jug per 106 cells per 24 hr). Human fetal lung fibroblasts transformed with simian virus 40, human bronchial epithelial cells transformed with the ras oncogene, fibrosarcoma cells (HT-1080), and a melanoma cell strain (A 2058), do not express this protease nor can the enzyme be induced in these cells by treatment with phorbol 12-myristate 13-acetate. Our data indicate that the expression and the possible involvement of secreted metalloproteases in tumorigenesis result from a specific interaction between the transforming factor and the target cell, which may vary in different species.
The extracellular matrix (ECM) of multicellular organisms plays an active role in the formation and maintenance of tissues. The meshwork of ECM macromolecules is deposited by resident cells and provides a substrate for cell adhesion and migration, as well as a permeability barrier in cell-cell communication. A number of processes in normal tissue maintenance [e.g., wound healing (1, 2), bone resorption (3), and uterine involution (4)] as well as certain pathologic processes [for example, rheumatoid arthritis (5, 6), epidermolysis bullosa (7, 8), corneal (9, 10) and gingival disease (11)] require connective tissue remodeling, involving a guided degradation of the preexisting ECM. Secreted proteases, required for initiation of degradation of the proteinaceous components of the ECM, also play an important part in processes requiring cell movement, such as tumor invasion and metastasis (12, 13). These enzymes constitute a family of structurally related metalloendoproteases. Previously, we characterized the expression, properties, and primary strucThe publication costs of this article were defrayed in part by page charge payment. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. §1734 solely to indicate this fact.
ture of human fibroblast collagenase (14-16), an enzyme responsible for the initiation of interstitial collagen degradation. In this report, we present the characterization of human fibroblast stromelysin, a proteoglycanase closely related to collagenase with a rather wide range of substrate specificities. The initial data on stromelysin expression in normal and tumor cell strains in vitro suggest that regulation of ECM metalloprotease expression is cell-type specific and may vary among species.
MATERIALS AND METHODS Cell lines used in this report have been described (15). In addition, human fetal lung fibroblast IMR-90 and IMR-90 transformed with simian virus 40 (SV40) (AG 2805) were from the Human Genetic Mutant Repository (Camden, NJ). Human fibrosarcoma cell line (HT-1080) was obtained from the American Type Culture Collection. Human keratinocytes, melanoma cell line A2058, and TBE-1 (17) were gifts from A. Pentland (Washington University), L. Liotta (National Institutes of Health), and G. Yoakum (National Institutes of Health), respectively. Cell culture and treatment with phorbol 12-myristate 13-acetate (PMA) and tunicamycin were described in a preceding publication (15). Cells were labeled for 16 hr with [3H]leucine at 20 ,uCi/ml (154 Ci/mmol; 1 Ci =37 GBq) in serum-free medium without leucine or with 200 ,Ci of [3H]mannose per ml in low glucose medium. All methods used in this report, including DNA sequencing, oligonucleotide synthesis, primer extension, in vitro translation, RNA blot and immunoblot analysis, colony screening, proenzyme activation, and digestion with endoglycosidases F and H were performed as described in reports on human fibroblast collagenase without significant modifications (15, 16). Amino-terminal protein sequences of prostromelysin and the activated 45-kDa form of stromelysin were determined as described (16), except NaDodSO4 gels were electroblotted on Polybrene-coated glass fiber filters in neutral pH buffer (18), and bands were visualized under UV light. Monospecific antiserum was prepared by injecting rabbits at weekly intervals (three times) with a NaDodSO4 gel slice containing -25 ,ug of the 57-kDa protein species emulsified with an equal volume of Freund's adjuvant in water. A simplified three-step procedure for the isolation of prostromelysin from human skin fibroblast (WU 80547) conditioned medium, yielding 50-100 ,ug of enzyme per liter was developed as outlined below. Collected serum-free medium was adjusted to 20 mM Tris-HCI (pH 7.5) and applied to a 2.5 x 15 cm zinc chelate Sepharose (Pharmacia) column equilibrated in 20 mM Tris'HCI, pH 7.5/5 mM CaC12 (Tris CaCI2 buffer) containing 150 mM NaCl. The column was Abbreviations: ECM, extracellular matrix; SV40, simian virus 40; PMA, phorbol 12-myristate 13-acetate.
6725
6726
Biochemistry: Wilhelm et al.
Proc. Natl. Acad. Sci. USA 84
washed with the same buffer (1 vol) and Tris CaCl2 buffer containing 250 mM glycine (Bio-Rad; 2-3 vol). Prostromelysin was eluted with 600 mM glycine in Tris CaCl2 buffer, dialyzed against 5 mM Tris-HCl, pH 7.5/0.1 mM CaC12, adjusted to Tris CaCl2 buffer and applied to a 1 x 4 cm DEAE-Sepharose column equilibrated with 0.005% Brij-35 in Tris CaCl2 buffer. The enzyme was eluted with 50 mM NaCl in the column buffer, applied directly to a 1 x 4 cm column of reactive red agarose (Sigma), and eluted with 400 mM NaCl and 0.005% Brij-35 in Tris CaCl2 buffer. Purified prostromelysin was stored at -700C after dialysis against 50 mM NaCl and 0.005% Brij-35 in Tris CaCI2 buffer. Enzyme activity was determined using a-[ 4C]casein (50,000 cpm/mg) as substrate according to the method of Cawston et al. (19). One unit of enzyme activity is defined as the amount of enzyme capable of digesting 1.0 ,ug of substrate per min at 37°C. Collagenolytic and gelatinolytic activities were determined as reported (14, 20). Proteoglycanase activity was measured using the polyacrylamide bead assay of Nagase and Woessner (21). Assays containing 15 ,ug of purified fibronectin or laminin (Bethesda Research Laboratories) were performed for 1, 3, and 12 hr at 37°C using an enzyme/substrate ratio of 1:60 in a final vol of 30,u. Type IV collagen degrading activity was measured at 32°C for 12 hr using [3H]prolinelabeled HT-1080 type IV collagen purified by ammonium sulfate precipitation and DEAE-cellulose chromatography (22).
RESULTS Cultured Human Skin Fibroblasts Constitutively Synthesize Stromelysin and a Limited Number of Other Metalloproteases. We have previously characterized human skin fibroblast collagenase, a major metalloprotease secreted in two forms, a 52-kDa form and a minor glycosylated 55-kDa species (15). We now have used a NaDodSO4 substrate gel electrophoresis (zymogram) to analyze the protease secretion pattern of various cell types and to monitor protease activity semiquantitatively. Enzyme purification can then be assessed in parallel by conventional NaDodSO4/PAGE of the same sample. Casein zymograms of human fibroblast conditioned media reveal four major bands of proteolytic activity at 60, 57, 55, and 52 kDa (Fig. 1, lane 1; see note in legend). Similarly, gelatin zymograms demonstrate a major band of activity at 68 kDa (unpublished data). Both gelatinolytic and caseinolytic activities were totally inhibited by EDTA (data not shown). The 55- and 52-kDa bands correspond to fibrokDa 94 68 -
.......... .....
(1987)
blast procollagenase as described (15), The caseinolytic metalloprotease of 60 and 57 kDa was purified (Fig. 1, lane 5) using the three-step purification procedure (as described in Materials and Methods). The results presented in Fig. 2A (lane 3) show that a monospecific antiserum raised against the major 57-kDa polypeptide species also recognizes a minor 60-kDa protein secreted from biosynthetically labeled human skin fibroblasts. Therefore, we investigated whether the 60-kDa protein is a glycosylated version of the 57-kDa major form of the enzyme similar to that observed for fibroblast procollagenase (15). [3H]Mannose was incorporated exclusively into the 60-kDa species ofthe enzyme (Fig. 2A, lane 1). Treatment of cultured fibroblasts with tunicamycin resulted in complete and selective inhibition of synthesis of the 60-kDa proenzyme (lane 5). The results of the experiment shown in Fig. 2B confirm that the 60-kDa proenzyme form contains N-linked complex oligosaccharides. The glycosylated 60-kDa polypeptide was sensitive to digestion with endoglycosidase F (24), while treatment with endoglycosidase H, an enzyme specific for high mannose type oligosaccharides (25), had no effect (Fig. 2B, lanes 2 and 4). Fig. 3A (lane 2) shows an in vitro cell-free translation of mRNA from human skin fibroblast strain WUN 80547, with a single immunoprecipitable band of 59 kDa, apparently due to the presence of an uncleaved signal peptide. The purified proenzyme was 95% inactive using [I4C]casein as substrate unless activated by limited proteolysis with trypsin or by treatment with the organomercurial paminophenylmercuric acetate as described for fibroblast procollagenase (26). As shown in Fig. 1 (lanes 6-8), enzyme activation by either trypsin or p-aminophenylmercuric acetate resulted in conversion of the major 57-kDa (lane 6) proenzyme species into a polypeptide of 45 kDa (lanes 7 and 8). The final specific activity of the 45-kDa active enzyme species against [14C]casein was 1200 units per mg of enzyme protein. Substrate Specificity of Human Fibroblast Stromelysin. To establish the range of substrate specificity of the isolated metalloprotease, we examined its action against several physiological substrates. Using the proteoglycan polyacrylamide bead assay of Nagase and Woessner (21) 1.0 ,ug of p-aminophenylmercuric acetate-activated stromelysin degraded 0.6,ug of proteoglycan per min. The activated enzyme was also capable of degrading fibronectin, laminin, and type IV collagen (data not shown). NaDodSO4/PAGE revealed that degradation of fibronectin initially produced a fragment of 200 kDa after reduction and subsequently two more stable products of 170 and 145 kDa. Basement membrane components were also degraded. Both chains of laminin were cleaved, resulting in degradation products of 185 and 165 kDa. Type IV collagen degradation fragments of 170 and 165 kDa produced at 30°C were similar to those produced by
Ile
45 -
kDa 68-
30-
-
45
A Amm
B .-
4.
.4%"j"'.
--
.--
-
1 2 3 4 5 6 7 8
1 2 3 4 5 FIG. 1. Skin fibroblast conditioned serum-free medium (lanes 1 and 4) and purified prostromelysin (lanes 2 and 5) were subjected to NaDodSO4/PAGE with (lanes 1 and 2) or without (lanes 4 and 5) a-casein (23). The proenzyme (lane 6) was activated by limited proteolysis with trypsin (trypsin/proenzyme ratio, 1:10, wt/wt) for 15 min at 25°C (lane 7) or by incubation with 1.0 mM p-aminophenylmercuric acetate (lane 8) for 2 hr at 37°C. The samples were reduced with 10 mM dithiothreitol and molecular size markers (Pharmacia) are shown in lane 3. Note that the appearance of the 60-kDa prostromelysin species, evident on the zymogram, is obscured in the photograph (lane 1) by the 57-kDa major species. All four bands are clearly revealed by immunostaining (see Fig. 5).
1
2
3 4
FIG. 2. Partial glycosylation of human stromelysin. (A) Fibroblasts were labeled with [3H]mannose (lanes 1 and 2) or [3H]leucine (lanes 3-5) and treated with tunicamycin (lanes 4 and 5). Samples were immunoprecipitated using monospecific anti-stromelysin antibody (lanes 1, 3, and 5) or nonimmune IgG (lanes 2 and 4) and subjected to NaDodSO4/PAGE. (B) Samples of fibroblast conditioned medium (lanes 1 and 3) were digested with endoglycosidase F (lane 2) or endoglycosidase H (lane 4) as described (15). The immunoblot was stained with specific antiserum to the 57-kDa prostromelysin. Arrow designates migration of glycosylated proenzyme species.
Biochemistry: Wilhelm et al. A
kDa
3 b
."I..
9468- _ 45 -
Proc. NatL. Acad. Sci. USA 84 (1987)
- 28S
i1. ..t
-18S
30-
1 2 3 4 5
1 2 3
FIG. 3. Cell-free translation and RNA blot analysis of fibroblast mRNA. (A) Total [35S]methionine-labeled in vitro translation products obtained with 1.0 ,ug of mRNA were subjected to NaDodSO4/ PAGE before (lane 5) or after immunoprecipitation using stromelysin-specific (lane 2) or nonimmune (lane 3) IgG. The biosynthetically labeled secreted prostromelysin (lane 1) is shown for comparison. Lane 4, mRNA omitted from the reaction. (B) RNA blot analysis of 5 ,ug of mRNA from skin fibroblast (lane 1), SV40transformed lung fibroblast (lane 2), and melanoma cells (lane 3).
pepsin (22). The pH optimum for this metalloprotease was 7.5-8.0, and the enzyme was inhibited by the metal chelators EDTA and 1,10-phenanthroline, by dithiothreitol, and by the 28-kDa tissue metalloprotease inhibitor (data not shown). Although the activated enzyme had negligible activity against collagen fibrils, it did exhibit some activity against [14C]gelatin (125 units/mg). This specific activity is at least 1/10th that reported for purified gelatinases (20,27). Taken together, these data on molecular mass, proenzyme activation, substrate specificity, and inhibition profile indicate that this metalloprotease is very similar, if not identical to stromelysin (proteoglycanase) isolated from rabbit synovial fibroblasts (23) and rabbit bone cultures (28) and the matrix metalloprotease 3, purified from human rheumatoid synovial cells
(29).
Complete Primary Structure of Human Stromelysin Suggests That this Protease Is an Analog of Rat Transin. A
preparation of prostromelysin and the active 45-kDa enzyme species were separated by NaDodSO4/PAGE and subjected to amino acid sequence analysis as described in Materials and Methods. The NH2-terminal amino acid sequences of the 57-kDa proenzyme and the 45-kDa active enzyme forms were found to be Tyr-Pro-Leu-Asp-Ala-Ala-Arg-Gly-Glu-Asp-Thr and Thr-Phe-Pro-Gly-Ile-Pro, respectively. The sequence Thr-Phe-Pro-Gly-Ile-Pro was reverse-translated and a 17base long oligonucleotide probe TGIAAAGGICITAATGG (SO10) containing 3 inosine residues was synthesized. A cDNA library constructed from WUN 80547 mRNA as reported (16) was screened with the SO10 oligomer. A number of clones were purified by colony hybridization and shown to be similar by restriction endonuclease mapping. The 5' end of the largest clone pSL45.5, which contained a 1675-base-pair (bp) insert, excluding the oligo(G) and poly(A) tails, was sequenced. A 20-base long oligomer, S012, was synthesized using a nucleotide sequence derived from the 5' end of clone pSL45.5. The cDNA library was rescreened with the S012 oligomer and clone pSL51.4, representing the complete mRNA sequence coding for preprostromelysin, was purified. The clone pSL51.4 contains an insert of 1802 bp and hybridizes to an mRNA species of 2.3 kilobases (kb) (Fig. 3B, lane 1). The complete sequence of the pSL51.4 has been determined and confirmed on both strands (30). The sequence is not presented in this report.* The insert consists of a 49-bp 5' untranslated leader, followed by an ATG methionine codon, and 1431 nucleotides coding for a 477-amino acid *This sequence is being deposited in the EMBL/GenBank data base (Bolt, Beranek, and Newman Laboratories, Cambridge, MA, and Eur. Mol. Biol. Lab., Heidelberg) (accession no. J03016).
6727
long preprostromelysin protein of a predicted size of 53,977 Da. The 3' untranslated region includes 321 bp between the termination codon (TGA) and the start of the poly(A) tail. The first 17 amino acid residues immediately following the initiating methionine constitute a hydrophobic signal peptide. The protein sequence between positions 18 and 29 corresponds to that obtained from the NH2 terminus of the proenzyme, which has the predicted size of 52,220 Da. Activation of stromelysin by limited proteolysis is accompanied by the loss of 84 amino acids and the sequence of the NH2 terminus of the activated stromelysin corresponds to positions 101-106 of the protein encoded by cDNA pSL51.4. The predicted molecular masses of the proenzyme and activated stromelysin are slightly lower than the apparent values obtained by NaDodSO4/PAGE. The two possible N-glycosylation sites (Asn120,Asn398) are contained within the trypsin-activated 42,533-Da enzyme species. In our previous report on the structure of human fibroblast collagenase (16), we have shown an extensive homology between collagenase and rat transin. Analysis of the transin coding sequence (31) and comparison with the partial sequence of rat fibroblast collagenase showed that transin is a secreted protein (with a 17-amino acid signal peptide) different from collagenase (16). Matrisian et al. (32) recently confirmed that rat transin is a secreted protease. A major feature of the presented comparison was two insertions of three and nine amino acids in rat transin. Two-dimensional analysis of the secondary structures showed that the latter insertion introduced two additional high probability f turns into the rat protein. The alignment of stromelysin with collagenase and transin presented here (Fig. 4) shows that human stromelysin is more closely related to rat transin (75% homology) than to human collagenase (55% homology). In addition, human stromelysin contains both the three- and nine-amino acid insertions that are present in the rat gene and required for alignment with collagenae. All cysteines (at positions 11, 92, 290, and 477) in human stromelysin are conserved in transin. A putative glycosylation site, asparagine-120, is also conserved. All three metalloproteases contain a triplet of basic amino acids (positions 54-56) immediately preceding the site of trypsin cleavage during zymogen activation (33). This sequence may serve an important MKSLPILLLLCV--AVCSAYPLDGAARGEDTSMNLVQKYLENYYDLKKDVKG G E H --SE AG EVL W T-G V H F P
S
50
K N N GRQ SHSP ATL--ETQEQDVD 100 FVRRKDSGPVVKKIREMQKFLGLEVTGKLDSDTLEVMRKPRCGVPDVGHF G N M L H KQ M TKKS E LQ E F K VEK RN F AE K KQ AQ
LFWGV
RTFPGIPKWRKTHLTYRIVNYTPDLPKDAVDSAVEKALKVWEEVTPLTFS L RES IER
150
RLYEGEADIMISFAVREHGDFYPFDGPGNVLAHAYAPGPGINGDAHFDDD T E IS I M E G GN VRGD R NS KVS Q FQ EQWTKDTTGTNLFLVAAHEIGHSLGLFHSANTEALMYPLYHSLTDLTRFR A H V K S L A
200
N IS S S VL E N R EQ R
R
R
D V
HR
NNFREY
RAD
E
FQL SN
T
LSQDDINGIQSLYGPPPDSPETPLVPTEPVPPEPGTPANCDPALSFDAVS KSNSLD E LPM SS TE DVLV VD IT Al RSQNPVQP---------IG Q KA SK T D A TLRGEILIFKDRHFWRKSLRKLEPELHLISSFWPSLPSGVDAAYEVTSKD NM TP I
V F
FYM TNPFYP V
VMF
GYF NF
V
250
SYTFSG VQ---
TDIG
S
L
H I
0
NR
300 350
FADR
N LE
L DQK - LE Q H EL TLVFIFKGNQFWAIRGNEVRAGYPRGIHTL-GFPPTVRKIDAAISDKEKNK L I KS L EENTG E RF KY VQ QN LH KD YSSF R KH
400
TYFFVEDKYWRFDEKRNSMELGFPKQIAEDFPGIDSKIDAVFEEFGFFYFF L KQ DPE RK N GT V A F
450
TGSSQLEFDPNAKKVTHTLKSNSWLNC--I G S F ---
477
AN
H TR
YK
Y
YKR DPSY
KT RILTLQ
A
M
F
H
RKN
GH V
HuSL RatT HuCL
MKD
FIG. 4. Comparison of protein sequence of human stromelysin (HuSL) and collagenase (HuCI) with rat transin (ratT). Stromelysin protein sequence as predicted from the cDNA sequence of clone pSL51.4. Only amino acids differing from stromelysin are shown in corresponding positions. Amino acids are identified by the singleletter code.
Proc. Natl. Acad. Sci. USA 84
Biochemistry: Wilhelm et al.
6728
function in the activation of these enzymes in vivo by other serine proteases such as plasmin and plasminogen activators (12, 34). These observations, together with the fact that rat transin is a secreted protease (32), lead to the conclusion that stromelysin is in all likelihood a human analog of transin. The Expression of Stromelysin Is Cell Type Specific and Does Not Necessarily Correlate with Tumorigenic Transformation. The expression of the transin gene in fetal rat fibroblasts is induced upon transformation with polyomavirus, Rous sarcoma virus, and the activated oncogene Ha-ras (31). In addition, Matrisian et al. (32) have recently shown that transin mRNA is present at significantly higher levels in chemically induced skin carcinomas than in normal skin or benign papillomas. The question arises, then, as to whether the expression of stromelysin is regulated by oncogene expression or treatment with PMA. We have addressed this question by measuring stromelysin and collagenase secretion by normal and transformed cell strains (Fig. 5). Cultured human skin fibroblasts (Fig. 5, AdSk) secrete a significant amount of stromelysin constitutively and the level of expression can be increased further by treatment with PMA (Fig. 5, AdSk+). The constitutive expression of stromelysin, as well as its stimulation by PMA, in fetal skin fibroblasts, human colon, cornea, and gingiva parallels the expression of collagenase (15). In contrast, human endothelial cells and the melanoma cell strain A2508 [Fig. 5 HuE (-) and A2058 (-)] secrete significant amounts of collagenase but not stromelysin (see Fig. 3B, lane 3); PMA treatment of these cells does not stimulate stromelysin expression (Fig. 5 HuE+, A2058+). Human keratinocytes cultured in vitro express a low amount of stromelysin, which was not stimulated by treatment with PMA (Fig. 5, Ker). The fibrosarcoma cell strain HT-1080 and SV40-transformed human fetal lung fibroblasts secrete low levels of collagenase and no detectable stromelysin [HT-1080(-), SV40(-), and Fig. 3B, lane 2]. Treatment of these cells with PMA causes an increased expression of collagenase but had no effect on stromelysin expression. The parental fetal lung fibroblast cells IMR-90 express low levels of collagenase and stromelysin that can be stimulated by PMA treatment but the effect of PMA on stromelysin expression by these cells is very limited. Of special interest is the fact that human bronchial epithelial cells transformed with activated Ha-ras oncogene (17) do not secrete either collagenase or stromelysin regardless of treatment with the tumor promoter PMA. Tight coordinate regulation of collagenase and stromelysin genes has recently PMA pro pro
Gi Pro/Act CL Cor FeSk AdSk Co +1Inr=-+I r +-1r +II-
SL: CLE
HuE PMA pro
IMR-9OSV-40 TBE
+r
Ker HT-1080 A2058
+
SLc
proCLE
an
FIG. 5. Expression of procollagenase (proCL) and prostromelysin (proSL) by normal and tumorigenic human cells. Samples of serum-free media from control (-) and PMA-treated (+) cells were subjected to immunoblot analysis as described (15). The immunoblot was initially stained for immunoreactive collagenase, rinsed in 0.2 M acetic acid, and restained using anti-stromelysin antiserum (1:400). FeSk, fetal skin; AdSk, adult skin; Col, colon; Cor, cornea; Gi, gingival fibroblast; HuE, human endothelial cells; Ker, keratinocytes. IMR-90, SV40, TBE, HT-1080, and A2058 cell lines are described in Materials and Methods.
(1987)
been observed in rabbit synovial cells, alveolar macrophages, and brain endothelial cells (35). Our observations suggest that the regulation of ECM protease expression, and consequently their individual roles in tissue remodeling or in tumor invasion (metastasis), is cell-type specific and possibly varies among species.
DISCUSSION Human skin fibroblasts in tissue culture constitutively secrete a number of metalloendoproteases that initiate the degradation of extracellular matrix components. These enzymes comprise a group of structurally related proteins with distinct, but in the case of stromelysin rather broad, substrate specificity. In addition, collagenase and stromelysin undergo partial posttranslational modification, and proenzyme activation leads to the appearance of active intermediates of different molecular size. These facts point out the importance of a systematic characterization of each member of the group by protein sequence analysis, complete structural analysis on a cDNA level, and the characterization of substrate specificity of the purified enzymes. This point is well illustrated by the recent report comparing the structure of the secreted rat protease transin and the putative protease transin 2 (36). The structural similarity between these two proteins is so extensive that identification on the level of cDNA hybridization or reaction with polyclonal antibody alone is likely to produce
confusing results. Stromelysin purified from human skin fibroblast conditioned medium is 95% inactive against a casein substrate. The NH2-terminal sequence of the proenzyme form of stromelysin is Tyr-Pro-Leu-Asp-Ala-Ala-Arg-Gly-Glu-Asp-Thr. Limited proteolysis in the presence of trypsin or treatment with the organomercurial compound p-aminophenylmercuric acetate results in activation of the enzyme concomitant with the loss of 84 amino acids from the NH2 terminus. This results in conversion of the zymogen to a 45-kDa active enzyme with the NH2-terminal protein sequence Thr-Phe-Pro-Gly-Ile-Pro. Activated stromelysin is capable of degrading a range of macromolecules present in ECM-i.e., proteoglycan, fibronectin, laminin, and type IV collagen. The enzyme does not degrade interstitial collagen (type I). Of particular importance is the observation that stromelysin is not capable of activating fibroblast procollagenase and therefore is different from the putative rabbit procollagenase activator of similar molecular mass (37).
Recently, there have been several reports of metalloproteases distinct from interstitial collagenase (38) and gelatinase (39), which degrade proteoglycans, gelatin, and other connective tissue components. Galloway et al. (28) purified a metalloprotease in an active form of 24 kDa, which degrades proteoglycans from rabbit bone culture medium. Antibodies to rabbit bone proteoglycanase immunoprecipitated a higher molecular mass proenzyme species of 51 kDa from PMA-treated rabbit synovial fibroblasts, which was named stromelysin or proteoglycanase. The substrate specificities, inhibition profiles, and pH optima of both the low molecular mass (28 kDa) and the high molecular mass (45 kDa) active matrix metalloprotease-3 forms (29), isolated from human rheumatoid synovial cells, appear to be identical to rabbit and human fibroblast stromelysin. Although the relationship of human stromelysin and metalloprotease-3 to a latent neutral proteoglycanase (68 kDa) purified from tissue extracts of human cartilage (40) remains to be elucidated, they appear to be distinct from a metalloprotease of 70 kDa isolated from tumor cells (41) and adherent macrophages (42), which degrades types IV but not type I collagen. The sequence of stromelysin from human skin fibroblasts presented here is in agreement with that of a putative stromelysin cDNA clone isolated from a human gingival fibroblast cDNA
Biochemistry: Wilhelm et al. library by cross-hybridization to a partial rabbit stromelysin cDNA clone (43). Our data indicate that a single mRNA species codes for both secreted proenzyme forms of human fibroblast stromelysin of 60 and 57 kDa. As with procollagenase (15), the minor 60-kDa polypeptide is the result of partial N-linked glycosylation, apparently through the addition of complex oligosaccharide(s) to the unmodified proenzyme species at one or both of the possible N-glycosylation recognition sites (Asn120,Asn31) found in the primary structure of the enzyme. Partial N-linked glycosylation appears to be a common feature in the biosynthesis of both collagenase and stromelysin by cultured cells. Although the physiological function of glycosylation is not known, it may be an important factor in targeting these metalloproteases to different compartments of the extracellular matrix. In a previous report we measured collagenase synthesis in a variety of normal cell strains (15) and compared this to enzyme production by two leukocyte tumor cell strains (44, 45). The results indicated that the expression of collagenase is tissue specific and regulated developmentally. Generally, a similar notion concerning the expression of human stromelysin can be forwarded based on the observations presented here. It is clear from our data that the specific pattern of human stromelysin expression is distinct from that of collagenase. In the cell strains of fibroblast origin, collagenase and stromelysin are secreted at similar rates, and both enzymes can be stimulated with PMA treatment. Treatment with PMA, however, failed to induce or stimulate stromelysin synthesis in normal human endothelial cells, keratinocytes, and several tumor cell lines, including melanoma, fibrosarcoma, SV40-transformed fetal lung fibroblasts, and Haras-transformed bronchial epithelial cells. These observations are of particular interest in view of the fact that the sequence homology between human stromelysin and rat transin and transin 2 genes strongly suggests that stromelysin is a human analog of transin. In addition, expression of the transin gene is readily stimulated in fetal rat fibroblasts by a variety of transforming agents with the exception of SV40, while more selective induction is observed for the transin 2 gene (36). The expression of the transin gene is significantly increased in experimentally induced rat carcinomas compared to normal skin or benign papillomas (32). This is consistent with the notion that increased expression of secreted proteases plays an important role in tumor invasion (12, 13). Our observations on the expression of collagenase and stromelysin indicate that the induction of metalloprotease synthesis is the result of interaction between a specific cell type and the transforming agent, which leads us to conclude that the specific role of each ECM protease in the progression of a variety of tumors has to be addressed individually. We thank Fred Woessner, Jr. (University of Miami Medical School) for the generous gift of the proteoglycan bead substrate. We
also thank Mari Teter for her excellent technical assistance and Rosemarie Brannan and Ginger Roberts for preparation of this manuscript. This study was supported by Grants AM12129, AR19537, and TO-AM07284 from the National Institutes of Health and in part by the Monsanto Company/Washington University Biomedical Research Agreement. Grillo, H. C. & Gross, J. (1967) Dev. Biol. 15, 300-317. Eisen, A. Z. (1969) J. Invest. Dermatol. 52, 449-453. Vaes, G. (1972) Biochem. J. 126, 275-289. Roswit, W. T., Halme, J. & Jeffrey, J. J. (1983) Arch. Biochem. Biophys. 225, 285-295. 5. Harris, E. D., Jr., Faulkner, C. S. & Brown, F. E. (1975) Clin. Orthop. Relat. Res. 110, 303-316. 6. Dayer, J. M., Russell, R. G. & Krane, S. M. (1977) Science 195, 1. 2. 3. 4.
181-183.
Proc. Natl. Acad. Sci. USA 84 (1987)
6729
7. Bauer, E. A. & Eisen, A. Z. (1978) J. Exp. Med. 148, 1378-1387. 8. Kronberger, A., Valle, K. J., Eisen, A. Z. & Bauer, E. A. (1982) J. Invest. Dermatol. 79, 208-211. 9. Brown, S. I. & Weller, C. A. (1971) Trans. Am. Acad. Ophthalmol. Otolaryngol. 74, 375-382. 10. Gordon, J., Bauer, E. A. & Eisen, A. Z. (1980) Arch. Ophthalmol. 98, 341-345. 11. Birkedal-Hansen, H. (1980) in Collagenase in Normal and Pathological Connective Tissues, eds. Woolley, D. E. & Evanson, J. J. (Wiley, New York), pp. 128-134. 12. Mignatti, P., Robbins, E. & Rifkin, D. B. (1986) Cell 47, 487-498. 13. Liotta, L. A., Rao, C. N. & Wener, U. M. (1986) Annu. Rev. Biochem. 55, 1037-1057. 14. Stricklin, G. P., Eisen, A. Z., Bauer, E. A. & Jeffrey, J. J. (1978) Biochemistry 17, 2331-2337. 15. Wilhelm, S. M., Eisen, A. Z., Teter, M., Clark, S. D., Kronberger, A. & Goldberg, G. (1986) Proc. Natl. Acad. Sci. USA 83, 3756-3760. 16. Goldberg, G. I., Wilhelm, S. M., Kronberger, A., Bauer, E. A., Grant, G. A. & Eisen, A. Z. (1986) J. Biol. Chem. 261, 6600-6605. 17. Yoakum, G. H., Lechner, J. F., Gabrielson, E. W., Korba, B. E., Malan-Shibley, L., Willey, J. C., Valerio, M. G., Shamsuddin, A. M., Trump, B. F. & Harris, C. C. (1985) Science 227, 1174-1179. 18. Vandekerckhove, J., Bauer, E. A., Puype, H., Damme, J. V. & Van Montagu, M. (1985) Eur. J. Biochem. 152, 9-19. 19. Cawston, T. E., Galloway, W. A., Mercer, E., Murphy, G. & Reynolds, J. J. (1981) Biochem. J. 195, 159-165. 20. Seltzer, J. L., Adams, S. A., Grant, G. A. & Eisen, A. Z. (1981) J. Biol. Chem. 256, 4662-4668. 21. Nagase, H. & Woessner, J. F., Jr. (1980) Anal. Biochem. 107, 385-392. 22. Crouch, E. & Bornstein, P. (1979) J. Biol. Chem. 254, 4197-4204. 23. Chun, J. R., Murphy, G. & Werb, Z. (1985) J. Biol. Chem. 260, 12367-12376. 24. Elder, J. H. & Alexander, S. (1982) Proc. Natl. Acad. Sci. USA 79, 4540-4544. 25. Tarentino, A. L. & Maley, F. (1974) J. Biol. Chem. 249, 811-817. 26. Stricklin, G. P., Jeffrey, J. J., Roswit, W. T. & Eisen, A. Z. (1983) Biochemistry 22, 61-88. 27. Hibbs, M. S., Hasty, K. A., Seyer, J. M., King, A. H. & Mainardi, C. L. (1985) J. Biol. Chem. 260, 2493-2500. 28. Galloway, W. A., Murphy, G., Sandy, J. D., Gavrilovic, J., Cawston, T. E. & Reynolds, J. J. (1983) Biochem. J. 209, 741-752. 29. Okada, Y., Nagase, H. & Harris, E. D., Jr. (1986) J. Biol. Chem. 261, 14245-14255. 30. Maxam, A. & Gilbert, W. (1977) Proc. Natl. Acad. Sci. USA 74, 560-564. 31. Matrisian, L. M., Glaichenhause, N., Gesnel, M. C. & Breathnach, R. (1985) EMBO J. 4, 1435-1440. 32. Matrisian, L. M., Bowden, G. T., Krieg, P., Furstenberger, G., Briand, J.-P., Leroy, P. & Breathnach, R. (1986) Proc. Natl. Acad. Sci. USA 83, 9413-9417. 33. Grant, G. A., Eisen, A. Z., Marmer, B. L., Roswit, W. T. & Goldberg, G. I. (1987) J. Biol. Chem. 262, 5886-5889. 34. Werb, Z., Mainardi, C., Vater, C. & Harris, E. D., Jr. (1977) N. Engl. J. Med. 296, 1017-1023. 35. Frisch, S. M., Clark, E. J. & Werb, Z. (1987) Proc. Natl. Acad. Sci. USA 84, 2600-2604. 36. Breathnach, R., Matrisian, L. M., Gesnek, M.-C., Staub, A. & Leroy, P. (1987) Nucleic Acids Res. 15, 1139-1151. 37. Vater, C. A., Nagase, H. & Harris, E. D., Jr. (1983) J. Biol. Chem. 258, 9374-9382. 38. Harris, E. D., Jr., Welgus, H. G. & Krane, S. M. (1984) Collagen Relat. Res. 4, 493-510. 39. Murphy, G., McAlpine, C. G., Poll, C. T. & Reynolds, J. J. (1985) Biochim. Biophys. Acta 831, 49-58. 40. Azzo, W. & Woessner, J. F., Jr. (1986) J. Biol. Chem. 261, 5434-5441. 41. Salo, T., Liotta, L. A. & Tryggnason, K. (1983) J. Biol. Chem. 258, 3058-3063. 42. Garbisa, S., Ballin, M., Gordini, D., Fastelli, G., Natmole, M., Nergo, A., Semenzato, G. & Liotta, L. (1986) J. Biol. Chem. 261, 2369-2375. 43. Whitham, S. E., Murphy, G., Angel, P., Rahmsdorf, H. J., Smith, B. J., Lyons, A., Harris, T. J. R., Reynolds, J. J., Herrlich, P. & Docherty, A. J. P. (1986) Biochem. J. 240, 913-916. 44. Bar-Shavit, Z., Tietelbaum, S., Stricklin, G. P., Eisen, A. Z., Kahn, A. J. & Welgus, H. G. (1985) Proc. Natl. Acad. Sci. USA 82, 5380-5384. 45. Welgus, H. G., Connolly, N. L. & Senior, R. M. (1985) Clin. Res. 33, 825 (abstr.).
