THE JOURNAL OF BIOLOGICAL CHEMISTRY (c) 1991 by The American Society for Biochemistry and Molecular Biology, Inc.

Vol. 266, No. 25, Issue of September 5, pp. 16485-16490,1991 Printed in U. S. A.

Complete Structure of the Human Gene for 92-kDa Type IV Collagenase DIVERGENT REGULATION OF EXPRESSION FOR THE 92- AND 72-KILODALTON ENZYME GENES IN HT-1080 CELLS*

(Received for publication, October 9, 1990)

Pirkko HuhtalaS, Ari Tuuttila$, Louise T. Chows, Jouko Lohill, Jorma Keski-Ojall, and Karl TryggvasonSll From the SBiocenter and Department of Biochemistry, University of Oulu, SF-90570 Oulu, Finland, the §Department of Biochemistry, University of Rochester School of Medicine and Dentistry, Rochester, New York 14642, and the VDepartments of Virology and Dermatology, University of Helsinki, SF-00290 Helsinki. Finland

The complete structure of the human gene for 92- kDa type IV collagenase was determined. Two over- lapping genomic clones spanning 26 kilobases (kb) of genomic DNA were shown to contain the entire 7.7-kb structural gene together with 15 and 3.5 kb of 5’-end and 3’-end flanking regions, respectively. The 92-kDa type IV collagenase gene contains 13 exons as does the 72-kDa type IV collagenase gene. All intron locations of the 92-kDa enzyme gene coincided with intron lo- cations in the 72-kDa enzyme gene. Exons 5,6, and 7 which were 174, 174, and 177 base pairs long, respec- tively, each encoded one complete internal repeat which resembles the collagen-binding domains of fi- bronectin. The sequence coding for a unique 48-residue segment in the 92-kDa type IV collagenase that has no counterpart in other metalloproteinases was not pres- ent in a separate exon, but was contained in exon 9 which also codes for sequences with homology to the other metalloproteinases. The initiation site for tran- scription was determined by primer extension analy- sis. Sequencing analysis of 599 base pairs of the 5‘- end flanking region showed that the promoter does not have a TATA motif, but a TTAAA sequence at position -29 to -25. A CAAT motif was not observed but there was one GC box. Two putative 12-0-tetradecanoyl- phorbol-13-acetate (TPA) response elements, that might serve as binding sites for the transcription factor AP-1 and a consensus sequence of a transforming growth factor j31 (TGF-j31) inhibitory element were found in the promoter region. Gelatinase assay of en- zyme secreted by cultured human fibrosarcoma cells (HT-1080) revealed only low levels of 92-kDa type IV collagenase activity, whereas considerable activity of the 72-kDa enzyme was present. Northern hybridiza- tion analysis confirmed these findings. Treatment of the HT-1080 cells with TPA resulted in induction of the secretion of 92-kDa type IV collagenase activity. This induction could not be significantly inhibited by

*This work was supported by grants from the Finnish Cancer Society, The Finnish Cancer Institute and the Academy of Finland, and United States Public Health Service Grant 36200. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked “aduer- tisernent” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

The nucleotide sequencefs) reported in this paper has been submitted to the GenBankTM/EMBL Data Bank with accession numberfs)

11 To whom correspondence should be addressed. Tel.: 358-81- M68343-M68355.

352361; Fax: 358-81-352356.

concomitant incubation with TGF-Bl. TPA and TGF- 81 did not markedly affect the activities of the 72-kDa enzyme. The activities of the secreted 92- and 72-kDa enzymes by HT- 1080 cells correlated with the amounts of mRNA as estimated by Northern analyses.

Mammalian extracellular metalloproteinases form a family of structurally related enzymes which are the products of related genes. These enzymes are capable of degrading various components of the extracellular matrix, but they vary with respect to their substrate specificity. The metalloproteinases are all similar in that they are secreted as proenzymes, they have a Zn2+-binding site, and they can be inhibited by the tissue inhibitors of metalloproteinases (see Refs. 1-9). The mammalian metalloproteinase family contains a t least two distinct interstitial collagenases (9, lo), two types of strome- lysin (transin) (5, 11-14), a small proteinase PUMP-1 (14), and the 72- (3,4,15) and 92-kDa (6,16) type IV collagenases. Structural analysis of the genes for interstitial collagenase, stromelysin, and 72-kDa type IV collagenase has clearly dem- onstrated the evolutionary relationship of these proteins (17- 20).

The two 72- and 92-kDa type IV collagenases resemble each other both with respect to primary structure and substrate specificity. The propeptide, amino-terminal domain, fibronec- tin-like collagen-binding domains, zinc-binding domain, and carboxyl end region are highly homologous (4, 6). However, the 92-kDa enzyme apparently contains a 54-residue proline- rich domain which has no counterpart in the 72-kDa enzyme or the other known metalloproteinases. This extra domain has sequence homology with a part of the helical domain of the type V collagen a2 chain, and corresponds in size to a 3- fold multiple of the 54-bp’ exons typically found in genes for fibrillar collagens (21). The 72- and 92-kDa type IV collagen- ases have identical substrate specificity profiles in uitro; they cleave native type IV collagen molecules at a single site into one-fourth and three-fourth size fragments and both enzymes degrade denatured collagen (gelatin) and type V collagen, but not type I collagen, proteoglycan, or laminin (3, 4, 6, 22). Despite this similarity, the genes for the 72- and 92-kDa type

’ The abbreviations used are: bp, base pair(s); kb, kilobase(s); PCR, polymerase chain reaction; TPA, 12-O-tetradecanoylphorbol-13-ace- tate; TGF-P1, transforming growth factor Pl; SDS-PAGE, sodium dodecyl sulfate-polyacrylamide gel electrophoresis; GAPDH, glycer- aldehyde-3-phosphate dehydrogenase; HEPES, 4-(2-hydroxyethyl)- 1-piperazineethanesulfonic acid.

16485

16486 FIG. 1. Structure of the human

92-kDa type IV collagenase gene. Top, diagram of the gene with exons depicted as boxes and introns and flank- ing region as interconnecting solid lines. The exons are numbered from the 5'- end of the gene. Cross-hatched boxes in- dicate exons coding for the three internal repeats and the black box the exon coding for the 48-residue sequence with no counterpart in other metalloproteinases (see text). Center, two overlapping ge- nomic clones with Hind111 ( H ) and BamHI ( B ) restriction sites. Bottom, scale in kilobases.

Human 92-kDa Type IV Collagenase Gene

1 2 3 4 5 e 7 0 9 l O \ 1 2 1 1

I1 nn n uu U 3'

H B B H H B H B B I I I 1 1 1 I I I

GL-3 H B B

GL-1

I ....... I T 6 ROC CTC ... ... CTG G C l l GIG g l g g g e ~ ~ ~ s 111 870 tht ser L." L.V 1 1 . GIU f 19 UI~PM.IOI.# I 46

2 tg~teeac.g mn TIC CTG ... ._ . ITC ncc TIT TO gtgagccggg 233 Glu Tyr L.U 11. Thr Tyr TP(P) 47 I24

420

a t ~ c g c ~ o c ~ g G nrc cnn nnc ... ... GGT GTC GCO G gtpegwegt 149 (T.)P 11. GI" G I ~ U ~ I R I ~ G I ~ U )

210

125 114

4 g t t t e t t ~ ~ g nG cnc GGR GRC ... ... nn6 GGC GTC G gttogmttct 129 ( 6 1 ) ~ HI. Gly nsP Lys Gly Uo l U d l )

220

175 217

s eetcctgcDg TG GTT ccn KT _ _ _ .. . ccc R G C G ~ G n gtgagtgegg 174 ( U ~ ) I MI pro n r PN sw GIU nr(g)

IO0

218 27s

6 clegccceog G R CTC T I C K C . . . . . . CCG K C C G l l G gtacclceae 174

276 313 (nr)g Ly. Tyr Th. Pro Th, nrg n l c d

325

7 ptctstcc.g CT Gnc TCG MO ... ... CCG G ~ C cnn G g10gg~gt0g 177 ( n l h RIP Ser Thr Pro I).D Gln bl(y)

11s-

334 392

0 e l e ~ ~ t e e o g GD TnC llGT TTG ... ... CnC CTC T I T G gtgag9cag9 IM

393 444

9 s l s l l l l t o g 01 CCT CGC CCT ... ... T I C I n 6 0111 GG glgog9@ggc 200

445 sa7

(Gl)y lye S.r Leu HI8 Leu l y r Gl(y) 100

(Gl)y Pro Rrg Pro Ph. Ly. 11.P GI(y) 113.

IO Q C ~ I ~ C ~ C O Q G nnc rnc TGG .. . ... TIC TIC TCT G g t t o g t t a ~ ~ 140 (Gl)y Ly. Tyr TPD Ph. Ph. Ser Gl(y)

246

530 504

I 1 tcseelgceg GG CGC CllG G I G ... ... CGC CTC TGG IG gtgopsgssg IS1 ( G I ) ~ nrg GI" UOI nrg L.U T W ne(()

96.

51s 634

I F 9 C G l

12 etgcccgsog G TIC we GIG ... ... CIG rnc cnn G g t g ~ g ~ ~ ~ l g 104 (Ilr)g Ph. ~ S P Val Gln Tyr Gln Gib)

I800

635 669

I3 leleetgcag llG MII GCC T l T ... ... CCT G l G GnC 100. .. 116 (GI)" L ~ S TVP P P ~ GIU n.0 *e* 6 I94 unlrmsle1.d

670 707

FIG. 2. Summary of the exon-intron structure of human 92- kDa type IV collagenase gene. Nucleotide sequences at the intron (lower case letters) and exon (upper case letters) junctions are shown. The derived amino acid sequence and the corresponding position in the polypeptide are displayed below. Amino acids encoded by split codons are listed twice and are partly in parentheses. Exons are numbered beginning from the 5'-end of the gene. The exon-intron structure was determined by heteroduplex analysis, nucleotide se- quencing, or size measurementsof PCR amplified introns as described under "Experimental Procedures." Introns sequenced are marked by an asterisk (*). The stop codon (TGA) is marked ***. The difference in nucleotide sequence of exon 12 found in the cDNA (Ref. 6, and this study) and the resulting amino acid substitution is shown.

IV collagenases seem to differ markedly with regard to tran- scriptional control. For example, alveolar macrophages, pol- ymorphonuclear leukocytes, and keratinocytes have been found to secrete the 92-kDa enzyme but not the 72-kDa enzyme (6) . In contrast, cultured human melanoma cells (A- 2058) and skin fibroblasts secrete the 72-kDa enzyme but not noticeable amounts of the 92-kDa enzyme (6).' Some trans- formed cells such as human fibrosarcoma cells HT-1080 have been reported to secrete both enzymes (4). The secretion of type IV collagenase is induced in cultured fibroblasts by TPA

* P. Huhtala, unpublished observations.

1 2 3 4 5

FIG. 3. Determination of the start site for transcription by primer extension. The mRNA start site was determined by primer extension analysis using poly(A) RNA from human fibrosarcoma (HT-1080) cells as described under "Experimental Procedures." Lams 1-4, results from a co-run of the sequencing reactions of cloned genomic DNA with guanidine (G), adenosine ( A ) , thymidine (T), and cytosine (C) indicated at the top. The oligonucleotide used for priming in the sequencing reactions was the same as that used in the primer extension experiment in lane 5.

(23). In these cells and fibrosarcomas, this increase has been shown to involve the 92-kDa type IV enzymes, whereas the 72-kDa enzyme is only slightly or not at all increased (6) . These data indicate that the two genes have different regu- latory elements and, therefore, that the two enzymes vary with respect to their developmental regulation and biological function. However, there is as yet little knowledge about the physiological role and relationship of the two type IV colla- genases.

Structural and functional analyses of the type IV collagen- ase genes are indispensable for the elucidation of their biolog- ical significance. In the present study we have determined the structure of the entire human 92-kDa type IV collagenase gene and compared it with the structure of the human 72- kDa enzyme gene. These collagenases are probably regulated in divergent fashion, and their activation and induction need to be under strict control. Evidence is presented showing that the two type IV collagenase genes are regulated in a divergent fashion by TPA in human HT-1080 cells.

EXPERIMENTAL PROCEDURES

Isolation and Characterization of cDNA and Genomic Clones-A Xgtll human fibrosarcoma cell (HT-1080) cDNA library (HT1048b, Clontech) and a X-Fix human genomic library (944201, Scratagene) were screened using two end-labeled oligonucleotides which were made according to the sequence of cDNA coding for the human 92- kDa type IV collagenase secreted by SV40-transformed lung fibro- blasts (6). The two synthetic oligonucleotides contained nucleotides 118-138 and 2013-2053 (see Ref. 6). Hybridizations and washes were performed using the sodium pyrophosphate method (24). One positive cDNA clone, HG-1, and one positive genomic clone, GL-1, were isolated. The cDNA insert was subcloned into pUCl9 and the result- ing recombinant plasmid, pHG-1, was isolated and labeled with 32P

FIG. 4. Nucleotide sequence at the 5‘-end of human 92-kDa type IV collagenase gene including coding sequence and the der ived amino acid sequence of f irst exon. The initiation site for transcription (bent arrow) was determined by primer extension. The numbering of nucleotides starts at the transcription initiation site. A TATA motif-like sequence has a double under- line. A GC motif, and TGF-/3 inhibitory element ( T I E ) and TPA response ele- ment (TRE) sequences are boxed. The alternating CA-rich sequence is under- lined.

” O l

J I

Id

A TPA

P cHx 0 ’ 7 , I

’ 0 ° 1 k I L \ I

f I v, B

O J , , I 1 0 1 E 1 2 2 4

n m [houml

FIG. 5. Effects of TPA on the mRNA level fo r 92- and 72- k D a t y p e I V collagenases in HT-1080 cells. The cells were cultured and treated for varying lengths of time in the presence of TPA (10 nm), as described under “Experimental Procedures,” fol- lowed by Northern hybridization and densitometric analysis of films exposed at a range. Relative changes in levels of mRNA for the 92- and 72-kDa enzymes are shown as a function of time. The results are calculated as the mean of three separate analyses & S.D. All calcula- tions were adjusted using the intensity obtained with the glyceralde- hyde-3-phosphate dehydrogenase in each lane as an internal stand- ard. A, effects on 92-kDa enzyme mRNA levels. Basic level arbitrarily set as 5. R, effects on 72-kDa enzyme mRNA levels. Basic level arbitrarily set as 100. CHX, cycloheximide.

by nick translation and used for the isolation of several additional genomic clones.

DNA Sequencing-Sequencing of the HG-1 cDNA insert and of exons, exon-intron junctions, and the 5’- and 3”flanking regions of the 92-kDa type IV collagenase gene was carried out by the dideoxy- nucleotide chain termination procedure (25) using Sequenase (United States Biochemical Corp.) and M13 universal primers or specific oligonucleotide primers derived from the sequence of the 92-kDa type IV collagenase cDNA (6). In order to accomplish sequencing of appropriate regions of the gene, fragments of the genomic clones were identified by Southern analysis and subcloned into M13 derivatives.

Heteroduplex Analysis-Genomic DNAs (GL-1 and GL-3) were purified from phage using standard methods. The HG-1 cDNA insert in pUC19 was linearized with SalI, which cuts in the vector. The

TPA (nfl) - 0.10.3 1 3 10 30 - - - - I 3 IO TGFB (nglrnl) - - - - - - - 0 . 1 0 . 5 2 IO 2 2 2

FIG. 6. Influence of TPA and TGE”P1 on 72- and 92 -kDa type IV collagenases secreted by HT-1080 cells. Confluent cultures of HT-1080 cells were incuhated in the presence of TPA or TGF-/31 or both at 37 “C for 12 h. Aliquots of the conditioned medium were analyzed on 4.5% SDS-PAGE gel containing gelatin (3 mg/ml) as described under “Experimental Procedures.” Zones of enzyme activity are indicated by negative staining. Concentrations for TPA and TGF-61 are shown above the gelatinogram.

asymmetric localization of the cDNA in the vector allowed an un- ambiguous determination of the orientation of exon-intron segments in the heteroduplex. The phage DNAs and the cDNA were denatured with NaOH, neutralized with Tris-HCI, and renatured in 50% (v/v) formamide, 0.1 M Tris-HCI buffer, pH 8.5, 0.01 M EDTA a t room temperature. The resulting heteroduplexes were mounted onto par- lodion grids, contrast enhanced, and examined in a Zeiss electron microscope as described previously (26, 27). Standards for length measurements were 6x174 (5386 bases) and renatured pHG-1 DNA in the same electron micrographs.

DNA Amplification with the Polymerase Chain Reaction-PCR was used to amplify regions of the 5’-end of the gene for determining the size of introns, which could not be determined by heteroduplex analysis as the pHG-1 cDNA clone lacked about 650 bp from the 5’- end. The reactions were performed in the amplification mixture recommended by the manufacturer (Perkin-Elmer Cetus) using 10 ng of DNA from the genomic clone GL-1. Alternatively, a fragment of the genomic clone subcloned in M13 was used as template. The primer pairs used contained sequences from opposite strands of two neighboring exons and were used a t a concentration of 100 pmol/ reaction. Each cycle consisted of heating a t 94 “C for 1.5 min, an- nealing a t 37 “C for 2 min, and polymerization a t 72 “C for 4 min. The reaction was carried out for 25 rounds after which the samples were incubated for 10 min at 72 “C to extend all incomplete chains. The PCR products were precipitated and loaded on a 1.0% agarose gel along with standards. The gels were blotted on nitrocellulose filters using standard procedures and the filters were hybridized with ”2P-end-labeled oligonucleotides that were derived from a known sequence located between that corresponding to the primers used in the PCR reactions.

Primer Extension-Two oligonucleotides complementary to the 92-kDa type IV collagenase mRNA sequence (nucleotides 118-138 and 70-90, Fig. 4, Ref. 6) were made and used as primers in two separate experiments. Each primer was end-labeled by [y:”P]ATP using T 4 polynucleotide kinase (24) and hybridized (100,000 cpm) with 8 pg of poly(A) RNA isolated from 92-kDa type IV collagenase producing HT-1080 cells (28). The reverse transcription reaction was carried out under standard conditions (24). The primer extended products were run on a sequencing gel along with sequencing reactions from the human 92-kDa IV collagenase gene prepared using the same oligonucleotides as used in the primer extension assay.

Cell Cultures and Treatment with TPA and TGF-@l--Human fibrosarcoma cells (HT-1080, CCL-121) were cultivated in RPMI- 1640 medium containing 10% heat-inactivated fetal calf serum (GIBCO), 100 IU/ml of penicillin, and 50 pg/ml of streptomycin. The

16488 Human 92-kDa Type IV Collagenase Gene

Hunun 92-kDa T y p IV Collagenase Gena

Rat Stromalysin Gena l$? 246 146 126 18s 146 134 160 104 lS6

FIG. 7. Comparison of exon structure of human 92-kDa type IV collagenase gene with those of human 72-kDa type IV collagenase (20), human interstitial collagenase (17), and rat stromelysin (13, 19). Exons are indicated by boxes and introns by interconnecting lines. Exon numbers are shown in the boxes with sizes in base pairs below. Interrupted vertical lines illustrate intron locations that are identical in the genes (Ref. 20, and this study). The considerably shorter exon 9 in the 70-kDa type IV collagenase gene and exons 6 in the interstitial collagenase and stromelysin genes as compared with exon 9 in the 92-kDa type IV collagenase gene are indicated by interrupted boxes.

P R P E P E P R P P T T T T P Q P T A P P T V C P T G P P T V H P S E R P T A G P T G P P S CLG4B

arxn aG StrCnelySjn

A G P T G P P T A G P S T A T T V P L S P V D D A C N V N 1 , G D A I A E I G N Q L Y L F K D A S P D I D L G T G P T P T L G P V T P E I C K Q D I V ~ ~ G I A Q I R G E I F F F ~ ~

R S Q N P V Q P I G P Q T P K A C D S K L T F O A I T T l R G E V M F F t D P P P D S P E T P L V P V P P E P G T P A N C D P A L S F D A V S T L R G E I L I F K D

FIG. 8. Comparison and alignment of amino acid sequences encoded by exon 9 in the two human type IV collagenase genes and exon 6 in the human interstitial collagenase and rat transin (stromelysin) genes. The amino acid sequence encoded by exon 9 and 5 amino acids encoded by the adjacent exons are shown in boxes connected by shaded bars representing connecting introns. Amino acids conserved in all four enzymes are highlighted by shading. CLG4A, 72-kDa type IV collagenase; CLGIB, 92-kDa type IV collagenase; CLG, interstitial collagenase.

cultures were incubated at 37 "C in a humidified 5% CO, atmosphere. Confluent cultures were washed with serum-free RPMI-1640 medium and then incubatedunder serum-free conditions for 18 h. The medium was then replaced by fresh RPMI-1640, containing 0.1-30 nM TPA (Sigma, St. Louis, MO) or 0.2-2 ng/ml of porcine-derived TGF-Dl (R&D Systems, Minneapolis, MN) as indicated. The cultures were then incubated for an additional 1-24 h. All experiments were carried out under serum-free conditions.

Assay for Gelatinolytic Actiuity-Confluent cultures of HT-1080 cells were incubated in the presence of TPA or TGF-B1 at 37 "C for 12 h. The media were then collected and clarified by centrifugation. Aliquots (50 pl) of the conditioned media were analyzed in 4.5% SDS- PAGE containing 3 mg/ml of gelatin (29). After electrophoresis the gel was washed in 50 mM Tris-HC1, 5 M CaC12, 1 p M ZnC12, 2.5% Triton X-100, pH 7.6, for 30 min to remove SDS and rinsed in the above buffer without Triton X-100. The gel was incubated at 37 "C for 12 h in a buffer containing 50 mM Tris-HC1, 5 mM CaCl,, 1 p M ZnCl,, 1% Triton X-100, 0.02% NaN3, pH 7.6. Areas of gelatinolytic activity were visualized by staining the gel with 0.1% Coomassie Brilliant Blue in 50% methanol followed by destaining in 10% meth- anol, 10% acetic acid.

Isolation of mRNA and Northern Hybridization Analyses-Con- fluent cultures of cells were incubated under serum-free conditions for 18 h and exposed to TGF-/31 (2 ng/ml) and TPA (20 nM = 12 ng/ ml). Total cellular RNA was then prepared using acid guanidinium thiocyanate-phenol-chloroform extraction (30) and quantified by ab- sorbance at 260 nm. For Northern hybridization analysis 20 pg of RNA was fractionated on 0.8% agarose gels containing 2.2 M form- aldehyde. Gels were treated with 20 X SSC (1 X SSC: 0.15 M sodium chloride, 15 mM sodium citrate, pH 7.0) for 45 min at room temper- ature and RNA was transferred to Biodyne membranes (Pall Corpo- ration, Glen Cove, NY). Prehybridizations were carried out at 42 "C for 2 h in a hybridization mixture containing Denhardt's solution (2% each of Ficoll, polyvinylpyrrolidone, and bovine serum albumin), 5 X SSC (0.75 M sodium chloride, 75 mM sodium citrate, pH 7.0),

50% formamide, 50 mM HEPES, 200 pg/ml of salmon sperm DNA, and 150 pg/ml of yeast tRNA. Membranes were hybridized to the cDNA probes specific for 72- and 92-kDa type IV collagenases. A cDNA probe specific for glyceraldehyde-3-phosphate dehydrogenase was used as internal control. The cDNA probes were labeled with 3zP using the nick translation DNA labeling system (Amersham, United Kingdom; Promega). After hybridization, membranes were washed several times at 65 "C in 1 X SSC, 0.1% SDS. The filters were exposed to Kodak XAR-5 film at -70 "C for 6-48 h. Relative amounts of radioactivity of the bands were quantitated from autoradiograms with a computing laser densitometer (Molecular Dynamics, Sunnyvale, CA) using area scanning.

RESULTS

Isolation of cDNA and Genomic Clones-Screening of the HT-1080 cell cDNA library with the 92-kDa specific oligo- nucleotide probes resulted in the isolation of one 1.7-kb cDNA clone (HG-1). Nucleotide sequencing of this clone showed (data not shown) that it contained nucleotides 654-2,326 in the 2,334-bp sequence reported by Wilhelm et al. (6). There- fore, this clone lacked the sequence coding for the 5"untrans- lated region and the first 212 amino acid residues of the polypeptide chain. Screening of the genomic library with the cDNA insert in pUC19 (pHG-1) as well as with the oligonu- cleotides resulted in the isolation of several clones. Of those, two clones, GL-1 and GL-3, were characterized in detail by restriction mapping, Southern analysis, nucleotide sequenc- ing, and heteroduplex analysis together with analysis of cer- tain PCR amplified regions. The two genomic clones were shown to span about 26 kb of genomic DNA, including the entire structural gene of 7.7 kb, 15 kb of the 5'-flanking region, and 3.5 kb of the 3"flanking sequences (Fig. 1).

Human 92-kDa Type IV Collagenase Gene 16489

Exon-Intron Structure-The nucleotide sequence of the exons was determined using synthetic oligonucleotide primers that were synthesized on the basis of the sequence known from the cDNA. The sizes of introns 5-12 were determined by electron microscopic analysis of heteroduplexes between the pHG-1 cDNA clone and the genomic clones, and introns 7, 9, and 11 were also sequenced. Additionally, the sizes of introns 1-4 were obtained after PCR amplification of each intron and parts of the flanking exons followed by size deter- mination from agarose gels.

These analyses showed that the human 92-kDa type IV collagenase gene contains 13 exons whose sizes vary between 104 and 310 (Fig. 2). The introns range in size between 96 and 1800 bp. Accordingly, the size of the structural gene is 7.7 kb. All translated exons, except for exon 2, start with the second or third base of a codon. Primer extension analysis (Fig. 3) revealed that exon 1 contains 157 bp, with only a short 19-bp 5”untranslated region and 138 bp of a translated sequence. Exons 5, 6, and 7 are 174, 174, and 177 bp, respec- tively, each of them coding for an internal repeat which resembles the fibronectin collagen-binding domain. Thus, the internal repeat encoded by exon 7 contains one more amino acid than the corresponding repeat in the 72-kDa type IV collagenase gene (4,ZO). Exon 9 is the largest translated exon, containing 280 bp. This exon contains 144 bp coding for a 48- residue sequence that does not have a counterpart in 72-kDa type IV collagenase. Additionally, this exon has 118 bp coding for an amino acid sequence homologous to the mammalian metalloproteinases. Sequencing of both strands of exon 12 in the genomic clone GL-3 revealed one base difference from the sequence in the HG-1 cDNA clone (this study, data not shown) and the cDNA sequence reported by Wilhelm et al. (6). The difference involved the second base A in the codon for glutamine in the genomic clone and the second base G in the CGA codon for arginine in the cDNA (Fig. 2). This apparent polymorphism led to a conservative amino acid substitution.

5’-End and -Flanking Region-The initiation site for tran- scription was determined by primer extension analysis. A 21- mer oligonucleotide complementary to nucleotides 118-138 in the first exon was used as primer. The primer extension resulted in a clear single band indicating a single site for the start of transcription of this gene (Fig. 3). This result was verified in a second experiment using an oligonucleotide com- plementary to nucleotides 70-90 as primer (data not shown).

Analysis of the sequence of the 5”flanking region (Fig. 4) revealed that the promoter has no TATA motif but there is a sequence TTAAA at position -29 to -25. There was no CAAT motif observed in the promoter region, but one GC box that can serve as binding site for the transcription factor Sp l (31) was found between positions -563 and -558. We located two sequences which have been shown to confer TPA responsive- ness and which can serve as binding sites for the transcription factor AP-1 (32). One sequence, TGAGTCA, at positions -533 to -527 and another one between positions -79 to -73 were found. A sequence, GGTTTGGGGA, which fulfills the criteria for the consensus sequence (GNNTTGGNGN) of a TGF-P1 inhibitory element (33) was found between -474 and -465. A segment of 42 nucleotides of alternating C and A residues was present at positions -131 to -90.

Influence of TPA and TGF-PI on mRNA Levels for 72- and 92-kDa Type IV Collagenase in HT-1080 Cells-Northern hybridization analysis revealed that cultured HT-1080 cells express very low amounts of 92-kDa type IV collagenase mRNA (data not shown). However, treatment of the cells with TPA resulted in a significant induction of 92-kDa type

IV collagenase-specific mRNA expressed (Fig. 5A). An in- crease was not observed after 1 h of incubation, but after 6 h the increase was about 12-fold, and after 12 h it was 17-fold. The presence of cycloheximide during the incubation blocked practically completely the effect of TPA, indicating that pro- tein synthesis was required for TPA induction. Because a putative TGF-/3 inhibitory element was found in the promoter region of the 92-kDa type IV collagenase gene, we examined the effect of TGF-Dl on the mRNA level for this enzyme. However, when TGF-Pl was added to the HT-1080 cell cul- tures alone or together with TPA the decrease in the levels of the 92-kDa enzyme mRNA was not statistically significant (data not shown).

In contrast to the observation for the 92-kDa type IV collagenase gene, the Northern analysis demonstrated that the HT-1080 cells had a considerable level of the 72-kDa enzyme mRNA without any treatment. Instead of increasing the level of mRNA for the 72-kDa enzyme, TPA reduced it roughly to one-half (Fig. 5B). This reduction was not affected by the addition of cycloheximide. TGF-Pl caused a slight induction (data not shown) on the expression of the 72-kDa enzyme gene after 6 h. When TGF-Dl and TPA were added together the results were similar as when TGF-01 was added alone.

Influence of TPA and TGF-@I on the Amounts of 72- and 92-kDa Type I V Collagenase (Gelatinase) Activity Secreted by HT-1080 Cells-The amounts of the two type IV collagenase proteins in the culture medium of HT-1080 cells were quan- tified by a gelatinase assay following treatment of the cells with TPA, TGF-Dl, or both for 12 h. The results revealed that the cells secrete very low amounts of 92-kDa type IV collagenase activity (Fig. 6). However, treatment of the cells with 0.1-30 nM TPA resulted in a significant elevation of enzyme activity secreted. TGF-Pl (0.1-10 ng/ml) alone did not seem to influence the level of activity nor did it affect the enzyme activity when added concomitantly with TPA (Fig. 6). As seen in Fig. 6, the cells secreted constitutively 72-kDa type IV collagenase. Significant changes in the level of activity by TPA, TGF-P1, or both were not visible in this assay system.

DISCUSSION

The 72- and 92-kDa type IV collagenases share many common features including identical substrate specificity (3, 4, 16, 22), but regulation of their gene expression appears to vary quite distinctly. We have recently reported the complete structure of the human gene for the 72-kDa enzyme (20). The present study, which provides the entire gene structure for the 92-kDa enzyme, demonstrates that the genes for the two enzymes are closely related but that the promoter regions differ markedly. The 92-kDa type IV collagenase gene has 13 exons which is the same as in the gene for the 72-kDa enzyme (Fig. 6). However, the 92-kDa enzyme gene is only about 7.7 kb while the gene for the 72-kDa enzyme is about 27 kb (20). The 92-kDa type IV collagenase gene is similar in size to the interstitial collagenase and stromelysin genes (13, 17-19). All these genes clearly belong to the same gene family based on their structural resemblance and the similarities of the gene products. For example, comparison of exon size pattern (Fig. 6) and alignment of sequences at intron boundaries demon- strate that all intron locations coincide in the 92- and 72-kDa type IV collagenase genes (Fig. 7; Ref. 20). Although the interstitial collagenase and stromelysin gene each has three fewer exons than the two type IV collagenase genes, their corresponding introns have identical locations with respect to the exons. Each of the three extra exons of the latter encode

16490 Human 92-kDa Type IV Collagenase Gene

a single fibronectin-like collagen-binding domain (20, this study),

The 92-kDa type IV collagenase protein was reported (6) to distinguish from the 72-kDa enzyme by the presence of a 54-amino acid residue proline-rich sequence located adjacent to the carboxyl-terminal side of the zinc-binding domain. It was reported to have between 30 and 50% identity with a part of a collagenous sequence in the collagen a2(V) chain (6). This sequence does not, however, have the Gly-X-Y-repeat pattern characteristic for collagen. It was proposed (6) that the 54-residue sequence might have been acquired through a recombination event involving insertion from a collagen gene of exon(s) with the typical 54-bp size (or multiple of 54 bp) into the 92-kDa enzyme gene. The present study on the gene provides a better possibility for the alignment of these se- quences and demonstrates that the nucleotide sequence en- coding the unique segment is 48 residues rather than 54 residues long. This sequence is encoded by exon 9 which also codes for amino acid sequences homologous to other mam- malian metalloproteinases (Fig. 8). It is quite possible that the 92-kDa type IV collagenase gene represents an ancient proteinase gene from which the other metalloproteinases have evolved by the loss of coding sequences. However, the possi- bility of an uptake of additional coding sequences into exon 9 of the 92-kDa enzyme cannot be excluded.

Sequencing analysis of the promoter region of the 92-kDa type IV collagenase gene revealed, that in many respects, it resembles more the promoters of the interstitial collagenase and stromelysin (transin) genes than that of the 72-kDa type IV collagenase gene. For example, the 92-kDa type IV colla- genase promoter has a TATA-like sequence (TTAAA) at positions -29 to -25. The human stromelysin gene also has a TTAAA sequence in the same region (34), and rat strome- lysin and rabbit and the human interstitial collagenase genes have true TATA motifs at the same location (13, 18, 32). In contrast, the 72-kDa type IV collagenase gene has no TATA motif-like sequence in the vicinity of the start site for tran- scription (20). Furthermore, the 92-kDa type IV collagenase gene has a TPA response element-like sequence at positions -79 to -73, similar to the rat stromelysin gene (13) and the rabbit and human interstitial collagenase genes (18, 32). A second TPA response element was found further upstream. These findings are in agreement with the induction of 92-kDa enzyme mRNA levels. No TPA response element has been observed in the 72-kDa type IV collagenase promoter (20). The reason for the decrease in mRNA levels for the following TPA treatment is not known. The results of the Northern analysis correspond well with the findings observed in the gelatinase assays that indicate levels of secreted enzyme.

TGF-B1 has been shown to inhibit growth factor and on- cogene induction of transin (33). This inhibition was mediated through a Fos-binding sequence, GAGTTGGTGA. A consen- sus sequence, GNNTTGGNGN, has been found in the pro- moter region of other TGF-Bl inhibited genes such as uroki- nase, elastase, interstitial collagenase, mitogen-regulated pro- tein-proliferin, and c-myc (see Ref. 33). One copy of this sequence was present in the 92-kDa type IV collagenase promoter analyzed. In this study we could not demonstrate that the TGF-Pl effects significantly the level of mRNA in the TPA treated or untreated HT-1080 cells. However, the

functional role of this and other potential regulatory se- quences of the promoter remain to be determined.

In conclusion, the two type IV collagenase enzymes, al- though having similar substrate specificity, differ consider- ably with respect to their gene regulation. Understanding of their biological function as well as role in pathological states such as in malignant transformation is still limited. The recent availability of cDNA and genomic probes allows more sophisticated approaches to explore the molecular biology and pathology of these enzymes.

Acknowledgment-We are grateful to Maire Jarva for expert tech- nical assistance.

REFERENCES 1. Tryggvason, K., Hoyhtya, M., and Salo, T. (1987) Biochirn. Biophys. Acta

2. Harris, E. D., Welgus, H. G., and Krane, S. M. (1984) Collagen Relat. Res.

3. Salo, T., Liotta, L. A,, and Tryggvason, K. (1983) J. Bid . Chem. 268 ,

907,191-217

4,493-512

m h c u m 4. Collier, I. E., Wilhelm, S. M., Eisen, A. Z., Marmer, B. L., Grant, G. A,,

Seltzer, J. L., Kronberger, A., He, C., Bauer, E. A., and Goldberg, G. I. (1988) J. Biol. Chem. 263.6579-6587

- . . - - - - -

5. Wilhelm,S.~M., Collier,-I. E., Kronber er, A., Eisen, A. Z., Marmer, B. L., Grant, G. A., Bauer, E. A., and Gol&rg, G. I. (1987) Proc. Natl. Acad. Scc. U. S. A. 84,6725-6729

6. Wilhelm, S. M., Collier, I. E., Marmer B. L., Eisen A. Z. Grant, G. A., and Goldberg, G. I. (1989) J. Biol. Cr$rn. 264 , 172i3-17i21

7. Murphy, G., Cawston, T. E., and Reynolds, J. J. (1981) Biochern. J . 195 ,

I~~ ~ ~

5. Wilhelm,S.~M., Collier,-I. E., Kronber er, A., Eisen, A. Z., Marmer, B. L., Grant, G. A., Bauer, E. A., and Gol&rg, G. I. (1987) Proc. Natl. Acad. Scc. U. S. A. 84,6725-6729

6. Wilhelm, S. M., Collier, I. E., Marmer, B. L., Eisen, A. Z., Grant, G. A., and Goldbere. G. I. (1989) J. Biol. Chern. 264. 17213-17221

I~~ ~ ~

7. Murphy, G., Ciwston,'T. E:, and Reynolds, J. J: (1981) Biochern. J . 195 , 167-170

8. Stetler-Stevenson, W. G., Krutzsch, H. C., and Liotta, L. A. (1989) J. Biol. Chem. 264,17374-17378

9. Goldberg, G. I., Wilhelm, S. M., Kronberger, A., Bauer, E. A., Grant, G. A.,

10. Hasty, K. A., Pourmotabbed, T. F., Goldberg, G. I., Thompson, J. P., and Eisen, A. 2. (1986) J. Biol. Chem. 261,6600-6605

Spinella, D. G., Stevens, R. M., and Mainardi, C. I. (1990) J. Biol. Chem.

11. Whitham, S. E., Murphy, G., Angel, P., Rahmsdorf, H.-J., Smith, B. J., 266,11421-11424

J. P. (1987) Biochem. J. 240,913-916 Lyons, A., Harris, T., Jr., Reynolds, J. J., Herrlich, P., and Docherty, A.

12. Matrisian. L. M.. Glaichenhaus. N.. Gesnel. M. C.. and Breathnach. R.

13.

14.

15.

16.

17.

18.

19.

20.

21.

22.

23.

24.

25.

26.

27.

28.

29.

31. 30.

32.

33. 34.

(1985) EMBO 3. 4,1435-1440 '

R. (1986) Mol. Cell. Biol. 6, 1679-1686

and Breathnach, R. (1988) Biochern. J. 263,187-192

Tryggvason, K. (1990) Genornics 6,554-559

Tryggvason, K. (1989) Biochern. J. 258,463-472

A. Z., and Goldberg, G. I. (1988) J. Biol. Chern. 263,10711-10713

I ~I ~ ~ ~ ~~~ ,~

Matrisian, L. M., Leroy, P., Ruhlman, C., Gesnel, M.-C., and Breathnacb,

Muller, D., Quantin, B., Gesnel, M.-C., Milon-Collard, R., Abecassis, J.,

Huhtala, P., Eddy, R. L., Fan, Y. S., Byers, M. G., Shows, T. B., and

Murphy, G., Ward, R., Hembry, R. M., Reynolds, J. J., Kuhn, K., and

Collier, E. I., Smith, J., Kronberger, A., Bauer, E. A., Wilhelm, S. M., Eisen,

Fini, M. E., Plucinska, I. M., Mayer, A. S., Gross, R. H., and Brincherhoff,

Breathnach, R., Matrisian, L., Gesnel, M.-C., Staub, A., and Leroy, P.

Huhtala, P., Chow, L. T., and Tryggvason, K. (1990) J. Biol. Chern. 266 ,

C. E. (1987) Biochemistry 26,6156-6165

(1987) Nucleic Aculs Res. 16,1139-1151

I 1~7-11082 Yamada, Y., Awidimento, V. E., Mudryj, M., Ohkubo, H., Vogeli, G., Irani,

Fessler, L., Duncan, K., Fessler, J. H., Salo, T., and Tryggvason, K. (1984) M., Pastan, I., and de Grombrugghe, B. (1980) Cell 22,887-892

Salo, T., Turpeenniemi-Hujanen, T., and Tryggvason, K. (1985) J. Biol. J. Bwl. Chem. 269,9783-9789

Ausubel, F. M., Brent, R., Kingston, R. E., Moore, D. O., Sedman, J. G., Chern. 260,8526-8531

Smith, J. A., and Struhl, K. (1989) Current Protocols in Molecular Biology, Wiley-Interscience, New York

Sanger, F., Nicklen, S., and Coulson, A. P. (1977) Proc. Natl. Acad. Sci. U. S. A. 74,5463-5467

Chow, L. T., Broker, T. R., and Lowis, J. B. (1979) J. Mol. Biol. 134,265- 303

Lo ata, M. A., Havercroff, J. C., Chow, L. T., and Cleveland, D. W. (1983) &ll 3 2 , 713-724

Alitalo, K., Vaheri, A., Krieg, T., and Timpl, R. (1980) Eur. J. Biochem. 109,247-255

Hibbs, M. S., Hasty, K. A., Seyer, J. M., Kang, A. H., and Mainardi, C. L.

Chomczynski, P., and Sacchi, N. (1987) Anal. Biochem. 162,156-159 (1985) J. Biol. Chem. 260,2493-2500

Lee, W., Mitchell, P., and Tjian, R. (1987) Cell 49 , 741-752 Angel, P., Imagawa, M., Chiu, R., Stein, B., Imbra, R. J., Rahmsdorf, H. J.,

Jonat, C., Herrllch, P., and K a r q M. (1987) Cell 4 9 , 729-739 Kerr, L. D., Miller, D. B., and Matrialan, L. M. (1990) Cell 6 1 , 267-278

umones, S., Saus, J., Otani, Y., Harris, E. D., Jr., and Kurkinen, M. (i989) J. Biol. Chem. 264,8339-8344

"."