Isolation and Characterization of the Human Tissue-type Plasminogen Activator Structural Gene Including Its 5 ’ Flanking Region *

mRNA specific for tissue type plasminogen activator (t-PA) is induced in HeLa cells by the tumor promoter phorbol myristate acetate (Waller, E. K., and Schleuning, W. D. (1985) J. Biol. Chem. 260, 6354-6360). To study the underlying mechanism, a cDNA library was constructed from phorbol myristate acetate stimulated HeLa cell mRNA and screened with two t-PA mRNA specific oligonucleotides (Edlund, T., Ny, T., Rhnby, M., Heden, L.-O., Palm, G., Holmgren, E., and Josephson, S. (1983) Proc. Natl. Acad. Sci. U. S. A. 80, 349-352). Two nearly full length double-stranded cDNA clones were obtained. Suitable restriction fragments from the cDNA were employed as probes for the isolation of three recombinant bacteriophages, containing overlapping fragments of the t-PA gene. By restriction analysis, heteroduplex mapping, and DNA sequencing it was determined that the three overlapping fragments contain the complete t-PA structural gene and that the 2658 bases long t-PA mRNA is encoded by a gene of approximately 29 kilobases overall length, which is interrupted by 13 introns. To characterize the presumptive control region, a subcloned gene fragment, containing the 5’ sequence of the cDNA, was sequenced, and the transcription initiation site was identified by nuclease S1 protection experiments. The putative transcription start site is located 24 base pairs (bp) downstream of a typical TATA consensus sequence. Two additional TATA motifs with hitherto unknown functions are found 93 and 226 bp upstream of the putative cap site. A recombinant plasmid was constructed, which accommodates the cap site including 475 bp of upstream sequences, fused to a double-stranded cDNA of t-PA mRNA which contains the complete translated and parts of the 5’ and 3‘ untranslated regions. This plasmid directs t-PA biosynthesis in Xenopus laevis oocytes after microinjection into the germinal vesicle.

To study the underlying mechanism, a cDNA library was constructed from phorbol myristate acetate stimulated HeLa cell mRNA and screened with two t-PA mRNA specific oligonucleotides (Edlund, T., Ny, T., 80, 349-352). Two nearly full length double-stranded cDNA clones were obtained. Suitable restriction fragments from the cDNA were employed as probes for the isolation of three recombinant bacteriophages, containing overlapping fragments of the t-PA gene. By restriction analysis, heteroduplex mapping, and DNA sequencing it was determined that the three overlapping fragments contain the complete t-PA structural gene and that the 2658 bases long t-PA mRNA is encoded by a gene of approximately 29 kilobases overall length, which is interrupted by 13 introns. To characterize the presumptive control region, a subcloned gene fragment, containing the 5' sequence of the cDNA, was sequenced, and the transcription initiation site was identified by nuclease S1 protection experiments. The putative transcription start site is located 24 base pairs (bp) downstream of a typical TATA consensus sequence. Two additional TATA motifs with hitherto unknown functions are found 93 and 226 bp upstream of the putative cap site. A recombinant plasmid was constructed, which accommodates the cap site including 475 bp of upstream sequences, fused to a double-stranded cDNA of t-PA mRNA which contains the complete translated and parts of the 5' and 3' untranslated regions. This plasmid directs t-PA biosynthesis in Xenopus laevis oocytes after microinjection into the germinal vesicle.
A better understanding of the biochemical mechanism which underlies the control of cell proliferation by polypeptide hormones and tumor promoters is essential for further insights into the molecular biology of regular and aberrant cellular growth and differentiation. Many growth-promoting ~~ * This work was supported by Grant 3.350-0.82 from Fonds National Suisse and by a grant from Biogen N.V., Cambridge, MA. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "aduertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
factors, e.g. the tumor promoter PMA' (1-4), the epidermal growth factor (5) and tumor angiogenesis factor (6) as well as the viral oncogene product pp60 src (7) induce fibrinolytic activity in specific target cells. It was recently demonstrated that in HeLa cells this increased fibrinolytic activity results from increased apparent levels of t-PA-specific mRNA and a corresponding increase of t-PA biosynthesis (8). t-PA is a highly specific protease, which catalyzes the conversion of the zymogen plasminogen to plasmin, in a reaction activated by fibrin (9). Plasmin, a trypsin-like protease, is the major fibrinolytic enzyme, and it is also required for intercellular matrix degradation, during tissue growth and remodeling (10). In order to initiate a study of the control mechanism involved in the regulation of the t-PA gene on a molecular level, we have cloned two nearly full length dscDNAs coding for t-PA and used appropriate restriction fragments and synthetic oligonucleotides for the isolation of three overlapping genomic clones, which span the complete structural t-PA gene including sequences 5' upstream of the cap site. Construction of a cDNA Library and Identification of t-PA Positive Clones-8 pg of size-selected (21-22 S) mRNA from PMA-treated HeLa cells were copied into double-stranded cDNA essentially as The abbreviations used are: PMA, phorbol myristate acetate; t-PA, tissue-type plasminogen activator; dscDNA, double-stranded cDNA; bp, base pair(s); kb, kilobase(s). 11223 described by Wickens et ai. (14). The dscDNA was nicked with S1 Tris/HCl, pH 7.4, 100 mM NaCI, 1 mM EDTA for 10 h a t 39,000 rpm nuclease and size selected on 5-20% (w/v) sucrose gradient in 10 mM in a Beckman SW 41 rotor. The gradient was fractionated from the bottom, and those fractions containing the largest size cDNA (about one-third of the total) were pooled and precipitated a t -20 "C by the addition of NaCl to 0.3 M, 5 pg/ml bovine serum albumin, and 2 volumes of ethanol. 25 ng of size-fractionated dscDNA were tailed with dC according to Deng and Wu (15) and annealed to pBR 322, tailed with dG at the PstI site (16). Competent Escherichia coli HB 101/LM 1035 were transformed and plated as described by Morrison (17). Transformants were plucked from isolated colonies into 96well microtiter plates containing 200 p1 of L broth, 10 pg/ml tetracycline, grown overnight a t 37 "C, and then stored frozen a t -80 "C after the addition of glycerol to 5% (v/v). Two synthetic oligonucleotide probes containing the sequences (Pl) d5"GGCAAA-GATGGCAGCCTGCAAG-3' and (P2) d5"GCTGTACTGTCT-CAGGCCGCAG-3' were synthesized and provided by Hans Rink (Ciba-Geigy Ltd., Basel, Switzerland). Oligonucleotides were labeled with [32P]ATP on the 5' terminus using T4 polynucleotide kinase to a specific activity of 3 X lo6 cpm/pmol (18).
Cultures were transferred from microtiter plates to Millipore HATF nitrocellulose filters, grown for 18 h, and the plasmids amplified as described by Grunstein and Hogness (19). Fixation of DNA to the filters and hybridization and washings were done as described by Wallace et al. (20) using the empirical relationship between the length of the oligonucleotide probe, its G + C content, and the T,,, that have been determined by Suggs et al. (21) and Smith (22). The microtiter wells that produced positive hybridization signals were identified and single colonies from these cultures rescreened as above. Colonies showing positive hybridization signals with both probes were grown and plasmids obtained using the alkaline lysis method of Birnboim and Doly (23).
Screening a Human X Phage Library-A human genomic X Ch4A phage library constructed by Lawn et al. (24) was screened with 32P nick-translated DNA probes by plaque hybridization (25) in two steps. By using the 472-bp EcoRI fragment of p P A l l 4B (Fig. 2a) as a probe, two overlapping recombinants, X PA 5.2 and X PA 8.3, were isolated from a screen of more than lo6 plaques. To isolate a phage containing the transcriptional start site and 5' flanking sequences, a second plaque hybridization was performed using a 282-bp HgiAI fragment from pPAll4B that included 5' noncoding sequences. This second hybridization yielded X PA 4, which, as described below, contains the first two exons and 5'-flanking sequences of the t-PA gene. Preparation and Analysis of Cloned DNA-Recombinant phage from large-scale lysates were purified by cesium chloride density gradient. centrifugation, and DNA was obtained from phage by phenol/chloroform extraction (18). The purified DNAs were cleaved with various restriction enzymes, either singly or in combination, and the DNA fragments transferred from gels to nitrocellulose as described by Southern (261, BamHIIEcoRI and EcoRI fragments were subcloned into the polylinker region of plasmid pUC 8 (as), and relevant colonies identified with suitable restriction fragments of p P A l l 4B by colony hybridization (19). pBG 198, a pUC 8 derivative, containing a 757-bp BamHIIEcoRI fragment of X PA 4, was identified by hybridization to a 282-bp HgiAI fragment of p P A l l 4B and PA XVI, a synthetic oligonucleotide containing the sequence d5"ACAG-GAGTCCAGGGCTGGAG-3' (kindly provided by Dr. Chandra Rascribed by Maxam and Gilbert (28). machandran, Biogen Inc.). DNA sequencing was carried out as de-Heteroduplex Analysis-Heteroduplex analysis was performed as previously described (291, hybridizing X PA 4, X PA 5.2, and X PA 8.3 to BamHI-cut pPAll 4B. Construction of pTPA 114, a Plasmid Containing t-PA dscDNA Fused to 476 bp of 5' Flanking Sequence-The 757-bp EcoRI-BamHI fragment from X PA 4 was inserted into the polylinker region of PUC 8 (27). The resulting plasmid was cut with BamHI, dephosphorylated with calf alkaline phosphatase, and a "t-PA-BamHI cassette"' ligated into the BamHI site, using T4 ligase. Briefly the "t-PA-BamHI cassette" was derived from pPAl1 4B, in which the most 5' HgiAI site of t-PA cDNA was converted into a BamHI site by blunting the protruding end and ligating it to the respective synthetic oligonucleotide linker. The 3' BglII site of p P A l l 4B was fused to the SV40 small t polyadenylation signal contained in pSV2DHFR (30). Thus, * R. Fisher, unpublished data. pTPA 114 contained 475 bp of genomic sequence upstream of the putative cap site, the first exon, 99 bp of intron a, the complete cDNA sequence from p P A l l 4B except the 25 5' nucleotides, and the sequences downstream of the BglII site in the 3' untranslated region. These sequences were replaced by the small t polyadenylation signal from pSV2DHFR.
Translation and Transcription in Xenopus laevis Oocytes-Toads were maintained as described (31), without any hormone injections. Ovaries were removed by dissection, teased apart with forceps, and treated for 2 h a t room temperature with 2 mg/ml Sigma type I collagenase. Stage 6 oocytes were selected by visual inspection, maintained in modified Barth's medium, and injected with mRNA from the sucrose gradient or pTPA 114 DNA essentially as described by Gurdon (31).
Nuclease SI Mapping-Nuclease S1 protection experiments were carried out essentially as described (32). The nuclease SI probe was obtained as follows. The subclone pBG 198 was digested with Hinfl, treated with calf intestine alkaline phosphatase, and 5' end labeled with [32P]ATP using T4 polynucleotide kinase. The 249-bp fragment was isolated by electrophoresis on a 5% polyacrylamide gel and strand separated on a 6% polyacrylamide gel. 5 pg of poly(A+)-selected RNA from PMA-stimulated HeLa cells were hybridized to 0.1-pg singlestranded fragment in a volume of 20 pl, diluted to 200 ~1 with S1 buffer, and digested for 30 min a t 30 "C with lo3 units of nuclease S1.
The solution was ethanol precipitated and analyzed by electrophoresis on a 6% acrylamide gel in TBE, 7 M urea. sequence d5'-CAGCAGAGCCCTCTCTTCATTG-3' (complemen-Primer Extension-An oligonucleotide (22 mer) containing the tary to t-PA mRNA just prior to the initiation codon) was generously provided by Dr. J. Jiricny, Friedrich Miescher Institute, Basel, Switzerland. The oligonucleotide was labeled with [32P]ATP on the 5' terminus using T4 polynucleotide kinase to a specific activity of 5 x lo6 cpm/pmol (28). 30 pmol of end-labeled oligonucleotide were annealed to 300 pg of HeLa poly(A+) RNA, enriched for t-PA sequences by size fractionation on a sucrose gradient as described above, and extended a t 42 "C for 60 min by the addition of 300 units of reverse transcriptase in a 500-pl solution containing 65 mM Tris-HC1, pH 8.3, 65 mM KC1, 8 mM MgC12, 0.5 mM each dNTP, 2 mM dithiothreitol, and 0.5 microunit/pl RNasin. The reaction was stopped by the addition of EDTA to 20 mM and the nucleic acids precipitated a t -70 "C by the addition of 2 volumes of ethanol. Nucleic acids were collected by centrifugation, washed in 70% v/v ethanol, dried, and dissolved in 500 pl of H20. NaOH was added to 10 mM and urea to 8 M, the solution heat denatured a t 90 "C for 3 min, chilled on ice, the cDNA products fractionated by electrophoresis on a 6% w/v denaturating polyacrylamide sequencing gel and recovered and sequenced as described (28).

Molecular
Cloning of t-PA-specific dscDNA from PMAinduced HeLa Cells-From one confluent culture flask (175 cm') -2 mg of total cellular RNA were obtained, 1.5% of which contained poly(A)-rich sequences. When this RNA was used to direct protein biosynthesis in a wheat germ-derived cell-free translation system in the presence of [35S]methionine, a protein of M, 63,000 could be immune precipitated by anti-HeLa t-PA IgG but not by nonimmune IgG (not shown). By comparing the radioactivity of immune-precipitable to trichloroacetic acid-precipitable material, it was estimated that t-PA-specific mRNA accounted for 0.01% of the total mRNA in PMA-stimulated HeLa cells. The presence of t -PA-specific mRNA in the HeLa cell RNA preparation was also demonstrated by microinjection of RNA into X . laevis oocytes as previously described (10). Zones of caseinolysis developed around oocytes injected with poly(A+) RNA. Caseinolysis was dependent on the presence of plasminogen in the medium (not shown) and blocked by the inclusion of anti-HeLa t-PA IgG (Fig. 1). The size of HeLa t-PA-mRNA was determined to be 21-22 S by sucrose gradient centrifugation, thiit included 3.5 kb of intron and the adjacent exon in X PA merit from X PA 4 which hybridized to the 5' noncoding 8 2 , described above and in Figs. 3 and 5. This interpretation synthetic probe, PA XVI (Fig. 3). The sequence of this EcoRIwas confirmed by electron microscopic analysis of heterodu-BamHI fragment revealed the donor site of the first intron in plc xes between X PA 4 with X PA 8.3 (results not shown) and the t-PA gene; the acceptor site resides in an EcoRI-BamHI by DNA sequence analysis of a 757-bp EcoRI-BarnHI fragfragment present also in X PA 8.  5.2 (left panel), pPA11 413  and A P A 8.3 (right panel). Interpretotive drnwings are represented in the lorrc'r pnnds. T h e lenkqhs ol introns am and exons 1-1.1 are given in Table 1 of' exons 1 and 2 on the insert of X PA 4 and exon 2 on the insert of X PA 8.3 indicated that the 5' noncoding sequence oft he t-PA gene is split by a single 9.5-kb intron, called intron a in Fig. 5 .
The sequence of the 757-hp fragment shown in Fig. 3 also revealed three consensus "TATA hoxes" which can determine downstream transcription initiation sites ( 3 5 ) . To define t,he t-PA gene transcription initiation site a 249-hp HinfI fragment from the 755-hp insert was 5' labeled with [:"I-']ATP using T4 kinase, strand separated, and used for nuclease SI protection experiments with mRNA from HeLa cells as described under "Experimental Procedures." The result ofthis analysis showed a major protected sequence of about 130 hp. In addition to this major protected sequence, a minor larger sequence ('LOO hp) was defined when a longer exposure of the gel was analyzed (not shown). If the major protected fragment was electrophoresed in parallel with a Maxam/Gilhert sequence ladder of the 'L49-hp HinfI fragment, it was shown that the major transcript ion initiation site of t-PA is located 24 nucleotides downstream from a consensus TATA box (Fig.  6). t-PA Expression by X . 1acui.s Oocytcs aftcr Intmnuclcar Injwlion of pl'A 1 /.!-After injection of 10 ng of pTPA 114 ( Fig. 7 ) into the germinal vesicle of X . lawis oocytes -50% of the treated oocytes produced plasminogen-dependent lysis zones in plasminogen casein agar. The development of these lysis zones could be inhibited hy addition of anti-t-PA IgG ( I O pg/ml) to the agar. As judged from a standard curve the average production rate was 20 pg of t-I'A/oocvte/24 h (Fig.   8).
An approximately equivalent production rate was ohtained when p'l'PA 25, a plasmid containing the SV40 large T promoter in equivalent position upstream o f the t-PA cl)NA, was injected. Only 2 0 5 of the oocytes produced t -P A after inject inn o f p T P A 1 I:i, a plasmid identical to pTI'A 1 14 except that the t -P A cDNA was inserted in reverse orientation (not shown).

I)IS('('SSION
We have cloned two nearlv complete cl)NAs coding for t - Assuming that we enriched X-fold for t -PA-specific sequences t)y sucrose gradient centrifugation, this valne is in good a g r w merit with the results obtained hy in r*itrfJ transhtion o f llnfractionated mHNA, o f which O . O l p ; were t-I'A specific.
The sequence o f both cDNAs in the 5' rlntranslated and the translated regions is identical t o the sequence reported Pennica ct al. ( 3 3 ) except for the differences indicated in Fig.  2. We have confirmed our cDNA structure hy primer extension and by sequencing a genomic fragment. Therefore, the minor differences could result from a cloning artifact. Restriction fragments of the nearly complete rDNA clone allowed us to isolate and characterize the structural gene encoding human t -P A . This gene extends over 29 kh and is interrupted by 14 introns. The largest exon is a O.62-kh fragment located at the 3' end of the gene which encodes mainly 3' untranslated sequences. The restriction maps of the three independently isolated clones overlap to fnrm a unique map. The human t-PA gene is, therefore, most likely a single copy gene. T h e first intron of 9.5 kh is unusually large. While we were unahle to ohtain cln unamhigunus hetemduplex map, we could still recognize that it contains at least one inverted repeat forming a stem-loop strrlcture.
An inverted repeat forming a stem-loop structure and a simplr inverted repeat are also found in intron h and intron with I'MA-treated He1,a cell mRNA and SI digestion. relative t o a 1)NA seqwnce latltler o f the same fragment. Note that the sequence ol'the anti-sense strand is given. The putative transrription start site is indicated I)v a n orrow at the sequence of interest. fragment of' the t -P A gene, starting 5' with a fragment of intron a and stretching downstream heyond the 3' end of the gene. Our results confirm and extend this report except with respect to exon 3 and exon 14 (Tahle I). A s exon 3 is very short (43 h p ) the limitat,ion of our experimental approach may account for the observed difference. However, the size of exnn 14 is, according to these authors, significantly larger (919 hp) than our results indicate (620 hp). We, therefore, conclude that X PA 5.2 does probably not contain the complete sites begin with an A following a C, we have tentatively assigned this function to the indicated nucleotide. In addition t o t h e 130-hp SI-resistant, fragment, a longer exposure of the gel revealed a -200 hp fragment, which could have heen protected hy a mRNA initiated at a possible transcription start site 73 hp upstream of the major cap site. T h e significance of t.his minor mRNA species is at present not known. X. lacuis oocytes, injected into the germinal vesicle with pTPA 114, synt,hesized t-PA activity, which could he quenched hy anti-t-PA antihodies. The results of these experiment,s indicate that pTPA 114 can serve as a n efficient templat,e for transcription in X. faeuis oocytes and that a mRNA is produced that can direct t-PA biosynthesis. It could he argued that oocyte DNA could provide signals for transrript.ion initiation. This appears unlikely, however, for two reasons: first, circular douhle-stranded DNA is rarely integrat.ed after injection into the nucleus, hut is conserved and assemhled into chromatin (38), and second, pTPA 115, which contains t-PA cDNA in inverse orientation to the putative promoter, did give significantly less positive signals after injection. At the time this report was written, we had not yet heen ahle to oht,ain unequivocal S1 nuclease maps of RNA, isolated from injected oocytes. prohahly because of low ahundance of the respective mRNA. I t is, therefore, not known whether the t-PA mRNA in the oocvte is initiated in the correct position. Other authors, however, have shown that mRNAs with authentic 5' termini were 20-40-fold more ahundant than read through transcripts when a RNA polymerase I1 gene was transcrihed from a circular template in oocyte nuclei (39). and that at least three sequence elements upstream of the mRNA initiation site were required for accurate and efficient initiation of transcription (40).
Thus, signals relevant for transcription initiation are prohahly contained within a 475-hp fragment upstream o f the putative cap site, and introns are not required to generate n functional mRNA. It remains to he estahlished whether other elements further upstream or contained within introns are required for high level expression.
A computer search for related sequences did not reveal any significant homologies to the 5' flanking regions nf other characterized genes. Future research will have to hear on the question whether there are such homologies in genes rekvlated in a similar fashion.