This invention relates to the construction of a recombinant DNA molecule and the expression of the recombinant DNA molecule in host cells. The invention also relates to a process for preparing human substitution- mutant proteins built up from a segment of the glycoprotein tissue type plasminogen activator (t-PA) and from a segment of the glycoprotein urokinase (u-PA). The invention further relates to pharmaceutical compositions having an anti-thrombotic activity. The theoretical concept underlying the present invention, namely,"on the basis of data about the chromo¬ somal structure of the t-PA gene (exon-intron distribution) and the supposed secondary structure of the t-PA protein, deleting, substituting and mutating autonomous structural and functional domains of t-PA, has already been laid down in a prior patent application (H. Pannekoek et al., EP-A-86200223.5 - t-PA deletions).

The human glycoprotein tissue type plasminogen activator (t-PA) is a serine protease, which is synthesized in vascular endothelial cells and also in various cell lines, such as the Bowes melanoma cell line ( if in et al., 1974; Rijken et al., 1980). t-PA performs an essential role in fibrinolysis, the process which is responsible for solubilizing and hence removing a blood clot. The enzyme t-PA catalyses the conversion of the zymogen plasminogen into active serine protease plasmin- Plasmin is the enzyme considered to be primarily respons¬ ible for the lysing of the most important protein component in blood clots, namely fibrine. In the absence of fibrin, t-PA only has a very low plasminogen activator activity.

while in the presence of fibrin the activity of t- PA is accelerated by a factor of 100 to more than 1000. Owing to this property of t-PA, plasmin is generated virtually exclusively on the fibrin surface of a clot, and a systemic formation of plasmin in the circulation is avoided. Such observations have drawn attention to t-PA as a useful anti-thrombolytic. However, the isolation on a large scale of t-PA from its natural environment (blood plasma) would meet with substantial practical and logistic problems owing to its low concentration (about 1 ng per ml). Recombinant DNA techniques can offer a solution here to produce relatively large quantitie It has indeed been shown that t-PA synthesized by means of these techniques represents a useful preparation for the treatment of patients suffering from myocardial infarction (Verstraete et al. , 1985; Williams et al., 1986) . However, the intravenous administration of an effective dose of t-PA in these studies requires relatively large quantities and, in a number of cases, may lead to the decomposition of components other than fibrin and to undesirable hemorrhagic tendencies. The need of administering large amounts of t-PA in an anti-thrombotic- therapy is specifically related to the rapid clearance of t-PA in vivo (its half-life is a few minutes). This phenomenon is attributed to interaction with an hepatic component through which t-PA is removed from circulation (Emeis et al. , 1985) and to inactivation through complexing with serine protease inhibitors, specifically with the endothelial plasminogen activator inhibitor PAI-1 (Sprenger and Kluft, 1987). It is accordingly desirable that molecula variants of t-PA be developed, which exhibit a more effective anti-thrombolytic activity than does t-PA, for example, by virtue of such variants being less effi¬ ciently inhibited by PAI-1. Before describing the construc- tion of such molecular variants and their properties, we will briefly summarize the structural and biological

properties of t-PA. t-PA is synthesized as a molecule consisting of one chain, and can be converted by plasmin into a molecule consisting of two chains by cutting a single arginine-isoleucine peptide bond (Wallen et al., 1980). Two-chain t-PA consists of an amino-terminal "heavy" chain (H; 38 kD) and a carboxy-terminal "light" chain (L; 34 kD) , which are kept together by a single disulfide bond. Starting from a purified t-PA preparation obtained from conditioned medium of cultured Bowes melanoma cells, the complete amino acid sequence and also the position of the asparagine-coupled glycosylation have been determin (Pohl et al., 1984) .

The construction and isolation of a recombinant DNA molecule carrying the complete genetic information for t-PA (so-called full-length t-PA copy DNA or cDNA) was first reported in 1983 (Pennica et al., 1983). The production of biologically active recombinant t-PA by transfected tissue culture cells was also reported (Goedd et al., 1983). In the former studies, it was found that t-PA is synthesized as a so-called precursor protein, consisting of 562 amino acids. The determination of the amino terminal of native t-PA from Bowes melanoma cells showed that t-PA occurs in two natural variants (Pohl et al., 1984). The S variant consists of 527 amino acids, while the L variant is composed of 530 residues. Concluding, an amino-terminal segment of either 32 or 35 amino acids is split off from the precursor protein to generate two types of t-PA. There are no indications that other properties are to be attributed to the S and the L variants.

On the ground of homology between the amino acid sequence of t-PA with that of other plasma proteins, a model has been proposed for the secondary structure of t-PA (Pennica et al., 1983; Banyai et al., 1983). In this model, several structural segments exhibiting

homology with other plasma proteins are assembled to form a composite molecule, possibly built up from autonomou structural domains. The clarification of the structure of the chromosomal t-PA gene (Ny et al., 1984) has supporte this hypothetical model. This study showed that the proposed structural domains are precisely encoded by either a single exon or by a set of adjacent exons. The H chain of the t-PA protein is then composed as follows (from the amino terminal) : a. The signal peptide, followed by the pro-sequence, segments similar to the pre-pro portion of, e.g., serum albumin (Patterson and Geller, 1977; Lawn et al., 1981). b. the so-called "finger" (F) domain, which is homolo¬ gous to the described type I fingers of fibronectin (Sekiguchi et al., 1981; Petersen et al., 1983). c. The "epidermal growth factor-like" (EGF or E) domain, which exhibits structural similarities to both human and mouse EGF (Savage et al., 1972; Gregory and Preston, 1977) . d. Two so-called ^' "kringle" domains (Kl and K2), which, among other things, are characterized by correspond¬ ing positions of three "internal" disulfide bonds. The kringle structure had already been described before for .plasminogen, which carries five of such structures (Sottrup-Jensen et al., 1978).

The carboxy-terminal L chain of t-PA exhibits homology with the "family of trypsin-like" serine proteases It has recently been shown in our laboratory that this part of the t-PA molecule contains the catalytic centre for the plasminogen activator activity and also is the principal interaction site for the physiological inhibitor of t-PA, the endothelial plasminogen activator inhibitor PAI-1 (MacDonald et al., 1985; Van Zonneveld et al., ^" 1986a) . The above model for the secondary structure of t-PA suggests that the proposed autonomous structures

could also represent autonomousfunctions. Studies carried out in our laboratory have given substantial support to this hypothesis (EP-A-86200223.5 H. Pannekoek et al., t-PA deletions; Van Zonneveld et al., 1986b). In this study, t-PA cDNA deletion mutants have been construct and the mutant proteins encoded by such t-PA cDNA*s have subsequently been expressed by means of transfection of so-called mouse Ltk cells with the corresponding t-PA cDNA, from which a defined part of the cDNA has been deleted. The results of this study showed that the possibility of highly accelerating the plasminogen activator activity of t-PA by the presence of fibrin is mediated by two discrete domains of the t-PA molecule, i.e., the ^" finger domain (F) and the kringle domain K2. we were also able to correlate the accelerating effect of fibrin on the t-PA activity with the fibrin-binding property of t-PA, which is also mediated by the above . two discrete domains. We were able to conclude that at least the H chain of the t-PA protein is built up from structural domains having anautonomous function. This conclusion opens up the possibility of combining specific properties of this molecule with other molecules thereby to produce novel substitution mutants in which a new combination of properties is stored and expressed in a functional manner.

This invention relates to the construction of cDNA's coding for substitution-mutants of t-PA, in which a constant portion of the t-PA cDNA constructed in our laboratory is fused with a different portion of a (partial) cDNA coding for the so-called B chain of the human urokinas (u-PA) (Verde et al., 1984). The object is to construct a recombinant DNA molecule coding for a fusion protein in which the properties of both t-PA and u-PA are repre¬ sented. Like t-PA, urokinase (u-PA) is a plasminogen activator and is isolated from urine or from conditioned medium of various cultured cell lines. Two different

for s of urinary u-PA have been identified, namely, a high-molecular (HMW u-PA; 54 kD) and a low-molecular (LMW u-PA; 33 kD) form, which both have plasminogen activator activity. LMW u-PA is regarded as a proteolytic degradation product of HMW u-PA. The amino acid sequence of both HMW u-PA and LMW u-PA, and also the position of asparaginine-coupled glycosylation sites have been completely explained (Gϋnzler et al., 1982a; Gϋnzler et al., 1982b; Steffens et al., 1982). HMW u-PA consists of two chains linked by a single disulfide bond. The amino-terminal chain (A chain) contains 157 residues (1 - 157) and the carboxy-terminal chain (B chain) contains 253 amino acids (159 - 411). LMW u-PA also consists of two chains linked by a disulfide bond in the same way and at the same position as HMW u-PA. The amino- terminal chain (Al) consists of 22 residues (136n - 157) and the carboxy-terminal B chain is identical to the B chain of HMW u-PA (-159 - 411) .

The construction and isolation of a recombinant DNA molecule carrying the complete genetic information- for human u-PA (so-called u-PA cDNA) and the expression of biologically active u-PA synthesized in Escherichia coli bacteria, transformed with a plasmid containing u-PA cDNA has been described (Heyneker et al., 1983). The nucleotide sequence of full-length u-PA cDNA was determined in these studies, so that the amino acid sequence could be derived from it. u-PA is synthesized as a precursor molecule consisting of 431 amino acids. The amino-terminal segment of 20 amino acids, which probably has a signal peptide function, is removed. The amino acid sequence of the remaining 411 residues correspon to the sequence determined in a u-PA preparation, purified from urine (Gϋnzler et al., 1982a and 1982b; Steffens et al. , 1982) . Independently of the above study, Blasi and colla¬ borators (Verde et al., 1984) have constructed a u-PA

cDNA carrying at least the genetic information for the B chain of human u-PA. This recombinant DNA molecule has been used in the present application for the construc of "t-PA substitution mutant cDNA's", while another portion has been derived from the full-length t-PA cDNA constructed in our laboratory (Van Zonneveld et al., 1986b). The "u-PA cDNA" construct (Verde et al., 1984) is not fully built up from cDNA, as there are also two ( "unspliced" ) introns present in this DNA. It is probable that the starting product in the construction of this

(partial) u-PA cDNA was a messenger RNA (mRNA) preparation containing mRNA still including introns. The presence of these two introns in the u-PA cDNA makes it necessary in expression studies to use eukaryotics or tissue culture cells of eukaryotics as host cells, as prokaryotics such as Escherichia coli will not be able to eliminate these introns from the corresponding mRNA. - ^• Like the t-PA protein, the u-PA protein can be regarded as a molecule built up from a number of discrete domains each performing an autonomous function. From the observation that both HMS u-PA and LMW u-PA has plasminogen activator activity, it can be concluded that the B chain carries the catalytic centre of u- PA. The conclusion is consistent with the observation that the B chain exhibits homology with the family of trypsine-like serine proteases, a conclusion which was also drawn with regard to the L-chain of t-PA. In addition it has been determined that the aminoterminal fragment (called ATF) extending from amino acid 1 to 135, has an autonomous function in the bonding of HMW u-PA to a receptor on the outer membrane of a monocytary cell line (Stoppelli et al., 1985).

Urokinase (u-PA) has been used on a large scale as an anti-thrombolytic. However, u-PA has a property which seriously hinders such uses. Unlike t-PA, u-PA has a substantial plasminogen activator activity in

the absence of fibrin- The result of this property is that when administered _iιι vivo both circulating plas¬ minogen and fibrin-bonded plasminogen are converted into plasmin without any distinction. The systemic formation of plasmin leads to the proteolytic degradation of fibrinogen and of the blood clotting factors Factor V and Factor VIII, possibly resulting in heavy bleeding complications.

The inventors have now succeeded in constructing two different human molecules (substitution-mutant cDNA's) and expressing these in host cells. These recombinant DNA molecules contain in addition to a vector portion a DNA sequence coding for the most important part of the H chain of t-PA and a DNA sequence coding at least for the complete B chain of u-PA. The object of this is to add the property that the plasminogen activator activity of t-PA i_> greatly accelerated by the interaction of the kringle 2 (K2) domain and the finger (F) domain of t-PA with fibrin , to the plasminogen activator activity of u-PA, which is located on the B chain of u-PA. In this connection it should once again be emphasized that, in the absence of fibrin t the plasminogen activator activity of u-PA ("basal activity") is considerably higher than the basal activity of t-FA. The invention is further embodied in a process for preparing human t-PA(u-PA) substitution-mutant proteins, which comprises culturing, in a known per se manner, host cells that have been transfected using such a recom¬ binant DNA, and recovering the human t-PA(u-PA) substitution mutants produced by the host cells.

In-vitro experiments have shown that, in the presence of fibrin , these human t-PA(u-PA) substitution- mutant proteins are more active plasminogen activators than t-PA, and in addition are less sensitive to the physiological inhibitor of t-PA, i.e., the endothelial plasminogen activator inhibitor (PAI-1) .

The invention is also embodied in a pharmaceutical composition having an effect on blood clotting and/or on fibrinolysis, comprising human t-PA(u-PA) substitution mutant proteins produced using such a process, and in host cells which, as a result of transfection by means of a recombinant DNA molecule as defined above, are capable of producing human t-PA(u-PA) substitution- mutant proteins.

We will now describe in detail how the human t-PA(u-PA) substitution-mutant cDNA's have been constructe and expressed. These two mutants have been constructed in a manner taking into account: a) The number of amino acids (arbitrarily 14 residues) which separate kringle 2 (K2) of t-PA from the arginine isoleucine peptide bond which is broken by plasmin, resulting in two-chain t-PA (Pennica et al., 1983), or b) The number of amino acids (arbitrarily 27 residues) separating the single kringle of u-PA from the lysine- isoleucine peptide bond which is broken by plasmin, ^' resulting in two-chain u-PA (Heyneker et al., 1983).

Construction of t-PA substitution-mutant cDNA's

Starting materials for the constructs were two recombinant DNA plasmids. Plasmid pSV2/t-PA carries the complete genetic information (cDNA) for human t-PA, and the nucleotide sequence of this plasmid has been fully determined (Mulligan and Berg, 1981; Van Zonneveld et al., 1986b). In addition, we used the plasmid pHUK-1, which contains a partial cDNA coding for at least the B chain of human u-PA (Verde et al., 1984). The nucleotide sequence of plasmid pHUK-1 has also been fully determined (Okayama and Berg, 1983; Verde et al., 1984). The numberin of restriction sites in t-PA cDNA has been taken from

Pennica et al. (1983), while the numbering of restriction sites in u-PA DNA and cDNA has been indicated by Verde et al (1984) .

The DNA of plasmid pSV2/t-PA was fully digested with the restriction endonuclease PstI and thereafter partially digested with the restriction endonuclease EcoRI, with the object not to "cut" the restriction site for EcoRI at position 1273. As a result of this digestion, we were able to isolate a restriction fragment corresponding to the t-PA cDNA segment from EcoRI (position 801) to PstI (position 1312) (Fragment A).

The DNA of plasmid pHUK-1 was fully digested with the restriction endonucleases PstI and EcoRI. As a result of this digestion we were able to isolate a restriction fragment corresponding to the u-PA (c)DNA segment from PstI (position 386) to EcoRI (position 727) (Fragment B) .

The DNA of the vector, being the replicative double-stranded form of the bacteriophage M13mp8 (Vieira and Messing, 1982) was fully digested with the restriction endonuclease EcoRI. The now linear vector was mixed with the Fragments A and B and, by means of the enzyme DNA ligase, a recombinant DNA construct was prepared consisting of the above three components, in which the PstI end of t-PA (position 1312) is fused with the PstI end of u-PA (position 386) and the two EcoRI ends (position 801 of t-PA and position 727 of u-PA) are fused with the two EcoRI ends of M13mp8. This construct was called Starting from the single-stranded form of the

DNA of the bacteriophage M13mp8/t-PA: :u-PA, and using the so-called "outlooping" method (Kramer et al., 1984), two deletion-mutants were constructed in accordance with the considerations set forth hereinbefore. For this purpose, two so-called "primers" were synthesized, which each have a length of 36 nucleotides. The nucleotide

sequence is set forth below:

Primer I 5' GATGTGCCCTCCTGCTCCCAGTGTGGCCAAAAGACT 3' (amino acids) D V P S C S Q C G Q K T

Primer II 5' GATGTGCCCTCCTGCTCCGGAAAAAAGCCCTCCTCT 3 ^» (amino acids) D V P S C S . G K K P S S

N.B. The corresponding amino acids are indicated by the one-letter code.

The 5' terminalhalves of the two primers (18 nucleo- tides) are identical to the base pairs 958 to 975 of t-PA cDNA coding for the amino acids asparagic acid (D), valine (V), proline (P) , serine (S), cysteine (C) and serine (S). The 3' terminal half of primer I correspon to the base pairs 677 to 698 of u-PA DNA coding for the amino acids glutamine (Q), cysteine (C) , glycine (G)V glutamine (Q) , lysine (K) and threonine (T) . The 3' terminal half of primer II corresponds to the base pairs 638 to 655 of u-PA and code for the amino acids glycine (G) , lysine (K) , lysine (K) , proline (P), serine (S) and serine (S).

Using the "outloop" method, in which single- stranded DNA of the bacteriophage M13mp8/t-PA: :u-PA was hybridized with either primer I or primer II and subsequently a double-stranded hetero-duplex was synthesiz in vitro, followed by transformation of suitable Escherich coli bacteria, the required deletions were made. Owing to this procedure, a segment t-PA cDNA was deleted which, among others, contains the EcoRI restriction site at position 1273 and the PstI restriction site at 1312. The segment u-PA deleted contains, among others, the restriction site for PstI at position 386. For both constructs it is now possible, after purification of the double-stranded form of the mutant Ml3mp8/t-PA: :u-PA, to isolate an EcoRI restriction fragment carrying the desired deletion. When primer I is used, the length

of this fragment is 222 base pairs (Fusion fragment A) and when primer II is used, this fragment is 261 base pairs long (Fusion fragment B) .

In the following steps of the constructions, the Fusion Fragments at the 5' side are f sed with the adjacent t-PA cDNA from the full-length t-PA cDNA, and at the 3* side the Fusion Fragments are fused with the adjacent u-PA (c)DNA from the plasmid pHUK-1. For this purpose the DNA of the plasmid pSV2/t-PA (Van Zonneveld et al., 1986b) was fully digested with the restriction enzymes Hindlll and EcoRI, as a result ^" of which we were able to isolate a restriction fragment extending from the Hindi!I restriction site on the vector portion of pSV2/t-PA and the EcoRI site at position 801 in the t-PA cDNA portion. Also, pHUK-1 DNA (Verde et al., 1984) was fully digested with the restriction endonuclease PstI and partially with the restriction endonuclease EcoRI, avoiding "cleavage" of the EcoRI restriction site at position 1364. In this way we were able to isolate an EcoRI PstI fragment corresponding to the positions

727 and 1969 for- EcoRI and PstI, respectively, on pHUK-1. The vector used for the following construction is the plasmid pUCl9, which was digested with the restriction endonucleases HindiII and PstI (Yanisch-Peron et al., 1985).

Fusion Fragment A (222 base pairs; EcoRI ends) was mixed with the Hindlll-EcoRI fragment from pSV2/t-PA, the EcoRI-PstI fragment from pHUK-1 and the digested vector pUC19. In the same way, the Fusion Fragment B (261 base pairs) was, separately, mixed with the same three fragments. After ligation and transformation of Escherichia coli strain DH1 (Maniatis et al., 1982), we were able to isolate the desired two different recom¬ binant DNA plasmids. These plasmids were called: pUC19/t- PA::u-PA-I and pUC19/t-PA: :u-PA-II.

The DNA of the plasmids pUC19/t-PA: :u-PA-I and

pUCl9/t-PA: :u-PA-II was fully digested with the restrictio endonuclease Hindlll and partially with the restriction endonuclease BamHI, avoiding "cleavage" of the BamHI site in the u-PA DNA portion at position 1654. In this way we were able to isolate two HindiII-BamHI fragments related to the original "outlooping" with either primer I or primer II, of which the Hindlll restriction site originates from the vector portion of pSV2/t-PA and the BamHI restriction site from the polylinker portion of the vector pUC19. In essence, the Hindlll-BamHI fragmen contain the fusions between t-PA cDNA, coding for the most important portion of the H chain of t-PA, and the B chain of u-PA (with the difference between the two fragments being expressed in that when primer II is used there are 13 amino acids more than when primer I is used). Finally, the two Hindlll-BamHI fragments are separately used to replace the full-length t-PA cDNA portion on the plasmid pSV2/t-PA. For this purpose the DNA of plasmid pSV2/t-PA was fully digested with the restriction endonucleases Hindlll and Bglll and the fragment containing the vector portion was isolated. As the restriction endonucleases BamHI and Bglll generate the same DNA ends, we were subsequently able to ligate the above Hindlll-BamHI fragments with the digested vector. After the transformation of Escherichia coli

DH1 bacteria, the desired two types of recombinant DNA plasmids were isolated and purified (Maniatis et al., 1982). The nucleotide sequence of the relevant sequences of these final plasmids, called pSV2/tPA: :u-PA-I and pSV2/t-PA: :u-PA-I.I, was fully determined by DNA sequencing (Sanger et al., 1977) and the relevant part of the fusion between t-PA cDNA and u-PA (c)DNA, and also the correspond¬ ing amino acid sequence, are shown in the accompanying Figure 1. The mutant proteins to be discussed in the following paragraphs were called t-PA::u-PA-I and t-

PA::u-PA-II, and are encoded by, respectively, pSV2/t-

PA::u-PA-I DNA and pSV2/tPA: :u-PA-II DNA. It is finally noted that, as a result of the constructions described, the plasmids last mentioned contain only one intron in the u-PA portion _r originating from the plasmid pHUK- 1. An E. coli K12 strain DHl pSV2/t-PA: :u-PA-II was deposited at the Centraal Bureau voor Schimmelcultures (CBS) at Baarn, The Netherlands, on April 28 1987, and was accorded number CBS 293.87. From the plasmid thus rendered accessible, the shorter plasmid pSV2/tPA::u- PA-I can be constructed with facility, for example, via the outlooping technique.

Expression of the mutant proteins t-PA::u-PA-I and t- PA: :u-PA-II

The mouse fibroblast cell line (mouse Ltk-) was cultured in so-called "Iscove's minimal medium", to which were added penicillin, streptomycin and 10% (v/v) foetal calf serum. Transfection of these cells was carried out as described (Lopata et al., 1984; Van Zonneveld et al., 1986b). After transfection the cells were incubated in the above medium, but without foetal calf serum. Five days after transfection, conditioned media were' harvested, whereafter Tween-80 and sodium azide were added to a final concentration of, respectively, 0.01%

(v/v) and 0.02% (w/v). Thereafter, the conditioned media were dialysed against a buffer consisting of 50 mM sodium phosphate (pH 7.4), 0.01% (v/v) Tween-80 and 0.02% (w/v) sodium azide at 4°C for 16 hours. Preparations treated in the above manner were then stored at 4°C.

Determination of the concentration of mutant proteins in conditioned media

The concentration of the secreted mutant proteins t-PA: :u-PA-I, t-PA: :u-PA-II, and also of control trans-

fection experiments with pSV2/t-PA DNA coding for recom¬ binant t-PA (rt-PA) was determined by means of an immuno- radiometric method (IRMA). The IRMA is based on the presence of kringle 1 (Kl) of t-PA in all proteins being investigated and the availability of two different mono¬ clonal anti-t-PA antibodies directed against Kl (Van Zonneveld et al., 1986c). These two antibody preparations have been called CLB-t-PA 72 and CLB-t-PA 16 (Van Zonnevel et al. , 1986c) . The monoclonal anti-t-PA antibody CLB-t-PA.72 was coupled to cyanogen bromide activated Sepharose and incubated in 10.5 mM sodium phosphate (pH 7.4), 150 mM sodium chloride (PBS), to which had been added 1% (w/v) bovine serum albumin, 0.1% (v/v) Tween-20 and various dilutions of the conditioned media of the transfec tissue culture cells, previously concentrated by means of Amicon centricon 30 filters by a factor of about 50. Subsequently, a 1 5ι_τ_ _a k _e n _e 3 χ--Q preparation (15,000 counts per minute) of monoclonal CLB-t-PA 16 was added. to the above mixture to measure the attachment of antigen to the immobilized monoclonal antibody. The incubations were carried out in a total volume of 0.5 ml at room temperature for 18 hours. During the incubations, the mixtures were rotated end-over-end. The Sepharose beads were washed five times with a buffer containing 0.9

M- sodium chloride, 0.1 % (v/v) Tween-20 and 10 mM EDTA and the bound radioactivity was measured by means of a radioactivity counter for gamma radiation. Dilutions of a standard Bowes melanoma t-PA preparation were used as a reference material for the concentration assay.

A characteristic concentration assay gave the following results: rt-PA 16 ng/ml; t-PA::u-PA-I 0.6 ng/ml and t-PA: :u-PA-II 0.4 ng/ml, being the concentrations in the unconcentrated conditioned media for transfected cells treated under comparable conditions.

The lower expression level of the mutant plasminoge

activator proteins, as compared with rt-PA, is possibly to be attributed to the presence of an intron in the u-PA portion of the t-PA substitution mutant. If splicing of this intron is a limitative step, this could result in a lower expression level for the mutant proteins as compared with the expression of rt-PA.

Analysis of mutant proteins by means of gelatin-plasminogen qel electroforesis

An analysis of the mutant plasminogen activator proteins by means of gelatin-plasminogen gel electrofor-^sis can lead to conclusions about two parameters of these proteins, i.e., a semi-quantitative insight into the plasminogen activator activity in the absence of fibrin and a relative molecular weight. The preparation of such gels (polyacrylamide, with copolymerized gelatin and plasminogen) , and also the principle on which this analysis system is based have been described before (Van Zonneveld- et al., 1986b). The results of this analysis are shown in the accompanying Figure 2.

The conclusion can be that both t-PA substitution mutants have plasminogen activator activity. This obser¬ vation is consistent with the determination of the nucleo- tide sequence of the mutant DNA's, which failed to show stop codons in the translation "reading frame" and with the length of the relevant coding portion of the mutant DNA's. The relative molecular weight of t-PA: :u-P - I and t-PA: :u-PA-II corresponds to the length of the relevant coding portion of pSV2/t-PA: :u-PA-I and pSV2/t- PA::u-PA-II, which corresponds to 527 and 540 amino acids, respectively.

Kinetics of plasminogen activation by the mutant proteins t-PA;:u-PA-I and t-PA: :u-PA-II

The activation of Glu-plasminogen to plasmin by t-PA::u-PA-I and t-PA: :u-PA-II (and by the controls rt-PA and HMW u-PA) was carried out at 37°C in a buffer consisting of 0.1 M TRIS-HC1 (pH 7.4), 0.1% (v/v) Tween- 80, 0.57 mM of the chromogenic substrate S-2251 and, if indicated, 120 /ug)ml fibrinogen fragments generated by digestion with cyanogen bromide (called "stimulator") (Verheijen et al., 1982). The total reaction volume was 250 /ul and the reactions were carried out in 96- well microtiter plates. A Titertek Multiscan spectrophoto- meter equipped with a thermostat was used to monitor the development of the optical density at 405 nm for 5 to 6 hours with intervals of 10 minutes. Initial reactio velocities were obtained by graphically plotting the relationship between the increase in optical density at 405 nm per minute against the incubation time. Under

< these conditions an increase in optical density at 405 nm per minute corresponds to 1 unit with a quantity of 40 pmoles plasmin in a volume of 250 /ul (concentration plasmin 160 nM) . In the absence of stimulator, for 0.2- 0.4 ng/ml t-PA::uPA-I, t-PA: :u-PA-II and HMW u-PA, the Glu-plasminogen concentration was varied from 0.11 to 0.55 /uM, while for 1.6-3.2 ng/ml rt-PA the Glu-plasminoge concentration was varied from 0.55 to 4.4 /uM. In the presence of stimulator (120 /ug/ml) the Glu-plasminogen concentration was varied from 1.7 to 27.5 nM for 0.08- 0.2 ng/ml of each of the mutant proteins, 0.32-0.64 ng/ml rt-PA and 0.1-0.2 ng/ml HMW u-PA. For each Glu- plasminogen concentration, both in the absence and in the presence of stimulator, the experiments were duplexed and repeated four times. By means of the results obtained, so-called "Lineweaver-Burk plots" were plotted, using the "weighed" linear regression analysis, whereafter we were able to determine the so-called Km and v(max) values. The so-called k(cat) value could be calculated by using the molecular weights of the various plasminogen

activators, namely, t-PA::u-PA-I 72 kD, t-PA: :u-PA- II 74 kD, rt-PA 70 kD and HMW uPA 54 kD. The results are listed in the following Table 1.

This kinetic analysis permits a number of conclusions relating to the affinity of the various plasminogen activators for the substrate Glu-plasminogen (the Km value being the standard) and the rate of conversion (the k(cat) value being the standard) in the absence and presence of the fibrin mimetic "stimulator".

In the absence of stimulator, the affinity of t-PA:u-PA-I and t-PA: :u-PA-II for Glu-plasminogenis in thesame order as HMW u This result follows from the fact that the mutant proteins and HMW u-PA carry the same catalytic centre, situated on the B chain of u-PA. The presence of different amino- terminal chains on the mutant proteins, and also as compared with HMW u-PA, has no or virtually no effect on the affinity for the substrate. The affinity of the mutant proteins for the substrate is at least a factor of 57 to 82 higher than is the affinity of rt-PA for the substrate. The rates of conversion for the mutant plasminogen activators, rt-PA and HMW u-PA, is in the same order of magnitude (0.25-0.65 sec.~l).

In the presence of fibrin, at least stimulator imitating the effect of fibrin, the affinity for the substrate Glu-plasminogen markedly increases for all plasminogen activators. For the mutant proteins t-PA::u-PA-I and t-PA: :u-PA-II, the affinity increases by a factor of 122 to 220 (Km ^* from 0.61-0.88 ιM to 0.005-0.004 ιM) , for rt-PA by a factor of at least 3125 (Km from > 50 ιM to 0.016 ιM) and for HMW u-PA by a factor of 30 (Km from 2.2 ιM) to 0 -073 ιM) ^' . From these values, the conclusion can be drawn that, in the presence of the stimulator, the affinity for the substrate Glu-plasminogen is increased

^" by a factor of 4.1 to 7.3 owing to the presence of the amino-terminal t-PA H chain, instead of the amino-terminal u-PA A chain. In the presence of stimulator, the rates of conversion of the various plasminogen activators are not essentially different: k(cat) 0.26 - 0.45 sec-l . The comparison of the enzymatic activity, represented by the quotient k(cat)/Km in the presence and in the absence of stimulator shows that this increases for rt-PA by a factor of at least 2100 (from < 0.01 sec ^-1 iM ^-1 to 21 sec ^-1 iM ^-1 ), for t-PA: :u-PA-I by a factor of 220

(from 0.41 sec ^-1 iM ^-1 to 90 sec ^-1 iM" ¹ ), for t-PA: :u-PA-I by a factor 288 (from 0.34 sec ^~ l ιM~l to 98 sec ^-1 /UM ^-1 ) , and for HMW u-PA by a factor of 12 (from 0.29 sec ^"1 _/ uM ^-1 to 3.6 sec~l -M ^-1 ) . The more efficient plasminogen activation by HMW u-PA in the presence of stimulator should be attributed to ,a conformation change of plasminog by the stimulator (Lijnen et al., 1984). The same phenomen will also occur in the reactions with the mutant plasminoge activators and rt-PA. From the comparison of the enzymatic activity C k(cat)/Km] value, therefore, it can be concluded that, in the presence of fibrin, the mutant proteins are 4 to 5 times more effective in plasminogen activation than rt-PA and about 25 times more effective than HMW u-PA. On the ground of the findings that the kinetic parameters of t-PA: :u-PA-I and t-PA: :u-PA-II are not essentially different in the presence and in the absence of "stimulator", the following experiments will be carried out with the mutant protein t-PA: :u-PA-II. As explained before, t-PA: :u-PA-II contains 13 amino acid residues more than t-PA: :u-PA-I .

Complex formation of the mutant plasminogen activators with the endothelial plasminogen activator inhibitor (PAI-1) and the inhibition of the mutant plasminogen activators by PAI-1

The endothelial plasminogen activator inhibitor (PAI-1) is regarded as the physiological inhibitor of the fibrinolytical process and therefore as an important regulator in vivo of the "nett fibrinolytic activity" (survey article: Sprengers and Kluft, 1987) . PAI-1 is synthesized, among other method, by cultured vascular endothelial cells. PAI-1 complexes very fast both with u-PA and with t-PA (second-order reaction constant in excess of 10 ⁷ M ^~ l sec ^-1 (Sprengers and Kluft, 1987). The complex of the plasminogen activators with PAI-1 is resistent to sodium dodecyl sulphate (SDS), so that the formation of a complex can be analyzed by means of SDS-polyacrylamide gel electrophoresis techniques. However, the activity of the plasminogen activators in complex with PAI-1 can only be demonstrated after treat ent with Triton X-100 (Loskutoff et al., 1983), possibly as a result of intrinsic plasminogen activator activity of the complex or as a result of dissociation of plasminogen activator from the complex. PAI-1, which is synthesized by and secreted by cultured endothelial cells, is found as a mixture of inactive, latent and active inhibitor. The latent form can be activated by treatment with certain chemicals, such as SDS.

Recently, a number of groups constructed PAI-1 cDNA, which harbours the complete genetic information for the human PAI-1 glycoprotein (Ny et al., 1986; Pannekoek et al., 1986; Ginsburg et al., 1986; Andreasen et al., 1986). The amino acid sequence of PAI-1 could be derived af er the elucidation of the complete nucleotide sequence. The amino acid sequence of PAI-1 shows significant homology with proteins belonging to the "family" of serine protease inhibitors (so-called serpins). The operating mechanism of the serpins, specifically that of the prototype alpha- 1-antitrypsin with the serine protease elastase, has been extensively studied (Loebermann et al., 1984; Carrell and Boswell, 1986). It is probable that the operating

mechanism of PAI-1 with t-PA and u-PA greatly resembles that of alpha-1-antitrypsin with elastase. On the ground of these considerations, it can be assumed that serpins, such as PAI-1, form a 1:1 molar complex with the "target" protease-

The complex formation of PAI-1 with the mutant plasminogen activator t-PA: :u-PA-II, and also with the controls rt-PA and HMW u-PA was determined as follows. 0.2 ng of the plasminogen activators was separately incubated with an excess of PAI-1. Thereafter, a buffer was added, consisting of 2% (w/v) sucrose, 5/ιg/ml bromo- phenol blue and 2.5% (w/v) SDS, and the samples were analyzed by means of SDS-polyacryl amide gel electrophoresi After electrophoresis, a fibrin-agarose indicator gel was prepared which also contained plasminogen (Granelli- Piperno and Reich, 1978) and used as an overlay for the SDS-polyacryl amide gel treated with Triton X-100. Through diffusion of plasminogen activator, either "free" or complexed with PAI-1, it is possible to determine both the activity of plasminogen activators semi-quantitati ly, and also to determine the molecular weight of the plasminogen activator. The results are shown in Fig. 3.

From the results of this analysis, two conclusions can be drawn. First, PAI-1 forms a complex with both t-PA: :u-PA-II and the control plasminogen activators t-PA and HMW u-PA. This conclusion can be drawn from the observation that the molecular weight of the plasminoge activator activity increases after pre-incubation with PAI-1. Second, it can be found that the molecular weight of the complex which PAI-1 forms with t-PA: :u-PA-II , rt-PA and HMW u-PA, for all plasminogen activators corres¬ ponds with a 1:1 molar complex. This conclusion is based on the observation that, after pre-incubation with PAI-1, the plasminogen activator activity migrates with a molecula weight corresponding with the sum of the molecular weights

of the plasminogen activators (about 70-74 kD) and of PAI-1 (about 52 kD) .

The effectiveness of inhibition of the mutant plasminogen activator, rt-PA and HMW u-PA by PAI-1 was then determined. For this purpose, the latent fraction of the PAI-1 preparation (total about 130 ng) was previousl activated with SDS (final concentration 0.1% (w/v)). Various dilutions of the activated starting preparation were treated with an equal volume (55 ιl) 0.2 M TRIS-HC1 (pH 7.4), 0.2% (v/v) Tween-80, 6% (v/v) Triton X-100 at room temperature for 15 minutes. Subsequently, one of the plasminogen activators was added in a volume of 30 ιl, which corresponded to 0.024 ng t-PA: :u-PA-II, 0.024 ng rt-PA or 0.024 ng HMW u-PA. The mixtures were incubated at room temperature for 30 minutes. The remaining plasminogen activator activity of the various activators was finally determined as follows. We added llO il of a mixture containing 0.1 M TRIS-HC1 (pH 7.4), 0.1% ^* (v/v) Tween-80, human Glu-plasminogen (final concentration 0.11 αM), chromogen substrate S-2251 (final concentration 0.54 mM) and "stimulator" (120 ιg/ml). The determination of the optical density at 405 nm, as a measure for the plasminogen activator activity, has already been described in a preceding paragraph. The results are shown in the accompanying Fig. 4.

From the results, the following conclusions can be drawn. When the quantity of PAI-1 needed to inhibit the plasminogen activator activity of the various plasmino activators by 50%, is compared, it can be found that about 5 times more PAI-1 is required for the mutant t-PA: :u-PA-II than for rt-PA or HMW u-PA. Consequently, the mutant plasminogen activator can be regarded as a molecule that is more resistent to the physiological inhibitor PAI-1 than are rt-PA and HMW u-PA. The dissociat constants of the various plasminogen activator-PAI-1 complexes could be determined by means of a Scatchard-type

analysis (Scatchard, 1949) . In accordance with the above conclusion, it can be concluded that the complex of rt-PA and HMW u-PA with PAI-1 dissociates less readily than does the complex of the mutant plasminogen activators with PAI-1. The calculated dissociation constants are listed below:

Plasminogen activator Kd

t-PA: :u-PA-II 4.1 x 10 ^_12 M rt-PA 0.6 x 10 ^"12 M HMW u-PA 0.24 x 10~ ¹² M

Materials and Methods

DNA analysis:

The isolation of plasmid DNA, the analysis of DNA digested with restriction endonucleases by means of agarose and polyacryl amide gel electrophoresis,

"Southern" blotting, the isolation of restriction fragment the radio-labelling of DNA, either by "nick translation" or with T4 polynucleotide kinase, and the technique of colony-hybridisation were carried out as described (Maniatis et al. , 1982). The determination of the sequence of bases in the DNA (DNA sequencing) was carried out by the dideoxy method (Sanger et al., 1977). The restricti endonucleases and other enzymes had been purchased from New England Biolabs (Beverley, Mass., USA) and radioactive labelled nucleotides from The Radiochemical Centre (Amersh England) .

Transfection of tissue culture cells:

The procedure for obtaining "transient" expressio by means of transfected mouse Lkt cells was carried out as described before (Lopata et al., 1984; Van Zonnevel

et al . , 1986b ) .

Gelatin-plasminogen gel electrophoresis:

The preparation of polyacryl amide gels containing gelatin and plasminogen, and the performance of the electrophoretic analysis of samples by means of these gels have been described before (Van Zonneveld et al., 1986b) .

Fibrin overlay technique:

The determination of plasminogen activator activity and the molecular weight associated with that activity was carried out using the fibrin overlay technique after SDS-polyacryl amide gel electrophoresis of samples, as described before (Granelli-Piperno and Reich, 1978).

Determination of plasminogen activator activity: in this determination, use is made of the chromo- genic substrate S-2251, which is specific for plasmin. The activation of Glύ-plasminogen by rt-PA, ^' Bowes melanoma t-PA and other t-PA derivatives is greatly accelerated by the addition of fission products of fibrinogen treated with cyanogen bromidei. This last component is called

"stimulator" (Verheijen et al., 1982). If the plasminogen activator activity is determined after pre-incubation with PAI-1, the PAI-1 preparation used is either conditioned (serum-free) medium of cultured human vascular endothelial cells (concentration PAI-1: 2.4 ιg/ml) or a purified preparation PAI-1 (100 ιg/,l; Lambers et al., 1987). The two preparations gave corresponding results.

Other materials: HMW u-PA (two-chained) was a gift of Dr. G.Cassani

(Lepetit, Milan, Italy). Bowes melanoma t-PA (two-chained) was purchased from Biopool (Umea, Sweden) . Glu-plasminogen, "stimulator" and the chromogenic substrate S-2251 were

obtained from KabiVitrum (Stockholm, Sweden).

TABLE 1 - stimulator + stimulato

PREPARATION Km k(cat (cat) /Km Km k(cat) k(c rt-PA > 50 0.49 < 0.01 0.016 0.34 > t-PA: :u-PA-I 0.61 0.25 0.41 0.005 0.45 t-PA: :u-PA-II 0.88 0.30 0.34 0.004 0.39

HMW U-PA 2.2 0.65 0.29 0.073 0.26

The Km values are given in iM. The k(cat) values are given in sec ^~ l, so that the quotient k(cat)/Km has the dimension ιM ^~ l.sec ^~ l.

Explanation of the figures: Fig. 1 Fig. 1 shows the nucleotide sequence of the cDNA area coding for the fusion portion of, respectively, t-PA::u-PA-I (top) and t-PA: :u-PA-II (bottom). Also shown are the corresponding amino acids of the t-PA portions and the u-PA portions. The numbering above the nucleotide sequences corresponds to that of the t-PA nucleotide sequence (955 etc.) and the u-PA nucleotide sequence (679 etc., 640 etc.). The numbering under the amino acid sequences corresponds to that of the t-PA amino acid sequence (256 etc.) and the u-PA amino acid sequence 147 etc., 134 etc.). The underligned nucleotide sequence indicates the synthetic primers used for the construction of the two mutants.

Fig. 2

Gelatin-plasminogen gel electrophoresis. The performance of this analysis is set forth in the text.

We successively analysed: 1.1 ng Bowes melanoma t-PA (control); 2. 0.1 ng Bowes melanoma t-PA; 3. 1 ng rt-PA; 4. 0.1 ng rt-PA; 5. 0.1 ng t-PA: :u-PA-I; 6. 0.1 ng t-PA: :u- PA-II; 7. 0.1 ng HMW u-PA. This analysis shows that the activity of the t-PA substitution mutants, in the absence of fibrin in this system, is comparable to HMW u-PA and higher than either (native) Bowes melanoma t-PA, or rt-PA. Indicated on the left are reference molecular weights 70,000 and 54,000 of t-PA and HMW —PA, respectively.

Fig. 3

Fibrin-overlay after SDS-polyacryl amide gel electrophoresis. This technique is described in the Materials and Methods section, while the details are given in the text. From left to right, we successively analysed the possible complex formation PAI-1 with one of the plasminogen activators, t-PA: :u-PA-II without PAI-1 (MW about 70,000); t-PA: :u-PA-II with PAI-1 (MW higher than 70,000); rt-PA without PAI-1 (MW about 70,000); rt-PA with PAI-1 (MW higher than 70,000); HMW u-PA without PAI-1 (MW about 54,000); HMW u-PA with PAI-1 (MW higher than 70,000); conditioned medium of cultured human endo¬ thelial cells (ECCM) , containing a complex of PAI-1 and t-PA; ECCM to which conditioned medium of cultured, non-transfected mouse Ltk~ cells had been added.

Fig. 4

The data about the test for the determination of the inhibition of the activity of the various plasminogen activators by (increasing amounts of) PAI-1 has been extensively described in the specification.

Ref erences

Pannekoek et al., European patent application 86200223.5 t-PA deletions.

Rifkin et al., J.Exp.Med. 139 (1974) 1317-1328.

Rijken et al., Thro b.Res. 18 (1980) 815-830. Verstraete et al., The Lancet 1 (1985) 842-847.

Williams et al., Circulation 72 (1986) 338-346.

Emeis et al., Thromb.Haemostasis 54 (1985) 661-664.

Sprengers and Kluft, Blood 69 (1987) 381-387.

Wallen et al., In Progress in Fibrinolysis (J.F. Davidson, I.M. Nilsson and B. Astedt, Eds.) Vol. 5 (1980) pp. 16-21,

Churchill Livingstone, Edinburgh.

Pohl et al., Biochemistry 23 (1984) 3701-3707.

Pennica et al., Nature 301 (1983) 214-220.

Goeddel et al., European Patent Office 0,093,619. 1983 September 11. 60p. Int.Cl. C12N15/00; Appl. 1983 April

5.

Banyai et al., FEBS Lett. 163 (1983) 37-41.

Ny et al., Proc.Natl.Acad.Sci. U.S. 81 (1984 ^" ) 5355-5359.

Patterson and Geller, Biochem.Biphys . Res.Commun. 74 (1977) 1220-1226.

Lawn et al., Nucleic Acids Res. 9 (1981) 6103-6114.

Sekiguchi et al. , J.Biol.Chem. 256 (1981) 6452-6462.

Petersen et.al., Proc.Natl.Acad. Sci. U.S. 80 (1983)

137-141. Savage et al., J.Biol.Chem. 247 (1972) 7612-7621.

Gregory and Preston, Int.J.Pept. Protein resp. 9 (1977)

107-118.

Sottrup-Jensen et al. , In Progress in Chemical Fibrinolysis and Thrombolysis (J.F. Davidson, R.M. Rowan, M.M. Samama and P.C. Desuoyers, Eds.) Vol. 3 (1978) pp. 191-209.

Raven, New York.

MacDonald et al., Gene 42 (1986) 59-67.

Van Zonneveld et al., J.Cell.Bioche . 32 (1986a) 169-178.

Van Zonneveld et al., Proc.Natl.Acad.Sci . U.S. 83 (1986b)

4670-4674 .

Verde et al., Proc.Natl.Acad.Sci. U.S. 81 (1984) 4727-4731).

Gϋnzler et al., Hoppe-Seyler' s Z. Physiol.Chem. 363

(1982a) 133-141. Gϋnzler et al. , Hoppe-Seyler' s Z. Physiol.Chem. 363

(1982b) 1156-1165.

Steffens et al., Hoppe-Seyler' s Z. Physiol.Chem. 363

(1982) 1043-1058.

Heyneker et al., European Patent Office 0.092.182. 1983 September 26. 43p. Int.Cl. C12N 15/00; Appl. 1983 (April

14).

Stoppelli et al., Proc.Natl.Acad.Sci. U.S. 82 (1985)

4939-4943.

Mulligan and Berg, Proc.Natl.Acad.Sci. U.S. 78 (1981) 2072-2076.

Okayama and Berg, Mol.Call.Biol. 3 (1983) 280-289.

Vieira and Messing, Gene 19 (1982) 259-268.

Kramer ete al., Nucleic Acids Res. 12 (1984) 9441-9456.

Yanisch-Peron et al.. Gene 33 (1985) ^' 103-119. Maniatis et al. In Molecular Cloning: a Laboratory Manual,

Cold Spring Harbor Laboratory, Cold Spring Harbor N.Y.

(1982) pp. 1-545.

Sanger et al., Proc.Natl. cad.Sci. U.S. 74 (1977) 5463-5467.

Lopata et al., Nucleic Acids Res. 12 (1984) 5705-5717. an Zonneveld et al., Thromb.Haemostasis 57 (1986c) 82-86.

Verheijen et al., Thromb. Haemostasis 48 (1982) 266-269.

Lijnen et al., Eur.J.Biochem. 144 (1984) 541-544.

Loskutoff et al., Proc.Natl.Acad.Sci. U.S. 80. (1983)

2956-2960. Ny et al., Proc.Natl.Acad.Sci. U.S. 83 (1986) 6776-6780.

Pannekoek et al., EMBO J. 5 (1986) 2539-2544.

Ginsburg et al. , J.Clin.Invest. 78 (1986) 1673-1680.

Andreasen et al., FEBS Lett. 209 (1986) 213-218.

Loebermann et al., J.Molec.Biol. 177 (1984) 531-536. Carrell and Boswell, In Proteinase Inhibitors (A.Barrell and J.S. Salvesen, Eds.), Elseviers Publishing Co, Amsterdam

(1986) pp. 403-420.

Granelli-Piperno and Reich, J.Exp.Med. 148 (1978) 223-234 Scatchard, Ann. N.Y. Acad.Sci. 51 (1949) 660-672. Lambers et al. (presented for publication).