Mutant and Chimeric Recombinant Plasminogen Activators PRODUCTION IN EUKARYOTIC CELLS AND PRELIMINARY CHARACTERIZATION*

Mutant urokinase­type plasminogen activator (u­ PA) gènes and hybrid gènes between tissue­type plas­ minogen activator (t­PA) and u­PA have been designed to direct the synthesis of new plasminogen activators and to investigate the structure­function relationship in thèse molécules. The following classes of constructs were made starting from cDNA encoding human t­PA or

PA)' and u rokinase-type plasminogen activator (u -PA), have been extensively stu died (for a review, see Réf. 2). The first one, tPA, found to be identical to blood plasminogen acti vator (3), has been isolated from human utérus (4). The second enzyme, uPA, has been identified in human urine and kidney cells (5,6).
Both proteins are serine proteases of 70,000 and 54,000 dallons, respectively, synthesized as singlechain polypeptides including a signal séquence involved in sécrétion (79). Single chain plasminogen activators are processed by plasmin to form active enzymes composed of two disulfidelinked poly peptides. tPA is cleaved at the Arg"*Ile"^ bond and single chain uPA (scuPA), primarily, between Lys'^ and Ile'^* residues. Secondary cleavages in the uPA molécule occur at the Arg ^ ^'Phe'" (with thrombin) (10) and Lys"'Lys"'' bonds, the latter event producing the low molecular size form of the enzyme (33,000 daltons) which has similar properties as the 54,000dalton species (11).
Although both enzymes activate plasminogen, tPA and u PA présent différent fibrinolytic properties. Indeed, plasmin ogen activation by tPA is highly fibrinspecific because the activator binds to the fibrin clot. Plasminogen then binds to both tPA and fibrin, thus forming a cyclic ternary complex with increased stability (12). Both singlechain andtwochain tPA have very similar fibrinolytic efficacy; this implies that the conversion of singlechain to twochain tPA at the sur face of the fibrin clot (13) has no physiological significance. On the contrary, twochain uPA displays little affinity for fibrin and activâtes free and fibrinbound plasminogen equally well. Singlechain urokinase (scuPA), which has been isolated recently by several groups (1417), is a plasminogen activator with better fibrin specificity than uPA (1820). scu PA thus displays intrinsic plasminogen activator properties (21,22).
A comparison of the amino acid and nucleotide séquences of tPA and uPA reveals extensive homology between their Bchains (COOHterminal régions) which carry the active site. The Achains (NHzterminal régions), however, differ in some significant aspects (79, 23,24). tPA contains two kringle domains, whereas uPA has only one. Thèse kringle

New Recombinant
Plasminogen Activators domains are highly homologous to équivalent structures of plasminogen involved in fibrin-binding (25,26). In addition, the NH2-terminal région of t-PA contains a finger-like domain similar in structure to the fibrin-binding régions of fibronectin (27,28). The  high  affinity  of  t-PA  for  fibrin  has  been  attributed  to  the  présence  of  both  the  finger  and  kringle  domains in the enzyme (29,30

RESULTS
Construction of Mutant u-PA and of Chimeric t-PA/u-PA Coding Séquences-In an effort to improve thrombolytic selectivity and fibrin specificity of plasminogen activators, a family of vectors carrying séquences encoding new plasminogen activators have been created through recombinant DNA technology. As outlined in the Miniprint Section, the starting material for the new constructs is carried by three plasmids: pULBlOOO and pULB1135 carry a preprourokinase cDNA (9) and pDSPl.lTPA25.BGH carries a t-PA precursor cDNA. The recombinant molécules derived from the manipulation of thèse DNAs share common features: they were ail obtained as //indlII-SacI cassettes carrying a 5'-terminal séquence coding for a signal peptide and a 3'-terminal séquence corresponding to the whole or partial B-chain of u-PA. In ail cases, the u-PA catalytic site has been maintained. The new constructs, however, differed either in the nature of the A-chain or in the séquence coding for the activation site of the proenzymes.
Full-length recombinant DNA molécules were obtained by subcloning various DNA fragments into the Hindlll and Sacl sites of plasmid pULB1221 (42). When necessary, séquences joining DNA fragments of différent origins were synthesized chemically and added to the ligation mixtures. The conformity of the recombinant DNAs to the expected séquences was then checked by DNA sequencing before proceeding to the insertion of iftndIII-SacI coding séquences into the eukaryotic transient expression vector pDSPl.lBGH (35), between the SV40 early promoter and the BGH polyadenylation signal (see Miniprint Section). Upon transfection in Chinese hamster R1610 or/and Cos I monkey cells, recombinant plasminogen activators were produced and secreted in the culture médium.
Characteristics of recombinant plasminogen activators are shown in Table 2 and the new enzymes schematically represented in Fig. 2. The products can be classified in three main groups. The first one consists of modified preprourokinase molécules ( Table 2). Two of thèse enzymes, ppUK.410 and ppUK.410/366, coded for by pULB9122 and pULB9134, carry amino acid substitutions in the B-chain as compared to pULBlOOO and pULB1135. Thèse modifications were introduced to assess their effect on enzymatic activity, in view of the reported discrepancies between the deduced amino acid séquence of cloned preprourokinase (9) and the séquence of the purified natural enzyme (23,24). Another set of constructs from the same group, Scupa n.c.410 and Scupa n.c.410/366 (coded for by pULB9129 and pULB9135), dérives from the former molécules; additional amino acid substitutions have been introduced at the physiological activation site in the proenzyme (Arg''^^ and Lys"^" are respectively replaced by threonine). The purpose of thèse constructions was to obtain prourokinase molécules with similar enzymatic properties as the natural single-chain species (scu-PA) (18)(19)(20), but résistant to cleavage by plasmin. The last constructs belonging to the first group, pULB9139 and pULB9152 (coding for ppUK.(410/366/131)del and Scupa n.c.(410/366/131)del), were designed to eliminate the secondary cleavage site of urokinase (Lys'^'^-Lys''"^) and to replace amino acid 131, tryptophan in pULBlOOO (9), by the cysteine residue found in the natural protein (23). This was achieved by deleting a stretch of amino acids (132-147) and replacing it with a shorter link (Ser-Thr) identical to the one found in t-PA at comparable positions of the enzyme. The product ppUK.(410/366/131)del, coded for by pULB9139, consists thus of a deleted but activable prourokinase, whereas Scupa n.c.(410/366/131)del, coded for by pULB9152, is deleted and non-activable. Finally, for comparison purposes, we constructed a recombinant preprourokinase DNA (pULB9154) identical to that described by Heyneker et al. (8).
The second group of constructions comprises four chimeric molécules. Taking into account the hypothesis of exon shuffling as a mechanism for protein évolution (48), we tried to recombine cDNA fragments, derived from t-PA and u-PA, corresponding as precisely as possible to exons in the gènes and to structural domains in the corresponding proteins. Fg.t-PA/UK.410 and Fg.t-PA/UK.410/366 (coded for by pULB9120 and pULB9124) resuit from the fusion of the finger domain of t-PA to the COOH-terminus of scu-PA. The two species are identical, except for the amino acid at position 366 in urokinase (glycine in pULB9124 and cysteine in pULB9120). Both molécules were designed to explore the potential rôle of the t-PA finger domain (29) in fibrin binding when associated with scu-PA. Another molécule, tPPUK.410/ 366, encoded by plasmid pULB9151, combines a larger portion of t-PA, the A-chain, to the B-chain of u-PA; it is designed to confer to scu-PA the fibrin specificity of t-PA, which appears to be associated to the A-chain moeity (49, 50). A similar product, tPKUK.410 (coded for by pULB9125), consists of the NH2-terminal part of t-PA containing the activation site, up to amino acid 313 in the B-chain, fused to the remaining COOH-terminal part of the B-chain of u-PA.
The single représentative of the third group, UK-K2.410/ 366, coded for by plasmid pULB9137, is a nearly complète scu-PA polypeptide wherein the kringle 2 région of t-PA has been inserted between the single kringle domain and the Bchain. It was designed to test the hypothesis that the kringle 2 région of t-PA behaves as an autonomous domain conferring fibrin binding ability to the enzyme (30).
Expression of Recombinant Plasminogen Activators in Cell Cultures-Eukaryotic cells transfected with the recombinant plasmids described above were cultivated for 3-5 days in the présence of aprotinin to prevent conversion of the recombi-  nant plasminogen activators secreted in the médium. Dosage of the recombinant polypeptides by ELISA using two monoclonal antibodies, AAU2 and AAU6 (46), reveals that ail recombinant plasmids, except pULB9125, direct the transitory expression of urokinase-like material (Table 3). Culture supernatants and extracts of cells transfected with pULB9125 (tPKUK.410) were consistently négative when assayed with anti-urokinase or anti-t-PA antibodies (data not shown).

New Recombinant Plasminogen Activators
Plasminogen Activation by Immobilized Recombinant Activators--The assay for plasminogen activation consisted of a two-step procédure. First, standard urokinase or cell culture supernatants were incubated with matrix-bound monoclonal antibody AAU2. Spécifie complexes were then exposed to plasminogen and to a plasmin-specific chromogenic substrate, D-Ile-Pro-Arg-p-nitroanilide. Any plasmin resulting from the activation of plasminogen will thus react with the substrate and release the paranitroaniline chromophor which can be monitored at 405 nm. In ail cases, typical sigmoidal curves were observed when plotting absorbance as a function of time (t) (data not shown). As previously described by Drapier et al.
(47), plotting of Afu;, as a function of squared time (i^) enables us to linearize the assay as long as initial conditions are valid (see Miniprint Section) (Fig. 3). The slope of thèse straight lines is almost proportional to the total u-PA concentration présent in the expérimental standard incubation mixtures at the moment of enzyme immobilization onto matrix-bound antibody (Fig. 3, inset). From this relationship it is concluded that, within the expérimental range tested (0-8 lU/ml), the amount of immobilized enzyme is proportional to the u-PA concentration in the upstanding solution. The dose-dependent plasminogen activation thus enables us to evaluate the enzymatic activity présent in initial incubation mixtures.
Compared to the curves obtained with control u-PA (Fig.  3), those for ail recombinant plasminogen activators and for purified natural scu-PA appeared biphasic. As shown for three différent dilutions of the recombinant ppUK.410/366 (pULB- 9134), the linear phase was preceeded by an exponential lag phase (Fig. 4A). Pretreatment of irnmobilized recombinant plasminogen activators with plasmin (or with trypsin) and careful élimination of the converting enzyme completely abolished this initial lag phase in subséquent plasminogen activation experiments (Fig. AB). Slopes measured in both expérimental conditions were identical, considering the steep part of the curves only. The data thus indicate that the initial phases, as observed in Fig. 4A, correspond to the activation of the irnmobilized recombinant activators which, under the expérimental conditions, are harvested essentially as singlechain molécules. This was confirmed by the fact that the same lag phase was observed with standard one-chain urokinase purified from Calu-3 cell line (data not shown). Therefore, the enzymatic activity présent in cell culture supernatants was determined by comparing the slopes of the Unear part of the curves obtained for recombinant enzymes (Fig. 4A) to the standard u-PA System (Fig. 3). The enzyme activities are presented in Table 3 for ail recombinant plasminogen activators tested; activities ranged from 0 to 4.5 lU/ml of culture supernatant. Apparent spécifie activities of the recombinant plasminogen activators were obtained by the ratio of measured activities to the amount of antigen (assuming that they display similar affinities for the monoclonal antibodies AAU2 and AAU6 as the standard 54,000-dalton u-PA). As seen in Table  3, values range from 35,000 and 100,000 lU per mg of 54,000dalton activator, except for non-activable scu-PA molécules. From thèse data, it can be concluded that two-chain recombinant enzymes activate plasminogen with catalytic efficiencies comparable to that of u-PA and, thus, that they have maintained a correct three-dimensional active site. On the other hand, no or only very weak activity has been found in the supernatants of respectively R1610 and Cos I cells although they expressed efficiently the non-activable scu-PA molécules. As expected, the modification of the activation site resulted in a single-chain product which cannot be transformed into the two-chain active species in the présence of plasminogen and chromogenic substrate or by plasmin (data not shown). Whether non-convertible scu-PA-like and natural scu-PA will be able to activate plasminogen directly in a freely diffusing System remains to be determined.

DISCUSSION
Thrombolytic agents lacking fibrin specificity, such as u-PA or streptokinase, induce thrombolysis but in association with generalized plasminogen activation and fibrinogen breakdown. t-PA induces thrombolysis with a high degree of clot selectivity due to a markedly higher rate of plasminogen activation at the surface of the fibrin clot, as compared to rates observed in the absence of fibrin (12). Efficient and fibrin-selective thrombolysis has also been obtained with scu-PA, the single-chain precursor of u-PA (22). The mechanism of this selectivity is not fully understood but appears distinct from that of t-PA.
One way to design improved thrombolytic agents would consist of the combination, in a single molécule, of two essential characteristics: high fibrin-mediated plasminogen activation and low fibrin-independent plasminogen activation. Such agents would be expected to display, in vivo, a fibrinolysis/fibrinogenolysis ratio at least equal or superior to that of t-PA or scu-PA.
In the présent study, three main groups of plasminogen activators have been produced; first, we constructed scu-PAlike molécules (Scupa n.c.410 and Scupa n.c.410/366) wherein conversion to two-chain urokinase was prevented by substituting two amino acids involved in the cleavage of the natural scu-PA molécule. This approach finds its rationale in the fact that the conversion of scu-PA into u-PA, in vivo, is not a prerequisite for thrombolysis, but leads to a loss in clot selectivity.
In a second approach, we recombined several domains derived from the A-chain of t-PA with u-PA (in part or in toto). Fg.t-PA/UK.410 and Fg.t-PA/UK.410/366 consist of the low molecular weight scu-PA carrying on its NH2 terminus the finger domain of t-PA; tPPUK.410/366 contains the A-chain of t-PA fused to the low molecular weight scu-PA, and tPKUK.410 is similar to tPPUK.410/366 but contains the cleavage site of t-PA. We hypothesized that some of thèse molécules might not only induce clot selectivity for plasminogen activation, via the mechanism of scu-PA, but also enhance fibrin specificity by binding to the clot via their t-PA structures. Another hybrid product, UK-K2.410/366, carries the putative fibrin binding domain of t-PA (kringle 2) inserted  into the nearly complète scu-PA molécule; this form is expected to yield a potent urokinase-like plasminogen activator showing high fibrin affinity if, indeed, the kringle 2 behaves as an autonomous domain.
Ail recombinant plasminogen activators, except tPKUK.410, were efficiently produced in cell cultures. In addition, spécifie activities of recombinant two-chain u-PA and of chimeric polypeptides were comparable to that of natural u-PA, indicating that the catalytic site carried by the urokinase moiety of the molécules has been maintained and is fully functional for plasminogen activation.
The recombinant uncleavable scu-PA molécules (Scupa n.c.410 and Scupa n.c.410/366), derived from transfected R1610 cells, did not show any activity in our assay System. However, the supernatants derived from transfected Cos I cells exhibited a slightly higher level of activity than the control. This is apparently due to the sécrétion by the cells of an endogenous plasminogen activator. Indeed, pretreatment of Cos I cell supernatant with plasmin confirmed this hypothesis (data not shown). We showed also that plasmin was unable to convert uncleavable scu-PA derived from transfected R1610 cells into an amidolytically active species. This observation supports the conclusion that Scupa n.c. proteins are effectively stable one-chain molécules. Whether the absence of the activation site in the Scupa n.c. molécule has any décisive influence on its biological in vitro and in uivo activities will be investigated in more détail once the recombinant product is obtained in large amounts and purified.