Coamplification and coexpression of human tissue-type plasminogen activator and murine dihydrofolate reductase sequences in Chinese hamster ovary cells.

Expression of human tissue-type plasminogen activator (t-PA) at high levels has been achieved in Chinese hamster ovary (CHO) cells by cotransfection and subsequent coamplification of the transfected sequences. Expression vectors containing the t-PA cDNA gene and dihydrofolate reductase (DHFR) cDNA gene were cotransfected into CHO DHFR-deficient cells. Transformants expressing DHFR were selected by growth in media lacking nucleosides and contained low numbers of t-PA genes and DHFR genes. Stepwise selection of the DHFR+ transformants in increasing concentrations of methotrexate generated cells which had amplified both DHFR genes and t-PA genes over 100-fold. These cell lines expressed elevated levels of enzymatically active t-PA. To optimize both t-PA sequence amplification and t-PA expression, various modifications of the original procedure were used. These included alterations to the DHFR expression vector, optimization of the molar ratio of t-PA to DHFR sequences in the cotransfection, and modification of the methotrexate resistance selection procedure. The structure of the amplified DNA, its chromosomal location, and its stability during growth in the absence of methotrexate are reported.

Thrombolytic therapy with tissue-type plasminogen activator: new modes and novel variant plasminogen activators.

Tissue-type plasminogen activator produced by recombinant DNA technology, has been established as an important thrombolytic agent in the treatment of acute myocardial infarction. New approaches to increase the effectiveness of this agent, including rapid high dose administration are being investigated. Several novel protein engineered variant forms of plasminogen activators have been produced that have increased thrombolytic potency in animal models and offer the potential of a more effective lower dose agent than can be administered clinically as a single bolus intravenous injection.

Soluble P-selectin glycoprotein ligand 1 inhibits ocular inflammation in a murine model of allergy.

PURPOSE: To assess the anti-inflammatory modality of a soluble extracellular form of P-selectin glycoprotein ligand 1 (sPSGL-1) in a mouse model of ocular allergic response. METHODS: Potential anti-inflammatory effects of sPSGL-1 were investigated in SWR/J mice sensitized by topical application of short ragweed pollen to the nasal mucosa followed by a challenge of the ocular mucosa with the same allergen. Five experimental groups were included in these studies: A, mice neither sensitized nor challenged with pollen (control group 1); B, animals sensitized but not challenged (control group 2); C, animals not sensitized but challenged (control group 3); D, animals sensitized and challenged; and E, sensitized animals treated with sPSGL-1 before pollen challenge. All experimental groups were evaluated for gross morphologic ocular changes, and histologic assessments were made to determine the onset/progression of inflammatory reactions and to look for evidence of eosinophil infiltration. RESULTS: Mice sensitized and challenged with pollen developed clinical signs consistent with human allergic conjunctivitis. These signs correlate with histologic changes in the conjunctival epithelium and stroma (e.g., edema and extensive eosinophil infiltration). Moreover, the ocular changes also correlated with evidence of eosinophil degranulation. However, sensitized and challenged mice concurrently treated with sPSGL-1 displayed no inflammatory ocular changes associated with a ragweed-induced type-1 hypersensitivity reaction. The lack of ocular changes included the absence of histologic late-phase inflammatory changes of the conjunctiva and a 97% reduction in the induced eosinophil infiltrate. CONCLUSIONS: The antagonistic intervention of cell- cell interactions through the blockade of selectin-dependent leukocyte adhesion may offer novel therapeutic strategies to modulate inflammatory responses. The potent inhibitory effects on eosinophil recruitment and late-phase inflammation suggest a role for sPSGL-1 in the treatment of ocular allergic diseases.

Variants of human tissue-type plasminogen activator. Fibrin binding, fibrinolytic, and fibrinogenolytic characterization of genetic variants lacking the fibronectin finger-like and/or the epidermal growth factor domains.

Tissue-type plasminogen activator (t-pa) is a serine protease comprising four different putative structural domains with homologies to fibronectin finger-like structures (finger), epidermal growth factor, kringle structures, and the active site of serine proteases. Only the finger and epidermal growth factor domain are each entirely encoded by unique single exons. We assessed the functional contribution of these two structural domains by making mutants precisely deleted for one or both of the relevant exons. The three mutant genes were expressed in monkey cells, and the variant proteins, purified from the culture medium, were characterized for their fibrinolytic activity, fibrinogenolytic potential, and affinity for fibrin. No significant difference in any biochemical property was observed among the variants. All three variants retained a catalytic dependence on cyanogen bromide fragments of fibrinogen which could not be distinguished from the wild-type enzyme. The activities of the variants were also very similar to that of wild-type t-pa, showing no detectable fibrinogenolytic potential in human plasma at activator concentrations of 500 IU/ml, or when their fibrinolytic activity was tested in human plasma using the 125I-labeled fibrin clot lysis assay at activator concentrations of 150 IU/ml or greater. However, the variants were markedly defective in fibrinolysis at low activator concentrations such that essentially no fibrinolysis was detected at 15 IU/ml. Measurement of fibrin binding showed that the variants lacked the high fibrin binding characteristic of wild-type t-pa. These results demonstrate that the fibrin specificity and fibrin-dependent activity of t-pa are independent of the protein's high affinity for fibrin. The implication of these results is that the t-pa variants would be ineffective activators at a physiological concentration of approximately 2 IU/ml but would be expected to behave similarly to wild-type t-pa at the steady-state plasma concentrations of 0.75-1.25 micrograms/ml (approximately 500 IU/ml) currently required for coronary reperfusion in patients receiving t-pa for acute myocardial infarction (Garabedian, H.D., Gold, H.K., Leinbach, R.C., Yasuda, T., Johns, J.A., and Collen, D. (1986) Am. J. Cardiol. 58, 673-679).

Stable hairpin structure within the 5'-terminal 85 nucleotides of poliovirus RNA.

The primary sequence of a 5'-terminal fragment of poliovirus type 1 RNA, generated by digestion with RNase III, has been determined. This sequence reveals the presence of a stable hairpin structure beginning nine nucleotides from the terminally linked protein VPg. The sequence does not contain (i) the initiation codons AUG or GUG or (ii) the putative ribosome-binding sequence complementary to the 3' end of eucaryotic ribosomal 18S RNA. The stem-and-loop structure identified can be drawn in either plus or minus RNA strands. It is unclear to which strand functional significance (if any) can be assigned. It is possible that the hairpin structure is involved in ribosomal recognition and translation or in RNA synthesis by interacting with replicase molecules.

The dissociation constants and stoichiometries of the interactions of Lys-plasminogen and chloromethyl ketone derivatives of tissue plasminogen activator and the variant delta FEIX with intact fibrin.

Active-site-blocked, fluorescent derivatives of tPA (Activase) and a variant (delta FEIX) which lacks the finger and epidermal growth factor-like domains and possesses Asn to Gln and Val to Met mutations at residues 117 and 245, respectively, were prepared. The binding of these to fibrin was studied by adding them at systematically varying concentrations to fibrinogen, at a fixed concentration, inducing clotting with thrombin, separating free and bound tPA or delta FEIX by centrifugation, and measuring the concentration of unbound material by extrinsic fluorescence. Similar studies were performed with Glu and Lys-plasminogen, using intrinsic fluorescence. epsilon-amino caproic acid (EACA) was utilized to distinguish kringle-dependent from finger-dependent binding. In the absence of EACA, delta FEIX-bound fibrin through a single class of sites with Kd = 0.69 microM and n = 1.34 delta FEIX/fibrin. The binding of delta FEIX was completely inhibited by EACA and 50% displacement occurred at [EACA] = 300 microM. Fibrin-bound tPA was only partially displaced with EACA. In the presence of 30 mM EACA, tPA binding reflected a single class of sites with Kd = 0.26 microM and n = 0.60 tPA/fibrin. In the absence of EACA, tPA binding was complex, typified by downwardly curved Scatchard plots, and was consistent with interactions of the two classes of sites, characterized by Kd = 0.13 microM, n = 0.60 and Kd = 0.61 microM, n = 1.23. These were attributed to finger and kringle-dependent interactions, respectively. Under the experimental conditions employed, Glu-plasminogen exhibited no binding to fibrin, whereas Lys-plasminogen bound to a single class of sites with Kd = 0.25 microM and n = 1.02 plasminogen/fibrin. This binding was completely inhibited by EACA and 50% displacement occurred at [EACA] = 28 microM. Competition experiments indicated that Lys-plasminogen does not displace either tPA or delta FEIX from fibrin. From these results the conclusions are drawn that tPA can interact with intact fibrin by two different and independent modes, involving, respectively, the finger and kringle 2 domains, and neither of these modes are competitive with the kringle-dependent binding of Lys-plasminogen.

Pharmacokinetic and distribution analysis of variant forms of tissue-type plasminogen activator with prolonged clearance in rat.

Human tissue-type plasminogen activator (t-PA) is a glycoprotein used currently in thrombolytic therapy. Because of its rapid half-life (T1/2) of approximately five minutes, intravenous (IV) infusion of large doses (approximately 100 mg) are required in patients treated for myocardial infarction. To identify the determinant(s) on t-PA responsible for such rapid clearance, metabolically labeled forms of recombinant t-PA were analyzed in rats following IV administration. The following seven forms of t-PA were tested: (a) natural or glycosylated wild-type t-PA; (b) nonglycosylated wild-type t-PA; (c) delta F t-PA, which lacks the fibronectin fingerlike domain; (d) delta E t-PA, which lacks the epidermal growth factor (EGF) domain; (e) delta FE t-PA, which lacks both the finger and EGF domains; (f) delta FE3X t-PA, a form of delta FE t-PA in which Asn-linked glycosylation is prevented at all known glycosylation sites (Asn-117, 184, and 448; replaced by Gln); and (f) delta FE1X t-PA, a form of delta FE t-PA in which high-mannose-type glycosylation is prevented at Asn-117. Both glycosylated and nonglycosylated wild-type t-PA cleared in an exponential biphasic manner, with an initial alpha-phase T1/2 of 0.8 and 1.9 minutes, respectively. This result demonstrates that carbohydrate is not the primary mediator of the rapid clearance of t-PA. The liver was the primary organ responsible for uptake of these molecules. All other proteins tested, except for delta E t-PA, demonstrated primarily monophasic clearance patterns with T1/2 ranging between 12 and 27 minutes, and reduced uptake in the liver. delta E t-PA however, cleared in a biphasic manner with an alpha-phase T1/2 of 2.1 minutes. Results presented suggest that the clearance of t-PA is mediated by two distinct mechanisms. The primary determinant(s) responsible for modulating the rapid clearance of t-PA appears to be resident within the polypeptide sequence encoding the finger and/or EGF domains, with emphasis on the finger domain. A second and less significant contribution to clearance is defined by the presence and type of glycosylation.

Pharmacokinetics and thrombolytic properties of a nonglycosylated mutant of human tissue-type plasminogen activator, lacking the finger and growth factor domains, in dogs with copper coil-induced coronary artery thrombosis.

The pharmacokinetics and thrombolytic properties of a variant of human tissue-type plasminogen activator (t-PA), obtained by deletion mutagenesis of the NH2-terminal fibronectin-like finger (F) and epidermal growth factor (E) domains, and substitution of the three known glycosylated Asn residues by Gln (t-PA-delta FE3X), were studied in dogs with a copper coil-induced thrombosis of the left anterior descending coronary artery. Bolus injections were given during 2 min to groups of three dogs. Injection of 0.15 mg/kg resulted in peak antigen levels in plasma of 1.58 +/- 0.72 micrograms/ml (mean +/- SEM) and caused reperfusion within 14 +/- 6 min. With 0.075 mg/kg, corresponding values of 0.81 +/- 0.20 micrograms/ml and 31 +/- 15 min were obtained. A bolus of 0.038 mg/kg yielded plasma peak levels of 0.43 +/- 0.20 micrograms/ml but did not cause coronary recanalization within 3 h. A bolus injection of natural t-PA (Mel-t-PA) at a dose of 0.1 mg/kg in four dogs resulted in plasma peak levels of 0.46 +/- 0.09 micrograms/ml and caused partial coronary artery reperfusion within 3 h in one of four dogs (after 31 min). None of these injections caused a significant decrease of the fibrinogen level. Pharmacokinetic parameters for t-PA-delta FE3X were alpha half-life (t1/2) 14-18 min, beta t1/2 72-125 min, and plasma clearance 21-36 ml/min. For Mel-t-PA, the corresponding values were 3 min, 8 min, and 520 ml/min. We conclude that the variant t-PA-delta FE3X has a markedly longer plasma t1/2 than does Mel-t-PA and, when administered as a bolus injection, a higher thrombolytic efficacy.

Functional effects of asparagine-linked oligosaccharide on natural and variant human tissue-type plasminogen activator.

The role of Asn-linked oligosaccharide in the functional properties of both human tissue-type plasminogen activator (t-PA) and a genetic variant of t-PA was studied. Nonglycosylated and glycosylated wild-type t-PA were produced in mammalian cells which express recombinant t-PA. These proteins were compared in fibrin binding and 125I-labeled fibrin clot lysis assays, using purified components. The nonglycosylated form showed higher fibrin binding, as well as higher fibrinolytic potency than the glycosylated form. Subsequently, prevention of glycosylation of a t-PA variant which lacked the finger and epidermal growth factor domains (delta FE), was carried out in an attempt to enhance its fibrinolytic activity. Glycosylation was prevented by changing Asn to Gln; at Asn-117 to produce delta FE1X t-PA, and at Asn-117, -184, and -448 to produce delta FE3X t-PA. All variants were similar to wild-type t-PA in their catalytic dependence on fibrinogen fragments, fibrinolytic activity in fibrin autography analysis, and plasminogen activator activity. In a clot lysis assay, using citrated human plasma, the fibrinolytic potency of the variants were comparable to that of wild-type t-PA at activator concentrations of 17-51 nM (approximately 1-3 micrograms/ml). At 0.5-5.1 nM (approximately 0.03-0.3 micrograms/ml), however, the variant proteins had lower fibrinolytic potency than wild-type t-PA. Fifty percent lysis in 1.5 h for wild-type, delta FE, delta FE1X, and delta FE3X t-PA, required 2.5, 10, 7.5, and 5.5 nM t-PA, respectively. The fibrinogenolytic activity in human plasma was measured for wild-type, delta FE, delta FE1X, and delta FE3X t-PA, and showed significant fibrinogen depletion after 3 h of incubation at 51 nM, decreasing to 11, 11, 50, and 72% of basal levels, respectively. These data indicate that partial or total nonglycosylated t-PA variants have a higher fibrinolytic versus fibrinogenolytic ratio than their fully glycosylated counterparts.

The structure of poliovirus replicative form.

The structure of polio replicative form (RF) has been investigated by 3' end labeling and the use of polynucleotide phosphorylase to now allow a complete composite of the RF structure. The evidence presented indicates that the 3' terminal sequence of the minus strand is an exact complement to the 5' end of polio RNA. This suggests that the 5' terminal U of polio RNA is genetically coded. Other data is presented to show that in addition to the genetically coded poly(A) tract of the plus strand in RF, a single-stranded poly(A) tail protrudes beyond the double-stranded RNA.

Identification of the initiation site of poliovirus polyprotein synthesis.

The complete nucleotide sequence of poliovirus RNA has a long open reading frame capable of encoding the precursor polyprotein NCVP00. The first AUG codon in this reading frame is located 743 nucleotides from the 5' end of the RNA and is preceded by eight AUG codons in all three reading frames. Because all proteins that map at the amino terminus of the polyprotein (P1-1a, VP0, and VP4) are blocked at their amino termini and previous studies of ribosome binding have been inconclusive, direct identification of the initiation site of protein synthesis was difficult. We separated and identified all of the tryptic peptides of capsid protein VP4 and correlated these peptides with the amino acid sequence predicted to follow the AUG codon at nucleotide 743. Our data indicate that VP4 begins with a blocked glycine that is encoded immediately after the AUG codon at nucleotide 743. An S1 nuclease analysis of poliovirus mRNA failed to reveal a splice in the 5' region. We concluded that synthesis of the poliovirus polyprotein is initiated at nucleotide 743, the first AUG codon in the long open reading frame.

Cleavage sites within the poliovirus capsid protein precursors.

Partial amino-terminal sequence analysis was performed on radiolabeled polio-virus capsid proteins VP1, VP2, and VP3. A computer-assisted comparison of the amino acid sequences obtained with that predicted by the nucleotide sequence of the poliovirus genome allows assignment of the amino terminus of each capsid protein to a unique position within the virus polyprotein. Sequence analysis of trypsin-digested VP4, which has a blocked amino terminus, demonstrates that VP4 is encoded at or very near to the amino terminus of the polyprotein. The gene order of the capsid proteins is VP4-VP2-VP3-VP1. Cleavage of VP0 to VP4 and VP2 is shown to occur between asparagine and serine, whereas the cleavages that separate VP2/VP3 and VP3/VP1 occur between glutamine and glycine residues. This finding supports the hypothesis that the cleavage of VP0, which occurs during virion morphogenesis, is distinct from the cleavages that separate functional regions of the polyprotein.

Poliovirus neutralization epitopes: analysis and localization with neutralizing monoclonal antibodies.

Two hybridomas (H3 and D3) secreting monoclonal neutralizing antibody to intact poliovirus type 1 (Mahoney strain) were established. Each antibody bound to a site qualitatively different from that to which the other antibody bound. The H3 site was located on intact virions and, to a lesser extent, on 80S naturally occurring empty capsids and 14S precursor subunits. The D3 site was found only on virions and empty capsids. Neither site was expressed on 80S heat-treated virions. The antibodies did not react with free denatured or undenatured viral structural proteins. Viral variants which were no longer capable of being neutralized by either one or the other antibody were obtained. Such variants arose during normal cell culture passage of wild-type virus and were present in the progeny viral population on the order of 10(-4) variant per wild-type virus PFU. Toluene-2,4-diisocyanate, a heterobifunctional covalent cross-linking reagent, was used to irreversibly bind the F(ab) fragments of the two antibodies to their respective binding sites. In this way, VP1 was identified as the structural protein containing both sites.

Tissue-type plasminogen activator binding to human endothelial cells. Evidence for two distinct binding sites.

The endothelium may contribute to fibrinolysis through the binding of plasminogen activators or plasminogen activator inhibitors to the cell surface. Using a solid-phase radioimmunoassay, we observed that antibodies to recombinant tissue-type plasminogen activator (rt-PA) and plasminogen activator inhibitor type 1 (PAI-1) bound to the surface of cultured human umbilical vein endothelial cells (HUVEC). HUVEC also specifically bound added radiolabeled rt-PA with apparent steady-state binding being reached by 1 h at 4 degrees C. When added at low concentrations (less than 5 nM), rt-PA bound with high affinity mainly via the catalytic site, forming a sodium dodecyl sulfate-stable 105-kDa complex which dissociates from the cell surface over time and which could be immunoprecipitated by a monoclonal antibody to PAI-1. rt-PA bound to this high affinity site retained less than 5% of its expected plasminogen activator activity. At higher concentrations, binding did not require the catalytic site and was rapidly reversible. rt-PA initially bound to this site retained plasminogen activator activity. These studies suggest that tissue-type plasminogen activator and PAI-1 are expressed on the surface of cultured HUVEC. HUVEC also express unoccupied binding sites for exogenous tissue-type plasminogen activator. The balance between the expression of plasminogen activator inhibitors and these unoccupied binding sites for plasminogen activators on the endothelial surface may contribute to the regulation of fibrinolysis.

Protein engineering of novel plasminogen activators with increased thrombolytic potency in rabbits relative to activase.

Human tissue-type plasminogen activator (t-PA) is a glycoprotein used currently in thrombolytic therapy for patients with acute myocardial infarction. Due to its rapid rate of clearance from the circulation, continuous intravenous administration of approximately 100 mg over 3 h is recommended. We have previously characterized novel thrombolytic variant forms of t-PA which offer the potential of administration by bolus injection and reduced dosage due to their slower rates of clearance, relative to t-PA. This study was undertaken to quantitatively compare the pharmacokinetics, thrombolytic activity, and hemostatic effects of two of these variant forms, called delta FE1X and delta FE3X plasminogen activator (PA), with commercially available recombinant t-PA (Activase). These evaluations were performed in rabbits after bolus intravenous injection of the proteins. Following injection of 0.25 mg of protein/kg of body weight, the rates of clearance for delta FE3X and delta FE1X PA antigen were decreased approximately 9- and 18-fold, respectively, relative to Activase. Plasma plasminogen activator activity was also measured and the rates of clearance of delta FE3X and delta FE1X PA activity were similarly decreased by approximately 9- and 22-fold, respectively, relative to Activase. To quantitate thrombolytic activity we used the rabbit jugular vein thrombosis model and demonstrated that approximately 50% thrombolysis was achieved with delta FE1X and delta FE3X PA at approximately an 8.6- and 3-fold lower dose than Activase, respectively. No major differences in fibrinogen and alpha 2-antiplasmin depletion were observed among the agents at doses required to produce 50% thrombolysis, indicating similarities in fibrin specificities among these agents. These results demonstrate a reciprocal relationship between thrombolysis and rate of clearance for these thrombolytic proteins. The 8.6-fold increase in potency of delta FE1X PA relative to Activase supports the future clinical testing of this novel engineered protein as a thrombolytic agent.

P-selectin and E-selectin. Distinct but overlapping leukocyte ligand specificities.

P-selectin on platelets and endothelial cells and E-selectin on endothelial cells are leukocyte receptors that recognize lineage-specific carbohydrates on neutrophils and monocytes. The proposed ligands for these receptors contain the Le(x) core and sialic acid. Since other investigators have shown that both E-selectin and P-selectin bind to sialylated Le(x), we evaluated whether E-selectin and P-selectin recognize the same counter-receptor on leukocytes. The interaction of HL60 cells with Chinese hamster ovary (CHO) cells expressing P-selectin or E-selectin was studied. To determine whether a protein component is required in addition to sialyl Le(x) for either P-selectin or E-selectin recognition, HL60 cells or neutrophils were digested with proteases, including chymotrypsin, elastase, proteinase Glu-C, ficin, papain, or thermolysin. Cells treated with these proteases bound E-selectin but not P-selectin. Fucosidase or neuraminidase treatment of HL60 cells markedly decreased binding to both E-selectin- and P-selectin-expressing CHO cells. Growth of HL60 cells in tunicamycin inhibited the ability of these cells to support P-selectin-mediated binding and, to a lesser extent, E-selectin-mediated binding. Purified P-selectin inhibited CHO:P-selectin binding to HL60 cells, but incompletely inhibited CHO:E-selectin binding to HL60 cells. However, purified soluble E-selectin inhibited CHO:P-selectin and CHO:E-selectin binding to HL60 cells equivalently and completely. COS cells, unable to bind to E-selectin or P-selectin, bound E-selectin but not P-selectin upon transfection with alpha-1,3-fucosyltransferase or alpha-1,3/1,4-fucosyltransferase. Similarly, LEC 11 cells expressing sialyl Le(x) bound E-selectin- but not P-selectin-expressing CHO cells. Sambucus nigra lectin, specific for the sialyl-2,6 beta Gal/GalNAc linkage, inhibited P-selectin but not E-selectin binding to HL60 cells. Although sialic acid and Le(x) are components of the P-selectin ligand and the E-selectin ligand, these results indicate that the ligands are related, having overlapping specificities, but are structurally distinct. A protein component containing sialyl Le(x) in proximity to sialyl-2,6 beta Gal structures on the P-selectin ligand may contribute to its specificity for P-selectin.

Primary structure, gene organization and polypeptide expression of poliovirus RNA.

The primary structure of the poliovirus genome has been determined. The RNA molecule is 7,433 nucleotides long, polyadenylated at the 3' terminus, and covalently linked to a small protein (VPg) at the 5' terminus. An open reading frame of 2,207 consecutive triplets spans over 89% of the nucleotide sequence and codes for the viral polyprotein NCVPOO. Twelve viral polypeptides have been mapped by amino acid sequence analysis and were found to be proteolytic cleavage products of the polyprotein, cleavages occurring predominantly at Gln-Gly pairs.

Replacement of finger and growth factor domains of tissue plasminogen activator with plasminogen kringle 1. Biochemical and pharmacological characterization of a novel chimera containing a high affinity fibrin-binding domain linked to a heterologous protein.

A novel triple-kringle plasminogen activator protein, PK1 delta FE1X, has been produced which is a genetic chimera between the fibrin binding kringle 1 domain of plasminogen and the two kringles and serine protease domains of naturally occurring wild-type tissue plasminogen activator (wt t-PA). This chimera also contains a modification to prevent high mannose type N-linked glycosylation on kringle 1 of t-PA. PK1 delta FE1X is biochemically and fibrinolytically similar to wt t-PA in vitro but retains the decreased plasma clearance rate characteristic of other t-PA variants which lack fibronectin finger-like and epidermal growth factor domains. The serine protease domain of PK1 delta FE1X exhibits the amidolytic activity characteristic of wt t-PA. In an indirect coupled plasminogen activator assay, the specific activity of PK1 delta FE1X is approximately 1.4 times greater than that of wt t-PA. In a fibrin film-binding assay, greater binding to untreated fibrin is observed with wt t-PA than with PK1 delta FE1X. However, following limited plasmin digestion of the fibrin film, PK1 delta FE1X binding increases to the level observed with wt t-PA. The incremental binding to plasmin-digested fibrin observed with PK1 delta FE1X is eliminated if plasmin digestion of the fibrin film is followed by carboxypeptidase B treatment. This result suggests that plasminogen kringle 1 binds plasmin-digested fibrin even after recombination with a heterologous protein. The fibrinolytic activity of PK1 delta FE1X in human plasma clot lysis assays was similar to that of wt t-PA at activator concentrations of approximately 1 microgram/ml. At substantially lower concentrations, approximately 0.1 microgram/ml, PK1 delta FE1X was only slightly less active than wt t-PA. Pharmacokinetic analysis showed that wt t-PA activity is cleared approximately 15 times as rapidly as PK1 delta FE1X following intravenous bolus injection. In a rabbit jugular vein clot lysis model, intravenous bolus injection of 0.06 mg/kg of PK1 delta FE1X showed greater thrombolytic potency than a similar administration of 0.5 mg/kg of wt t-PA. Thus it appears that in vitro exon shuffling techniques can be used to generate novel fibrinolytic agents which biochemically and pharmacologically represent the combination of individual domains of naturally occurring proteins.

Site-directed mutagenesis in human tissue-plasminogen activator. Distinguishing sites in the amino-terminal region required for full fibrinolytic activity and rapid clearance from the circulation.

Recombinant variants of tissue plasminogen activator (t-PA) containing either substitutions or deletions of amino acids within the fibronectin finger-like domain (residues 6-50) were found to exhibit widely varying in vivo clearance profiles in rats and fibrinolytic activity in 125I-fibrin clot lysis assays. Clearance was not significantly affected by changes in the densely charged region of amino acid residues 7-10. Deletions or substitutions of amino acids in the region 14-32 decreased both fibrinolytic activity and the clearance of the enzyme. Modifications within the predicted omega loop of residues 37-41 affected clearance only to a small degree, whereas amino acid alterations in the region of residues 42-49 resulted in as much as a 6-fold decrease in the rate of clearance with only relatively minor decreases in the fibrinolytic activity of the variants. The cumulative results distinguish discrete sections of the NH2-terminal region of the enzyme as determinants of in vivo clearance and fibrinolytic activity of t-PA. In addition, the fibrinolytic activity of a variant containing the substitutions Gln42----Asn, His44----Glu, and Asn117----Gln, when compared with wild-type t-PA in an in vivo rabbit venous clot lysis model, was found to have similar lytic efficacy at approximately one-fourth the dose. We conclude that decreases in the in vivo clearance of t-PA can result in more potent thrombolytic agents in vivo, even though the in vitro fibrinolytic activity of the enzyme may be somewhat impaired.