Sep 29, 1990 - Biochemical and Functional Characterization of Human Tissue-type. Plasminogen Activator Variants with Mutagenized Kringle Domains*.
THE JOURNAL OF BIOLOGICAL. CHEMIST 0 1990 by The American Society for Biochemistry
Biochemical Plasminogen
Vol. 265, No. 21, Issue of July 25, pp. 121’34-12191, 1990
Printed in U.S.A.
and Molecular Biology, Inc.
and Functional Characterization of Human Tissue-type Activator Variants with Mutagenized Kringle Domains* (Received
D&sir6 Collen$, Henri Alexander Tulinskye,
Roger Lijnen, Frank and Luc Nelles
From the Center for Thrombosis and SDepartment of Chemistry, Michigan
Bulens,
Vascular Research, State University,
Anne-Mieke
University of kuven, East Lansing, Michigan
The cDNA encoding full-length human tissue-type plasminogen activator (t-PA) and five variant cDNAs, constructed by in vitro site-directed mutagenesis, were cloned and expressed in Chinese hamster ovary cells. The variant cDNAs were designed to increase the fibrin affinity of t-PA by mutagenesis in the kringle domains of specific amino acids which are assumed to constitute the lysine-binding site. These amino acids were replaced with the corresponding amino acids present in kringle 1 of plasminogen, which has a high affinity for lysine analogues. The mutants included: rt-PA-Arg”’ with a Pro12’ + Arg mutation; rt-PAArg’64,Tyr’65 with Ser’64,Ser*e5 + Arg,Tyr; rt-PAArg’2S,Arg’64,Tyr’s5 with Pro12s,Ser1e4,Ser1ee -* Arg, Arg,Tyr; rt-PA-Arg213 with Va1213 + Arg; and rt-PAArg2b2 with Thr2b2 -) Arg. Compared to wild-type recombinant t-PA (rt-PA), the catalytic efficiency for plasminogen activation was enhanced 4-fold for rt-PAArg12’, and 3-fold for rt-PA-Arg2b2 while stimulation of plasminogen activation by CNBr-digested fibrinogen was comparable to wild-type rt-PA for rt-PAArg12’ and a-fold enhanced for rt-PA-Arg2S2. All rtPA moieties showed a similar concentration-dependent and nearly quantitative binding to fibrin as well as to lysine-Sepharose and induced a similar timeand concentration-dependent lysis of a 12’1-fibrin-labeled plasma clot immersed in human plasma. Equieffective concentrations (causing 50% clot lysis in 2 h) were 0.17 pg/ml for rt-PA-Arg”‘, and 0.31 ccg/ml for r&PAArg252 as compared to 0.55 rg/ml for rt-PA. The initial plasma half-life following intravenous bolus injection of 0.25 mg/kg in hamsters was 1.2-2.6 min, not significantly different from wild-type rt-PA (2.4 min). Continuous infusion over 60 min in hamsters with a 12’1-fibrin-labeled pulmonary embolus produced 50% clot lysis over background with a dose of 0.9-1.8 mg/ kg, which is not markedly superior to wild-type rt-PA (2.1 mg/kg). It is concluded that these variants, designed to mimic the high affinity fibrin-binding site of plasminogen, are not endowed with a markedly improved thrombolytic potency.
* The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. $ To whom correspondence should be addressed: Center for Thrombosis and Vascular Research, K. U. Leuven, Campus Gasthuisberg, 0 & N Herestraat 49, B-3000 Leuven, Belgium.
for publication,
September
29, 1990)
Vandamme, B-3000 48824
Leuven,
Belgium
and the
Tissue-type plasminogen activator (t-PA)’ is a fibrin-selective thrombolytic agent which, when given intravenously, reperfuses approximately 70% of occluded coronary arteries in patients with acute myocardial infarction, thereby preserving left ventricular function and reducing mortality (for references, see Ref. 1). It is composed of multiple structuralfunctional domains including a finger-like domain, an epiderma1 growth factor domain, two disulfide-bonded triple-loop structures commonly called kringles, and a serine protease domain (2, 3). The fibrin selectivity of t-PA is mediated via its affinity for fibrin (4), which in turn is supported by the finger-like and the second kringle domains (5) and potentially also by the first kringle domain (6). The fibrin affinity of the kringle(s) of t-PA is probably mediated via its lysine-binding site which endows the molecule with affinity for COOHterminal lysine residues (5). The three-dimensional structure of the lysine-binding site of the first and fourth kringles of plasminogen (PGK1 and PGK,) have recently been modeled on the basis of the known three-dimensional structure of prothrombin kringle 1 (7, 8) and compared with corresponding regions of modeled kringles of t-PA (PAK1 and PAK2) (8). The lysine-binding site of PGK1, which has the highest affinity for lysine analogues, can be defined by residues 31-35, 54-58, 61-64, and 71-75 based on the numbering system used by Tulinsky et al. (8). It is characterized by an apparent dipolar surface, consisting of Asp55and Asp57as the anionic center, and Arg34 and Arg71 as the cationic center, separated by a hydrophobic region of highly conserved aromatic residues (Phe35,Trp6*, Phe64,Trp7’, and Ty?). Both t-PA kringles possessthe Asp55and Aspb7of the lysine-binding site but neither has an Arg at positions 34 or 71 of the cationic center: PAK, has Pro34 and Ser71 and does not bind lysine, whereas PAKz has VaP4 and Thr71 but has some fibrin affinity. In addition, PAKl has Ser7’, whereas all other lysine-binding kringles have an aromatic residue in ’ The abbreviations used are: t-PA, tissue-type plasminogen activator; r&PA, recombinant t-PA obtained by expression of cDNA in Chinese hamster ovary cells; rt-PA-Arglz5, rt-PA obtained by sitespecific mutagenesis of Pro”” to Arg; rt-PA-Arg’“,Tyr’65, rt-PA obtained by site-specific mutagenesis of Serle4 to Arg and of Serle5 to Tyr; rt-~A-Argi25,Arg’G’,Tyri65, rt-PA obtained by-site-specific mutagenesis of Pro”” to Are. of SerlG4 to Arp., and of Serls5 to Tyr: rtPi-Ar$13, r&PA obtain& by site-specific. mutagenesis of VaiZi3 to Arg; r&PA-A@‘, rt-PA obtained by site-specific mutagenesis of Th? to Arg; S-2251, D-valyl-leucyl-lysine-p-nitroanilide; D-&-ProArg-CH,C1; b-isoleucyl-proiyl-arginine-chloromethylketone; IU, international units: SDS-PAGE, sodium dodecyl sulfate-polyacrylamide gel electrophoresis; CHG, Chinese hamster ovary; l?LISA, enzyme-linked immunosorbent assay; 6-AHA, 6-aminohexanoic acid; PGK, plasminogen kringle; PAK, tissue-type plasminogen activator kringle.
12184
Kringle Mutants TABLE I Amino acid residues of plasminogen kringle 1 fPGK4 and of tissuet.ype plasminogen activator kringle 1 (PAKJ and kringle 2 (PAK?) which may be implicated in the interaction witklysine Binding
pocket
Anionic
center
Cationic
center
Hydrophobic
a Amino b Amino
acid acid
site
PGK,”
PAKlb
PAKsb
lining
Asp= Asp5’ Arg3 Argr’ Phe”” Trp= Phe6* TIp72 Tyr’4
Asp”’ Asp”’ Pro’= Serrc4 Ty? TIJI’~~ Tyr’56 Ser’65 Phe”’
Asp’= Asp=’ Va1213 Th? Tyr”” Trp”’ His24~
numbering numbering
according according
to Tulinksy to Pennica
Trp253 Tyr255 et al. (8). et al. (2).
this position which contributes to the hydrophobic platform of the lysine-binding site (8). We hypothesized that site-specific mutagenesis of these variant residues in PAK, and PAKz into the residues found in PGKl might increase the fibrin affinity of t-PA and consequently alter its fibrin selectivity and thrombolytic potency. Therefore we have constructed and expressed t-PA cDNA variants with the codon for Prolz5 replaced with a codon for Arg, that for Ser’64 with one for Arg, Ser”” with Tyr, Va12i3 with Arg, and Thrz5’ with Arg (numbering system according to Pennica et al. (2)). The numbering system for these residues and the substitutions performed in the present study are defined in Table I. The variants were purified and their biochemical and biological properties determined. MATERIALS
AND
METHODS
Proteins and Reagents-Recombinant single-chain t-PA (Activasea) was obtained from Genentech. Monoclonal antibodies blocking the enzymatic activity of single-chain t-PA (MA-2G6) or the fibrin binding of t-PA (MA-1C8) were obtained and characterized as described (9). Human plasminogen, fibrinogen, and CNBr-digested fibrinogen were obtained and characterized as described elsewhere (10). Aprotinin (TrasylolR) was purchased from Bayer, D-Val-LeuLys-p-nitroanilide (S-2251) from KabiVitrum, and D-Ile-Pro-ArgCH&l from UCB. Restriction and DNA modifying enzymes were obtained from Gibco/BRL, Boehringer Pharma, or New England Biolabs. The plasmids pNeo (Pharmacia), pSV328A’ (provided by Van Heuvel (ll)), pSV328DHFR (constructed as described elsewhere (12)), pUC18 (provided by Yanisch-Perron et al. (13)) and pRR23 (obtained from Boshart et al. (14)) were used for the construction of the expression plasmid pCMpNeo, as described elsewhere (15). Construction and Expression of t-PA cDNA-The isolation of tPA cDNA and the construction of pULt-PA, containing the complete coding sequence, 127 nucleotides of 5’untranslated sequence, and 391 nucleotides of 3’untranslated sequence, has been described elsewhere (16). From this plasmid a BamHl restriction fragment containing the entire t-PA cDNA sequence was transferred to the unique BamHl restriction site of the expression vector pSV328DHFR, resulting in pSVt-PADHFR. Dihydrofolate reductase-deficient Chinese hamster ovary (CHO) cells (17) were transfected with this plasmid according to Graham and van der Eb (18), and selected for dihydrofolate reductase production as described elsewhere (12). Isolated dihydrofolate reductase+. colonies were screened for t-PA antigen secretion using a specitic ELISA (19). Large scale production of t-PA was performed as described previously (20). Mutagenesis of Kringle Domains in t-PA-Mutagenesis and expression of t-PA variants were carried out using a t-PA cDNA with truncated 5’- and 3’-untranslated ends (pt-PAsh) which was obtained as follows. An 850-base pair EcoRl restriction fragment covering the 5’untranslated sequence and the NH*-terminal codons of the t-PA cDNA (2, 16) and an 888-base pair EcoRl restriction fragment containing the COOH-terminal coding region and 391 base pairs of 3’untranslated nucleotides, were separately inserted in the EcoRl restriction site of pUC18 in such a way that the untranslated regions faced the Hind111 restriction site of the multiple cloning site. Both resulting plasmids were linearized with Sal1 and Sphl and subjected
of t-PA
12185
to unidirectional deletion with exonuclease Ill and Sl nuclease according to Henikoff (21). The deleted plasmids were religated and, after transformation, colonies were screened for suitable deletions. For each of the two initial constructs, a clone was selected with most of the untranslated sequence removed. With these clones the complete coding sequence was reconstructed in pUC18 and the resulting ptPAsh, containing 9 nucleotides of 5’-untranslated and 61 nucleotides of 3’untranslated sequences, flanked by two Hind111 restriction sites were obtained. The Hind111 restriction fragment of pt-PAsh, containing the complete coding sequence of t-PA, was subcloned in M13mp18 (13), and template DNA was used for mutagenesis as described (12). The oligodeoxynucleotides used for site-directed mutagenesis are listed in Table Il. Mutants were identified by differential plaque hybridization with “‘P-labeled oligodeoxynucleotides as described (12). The identity was confirmed by nucleotide sequencing. The Hind111 restriction fragment containing the mutant t-PA cDNA was then transferred into the unique Hind111 restriction site of pCM@Neo. The resulting plasmids were introduced in CHO cells by the calcium phosphate co-precipitation method (18) and cells were selected in minimal essential amedium (Gibco) with 400 pg/ml G-418 (Gibco). G-418-resistant colonies were tested for t-PA antigen secretion and a suitable cell line was chosen for large scale production in 850~cm2 roller bottles essentially as described elsewhere (20). Purification of r&PA Moieties from Conditioned Cell Culture Media-The rt-PA moieties characterized in the present study were purified by chromatography on zinc chelate-Sepharose essentially as described (22), followed by immunoadsorption on an insolubilized murine monoclonal antibody against t-PA (MA-lC8) (9). Elution was performed with 1.6 M KSCN and the fractions containing t-PArelated antigen were pooled and dialyzed against 0.3 M NaCl, 0.1 M arginine, 0.02 M Tris-HCl buffer, pH 7.4, containing 0.01% Tween 80 and 10 kallikrein inhibitor units/ml aprotinin. Before use, aprotinin was removed by extensive washing on a Centricon 30 microconcentrator (Amicon) with 0.05 M Tris-HCl buffer, pH 7.4, containing 0.038 M NaCl, 0.1 M arginine, and 0.01% Tween 80. Assay Techniques-Amino acid analysis, after removal of arginine by extensive washing with 10% acetic acid on a Centricon 30 microconcentrator, was performed on a Beckman 119CL amino acid analyzer after 20 h hydrolysis in 6 M HCl at 110 “C in uucuo. Automated NH?-terminal amino acid sequencing was performed at the Department of Biochemistry, University of Vermont, Burlington, VT (courtesy of Dr. K. G. Mann). Total t-PA antigen was measured with a specific ELISA (19) calibrated with rt-PA moieties of which the protein concentration was determined by amino acid analysis. Alternatively, t-PA protein concentration was determined with the Bradford assay (23), calibrated with rt-PA. Specific fibrinolytic activities were determined on bovine fibrin plates (24) by comparison with the 2nd International Reference Preparation for t-PA (25) (code 86/670), obtained from the National Institute for Biological Standards and Control (London, United Kingdom). SDS-PAGE was performed on 10 to 15% gradient gels without reduction or after reduction of the samples with dithioerythritol, using the Phast System’” (Pharmacia). lmmunoblotting of nonreduced SDS-PAGE on nitrocellulose sheets (26) was performed using a polyclonal antiserum against t-PA. Plasminogen activator inhibitor-l antigen was determined with a specific monoclonal antibody-based ELISA (27). Activation of Plasminogen-Activation of native plasminogen (final concentration 25-130 pM) by the single-chain r&PA moieties (final concentration between 25 nM for Activase and 1200 nM for rt-PAA$]“) was measured by incubation at 37 “C in 0.05 M Tris-HCl buffer, pH 7.4, containing 0.038 M NaCl and 0.01% Tween 80. The generated plasmin at different time intervals (O-5 min) was measured TABLE II Oligodeoxynucleotides used for mutagenesis The underlined nucleotides refer to the mutagenized codon; the numbering is according to Pennica et al. (2). Mutagenesis of &PAArg’25,Arg’M,Tyr’65 was performed using the oligodeoxvnucleotides for rt-PAIArg’i” and for rt-PA-Arg’64,Tyr’65. Mutant
Oligodeoxynucleotide
rt-PA-Ars? 574-GCCCGCTGTATCTCTTCTGGGC-553 rt-PA-Argl64,Tyr’GS693-GCAGAACTCGTATCTGTACTTCCC-670 rt-PA-Arg213 837-TGCTGTGTATCTCTTGCCTAT-817 rt-PA-A$” 954-GTACTCCCATCTCAGCCTGCG-934
12186
Kringle
Mutants
with S-2251 (final concentration 1 mM) after 50- to loo-fold dilution of the sample. Initial activation rates were obtained from plots of the concentration of generated plasmin versus incubation time. The kinetic constants (K, and Iz2) were determined by linear regression analysis of the data after Lineweaver-Burk transformation. The effect of fibrin on the activation rate of plasminogen by the rt-PA moieties was evaluated by incubation of plasminogen (final concentration 1 KM) at 37 “C in 0.05 M Tris-HCl buffer, pH 7.4, containing 0.038 M NaCl and 0.01% Tween 80 with CNBr-digested fibrinogen (final concentration 0.1-1.0 wM) prior to addition of the enzyme (final concentration 11 nM for rt-PA-A$“* to 175 nM for rtPA-Arg2i3). Before use, arginine was removed from the r&PA moieties by extensive dialysis. The plasmin generated was measured as described above after 20-fold dilution of the sample, and initial activation rates were expressed as nanomolar/s. Binding to Purified Fibrin-Plasminogen-depleted human fibrinogen (final concentration O-3.2 mg/ml) in 0.05 M Tris-HCl buffer, pH 7.4, containing 0.038 M NaCI, 0.01% Tween 80, and 1 mg/ml bovine serum albumin was mixed with the rt-PA moieties (final concentration 200 rig/ml). The mixture was clotted by addition of thrombin to a final concentration of 20 NIH units/ml. Following a lmin incubation at 37 “C, thrombin was inactivated by the addition of rr-Ile-Pro-Arg-CHsC1 (final concentration 10 NM), and the samples were centrifuged for 1 min at 10,000 X g. The concentration of rt-PA related antigen in the supernatants was determined by ELISA. Binding to Lysine-Se&arose-The different r&PA moieties (final concentration 50 rig/ml), dialyzed extensively to remove arginine, were incubated on a tilting table for 2 h at room temperature in phosphate-buffered saline containing 1 mg/ml albumin and 0.01% Tween 80 with different amounts (O-100 ~1) of a suspension of lysineSepharose (0.3 g wet Sepharose beads/ml) in a total volume of 0.5 ml. After centrifugation, residual rt-PA antigen in the supernatant was measured by ELISA. The affinity of the r&PA moieties for lysine-Sepharose was determined by competition with 6-aminohexanoic acid (6-AHA). Therefore, 6-AHA (final concentration lo-2,500 pM) was incubated with the different rt-PA moieties (final concentration 50 rig/ml) prior to addition of 20 ~1 of 1ysineSepharose suspension. Binding was quantitated by measuring the residual r&PA antigen in the supernatant by ELISA as described above. Fibrin Clot Lysis in a Plasma Milieu-The relative fibrinolytic potency and the fibrin specificity of the rt-PA moieties (final concentration 0.05-3.2 fig/ml) was measured in a system composed of a 0.2ml ‘2”I-fibrin-labeled human plasma clot, suspended in 1 ml of titrated human plasma as described elsewhere (12). In Viva Clot Lysis-The thrombolytic potencies of the rt-PA variants were compared with those of wild-type r&PA and Activasea in hamsters with experimental pulmonary embolism (28). In brief, a 25~1 ‘2SI-fibrin-labeled human plasma clot was injected via the left jugular vein and clot lysis was performed by intravenous infusion over 60 min of plasminogen activator at a dose of 0.1-2 mg/kg, using a constant rate infusion pump. Thirty minutes after the end of the infusion, the extent of clot lysis was determined as the difference between the radioactivity initially incorporated in the clot and that recovered ex uiuo in the heart and lungs. The relative thrombolytic potency of the r&PA variants was expressed in the following way. The mean values of the dose-response data were corrected for background lysis (25%) and the corrected data were fitted with a regression line forced through the origin. The concentration of compound producing 50% lysis of the clot after background correction (i.e. (loo25)/2 + 25 = 62.5%) was determined from the slope of the linear regression line. Blood samples of 0.2 ml were drawn into trisodium citrate (0.011 M final concentration) for measurement of fibrinogen, olz-antiplasmin, and t-PA-related antigen as described elsewhere (28). Pharmacokinetics-The pharmacokinetics of the rt-PA variants were compared with those of wild-type rt-PA and Activasea following bolus injection of 0.25 mg/kg in groups of three hamsters. The initial half-life of t-PA-related antigen in plasma was determined as well as the plasma clearance, using standard pharmacokinetic procedures as outlined elsewhere (28). The clearance (in miIliIiter/min) during the steady state phase accompanying continuous intravenous infusion of the variants was also calculated from the ratio between the infusion rate (in nanograms/min) and the steady state plasma concentration (in nanograms/ml), assuming a body weight of 100 g.
of t-PA RESULTS
Construction and Expression of cDNA Encoding Wild-type and Variant rt-PA-A t-PA cDNA was isolated (16) and inserted into the eukaryotic expression vector pSV328DHFR (12). A stable CHO cell line, expressing rt-PA, was established and used for large scale production as described (20). For easier manipulation of the t-PA cDNA during construction and expression of variant t-PA cDNAs most of the untranslated region of t-PA cDNA was removed as described under “Materials and Methods,” resulting in pt-PAsh comprising nucleotides 76-1831 (2). This shorter cDNA, which is flanked by Hind111 restriction sites, was used for further in vitro mutagenesis and for expression of the variants in pCM@Neo. The following substitutions were performed by in uitro sitedirected mutagenesis: in kringle 1: Proiz5 to Arg (rt-PAArglz5), Ser’64 and Serle5 to Arg and Tyr (rt-PA-Arg”j4,Tyr”j5), or Prolz5 to Arg and Ser’” to Arg and Ser’65 to Tyr (&PAArg’25,Arg164,Tyr’65). Substitutions in kringle 2 included Va1213 to Arg (rt-PA-A&13) or Thr252 to Arg (rt-PA-Arg252). The oligodeoxynucleotides used for these mutations are listed in Table II. pCMpNeo was constructed using pUC18 to obtain a high level expression vector that can be grown in Escherichia coli as a high copy number plasmid (15). This vector drives transcription via the human cytomegalovirus immediate-early promotor, while the selection marker, neomycin phosphotransferase is controlled by the SV40 early promotor. All these elements reside in a pUC environment. The vector has a single Hind111 restriction site for insertion of cDNAs. All mutant t-PA cDNAs were inserted in pCMpNeo and transfected in CHO cells. The transfected cells were selected for resistance against G-418 and stable cell lines, secreting t-PA antigen, were established and scaled up as described elsewhere
cm Purification of Wild-type and Variant rt-PA from Conditioned Cell Culture Media-Chromatography of 15-30 liters of conditioned medium on zinc chelate-Sepharose yielded an approximately loo-fold volume reduction. After immunoadsorption on MA-lC8Sepharose, the yields (mean of two or three preparations) were 500 pg/liter for wild-type &PA, 320 pg/liter for rt-PA-Arglz5, 270 fig/liter for rt-PA-Arg’64,Tyr165, 140 pg/liter for rt-PA-Arg’25,Arg’64,Tyr165, 250 rg/liter for rtPA-Arg213, and 280 pg/liter for rt-PA-Arp2. Characterization of rt-PA Variants-Nearly homogeneous preparations with M, about 70,000 were obtained for all rtPA moieties as shown by nonreduced SDS-PAGE (Fig. lA). The purified proteins were obtained primarily as single-chain forms except for rt-PA-Arg213 which showed significant conversion to a two-chain form, as evidenced by SDS-PAGE after reduction with dithioerythritol (Fig. 1B). All stained bands on nonreduced SDS-PAGE (Fig. lA) react with t-PA antibodies (Fig. 2), but not with antibodies directed against plasminogen activator inhibitor-l (not shown), suggesting that the weaker bands at higher M, probably represent aggregated rtPA. This was further supported by the finding of less than 1% of protein cross-reacting in the plasminogen activator inhibitor-l-specific ELISA. NHz-terminal amino acid sequence analysis on 200 pmol of wild-type rt-PA revealed only one major sequence, with the following yields in pica mole: Ser242-Tyr'~2-Gln106-Va11~2~Ile74-X-Arg46-Asp67-Glu64-Lyszz,
confirming that the translation product of the transfected cDNA is correctly processed by the CHO cells. The amino acid compositions of the rt-PA moieties were compatible with the published amino acid sequence (not shown). The specific fibrinolytic activities measured on fibrin plates were (mean f S.E.; n = 8-11): 180,000-250,000 IU/mg for different batches
Kringle Mutants
of
,
-
34 ,_
12187
123456
A 12
t-PA
5
6
_
76 .-.-
6
7
~.
B 3
4
5
6
FIG. 2. Immunoblotting on a nitrocellulose sheet of SDSPAGE without reduction, using a polyclonal rabbit antiserum raised against t-PA. Lane 1, rt-PA; lane 2, rt-PA-Arg’GI,Tyr’Gi; lane 3, rt-PA-Arg252; lane 4, rt-PA-Arg’““; lane 5, rt-PA-A&“‘; lane 6, rtPA-Arg”S,Arg’G’,Tyr’G”. TABLE III of the actiuation ojplasminogen in the absence of fibrin The correlation coefficients determined by linear sis with six experimental points were 20.99. Kinetic
I
FIG.
‘\
parameters
0
1. SDS-PAGE on lo-15 gradient gels without reduction (A) or after reduction with dithioerythritol (B). Lane I, Activasea; lane 2, rt-PA; lane 3, rt-PA-Arg’64,Tyr’G”; lane 4, rt-PAA@‘; lane 5, rt-PA-Arg”“; lane 6, rt-PA-Arg”“; lane 7, rt-PAArg’2”,Arg’64,Tyr’6”; lane 8, protein calibration mixture consisting of phosphorylase b (M, 97,000), albumin (Mr 67,000), ovalbumin (M, 45,000), carbonic anhydrase (M, 30,000), trypsin inhibitor (M, 20,100), and a-lactalbumin (M, 14,400).
K, rt-PA rt-PA-Arg”” rt-PA-Arg’64,Tyr’G” rt-PA-Arg”S,Arg’64,Tyr’65 rt-PA-A$” rt-PA-A#* Activases
PM 400 800 500 400 330 800 500
k, s-1 0.12 0.92 0.41 0.18 0.007 0.69 0.25
by r&PA
moieties
regression
analy-
k,lK, PM-’
s-’
0.0003 0.0012 0.0008 0.0005 0.00002 0.0009 0.0005
of rt-PA; 280,000 + 20,000 IU/mg for rt-PA-ArglZ5; 180,000 f tration-dependent stimulation of the activation rate of plas14,000 IU/mg for rt-PA-Arg’64,Tyr16”; 87,000 f 4,000 IU/mg minogen by all rt-PA variants (Fig. 3). The activation rate at for rt-PA-Arg’2”,Arg’64,Tyr’65; 28,000 f 4,000 IU/mg for rt- infinite concentration of fibrin is determined from the inverse PA-Arg213;210,000 + 22,000 IU/mg for rt-PA-Arg252, as com- of the ordinate intercept in a double-reciprocal plot of the pared to 460,000 f 77,000 IU/mg for ActivaseR. activation rate uersus the concentration of CNBr-digested Actiuation of Plasminogen-The activation of plasminogen fibrinogen (not shown). Table IV summarizes these data and by all rt-PA moieties obeyed Michaelis-Menten kinetics, as also shows the stimulation factor of CNBr-digested fibrinogen evidenced by linear double-reciprocal plots of the initial ac- on plasminogen activation, obtained as the ratio of the initial tivation rate versus the plasminogen concentration (not activation rate in the presence of infinite fibrin concentration shown). At the final concentrations used, no effect of arginine and the initial activation rate in the absence of fibrin. was observed on the activation rate of plasminogen by rt-PA Binding of rt-PA Variants to Fibrin-All rt-PA moieties or on the hydrolysis of S-2251 by plasmin. The kinetic con- revealed a concentration-dependent binding to fibrin (Fig. 4). stants obtained by linear regression analysis from the LineSpecificity of the binding is evidenced by the complete absence weaver-Burk plots are summarized in Table III. The affinity of binding of single-chain urokinase-type plasminogen actifor plasminogen of all &PA moieties evaluated is low, with vator. Saturation of binding is observed at a comparable fibrin K,,, ranging between 330 PM for rt-PA-Arg213 and 800 PM for concentration for all rt-PA moieties. At a fibrin concentration both rt-PA-Arg2”’ and rt-PA-Arg’25. The cause of the apparent of 1.7 mg/ml, binding was (mean f SD.; n = 3) 90 f 2% for 5-fold higher K,,, for wild-type rt-PA and for Activase”, rela- wild-type rt-PA, 78 + 5% for rt-PA-Arg’25, 77 f 4% for rttive to the results obtained by Hoylaerts et al. (4), is unclear. PA-Arg”j4,Tyr”“, 65 f 9% for rt-PA-Arg’25,Arg164,Tyr’65,72 + The catalytic efficiencies for plasminogen activation (k2/Km) 9% for rt-PA-Arg213, and 88 + 3% for rt-PA-Arg2”‘, as comrange between 0.00030 and 0.0012 PM-’ s-’ for all rt-PA pared to 84 f 3% for ActivaseR. At a fibrin concentration of moieties, except for rt-PA-Arg213, which has a lower catalytic 0.0125 mg/ml, binding was 43 f 1% for rt-PA, 38 f 18% for efficiency (0.00002 PM-’ s-‘) mainly due to its much lower k2. rt-PA-Arg’*“, 33 f 4% for rt-PA-Arg164,Tyr’65, 22 f 0% for rtAddition of CNBr-digested fibrinogen resulted in a concen- PA-Arg’25,Arg’64,Tyr’65, 23 f 6% for rt-PA-Arg?13, and 30 +
Kringle Mutants
of t-PA
1.5
) FIBRIN
1 ( mg /ml )
FIG. 4. Binding of rt-PA moieties to fibrin clots. 0, rt-PA; A, rt-PA-Arg lz5., 0, rt-PA-Arg’G4,Tyr’65; V, rt-PA-Arg’2S,Arg’M,Tyr165; +, rt-PA-Arg213; *, rt-PA-A@*; n , Activasea. The results, expressed as percent r&PA antigen bound to the clot, represent mean values of three separate experiments.
1.0
0.5
0.5 O
I CNBr
1.0
- digested
fibrinogen
1 ( PM )
FIG. 3. Stimulation by CNBr-digested fibrinogen of the activation of plasminogen by rt-PA moieties. The initial activation rate of plasminogen (final concentration 1 wM) by the different rt-PA moieties is plotted against the concentration of CNBr-digested fibrinogen (O-l PM). 0, 35 nM rt-PA; A, 20 nM rt-PA-Arglz5; 0, 20 nM rtPA-Arg”‘,Tyr le5; ‘I, 30 nM rt-PA-Arg’25,Arg’~,Tyr’65; +, 175 nM rtPA-Arg213; *, 11 nM rt-PA-Arg252; W, 20 nM Activases. TABLE
IV
Effect of CNBr-digested fibrinogen (fibrin) on the activation rate of plosminogen by r&PA moieties The initial rate for plasminogen activation was determined in the absence of fibrin and at saturating concentrations of fibrin (m fibrin). The correlation coefficients determined by linear regression analysis on double-reciprocal plots of the initial activation rate uersus the concentration of CNBr-digested fibrinogen were 20.983. Initial rate Enzyme Stimulation No m concentration factor fibrin Fibrin nhf
rt-PA rt-PA-Arg’*’ rt-PA-Arg’64,Tyr’65 rt-PA-Arg’25,Arg’“4,Tyr’s5 rt-PA-Arg13 rt-PA-Arg252 ActivaseR
35 20 20 30 175 11 20
nM.S-l 0.011 0.023 0.015 0.014 0.004 0.009 0.010
1.7 2.3 3.6 4.5 1.3 2.8 3.1
LYSINE
- SEPHAROSE
( ~1)
160 100 240 320 370 310 310
1% for rt-PA-Arg252, as compared to 50 + 1% for ActivaseR. Binding of rt-PA Variants to Lysine-Sepharose-Fig. 5A shows that all &PA moieties bind in a concentration-dependent way to lysine-Sepharose with a maximal binding of 75-90% at a comparable concentration of lysine-Sepharose (100 ~1 of gel suspension). Specificity of the binding is evidented by the absence of binding of as-antiplasmin. Fig. 5B shows the influence of 6-AHA (o-2500 PM) on the binding of a fixed amount of r&PA variant (50 rig/ml) to 20 gl of lysine-Sepharose suspension. The binding of all rt-PA variants to lysine-Sepharose was reduced to 50% at a 6-AHA concentration ranging between 300 and 800 pM.
FIG. 5. Binding of rt-PA variants to lysine-Sepharose. A, binding of rt-PA moieties (50 rig/ml) as a function of the volume of the lysine-Sepharose suspension; B, competition of 6-AHA (lo-2500 pM) for binding of rt-PA moieties (50 rig/ml) to lysine-Sepharose (20~1 suspension). 0, rt-PA; A, rt-PA-Arglz5; 0, rt-PA-Arg’64,Tyr’?
v, ~~~~~~~~~~~~~~~~~~~~~~~~~~ +, rt-p~-~r2~3; *, ,+PA-A+z; n, A c t’IvaseR. The results represent means of two or three separate experiments.
Kringle Mutants Fibrin
Clot Lysis and Fibrinogenolysis
in a Plasma
Milieu-
All r&PA variants tested induced a time- and concentrationdependent lysis of a ‘251-fibrin-labeled human plasma clot immersed in human plasma. Equi-effective concentrations (causing 50% lysis in 2 h) (Fig. 6) were 0.55 rg/ml for wildtype r&PA, 0.17 pg/ml for rt-PA-ArglZ5, 0.58 pg/ml for &PAArg’64,Tyr’65, 0.90 pg/ml for rt-PA-Arg125,Arg’64,Tyr165, 0.74 @g/ml for rt-PA-Arg213, and 0.31 Kg/ml for rt-PA-Arg252, as compared to 0.28 Fg/ml for ActivaseR. At concentrations of plasminogen activator up to 3.2 pg/ml, no fibrinogen breakdown after 2 h was observed for any of the rt-PA variants (not shown). In Vivo Thrombolysis-Dose-response curves following intravenous infusion over 60 min of ActivaseR, wild-type &PA, and the rt-PA variants in hamsters with pulmonary embolism are summarized in Table V. Systemic infusion resulted in a dose-dependent degree of clot lysis over a background value of spontaneous lysis of 25 + 1%. Fifty percent clot lysis over background was obtained with 2.1 mg/kg &PA, 0.9 mg/kg rtPA-ArglZ5, 1.0 mg/kg rt-PA-Arg’64,Tyr’65, 1.8 mg/kg r&PAArg12s,Arg164,Tyr165, 3.5 mg/kg rt-PA-Arg213, 1.2 mg/kg r&PAArg’“‘, and 0.8 mg/kg ActivaseR. None of the agents induced systemic fibrinogen degradation or a*-antiplasmin consumption at the dose which produced 50% clot lysis. Pharmacokinetic Properties-Following bolus injection of 0.25 mg/kg in groups of three hamsters, the initial half-life of t-PA-related antigen in plasma (tlha) ranged between 1.2 min for rt-PA-Arg252 and 2.6 min for rt-PA-Arglz5, as compared to 2.4 min for r&PA and 1.5 min for ActivaseR (Table V). Plasma clearances derived from the rt-PA antigen disappearance after bolus injection ranged between 1.6 ml/min for rt-PA-Arg252 and 4.0 ml/min for rt-PA-Arg213, as compared to 5.1 ml/min for rt-PA and 2.8 ml/min for ActivaseR. Plasma clearances derived from the steady-state t-PA-related antigen levels during the 60-min infusion ranged between 1.8 ml/min for rtPA-Arg”52 and 3.0 ml/min for rt-PA-Arglz5, as compared to 5.9 ml/min for wild-type r&PA and 7.9 ml/min for ActivaseR. DISCUSSION
The fibrin specificity of thrombolysis with tissue-type plasminogen activator (t-PA) is mediated via fibrin stimulation of t-PA-induced plasminogen activation. The isolated protease part of two-chain t-PA is fully active towards low M, substrates and plasminogen, but is not stimulated by fibrin, indicating that the structures involved in fibrin stimulation are comprised within the NH,-terminal region (9,29-31). The
FIG. 6. Lysis of 0.2 ml “‘I-fibrinlabeled human plasma clots immersed in 1 ml of titrated human plasma. 0, rt-PA; A, rt-PA-Arg”“; 0, rt-PA-ArglG4,TyrlG6; V, rt-PA-Arg’*‘, Arg’“,Tyr”“; +, rt-PA-Arg213; *, r&PAArg5’; W, Activasea. The results, plotted as percent clot lysis after 2 h uersus the concentration of plasminogen activator, represent means of three separate experiments.
; Q v) F J g
100
6 a z
5o
of t-PA
12189
NH?-terminal region of the molecule also mediates the highaffinity binding of t-PA to fibrin via structures located in the finger domain and in the second kringle domain (3, 29, 30). The kinetic data of Hoylaerts et al. (4) have suggested that the fibrin stimulation of plasminogen activation by t-PA occurs by sequential ordered addition of t-PA and plasminogen to fibrin, producing a thermodynamically more stable cyclic ternary complex. This would imply that the fibrin affinity of t-PA and the fibrin stimulation of plasminogen activation are causally related, and that both properties would evolve in parallel. However, some recent evidence, obtained with deletion mutants, has suggested that there may be no strict parallelism between fibrin affinity and fibrin stimulation (29, 30, 32). In the present study we have produced variant rt-PAS by in vitro site-specific mutagenesis of homologous amino acids supposed to be involved in the constitution of the high-affinity lysine/fibrin-binding site of the first kringle domain in plasminogen. It was hoped that these mutations would endow the variant rt-PAS with a higher fibrin affinity, increased fibrin stimulation, and enhanced thrombolytic potency. The mutations comprised replacements which reconstituted, in the kringle domains of t-PA, one or both Arg residues believed to form the cationic center of the lysine-binding sites (8), and in addition, to reintroduce a missing aromatic residue in the hydrophobic region of the lysine-binding site in the first kringle of t-PA. Thus five mutant cDNAs were constructed, expressed, and purified to near homogeneity by zinc chelate and immunoadsorption chromatography. The relevant biochemical and biological properties of these variants are summarized in Table VI and are compared with those of commercially available recombinant t-PA (ActivaseR). These data reveal that, when measured in a standard bovine fibrin plate assay, our wild-type rt-PA had a specific activity of 180,000250,000 IU/mg when calibrated against the international reference preparation of t-PA. This is lower than the specific activity of the standard itself (approximately 500,000 IU/mg) (25) and that of ActivaseR. The reason for this discrepancy is not clear; it may relate to differences in the expression system (although both use CHO cells) or in the purification procedures. NH*-terminal amino acid sequence analysis, however, revealed correct processing of the translation product in our expression system. Because of these discrepancies, all t-PA variants were compared both to our wild-type rt-PA and to ActivaseR. The discrepancies between wild-type rt-PA and ActivaseR were less pronounced with respect to catalytic effi-
2 0
0
50
100
200
400
Irt-PA(
800
(rig/ml)
1600
3200
Kringle Mutants
12190
The
Thrombolytic data represent
Agent
Clot
Residual
lysis
fibrinogen
%
wlkg Saline r&PA
rt-PA-Arg””
rt-PA-Arg”‘,TyrrB5
rt-PA-Arg’25,Arg’64,Tyr165
rt-PA-A@*
rt-PA-Arg213
0.14 0.28 0.56 1.10 2.20 0.11 0.22 0.45 0.90 0.15 0.29 0.59 1.20 0.22 0.44 0.88 1.80 0.26 0.52 1.0 2.1 0.56 1.1
2.3 0.13 0.25 0.50 1.00
ActivaseR
TABLE V of rt-PA variants in hamster with pulmonary of experiments indicated between brackets.
and pharmacokinetic properties mean + SE. of the number Dose
25 38 37 64 62 81 40 39 75 82 37 57 82 91 33 57 51 84 32 72 71 72 32 50 64 52 61 78
Residual m-antiplasmin
%
-t k + f + + * + +f
1 3 4 6 6 8 9 5 6 4
f
11 (3)
* + + + f +
7 (3) 1 (3) 2 (2) 5 (3) 10 (3) 9 (3)
(87) (8) (9) (6) (6) (4) (3) (4) (4) (3)
+ 1 (2)
+ + k + + + f + & f
of t-PA
2 (3) 5 (3) 13 (3) 14 (3) 9 (3) 3 (3) 4 (3) 8 (4) 9 (5) 5 (4)
91 + 3 (4)
Antigen
2 IT + f 5 f +
5 (61) 4 (3) 13 (5) 21 (5) 11 (6) 13 (4) 19 (2)
96 130 110 150 140 80 120 160 140 110 120 110 140 110 140 140 150 70
+ 4 (3) 2 9 (2) f 0 (2) f 10 (3) AZ 9 (4) f 23 (2) -c 9 (3) f 21 (3) +- 16 (3) + 4 (2) & 26 (3) * 7 (3) -c 5 (2) f 16 (3) + 10 (3) + 13 (3) -c 17 (2) +- 4 (2)
110 100 100 100
110 -t 3 (2)
94 -+ 8 (3) 110 -+ 6 (4)
f + f +
at end
Clp”
Ma)”
Clpb
ml/min
min
mllmin
5.9 + 0.4
2.4
5.1 + 0.7
3.0 f 0.6
2.6
3.3 * 0.4
2.0 * 0.1
2.0
2.5 + 0.4
2.4 + 0.1
2.2
2.2 k 0.1
2000 k 590 (2) 430 + 110 (3) 540 * 53 (3)
1.8 + 0.2
1.2
1.6 + 0.03
1390
of infusion
%
140 110 110 150 130 130 140
whl
3 (68) 11 (5) 15 (5) 5 (6)
110 f
11 (6)
100 110 110 82 100 160 97 95 83 140 91 89 140 140 160 100 67 130 120 93 120 120 110 93
f f + + ++ f + + + + k -+ * + + f f f -t
6 (4) 15 (3) 2 (3) 8 (4) 16 (3) 12 (3) 9 (3) 2 (3) 32 (2) 28 (3) 8 (3) 7 (3) 9 (2) 19 (3) 4 (3) 3 (3) 4 (3) 21 (3) 13 (3) 12 (3)
f
11 (4)
43 64 150 340 740 79 130 330 320 130 280 400 990 160 350 570 1100 190 540
+ + rt + f + f + + k t2 -t + -t + * & +
5 (5) 12 (6) 13 (6) 76 (5) 110 (4) 9 (3) 20 (4) 36 (4) 120 (3) 53 (2) 70 (3) 13 (2) 120 (2) 20 (3) 31 (2) 100 (3) 290 (2) 2 (2) 120 (2)
1150
f
170 (3)
24 56 140 180
+ 11 (4) + 3 (3) + 3 (4)
f
230 (3)
2.8 f 0.3
1.6
4.0 f 0.1
f k + k
11 (3) 8 (3) 11 (3) 33 (3)
7.9 + 0.8
1.5
2.8 -+ 0.2
“The plasma clearance (Clp) was calculated as the ratio of the infusion rate (ng/min) plasma concentration of antigen (rig/ml) at the end of the infusion, assuming an average The results represent mean + S.E. of the values obtained with the different doses. * Calculated from the pharmacokinetic studies in groups of three hamsters.
The
results
Variant
rt-PA rt-PA-Arglz5 rt-PA-Arg’64,Tyr’65 rt-PA-Arg’25,Arg164,Tyr’65 rt-PA-Arg213 rt-PA-Argz5’ Activases
represent
Comparative mean values
TABLE VI and biological properties
biochemical f S.E.
activity
Iulmg 250,000 2 280,000 + 180,000 * 87,000 k 28,000 -t 210,000 * 460,000 +
of rt-PA Fibrin
W-L
Specific No fibrin
m Fibrin”
0.0003 0.0012 0.0008 0.0005 0.00002 0.0009 0.0005
1.7 mg/ml %
0.05 0.12 0.18 0.15 0.008 0.26 0.16
43 38 33 22 23 30 50
f +f +f +-
1 18 4 0 6 1
f
1
and body
the steady state weight of 100 g.
variants
binding
0.0125 mg/ml
PM--’ s-’
5,000 20,000 14,000 4,000 4,000 22,000 77,000
embolism
90 78 77 65 72 88 84
++f f f +f
2 5 4 9 9 3 3
50% Clot lysis in 2 h
Thrombolytic potency in uiuob
dnl
70 lysisfmg
0.55 0.17 0.58 0.90 0.74 0.31 0.28
30 73 65 34 18 54 79
k + + f k + +
6’ 20d 10 8 4 29 26
a kJK,,, values were obtained by dividing the initial activation rate by the concentration of enzyme and substrate used. These values may be underestimated by a factor that depends on the K,,, in the presence of fibrin, according to the equation: u/[E][S] = kJ(K, + [S]). b Slope of the linear regression line of the dose-response curve in hamsters with pulmonary embolism (percent lysis/mg of compound). ‘p = 0.06 versus Activases. dp = 0.04 versus rt-PA and p = 0.9 versus Activases.
ciency in the absence or presence of fibrin, and with respect to thrombolytic potency in a plasma milieu, whereas the extent of fibrin binding both at low and high fibrin concentrations was identical. The thrombolytic potency of ActivaseR in vivo was somewhat superior to that of our wild-type rt-PA although the difference was not significant (p = 0.06). Compared with wild-type &PA, the three mutants, rt-PAArglz5, rt-PA-Arg’64Tyr’65, and rt-PA-Arg5’, appeared to have a comparable specific activity and a comparable or increased thrombolytic potency in a plasma milieu, although their fibrin
affinity is not significantly different (Table VI). The apparent lack of a strict correlation between specific activity and fibrinolytic potency in a plasma milieu is unclear but may be due to the different dynamics of both systems, such as better diffusion of components in the plasma system. The variants rt-PA-Argiz5, rt-PA-Arg164,Tyr’65, and rt-PAArg25* were somewhat more potent than wild-type rt-PA, but only the difference between &PA and rt-PA-Arglz5 was borderline significant (p = 0.04). None of the variants had a thrombolytic potency higher than that of ActivaseR. These
Kringle Mutants tindings indicate that the mutations introduced in the kringle domains of t-PA have not markedly altered their biochemical properties, nor their in uiuo thrombolytic potency. The relatively poor performance of &PA-A&l3 relative to the other variants might not only be due to the intrinsic properties of the molecule but also, at least in part, to more extensive degradation of the molecule as revealed by SDS-gel electrophoresis. Our approach to reconstitute in the K1 or KB domain of rtPA a lysine-binding site similar to that in K1 of plasminogen thus has apparently not yielded &PA mutants with significantly enhanced fibrin affinity and/or thrombolytic potency. There may be several explanations for this finding. First, the lysine-binding site in K1 of plasminogen has a relatively weak affinity for fibrin, and reconstitution of this structure in the K1 or Kz domain of t-PA may not result in additional fibrin affinity over the intrinsic affinity of the wild-type molecule. Second, mutation of single amino acids in K1 or KP may not have reconstituted the proper conformation of the lysinebinding site. In conclusion, reconstitution of the presumed cationic center in the lysine-binding sites of the kringle domains of t-PA has resulted in a relatively marginal increase in thrombolytic potency. Therefore the potential clinical utility of such mutants is doubtful. REFERENCES 1. Collen, D., Lijnen, H. R., Todd, P. A., and Goa, K. L. (1989) Drugs 38, 346-388 2. Pennica, D., Holmes, W. E., Kohr, W. J., Harkins, R. N., Vehar, G. A., Ward, C. A., Bennett, W. F., Yelverton, E., Seeburg, P. H., Heyneker, H. L., Goeddel, D. V., and Collen, D. (1983) Nature 301,214-221 3. Banyai, L., Varadi, A., and Patthy, L. (1983) FEB.9 Lett 163, 37-41 4. Hoylaerts, M., Rijken, D. C., Lijnen, H. R., and Collen, D. (1982) J. Bid. Chem. 257, 2912-2919 5. van Zonneveld, A. J., Veerman, H., and Pannekoek, H. (1986) J. Biol. Chem. 261,14214-14218 6. Gething, M. J., Adler, B., Boose, J. A., Gerard, R. D., Madison, E. L., McGookey, D., Meidell, R. S., Roman, L. M., and Sambrook, J. (1988) EMBO J. 7, 2731-2740 7. Park, C. H., and Tulinsky, A. (1986) Biochemistry 25,3977-3982
12191
of t-PA
8. Tulinsky, A., Park, 3,85-96 9. Holvoet, P., Lijnen,
C. H., Mao,
B., and Llinas,
H. R., and Collen,
D. (1986)
M. (1988) Eur.
Proteins
J. Biochem.
158,173-177 10. Zamarron, C., Lijnen, H. R., and Collen, D. (1984) J. Biol. Chem. 259,2080-2083 11. Van Heuvel, M., Bosveld, I. J., Mooren, A. A., Trapman, J., and Zwarthoff. E. C. (1986) J. Gen. Viral. 67.2215-2222 12. Nelles, L., Lijnen, H. R.,‘Collen, D., and Holmes, W. E. (1987) J. Biol. Chem. 262,5682-5689 13. Yanisch-Perron. C., Vieira. J., and Messing. -. J. (1985) Gene (Am&.) 33,103-ii9 14. Boshart. M.. Weber. F.. Jahn. G.. Dorsh-Haesler. K.. Fleckenstein, ‘B., and Schaffner, W. (1985) Cell 41, 521-530’ 15. Lijnen, H. R., Nelles, L., Van Hoef, B., De Cock F., and Collen, D. (1990) J. Biol. Chem. 265, 5677-5683 16. Nelles. L.. Liinen, H. R.. Collen. D., and Holmes, W. E. (1987) J. Biol: Chem. 26& 10855-10862 17. Urlaub. G.. and Chasin. L. A. (1980) Proc. N&l. Acad. Sci. U. S. A. 7j, 4216-4220 ’ ’ 18. Graham, F. L., and van der Eb, A. J. (1973) Virology 52, 456467 19. Holvoet, P., Cleemput, H., and Collen, D. (1985) Thromb. Haemost&s 54,684-687 20. Lijnen, H. R., Nelles, L., Holmes, W. E., and Collen, D. (1988) J. Biol. Chem. 263,5594-5598 21. Henikoff, S. (1984) Gene (Am&) 28,351-359 22. Rijken, D. C., and Collen, D. (1981) J. Biol. Chem. 256, 70357041 23. Bradford, M. (1976) Anal. Biochem. 72, 248-254 24. Astrup, T., and Mullertz, S. (1952) Arch. Biochem. Biophys. 40, 346-351 25. Gaffney, P. J., and Curtis, A. D. (1987) Thromb. Haemostasis 58, 1085-1087 26. Towbin, H., Staehelin, T., and Gordon, A. (1979) Proc. N&l. Acad. Sci. U. S. A. 76,4350-4354 27. Declerck, P. J., Alessi, M.-C., Verstreken, M., Kruithof, E. K. O., Juhan-Vague, I., and Collen, D. (1988) Blood 71,220-225 28. Stassen, J. M., Vanlinthout, I., Lijnen, H. R., and Collen, D. (1990) Fibrinolysk, in press 29. Verheijen, J. H., Caspers, M. P. M., Chang, G. T. G., De Munk, G. A. W., Pouwels, P. H., and Enger-Valk, B. E. (1986) EMBO J. 5,3525-3530 30. Van Zonneveld, A. J., Veerman, H., and Pannekoek, H. (1986) Proc. Natl. Acad. Sci. U. S. A. 83. 4670-4674 31. Rijken, D. C., and Emeis, J. J. (1986) Biochem. J. 2313~ 643-646 32. Larsen, G. R., Henson, K., and Blue, Y. (1988) J. Bzol. Chem. 263, 1023-1029