Background of the Invention Human tissue-type plasminogen activator (t-PA), a single chain serine protease of Mr 68,000, is a key physiological regulator of fibrinolysis. It converts the zy ogen plasminogen into plasmin, the enzyme which degrades the fibrin network of the thrombus (Collen (1980) Thromb. Haemosta. 43:77-82: Rijken and Collen (1981) J. Biol. Chem. 256:7035-7041). Apparently, in the presence of a clot, both t-PA and plasminogen bind to fibrin and form a ternary complex in which plasminogen is efficiently activated (Hoylaerts- et al^. (1982) J. Biol. Chem. 257:2912-2919; Ranby (1982) Biochi . Biophvs. Acta 704:461-469). The affinity for fibrin makes t-PA clot-specific, and useful as a therapeutic agent for fibrinolytic therapy in man (Van de Werf et aL. (1984) Circulation 69:605-810). The principal physiological regulator of t-PA appears to be a specific, fast-acting, plasminogen activator inhibitor (PAI-1). PAI-1 is a protein of Mr 50,000 which binds to t-PA in a 1:1 complex, and inactivates the serine protease (Van Mourik et al^ (1984) J. Biol. Chem. 259:14914-14921; Colucci et a (1985) i Clin. Invest. .75:818-824; Aimer and Ohlin (1987) Thromb. Research 47:335-339). A number of recent clinical studies suggest that elevated levels of PAI-1, by reducing the net endogenous fibrinolytic capacity, may contribute to the pathogenesis of various thrombotic disorders, including myocardial infarction, deep vein thrombosis, and disseminated intravascular coagulation (Hamsten et aJL. (1985) New England Journal of Medicine 313:1557-1563; Wi an et a _. (1985) J. Lab. Clin. Med. 105:265-270). These observations suggest that the t-PA:PAI-l interaction may be a suitable target for pharmacologic intervention aimed at enhancing endogenous fibrinolytic activity.

On the other hand, induction of endogenous PAI-1 by administration of endotoxin in rabbits failed to alter thrombolysis by infusion of t-PA. M. Colucci et aL. (1986) LL. Clin. Invest. 78:138-144. This suggests that administration of exogenous PAI-1 would not alter fibrinolysis jjn vi o.

The isolation of human PAI-1 cDNA has previously been reported (Ny et aK, (1986) Proc. Nat! . Acad. Sci . USA 83:6776-6780; Pannekoek et & (1986) EMBO J. 5:2539-2544; Ginsburg et ak. (1986) J. Clin. Invest. 78:1673-1680; un and Kretzmer (1987) FEBS Letters 210:11-16).

Ny et al_j. (supra) disclosed the expression of functional rPAI in E. coli. using a phage lambda gtll-derived vector. rPAI was expressed as a beta-galactosidase-PAI fusion protein of about 180 kDa, with the PAI coding sequence fused to the E. coli beta- galactosidase coding sequence. A PAI-derived fragment of about 40 kDa was also detected by im unoblot and by reverse fibrin autography assay. This 40 kDa form of rPAI was not characterized by Ny et aK., but likely arises by translation initiation at Met/aa #3, since an E. coli translation initiation start signal (AGGA) fortuitously occurs 5' to this Met (see Table 1). Wun and Kretzmer (supra) similarly used a lambda gtll expression vector to express functional rPAI in E. coli. Pannekoek et a . (supra) reported the expression of a functional 43 kDa form of rPAI in g^ coli using a pUC8-derived vector, as assayed by reverse fibrin autography.

Although Ny et aL. (supra) , Pannekoek et a ^ (supra) . and Wun and Kretzmer (supra) report the expression of functional rPAI-1 in E. coli, they do not report the purification of the E. coli- expressed rPAI and do not quantitate the specific activity of the crude preparations of rPAI. It is, therefore, not clear from these references whether the rPAI could be purified from E. coli in a fully functional form, or whether, alternatively, the purified E. coli-expressed PAI would have reduced activity relative to mammalian cell-expressed PAI or would require an activation step to attain full activity. PAI-1 isolated from

many mammalian cell types is obtained in a partially active latent form; the latent form can be converted to a fully active form, with about 6-fold increased specific activity relative to the latent form, by treatment with denaturants such as sodium dodecylsulfate (SDS) (Hek an and Loskutoff (1985) J. Biol. Chem. 260:11581). In fact, a recent report from the Pannekoek lab states that rPAI-1 is expressed in E. coli almost exclusively in an inactive, latent form (Lambers et cT . Fibrinolysis (1988) 2, Supp. 1:33). Apparently the 43 kDa form reported by Pannekoek et < L (supra) is predominantly inactive. Another recent report states that rPAI-1 expressed in E. coli has biological activity toward urokinase almost equal to that of human. fibrosarcoma PAI-1 (Lawrence et al . Fibrinolysis (1988) 2, Supp.1:54).

Brief Summary of the Invention

The invention provides substantially pure, biologically functional, nonfused mature form E. coli-expressed human PAI-1 protein having a specific activity of about 0.3 unit/ng, where a unit is defined as the amount of protein required to neutralize 1 international unit of t-PA in an S2288 chromogenic assay, where the a idolytic activity of t-PA is measured.

The invention also relates to substantially pure, biologically functional, nonfused mature form E. coli-expressed human PAI-1 having a specific activity, as defined above, the biological activity of which is not significantly enhanced by treatment with protein denaturants.

The invention relates further to a recombinant plasmid expression vector capable of expression in E. coli of soluble, biologically functional human PAI-1, wherein the DNA coding sequence for mature PAI-1 is operatively linked to a phage lambda P|_ promoter or to a tac promoter, preferably to a phage lambda, P _| _, promoter, which gives particularly high level expression. Preferred is a plasmid which is or which has the identifying characteristics of pCE1200. The invention also relates to an E coli cell line transformed with such a plasmid, preferably an g

coli cell l ne that is or that has the identifying characteristics of E. coli strain TAP 106. The invention relates further to biologically active recombinant mature human PAI-1 expressed by such a cell line. The invention also provides a method for identifying inhibitors of the binding of t-PA and PAI-1, comprising the steps of:

1) providing in a container, under conditions which permit the binding of t-PA and PAI-1: a) a surface on which is immobilized a known quantity of PAI-1 protein, the PAI-1 protein being attached to the surface by a monoclonal antibody specific for PAI-1, said monoclonal antibody characterized in that it binds to PAI- 1 at a domain not involved in the interaction of t-PA and PAI-1; b) a known quantity of labeled t-PA; and, c) a known quantity of a compound to be evaluated for inhibition of t-PA-PAI-1 binding; and 2) measuring the amount of labeled t-PA bound to PAI-1 and determining the extent of inhibition of t-PA- PAI-1 binding. Preferred is a method for identifying inhibitors of the binding of t-PA and PAI-1 wherein the surface on which the PAI-1 is immobilized is a microtiter plate. Also preferred is a method wherein the labeled t-PA s radiolabeled.

The invention relates further to a pharmaceutical composition comprising, in a pharmaceutically acceptable vehicle, an amount of rPAI-1 of the invention, that is therapeutically effective to counteract excessive or inappropriate fibrinolysis. Also included is a method of treating inappropriate or excessive fibrinolysis in a mammal suffering from such a condition, comprising administering to the mammal an effective dose of I coli-expressed rPAI of the invention.

The invention relates further to a recombinant human PAI-1 peptide having an amino acid sequence corresponding to that of Table 1, with an N-terminal sequence of Met-Ser-Ile-Val-His or Ser-Ile-Val-His, where Val is a ino acid #24 of the PAI-1 precursor protein, and to a recombinant human PAI-1 peptide having an amino acid sequence corresponding to that of Table 1, and having an N-terminal sequence of Met-Val-His or Val-His, where Val is amino acid #24 of the PAI-1 precursor protein.

Brief Description of the Drawing

Figure 1 - Scatchard analysis of binding data for the binding of ¹²⁵ I-t-PA to PAI-1 in the assay of the invention, characterized in that PAI-1 is bound to a solid support using a PAI-1-specific monoclonal antibody. PAI-1-containing wells, prepared as described in Example 5 were incubated with 100 fmoles of -2-*I-t-Pι\ and increasing concentration of unlabeled t-PA (ranging from 10 fmoles to 10 nmoles). The total t-PA bound in each well was determined, and was plotted versus the bound/free ratio. Each point represents the average value of triplicate samples.

Detailed Description of the Invention We have expressed in E. coli two forms of rPAI-1 of about 43 kDa that closely correspond to the two mature forms of PAI-1 (Andreasen et a . (1986) FEBS Letters 209:213-218). It was found that mature form rPAI could be expressed at high levels in £;. coli , i.e., at levels of 10 to 15% of total E. coli protein using a plasmid vector employing the phage lambda P|_ promoter. Use of the PL promoter-containing plasmid resulted in significantly higher levels of expression in E. coli than a plasmid vector employing the "tac" promoter. Whereas the P[_ promoter-containing vector resulted in rPAI accumulation of 10 to 15% of total cell protein, the "tac" promoter-containing expression vector only produced rPAI at levels of about 1% of total cell protein. We have in addition found that E. coli- expressed nonfused mature

form rPAI purified to homogeneity exhibits full biological activity. Specifically, the purified E. coli-expressed rPAI exhibited specific functional activity exceeding that of purified human fibrosarcoma cell PAI by 5 to 10-fold, depending on the assay employed. Moreover, E. coli-expressed purified rPAI did not require an activation step to convert the protein from a less active to a more active form.

Our results contrast with those of previous researchers. Lawrence et a_L (Fibrinolysis (1988) 2, Supp.1:54) report J _. coli-expressed rPAI with activity equal to that of human fibrosarcoma PAI. In contrast, rPAI obtained using our process exhibits functional activity exceeding that of human fibrosarcoma PAI by 5 to 10-fold. Other recent reports (La bers et aL _. (1988) Fibrinolysis 2, Supp.1:33) state that rPAI is obtained from E^ coli almost exclusively in an inactive latent form. As discussed above, rPAI obtained by the methods of the invention exhibits full functional activity that is not enhanced by and does not require any activation procedures.

A further aspect of the invention involves a process for detecting inhibitors of the interaction of PAI and t-PA. This process employs rPAI which is immobilized to a solid support using a PAI-specific monoclonal antibody which is first attached to the solid support. The binding of radio!abeled t-PA to the immobilized PAI is quantitated by separating the bound and free t-PA. Previous assays for the quantisation of PAI and t-PA binding which have used immobilized PAI (PCT WO 88/01273) did not utilize a PAI-specific monoclonal antibody to capture the PAI on the solid support and present it for binding to t-PA. We have found that the use of such a capture antibody markedly improves the sensitivity of the binding assay.

Example 1 Isolation of Human PAI-1 cDNA A human umbilical cord endothelial cell (Gi browne et al . (1974) J. Cell Biol. 60:673) cDNA lambda gtll library was constructed using published methods (Gubler and Hoffman (1983)

Gene 25:263) . Oligonucleotide probes to screen the library were designed based on published PAI-1 sequences (Ny et al (1986) Proc. Natl. Acad. Sci. USA 83:6776-6780; Pannekoek et aL. (1986) EMBO J. 5:2539-2544). A mixture of the following probes was used:

1. 5' GAATTCCTGCAGCTCAGCAGC 3'

2. 5' TGGACCAGCTGACACGGCTGGTGC 3'

3. 5' TCTGAGGCCAGGTGGCCACTAGATGGGGGATGG 3' 4. 5 ¹ AGAGAGGCACCTCTTTTTCATAAGGGGCAGCAA 3'

Approximately 3 X 10^ recombinant lambda plaques were screened for PAI-1 sequences. Four strongly reactive positives and 9 weaker positives were isolated. The insert size, removed from lambda vector by EcoRI digest, of the stronger positives was found to be approximately 2100 to 3100 bp long and was therefore, consistent with the size expected for a PAI-1 cDNA (Ny et al . , supra: Ginsburg et aL. (1986) jL. Clin. Invest. 78:1673-1680). The 2.1 kbp cDNA insert from one of the lambda isolates, designated ECE3-1, was inserted into the EcoRI site of the pTZ19R vector (Pharmacia, Piscataway, NJ) to yield plasmid pECE3-l. Restriction analyses and DNA sequencing of pECE3-l revealed that it contained a human PAI-1 gene cDNA and coded for a 44 kDa protein of identical amino acid sequence to that previously described (Table 1; Ny et ajk., supra: Pannekoek et a_L, supra: Ginsburg et aL _. , supra) . The nucleic acid sequence of PAI-1 clone ECE3-1 only differed from that of Pannekoek et aL. within the untranslated region of the cDNA (Table 2). A translation stop signal (TGA) terminates translation with amino acid #402, proline.

Table 1

PAI-1 cDNA Clone ECE3-1 DNA Sequence and Encoded Amino Acid Sequence:

EcoRI G IAATTCCTGCAGCTCAGCAGCCGCCGCCAGAGCAGGACGAACCGCCAATCGCAAGGCACC

1 +—, + + + -+ + 60

CTTAAGGACGTCGAGTCGTCGGCGGCGGTCTCGTCCTGCTTGGCGGTTAGCGTTCCG TGG

TCTGAGAACTTCAGGATGCAGATGTCTCCAGCCCTCACCTGCCTAGTCCTGGGCCTG GCC

61 + + --+-— +---- + --+ 120

AGACTCTTGAAGTCCTACGTCTACAGAGGTCGGGAGTGGACGGATCAGGACCCGGAC CGG

M Q M S P A L T C L V L G L A 1 3

ApaLl CTTGTCTTTGGTGAAGGGTCTGCTG ITGCACCATCCCCCATCCTACGTGGCCCACCTGGCC

121 + —+ + + + —+ 180

GAACAGAAACCACTTCCCAGACGACACGTGGTAGGGGGTAGGATGCACCGGGTGGAC CGG

Table 1 (continued)

L V F G E G S A V H H P P S Y V A H L A

22 24 TCAGACTTCGGGGTGAGGGTGTTTCAGCAGGTGGCGCAGGCCTCCAAGGACCGCAACGTG 181 + + + + —+ + 240

AGTCTGAAGCCCCACTCCCACAAAGTCGTCCACCGCGTCCGGAGGTTCCTGGCGTTG CAC

S D F G V R V F Q Q V A Q A S K D R N V

GTTTTCTCACCCTATGGGGTGGCCTCGGTGTTGGCCATGCTCCAGCTGACAACAGGA GGA 241 + + + + + —+ 300

CAAAAGAGTGGGATACCCCACCGGAGCCACAACCGGTACGAGGTCGACTGTTGTCCT CCT

S

V F S P Y G V A S V L A M L Q L T T G G

GAAACCCAGCAGCAGATTCAAGCAGCTATGGGATTCAAGATTGATGACAAGGGCATG GCC 301 +-— + + + —+ + 360

CTTTGGGTCGTCGTCTAAGTTCGTCGATACCCTAAGTTCTAACTACTGTTCCCGTAC CGG E T Q Q Q I Q A A M G F K I D D K G M A

Table 1 (continued)

Rsal CCCGCCCTCCGGCATCTGT IACAAGGAGCTCATGGGGCCATGGAACAAGGATGAGATCAGC

361 + + + + + + 420

GGGCGGGAGGCCGTAGACATGTTCCTCGAGTACCCCGGTACCTTGTTCCTACTCTAG TCG

P A L R H L Y K E L M G P W N K D E I S

ACCACAGACGCGATCTTCGTCCAGCGGGATCTGAAGCTGGTCCAGGGCTTCATGCCC CAC 421 + + + + + + 480

TGGTGTCTGCGCTAGAAGCAGGTCGCCCTAGACTTCGACCAGGTCCCGAAGTACGGG GTG

T T D A I F V Q R D L K L V Q G F M P H

TTCTTCAGGCTGTTCCGGAGCACGGTCAAGCAAGTGGACTTTTCAGAGGTGGAGAGA GCC

481 + + + + + + 540

AAGAAGTCCGACAAGGCCTCGTGCCAGTTCGTTCACCTGAAAAGTCTCCACCTCTCT CGG

F F R L F R S T V K Q V D F S E V E R A

Table 1 (continued)

AGATTCATCATCAATGACTGGGTGAAGACACACACAAAAGGTATGATCAGCAACTTG CTT 541 + + + + + —+ 600

TCTAAGTAGTAGTTACTGACCCACTTCTGTGTGTGTTTTCCATACTAGTCGTTGAAC GAA

R F I I N D W V K T H T K G M I S N L L

GGGAAAGGAGCCGTGGACCAGCTGACACGGCTGGTGCTGGTGAATGCCCTCTACTTC AAC 601 — —+-- + + + -+ + 660

CCCTTTCCTCGGCACCTGGTCGACTGTGCCGACCACGACCACTTACGGGAGATGAAG TTG

G K G A V D Q L T R L V L V N A L Y F N

GGCCAGTGGAAGACTCCCTTCCCCGACTCCAGCACCCACCGCCGCCTCTTCCACAAA TCA 661 —- + +— + + + + 720

CCGGTCACCTTCTGAGGGAAGGGGCTGAGGTCGTGGGTGGCGGCGGAGAAGGTGTTT AGT

G Q K T P F P D S S T H R R L F H K S

GACGGCAGCACTGTCTCTGTGCCCATGATGGCTCAGACCAACAAGTTCAACTATACT GAG 721 + + + —+ + + 780

CTGCCGTCGTGACAGAGACACGGGTACTACCGAGTCTGGTTGTTCAAGTTGATATGA CTC

Table 1 (continued)

D G S T V S V P M M A Q T N K F N Y T E

TTCACCACGCCCGATGGCCATTACTACGACATCCTGGAACTGCCCTACCACGGGGAC ACC 781 - + + +- + + + 840

AAGTGGTGCGGGCTACCGGTAATGATGCTGTAGGACCTTGACGGGATGGTGCCCCTG TGG

F T T P D G H Y Y D I L E L P Y H G D T CTCAGCATGTTCATTGCTGCCCCTTATGAAAAAGAGGTGCCTCTCTCTGCCCTCACCAAC

&m + —+ +- + + + 900

GAGTCGTACAAGTAACGACGGGGAATACTTTTTCTCCACGGAGAGAGACGGGAGTGG TTG

L S M F I A A P Y E K E V P L S A L T N

ATTCTGAGTGCCCAGCTCATCAGCCACTGGAAAGGCAACATGACCAGGCTGCCCCGC CTC 901 - + + --+ + —+ + 960

TAAGACTCACGGGTCGAGTAGTCGGTGACCTTTCCGTTGTACTGGTCCGACGGGGCG GAG

I L S A Q L I S H W G N M T R L P R L

Table 1 (continued)

CTGGTTCTGCCCAAGTTCTCCCTGGAGACTGAAGTCGACCTCAGGAAGCCCCTAGAG AAC

961 +— + —+ + + + 1020

GACCAAGACGGGTTCAAGAGGGACCTCTGACTTCAGCTGGAGTCCTTCGGGGATCTC TTG

L V L P K F S L E T E V D L R K P L E N

CTGGGAATGACCGACATGTTCAGACAGTTTCAGGCTGACTTCACGAGTCTTTCAGAC CAA

1021 + + —+ —+ -+— + 1080

GACCCTTACTGGCTGTACAAGTCTGTCAAAGTCCGACTGAAGTGCTCAGAAAGTCTG GTT

L G M T D M F R Q F Q A D F T S L S D Q

GAGCCTCTCCACGTCGCGCAGGCGCTGCAGAAAGTGAAGATCGAGGTGAACGAGAGT GGC

1081 + + + +— -+- + 1140

CTCGGAGAGGTGCAGCGCGTCCGCGACGTCTTTCACTTCTAGCTCCACTTGCTCTCA CCG

E P L H V A Q A L Q K V K I E V N E S G

Table 1 (continued)

ACGGTGGCCTCCTCATCCACAGCTGTCATAGTCTCAGCCCGCATGGCCCCCGAGGAG ATC

1141 + + + + + + 1200

TGCCACCGGAGGAGTAGGTGTCGACAGTATCAGAGTCGGGCGTACCGGGGGCTCCTC TAG

T V A S S S T A V I V S A R M A P E E I

ATCATGGACAGACCCTTCCTCTTTGTGGTCCGGCACAACCCCACAGGAACAGTCCTT TTC

1201 + + + +-— + + 1260

TAGTACCTGTCTGGGAAGGAGAAACACCAGGCCGTGTTGGGGTGTCCTTGTCAGGAA AAG

1 M D R P F L F V V R H N P T G T V L F

ATGGGCCAAGTGATGGAACCCTGACCCTGGGGAAAGACGCCTTCATCTGGGACAAAA CTG

1261 + + + + +— + 1320

TACCCGGTTCACTACCTTGGGACTGGGACCCCTTTCTGCGGAAGTAGACCCTGTTTT GAC

M G Q V M E P * 402

Table 1 (continued)

Nsil GAGATGCA ITCGGGAAAGAAGAAACTCCGAAGAAAAGAATTTTAGTGTTAATGACTCTTTC

1321 + + + +— +- + 1380

CTCTACGTAGCCCTTTCTTCTTTGAGGCTTCTTTTCTTAAAATCACAATTACTGAGA AAG

Bglll TGAAGGAAGAGAAGACATTTGCCTTTTGTTAAAAGATGGTAAACCA IGATCTGTCTCCAAG 1381 + +— + + + + 1440

ACTTCCTTCTCTTCTGTAAACGGAAAACAATTTTCTACCATTTGGTCTAGACAGAGG TTC

ACCTTGGCCTCTCCTTGGAGGACCTTTAGGTCAAACTCCCTAGTCTCCACCTGAGAC CCT

1441 —- + + + + + + 1500

TGGAACCGGAGAGGAACCTCCTGGAAATCCAGTTTGAGGGATCAGAGGTGGACTCTG GGA

GGGAGAGAAGTTTGAAGCACAACTCCCTTAAGGTCTCCAAACCAGACGGTGACGCCT GCG

1501 + + + + +— + 1560

CCCTCTCTTCAAACTTCGTGTTGAGGGAATTCCAGAGGTTTGGTCTGCCACTGCGGA CGC

Table 1 (continued)

GGACCATCTGGGGCACCTGCTTCCACCCGTCTCTCTGCCCACTCGGGTCTGCAGACC TGG

1561 + + + + +- + 1620

CCTGGTAGACCCCGTGGACGAAGGTGGGCAGAGAGACGGGTGAGCCCAGACGTCTGG ACC

TTCCCACTGAGGCCCTTTGCAGGATGGAACTACGGGGCTTACAGGAGCTTTTGTGTG CCT

1621 + + + + +— + 1680

AAGGGTGACTCCGGGAAACGTCCTACCTTGATGCCCCGAATGTCCTCGAAAACACAC GGA

Rsal i GGTAGAAACTATTTCTGTTCCAGCTCACATTGCATCACTCTTGTACTGCCTGCCACCGCG

1681 + + + + + + 1740

CCATCTTTGATAAAGACAAGGTCGAGTGTAACGTAGTGAGAACATGACGGACGGTGG CGC

GAGGAGGCTGGTGACAGGCCAAAGCGAGTGGAAGAAACACCCTTTCATCTCAGAGTC CAC

1741 + + + + +- + 1800

CTCCTCCGACCACTGTCCGGTTTCGCTCACCTTCTTTGTGGGAAAGTAGAGTCTCAG GTG

Table 1 (continued)

Rsal TGTGGCACTGGCCACCCCTCCCCAGT IACAGGGGTGCTGCAGGTGGCAGAGTGAATGTCCC

1801 -+ +— + + + + 1860

ACACCGTGACCGGTGGGGAGGGGTCATGTCCCCACGACGTCCACCGTCTCACTTACA GGG

CCATCATGTGGCCCAACTCTCCTGGCCTGGCCATCTCCCTCCCCAGAAACAGTGTGC ATG

1861 + + +— + +— + 1920

GGTAGTACACCGGGTTGAGAGGACCGGACCGGTAGAGGGAGGGGTCTTTGTCACACG TAC

GGTTATTTTGGAGTGTAGGTGACTTGTTTACTCATTGAAGCAGATTTCTGCTTCCTT TTA

1921 + —+- -+ + + + 1980

CCAATAAAACCTCACATCCACTGAACAAATGAGTAACTTCGTCTAAAGACGAAGGAA AAT

TTTTTATAGGAATAGAGGAAGAAATGTCAGATGCGTGCCCAGCTCTTCACCCCCCAA TCT

1981 —+ --+ + + +— + 2040

AAAAATATCCTTATCTCCTTCTTTACAGTCTACGCACGGGTCGAGAAGTGGGGGGTT AGA

Table 1 (continued)

Rsal CTTGGTGGGGAGGGGTGT IACCTAAATATTTATCATATCCTTGCCCTTGAGTGCTTGTTAG

2041 + + + + + + 2100

GAACCACCCCTCCCCACATGGATTTATAAATAGTATAGGAACGGGAACTCACGAACA ATC

EcoRI AGAGAAAGAGAACTACTAAGGAAAAGG IAATTC

2101 —— + +— +-- 2132

TCTCTTTCTCTTGATGATTCCTTTTCCTTAAG

Table 2

Comparison of the Sequence of PAI 3' Untranslated Region in PAI-1 cDNA Clone ECE3-1 (upper row) and that of Pannekoek et aL (EMBO J. (1986) 5:2537) (lower row):

1667 GCTTTTGTGTGCCTGGTAGAAACTATTTCTGTTCCAGTCACATTGCCATC 1716

1751 G 1C1T1T1T1T1G1T1G1TiG1C1C1T1G1G1T1A1G1AiA1A1C1T1A1T1T1TfCITEG 1T1T1CiCfA1G1CTCACATITGCMATMC 1800 1717 ACTCTTGTACTGCCTGCCACCGCGGAGGAGGCTGGTGACAGGCCAAAGGC 1766

1801 A MCTMCTMTGMTAMCTMGCMCTMGCMCAMCCMGCMGGMAGMGAMGGMCTMGGITMGAMCAM GGMCCMAAMA.GMC 1849 1767 CAGTGGAAGAAACACCCTTTCATCTCAGAGTCCACTGTGGCACTGGCCAC 1816

1850 GA MGTMGGMAAMGAMAAMCAMCCMCTMTTMCAMTCMTCMAGMAGMTCMCAMCTMGTMGGMCA MCTMGGMCCMAC! 1899

Example 2 Expression of Recombinant PAI-1 in Bacteria

Expression of Full Length PAI-1 in E.coli Immunoblots and functional assay by reverse fibrin autography (Loskutoff et aL. (1983) Proc. Nat! . Acad. Sci . USA 80:2956) of lysates of heat-induced lambda gtll lysogens demonstrated the presence of PAI-1. Functional PAI-1 expressed using the lambda gtll vector had an estimated molecular weight of approximately 50 kDa on SDS-polyacrylamide gel electrophoresis (SDS-PAGE) gels.

The 2.1 kbp PAI-1-containing EcoRI to EcoRI fragment was inserted at the EcoRI site and 3' to the trfi promoter in a plasmid derivative of pKGP36-trp (Ivanoff et al^ (1986) Proc. Nat! . Acad. Sci. USA 83.:5392-5396; ATCC No. 39413) which contains a unique EcoRI site 3' to the trfi promoter. The resultant plasmid, which contains the PAI coding sequence inserted in the correct orientation with respect to the trβ promoter, is designated pPAI2.1. E. coli containing pPAI2.1 expressed an approximately 50 kDa form of rPAI-1 which was detected by reverse fibrin autography assay (Loskutoff et aL.. supra) . However, rPAI-1 protein levels produced in E. coli containing pPAI2.1 were not detectable (estimated to be less than 0.1% of total E. coli protein) by Coomassie blue staining of SDS-PAGE gels. A fortuitous E. coli translation start signal (AGGA) 5' to the start of the PAI coding sequence is likely the start signal of translation, with the second ATG (Met aa #3; Table 1) in the coding sequence acting as the start site of translation. Thus, the approximately 50 kDa rPAI-1 product expressed by pPAI2.1 appears to closely correspond to full length pro-form PAI-1, only differing by the deletion of aa residues #1 (Met) and #2 (Gin)

(see Table 1). This form of PAI-1 is not mature form PAI-1 since it contains most of the leader sequence.

Expression of Mature PAI-1 in E. coli

The N-terminal leader peptide of the PAI-1 coding region was removed by a partial ApaLI and Nsil digest of Haelll methylase treated pECE3-l DNA. The approximately 1800 bp fragment containing the PAI coding sequence was modified by adding a synthetic DNA adaptor to the 5' end of the fragment as shown below:

Synthetic Adaptor PAI-1 Gene Coding Seguence

Bam Eco Nco ApaL I I I

5' GGATCCGAATTCC AJG G TGCAC CAT

3' CCTAGGCTTAAGG TAC CACGT G GTA ...

Met Val His His ...

This fragment, following digestion with the restriction enzyme Ncol, was inserted at the Ncol and Pstl sites within the polylinker of the E. coli expression vector pKK233-2 (Amann and Brosius (1985) Gene 40:183; Straus and Gilbert (1985) Proc. Natl. Acad. Sci. USA 82:2014), to yield plasmid ptac-PAI. In ptac-PAI mature form rPAI-1 is expressed in E. coli under control of the strong "tac" promoter (Amann and Brosius (1985) Gene

40:183). E. coli strains HB101 or JM109 ( crB ^" ) containing ptac- PAI produced functional mature PAI-1 (i.e., PAI-1 lacking N- ter inal leader peptide that is normally removed during translocation in the eukaryotic cell) as determined by SDS-PAGE, reverse fibrin autography, amidolytic assay, t-PA binding activity, and immunoblot analysis (see below for details of these assays). In either host, cells containing this plasmid produced recombinant mature form PAI-1 at a level of approximately 1% of total E. coli protein, as determined by Coo assie brilliant blue staining of SDS-PAGE gels. The apparent molecule weight on SDS- PAGE gels was 45 kDa. The PAI-derived product of ptac-PAI corresponds to one of the two forms of mature PAI-1 (the two- residue shorter form) found in mammalian cells. In mammalian cells, two forms of mature PAI-1 are found in approximately equal amounts: a form with an N-terminus of Ser-Ala-Val-His-His and a

two-residue shorter form with N-terminus Val-His-His (Andreasen et aL. (1986) FEBS Letters 209: 213-218) (see Table 1). Thus, the two forms of mature PAI-1 begin with aa #22 (Ser) and #24 (Val) (see Table 1). The leader sequence, amino acids #1-21 or #1-23, is removed in mammalian cells to yield the mature form PAI-1.

We also constructed a plasmid pCE1200 designed to express in E. coli the mature form of PAI with the N-terminus Ser-Ala-Val- His-His. The construction of plasmid pCE1200 is described below. The 2.1 kbp PAI EcoRI to EcoRI fragment, isolated from the plasmid pECE3-l, was digested with restriction enzyme ApaLI and the ends were filled-in with DNA polymerase I Klenow fragment and the four deoxyribonucleotides. This DNA was then digested with Bglll and the 1206 bp fragment containing a 5 ^l filled-in ApaLI site (blunt end) was recovered. The plasmid expression vector pCE31 (described below) was digested with Clal and the overhang was filled-in with DNA polymerase I Klenow fragment and the four deoxyribonucleotides. The vector was digested with BamHI and the resulting large fragment containing the lambda P|_ promoter was purified from an agarose gel. The 1206 bp PAI DNA fragment was then ligated into pCE31 at the filled-in Clal and BamHI sites, to yield plasmid pCE1200.

The nucleotide sequence of pCE1200 encoding the N- and C- terminal amino acid sequence of the PAI-1-derived protein was determined and is shown below.

5' ATG AGT ATC GTG CCC TAG

M S I V P END

#22 #402

The predicted molecular weight of the protein expressed from pCE1200 is approximately 42 kDa based on an average of 110 Da/amino acid. The product expressed by pCE1200 differs from the longer mature form of PAI by the conservative substitution of Ala residue #23 with He.

The expression of rPAI-1 encoded by pCE1200 is under the control of the phage lambda PL promoter. The plasmid pCE1200 in E. coli host TAP106 (deposited with the American Type Culture Collection, Rockville, MD; ATCC accession number 67911) contains a defective lambda prophage including a temperature sensitive, mutant repressor gene (cI857) . At low temperatures (32°C) , the mutant repressor is active and transcription from P|_ is shut off; however, when the temperature is raised to 42°C the repressor is inactivated and transcription initiating at the P|_ promoter can proceed (Reznikoff and Gold, Maximizing Gene Expression,

Butterworths, 1986). The genotype of the TAP106 host is leu ^" , bio ^" , gal K (am), bl ^" , ΔlacU.169, rpsL, Sup E44, λdef [ΔBamHI, N ^" , Kan ^r , CI857, Δbio].

Production of PAI-1 involves growing the culture under conditions at which the repressor is active, preferably at temperature of 32°C. The culture may then be induced to produce PAI-1 by changing to conditions which inactivate the repressor and allow expression from the Pj_ promoter, e.g., raising the temperature to 42°C. Other conditions can be used to inactivate the repressor, e.g., adding nalidixie acid or mitomycin C.

Typical large scale induction conditions require growth at 32°C in LB media supplemented with ampicillin (100 g/mL) and glucose (3%) until mid-log growth phase and then the culture is induced by raising the temperature of the fermenter to 42°C (Sisk et aL. (1986) Gene 48:183-193). The culture is grown at the elevated temperature for 4 to 8 hours, after which time the cells are harvested by centrifugation and frozen at -70°C prior to PAI-1 purification. The yields of recombinant PAI-1 in E. coli using the CE1200 expression vector are about 10 to 15% of total E. coli protein, as estimated by Coomassie blue staining of SDS- PAGE gels. As defined herein, "high level" expression of recombinant PAI-1 means yields of at least 5%, and preferably 10% or greater. The E. coli-expressed PAI-1 protein of the invention was not only produced at high levels, but was also in a soluble, easily purified form.

Plasmid pCE1200 in E. coli K12, TAP 106 (host) was deposited on March 15, 1989 in the American Type Culture Collection (ATCC) , Rockville, MD, in accordance with the provisions of MPEP 608.01 (p) (C) (1) (2) and (3) and the Budapest Treaty. The ATCC accession number is 67911. Access to the culture will be available during pendency of the patent application to one determined by the Commissioner to be entitled thereto under 37 CFR 1.14 and 35 USC 122. Upon granting of a patent all restrictions on the availability of the culture to the public will be irrevocably removed.

Construction of Expression Vector pCE31 Plasmid pCE31 was derived from plasmid pJL6, which is described by Lautenberoer et al . (Gene 23:75-84. 1983). Plasmid pJL6 was digested with restriction enzymes BstXI and Clal and a synthetic DNA sequence was inserted containing a consensus ribosome binding site (SD) and an ATG initiation codon. The sequence of the synthetic oligonucleotide used was as follows:

BstXI Clal SD M S

5' CCAACCTCTGGACATTGCAAGGAGTTTATAAATG AGT ATC GAT 3' 3' GGTTGGAGACCTGTAACGTTCCTCAAATATTTAC TCA TAG CTA 5'

As shown above, there is a three amino acid-encoding sequence 5' to the Clal site (Met-Ser-Ile). Plasmid pCE31 provides the Clal site 3' to a translation start site allowing DNA coding sequences to be inserted and operatively linked to the phage lambda (P[_) promoter.

Example 3

Purification of PAI Expressed in E. coli E. coli cells containing the PAI protein were suspended in ten volumes of 50 mM Na phosphate pH 6 and disrupted using a sonicator (Heat Systems Ultrasonics). The sonication was carried out for about 5 min using a medium tip probe. The sample was

kept cold in an ice bath during the sonication. The sonicated sample was then centrifuged at 10,000 rpm for about 20 min and the supernatant saved. The pellet obtained was resuspended in 5 volumes of 50 mM Na phosphate buffer and resonicated using similar conditions as for the first sonication. The material obtained after the second sonication was centrifuged at 10,000 rpm as before. The supernatants obtained after the two centrifugation steps were pooled, filtered through a 5 micron filter and pumped onto a 4.4 x 30 cm Q Sepharose fast flow (Pharmacia) column, equilibrated with 50 mM Na phosphate pH 6.0, at a flow rate of about 10 mL/min. The absorbance of the effluent was monitored at 280 nm, the column was then washed with 50 mM Na phosphate buffer until there was very little 280 nm absorbing material in the effluent (about 0.5 column volumes). The effluent from the Q Sepharose column was loaded onto a S Sepharose column equilibrated with 50 mM Na phosphate pH 6 at about 10 mL/min. The absorbance of the effluent was monitored at 280 nm. The column was washed at 10 mL/min with 50 mM Na phosphate buffer until the 280 nm absorbance was close to the baseline obtained with buffer alone (about 1 column volume). The protein was then eluted from the column at the same flow rate using a 0 to 1 M NaCl gradient in 50 M Na phosphate pH 6. The fractions were assayed for PAI using sodium dodecyl sulfate polyacrylamide gel electrophoresis (SDS-PAGE) and analytical HPLC, using the methods outlined below. The PAI was found to be about 85 to 90% pure by SDS-PAGE and by analytical HPLC. The fractions containing PAI were pooled, diluted about 2.5 fold with water and loaded onto another 4.4 x 30 cm S Sepharose column equilibrated with 50 mM Na phosphate buffer pH 6. The column was washed with about 2 column volumes each of starting buffer and 50 mM Na phosphate buffer pH 8.0 successively. The PAI was eluted using 150 mM Na phosphate pH 8.6. The PAI obtained at this stage was found to be about 98% pure by SDS-PAGE and by analytical HPLC. PAI of greater than about 95% purity is considered "substantially pure".

Other standard protein purification procedures may be used, as known to those skilled in the art. Generally speaking, the cells are disrupted, preferably by sonication. The soluble supernatant is then isolated, and the PAI-1 is purified from the soluble supernatant using column chromatography. In the method described above, Q Sepharose is an anion exchange column; S Sepharose is a cation exchange column. The described purification procedure takes advantage of the finding that rPAI-1 does not bind to the anion exchange column and does bind to the cation exchange column. Those of skill in the art can devise other purification methods without undue experimentation. The fractions were assayed for PAI by SDS-PAGE and by analytical reverse phase HPLC. The SDS-PAGE was carried out using the Pharmacia Phast system (Pharmacia, Piscataway, NJ) utilizing the procedures outlined in the users manual. An 8 to 25% polyacrylamide gradient gel was used and the proteins were visualized by staining with Coomassie blue. Analytical HPLC was carried out using an HP 1090 M HPLC system. A 4.6 x 50 mm C4 Vydac column was used. The column was equil brated in 0.1% TFA and the samples were applied to the column in the same solvent.

The samples were eluted using an acetonitrile gradient. Purified PAI was used to establish the retention time under the conditions of the assay.

The N-terminal sequence of purified E. coli-expressed rPAI-1 expressed from plasmid pCE1200 was determined to begin Ser-Ile- Val-His-His. Thus, the N-terminal Met residue of the nascent protein appears to be efficiently removed by amino peptidases in E. coli.

It will be recognized that short form mature PAI-1 (aa #24-402) or long form mature PAI-1 (aa#22-402) described above may be produced in the same or similar manner by introducing the desired modifications in the PAI-1 DNA coding sequence, as is known to those skilled in the art. Mature PAI-1 as defined herein includes both long form and short form proteins, from which the N-terminal leader sequence has been removed. Also included are mature PAI-1 proteins having substantially the same

amino acid sequence and substantially the same activity.

Example 4 Functional Assays for PAI-1 The relative activities of purified recombinant E. coli- expressed human rPAI-1 products of both pCE1200 and ptac-PAI (98% pure, as described in Example 3), and purified nonrecombinant human PAI-1 expressed in a human fibrosarcoma cell line, designated hPAI-1, were compared using the assays described below. Purified hPAI-1 was obtained from American Diagnostica Inc. (New York, NY), with a purity of about 98%, as estimated by SDS-PAGE and Coomassie staining of SDS-PAGE gels.

The activities of the purified recombinant and nonrecombinant PAI-1 preparations were compared in three different assays: 1) The S2251 assay measures the enzymatic ability of tissue plasminogen activator (t-PA) to generate plasmin from its plasminogen precursor by determining the level of amidolytic activity of generated plasmin on the chromogenic substrate D-Val-Leu-Lys-p-nitroanilide (S2251); inhibition of the activity by PAI-1 is determined; 2) The S2288 assay measures the amidolytic activity of t-PA on the chromogenic substrate D-Ile-Pro-Arg-NH-nitroanilide (S-2288) and inhibition of this activity by PAI-1 is determined;

3) The binding of iodinated t-PA ( ¹²⁵ I-t-PA) by PAI-1 captured onto microtiter wells with a specific anti-PAI-1 monoclonal antibody.

The rPAI-1 products of both pCE1200 and ptac-PAI were purified, as detailed in Example 3, and shown to have similar specific activities in the assays here described.

In the S2251 assay, 50% inhibition of 10 international units (IU) of t-PA activity was observed with 2.5 x 10~ ⁴ mg/mL of hPAI-1, and with 5.3 x 10~ ⁵ mg/ml of rPAI-1. The activity of t-PA is expressed in IU by comparison with the International Reference Preparation for t-PA. Accounting for an assay volume of 100 μλ , it is calculated that 25 ng of hPAI-1 reduces the

activity of 10 IU of t-PA by one-half. Defining a unit of PAI-1 as the amount of protein needed to neutralize 1 IU of t-PA, we determined the specific activity for the hPAI-1, in terms of units/ng, of 5/25 or 0.2 units/ng. A similar calculation with the rPAI-1 yields a specific activity of approximately 1 unit/ng. The rPAI-1 product of either pCE1200 or ptac-PAI was, therefore, approximately 5-fold more active than the hPAI-1 in this particular assay. The substantially pure E. coli expressed rPAI- 1 of the invention will have a specific activity in the S2251 chromogenic assay of about lu/ng. Those skilled in the art will recognize that some variation in specific activity, e.g., in the range of +10%, is to be expected.

In the S2288 assay, 50% inhibition of 10 IU of t-PA activity was observed with 1.6 x 10 ^"3 mg/ml of hPAI-1, and with 1.7 x 10 ^"4 mg/ml of rPAI-1. These values yield specific activities of 0.03 and 0.3 units/ng for the hPAI-1 and rPAI-1, respectively. The rPAI-1 is therefore approximately 10-fold more active than the hPAI-1 in this assay. The lower specific activities for the PAI-1 in the S-2288 assay, in comparison with the S-2251 assay, most likely reflect the fact that t-PA's activity in the S-2251 assay is dependent upon and greatly enhanced by fibrin.

In the t-PA binding assay (see below, Example 5), 0.002 mg/ml of rPAI-1 bound approximately 6-fold more iodinated t-PA than did 0.002 mg of hPAI-1. Thus, again, rPAI-1 exhibited a significantly greater specific activity than did hPAI-1.

The PAI-1 isolated from many cell types is obtained in a partially inactive or latent form; the activity can be induced or increased by treatment with protein denaturants such as sodium dodecylsulfate (SDS) (Hekman and Loskutoff (1985) J. Biol. Chem. 260:11581). In the S2251 chromogenic assay, the activity of hPAI-1 from human fibrosarcoma cells was noted to be increased approximately 5.6-fold following an activation procedure which involved incubation of the protein for 1 h at 37° in the presence of 0.1% SDS, and dialysis of the material overnight against 150 mM Na2HP0 . This result indicates that the hPAI-1 contains a

certain content of inactive or latent species. In contrast, similar treatment of the E. coli-expressed rPAI-1 of the invention did not increase its activity. E. coli expressed rPAI- 1, the activity of which is not significantly increased by treatment with protein denaturants, e.g., less than about 10%, is considered to be within the invention. Our results establish an advantage of our rPAI-1 composition in comparison with the hPAI-1, based on its higher specific activity in all three functional assays and on the absence of any latent species or need for an activation step to obtain fully active PAI-1. The present results are surprising since they contradict those of La bers et aL. (the Pannekoek lab) who reported that rPAI-1 expressed in E. coli was expressed almost exclusively in an inactive, latent form (Fibrinolysis (1988) 2, Supp.l, 33).

Example 5 PAI:t-PA Binding Assay To identify potential compounds capable of preventing the binding of t-PA to PAI-1, and to better understand the nature of the interaction between these two proteins, we have developed a binding assay utilizing recombinant PAI-1 and radiolabeled t-PA, ¹²⁵ I-t-PA, for measuring the interaction between these two proteins. This assay relies on the use of a PAI-1-specific monoclonal antibody to capture the recombinant PAI-1 onto microtiter wells, and measuring the amount of radioactivity ( -ZSj-t-Ph) bound to the well following incubation with ¹²⁵ I-t-PA. Our studies with this assay indicate it is specific, it mimics the physiological reaction between t-PA and PAI-1, and is suitable for screening large numbers of compounds for their activity in inhibiting the binding of t-PA to PAI-1.

Furthermore, the assay will prove useful in identifying specific domains on the t-PA and PAI-1 involved in their interaction, an important first step in the process of rationally designing inhibitors of binding of the t-PA and PAI-1. Human, one chain t-PA, iodinated using the iodogen method, was purchased from

Du Pont-New England Nuclear (Billerica, MA), and is designated ¹²⁵ I-t-PA.

PAI:t-PA Binding Assay: 50 μ] of a urine monoclonal antibody specific for PAI-1 (#379, American Diagnostica Inc.), resuspended in 0.1 M Na2Cθ3 buffer to a concentration of 2 ng/ l , was added to wells of 96-well icrotiter plates made using Removawell strip holders (Dynatech Labs). The plates were incubated at 4°C overnight, after which time the wells were washed with phosphate buffered saline (PBS), dried by patting on a paper towel, and incubated with 5% bovine serum albumin in PBS for 2 h at room temperature, or overnight at 4°C. Plates may be stored in this condition for weeks at 4°C. Following three washes with PBS-Tween 10 (0.05%) and 1 wash with PBS, 50 μl of the PAI-1-containing lysate from E. coli cells (containing approximately 100 ng of rPAI-1) and 30 μ] of 0.01 M Tris buffer, pH 8.0, was added to each well for a 2 h incubation at room temperature, or an overnight incubation at 4°C. Following another 3 washes with PBS-Tween 20 and 1 with PBS, 50 μλ of Tris buffer containing 50,000 cpm of ¹²⁵ I-t-PA (approximately 100 fmoles), and test compounds, where appropriate, were added to the wells containing immobilized PAI-1, for a 1 h incubation at room temperature. Wells were then washed 3 times with PBS-Tween 10, patted dry on a paper towel , transferred to 12 x 75 mm test tubes, and counted for radioactivity in a gamma counter. Data is expressed as the percent of added ¹²⁵ I-t-PA which was bound to the wells.

We analyzed the effects of varying the PAI-1-specific monoclonal antibody concentration used in the PAI:t-PA binding assay. As mentioned above, the purpose of this antibody is to selectively capture the rPAI-1 onto the wells in such a way as to facilitate PAI-t-PA binding, i.e., to appropriately present the t-PA binding site for PAI. Optimal binding, expressed as the % of the total t-PA input bound by the PAI-1, was obtained with 100 ng of the antibody per well (Table 3). This antibody concentration was therefore utilized routinely. As is also

evident from Table 3, binding was substantially lower when this capturing antibody was omitted. This antibody therefore captures PAI-1 on the plastic well in a conformation which allows t-PA to bind to it. The use of a monoclonal antibody specific to PAI-1 to capture PAI-1 distinguishes this assay from other t-PA-PAI-1 binding assays that have been described. While the capture monoclonal antibody utilized in the described procedure is commercially available (#379 from American Diagnostica, NY, NY), the state of the art of hybrido a technology is such that generation of other PAI-1 monoclonal antibodies which have substantially the same characteristics and which can be used to serve the same purpose as #379 is a straightforward exercise. The major criterion for identifying such antibodies is that they recognize and bind determinants on PAI-1 not involved in its interaction with t-PA.

Table 3 Effect of Increasing the PAI-1-capturing Monoclonal Antibody on the Binding of ¹²⁵ I-t-PA to PAI-1

ng of Antibody/Well co ¹²⁵ I-t-PA bound

0 297 25 4768

50 6603

100 10334

250 9740

500 9489

As shown in Table 3, the binding of t-PA to immobilized PAI was substantially lower when the capturing antibody was omitted. The antibody therefore captures PAI-1 in a conformation which permits t-PA to bind to it.- The use of a monoclonal antibody to

capture PAI-1 in a conformation which allows it to bind t-PA distinguishes the assay from other t-PA-PAI-1 binding assays which have been described. The PAI-specific capture antibody utilized in this Example is commercially available (#379 from American D agnostica) . The state of the art of hybridoma technology is such that production of other PAI-1-specific monoclonal antibodies to serve the same purpose is a straightforward exercise. For example, Table 4 shows the % of the total t-PA input bound by immobilized PAI-1, where the PAI-1 was captured with different PAI-1 monoclonal antibodies. Six different monoclonal antibodies were tested, the first three of which were from the commercial vendor American Diagnostica (379, 380, and 3783) and the last three of which were generated in our laboratory using hybridoma technology known to those of skill in the art (FCDIO, BBA4 and FAG9). All significantly increased the binding of t-PA to PAI-1 in comparison with a control test where no antibody was utilized. All the monoclonal antibodies in this experiment were utilized at a concentration of 1 μg per well.

Table 4

% Binding of ¹²⁵ I-t-PA to PAI-1

Captured on M crotiter Wel ls with Di fferent

PAI-1 Monoclonal Antibodies

Monoclonal Antibody % t-PA Captured

(none) 2

379 42

380 66

3783 31

FCD10 19

BBA4 49

FAG9 30

Increasing the amount of PAI-1-containing lysate up to 50 l per well resulted in increased t-PA binding. ELISA studies

revealed that approximately 100 ng of PAI-1 is the maximum amount that is immobilized by the 100 ng of capturing antibody.

The specificity of the t-PA-PAI-1 interaction was demonstrated in an experiment where unlabeled t-PA, in increasing concentrations, was found to compete with a fixed amount of ¹²⁵ I-t-PA for PAI-1 binding (Table 5).

Table 5 Effect of Increasing the t-PA Concentration on the Binding of ¹²⁵ I-t-PA to PAI-1

rt-PAT . aα/well % Bound

0.0 100.0

0.01 62.7 0.10 56.5

1.0 34.4

10.0 11.9

50.0 4.8

The kinetics of t-PA binding to PAI-1 were examined by incubating ¹²⁵ I-t-PA for various times on PAI-1 coated wells, and examining the total cpm bound at each time point. As shown in Table 6, this time-course experiment yielded a biphasic curve displaying a very rapid initial phase of binding, followed by a second, slower phase of binding. This kinetic analysis is consistent with the physiological reaction between t-PA and PAI, which has been characterized as involving a two-step reaction: a very fast reversible phase followed by a slower irreversible phase (Hekman and Loskutoff (1988) Arch. Biochem. Biophys. 262:199-210).

Table 6 Kinetics of ¹²⁵ I-t-PA Binding to PAI-1

time (min) cpm

0.25 893

0.50 627

0.75 532

1.00 592

1.5 927

2.0 1146

2.5 968

3.0 1125

3.5 1191

4.0 1273

5.0 1954

7.5 2416

10.0 2189 12.5 2508

15.0 2931

25.0 3595

30.0 3777

40.0 5392

50.0 3840

60.0 5063

70.0 4503

80.0 4935

90.0 5699

120.0 5924

135.0 5499

150.0 5489

Scatchard analysis (Scatchard (1949) Ann. N.Y. Acad. Sci. 51:660-668) of binding data for the binding of t-PA to PAI revealed a curvilinear plot (Figure 1). This data supports the notion of a two-site interaction between t-PA and PAI: one high affinity, low capacity component; and a second lower affinity, high capacity component. That two sites on t-PA are involved in its interaction with PAI-1 in our assay system is further supported by the data in Table 7 and Figure 1. Table 7 indicates that compounds directed to the catalytic site of t-PA (the tripeptide D-Phe-L-Pro-L-Arg chloromethylketone and the monoclonal antibody CD2) , as well as compounds directed to lysine

binding sites on the kringle 2 domain of t-PA (lysine and its analog epsilon-amino caproic acid), are both capable of inhibiting the binding of t-PA to immobilized PAI-1. An irrelevant amino acid, glycine, had no inhibitory effect.

Table 7

Effect of Compounds Di ected to Different

Domains of t-PA on the Binding ¹²⁵ I-t-PA to PAI-1

Compound % Inhibition

CD2 89

D-Phe-L-Pro-L-Arg Chioromethylketone 85

Lysine 65 e-amino caproic acid 55

Glycine 0

Compounds were incubated with ¹ ^I-t-PA on PAI-1-containing wells for 1 h, as described above. CD2 and chloromethyl etone were used at 20 μg per well, while lysine, e-amino caproic acid, and glycine were used at 0.1 M per well.

Table 8 compares the inhibitory effect of the purified heavy and light chains of t-PA (prepared by reduction of two-chain t-PA with dithiolthreitol) with that of intact t-PA on the binding of ¹² I-t-PA to PAI-1. The purified light chain, which contains the catalytic moiety of the t-PA, was a potent inhibitor of binding as would be predicted based on our knowledge that the catalytic site of a serine protease is involved in binding to its specific inhibitor. However, the purified heavy chain also displayed

inhibi tory properties suggesti ng the presence of a domai n on thi s chain that al so interacts with PAI-1.

Table 8 Effects of Intact 2-Chain t-PA,

Purified Heavy Chain, and Purified Light Chain on the Binding of ¹²⁵ I-t-PA to PAI-1

% Bound % Bound % Bound t-PA. UQ A chai n B chai n intact 1-chai n

0.0 100 100 100

0.005 90 62 50

0.01 94 58 53

0.05 95 40 31

0.1 83 37 30

0.5 70 30 26

1.0 57 32 22

A growing body of cl nical data (Ha sten et aL. (1985) New England Journal of Medicine 313:1557-1563; Wiman et aL. (1985) J_^ Clin. Med. 105:265-270) suggests that elevated levels of PAI-1, by binding to and inactivating t-PA, contribute to the pathogenesis of various thrombotic disorders including myocardial infarction, deep vein thrombosis and stroke. This recognition of PAI-l's pathological significance has led to interest in identifying novel compounds which prevent PAI's binding to, and inactivation of, t-PA.

The microtiter well environment of the assay will permit the screening of large numbers of test compounds for the desired activity, that is, inhibition of the binding of t-PA to PAI-1. PAI-1-coated wells may be prepared weeks in advance, thereby reducing the time, and increasing the efficiency of drug

screening. Furthermore, our data indicates that in the assay of the invention the interaction between immobilized PAI-1 and ¹² ^I-t-PA mimics the physiological reaction between these two proteins. This evidence includes the following: 1) binding of ¹²⁵ I-t-PA to immobilized PAI-1 is inhibited by increasing concentrations of unlabeled t-PA; 2) the kinetics of binding in our assay system yielded a biphasic curve consistent with the physiological reaction between t-PA and PAI-1 (Hekman and Loskutoff (1988) Arch. Biochem. Bioohvs. 262:199-210): and 3) Scatchard analysis of binding data, as well as studies utilizing various binding inhibitors, suggest that two sites on t-PA are involved in its interaction with PAI-1, as has been described in the interaction of other serine proteases with their specific inhibitors (Wiman and Collen (1978) Eur. J. Biochem. 84:573-578). In summary, our results support the validity of utilizing this t-PA-PAI-1 binding assay as a first step for identifying compounds which may ultimately be active, in vivo, in enhancing fibrinolytic activity.

Example 6 Therapeutic Use of rPAI-1 in the Control of Fibrinolysis

The destruction of blood clots is mediated by plasmin breakdown of fibrin. Normally this process of fibrinolysis is terminated by the inhibitory action of anti-plasmin. However, there exist clinical conditions characterized by excessive fibrinolysis or bleeding caused by either defects in fibrinolytic inhibitors, or as a side-effect of therapeutic administration of fibrinolytic agents such as t-PA, urokinase or streptokinase. One specific clinical example is increased fibrinolysis and bleeding which occurs in patients during liver transplantation and which has been related to increased circulating plasma levels of t-PA (Dzik et ^ Blood 71:1090 (1988)). Pharmacological inhibitors of t-PA may represent important therapeutic agents for controlling excessive fibrinolysis during liver surgery. On the other hand, Colucci et aL. (1986) J. Clin. Invest. 78:138-144. report that induction of endogenous PAI-1 in rabbits did not

alter thrombolysis by infusion of t-PA. This suggests that administration of exogenous PAI-1 would not inhibit fibrinolysis in vivo. The lack of pharmacokinetic data lends further uncertainty as to the in vivo effect of exogenously administered PAI. It is also unclear whether unglycosylated rPAI-1 expressed by a prokaryotic cell would exhibit diminished activity in a mammal relative to the endogenous glycosylated peptide. We therefore investigated the activity of rPAI-1 in inhibiting fibrinolysis in vivo for the purpose of evaluating its therapeutic potential for eventual use in humans as a t-PA inhibitor. For this study, human fibrin clots were injected into rabbits (24 mg/kg) , which were infused with either saline or purified rPAI-1 at 1J3 g/kg/min. At various times, serum samples were collected and assayed for specific fibrin degradation products (D-dimer) by means of a D-dimer ELISA (American Diagnostica). As shown in Table 9, fibrin degradation, as indicated by the D-dimer concentration, was significantly reduced in those rabbits infused with rPAI-1 in comparison with those animals infused with saline. This study reveals the in vivo activity of rPAI-1 in inhibiting fibrinolysis, and suggests an important role for the protein as an inhibitor of fibrinolysis during liver transplant surgery and other clinical conditions characterized by excessive fibrinolysis.

Table 9

Effect of rPAI-1 on Fibrin Degradation In Vivo

D-dimer Concentration time ( in) Saline infusion PAI-1 infusion

0 15 + 2.7 14 + 3.9

30 15 + 2.6 14 + 3.0

45 19 + 0.6 13 + 2.8

60 40 + 3.5 13 + 0.6

75 48 + 3.3 16 + 1.2

105 80 + 6.5 18 + 0.1

Inttmation*! Application No: PCT/