Proc. Nati. Acad. Sci. USA Vol. 91, pp. 3670-3674, April 1994 Biochemistry A faster-acting and more potent form of tissue plasminogen activator BRUCE A. KEYT*, NICHOLAS F. PAONI*, CANIO J. REFINO*, LEA BERLEAU*, HUNG NGUYEN*, ALICE CHOW*, JADINE LAI*, LUIS PENA*, CHERYL PATER*, JOHN OGEZt, TINA ETCHEVERRYt, DAVID BOTSTEIN§¶, AND WILLIAM F. BENNETrt Departments of *Cardiovascular Research, tCell Culture Research, and tProcess Sciences, 1Genentech, Inc., 460 Point San Bruno Boulevard, South San Francisco, CA 94080 Contributed by David Botstein, December 27, 1993 ABSTRACT Current treatment with tissue painogen generation causes decreased levels of circulating plasmino- activator (tPA) requires an Intravenous infusion (1.5-3 h) gen, fibrinogen, and a2-antiplasmin. An undesirable conse- became the clearance of tPA from the circulation is rapid (t1/2 quence of systemic activation is bleeding that may be related 6 min). We have developed a tPA variant, T103N,N117Q, to plasmin generation rather than fibrinogen depletionper se; KHRR(296-299)AAAA (TNK-tPA) that has substantially both peripheral and intracranial hemorrhage are associated slower in vivo clearance (1.9 vs. 16.1 ml per min per kg for tPA with systemic activation (11). One way to reduce systemic in rabbits) and near-normal fibrin binding and plasma clot lysis activation is to make tPA even more fibrin-specific-i.e., activity (87% and 82% compared with wild-type tPA). TNK- reduce its activity in the absence of clotted plasma. System- tPA exhibits 80-fold higher resistance to plashlogen activator atic mutagenesis was applied to tPA with the hope ofincreas- iuhibltor 1 than tPA and 14-fold enhanced relative fibrin ing the fibrin specificity oftPA. Fortunately, mutations in the specificity. In vitro, TNK-tPA is 10-fold more effective at protease domain were found that have this property (12); the conserving fibrinogen in plasma compared to tPA. Arterial best characterized example is a tetraalanine substitution at venous shunt models of fibrinolysis in rabbits indicate that positions 296-299 [KHRR(2%-299)AAAA (abbreviated as TNK-tPA (by bolus) Induces 50% lysis in one-third the time K)] (13, 14). required by tPA (by infusion). TNK-tPA is 8- and 13-fold more In our mutagenesis studies, we found variants that exhibit potent in rabbits than tPA toward whole blood clots and reduced plasma clearance of tPA sufficient to provide effec- platelet-enriched clots, respectively. TNK-tPA conserves fi- tive thrombolysis when the agent is administered as a bolus. brinogen and, because of its slower clearance and normal clot It was an unexpected type of mutation, T103N (T) (which lysis activ, is effective as a thrombolytic agent when given as exhibits an additional glycosylation site on kringle 1), that a bolus at a relatively low dose. produced a variant (T-tPA) with the most suitable pharma- cokinetic profile. That mutation in combination with the Tissue plasminogen activator (tPA) is a multidomain serine tetraalanine substitution at positions 296-299 yielded a tPA protease of the plasminogen family (1). In the early 1980s, it variant (TK-tPA) with the desired clearance rate and en- was discovered that tPA is capable of fibrin-stimulated clot hanced fibrin specificity (15, 16). However, this combination lysis (2) and that it could be used as an agent in the treatment variant still did not yield full in vitro or in vivo fibrinolytic of acute thrombotic disorders such as myocardial infarction activity when compared with wt tPA. In the present study, (3, 4). Evolutionary pressures that shaped tPA conferred we show that through the appropriate combination of muta- properties on the molecule that make it optimal for endoge- tions at three distinct sites on tPA, it is possible to reduce the nous fibrinolysis. When tPA is used as a thrombolytic agent, clearance while retaining full fibrinolytic activity. The key additional properties may be desirable as well. Due to its factor appears to be maintenance of fibrin affinity of the rapid clearance from the circulation, tPA must be infused to molecule that we achieved by an additional mutation, N117Q achieve thrombolysis. Front-loaded dosing with increased (N). The resultant variant (TNK-tPA) is substantially more initial concentrations of tPA have demonstrated more rapid potent than wt tPA. and complete lysis compared to the standard infusion proto- col (5, 6), and early patency is correlated with significantly MATERIALS AND METHODS improved survival rate (7). Bolus administration might fur- ther improve the lytic rate by quickly exposing the target clot tPA variants were constructed by oligonucleotide-directed to a higher concentration of the enzyme, but single bolus mutagenesis as described (12). Expression and purification of administration of wild-type (wt) tPA is not generally used in the tPA variants was as described (15). Briefly, CHO cells the clinic due to its clearance. were stably transfected with plasmids containing the tPA Many investigators have produced longer half-life versions gene, amplified in the presence of methotrexate, and grown oftPA that could be administered as a bolus (for reviews, see in serum-free medium for 6 days (17). The conditioned cell refs. 8 and 9). However, virtually all ofthe tPA variants with culture medium was concentrated and diafiltered (18), and reduced clearance turned out to have significantly decreased the tPA variants were purified using lysine affinity chroma- fibrinolytic activities (10). There is a potential problem that tography. Quantitation of tPA and variants after purification must be addressed when considering bolus administration of was accomplished by a dual monoclonal assay sensitive to a fully active tPA: when tPA is given as a bolus (or even as epitopes in the kringle 2 and the protease domains. The an infusion at high doses), the plasma levels of enzyme mutations evaluated in this study did not exhibit altered rapidly increase and plasminogen becomes activated system- ically and on the surface of the clot. This systemic plasmin Abbreviations: tPA, tissue plasminogen activator; PAI1, plasmino- gen activator inhibitor 1; wt, wild type. Mutations in tPA are abbreviated as follows: T, T103N; N, N117Q; K, KHRR(296- The publication costs of this article were defrayed in part by page charge 299)AAAA. payment. This article must therefore be hereby marked "advertisement" §Present address: Department of Genetics, Stanford University, in accordance with 18 U.S.C. §1734 solely to indicate this fact. Stanford, CA 94305. 3670 Downloaded by guest on October 1, 2021 Biochemistry: Keyt et al. Proc. Natl. Acad. Sci. USA 91 (1994) 3671 Table 1. Pharmacokinetic and fibrinolytic activities, fibrin binding, fibrin specificity, and inhibitor resistance tPA Plasma Plasma Fibrin Activity Activity in Resistance variant clearance clot lysis binding in plasma plasma clot to PAI T 0.10 0.28 0.34 0.68 0.56 0.9 K 1.18 0.85 0.93 0.13 1.01 90 TK 0.16 0.34 0.33 0.13 0.65 90 TN 0.29 0.47 1.00 1.13 1.17 ND NK 0.86 0.98 1.13 0.16 1.38 ND TNK 0.12 0.82 0.87 0.06 0.85 80 Activities are normalized to those of wt tPA. tPA mutations: T, T103N; K, KHRR(2%-299)AAAA; N, N117Q. In vivo plasma clearance in rabbits is expressed as a ratio of area under the curves for wt and variant tPA. Clot lysis is the normalized value for each sample in the plasma clot lysis assay. Fibrin binding results were determined with the variants in the two-chain (plasmin converted) form. Apparent Kd values for the variants were calculated at the fibrin concentration for half-maximal binding, which was normalized to that ofwt tPA. Plasminogen activation activity oftPA variants in plasma normalized to that of wt tPA. Plasmin generation was determined by chromogenic assay using S2251 as substrate. Plasminogen activation activity of tPA variants in the presence of clotted plasma, normalized to that of wt tPA. PAI resistance is expressed as the second-order rate constant of wt tPA divided by those ofthe tPA variants. The rate constant for wt tPA was determined as 1.2 x 107 M-1sec-1. ND, not done. interactions with the monoclonal antibodies against wt tPA mutants in this region (38), K-tPA is -90-fold more resistant that were used in the assay. to inhibition by PAI1 than wt tPA, as determined by the Pharmacokinetic analysis oftPA variants was done in New second-order inhibition rates (15). To create a variant oftPA Zealand White rabbits (weight, 3 kg; three rabbits per group) with reduced clearance, enhanced fibrin-specificity, and with a variation ofpreviously described procedure (15). Each PAI1 resistance, the mutations at three loci, T103N, N117Q, sample of tPA or variant (0.3 mg/kg of body weight) was and KHRR(296-299)AAAA, were combined in TNK-tPA. administered by bolus injection via the ear vein and blood Pharmacokinetic Analysis. The clearance of T-containing samples were drawn over a 2-h period. Blood samples (0.2 variants were compared to that of wt tPA in rabbits after ml) were collected on EDTA with Phe-Pro-Arg chloromethyl bolus intravenous injections (Table 1). For TNK-tPA, the ketone, processed to plasma, stored at -70°C, and quantified area under the curve was 8.4-fold greater than for wt tPA with an ELISA employing two monoclonal antibodies. (Fig. 1); yielding a clearance of 1.9 ml per min per kg for Fibrin binding, plasma clot lysis, and plasminogen activa- TNK-tPA compared with 16.1 ml per min per kg for wt tPA. tion assays were performed as described (12). The fibrin The data was fitted to biexponential and triexponential equa- binding studies were done with 125I-labeled Tyr-Pro-Arg- tions for TNK-tPA and wt tPA, respectively.