Functional Properties of the Recombinant Kringle-2 Domain of Tissue Plasminogen Activator Produced in Escherichia coZi*

The kringle-2 domain (residues 176-262) of tissue-type plasminogen activator (t-PA) was cloned and ex- pressed in Escherichia The recombinant peptide, which concentrated in cytoplasmic inclusion bodies, was isolated, solubilized, chemically refolded, and purified by affinity chromatography on lysine-Sepharose to apparent homogeneity.

The kringle-2 domain (residues 176-262) of tissuetype plasminogen activator (t-PA) was cloned and expressed in Escherichia coli. The recombinant peptide, which concentrated in cytoplasmic inclusion bodies, was isolated, solubilized, chemically refolded, and purified by affinity chromatography on lysine-Sepharose to apparent homogeneity.
[3BS]Cysteine-methioninelabeled polypeptide was used to study the interactions of kringle-2 with lysine, fibrin, and plasminogen activator inhibitor-l.
The kringle-2 domain bound to lysine-Sepharose and to preformed fibrin with a Kd = 104 f 6.2 MM  to structures in other proteins. These domains are from the NH, terminus to the COOH terminus: a finger domain homologous to the finger structures in fibronectin (2), next an EGF domain homologous to the epidermal growth factor (2), two so-called kringle structures, kringle-1 and kringle-2, homologous to structures in plasminogen, urokinase, prothrombin, and Hageman factor (3)(4)(5)(6), and the serine protease domain showing varying degrees of homology with numerous other serine proteases (7).
Subsequently, the delineation of exon-intron junctions in the t-PA gene by Ny et al. (8) suggested that separate exons or sets of exons encode such discrete structural domains in the t-PA molecule. Thus, t-PA appears to be a mosaic protein that has evolved through exon shuffling, which suggests that the structural domains may be functionally autonomous. To test this hypothesis several groups have created sets of deletion mutants lacking various domains of wild-type t-PA (9-13). However, this approach has not resulted in a clear understanding of the role of each domain. First, it is unclear whether t-PA binding to fibrin can be attributed to specific domain(s) in the t-PA molecule, and, if so, which structures are involved. There is a general consensus now that the finger and the kringle-2 domains possess fibrin binding properties (g-13), but the nature of this interaction is not understood. It is assumed that the binding of the kringle-2 domain to fibrin is mediated through the lysine binding site, but it is not established whether binding via such a site requires new COOH-terminal lysine residues in fibrin which are generated during limited plasmin hydrolysis. Secondly, it has been postulated but never rigorously proven that kringle-2 serves a role in "docking" t-PA to its major inhibitor, plasminogen activator inhibitor-l (PAI-l), thereby facilitating the formation of the classical protease-serpin acyl intermediate (14). In an attempt to clarify the functions of kringle-2 in the interaction of t-PA with fibrin and PAI-1, we cloned and expressed the isolated kringle-2 polypeptide. The purified polypeptide was used in binding experiments to study the interaction with lysine, fibrin, and PAI-as an alternative approach toward the elucidation of the structure-function relationships of t-PA/fibrin and t-PA/PAI-1 interactions.   (1)). The folded recombinant kringle-2 was separated from misfolded kringle-2 protein and the remaining cell proteins by affinity chromatography on lysine-Sepharose and elution with EACA. The resulting protein electrophoresed as a homogeneous band on SDS-PAGE following reduction ( Fig.  2), migrating with an apparent molecular weight of 11,000. The EACA eluate was also analyzed by reverse-phase HPLC to establish homogeneity (Fig. 2). The chromatographic profile revealed a major protein peak, eluted with 37% acetonitrile. The amino acid composition analysis listed in Table I corresponded to the expected composition of the t-PA kringle-2 domain. The NH,-terminal amino acid sequence determination showed the expected starting sequence of the kringle-2 domain preceded by the dipeptide Met-Tyr at the NH2 terminus, which was present in the t-PA derivative used to construct the expression plasmid in Fig. 1 (18). Peptide mapping of the purified kringle-2 domain revealed that the disulfide bridges were positioned as originally proposed from the primary structure. 3 The Interaction of the Kringle-2 Domain with Lysine-Sepharose-The association of the kringle-2 domain with lysine-Sepharose was studied quantitatively. Different concentrations of [?S]methionine-cysteine-labeled kringle-2, ranging from 50 to 1250 nM, were incubated with lysine-Sepharose containing 200 nM immobilized lysine-HCl. A least square regression analysis was used to determine a dissociation constant of 104 + 6.2 @M for the binding of the kringle-2 domain to immobilized lysine, fitting most accurately to a one-binding site model (n = 0.85 + 0.012) (Fig. 3).
Binding to Fibrin (30). The intensity of the intact BP-chains is diminished, indicating partial degradation to PAl-42 having the NHz-terminal amino acids 1 to 42 removed (30). As shown in Fig. 6, kringle-2 bound to preformed fibrin formed from intact fibrinogen fraction I-4 and equally well to fibrin clots formed from mildly plasmin-degraded fibrinogen fraction I-8.
However, it bound substantially less to preformed fibrin formed from more extensively plasmin-degraded fibrinogen, fragment X. Thus, the fibrin binding site for kringle-2 is present in intact fibrin and absent in fibrin formed from fragment X.
Inhibition of t-PA-mediated Plasminogen Actiuation-The role of the kringle-2 domain in t-PA-mediated plasminogen activation was studied in the presence of different fibrin oligomers.
Kringle-2 strongly inhibited when intact fibrin oligomers I-4 were used as accelerators, inhibited less when mildly plasmin-degraded fibrin oligomers I-8 were the accelerators, and did not show any inhibition when more extensively plasmin-degraded X oligomers were used as accelerators (Fig. 7). Kringle-2 had no effect on plasmin activity or the hydrolysis of the synthetic substrate S-2251, since no inhibition could be observed when X oligomers were used as the accelerators.
Binding to Plasminogen Activator Inhibitor-l-To study the interaction between kringle-2 and PAI-1, we used three independent experimental approaches. First, '251-labeled guanidine-activated PAI-and [35S]methionine-cysteine-labeled kringle-2 were mixed and passed over a column of polyclonal t-PA antibody immobilized on protein-A. The portion of PAI-1 which bound to kringle-2 immobilized to the t-PA antibody-Sepharose could be eluted with 0.2 M EACA, whereas kringle-2 was eluted from the column only after application of 6 M guanidine HCl which dissociates the antigen-antibody complex (Fig. 8). In control experiments, PAI-did not react with the antibody matrix, since the total amount of PAI-applied could be recovered in the flow-through fraction when kringle-2 was omitted (not shown). Secondly, the amidolytic activity  strongly by PAI-1. In this assay system, both the isolated kringle-2 peptide and EACA interfered with the interaction between the activator and the inhibitor and partly restored t-PA amidolytic activity (Fig. 9). In the third approach, the binding of varying amounts of [35S]methioninecysteine-labeled kringle-2 to a constant amount of PAI-was investigated.
As shown in Fig. 10, the binding becomes saturated with 8000 ng of kringle-2 added. The presence of unlabeled kringle-2 in a 50-fold molar excess displaced approximately 75% of labeled kringle-2. Nonlinear regression analysis of the binding data of three separate experiments yielded a Kd of 0.51 f 0.055 PM, involving 0.36 + 0.026 binding site. DISCUSSION Deletion mutagenesis has been used extensively to study the structure-function relationships of t-PA (9-13). However, this approach has several shortcomings; the expression of recombinant mutant proteins in different eukaryotic cell lines probably results in different post-translational modifications such as glycosylation and may lead to different molecules. Since the disulfides in the t-PA variants have not been characterized, it is not clear whether the mutated molecules are properly folded. In addition, the deletion of one or several domains and joining together of domains which are normally separated could conceivably lead to steric hindrance of one domain or another or significant conformational changes of neighboring domains. Based on the hypothesis (9) that separate exons or sets of exons encode "structural domains" in the t-PA molecule with autonomous functions, we took a different approach to clarify further the structure-function relationships of t-PA by cloning and expressing individual domains in E. coli.
The capability of the kringle-2 peptide to bind to lysine and to fibrin supports the observations of van Zonneveld et al.
(31) that beside the finger domain, the kringle-2 domain also contributes to the fibrin-directed properties of t-PA. They observed that a deletion derivative consisting of the kringle-2 and serine protease domains bound to immobilized lysine and to intact fibrin. These results have been substantiated by other reports (10,13) showing the involvement of the kringle-2 domain in fibrin binding of t-PA. We extended these experiments by examining the affinities of these interactions.
Our calculated dissociation constant for the interaction of kringle-2 with lysine of 104 pM is in good agreement with the 100 PM recently published (32). The significant differences of the dissociation constants of the kringle-2 domain for lysine and preformed fibrin (& = 4.2 pM) of more than 20-fold raise doubts about the theory that binding of kringle-2 to fibrin is exclusively mediated through a lysine binding site. Since the kringle-2 domain represents a region of high hydrophobicity within the t-PA molecule, the additional involvement of hyprophobic binding cannot be excluded.
Controversies, however, exist whether plasmin digestion of the fibrin substrate and exposure of new COOH-terminal lysine residues are required for kringle-2 binding as hypothesized by van Zonneveld et al. (9). This view is in contrast to Higgins and Vehar (33) who reported that EACA inhibited the binding of t-PA to intact fibrin, but not to plasmindegraded fibrin. These authors concluded that the lysine binding site in t-PA is required for the interaction with intact fibrin and not with plasmin-degraded fibrin. Furthermore, Pannell and co-workers (34) demonstrated that plasmin treatment and subsequent carboxypeptidase B treatment of the fibrin substrate did not affect t-PA-mediated clot lysis. It is firmly established that plasminogen binds to COOH-terminal lysine residues upon plasmin digestion (35). Therefore, if t-PA would share the same binding sites, competition would arise between the two molecules. However, this does not appear to be the case since the presence of plasminogen increased the binding of t-PA mutants containing the kringle-2 structure (10).
Our findings that kringle-2 fails to bind significantly to polymerized fragment X and does not inhibit t-PA-mediated plasminogen activation when X oligomers were used as accelerator are supported by the observation of Niewenhuizen et al. (36) that t-PA binds to FCB-24 but not to fragment X, indicating that the binding site is masked in fragment X and exposed upon CNBr digestion.
In fibrinogen I-4 and I-8, exposure of the binding site may be induced by conformational changes upon polymerization.
Clearly, our results indicate that plasmin degradation and exposure of new COOH-terminal lysine residues are not a prerequisite of the kringle-2 domain to bind to fibrin. However, since our kringle-2 domain was produced in E. coli, it did not contain a carbohydrate side chain at position Asn-184. Therefore, it cannot be ruled out that glycosylation plays a role in fibrin binding as suggested by Hansen and co-workers (37). This is supported by the finding that a t-PA variant, lacking the carbohydrate side chain at position Asn-184, possessed a 30% to 50% greater specific fibrinolytic activity than t-PA containing this carbohydrate structure (38). Previously, we reported that the affinity of a mutant consisting of the kringle-2 domain and serine protease domain significantly decreased upon limited plasmin digestion of fibrin (39). This supports our results reported on this occasion and argues against the possibility that the lack 4 FCB-2, cyanogen bromide fragment FCB-2 of fibrinogen, which consists of fibrinogen chain fragments Aa 148-208, BP 191-224, 225-242, 243-305, and y 95-265, held together by disulfide bonds. of carbohydrate side chain attachment alters the interaction of the kringle-2 domain to intact fibrin and to limited plasmindigested fibrin with respect to the structural requirements for binding. However, the affinities might be different.
It has been suggested that not only is the light chain of t-PA the target of PAI-1, but also that the t-PA heavy chain is involved in the complex formation. This hypothesis was based on the observation that a t-PA deletion mutant consisting only of the finger, EGF, and serine protease domains is less efficiently inhibited by PAI-than the parent t-PA molecule (14). Heckman and Loskutoff (40) investigated the kinetics of the t-PA/PAI-1 interactions, and their results indicate that this interaction may involve two binding sites in the t-PA molecule. Recently, de Vries et al. (41) investigated the inhibition pattern of a hybrid protein consisting of the heavy chain of t-PA and the light chain of urokinase and demonstrated that the hybrid protein is inhibited to a greater extent by PAI-than urokinase, suggesting the involvement of the heavy chain of t-PA in the interaction with PAI-1. The kinetics of the inhibition of t-PA by PAI-in the presence of fibrin showed that the inhibitor interferes with the binding of t-PA to fibrin in a competitive manner (42). It was concluded that one binding site of PAI-to the t-PA molecule is a sequence in or close to the fibrin binding domain of t-PA. We now present evidence that the kringle-2 domain is involved in the binding of PAI-to t-PA. Based on our results and the aforementioned observations, we propose the following model for the interaction of t-PA with its primary inhibitor. The reaction of t-PA with PAI-occurs in a two-phase reaction. The first phase is a reversible one, consisting of a reversible binding of PAI-to the kringle-2 domain and also a reversible interaction with the active center of t-PA. The second phase is an irreversible binding of PAI-with the light chain of t-PA resulting in the SDS-stable complex. This model is analogous to the plasmin/a-2 antiplasmin interaction representing also a two-step reaction with a very fast reversible second order reaction followed by a slower irreversible first order transition (43). In conclusion, the demonstration that an isolated domain in t-PA possesses biological activities supports the theory proposed by Rogers (44)