Colin Longstaff, Biotherapeutics Group, National Institute for Biological Standards and Control, South Mimms, Herts, EN6 3QG, UK. E-mail: email@example.com
Summary. Background: Binding of tissue-type plasminogen (Pgn) activator (t-PA) and Pgn to fibrin regulates plasmin generation, but there is no consistent, quantitative understanding of the individual contribution of t-PA finger and kringle 2 domains to the regulation of fibrinolysis. Kringle domains bind to lysines in fibrin, and this interaction can be studied by competition with lysine analogs and removal of C-terminal lysines by carboxypeptidase B (CPB).
Methods: High-throughput, precise clot lysis assays incorporating the lysine analog tranexamic acid (TA) or CPB and genetically engineered variants of t-PA were performed. In particular, wild-type (WT) t-PA (F-G-K1-K2-P) and a domain-switched variant K1K1t-PA (F-G-K1-K1-P) that lacks kringle 2 but retains normal t-PA structure were compared to probe the importance of fibrin lysine binding by t-PA kringle 2.
Results: WT t-PA showed higher rates of fibrinolysis than K1K1t-PA, but the inhibitory effects of TA or CPB were very similar for WT t-PA and the variant t-PA (< 10% difference). Urokinase plasminogen activator (u-PA)-catalyzed fibrinolysis was also inhibited by TA, even though Pgn activation could be stimulated. Fibrin treated with factor XIIIa (FXIIIa) generates crosslinked degradation products, but these did not affect the results obtained with WT t-PA and K1K1t-PA.
Conclusions: t-PA kringle 2 has a minor role in the initial interaction of t-PA and fibrin, but stimulation of fibrinolysis by C-terminal lysines (or inhibition by carboxypeptidases or TA) operates through Pgn and plasmin binding, not through t-PA. This is also true when fibrin is crosslinked by treatment with FXIIIa.
Fibrin binding by tissue-type plasminogen (Pgn) activator (t-PA) and Pgn is key to the initiation of fibrinolysis. t-PA is believed to bind to fibrin primarily via finger and kringle 2 domains  and Pgn via kringle domains . Plasmin is generated, and cleaves fibrin, producing new C-terminal lysines, which serve to mediate positive feedback in the fibrinolytic cascade by: (i) promoting the binding and activation of Pgn and t-PA; (ii) promoting the plasmin-mediated conversion of Glu-Pgn to Lys-Pgn, which is a better substrate for Pgn activators; and (iii) by binding plasmin and thus protecting it from consumption by its major plasma inhibitor, α2-antiplasmin . Factor XIII (FXIII), a transglutaminase, acts in the final stage of coagulation and introduces crosslinks into fibrin, which may stabilize the clot structure. Crosslinked fragments of fibrin may also play an important role in stimulating fibrinolysis, specifically via the DDE fragment, which is believed to act as a template binding Pgn and t-PA via kringle domains . Downregulation of fibrin binding and hence fibrinolysis in vivo are also known to occur through the action of carboxypeptidases, which remove C-terminal lysines, the principal player in vivo being thrombin-activatable fibrinolysis inhibitor (TAFIa) (carboxypeptidase U) . The importance of fibrin binding in the regulation of fibrinolysis is further highlighted by the application of lysine analogs, such as aminohexanoic acid or tranexamic acid (TA), as antifibrinolytic drugs . TA has a long history of use as an antifibrinolytic, but a recent clinical trial indicated an increase in mortality if TA was given later than 3 h after trauma . Lysine analogs are able to stimulate Pgn activation by converting native Glu-Pgn to a more active Lys-Pgn-like conformation , but we do not have a complete knowledge of the mechanism of antifibrinolytics in the presence of fibrin.
There has been recent interest in computer modeling of the pathways of blood coagulation and fibrinolysis [9,10], which requires detailed kinetic knowledge of all the steps in the pathway, but may begin with identification of those steps that are most critical to the overall regulation of the pathway (sensitivity analysis). Currently, there are conflicting views on the relative importance of t-PA and Pgn binding to native fibrin and to the C-terminal lysines of partially degraded fibrin, which also means that we have an incomplete understanding of the mechanism of action of inhibitors of fibrinolysis such as TAFI and TA. There are also questions surrounding the role of the conversion of Glu-Pgn to Lys-Pgn in the regulation of fibrinolysis. To address these questions, we have performed systematic kinetic studies, using precise, high-throughput assay methods to investigate both Pgn activation and fibrinolysis catalyzed by urokinase Pgn activator (u-PA) or t-PA and engineered variants with native Glu-Pgn or Lys-Pgn. Detailed comparisons of wild-type (WT) t-PA (F-G-K1-K2-P) and a variant K1K1t-PA (F-G-K1-K1-P) permits dissection of the roles of t-PA finger and kringle 2 binding to fibrin. An understanding of these fundamentals will lead to a better understanding of the regulation of fibrinolysis in vivo.
Materials and methods
Generation and purification of t-PA variants and other activators
Three t-PA domain variants – WT t-PA, K1K1t-PA and t-PA protease domain (P) – were expressed by use of an insect cell system, as previously described . To achieve this, derivatives of pFastBac-t-PA were generated: pFastBac-t-PA:k1k1 was generated as previously described ; pFastBac-t-PA:P, which lacks the 5′ 774 bp sequence that encodes the F-E-K1-K2 region of t-PA (Val5–Ser262) was generated by site-directed mutagenesis with a QuikChange kit (Stratagene, La Jolla, CA, USA), according to the manufacturer’s protocol. u-PA was 1st International Standard for high molecular weight urokinase, 87/594 (NIBSC, South Mimms, UK). WT t-PA and variants were characterized by SDS-PAGE, and their molecular masses were in accordance with the expected values. K1K1t-PA was found to show negligible binding to lysine-Sepharose, as expected.
Pgn activation and fibrin lysis assays were performed as previously described, whereby activator was added to preformed clots (clotted for 30 min at 37 °C) made by mixing 9 μm fibrinogen (human, Pgn-depleted; Calbiochem, La Jolla, CA, USA) with 10 nmα-thrombin (NIBSC reagent 01/578) and 165–550 nm Glu-Pgn (Chromogenix, Milan, Italy, or Hypen Biomed France) or Lys-Pgn (Immuno, Vienna, Austria) . Pgn purity was checked by SDS-PAGE and quantitated by A280 nm, assuming A1% = 16.9. t-PA concentrations were determined by Coomassie binding assays with a standard traceable to a t-PA preparation whose concentration was determined by amino acid analysis. Effector molecules such as TA (Sigma-Aldrich, Poole, UK) and porcine pancreatic carboxypeptidase B (CPB) (Calbiochem, Merck Biosciences, Nottingham, UK) were added to the fibrinogen solution before clotting. Pgn activation was monitored (SpectraMax M5 plate reader; Molecular Devices, Stanford, CA, USA) by including plasmin chromogenic substrate (S-2251, Val-Leu-Lys-p-nitroaniline [pNA]; Chromogenix) with activator added to the clot, and monitoring pNA release at 405 nm. Initial rates of Pgn activation were calculated from plots of absorbance vs. seconds squared for absorbance values < 0.2, to give results in pm s−1 plasmin generation, as previously described . To follow fibrinolysis, parallel plates were prepared, but chromogenic substrate was omitted and changes in fibrin opacity were followed at 405 nm. An additional method was used that involved monitoring clot formation and lysis profiles in microtiter plates by mixing 2.5 nm thrombin, 100 nm Pgn, 8 μm fibrinogen, and a matrix of activator and effector concentrations. Typically, three replicate doses of activator were included in all assays for calculation of the apparent potency of the activator relative to the dose range without effector, which was set to 100%. Parallel line bioassay analysis  was used to provide statistics on relative potency and 95% confidence intervals at each effector concentration. Selectable endpoints for analysis (such as clotting maximum absorbance, time to clotting maximum, time to 50% lysis, time to 100% lysis, time to 50% lysis from clotting maximum, time to minimum first derivative [maximum rate of lysis], and area under the curve) were extracted from absorbance vs. time data, and analyzed with bespoke software, written using the free statistical software package r  (script available on request). For assays in the presence of crosslinked fibrin, FXIII (NIBSC; FXIII concentrate reagent 02/170) was included in clots at 2, 4 or 8 IU mL−1 following preactivation to FXIIIa, achieved by incubating FXIII with 5.5 nmα-thrombin and 5 mm CaCl2 for 30 min at 37 °C.
Model of Pgn activation on fibrin with TA
A simple scheme of Pgn activation by t-PA was developed to simulate the initial rates of plasmin generation and hydrolysis of S-2251 with the simulation software gepasi [17,18], as shown below.
1 S + E = ES (S-2251 reacting with plasmin)
2 ES → E + P (generation of pNA)
3 F + A = AF (binding of activator to fibrin)
4 AF + G = AFG (binding of Pgn to activator–fibrin)
5 F + G = FG (binding of Pgn to fibrin)
6 FG + A = AFG (binding of activator to Pgn–fibrin)
7 AFG → AF + E (generation of plasmin from ternary complex)
8 E + F = EF (fibrin is a competitive substrate with S for plasmin)
9 G + T = GT (equilibrium between Pgn and TA)
Initial estimates for the parameters used in the model were taken from the literature (e.g. [4,19] for Pgn and t-PA with fibrin, and  for TA with Pgn), and adjusted to optimize the fit of simulated data to observed data.
The effect of TA on Pgn activation in fibrin highlights the different mechanisms of t-PA and u-PA, as shown in Fig. 1. t-PA activation requires binding of t-PA and Pgn to fibrin, and disturbance of this ternary complex inhibited activation with half-maximal effects (IC50) of < 10 μm and ∼ 30 μm TA for Glu-Pgn (Fig. 1A) and Lys-Pgn (Fig 1B), respectively. A completely distinct mechanism operates with u-PA, whereby Glu-Pgn activation is stimulated with a half-maximum of ∼ 300 μm TA, reflecting a conformational change from the globular to the extended form of Pgn . Crucially, t-PA is insensitive to this conformational change, and this is not always appreciated. Very similar results, obtained in the absence of fibrin, have been previously published, also showing that t-PA is not sensitive to the conformational change in Glu-Pgn that occurs at 2 mmε-aminocaproic acid, which does result in stimulation of activation catalyzed by u-PA .
When Pgn activation (assessed with S-2251) and fibrinolysis (fibrin turbidity) were measured simultaneously in fibrin, TA was seen to inhibit the rate of Pgn activation by t-PA, and this was mirrored by the rate of fibrinolysis (Fig. 2A). Again, the different mechanism of u-PA was apparent, and there was a disconnect between Pgn activation and fibrinolysis. As shown in Figs 1A and 2B, at > 100 μm TA, the Glu-Pgn conformation was changed to create a substrate that was more rapidly activated by u-PA. However, this was not translated into fibrin lysis, emphasizing the importance of plasmin binding to fibrin as a key feature in the regulation of fibrinolysis and the action of lysine analogs and C-terminal lysines. Figure 2B shows that low concentrations of TA, up to 20 μm, inhibited Pgn activation by u-PA, but the mechanism is not clear.
To investigate the mechanism of t-PA binding and the importance of kringle 2, WT t-PA was compared with K1K1t-PA in the same assay system over a range of TA concentrations incorporated into the fibrin clot, and the results are presented in Fig. 3. K1K1t-PA had lower activity than WT t-PA (Fig. 3, inset), suggesting that kringle 2 is involved in initial binding to fibrin. Simulations with gepasi suggested that this difference in activities could be simply accounted for by weaker binding of K1K1t-PA to fibrin. Significantly, the model used to simulate the data shown in Fig. 3 (lines) only includes terms for TA blocking Lys-Pgn binding, and is sufficient to describe the observed data (points). Thus, although t-PA kringle 2 is involved in the initial binding to fibrin, inhibition by TA appears to operate largely through plasmin(ogen) kringles rather than kringle 2 of t-PA.
All other studies performed in this or other systems and with Glu-Pgn or Lys-Pgn showed the same minimal differences between WT t-PA and K1K1t-PA.
The techniques used to obtain the data in Figs 1–3 all involve kinetic methods that rely on the initial rates of Pgn activation and limited digestion of fibrin. In order to investigate the whole time course of fibrinolysis, a classical clot lysis method was included, in which fibrinogen, thrombin, Pgn activator, Pgn and effectors were mixed, and the absorbance profile of clotting and lysis was monitored. Software was developed specifically to rapidly explore ranges of TA or CPB, activator and other reactant concentrations. Endpoints could be easily selected, and representative analysis is presented as time to 50% lysis or time to 100% lysis, when more fibrin degradation products and more C-terminal lysines are present. Results are expressed as a potency (and 95% confidence interval) of activator, calculated with a parallel line bioassay statistical package, over a three-point, four-fold concentration range, relative to activity without TA. Figure 4 shows results from such an assay system with either Glu-Pgn or Lys-Pgn, and both WT t-PA (Fig. 4A) and u-PA (Fig. 4B), where potencies were calculated by using time to 50% lysis or time to 100% lysis. The curves appear similar with both activators, and are consistent with the results shown in Figs 1–3. The IC50 values for TA with t-PA in Fig. 4A are 3.4 and 26.9 μm for Glu-Pgn and Lys-Pgn, respectively, and those for TA with u-PA in Fig. 4B are 3.1 and 23.3 μm for Glu-Pgn and Lys-Pgn, respectively, highlighting the similarity between t-PA and u-PA, and supporting a plasmin(ogen)-dependent mechanism of inhibition. The small reproducible shift between results from 50% lysis and results from 100% lysis indicate more binding sites, but the consistent separation indicates that there was little conversion of Glu-Pgn to Lys-Pgn during the assay.
CPB removal of C-terminal lysines
TA blocks fibrin–kringle binding, and carboxypeptidases also downregulate fibrinolysis by cleaving C-terminal lysines. Clot lysis experiments, as performed above with TA, were repeated with WT t-PA and K1K1t-PA in the presence of CPB. CPB is frequently used as a model for other carboxypeptidases, and its activity is not complicated by thermal instability, like that of TAFIa. There was a dose–response effect in this system of CPB on clotting and lysis profiles, and representative curves are shown in Fig. 5, here with concentrations of activator that gave matched rates. Again, there were only small differences between the responses of WT t-PA and K1K1t-PA to a range of CPB concentrations up to 2.5 U mL−1. Very similar results were observed with times to 50% or 100% lysis, and with Glu-Pgn or Lys-Pgn as substrate. Taken together, these results support the notion that CPB, like TA, works primarily by reducing plasmin(ogen) binding, not t-PA binding.
It has been proposed that FXIIIa-crosslinked FDP stimulates t-PA, and also that t-PA binds to and is stimulated by DDE via kringle 2 . The significance of crosslinked FDP was investigated in this clot lysis system with and without FXIIIa, with WT t-PA and K1K1t-PA. The potencies of WT t-PA and K1K1t-PA (over a dose range of 40, 20 and 10 ng mL−1) were measured over a range of TA concentrations (0–111 μm, where inhibition is observed). If crosslinked fibrin and FDP were more important for binding t-PA through kringle 2, then WT t-PA would require higher concentrations of TA than observed in non-crosslinked fibrin (in the absence of FXIIIa). Initial studies showed that lysis was slowed in clots incorporating 2, 4 or 8 IU mL−1 FXIIIa, suggesting that FXIIIa was crosslinking fibrin, and the results obtained with 2 IU mL−1 FXIIIa and TA are shown in Fig. 6. However, the effects of TA on the potencies of WT t-PA or K1K1t-PA were very similar with or without 2 IU mL−1 FXIIIa, as shown in Fig 7, with time to 50% clot lysis (similar results were obtained for times to 100% clot lysis). Again, t-PA kringle 2 appeared to be relatively unaffected by TA. Careful analysis showed a small but statistically significant difference of 4.1% between WT t-PA and K1K1t-PA, which could be ascribed to t-PA kringle 2.
The consensus view on fibrin binding and regulation of fibrinolysis supposes that initial binding of t-PA and Pgn results in the generation of plasmin and then C-terminal lysines, which provide new binding sites and stimulation of fibrinolysis. Conversely, carboxypeptidases, such as TAFI, are able to remove C-terminal lysines and inhibit fibrinolysis. A summary of current beliefs is presented in the excellent review on TAFI by Morser et al. : ‘With the removal of these lysines there is less incorporation of Pgn and tissue plasminogen activator (tPA) into the clot. In addition Glu1-plasminogen is not converted into Lys78-plasminogen, a better tPA substrate’. The current work addresses two claims in this proposal: (i) a role for Lys-Pgn formation to create a better substrate for t-PA; and (ii) C-terminal lysines increasing the binding of t-PA.
Lys-Pgn does have a higher affinity for fibrin, so higher rates of activation by t-PA are simply explained as resulting from more ternary complex formation. However, there is no effect of the globular or the extended Pgn conformation  on activation rates with t-PA, as shown in Fig. 1, so this cannot be part of any mechanism involving t-PA. We found no evidence for major conversion of Glu-Pgn to Lys-Pgn in our assay systems, as there was a persistent difference between Glu-Pgn and Lys-Pgn activation responses to TA. The IC50 values with TA remained consistent for Glu-Pgn or Lys-Pgn, whatever the activator, and mixtures of Glu-Pgn and Lys-Pgn gave intermediate IC50 values with TA. The results with endpoints of 50% lysis were very similar to the results with endpoints of 100% lysis. Previous work by Wang et al.  has demonstrated, with SDS-PAGE, some accumulation of Lys-Pgn, and these authors suggested that, in the presence of TAFIa, the generation of Lys-Pgn was slowed, which could in part explain the antifibrinolytic activity of TAFIa. However, in their work, the amount of Lys-Pgn detected was a minor proportion of the Glu-Pgn or plasmin present in the system even at full lysis, and an even lower proportion at 50% lysis. We cannot exclude the possibility that some Lys-Pgn was generated during our clot lysis experiments, but this does not seem to be significant in these very sensitive and precise assays. Lys-Pgn formation is proposed as an important intermediate in cell surface activation, which may be different .
Additional t-PA binding as a mechanism of stimulation by C-terminal lysines (or inhibition by TAFI) is not indicated by the data presented in Fig. 3. These data demonstrate that WT t-PA and K1K1t-PA have similar responses to TA, and this is adequately explained by TA blocking Pgn binding. The simple model developed above is analogous to the steady-state template model previously described . Similar values for kcat/Km for t-PA with Lys-Pgn can be calculated with the analytical approach described in  (here, 2.4 vs. 2.69 μm−1 s−1 in ), but the numerical simulation approach of gepasi avoids the derivation of multiple equations as in . The lower activity of K1K1t-PA in the absence of TA (Fig. 3, inset) is accounted for in the model by weaker fibrin binding (WT t-PA Kd = 0.05 vs. 0.13 μm for K1K1t-PA), so there is a small effect of initial fibrin binding of t-PA kringle 2 in the absence of TA. Thus, these results indicate that t-PA kringle 2 plays a small part in the formation of the t-PA–fibrin–Pgn ternary complex, but inhibition of complex formation by extraneous TA is largely driven through blocking of Pgn binding. The data presented in Figs 2 and 4 also illustrate the importance of Pgn and plasmin binding via lysines, but not t-PA–lysine binding. Previous work has highlighted the ability of TAFIa to slow down thrombus lysis in the presence of single-chain u-PA and u-PA, and not just t-PA, stressing the important role of Pgn–fibrin binding . However, we have extended this observation to show that, even when u-PA activity is stimulated by TA within a fibrin clot, as in Fig. 2B, fibrinolysis is still inhibited by the blocking of plasmin binding.
Various physical studies and assorted assay systems have shown that t-PA kringle 2 can bind to lysines in fibrin, and there are wide-ranging estimates for Kd values . However, it is generally accepted that the t-PA finger domain has a higher binding affinity for fibrin than kringle 2, and the finger has the dominant role in driving t-PA binding and promoting enzyme activity throughout the process of clot lysis . Furthermore, t-PA kringle binding may compete with Pgn kringle binding, and it may be significant that in the assay systems used in the current study t-PA is in the physiologic range, ∼ 70 pm, about three orders of magnitude lower than Pgn. Most of the studies on the regulation of fibrinolysis that have included TAFI or other carboxypeptidases have involved t-PA as the activator, but conflicting results have been obtained, suggesting that fibrinolysis is sensitive to carboxypeptidases when the t-PA concentration is low [27,28], high , or varies over a wide range . The current work adds to our understanding of the fundamentals of how carboxypeptidases act through the removal of C-terminal lysines to inhibit fibrinolysis very largely through effects on Pgn and plasmin.
FXIIIa added to clots caused an apparent slowing of fibrinolysis, as demonstrated in Fig. 6, although recent work by Mutch et al. [30,31] has suggested that FXIIIa works by crosslinking α2-antiplasmin, and has no direct effect on fibrin stabilization. The difference between the results presented here in Fig. 6 (where there is an effect of FXIIIa but there is no α2-antiplasmin present) and their findings may be attributable to flow, which was part of their experimental system and warrants further investigation. Figure 7 clearly shows similar behavior of WT t-PA and K1K1t-PA with or without FXIIIa. If crosslinked FDP and DDE provided a high-affinity binding template for t-PA via kringle 2 , higher concentrations of TA would be needed to inhibit WT t-PA stimulation by DDE produced in the presence of FXIIIa, but this is not seen.
Because TA blocks plasmin binding and activity on fibrin, it should be a potent antifibrinolytic in all circumstances. However, it is possible that treating patients with TA in the presence of u-PA would stimulate plasmin generation and at the same time protect plasmin from inhibition by α2-antiplasmin. Thus, the role of u-PA and plasmin activity in patients given TA later than 3 h after trauma could be investigated as a possible explanation for increased mortality .
This work was supported by the Wellcome Trust (083174) and the British Heart Foundation (F5/06/021).
Disclosure of Conflict of Interests
The authors state that they have no conflict of interest.