When thrombi are formed within blood vessels, cells (platelets and leukocytes) infiltrate the fibrin matrix and later profoundly affect proteolytic dissolution by plasmin (reviewed in [¹,²]). The recently described thrombolytic barrier function of phospholipids [³] is based on diffusion limitations and intermolecular interactions with fibrinolytic enzymes. In a phospholipid environment, both the conversion of plasminogen to plasmin by plasminogen activators [e.g. tissue-type plaminogen activator (t-PA)] and the proteolytic action of plasmin are significantly retarded. Although there are only traces of free fatty acids in resting platelets [⁴], as part of their activation mechanism phospholipase A₂ (PLA₂) stored in secretory granules is released and activated [⁵?⁷]. Thus, PLA₂ hydrolyzes the ester bond at the sn-2 position of the phospholipids, releasing a free fatty acid and lysophospholipid. Considering the millimolar concentration of phospholipids in arterial thrombi [³], PLA₂ from entrapped cellular elements (platelets and leukocytes) could, hypothetically, release large quantities of free fatty acids (mainly oleic acid, because it occupies more than 54% of the sn-2 ester bonds in platelet phospholipids [⁴]).

The data on the impact of free fatty acids on the fibrinolytic system are scarce and controversial. The amidolytic activity of plasmin on small synthetic substrates is reported to be stimulated by oleic acid [⁸,⁹], whereas plasmin digestion of macromolecules (fibrin, prostromelysin-1) is inhibited [⁹,¹⁰]. The modulation of the plasmin activity is probably mediated by the binding of oleic acid to the kringle-5 domain of the enzyme [⁹]. Among the examined fatty acids with varying numbers of carbon atoms and double bonds, oleic acid proved to be the most potent in its effect on the action of plasmin [⁹,¹⁰]. The influence of fatty acids on plasminogen activation is even less well characterized. Only one plasminogen activator (urokinase) has been evaluated in this respect, and stimulation of its activity by oleic acid has been reported [⁹,¹⁰]. No data on the modulation of other activators and the inhibitors of fibrinolysis are available.

The thrombolytic agents currently approved for treatment of myocardial infarction and ischemic stroke are all plasminogen activators with more or less frequent hemorrhagic side-effects, which led to a search for new fibrin-specific plasminogen activators. However, administration of fibrin-binding thrombolytics (t-PA) did not overcome the bleeding complications [¹¹]. Reteplase is a recombinant variant of t-PA, which consists of the kringle-2 and the protease domains of the wild-type activator (reviewed in [¹²]). As expected from the lack of the finger domain, reteplase has lower affinity for fibrin, but clinical trials have shown that it produces faster reperfusion than t-PA, even at lower doses, and without any increase in the rate of bleeding side-effects, despite a moderate decrease in fibrinogen levels [¹³]. This apparent discrepancy between the clinical profile and the lack of fibrin selectivity of reteplase can be partially explained by its slower clearance in circulation, but, as our present study shows, evaluation of the activator in models that approach the complex composition of thrombi provides better understanding of its in vivo thrombolytic properties.

Surgically removed human thrombi were examined for the presence of free fatty acids and estimation of their concentration. The probe ADIFAB indicated highly variable amounts of free fatty acids in the thrombectomy samples. In five of eight thrombi examined by us, the free fatty acids present in the thrombus sections completely quenched the fluorescence of the applied ADIFAB (Fig. 1A), which is equivalent to 20 ?mol L^?1 sodium oleate in the same volume of ADIFAB (Fig. 1C). The original fluorescence of ADIFAB (Fig. 1D), however, was not changed in three of the examined thrombi (Fig. 1B). Thus, the amount of free fatty acids in thrombi spans a range from undetectable levels up to millimolar concentrations (fatty acids from 1 ?L of thrombus saturate 0.2 nmol L ADIFAB).

The effects of fatty acids on fibrinolysis were analyzed with in vitro assay systems. We modeled the overall course of thrombolysis induced by t-PA under the changing conditions of ongoing hydrolysis of phospholipids with fibrin?phospholipid clots containing PLA₂ (at a concentration releasing 1.97 ?mol L^?1 min^?1 fatty acid from LUVs). In order to identify the effects of the phospholipid hydrolytic products, which are not related simply to the elimination of the phospholipids, we assembled fibrin clots with a poPCPS mixture, the melting temperature of which is 33 ?C [²⁴]. This phospholipid only mildly affected the course of fibrinolysis at 37 ?C, in agreement with our previous findings that an ordered gel phase is required for the inhibitory effects of phospholipids on clot dissolution [³]. However, when the concentration of free fatty acid released by PLA₂ increased to over 0.24 mmol L^?1 (measured with ADIFAB), the dissolution process was significantly accelerated. Because LUVs containing lysophospholipid (the other product of the PLA₂ action on phospholipids) at the same concentration did not affect the clot dissolution, we restricted our further experiments to the effects of fatty acids.

Using an assay specific for plasminogen activation on a fibrin surface [¹⁹], significant acceleration of plasmin generation was measured in the presence of oleic acid with the fibrin-binding activators t-PA and reteplase (Fig. 2), but not with the non-fibrin-binding activator urokinase (not shown). The activation with reteplase was more sensitive to the effect of the oleic acid than that with t-PA; identical fatty acid concentrations caused greater stimulation of the reaction rate with reteplase (Fig. 2, insets; Table 1). Because Spectrozyme-PL is known to affect plasminogen activation under certain conditions, whereas S-2251 does not [²⁵], the experiments shown in Fig. 2 were repeated with this alternative plasmin substrate. At concentrations of oleic acid up to 0.5 mmol L^?1, the assay with S-2251 showed the same pattern of acceleration (data not shown), indicating that the observed stimulation does not depend on the type of detection substrate. At higher oleic acid concentrations, however, the absorbance progress curves reached plateaus at much lower values than in the Spectrozyme-PL system, resulting in a loss of sensitivity, which can be attributed to the different K_m values of the two substrates for plasmin (266 ?mol L^?1 for S-2252 [¹⁹] vs. 10 ?mol L^?1 for Spectrozyme-PL [¹⁴]).

The template function of fibrin seems to be essential for the stimulatory effect of oleic acid, because no activation was seen in the absence of template (Fig. 3A), and only a transient plasmin activity could be detected in the presence of a soluble template, cyanogen bromide human fibrinogen fragment (FgDP) (Fig. 3B). The loss of plasmin activity in the experimental setting of Fig. 3B, as well as in the later stages of the fibrin-dependent plasminogen activation assay (Fig. 2), was a surprising result in the light of earlier reports on stimulation of the plasmin amidolytic activity on small substrates by oleic acid [⁸,⁹], and prompted evaluation of the activator and plasmin activities in fibrin-free systems.

A spurious stimulation of the amidolytic activity of plasmin by oleic acid was seen when the action of 1 nmol L^?1 plasmin on 0.1 mmol L^?1 Spectrozyme PL was monitored for a period of 30 min in an assay system free of additional proteins and detergents. This effect, however, is related to adsorption of the enzyme to the plasticware, which is prevented by the detergent effect of the fatty acid. When shorter incubation with a higher concentration of plasmin (25 nmol L^?1) was used in the amidolytic assay, the magnitude of this interference was negligible, and a definite inhibition of the plasmin activity was observed (Fig. 4). Plasmin inhibition could be reversed slowly by BSA, which binds the free fatty acids, or by dilution of the plasmin?oleic acid mixture (Fig. 4, inset). Similar (but weaker) effects on the amidolytic activity of plasmin are produced by arachidonic acid (data not shown). The fibrinolytic activity of plasmin, however, was not affected to the same degree as its amidolytic activity (Fig. 5). The presence of oleic acid in the fibrin clot did not affect the lysis rate with plasmin applied to its surface without oleic acid (data not shown).

The combination of plasminogen activation in fluid phase and on a fibrin template in a single assay provided further evidence for the role of fibrin in the modulating effects of oleic acid on plasminogen activation and plasmin activity described above. When fibrinogen clotting and plasminogen activation were initiated simultaneously (Fig. 6), during the ascending phase of the turbidity curves plasminogen activation occurred in a fluid fibrinogen environment, where plasmin is susceptible to the action of oleic acid. Thus, at increasing oleic acid concentrations, more fibrinogen was spared from plasmin digestion, resulting in higher values of absorbance (oleic acid on its own does not affect the clot turbidity in the absence of plasmin formation; data not shown). During the descending phase of the curves, when most of the fibrinogen had already been converted to fibrin, the fibrinolytic rate was hardly affected by oleic acid (lysis time after maximal A₃₄₀ in Fig. 6, insets). The effect of oleic acid depended on the type of activator; although less efficient in the fluid phase [²⁶], reteplase showed fibrinolytic efficiency similar to that of t-PA in the fibrin-dependent stage of the assay (lysis time after maximal absorbance in Fig. 6A,B). In agreement with earlier reports [⁹,¹⁰], oleic acid stimulated plasminogen activation by urokinase in this assay (data not shown). If 3 ?mol L^?1 plasminogen and 30 nmol L^?1 t-PA or reteplase were added together with oleic acid (in the range 0.01?0.4 mmol L^?1) to the surface of plasminogen-free fibrin clots, a concentration-dependent prolongation of the lysis time was detected (from 74.0 ? 4.9 min in the absence of oleic acid to 87.6 ? 2.6 min at the highest oleic acid concentration with t-PA, and from 81.5 ? 5.0 min to 92.5 ? 3.0 min for reteplase).

Despite the decrease in the amidolytic and fibrinolytic activity of plasmin caused by oleic acid (Figs 4 and 5), the susceptibility of the protease to its major plasma inhibitor, ?₂PI was not modified significantly in the fluid phase (Fig. 7, inset). As a result, oleic acid and ?₂PI (at the concentration detected in human thrombi [²⁷]) yielded additive inhibition of plasmin in fibrin clots (Fig. 7).

Considering the known phospholipid content and PLA₂ release from platelets and inflammatory cells in thrombi [³,⁵?⁷], the thrombus localization of free fatty acids is not a surprising finding, but their direct demonstration in human thrombi (Fig. 1) is important evidence for the in vivo relevance of their effects described by others [⁸?¹⁰] or reported now with respect to thrombolysis. In addition, because of known inhibition of other proteases by oleic acid (leukocyte elastase, gelatinases A and B released by inflammatory cells) [²⁸,²⁹], the identification of free fatty acids in the structure of thrombi at varying (up to millimolar) concentrations supports their in vivo role in the modulation of inflammation-related proteolysis in this compartment. Because of the abundance of oleoyl side chains in platelet phospholipids [⁴] and the reported data for the role of unsaturated long-chain (more than 16 carbon atoms) fatty acids in the modulation of plasmin and other proteases [⁹,¹⁰,²⁸], oleic acid was chosen as a model molecule.

The overall concept that emerges from our results is that oleic acid is a factor that contributes to the localized action of plasmin in fibrin clots; it favors plasmin generation on its surface, and at the same time prevents the dissemination of protease activity in the circulation because of its inhibitory effects in solution. Using an assay that selectively monitors plasminogen activation at the interface of the fluid phase (containing activator) and fibrin (containing plasminogen) [¹⁹], significant acceleration of plasmin generation by t-PA can be demonstrated in the presence of oleic acid (Fig. 2A). A tenfold increase in the activation rate by t-PA can be achieved within the range of oleic acid concentrations relevant for thrombi (Table 1). This accelerating effect does not require the fibronectin finger-like, the epidermal growth factor and kringle-1 domains of t-PA. Reteplase, a recombinant variant of t-PA containing the kringle-2 and protease domains, is stimulated more strongly by oleic acid on a fibrin surface (Fig. 2B, Table 1). The facts that kringle-2 is involved primarily in interactions with lysine residues [³⁰], which are continuously exposed when plasmin digests fibrin, and that the effect of oleic acid is seen only in the presence of fibrin suggest that the acceleration of plasmin generation is based on optimized formation of the ternary plasminogen?fibrin?activator complex. The inhibition of plasminogen activation in the fluid phase (Fig. 3A) is reversed on the fibrin surface (Fig. 2). Such an effect can be expected if the modulator (oleic acid) causes the template (fibrin) to have higher affinity for the activator, so that more t-PA or reteplase is mobilized in the reactive interfacial layer, and in addition the bound form of the activator is less susceptible to the inhibitory effect of oleic acid. This interpretation is in line with the lack of an accelerating effect of oleic acid on plasminogen activation by urokinase, which does not require a template. Thus, the presence of oleic acid in thrombi may contribute to the therapeutic advantage of fibrin-specific plasminogen activators. On the other hand, because the oleic acid effects vary between fibrin-dependent activators, its role as a fibrinolytic modulator in vivo may explain the lack of strict correlation between fibrin binding and therapeutic effectiveness of thrombolytics, in addition to the role of other known factors, such as different pharmacokinetics of the activators in vivo.

In contrast to the situation with urokinase [¹⁰], t-PA-catalyzed plasminogen activation is completely blocked by oleic acid in fibrin-free solution (Fig. 3A), which is related to the effect of oleic acid on t-PA (its amidolytic activity on small synthetic substrates is also inhibited; data not shown). CNBr digestion of fibrinogen exposes the binding sites for t-PA and plasminogen that are necessary for its template function in plasminogen activation (reviewed in [³¹]), and consequently FgDP can at least partially relieve this inhibitory effect (Fig. 3B). However, the already formed plasmin is also inhibited by oleic acid in its amidolytic action in the fluid phase (Fig. 4) and to a lesser extent in its fibrinolytic action (Fig. 5). In the absence of an extensive fibrin matrix, FgDP cannot protect the already formed plasmin against the inhibitory effect of oleic acid, and consequently plasmin is not detected in the late stages of the activation assay with this soluble template. A similar phenomenon is observed in the later stages of the clot-dependent plasminogen activation assay (Fig. 2), when the generated plasmin completely digests the interfacial fibrin layer to soluble products. The loss of enzyme activity cannot be attributed to denaturation as a result of the detergent effects of oleic acid, because the reactivity of plasmin with macromolecular inhibitors is retained. In the presence of oleic acid, the sensitivity of plasmin to its natural inhibitor ?₂PI is preserved not only in free solution (Fig. 7, inset), but even on the fibrin surface (Fig. 7), which on its own is known to protect the protease against ?₂PI [²¹,³²]. Plasmin inhibition is reversible (Fig. 4, inset), but the exact type of inhibition requires further investigation. Probably, plasmin inhibition is based on conformational changes in the protease domain induced by the known binding of oleic acid to kringle-5 [⁹], whereas the kringle-1-dependent interaction with ?₂PI [³³] is not affected.

The effects of oleic acid on plasminogen activation and plasmin activity contribute to a better understanding of the in vivo data for fibrinolytic potency and fibrin specificity of reteplase, and emphasize the need to perform enzymological evaluation of thrombolytics in an environment that contains not only fibrin but also other modulators identified in thrombi. Thus, reteplase is reported to have 2-fold to 4-fold lower activator efficiency than the wild-type t-PA and to bind weakly to native fibrin because of the lack of the finger domain [²⁶,³⁴], but as a thrombolytic agent, it produces identical or even improved clinical outcome [¹³]. Our data demonstrate a 2-fold lower apparent reaction rate constant for plasminogen activation by reteplase as compared to t-PA in plasminogen activation on a fibrin surface (Fig. 2, insets), but this difference is reversed in favor of reteplase in the presence of 1.4 mmol L^?1 oleic acid (Table 1). When the action of the activators is evaluated in a model system (Fig. 6), which monitors the consequences of plasminogen activation with respect to both fibrinogen and fibrin, oleic acid causes reteplase to have higher fibrin specificity and fibrinogen-sparing properties as compared to t-PA, because its common inhibitory effect on plasmin is combined with more pronounced fibrin-dependent stimulation of plasminogen activation by reteplase. This conclusion is based on the changes in lysis time at increasing concentrations of oleic acid (Fig. 6, insets); because reteplase is less efficient in solution [²⁶], the smaller amounts of plasmin generated by it are easily blocked by oleic acid (compare the ascending phase of the turbidity curves in Fig. 6A,B), whereas there is hardly any difference in the fibrinolytic rate with the two activators when plasminogen is activated predominantly on partially degraded fibrin, to which both activators can bind through their kringle-2 domains (descending phase of the curves in Fig. 6). The experimental setting of Fig. 6 also models the action of plasminogen activators homogeneously entrapped in newly formed hemostatic plugs in the course of thrombolytic therapy. Because the premature dissolution of fibrin clots is probably responsible for the bleeding side-effects of thrombolytics, the similarity of t-PA and reteplase in this assay may explain the identical rate of hemorrhagic complications with these two activators [¹³].

In conclusion, our report provides evidence for the presence of free fatty acids in thrombi, which optimize the fibrin matrix as a template for plasminogen activation by t-PA and thus accelerate fibrinolysis with activators that approach the clot from the fluid phase. Fibrin also partially protects plasmin and t-PA against inhibition by oleic acid. In addition, a variant of t-PA (reteplase), which is a less efficient plasminogen activator in a pure fibrin environment, has an advantage over the wild-type molecule in the presence of fatty acids. Analogous evaluation (using assay formats that approach the complex composition of thrombi) may explain why there is no obvious correlation between fibrin binding and the in vivo fibrinolytic efficiency of recombinant plasminogen activators. Finally, the variability in the free fatty acid content of thrombi may contribute to the variable therapeutic outcome of thrombolysis in different patients.

The technical assistance of Gy?rgyi Oravecz is highly appreciated. The authors are grateful to K. Tenekedjiev for mathematical support in the evaluation of the experimental data. This work was supported by grants OTKA T-031 891, OTKA K60123, ETT 385/2006, NKFP-1A/0023/2002 and the Wellcome Trust (069 520/Z/02/Z).

The authors state that they have no conflict of interest.