Lytic resistance of fibrin containing red blood cells

Abstract

Objective

Arterial thrombi contain variable amounts of red blood cell (RBC), which interact with fibrinogen through an eptifibatide-sensitive receptor and modify the structure of fibrin. Here we evaluate the modulator role of RBCs in the lytic susceptibility of fibrin.

Methods and Results

If fibrin is formed at increasing RBC counts, scanning electron microscopy evidenced a decrease in fiber diameter from 150 nm to 96 nm at 40 %(v/v) RBC, an effect susceptible to eptifibatide inhibition (restoring 140 nm diameter). RBC prolonged the lysis time in a homogeneous-phase fibrinolytic assay with tissue plasminogen activator (tPA) by up to 22.7±1.6 %, but not in the presence of eptifibatide. Confocal laser microscopy using green fluorescent protein (GFP)-labeled tPA and orange fluorescent fibrin showed that 20-40 %(v/v) RBC significantly slowed down the dissolution of the clots. tPA-GFP did not accumulate on the surface of fibrin containing RBC at any cell count above 10 %. The presence of RBC in the clot suppressed the tPA-induced plasminogen activation resulting in a 45 % less plasmin generated after 30 min activation at 40 %(v/v) RBC.

Conclusion

RBCs confer lytic resistance to fibrin resulting from modified fibrin structure and impaired plasminogen activation through a mechanism that involves eptifibatide-sensitive fibrinogen-RBC interactions.

A reverse correlation between bleeding time and red blood cell (RBC) count was observed over a century ago.¹ When anemic and thrombocytopenic patients were treated with transfusion, the correction of the bleeding time correlated with the corrected hematocrit despite the fall in the platelet count to pre-transfusion levels. Later studies² confirmed that the prolonged bleeding time of severely anemic patients can be corrected with washed RBC transfusion (thus excluding the effect of plasma factors) at essentially unchanged platelet counts. The mechanism of RBC contribution to the hemostatic function is still open to question notwithstanding their known role in the maintenance of blood viscosity, the chemical signaling of platelet activation and the provision of phospholipid support for the activation of coagulation factors (reviewed by Wohner³). The solid matrix of hemostatic and pathological blood clots is composed of fibrin and variable cellular elements, but the incorporation of RBCs appears to be mediated through a less specific entrapment than that of platelets.⁴ Based on binding data approximately 2 % of fibrinogen, the plasma precursor of fibrin monomers, is estimated to circulate in vivo in association with erythrocytes.⁵ This interaction is not simply a non-specific protein adhesion to the cell membrane, but it involves an erythrocyte receptor⁶ and a specific domain around residues 207-303 of fibrinogen Aα-chains.⁷ Recently the RBC receptor for fibrinogen has been identified as an integrin related to the platelet α_IIbβ₃ receptor with similar eptifibatide-sensitivity and impaired function in Glanzmann thrombastenia.⁸ When fibrinogen is converted to fibrin, the presence of RBCs modifies the structural and viscoelastic characteristics of plasma clots.^9,10 Because fibrin structure profoundly affects the subsequent removal of blood clots from the vasculature by tissue plasminogen activator (tPA)-dependent proteolysis as a basic fibrinolytic mechanism (reviewed recently by Weisel¹¹ and Lord¹²), the presence of RBC may change the lytic susceptibility of thrombi, but no direct evidence for such modulation of fibrin dissolution is currently available. The present study was undertaken in attempt to characterize the impact of RBC on two distinct stages of the fibrinolytic process, plasminogen activation by tPA and fibrin degradation by the generated plasmin, and correlate it to the observed structural changes caused by the incorporation of RBC in pure fibrin clots.

Methods

Isolation and labeling of RBCs

RBCs were isolated from citrated whole blood collected from healthy volunteers with venipuncture. Within 1 h after collection platelet-rich plasma was removed following centrifugation at 150g for 10 min and the cell pellet was further centrifuged at 2,000g for 10 min to sediment the RBCs. The RBC pellet was washed 3 times with 10 volumes of 1.5 mM KH₂PO₄, 8.1 mM Na₂HPO₄ buffer pH 7.4 containing 137 mM NaCl, 2.7 mM KCl and 5 mM glucose (PBS) by resuspension and centrifugation at 1,200g for 5 min. RBCs were counted in hematology analyzer Abacus Junior B (Diatron GmbH, Vienna, Austria) and the hematocrit of the RBC suspension was adjusted to 0.8 with PBS. RBCs were stored up to 2 h at room temperature before use. For the measurements using confocal laser microscopy washed RBCs were labeled with Vybrant DiD cell-labeling solution (Invitrogen Life Technologies, Budapest, Hungary) by mixing 5 ml RBC suspension diluted to hematocrit of 0.016 in PBS with 25 μl Vybrant DiD solution for 30 min followed by 3 wash cycles (resuspension/centrifugation at 1,200g for 5 min) with PBS. Based on the cell count, the hematocrit of the Vybrant-labeled RBC in the final pellet was adjusted to 0.8.

Ball sedimentation assay of fibrin dissolution

The assay was designed as an alternative of the reference method from the European Pharmacopoeia for determination of tPA potency¹³ adapted for the high opacity of clots containing RBCs. Two experimental setups were used, in both of which the total amount of fibrin in the clot was constant, but in Model 1 this was achieved with increasing extracellular concentrations of fibrinogen (from 7.4 up to 12.4 μM human fibrinogen, plasminogen-depleted, Calbiochem, LaJolla, CA) compensating for the rising RBC occupancy (from 0 up to 40 %(v/v)) in a total volume of 2 ml, whereas in Model 2 extracellular concentrations of fibrinogen was constant (7.4 μM) in a clot volume increasing from 0.8 ml in the absence of RBC up to 1.35 ml at 40 %(v/v) RBC. The RBC-fibrinogen mixtures were supplemented with varying amounts of plasminogen isolated from human plasma,¹⁴ CaCl₂ and in certain cases with eptifibatide (GlaxoSmithKline Kft., Budapest, Hungary), so that the final extracellular concentrations in the assay clots were always 0.1 μM plasminogen, 1 mM CaCl₂ and 20 μM eptifibatide. For all assays thrombin of low specific activity (Serva Heidelberg, Germany) was further purified by ion-exchange chromatography on sulfopropyl-Sephadex yielding preparation with specific activity of 2100 IU/mg¹⁵ and 1 IU/ml was considered equivalent to approximately 10.7 nM by active site titration.¹⁶ Clotting and fibrinolysis were initiated simultaneously in transparent reaction tubes of 0.8 cm diameter by mixing RBC-fibrinogen suspensions with PBS containing thrombin and tPA (Boehringer Ingelheim, Germany) at concentrations needed to reach final extracellular values of 85 nM and 1 nM, respectively. After 15 s a steel ball of 2 mm diameter and 0.13 g weight was placed on the surface of the clot and the time elapsed until the ball reached the bottom of the tube (lysis time) was measured as an indicator of the collapse of identical quantities of fibrin.

Plasminogen activation assays

For the plasminogen activation assay on the surface of the clot RBC suspensions in fibrinogen were prepared as described above for the fibrinolytic assay (only plasminogen was applied at a different concentration, so that the final extracellular value in the clots was always 0.5 μM). The RBC-fibrinogen mixtures were clotted by 16 nM thrombin in a total volume of 75 μl in 96-well microplates for 30 min. Thereafter 150 μl 15 nM tPA was added to the surface of the clots and after 30 min the fluid phase was removed in ice-cold tubes and centrifuged at 2,000g for 10 min at 4 °C to remove cell debris. The plasmin activity in the supernatant was determined by mixing 100 μl supernatant and 100 μl 0.2 mM Spectrozyme-PL (H-D-norleucyl-hexahydrotyrosyl-lysine-p-nitroanilide, American Diagnostica, Pfungstadt, Germany) and measurement of absorbance at 405 nm (activity was expressed in ΔA₄₀₅/min). To evaluate the effect of eptifibatide on plasminogen activation in the absence of fibrin a two-stage activation assay was used, as previously described.¹⁷ Briefly, 3 μM plasminogen containing no or 20 μM eptifibatide was mixed with 70 nM tPA, samples were taken at intervals and the amidolytic activity of the generated plasmin was measured using 0.1 mM Spectrozyme-PL.

Expression of fluorescent chimeric tPA variant (tPA-GFP)

Recombinant human tPA -jelly fish green fluorescent protein (GFP) was constructed and expressed using the Bac-toBac baculovirus expression system as a tPA-C-terminal fusion with Enhanced Green Fluorescent Protein (EGFP) isolated from the pEGFP plasmid (Clonetech, Mountain View, CA, USA), as described previously.^18,19

Confocal microscopic imaging in the course of fibrinolysis

RBC suspensions in fibrinogen were prepared as described for the plasminogen activation assay using Vybrant-labeled RBCs and replacing 1 % of the fibrinogen with Alexa Fluor® 546-conjugated fibrinogen (Invitrogen Life Technologies, Budapest, Hungary). Fibrinogen was clotted with 16 nM thrombin for 30 min in 0.5-mm high chambers constructed from glass slides. Thereafter 60 nM tPA -GFP was added to the edge of the clot and the fluorescence (excitation wavelength 488 nm, emission wavelength 525 nm for tPA-GFP detection; excitation wavelength 543 nm, emission wavelength 575 nm for Alexa⁵⁴⁶-fibrinogen detection; excitation wavelength 633 nm, emission wavelength 650 nm for Vybrant-labeled RBC detection) was monitored with Confocal Laser Scanning System LSM510 (Carl Zeiss GmbH, Jena, Germany) taking sequential images of the fluid-fibrin interface at a distance of approximately 50 μm from the glass surface with identical exposures and laser intensities using Plan-Neofluar 20x/0.5 objective.

Scanning electron microscope (SEM) imaging of thrombi and fibrin

The RBC-fibrinogen mixtures were clotted with 16 nM thrombin in a total volume of 100 μl and after 30 min clots were placed into 10 mL 100 mM Na-cacodylate pH 7.2 buffer for 24 h at 4 °C. Following repeated washes with the same buffer, samples were fixed in 1 %(v/v) glutaraldehyde for 16 h. The fixed samples were dehydrated in a series of ethanol dilutions (20 – 96 %(v/v)), 1:1 mixture of 96 %(v/v) ethanol/acetone and pure acetone followed by critical point drying with CO₂ in E3000 Critical Point Drying Apparatus (Quorum Technologies, Newhaven, UK). The specimens were mounted on adhesive carbon discs, sputter coated with gold in SC7620 Sputter Coater (Quorum Technologies, Newhaven, UK) and images were taken with scanning electron microscope EVO40 (Carl Zeiss GmbH, Jena, Germany).

Morphometric analysis of fibrin structure and statistical procedures

The SEM images of thrombi and fibrin were analyzed to determine the diameter of the fibrin fibers using self-designed scripts running under the Image Processing Toolbox v. 7.0 of Matlab 7.10.0.499 (R2010a) (The Mathworks, Natick, MA). For the diameter measurements a grid was drawn over the image with 10-15 equally-spaced horizontal lines and all fibers crossed by them were included in the analysis. The diameters were measured by manually placing the pointer of the Distance tool over the endpoints of transverse cross-sections of 300 fibers from each image (always perpendicularly to the longitudinal axis of the fibers). The distribution of the data on fiber diameter was analyzed using an algorithm described previously for identification of a theoretical distribution that gives the best global fit to several empirical data sets.^20,21 The best fitted distributions for different samples were compared using Kuiper test and p-values were calculated with Monte Carlo simulation procedures. When statistically significant difference between two distributions was established,^20, the numerical characteristics of the central tendency and variance were considered to be statistically significant. The statistical hypothesis testing for differences in other experimental measurements in this report was performed with Kolmogorov-Smirnov test (Statistical Toolbox 7.3 of Matlab).

Results

Thrombi formed in vivo under similar rheological conditions in large arteries showed remarkable differences in their RBC content (online Supplemental Material, Fig. I, please see http://atvb.ahajournals.org). Even within a single thrombus, regions of high and low RBC count could be identified indicating the prevalence of local factors (blood vessel geometry, flow pattern) as determinants of thrombus composition. Thus, a correct understanding of the impact of RBC on the lytic susceptibility of such compartmental architecture should be based on data modeling the structure and lysis of fibrin over a range of relevant RBC counts. The present study was restricted to clots formed from purified fibrinogen and washed RBCs. Because of the known compartmentalization of fibrin and cellular components within thrombi^10,22 two extreme model states were evaluated (Fig. 1). At increasing RBC occupancy, identical amounts of fibrin were present either in clots with constant total volume (and consequently increasing extracellular fibrin concentration; Model 1) or in clots with volume expanding proportionally to the increase of RBC content (preserving constant extracellular fibrin concentration; Model 2). In all cases the intercellular fibrin was composed of thinner fibers and smaller pores resulting in a denser network compared to the cell-free fibrin as illustrated for Model 1 in Fig. 2 (online Supplemental Material, Fig. II, please see http://atvb.ahajournals.org).

Fig. 1.

Schemes of the experimental setups used in the evaluation of fibrinolysis in the presence of RBC. The total volume (V_t) of the clot was partially occupied by RBC, whereas fibrinogen (converted to fibrin) occupied the extracellular volume (V_ec) at concentration [Fg]_ec. In both models the total amount of fibrin (V_ec.[Fg]_ec) was identical at any RBC occupancy, which was maintained by the indicated increase in [Fg]_ec at decreasing V_ec in Model 1 or at constant [Fg]_ec and V_ec in Model 2. Because in Model 1 two variables (fibrinogen concentration and RBC occupancy) were changed in parallel, separate cell-free clots were also prepared at identical fibrinogen concentrations and used as a reference for each RBC occupancy to isolate the RBC effects in the experiments performed according to this model.

Fig. 2.

Changes in the fibrin network structure caused by red blood cells and eptifibatide. The SEM images in Panel A illustrate the fibrin structure in clots of identical volume and fibrinogen content in the absence or presence of 20 % RBC (note that expulsion of fibrinogen by RBCs increases the extracellular fibrinogen concentration in this model). Panel B shows fiber diameter measured from the SEM images for a range of RBC-occupancy in the same clot model. Probability density functions (PDF) of the empiric distribution (black histogram) and the fitted lognormal theoretical distribution (grey curves) are presented with indication of the median and the interquartile range (in brackets) of the fitted theoretical distributions. In the presence of RBC the parameters of the fitted distributions of the eptifibatide-free and eptifibatide-treated fibers differ at p<0.001 level (for the RBC-free fibrins the eptifibatide-related difference is not significant, p>0.05). Four replicate samples of each clot type were evaluated in a single global statistical procedure.

The role of the integrin-dependent RBC-fibrinogen interaction in the modification of fibrin structure was approached with the addition of the integrin-blocker eptifibatide (Fig. 2A) at a concentration that has been shown to be efficient in the inhibition of fibrinogen binding to RBC.8 Because the most striking changes in the fibrin structure related to the presence of RBC and eptifibatide appeared to be in the fiber diameter, this parameter was quantitatively evaluated using the morphometric approach illustrated in Fig. 2B. In all cases the increase in the clot occupancy by RBC was coupled to a decrease in the fiber diameter (Table 1). This effect was more pronounced in Model 1 (at increasing extracellular fibrinogen concentrations), where 40 %(v/v) RBC occupancy resulted in a more than 2-fold diameter reduction. Based on these results we propose that in Model 1 the fiber diameter reflects the outcome of two opposing effects; the fiber-thickening effects of increasing fibrinogen concentrations in the absence of RBC (reference values in column 2, Table 1) and the fiber-thinning effect of increasing RBC occupancy at constant fibrinogen concentration (Model 2, Table 1). Although it had no effect on the architecture of clots formed from purified fibrinogen, eptifibatide at 20 μM completely reversed the changes in fiber diameter caused by RBC in Model 2 supporting the role of the specific RBC receptor – fibrinogen interaction in the assembly of the final fibrin structure. In Model 1 the influence of eptifibatide was less pronounced resulting in up to 40 % reduction in the fiber-thinning effect of RBC probably related to the stronger competition on behalf of the increasing fibrinogen concentration in this experimental setup.

Table 1.

Fiber diameter (nm) of fibrin clots with varying red blood cell content

	no eptifibatide			20 μM eptifibatide
	Model 1		Model 2	Model 1		Model 2
volume occupancy	+RBC	reference	+RBC	+RBC	reference	+RBC
10 %	110† (29)	170† (91)	160† (90)	140† (70)	170† (86)	170 (83)
20 %	90* (26)	170 (99)	100* (50)	160* (95)	180 (86)	180* (88)
40 %	96* (31)	200* (100)	92* (33)	140* (87)	210* (110)	170 (86)

Fibrin was prepared and the SEM images were analyzed as illustrated in Fig. 2. The median and interquartile range (in brackets) of the diameter distributions are presented in nm units (4 replicate samples of each clot type were evaluated in a single statistical procedure). Reference values refer to cell-free fibrin prepared from fibrinogen at concentrations equivalent to the extracellular concentrations resulting from the expulsion of fibrinogen by the respective RBC occupancy as described for Model 1 in Methods (8.2, 9.2 and 12.4 μM in the increasing order of RBC occupancy). For Model 2 the extracellular concentration of fibrinogen was always 7.4 μM and thus a single reference value of 150 (71) nm was used in the absence of eptifibatide and 160 (83) nm in the presence of 20 μM eptifibatide. All differences between fiber diameters in the presence of RBC and the respective reference values in Model 1 are significant at p<0.001 level. Additional significant (p<0.001) differences are also indicated between two subsequent values along the columns from top to bottom (*) or in comparison with the cell-free 7.4 μM fibrin (†).

When fibrinogen containing plasminogen is mixed with thrombin and tPA, following clotting the plasminogen activator is uniformly dispersed in the total volume of the clot and the generated plasmin dissolves the fibrin matrix. As a result of proteolysis fibrin loses its mechanical stability and thus a steel ball placed on the top of the clot will descend in response to gravity. This global lytic assay system is intimately related to the biological function of fibrin, because it reflects the changes in the mechanical stability of the fibrin matrix. The time to achieve complete collapse of the clot can be used as a global end-point indicator of this assay. If identical quantities of fibrin are degraded, this lysis time depends on the combined effect of several factors (stability of the initial fibrin network, rate of plasminogen activation, catalytic efficiency of plasmin in various fibrin structures). Although the constant amount of fibrin was maintained differently in our Models 1 and 2 defined above, in both situations the presence of 10 and 40 %(v/v) RBC slowed down the fibrinolysis induced by incorporated tPA and eptifibatide partially or completely reversed the RBC effect (Fig. 3). In view of our recent data on the differential impact of fibrin structure on plasminogen activation and fibrin dissolution¹⁹ the discordant response of the 20 %(v/v) RBC-fibrin clots in this complex fibrinolytic assay indicated the necessity to dissect the role of RBC in these two stages of the process.

Fig. 3.

Fibrinolysis induced by clot-incorporated tPA in the presence of red blood cells. Clots of various RBC-occupancy contained 5 mg fibrin at constant volume and increasing extracellular concentrations (A) or 2 mg fibrin at increasing volumes and constant extracellular concentration (B) as well as plasminogen and tPA at identical extracellular concentrations as detailed in Methods. After 15-s clotting a steel ball was placed on the surface of the clots and the lysis time was measured in the absence (black bars) or in the presence of eptifibatide (grey bars). Data are presented as mean and SD (n=4), asterisks indicate significant (p<0.05) difference from the fibrin without RBC, cross signs indicate significant (p<0.05) difference of eptifibatide-free and eptifibatide-treated fibrins with identical RBC content.

When tPA was applied to the surface of fibrin containing plasminogen and increasing number of RBC, plasmin was generated on a fibrin surface with decreasing area (part of the interface was occupied by RBC) and varying structure (Fig. 2, Table 1), which is also known to affect plasminogen activation.¹⁹ Under the conditions of Model 1 after 30-min activation on the surface of fibrin containing 10 - 20 %(v/v) RBC, the plasmin level peaks up at 36.6 - 44.6 % over the cell-free control values, but further increase in the RBC content moderates this stimulation to 25.6 % at 40 %(v/v) RBC (Fig. 4A). It is noteworthy that on the surface of cell-free fibrin formed from fibrinogen at concentrations equivalent to the extracellular values in Model 1, plasmin generation decreased with increasing fibrinogen concentrations (at 12.4 μM fibrinogen 22.2 % less plasmin formed than at 7.4 μM fibrinogen, data not shown). In contrast to the bell-shaped dependence of plasminogen activation on the RBC count in Model 1, a linear decrease in the amount of generated plasmin could be observed at increasing RBC occupancy and constant extracellular fibrinogen concentration (Fig. 4B). Because the RBC-occupied area of the activation interface is the same in both models for identical volume occupancy, the differences in plasmin generation can be attributed to the variations in fibrin structure. In line with this eptifibatide at 20 μM moderated the effect of the highest RBC counts in both models of fibrin-dependent plasminogen activation (in the absence of fibrin eptifibatide modestly stimulated the tPA-induced plasminogen activation resulting in 20 % more plasmin at 30 min, data not shown).

Fig. 4.

Plasminogen activation on the surface of fibrin containing red blood cells. Plasmin activity (ΔA₄₀₅/min) was measured after 30 min activation by tPA applied to the surface of clots with embedded plasminogen as described in Methods. The same RBC-fibrin model clots were used as in Fig. 3 (A, varying extracellular fibrinogen concentration; B, constant fibrinogen concentration). In panel A only the cell-free control for 9.2 μM fibrin concentration corresponding to 20 % RBC clot-occupancy is shown, whereas in panel B the control refers to 7.4 μM fibrin. Data are presented as mean and SD (n=4) in the absence (black bars) or presence of 20 μM eptifibatide (grey bars), asterisks indicate significant (p<0.05) difference from the fibrin without RBC, cross signs indicate significant (p<0.05) difference of eptifibatide-free and eptifibatide-treated fibrins with identical RBC content.

The interplay of tPA, fibrin and RBC at the fluid phase-clot interface was approached with confocal microscopic observation of tPA-GFP penetration in the clot, propagation of the lytic front of orange fluorescent labeled fibrin and release of fluorescent labeled RBC from the clot. In agreement of our and others’ earlier findings^19,23 within 10 min after application of tPA-GFP to the surface of pre-formed fibrin a sharp zone of intense tPA-related fluorescence was formed on the surface of cell-free fibrin (Fig. 5A) and preserved throughout the lysis observation period (Fig. 5B). In contrast, the presence of RBC prevented this accumulation of tPA (Fig. 5A&B). The progress of fibrin dissolution could be monitored based on the position of the boundary layer of the fluorescent fibrin phase (Fig. 5C). The presence of RBC at 20-40 % in the clot significantly slowed down the dissolution of fibrin after 25-min exposure to tPA with the greatest difference at 30 min (3.5-fold shorter distance run by the lytic front) followed by a catch-up phase, but the difference remaining almost 2-fold at 55 min. Eptifibatide did not affect the progress of lysis in the absence of RBC, but it moderated the inhibiting effect of RBC. The release of RBC from the clot occurred in parallel with the progress of the lytic process as evidenced by the lack of RBC in the fluid phase in the early stage of dissolution (Fig. 5A) and their abundance out of the clot at later stages (Fig. 5B).

Fig. 5.

Effect of red blood cells on fibrinolysis induced by tPA applied to the surface of clots. Fibrin clots were prepared from fibrinogen containing Alexa⁵⁴⁶-labeled fibrinogen, Vybrant DiD-labeled RBCs and plasminogen. Following clotting tPA-GFP was added to fibrin and the fluid/fibrin interface was monitored by confocal laser scanning microscopy using triple fluorescent tracing at the indicated wavelengths. The images were taken 5 (A) and 40 (B) min after the addition of tPA-GFP (the tPA signal is shown in green, the fibrin in red and the RBC in white). Panel C shows the distance run by the fibrin boundary layer in the course of lysis (mean and SD of three measurements). The numbers next to the curves in the same color indicate the volume occupancy by RBC. In some experiments fibrin contained 20 μM eptifibatide (dashed lines).

Discussion

The hemostatic function of whole blood is a delicate balance of coagulation and fibrinolytic pathways, which is profoundly affected by RBC. While severely anemic patients show a bleeding trend, which can be reversed with transfusion of washed RBC,² the increased RBC mass of polycythemia vera patients is coupled to frequent (in more than one third of the patients) thrombotic complications.²⁴ However, the local amount of RBC in thrombi does not necessarily correlate with the systemic RBC count, because some complicating factors such as flow conditions and variable geometry, especially at stenotic sites, will influence the distribution of the cellular elements in the lumen of blood vessels.²⁵ Thus not surprisingly, variable amounts of RBCs can be detected in thrombi removed surgically from large arteries (online Supplemental Material, Fig. I, please see http://atvb.ahajournals.org). The major goal of the present study was to evaluate the contribution of RBCs to clot stability from the aspect of fibrinolysis. Because in a plasma environment RBCs could affect the final clot structure through interactions at multiple steps of the coagulation process (e.g. exposing negatively charged phospholipids that provide surface for the assembly of coagulation factor complexes)²⁶ and in order to avoid the interference of plasma inhibitors, we evaluated clots prepared from purified fibrinogen and washed RBC. Thus, initiating dissolution with plasminogen and tPA at precisely known concentrations a relatively limited number of variables needed to be considered in the evaluation of the fibrinolytic side of the hemostatic balance. The study of clot dissolution was approached in two experimental setups. In the model with incorporated tPA, the activator was uniformly dispersed within the RBC-fibrin clots at concentration (1 nM), which is relevant to the amount of tPA measured in human arterial thromb,i²⁷ thus mimicking the situation in hemostatic plugs at sites of vascular injury when coagulation and fibrinolysis are initiated simultaneously. In the second setup tPA was applied to the surface of pre-formed RBC-fibrin clots at concentration (60 nM), relevant to the values maintained in blood in the course of enzymatic thrombolysis,²⁸ thus mimicking the therapeutic systemic administration of plasminogen activators. Despite the differences in the experimental approaches, in both settings the effect of RBCs was consistent; increasing RBC occupancy of the clot resulted in resistance to lysis (Fig. 3, Fig. 5C). This anti-fibrinolytic effect of RBCs, which is expressed at physiologically relevant RBC counts, shifts the balance of coagulation/fibrinolysis in favor of clotting and may prevent the premature dissolution of hemostatic plugs or contribute to the thrombotic complications in polycythemia vera. In addition, the heterogeneous RBC composition of thrombi (online Supplemental Material, Fig. I, please see http://atvb.ahajournals.org) may be one of the factors contributing to the variability in the lytic susceptibility during thrombolytic therapy of myocardial infarction and stroke.

The mechanism of RBC-related fibrinolytic resistance was approached using with ultrastructural and kinetic methods. SEM imaging of the fibrin structures formed in the presence RBCs evidenced thinner fibers in the RBC-rich areas (Fig. 2). These measurements were performed in two extreme settings; either keeping the total amount of fibrin constant with consequent increase in the extracellular fibrinogen concentration in parallel with the RBC-excluded volume or at constant extracellular fibrinogen concentration for any RBC occupancy (Fig. 1). In the second setting a dose-dependent decrease in fiber diameter was observed at increasing RBC occupancy of the clot (Table 1), whereas in the first setting this effect was counteracted by the thickening effect of rising fibrinogen concentration (Table 1, reference values) with the outcome seen in Fig. 2B and Table 1 (Model 1, +RBC values). The lysis time measured in the ball sedimentation assay (Fig. 3) inversely correlates with the established changes in fiber diameter and not directly with the RBC occupancy of the clots. When eptifibatide, a blocker of the RBC fibrinogen receptor, reverses the RBC effect on the fiber structure (Fig. 2, Table 1), it also reverses the RBC-related inhibition of fibrinolysis (Figs. 3, 5C). Thus, RBC-induced modification of fibrin structure appears to be a major mechanism in the lytic resistance of clots with RBC content. These results are in agreement with earlier findings that if the fiber diameter is reduced by different factors, the overall rate of fibrin dissolution is slower in similar lytic models with clot-incorporated^19,29 or surface-applied³⁰ activators.

Because the lytic assays discussed above reflect a global measure of the combined outcome of the two stages of fibrinolysis (plasminogen activation and proteolysis of fibrin), separate kinetic and confocal microscopic measurements were performed in attempt to identify discrete effects of RBC. The plasmin generated in the course of plasminogen activation on the surface of clots in Model 1 increased in the presence of RBCs in the clot (Fig. 4A), which is surprising in view of the RBC-related suppression of fibrin dissolution. Furthermore, plasminogen activation on the surface of clots in Model 2 shows a dose-dependent decreasing trend at rising RBC occupancy (Fig. 4B). These discordant findings can be explained by the simultaneous action of two opposing factors. Firstly, greater RBC occupancy of clots results in thinner fibers (Table 1), which are a better template for tPA-catalyzed plasminogen activation.¹⁹ Secondly, more RBCs in the clot volume result in more cells on the clot surface with a consequent relative increase in the area devoid of plasminogen and inert in terms of plasmin generation. The reduction in interface area occupied by plasminogen-presenting fibrin translates into less plasmin generation. In the setting of the Fig. 4A assay, the thinner fiber network gains higher density at increasing fibrinogen concentrations and thus, the effect of template quality coupled to higher cofactor concentration dominates. In the setting of a constant fibrinogen concentration (Fig. 4B), increasing the surface taken by RBC limits plasminogen activation despite the thinner fibers. Independently of the variations in plasmin generation, in both setups RBCs slow down fibrin dissolution (Fig. 3), which suggests that the plasmin-catalyzed proteolytic stage prevails in the control of fibrinolysis at increasing RBC occupancy. The confocal images of fibrin exposed to tPA-GFP (Fig. 5A&B) provide direct evidence for the limited access of tPA to activatable substrate; the accumulation of tPA in the interfacial fibrin layer is definitely retarded in the presence of RBCs.

This interpretation of the observed changes in plasminogen activation on the surface of RBC-fibrin clots is further supported by the effects of eptifibatide in these assays (Fig. 4). This blocker of the RBC fibrinogen receptor increases the fiber diameter (Table 1) and thus eliminates RBC-related changes in the structure of the fibrin template that favor plasminogen activation¹⁹ leaving unopposed the effect of the expanding RBC-occupied surface with consequent decline in plasminogen activation. In the absence of fibrin, eptifibatide stimulates plasminogen activation (data not shown) and this effect persists if activation occurs on the surface of cell-free fibrin with less efficient template function (thicker fibers at 8.2 μM or higher fibrinogen concentration, Fig. 4A), but not on the thinner fibers with better cofactor function in the tPA-catalyzed plasminogen activation (Fig. 4B). The reported observations for the generation of plasmin in the presence of RBCs indicate that binding of the receptor blocker to RBCs eliminates its direct effect on plasminogen activation and likewise moderates the RBC effects. The data on the effects of eptifibatide in plasminogen activation (Fig. 4) and fibrinolytic (Fig. 3) assays suggest that the RBC fibrinogen receptor mediates at least in part the RBC-related lytic resistance of fibrin.

In summary, our present report unravels a clot-stabilizing function for RBCs based on suppressed tPA-induced fibrinolysis in RBC-modified fibrin structures. In addition to steric hindrance, RBCs modulate fibrinolysis through a specific fibrinogen receptor as evidenced by the effects of eptifibatide, a cyclic heptapeptide that reversibly inhibits the binding of fibrinogen to α_IIbβ₃ integrin. This α_IIbβ₃ antagonist is used in clinical practice as an antiplatelet agent,³¹ but its newly described role in RBC-containing clots extends our understanding of its favorable antithrombotic action by overcoming the resistance to lysis of RBC-rich thrombi.