Antiadhesive effect of fibrinogen: a safeguard for thrombus stability

The recruitment of phagocytic leukocytes to sites of injured vessel wall plays an important role in thrombus remodeling during normal vascular repair and in the pathophysiology of thrombosis. Fibrinogen and fibrin, present in the thrombus, function as potent adhesive substrates for neutrophils and monocytes. They support cellular attachment by binding cell-surface receptors that belong to the β₂ subfamily of integrins, including α_Mβ₂ (Mac-1, CD11b/CD18), α_Xβ₂ (CD11c/CD18), and α_Dβ₂ (CD11d/CD18).^1-3  Furthermore, integrin α₅β₁ on neutrophils contributes to adhesion to fibrin.⁴  Numerous in vitro studies performed under static and flow conditions demonstrated that either stimulated or unstimulated neutrophils and monocytes adhere to the surface-bound fibrinogen and fibrin with alacrity.^5-7  Furthermore, stimulated cells are capable of binding soluble plasma fibrinogen, whereas integrins expressed on resting leukocytes do not bind soluble protein.^1,8,9  The prerequisite for its binding is activation of leukocytes with agonists that induce conformational changes in integrins and their transition into a high-affinity state.¹⁰  The ability of surface-bound fibrinogen to support adhesion of resting leukocytes, as well as of other cells such as platelets, is incompletely understood but is believed to derive from its capacity to induce integrin clustering. In addition, the conformational changes in fibrinogen, which occur on its immobilization or transformation into fibrin, and which are associated with unmasking of integrin binding sites, can contribute to adhesion of unstimulated cells.¹¹ 

Although fibrin clots and deposits of fibrinogen are highly adhesive for resting leukocytes in vitro, they seem to be much less adhesive in vivo, at least in the first hours after thrombus formation. Several studies using experimental models of thrombosis have shown that few leukocytes adhere and invade the thrombus within the first hours following its formation.^12,13  Neutrophil infiltration was found after 3 hours and slowly increased during the next 24 hours. Notably, first monocytes arrived after 1 day, and then their numbers increased steadily. Complete investment of thrombi by flattened monocytes occurred after 4 days.¹²  It is well documented that neutrophils and monocytes possess a high fibrinolytic potential. They contain nonplasmin fibrinolytic proteases^14,15  and also assemble on their surface the components of the plasminogen activator system which leads to the elaboration of plasmin.¹⁶  Moreover, neutrophils and monocytes actively phagocytose fibrin and fibrinogen in a process partially dependent on α_Mβ₂.^17,18  Consequently, if adhesion of circulating neutrophils and monocytes to the thrombus occurred rapidly, it would likely contribute to its premature dissolution when the plug stability is critical for arresting bleeding and wound repair. Therefore, mechanisms protecting the hemostatic plug from premature weakening might exist. However, although the search for effective ways to disintegrate thrombi or prevent their formation has been the main theme of numerous studies, the possibility of physiologic mechanisms protecting thrombi from premature weakening has never been envisioned.

Adhesive interactions of neutrophils and monocytes with polymerized fibrin in vivo occur in the presence of high concentrations of circulating plasma fibrinogen (∼2-4 mg/mL). One important property of fibrinogen that would have a major bearing on leukocyte adhesion is its capacity to form complexes with fibrin molecules.¹⁹  Therefore, by virtue of its binding to fibrin and/or immobilized fibrinogen, soluble plasma fibrinogen can influence leukocyte adhesion to these substrates. In this study, we show that soluble fibrinogen has a potent antiadhesive effect. We also provide evidence that it inhibits leukocyte adhesion by binding to the fibrin clot and immobilized fibrinogen but not to leukocyte β₂ integrins. The antiadhesive effect of fibrinogen is shown with various β₂-integrin–expressing cells, including neutrophils and monocytes. Furthermore, soluble fibrinogen protects the fibrin clot from leukocyte attachment under both static and flow conditions. Thus, a new mechanism regulating leukocyte adhesion has been identified which may have broad pathophysiologic implications.

Human thrombin, factor XIIIa, and fibrinogen, depleted of fibronectin and plasminogen, were obtained from Enzyme Research Laboratories (South Bend, IN). Fibrinogen was treated with 40 mM iodoacetamide for 30 minutes at 37°C to inactivate the residual factor XIII and then dialyzed against phosphate buffered saline (PBS). Fibrinogen was labeled with 125-Iodine using IODO-GEN (Pierce, Rockford, IL), dialyzed against PBS and stored at −20°C. Fibrin-monomer devoid of fibrinopeptides A and B was prepared by clotting fibrinogen with thrombin and dissolving the fibrin clot in 0.02 M acetic acid.²⁰  Fibrinogen and fibrin-monomer concentrations were determined by spectophotometry at 280 nm using the adsorption coefficient 1.56 at 1 mg/mL. Phenylalanyl-l-prolyl-l-arginine chlormethyl ketone (PPACK) and polyvinylpyrrolidone (PVP) were from Sigma (St Louis, MO). The chromogenic thrombin substrate S-2238 was from Chromogenix Diapharma Group (Franklin, OH). Calcein AM was purchased from Molecular Probes (Eugene, OR). mAb IB4 against the β₂-integrin subunit was purified from conditioned media of the hybridoma cell line obtained from ATCC (Manassas, VA). Anti-α₅β₁ polyclonal antibody (1950) was from Chemicon (Temecula, CA). mAb CBRM1/5 against the activation epitope in the α_MI-domain of α_Mβ₂ was from eBioscience (San Diego, CA).

Human embryonic kidney (HEK) cells expressing α_Mβ₂ were described previously.²¹  Cells were maintained in DMEM/F-12 (BioWhittaker, Walkersville, MD) supplemented with 10% FBS and 25 mM HEPES. U937 monocytoid cells and mouse macrophage cell line IC-21 were obtained from ATCC and cultured in RPMI 1640 supplemented with 10% FBS. Neutrophils were isolated under sterile conditions from peripheral blood obtained from consenting volunteers and anticoagulated with acid-citrate-dextrose. Isolations were performed using density centrifugation on Ficoll-Hypaque Plus (Amersham Biosciences, Piscataway, NJ), followed by dextran sedimentation of erythrocytes and hypotonic lysis of residual erythrocytes.²²  The study was approved by the Cleveland Clnic Foundation.

The wells of 96-well polysterene microtiter plates (Immulon 4HBX; ThermoLabsystems, Franklin, MA) were coated with various concentrations of fibrinogen or fibrin-monomer for different times at 37°C and coated afterward with 1.0% PVP for 1 hour at 37°C. The fibrin gels were formed in siliconized Immulon 2 96-well format strips by mixing 100 μL of 2 mg/mL fibrinogen in Hanks balanced salt solution (HBSS) with 0.5 U/mL thrombin for 2 hours at 37°C. The cells were labeled with 10 μM calcein (Molecular Probes) for 30 minutes and washed twice with HBSS + 0.1% BSA, and assays were performed as described previously.²¹  Briefly, aliquots (100 μL) of the labeled cells (3-5 × 10⁴/mL) were added to each well. For inhibition experiments, cells were mixed with soluble fibrinogen before they were added to the wells. Alternatively, the wells coated with immobilized fibrin(ogen) or containing fibrin gels were incubated with different concentrations of soluble fibrinogen for 15 minutes at 37°C, after which fibrinogen solutions were aspirated and cells were added. After 30 minutes of incubation at 37°C, the nonadherent cells were removed by 2 washes with PBS. Fluorescence was measured in a CytoFluorII fluorescence plate reader (Applied Biosystems, Foster City, CA).

Perfusions with monocytes and neutrophils were performed in a flow chamber consisting of a transparent glass tube (50 × 2.5 mm) lined with the fibrin gel which forms a fabricated capillary (flow tube). To produce this configuration, the fibrin gel was formed by pouring a mixture consisting of 2 mg/mL fibrinogen and 0.5 U/mL thrombin in the glass tube with an inserted metal rod (70 × 0.9 mm). After 2 hours at 22°C, the metal rod was removed, and the tubes with the preformed capillary inside the fibrin gel were filled with 20 μM PPACK for 30 minutes at 22°C to inhibit thrombin, followed by flushing with HBSS buffer containing 0.1% BSA. Suspensions of calcein-labeled U937 cells or neutrophils (10⁶/mL) in HBSS at 37°C were drawn from a reservoir through the fibrin flow tube by a Harvard syringe pump (Harvard Apparatus, South Natick, MA). After perfusions, the system was rinsed with HBSS + 0.1% BSA, and fluorescence of adherent neutrophils or monocytes was measured. To accomplish this, the flow tube was placed in a holder for mounting of 96-well microtiter well strips, and fluorescence was detected using a CytoFluorII plate reader (Applied Biosystems). The wall shear rate was calculated as described.²³  Adherent cells within the flow tubes were visualized using a fluorescence microscope (Leica DMIRB, Allendale, NJ). To determine neutrophil and U937 cell activation, cells perfused through the fibrin flow tubes were collected by centrifugation and incubated 30 minutes on ice with 10 μg mAb CBRM1/5. Cells were washed and stained with Alexa 488–conjugated goat antimouse secondary antibody. Cells were analyzed by flow cytometry using a FACScan instrument (Becton Dickinson, Franklin Lakes, NJ). Total β₂-integrin levels were measured using mAb IB4.

The dissociation constants of the fibrin-fibrinogen and fibrinogen-fibrinogen interactions were measured by using a Biacore 3000 SPR-based biosensor (Biacore AB, Uppsala, Sweden). Intact fibrinogen was immobilized on the CM5 chips (Biacore) at a density of approximately 1000 to 1200 response units according to the manufacturer's protocol. To convert fibrinogen to fibrin, 1 U/mL thrombin was flowed over fibrinogen for 30 minutes at 5 μL/minute followed by thrombin inhibitor (PPACK, 30 minutes at 5 μL/minute). To prepare its conformationally altered form, fibrinogen was denatured by incubating with 6 M urea for 30 minutes at 22°C and then coupled to the chip. Alternatively, fibrinogen immobilized on the chip was modified by two 3-minute pulses of 0.5% SDS. Different concentrations of fibrinogen in HBS-P buffer (Biacore) supplemented with 1 mM CaCl₂ were flowed over flow cells coated with fibrinogens. All data were corrected for the response obtained using a blank reference flow cell that was activated with N-ethyl-N′-(dimethylaminopropyl)carbodiimide/N-hydroxycuccinimide and then blocked with ethanolamine. Steady-state experiments were performed by injecting fibrinogen at 10 μL/mL for 8 minutes. The chip surface was regenerated with 10 mM citrate buffer, pH 4.0, which dissociates fibrin-fibrinogen complexes.²⁴  Data were analyzed using the program BIAevaluation 4.01 (Biacore). Binding constants were calculated from the binding isotherms which were constructed using the equilibrium portions of SPR sensograms.

Previous studies with leukocytes and the α_Mβ₂-expressing HEK 293 cells have demonstrated that fibrinogen adsorbed on surfaces is capable of supporting adhesion of unstimulated cells by β₂ integrins with α_Mβ₂ playing the central role.^2,5,25  Although soluble fibrinogen does not bind resting cells in suspension,^8,9  it was an efficient inhibitor of adhesion of α_Mβ₂-bearing cells to immobilized fibrinogen (Figure 1A). At 100 μg/mL (0.29 μM), fibrinogen inhibited adhesion of U937 monocytoid and the α_Mβ₂-expressing HEK 293 cells by greater than 95% and 60%, respectively. It also completely blocked cell adhesion to immobilized fibrin-monomer (not shown). Because nonactivated integrins do not bind soluble fibrinogen, we have sought the molecular basis for its blocking effect. It is well documented that fibrinogen forms complexes with fibrin-monomer and polymerized fibrin by the binding between complementary sites present in the D and E domains of these molecules.^19,26  Because immobilized fibrinogen is believed to mimic fibrin, we have hypothesized that binding of soluble fibrinogen to the fibrin(ogen) substrate, rather than to receptors on the cell surface, is responsible for its capacity to inhibit adhesion. To explore this possibility, the microtiter wells were coated with fibrinogen or fibrin-monomer, coated afterward with PVP, and then incubated with different concentrations of soluble fibrinogen. After 15 minutes at 37°C, fibrinogen solutions were aspirated, and adhesion of U937 or α_Mβ₂-expressing HEK 293 cells was assessed. Increasing concentrations of soluble fibrinogen added to immobilized fibrin(ogen) progressively inhibited cell adhesion (Figure 1B).

Fibrinogen produced 50% inhibition of cell adhesion to immobilized fibrinogen and fibrin-monomer at approximately 10 and 2.5 μg/mL, respectively. The inhibitory effect was also observed with neutrophils: preincubation of immobilized fibrinogen with 50 μg/mL soluble fibrinogen inhibited neutrophil adhesion by 65% (not shown). Figure 2 shows that there was a reverse time-dependent correlation between binding of soluble fibrinogen to immobilized fibrinogen and the loss of cell adhesion; that is, the increase in the amount of bound ¹²⁵I-labeled fibrinogen paralleled a decrease in the number of adherent U937 cells. Thus, these results suggest that the interactions between soluble fibrinogen and its immobilized form or fibrin-monomer blocked cell adhesion.

Although the ability of soluble fibrinogen to form complexes with fibrin and its capacity to undergo self-association in solution under special conditions is well characterized,^19,27  little is known about the capacity of immobilized fibrinogen to form a complex with soluble fibrinogen. Numerous studies demonstrated that fibrinogen adsorbed on the surfaces or deposited in the extracellular matrix undergoes conformational changes and acquires properties of fibrin manifesting in binding of fibrin-specific mAbs and other molecules.^28-30  Unfolding of fibrinogen on its immobilization may also generate the fibrinogen-binding activity. However, despite the prevalence of this premise, there has been no investigation of fibrinogen binding to its conformationally altered form. To characterize fibrinogen interactions, we determined the dissociation constants of its complexes using SPR. Intact fibrinogen or fibrin were coupled to the chip, and the SPR profiles across a range of fibrinogen concentrations (0-25 μM) flowed over the protein surfaces were examined. No binding of soluble fibrinogen to intact fibrinogen was detected. At the same time, fibrin on the chip bound soluble fibrinogen in a dose-dependent and saturable manner with a K_D of 1.47 ± 0.26 μM as determined from the binding data (Figure 3A-B).

Because protein coupling to the dextran surface of the chip does not induce gross conformational alterations, fibrinogen on the chip was treated with 0.5% SDS, and its ability to bind intact soluble fibrinogen was tested. The treatment of fibrinogen rendered it competent to bind soluble fibrinogen (not shown). Analyses of the binding data demonstrated that fibrinogen bound with a K_D of approximately 2.1 ± 0.4 μM. Similar results were obtained when fibrinogen was treated with 6 M urea and then coupled to the chip (K_D = 1.1 ± 0.3 μM). Although the conformational changes in fibrinogen induced by its interactions with plastic can vary from those induced by denaturating agents, these studies indicate that the unfolded molecule is capable of associating with its intact soluble form, and that binding parameters are consistent with those for binding of fibrinogen to fibrin.

We³¹  and others³²  have previously demonstrated that the pattern of adhesion of the α_Mβ₂-expressing HEK 293 cells to increasing coating concentrations of fibrinogen is anomalous. Whereas cell adhesion was dependent on fibrinogen coating concentrations at low doses and reached a maximum at 1 to 3 μg/mL, further increases in fibrinogen concentration resulted in a precipitous decline in the number of adherent cells, and surfaces coated with fibrinogen above 5 to 10 μg/mL were nonadhesive. To test whether natural α_Mβ₂-expressing leukocytes exhibit the same behavior, adhesion of neutrophils, U937 monocytoid cells, and mouse macrophage cell line IC-21 was examined. Microtiter wells were coated with different concentrations of fibrinogen for 3 hours at 37°C, and cell adhesion was determined. Figure 4A shows that natural leukocytes adhered to increasing concentrations of fibrinogen in a way similar to that of recombinant α_Mβ₂-expressing HEK 293 cells. These experiments demonstrated that abnormal adhesion at high-coating concentrations of fibrinogen is not a property of the α_Mβ₂-transfected cells and indicated that the mechanism underlying this phenomenon does not depend on the source of adhesion receptors. Also, this pattern was unique for fibrinogen because cell adhesion to other proteins follows a normal adsorption isotherm as a function of their coating concentration (shown in Figure 4A, adhesion of U937 cells to fibronectin). Moreover, when the amount of fibrinogen bound to the surface was quantified using radiolabeled protein, the results demonstrated a normal dose-dependent adsorption of fibrinogen on the plastic (Figure 4B). Notable, however, was the lack of clear saturation of fibrinogen adsorption. Thus, these results indicate that progressive binding of fibrinogen is correlated with the dramatic loss of cell adhesion.

Collectively, the results shown in Figures 1 to 4 suggested that one explanation for the antiadhesive effect of fibrinogen immobilized at higher coating concentrations is that, during adsorption, it binds not only to plastic but also forms complexes with the molecules tightly attached to plastic. Such fibrinogen-fibrin(ogen) complexes are then incapable of sustaining firm adhesion when shearing forces, generated during washing procedures typically used in static adhesion assays, are applied. To address the possibility that complex formation is responsible for the observed effect, we measured cell adhesion to 2 different concentrations of fibrinogen, “nonadhesive” (5 μg/mL) and “adhesive” (2 μg/mL), coated onto the plastic surface for various periods of time. If complex formation occurs, it should be a time-dependent process.

As shown in Figure 5A, the coating concentration of 5 μg/mL became adhesive when the protein was allowed to adsorb for 30 minutes. However, the prolonged coating times resulted in the gradual decrease of cell adhesion, and immobilization for 3 hours rendered the substrate essentially nonadhesive. Likewise, the prolonged immobilization of the adhesive concentration of fibrinogen (2 μg/mL) decreased its adhesive potency. These results indicate that a position of the adhesion peak depends not only on the coating concentration but on the coating time as well. They further suggest that short incubations of fibrinogen allow direct and firm attachment of molecules to plastic, whereas prolonged times lead to binding of additional molecules that associate with those that were preadsorbed to the surface.

To further investigate the effect of fibrinogen self-association on cell adhesion, we immobilized fibrinogen in the presence of agents which prevent complex formation. One could expect that immobilization of fibrinogen in the presence of dissociating agents should allow only direct adsorption of molecules on plastic and, thus, lead to normalization of the pattern of cell adhesion. Indeed, coating of fibrinogen in the presence of 0.015% SDS or 6 M guanidine hydrochloride resulted in the disappearance of the peak and produced a dose-dependent and saturable isotherm of adhesion (Figure 5B). Thus, these results suggest that fibrinogen molecules that were firmly immobilized on plastic are capable of supporting normal concentration-dependent adhesion.

To rule out the possible influence of plastic on the conformation of adsorbed fibrinogen and to explore a model which is more representative of physiologic conditions, we examined whether soluble fibrinogen exerts its antiadhesive effect on cell attachment to the fibrin clot. In these experiments, the fibrin gels were prepared from fibrinogen in which traces of factor XIII were inactivated with iodacetamide, whereas after polymerization the thrombin activity was quenched by addition of PPACK. Thrombin inactivation was verified with its chromogenic substrate S-2238. The fibrin gels were overlayed with solutions containing different concentrations of soluble fibrinogen for 15 minutes at 37°C, solutions above the gels were aspirated, and adhesion of U937 cells was examined. The increasing concentrations of fibrinogen inhibited cell adhesion to the fibrin gel in a dose-dependent manner (Figure 6A, closed circles). At 100 μg/mL, added fibrinogen blocked adhesion by approximately 70% compared with nontreated gels. We next sought to determine whether the antiadhesive effect of fibrinogen resulted from its loose association with the surface of a fibrin matrix. Factor XIIIa is known to introduce covalent cross-links between fibrinogen and fibrin molecules.³³  If the highly adhesive fibrin clot becomes “nonsticky” as a result of the fragile layer of loose fibrinogen-fibrin complexes formed on its surface, then factor XIIIa should eliminate the antiadhesive effect of soluble fibrinogen. To verify this, factor XIIIa was included in the mixture for fibrin polymerization, and, after the gel was formed, exogenous soluble fibrinogen was added. After incubation for 15 minutes at 37°C, fibrinogen solutions above the gels were removed, and the gels were incubated for an additional 2 hours. Incubations of fibrinogen with fibrin in the presence of FXIIIa resulted in cross-linking of fibrinogen as demonstrated by formation of the γ-γ dimer and cross-linked α/γ-polymers (Figure 6B). As shown in Figure 6A (open circles), no significant inhibition of U937 cell adhesion in the presence of factor XIIIa was found. These experiments suggest that stable covalent complexes between soluble fibrinogen and the surface of the fibrin clot prevent cell detachment and that they are, in fact, as adhesive as fibrin itself.

A critical difference between adhesion assays in the experiments described above and cell adhesion in vivo is that cell interactions with the adhesive substrate in blood vessels take place under permanent shear forces developed in circulation. In static adhesion assays, a shear force is applied during the washing procedure only. Therefore, a transient shear force affects cells that have already adhered to or are in the process of their attachment to the substrate. To approximate our model to physiologic conditions, we used a flow chamber which consists of a capillary preformed within the fibrin gel. In this assay, calcein-labeled U937 cells or neutrophils were perfused through the capillary for 20 minutes at a wall shear rate of 50 sec⁻¹ or for 3 minutes at 100 sec⁻¹, respectively. Cell adhesion was observed in a microscope and quantified by measuring fluorescence. Consistent with previous studies, cell adhesion under these conditions was dependent on β₂⁶ and β₁⁴ integrins as the combination of anti-β₂ mAb IB4, and anti-α₅β₁ polyclonal antibody 1950 inhibited adhesion completely (not shown). Figure 7A shows fluorescent microscopy of the gel capillaries after perfusion of labeled U937 cells and neutrophils in either the absence or presence of fibrinogen in the cell suspension. At the physiologic concentration of 2 mg/mL, fibrinogen decreased the number of adherent neutrophils and monocytes by approximately 4- and 10-fold, respectively (Figure 7B). Flow cytometry analyses of cells before and after perfusions detected no activation of α_Mβ₂ as determined with the conformation-sensitive mAb CBRM1/5. Likewise, there was no up-regulation of β₂ integrins as determined using mAb IB4, which recognizes the common β₂ integrin subunit.

In this study, we demonstrate that soluble fibrinogen has a powerful antiadhesive effect and propose a mechanism whereby this protein inhibits integrin-mediated adhesion of leukocytes to the fibrin clot and immobilized fibrinogen. Accordingly, soluble fibrinogen reduces leukocyte adhesion by forming complexes with fibrin(ogen) substrates rather than blocking leukocyte integrins. The inhibitory effect of soluble fibrinogen is demonstrable both under static and flow conditions. Several lines of data support the concept that the antiadhesive effect of fibrinogen is mediated by its interactions with the substrate. First, the pretreatment of the immobilized fibrinogen and fibrin clot with soluble fibrinogen inhibited leukocyte adhesion. Note that this effect was more pronounced when soluble fibrinogen was added to immobilized fibrin(ogen) before cells were added; hence, it was demonstrable even when soluble fibrinogen was not present in the system (Figure 1B). Second, fibrinogen coated on plastic at high concentrations did not support cell adhesion, whereas its low-coating concentrations were highly adhesive. On the basis of the analyses of a “peak of adhesion” (Figures 4 and 5), it is reasonable to propose that fibrinogen adsorbed at low concentrations makes direct contact with the plastic surface and, therefore, binds firmly. In contrast, fibrinogen adsorbed at higher concentrations would bind not only to the plastic surface but also would form bonds with other fibrinogen molecules that have already fastened to plastic. Accordingly, cells that are capable of engaging the surface-bound fibrinogen do not detach. As the coating concentration increases, cells adhere to a “fake” adhesive layer consisting of loosely bound fibrinogen. Therefore, such cells would be less capable of consolidating their grip by spreading on the substrate when shearing force is applied and likely detach. Third, stabilization of the fibrinogen-fibrin substrate by cross-linking with FXIIIa eliminated the antiadhesive effect of fibrinogen (Figure 6). Finally, the direct interaction between soluble fibrinogen and its unfolded immobilized form and with fibrin was shown by SPR.

Figure 8 schematically depicts the mechanism that protects fibrin clots by preventing the accumulation of leukocytes on their surface. Under flow, circulating fibrinogen forms complexes with polymerized fibrin. These structures on the surface of the clot can produce an unstable, easily assembling/disassembling layer. It might break under the shear forces of the bloodstream when attached cells attempt to consolidate their grip on the surface; subsequently, cells slip off the clot. Another, although not mutually exclusive, interpretation is that this soft feeble substrate can limit the ability of cells to spread and attach firmly by failing to induce sufficient signaling. In this regard, the role of rigidity of adhesive substrates and forces generated at sites of integrin-extracellular matrix contacts in control of intracellular signaling has become an important theme in cell mechanobiology.^34,35 

The finding of the present work is consistent with previous studies by Kuijper et al.⁶  Using a parallel-plate perfusion chamber, those researchers demonstrated that soluble fibrinogen inhibited adhesion of resting neutrophils to fibrin. However, their interpretation of this phenomenon was different; that is, low-affinity binding of fibrinogen to β₂ integrins was sufficient to inhibit leukocyte adhesion to the fibrin gel. Although we cannot exclude the possibility that association of fibrinogen with resting cells might contribute to the observed effects, our neutrophil isolations were performed under sterile conditions, and no up-regulation or activation of α_Mβ₂ before and after perfusions were detected. Furthermore, neutrophils and U937 monocytoid cells in suspension do not bind soluble protein in the absence of stimulation.^1,9,36  Thus, the simplest interpretation of these findings, which is also supported by our data from static adhesion assays, is that inhibition of leukocyte adhesion under flow conditions is enacted by the mechanism that involves the formation of an antiadhesive fibrinogen layer on the fibrin surface.

Previous studies demonstrated that deposition of fibrinogen on plastic at different densities exhibits a differential availability for mAb binding.³⁷  This was proposed to arise from its modes of orientation on the surface, resulting in various conformations. Therefore, it could have been argued that immobilization of fibrinogen molecules on plastic might alter exposure of integrin binding sites and render them nonadhesive. But, as we showed previously, fibrinogen bound to plastic and fibrin in the extracellular matrix acquire the conformations that expose the cryptic γ390 to 395 sequence,¹¹  the major binding site for integrin α_Mβ₂.^25,32  The accessibility of this sequence does not depend on the coating concentration, up to 100 μg/mL.³⁷  Moreover, integrin α_Mβ₂ has broad specificity and is capable of engaging multiple binding sites in the fibrinogen molecule.^38,39  Therefore, it seems unlikely that small changes in the coating concentration could convert the molecule into a nonadhesive structure. Thus, although availability of integrin-binding sites might contribute to leukocyte adhesion, our data imply that the poor adhesive properties of fibrinogen coated at high concentrations are due to its complex formation and the inability of this loosely packed layer to ensure cell fastening and spreading.

Several plasma proteins, including fibronectin, thrombospondin, vitronectin, and plasminogen, can support integrin-mediated leukocyte adhesion,^40-42  and they are also capable of interacting with fibrin(ogen).^30,43-45  Hence, they potentially can exert the antiadhesive effect. However, soluble fibrinogen appears to be superior to other proteins in its ability to block leukocyte adhesion to thrombus-associated fibrin. The important feature that enables fibrinogen to efficiently compete with leukocytes for binding to the fibrin clot under flow is its physiologic plasma concentration (∼6-10 μM), which is much higher than that of other proteins. Nevertheless, other plasma proteins with the ability to bind fibrin(ogen) might potentially exhibit antiadhesive properties, and their contribution remains to be defined.

The entire significance of the identified mechanism remains to be determined. But it is obvious that in vivo, it could be related to the processes of vascular repair. The integrity of the vascular plug appears to be indispensable at the early stages of vessel wall damage when it is required to prevent blood loss and to provide a mechanical scaffold in reparative processes. Therefore, the mechanisms that limit premature fibrinolysis by proteases associated with infiltrating phagocytes might initially be beneficial. As wound healing proceeds, release and/or elaboration of fibrinolytic proteases from accumulating phagocytes, as well as fibrin phagocytosis, would be advantageous to thrombus resolution. Furthermore, implications of our findings may go beyond explaining how fibrinogen protects the clot from an undesirable leukocyte accumulation but may also provide insight into the mechanisms controlling thrombus growth. It appears that the principle of the “fibrinogen shield” (ie, formation of a nonadhesive protective coat consisting of fibrin(ogen) complexes) can be applied to platelets. In this regard, previous studies demonstrated that after approximately half an hour after injury, circulating platelets cease to adhere to the lumenal surface of thrombi.^46-48  Because fibrinogen and fibrin are deposited in close association with platelets, the antiadhesive layer formed on the exposed surface of thrombi may potentially limit thrombus growth under flow conditions. Moreover, reversible binding of fibrinogen to its platelet receptor GPIIbIIIa, present on the upper surface of adhering platelets, has been proposed as a mechanism responsible for the arrest of thrombus growth.⁴⁷  Whether formation of complexes between soluble fibrinogen and its platelet receptor is part of the general antiadhesive mechanism is an intriguing possibility. Finally, because plasma fibrinogen is rapidly adsorbed on implanted vascular grafts and the accumulation of leukocytes and platelets on this fibrin(ogen) substrate affects the biomaterial's performance, it may also be important to re-examine the adsorption effects of fibrinogen on blood cell recruitment in biomaterial applications.

Contribution: V.K.L. and T.P.U. designed the research; V.K.L. performed the research; T.B. contributed to the interpretation of data; V.K.L., T.B., and T.P.U. wrote the paper.

Conflict-of-interest disclosure: The authors declare no competing financial interests.

Correspondence: Tatiana P. Ugarova, Center for Metabolic Biology, Arizona State University, ISTB-1 (Mail code 4501), 530 East Orange Street, Tempe, AZ 85287; e-mail: Tatiana.Ugarova@asu.edu.

We thank Dr John Shainoff for the generous gift of the syringe pump and Judy Drazba of the Imaging Core of the Lerner Research Institute for assistance in setting up the conditions for microscopic examination of the flow tubes. We thank Drs J. Weisel, B. Coller, J. Degen, and M. Flick for insightful comments. We thank the Molecular Biotechnology Core Lab for making the Biacore 3000 available for these studies. The Biacore 3000 was purchased through a Shared Instrumentation Grant RR016789-01A1 from NIH.

This work was supported by the National Institutes of Health and the American Heart Association.