Fibrinogen and Fibrin in Hemostasis and Thrombosis

Despite its recognition as a key component of blood clots, the roles of fibrinogen and fibrin (collectively fibrin[ogen]) in hemostasis and thrombosis are insufficiently understood. Consequently, fibrin(ogen) remains an active focus of investigation at all levels of the research spectrum, including fundamental basic/discovery science, epidemiology, and clinical practice and applications. This article briefly reviews basic biology and biochemistry of fibrinogen and fibrin formation, structure, and stability and highlights recent studies published in Arteriosclerosis, Thrombosis, and Vascular Biology and elsewhere. These have enhanced our understanding of fibrin(ogen) and revealed new potential applications for fibrin detection in thrombosis.

The fibrinogen molecule is a 340-kDa homodimeric glycoprotein consisting of 2Aα, 2Bβ, and 2γ polypeptide chains linked by 29 disulfide bridges. Fibrinogen synthesis occurs primarily in hepatocytes (Figure 1). Assembly of the 6 chains takes place in a stepwise manner in which single chains assemble first into Aα-γ and Bβ-γ complexes, then into Aα/Bβ/γ half-molecules, and finally into hexameric complexes (Aα/Bβ/γ)₂.¹ All 6 fibrinogen chains are assembled with their N termini located in a central E nodule and extend outward in a coiled-coil arrangement. The Bβ and γ chains terminate in globular regions known as βC and γC modules, respectively. These regions collectively comprise the so-called D nodule. The Aα chains are the longest; at the end of the coiled-coil region, each chain extends into a highly flexible series of repeats followed by a globular αC region. Using high-resolution atomic force microscopy, Protopopova et al² obtained striking images of fibrinogen that visualize each of these structural components.

In healthy individuals, fibrinogen circulates in plasma at high concentrations (2–5 mg/mL). However, fibrinogen is an acute phase protein, and during acute inflammation, plasma fibrinogen levels can exceed 7 mg/mL. The fibrinogen chains are encoded in 3 genes that are thought to have arisen through gene duplication. Mechanisms that regulate expression of the fibrinogen genes are still largely undetermined. Genome-wide association studies have identified single-nucleotide polymorphisms within the fibrinogen genes,³ as well as loci distinct from fibrinogen that implicate transcription factors (eg, hepatocyte nuclear factors 1 and 4 [TCF1 and HNF4], signal transducer and activator of transcription 3 [STAT3])^3–5 and inflammatory signaling pathways downstream of interleukin-6⁵ in fibrinogen gene expression. In addition, microRNAs (miR) in the hsa-miR-29 family and hsa-miR-409-3p downregulate fibrinogen expression in hepatoma cells in vitro,⁶ revealing mechanisms that may fine-tune fibrinogen levels in response to environmental cues.

During coagulation, fibrinogen is converted into insoluble fibrin (Figure 1). Fibrin formation involves thrombin-mediated proteolytic cleavage and removal of N-terminal fibrinopeptides from the Aα and Bβ chains. Insertion of these newly exposed α- and β-knobs into a- and b-holes in the γC and βC regions of the D nodule, respectively, on another fibrin monomer permits the half-staggered association of fibrin monomers into protofibrils. Subsequent aggregation of protofibrils into fibers yields a fibrin meshwork that is essential for blood clot stability. This process has been extensively reviewed.^7–11

Clot formation, structure, and stability are strongly influenced by the conditions present during fibrin generation. These include the concentrations of procoagulants, anticoagulants, fibrin(ogen)-binding proteins, molecules,^12–20 and metal ions,^21,22 as well as contributions of blood and vascular cells, cell-derived microvesicles,^23–30 and presence of blood flow^31,32 (Figure 2). Many of these mechanisms have been reviewed.^33,34

The contribution of thrombin concentration to fibrin formation and structure has received considerable attention. High thrombin concentrations produce dense networks of highly branched fibrin fibers, and these clots are relatively resistant to fibrinolysis. In contrast, low thrombin concentrations produce coarse networks of relatively unbranched fibrin fibers, and these clots are relatively susceptible to fibrinolysis.^12,17,35,36 Most studies have reported that compared with fibers formed from low thrombin concentrations, fibers generated by high thrombin concentrations are thinner. However, turbidimetry and microscopy studies of fully hydrated clots suggest that high thrombin concentrations decrease the average protofibril content per fiber but only slightly decrease the fiber size, leading to a generally less compact fiber.³⁷ Thus, the substantially thinner fibers observed at high thrombin concentrations in earlier studies may reflect fiber compaction or shrinkage that occurs during dehydration. Regardless, the association of fibrin clot structural parameters with clinical pathologies—dense networks of thinner/compact fibers with increased thrombotic risk^38–40 and coarse networks of thicker/less compact fibers with increased bleeding risk^36,41—suggests that fibrin structure is a critical determinant of hemostasis and thrombosis.

Given the prominent role of thrombin concentration in determining fibrin network formation and structure, it is not surprising that antithrombotic agents that reduce thrombin activity reduce fibrin deposition and consequently thrombus formation. Because factor XI (FXI[a]) augments thrombin generation, in part by synergizing tissue factor–mediated procoagulant activity,⁵⁰ FXI inhibition strategies to reduce thrombosis have received considerable attention.⁵¹ These approaches include conventional anti-FXI inhibitory antibodies, as well as technology in which antisense oligonucleotides (ASOs) result in the specific degradation of a target mRNA and corresponding reduction in target protein level. These studies reveal surprisingly specific effects of FXI inhibition in models of thrombosis and bleeding. Pharmacological FXI inhibition does not reduce local platelet adhesion in tissue factor and collagen-coated capillary tubes but reduces platelet activation and aggregation downstream of the growing thrombus.⁵² Similarly, in an arteriovenous shunt model of thrombosis in nonhuman primates, neither anti-FXI antibodies nor ASOs alter platelet deposition on a collagen-rich segment of graft, but both decrease thrombus propagation (platelet accumulation and fibrin deposition) downstream of the collagen-rich region.^53,54 A promising phase I clinical trial demonstrated success of anti-FXI ASO treatment in humans undergoing elective total knee arthroplasty; ASO-mediated reduction of plasma FXI levels decreased symptomatic or asymptomatic venous thrombosis/thromboembolism (VTE) incidence, and the higher ASO dose tested was superior to enoxaparin.⁵⁵ In addition to its role in VTE, FXI also seems to contribute to atherogenesis in mice. Mice with deficiency in apolipoprotein E (Apoe^−/−) spontaneously develop atherosclerotic lesions, but Apoe^−/− mice with genetic FXI deficiency show reduced atherosclerosis progression.⁵⁶ Moreover, anti-FXI ASO treatment reduces thrombus formation and fibrin deposition in a model of plaque rupture in Apoe^−/− mice.⁵⁷ Thus, FXI inhibition may also be effective for reducing arterial thrombosis in humans. Notably, FXI reduction has not been associated with increased bleeding in any of these studies, suggesting that FXI antagonism may be safer than current antithrombotics. However, given that bleeding occurs in a subset of patients with FXI deficiency^58,59 and the finding of altered structure and stability of plasma clots from these patients,⁶⁰ the safety of FXI inhibition should be carefully monitored in future trials. Regardless, these findings collectively support continued efforts to investigate and advance FXI inhibition strategies into the clinic.

Heparin and heparin-like compounds are used to prevent thrombosis, presumably because of their ability to reduce thrombin activity. However, heparin binds to the central E nodule of fibrin,⁶¹ and both unfractioned heparin and low molecular weight heparin can also directly alter fibrin structure in an antithrombin-independent manner.⁶² Observed changes include effects on fibrin fiber thickness, as well as porosity. These changes are not observed with the pentasaccharide, fondaparinux. Demonstration of these direct effects of unfractioned heparin and low molecular weight heparin on fibrin structure suggests tests that assess efficacy based solely on thrombin inhibition may not fully capture therapeutic effects of these drugs. Global assays that assess both thrombin generation and fibrin formation⁶³ may more closely reflect therapeutic effects of these drugs.

The common heparin reversal agent, protamine, also modulates fibrin network structure and stability. Protamine interacts directly with fibrinogen and is incorporated into clots, resulting in the production of thicker fibrin fibers in clots that are more susceptible to fibrinolysis.⁶⁴ Recently, Kalathottukaren et al⁶⁴ characterized a synthetic polycation they termed UHRA (universal heparin reversal agent) as an alternative to protamine. Universal heparin reversal agent can neutralize both heparin anticoagulant activity and polyphosphate procoagulant activity without the off-target effects on fibrin quality observed with protamine.⁶⁴ Further studies to evaluate the therapeutic potential of universal heparin reversal agent are anticipated.

Clot quality is heavily influenced by cells and cell-derived components present at the injury site.^23–30 Recent studies have revealed previously unrecognized effects of red blood cells (RBCs) and neutrophil extracellular traps (NETs) on fibrin formation, structure, and stability. These may have substantial implications for understanding coagulation disorders.

RBCs are present in hemostatic and thrombotic clots, but their ability to influence clot formation or function has been unclear. RBCs can support thrombin generation^65–67 and thereby alter procoagulant activity at the site of clot formation. The presence of RBCs during clot formation also increases fibrin network heterogeneity, but whether RBCs increase²⁸ or decrease²⁷ fibrin fiber thickness is unclear. Once in the clot, RBCs alter clot viscoelastic properties²⁸ and, by reducing plasminogen activation, increase resistance of clots to fibrinolysis.²⁷ Effects of RBCs on fibrin structure are reduced in the presence of eptifibatide, suggesting that these effects are modulated in part by an interaction between fibrin(ogen) and a cell surface integrin.²⁷ This is consistent with observations suggesting that fibrin(ogen) binds to RBCs via a β₃-like molecule on the RBC surface,⁶⁸ although neither the binding site on fibrin(ogen) nor the putative RBC receptor has been identified. Studies of contracted whole blood clots show profound compression of resident RBCs into structures termed polyhedrocytes.^25,69 Adoption of this tight packing arrangement reduces clot permeability and may explain the resistance of older, compact clots to thrombolysis.²⁵

NETs are composed of DNA, histones, and antimicrobial proteins and have surfaced as an intriguing link between inflammation and coagulation. NETs can be detected within venous thrombi,⁷⁰ and levels of cell-free DNA (CFDNA) in plasma are increased in patients with deep vein thrombosis, suggesting that NETs contribute to thrombosis pathogenesis.⁷¹ However, specific effects of NETs and NET components on coagulation are complex. Briefly, NETs can interact with cells and coagulation factors and influence their activation and activity. NET components promote thrombin generation by activating the intrinsic pathway of coagulation and by inducing platelet-dependent mechanisms in toll-like receptors-2 and -4–dependent mechanisms.^72–74 Histones also enhance activated protein C generation by thrombin/thrombomodulin in vitro and in mice.⁷⁵ By altering these procoagulant and anticoagulant pathways, NETs may alter local thrombin levels and indirectly alter fibrin formation and quality. In vitro studies suggest that CFDNA promotes formation of densely packed networks of thick fibrin fibers.⁷⁴ This observation is interesting, given that dense fibrin networks are more typically associated with decreased fiber thickness/compaction. Thus, this finding suggests that CFDNA, like its highly charged cousin polyphosphate, modulates fibrin structure at least partly through a direct interaction with fibrin(ogen). In addition, clots formed in the presence of CFDNA exhibit delayed fibrinolysis via a mechanism that involves a CFDNA-dependent reduction in plasmin fibrinolytic activity.⁷⁴ These effects may have substantial clinical implications. For example, CFDNA levels are elevated in patients with sepsis, and the effects of CFDNA on fibrin structure and fibrinolysis are also observed in plasmas from sepsis patients.^74,76 The role of NETs and potential use of DNA-dissolving treatment (eg, DNase) remains an active area of investigation.

Multiple alternatively spliced forms of fibrinogen can be detected in plasma. Of these, an alternatively spliced form of the γ chain (γ′) is the most prevalent and has received the most attention. The γ′ chain has the final 4 amino acids of the native γ chain replaced with 20 amino acids that add substantial negative charge.^77–79 Molecules containing the γ′ chain circulate as a heterodimer with the γA chain (2Aα, 2Bβ, and γA/γ′) and comprise 8% to 15% of total fibrinogen in healthy individuals.^79,80

Epidemiological studies have associated altered arterial and venous thrombosis risk with the level of circulating γA/γ′ fibrinogen. For example, elevated levels of γA/γ′ fibrinogen have been associated with increased incidence of coronary artery disease,⁸¹ myocardial infarction,⁸² and ischemic stroke,⁸³ leading to the hypothesis that γA/γ′ fibrinogen promotes arterial thrombosis. However, a recent prospective study showed that although γA/γ′ fibrinogen is associated with increased incidence of cardiovascular disease, peripheral arterial disease, and heart failure, this association is lost when the analysis is adjusted for total fibrinogen and C-reactive protein.⁸⁴ Thus, the association of elevated fibrinogen with these pathologies may be mediated, at least in part, by a coexisting inflammatory reaction. In contrast, reduced levels of γA/γ′ fibrinogen and decreased γ′/total fibrinogen ratio have been fairly consistently associated with increased risk of VTE⁸⁵ and thrombotic microangiopathy.⁸⁶ These findings suggest that γA/γ′ fibrinogen is protective against venous thrombosis and raise interesting questions about the operant mechanism.

In vitro studies investigating the relative role(s) of the fibrinogen isoforms during clot formation indicate that both γA/γA and γA/γ′ isoforms are incorporated into the fibrin network; however, the γA/γ′ isoform has unique properties that modify its role during clotting and subsequently the function of the fibrin clot. Studies have generally reported thinner fibrin fibers in clots containing the γA/γ′ isoform and associated this effect with increased resistance to fibrinolysis.^87,88 However, Domingues et al³⁷ detected reduced packing of γA/γ′ molecules in fully hydrated protofibrils, suggesting that reduced packing results in the appearance of decreased fiber diameter in dehydrated clots. Moreover, in contrast to clots formed by γA/γA fibrinogen, characteristics of fibers formed by γA/γ′ fibrinogen are relatively unaffected by the thrombin concentration.³⁷

The γA/γ′ isoform supports high-affinity binding to thrombin exosite II^89,90 that led to its recognition as antithrombin I. Thrombin binding to the γ′ chain competitively inhibits thrombin-mediated platelet activation,⁹¹ reduces thrombin-mediated fibrinopeptide B cleavage,⁹² and decreases factor VIII⁹³ and V⁹⁴ activation. In in vitro microfluidic models, γA/γ′ fibrinogen reduces clot growth primarily at venous, but not arterial, wall shear rates, suggesting that the impact of its antithrombin I activity depends in part on the location of the thrombotic event (vein or artery).⁹⁴ Studies with mice have documented antithrombotic effects of γA/γ′ in models of both venous and arterial thrombosis. Expression of the human γA/γ′ fibrinogen isoform in mice that are heterozygous for the factor V Leiden mutation reduces thrombus volume after electrolytic injury to the femoral vein.⁹⁵ In a model of FeCl₃-induced carotid artery injury, healthy mice infused with unfractionated human fibrinogen have a shortened time to vessel occlusion, and this effect is recapitulated by infusion of γA/γA, but not γA/γ′, fibrinogen. Although γA/γ′-infused mice are not protected against thrombus formation in this model, they do have lower levels of circulating plasma thrombin-antithrombin complexes compared with γA/γA-infused mice, consistent with increased thrombin-binding capacity of γA/γ′ fibrin(ogen).⁹⁶ Collectively, these studies suggest that γA/γ′ has generally antithrombotic roles during coagulation, and its expression may serve to downregulate inflammation-induced prothrombotic activity.

Covalent crosslinking of fibrin chains is a critical determinant of fibrin stability. Crosslinking is mediated predominantly by transglutaminase factor XIII (FXIII) found in plasma and platelets. Plasma FXIII is a 320-kDa heterotetrameric zymogen (FXIII-A₂B₂) composed of 2 catalytic subunits (FXIII-A₂) tightly associated (K_d≈10⁻¹⁰ mol/L)⁹⁷ with 2 noncatalytic subunits (FXIII-B₂). FXIII-A₂B₂ circulates at ≈70 nmol/L (14–28 μg/mL)⁹⁸ in complex with fibrinogen. Although early data suggested that FXIII-A₂B₂ preferentially binds the alternatively spliced fibrinogen γ′ chain, more recent studies have localized binding to γ-chain residues 390 to 396 with additional contributions from the Aα-chain.^99–101

Catalytically active FXIII (FXIIIa) introduces ε-N-(γ-glutamyl)-lysyl crosslinks between glutamine and lysine residues on fibrin γ- and α-chains, yielding γ-γ dimers and high molecular weight species (γ-multimers, α-polymers, and αγ-hybrids). FXIII can also crosslink other plasma proteins (eg, α₂-antiplasmin and fibronectin) to fibrin. Covalent crosslinking of α₂-antiplasmin to fibrin prevents expulsion of α₂-antiplasmin from the clot during clot compression or contraction¹⁰² and is essential for clot stability. For example, in a mouse model of middle cerebral artery occlusion in which plasma clots formed ex vivo are placed into α₂-antiplasmin–deficient mice, clots made from α₂-antiplasmin–sufficient plasma are more resistant to dissolution than clots made from α₂-antiplasmin–deficient plasma.¹⁰³ The importance of FXIIIa-mediated crosslinking for clot stability has been reviewed.^104–106

Recently, we discovered that FXIIIa-mediated fibrin crosslinking also promotes RBC retention in clots, exposing a newly recognized role for this activity during VTE.^99,107 Briefly, compared with FXIII-sufficient mice (F13a^+/+), FXIII-deficient (F13a^−/−) mice produce thrombi that have reduced RBC retention and consequently are smaller.⁹⁹ This effect of FXIII on RBC retention in clots is mediated specifically by fibrin α-chain crosslinking.¹⁰⁷

The timing of fibrin crosslinking also seems to be integral to RBC retention in clots. Mice that have reduced binding of FXIII to fibrinogen and delayed FXIII activation and fibrin crosslinking (Fibγ^390–396A) show significantly decreased RBC retention and thrombus size, similar to that seen in F13a^−/− mice.⁹⁹ Interestingly, the FXIII Val34Leu polymorphism that exhibits accelerated FXIII activation paradoxically conveys moderate protection against VTE by modulating clot structure in a fibrinogen concentration-dependent manner.¹⁰⁸ These interesting and apparent paradoxical findings raise important and clinically relevant questions on the role of FXIII activation kinetics and fibrin crosslinking in thrombosis. Duval et al¹⁰⁹ tested the contribution of the Val34Leu polymorphism to thrombus formation in mice. Despite observing increased FXIII activation and crosslinking in vitro and in vivo, F13a^−/− mice infused with recombinant FXIII-Leu34 showed no difference in thrombus size compared with FXIII-Val34–infused mice in the FeCl₃ model of femoral vein thrombosis.¹⁰⁹ These data suggest that the Leu34 variant does not alter thrombus size; however, because FeCl₃ injury induces rapid formation of platelet-rich thrombi, effects of the Val34Leu polymorphism on the slow process of RBC- and fibrin-rich venous thrombus formation remain unknown. Further investigations on the contribution of the FXIII-Leu34 polymorphism to fibrin formation and thrombosis are warranted.

An essential function during coagulation is the platelet-mediated consolidation of clots in a process known as clot contraction (or retraction). This process involves fibrin(ogen) binding to platelet integrin receptor α_IIbB₃ and is influenced by both platelet and fibrin(ogen) concentrations.¹¹⁰ Although recognized as a fundamental process during coagulation, clot contraction has received little attention, particularly in a clinical setting. This gap is noteworthy given findings that associate platelet aggregation and clot contraction with decreased clot permeability and increased resistance to fibrinolysis, 2 parameters thought to impact thrombosis risk.^25,111 Tutwiler et al⁶⁹ evaluated the kinetics of clot contraction in blood samples collected from patients with recent acute ischemic stroke and correlated parameters with hemostatic and hematologic laboratory characteristics. Surprisingly, compared with clots from healthy individuals, whole blood clots from patients with recent ischemic stroke exhibit reduced clot contraction.⁶⁹ However, because samples were collected after symptom onset, these changes may reflect a consequence, rather than cause, of the thrombotic event. Ischemic stroke patients had quantitative and qualitative defects in circulating platelets (decreased platelet count, shape change, and P-selectin exposure in unstimulated platelets and decreased fibrinogen-binding capacity of activated platelets),⁶⁹ suggesting that the ischemic event may consume platelets and induce a refractory phenotype in circulating platelets that are not incorporated into the thrombus. Prospective analysis of blood samples before stroke onset is necessary to determine if altered contraction promotes occlusive thrombus formation.

Production of clots with abnormal structure and stability has been demonstrated in plasma samples from patients with increased cardiovascular disease risk.^38–40 After percutaneous coronary intervention, patients who develop in-stent thrombosis demonstrate abnormal plasma clot characteristics (eg, permeability, turbidity, and lysis time) compared with patients who did not develop in-stent thrombosis.¹¹² Similarly, compared with healthy controls, plasma clots from patients with abdominal aortic aneurysm have more densely packed fibrin networks with smaller pores and were more resistant to lysis.¹¹³ Moreover, effects are aneurysm size dependent; patients with larger aneurysms have more densely packed fibers compared with patients with smaller aneurysms.¹¹³ In both in-stent thrombosis and abdominal aortic aneurysm patients, these effects on clot properties are independent of total fibrinogen levels but may be related to effects of other plasma proteins on fibrin formation.^112,113 It remains unclear whether these fibrin clot abnormalities are only a biomarker for an operant pathophysiologic mechanism, or whether abnormal fibrin clot structure is causative in the disease.

VTE diagnosis includes imaging technologies such as Doppler ultrasound or computed tomography to detect deep vein thrombosis or pulmonary embolism, respectively.^123,124 These technologies show vascular abnormalities and flow disturbances around the thrombus but do not reveal information about thrombus composition. Development of technologies that can detect thrombus composition may have clinical use. Notably, whereas early thrombi have substantial crosslinked fibrin content, this fibrin is replaced with collagen during thrombus resolution.¹²⁵ Because fibrin-rich thrombi are more susceptible to fibrinolysis than collagen-rich thrombi,¹²⁶ distinguishing early, fibrin-rich thrombi from older, collagen-rich thrombi may aid in identifying thrombi that are susceptible to fibrin-degrading thrombolytic therapy. Currently, assessment of thrombus age is highly subjective and only poorly able to identify patients who may respond to thrombolytic treatment.¹²⁶ However, 2 recent studies of thrombosis detection in rodents have advanced methods to detect intravascular thrombi and reveal information about thrombus fibrin content. Blasi et al¹²⁷ demonstrated the ability of a fibrin-binding probe,⁶⁴ Cu-FBP8, to detect both venous and arterial thrombi in a single whole-body positron emission tomographic scan. Probe uptake correlated positively with fibrin content in both arterial and venous clots, distinguishing young (high probe uptake) from old (low probe uptake) thrombi.¹²⁷ This positron emission tomography–based imaging method enables imaging of multiple thrombi in one examination and may be a noninvasive and sensitive approach to assess changes in thrombus composition over time. Similarly, the spatial and temporal uptake of a gadolinium-based fibrin-specific magnetic resonance imaging contrast agent, EP-2104R, also correlates positively with time-dependent changes in thrombus fibrin content.¹²⁸ Furthermore, thrombi that exhibit high EP-2104R uptake are more susceptible to tissue-type plasminogen activator–mediated dissolution, suggesting that EP-2104R can be used to identify thrombi that are susceptible to thrombolytic therapy.¹²⁸ Additional studies are warranted to determine whether these methods can be used to identify human patients with greatest potential benefit of thrombolytic therapy.

This review has highlighted both established and newer findings on fibrin(ogen) expression and function that demonstrate its central role in clot formation during hemostasis and thrombosis. Additional studies beyond the scope of this review have exposed intriguing roles for fibrin(ogen) in inflammation, infection, neurological disease, cancer, and other pathologies. Collectively, these discoveries have uncovered critical links between disease pathways and rationalize the significant association of many diseases with increased bleeding and thrombosis risk. Identification of these pathways may yield new therapeutic targets with enhanced specificity and safety. Consequently, efforts to advance both basic research in fibrin(ogen) genetics, biology, biochemistry, and biophysics and translational applications for fibrin(ogen) detection and altering fibrin(ogen) function are likely to have broad impact on health and disease.

We thank Joan D. Beckman and Chase B. Brandner for reading the article.