Fibrin-based biomaterials: Modulation of macroscopic properties through rational design at the molecular level

Abstract

Fibrinogen is one of the primary components of the coagulation cascade and rapidly forms an insoluble matrix following tissue injury. In addition to its important role in hemostasis, fibrin acts as a scaffold for tissue repair and provides important cues for directing cell phenotype following injury. Because of these properties and the ease of polymerization of the material, fibrin has been widely utilized as a biomaterial for over a century. Modifying the macroscopic properties of fibrin, such as elasticity and porosity, has been somewhat elusive until recently, yet with a molecular-level rational design approach can now be somewhat easily modified through alterations of molecular interactions key to the protein’s polymerization process. This review outlines the biochemistry of fibrin and discusses methods for modification of molecular interactions and their application to fibrin based biomaterials.

Introduction

Fibrin is one of the classical biomaterials and has been widely utilized for a variety of applications [1–4] since it was first purified in large quantities in the 1940s [5]. Fibrin is at the crux of the coagulation cascade and is formed through the polymerization of the soluble precursor molecule fibrinogen, a process that is initiated by the serine protease thrombin, which is activated in response to injury (Reviewed by [6]). In addition to its important role in hemostasis, following cessation of bleeding, fibrin serves as a scaffold for tissue repair following injury. Furthermore, fibrinopeptides, which are released during polymerization of fibrin, are bioactive themselves and contribute to tissue repair due to their mitogenic, chemotactic and proangiogenic activities [7–9]. Fibrin degradation products are also known activators of wound repair [10–12]. Because of these properties, fibrin serves as a seminal “matrikine”, such that the molecule forms an insoluble matrix and also provides chemokines to stimulate surrounding cells.

The fibrin network provides a physical support for neutrophil, macrophage, and fibroblast infiltration, which will ultimately lay down fibronectin, collagen and other extracellular matrix (ECM) components to rebuild the damaged tissue. However, like all ECM components, fibrin is not merely a scaffold and provides a rich wealth of signals and cues to direct cell behaviors following injury through its numerous binding sites for growth factors, integrins and additional ECM components, such as fibronectin, fibulin, thrombospondin [13], and SPARC (secreted protein, acidic and rich in cysteine) [14]. Because of its role in hemostasis and wound repair, fibrin has been used extensively in hemostatic materials, such as fibrin glues, and wound dressings [15–18]. Fibrin has also been utilized for the development of cell instructive scaffolds and is widely utilized for differentiation of stem cells [19, 20], stem cell delivery [21] and induction of angiogenesis [22, 23].

This wide range of applications of fibrin is due to its rich bioactivity as well as the ease of manipulation of material properties of resulting fibrin gels. Modification of fibrin polymerization dynamics directly affects the porosity, fiber thickness and degree of branching of the polymerized gel, which in turn affects the mechanical properties [24, 25]. Fibrin polymerization is a mostly well-understood phenomenon and many groups have exploited the molecular details of this process in order to obtain biomaterials with desired properties for specific applications. Alteration of fibrin polymerization at the molecular level has been achieved through a number of methods ranging from alterations in pH, salt and thrombin concentration [26, 27] to incorporation of PEG [28] and other polymers [29, 30] that can be simply space filling or interact directly with the fibrin molecule [31]. This review will discuss the biochemistry of fibrin structure and polymerization and highlight approaches to exploiting the molecular mechanisms of fibrin gelation to create biomaterials with a wide range of material properties and applications.

Fibrin structure and polymerization

Biochemistry

Fibrinogen, the circulating inactive precursor of fibrin monomers, is a 340 kD dimeric glycoprotein, with each dimer being comprised of three distinct chains designated as the Aα, Bβ, and γ chains (Figure 1). Thrombin selectively cleaves fibrinogen to release two sets of peptides 16 and 15 amino acids long, known as fibrinopeptides A and B respectively, from the N termini of the Aα and Bβ chains, resulting in a fibrin molecule now comprised of two α, β, and γ chains [32]. Each set of α, β and γ chains meet in the central domain of fibrinogen and are linked by disulfide bonds at their N-terminal regions. The α chains are linked by one interchain disulfide bond and the γ chains are linked by two interchain disulfide bonds [33]. The two β chains are not directly linked to one another; a disulfide bond links the β chain of one subunit to the α chain on the other subunit while the second β chain is linked to the γ chain of the opposite subunit through an additional disulfide bond. This region where the chains are linked is known as the central domain and is found within the “E” region of fibrinogen. The central domain is flanked on either side by α-helical coiled-coil regions, each stabilized by a pair of disulfide ring structures at their ends, which extend into the distal “D” regions of the molecule. There are two “D” and one “E” region per fibrinogen molecule, with these regions corresponding the proteolytic fragments obtained through complete plasmin degradation [34]. The coiled-coil regions are split approximately equally between the E and D domains. The distal portions of the D regions are comprised of the C-termini of the β and γ chains (known as the βC and γC domains, respectively) [35]. The C-termini of the α chains (known as the αC domain) account for approximately two thirds of the α chains and are comprised of a globular and a linear portion. The αC domains are located near the E region of the molecule and can interact both intramolecularly and intermolecularly [36]. These interactions are discussed subsequently in further detail.

Figure 1. Fibrinogen Structure.

Aα chains are shown in blue, Bβ chains are shown in green, and γ chains are shown in red. Interchain disulfide bridges connecting the six polypeptide chains in the central domain are shown in orange and disulfide rings stabilizing the coiled coil regions are shown in yellow.

Polymerization: knob A/knob B

Fibrin polymerization begins when thrombin cleaves fibrinopeptides A and B from the central domain of fibrinogen. This enzymatic cleavage exposes peptide sequences at the N-termini of the α and β chains, termed knobs ‘A’ and ‘B’, respectively, which are then available to interact with the complementary holes ‘a’ and ‘b’ located at the C termini of the γ and β chains, respectively (Figure 2). The holes a and b do not require enzymatic cleavage to bind to their respective peptides and are always available for interactions. Knob A is comprised of the N-terminal Gly-Pro-Arg (GPR) motif and is complementary to hole ‘a’ located in the γ chains [37]. These knob A:hole a interactions are high affinity and appear to be the primary contributor to fibrin polymerization [38]. Indeed, fibrin polymerization is inhibited in the presence of high concentrations of a synthetic knob A peptide variant of the sequence Gly-Pro-Arg-Pro (GPRP) [39]; for reference, human knob A contains the sequence Gly-Pro-Arg-Val (GPRV). The knob B motif is comprised of the N-terminal Gly-His-Arg-Pro (GHRP) motif and is complementary to hole b located in the β chains. These knob B:hole b interactions appear to be less crucial than knob A:hole a interactions in clot formation and fibrin clots can be formed in the absence of knob B cleavage [40]. Studies using snake venom enzymes which only cleave the knob A while leaving knob B intact demonstrate that clots formed in the absence of knob B interactions displayed thinner fibers than control clots [41]. There is some debate over the physiological relevance and specific functional role of knob B:hole b interactions, however these and other studies suggest that knob B:hole b interactions promote lateral aggregation and play a role in determining clot stability and susceptibility to degradation [42, 43].

Figure 2. Fibrin Polymerization.

α chains are shown in blue, β chains are shown in green, and γ chains are shown in red. αC domains are shown in gray.

The energetics of knob A:hole a and knob B:hole b interactions vary greatly. In particular, the affinity of knob B for fragment D is approximately 5 times lower than knob A, and the strength of knob B:hole b interactions, as determined through laser tweezing based experiments, is approximately 6 times lower than that of knob A:hole a interactions [44, 45]. Furthermore, the initial release of fibrinopeptide A (and thus exposure of knob A) by thrombin cleavage is significantly faster than the release of fibrinopeptide B and exposure of knob B [46], but as polymerization proceeds, the rate of knob B exposure increases, a process which is thought to be driven by conformation changes within fibrinogen. Interestingly, studies utilizing surface bound fibrinogen rather than soluble fibrinogen found that fibrinopeptide B was cleaved faster than fibrinopeptide A [47]. In addition to the classical knob A:hole a and knob B:hole b interactions, there is evidence that knob A:hole b interactions occur while knob B:hole a interactions do not [38].

Upon initial cleavage by thrombin, knob A:hole a interactions occur between two adjacent molecules to form non-covalent bonds between the E region of one molecule and the D region of another molecule, resulting in the formation of half-staggered dimer, to which additional molecules are added to give rise to double-stranded protofibrils [48]. It is currently debatable whether knob B:hole b interactions occur within or between protofibrils [49] however, recent studies support the hypothesis that B:b occur within protofibrils [50, 51]. Protofibrils extend to a length of approximately 600–800 nM and then associate laterally with other protofibrils to form fibrin fibers. The process of lateral aggregation is enhanced by interactions between αC domains of adjacent molecules. Fibers then go on to branch, leading to the formation of a space-filling gel. The process of fiber branching is poorly understood, however thrombin concentrations are known to affect the degree of branching. Low thrombin concentrations lead to less branched networks compared to clots formed in the presence of high thrombin concentrations [52, 53]. Recent in situ studies by Fogelson et al. propose that branch formation is dependent on monomer (activated fibrinogen) supply rate, therefore this effect is due to a higher rate of monomer formation in the presence of high thrombin concentrations [53]. Several different types of branch points have been observed, however the intersection of three fibers at a node is most common [54]. Two distinct molecular mechanisms have been described which give rise to two distinct types of junctions known as bilateral and trilateral junctions [6]. Bilateral junctions arise when two protofibrils laterally aggregate to form a four-stranded fibril, which then goes on to diverge again into two separate protofibrils [55]. Trilateral junctions occur when a fibrin molecule binds to only one end of a protofibril, therefore both the attached molecule and the original protofibril can continue to elongate [6].

Formation of monomer assembly through knob:hole interactions is rapid, however these noncovalent interactions alone do not contribute significant mechanical strength or resistance to fibrinolysis. Covalent crosslinks that are introduced at a slower rate, minutes to hours, through the transglutaminase Factor XIIIa, enhance the stability of fibrin clots. Recent studies by Weisel et al. utilizing FRAP techniques demonstrate that prior to FXIIIa crosslinking, these knob:hole interactions are not as stable as previously thought and large rearrangements of clot structure can occur during early clot formation [56].

Fibrin crosslinks

Crosslinking of fibrin clots by Factor XIIIa significantly enhances clot stability through increasing clot stiffness and resistance to deformation as well as through decreasing susceptibility to degradation [24, 57–59]. Factor XIII, the inactive precursor, circulates in plasma at a concentration of 14–28 μg/mL and, like fibrinogen, is activated through proteolytic cleavage by thrombin and thereafter denoted Factor XIIIa. This cleavage however, is approximately 40 times slower than cleavage of fibrinopeptide A [60], therefore release of fibrinopeptides seem to play a larger role in overall fibrin structure than Factor XIIIa crosslinks. Electron microscopy experiments demonstrate that FXIIIa crosslinking is not detectable under electron microscopy and does not appear to affect overall clot organization [52]. The zymogen form of FXIII has been shown to have a high affinity for fibrinogen (K_D ~ 10⁻⁸ M) and is found mainly in association with the molecule [61]. Fibrinopeptide release enhances FXIII activation, leading to an interesting positive feedback loop. FXIIIa crosslinks both the α and γ chains of both fibrinogen and fibrin through intermolecular ε(γ-glutamyl)lysyl bonds to form γ-γ, α-α and α-γ crosslinks (Reviewed by [57]). γ-γ crosslinks are formed almost instantly following activation of FXIII, while α-α and α-γ form much slower. FXIIIa mediated γ-γ crosslinking of fibrin occurs much faster than γ-γ crosslinking of fibrinogen molecules. The precise contribution of each type of crosslink remains a topic of investigation, but several studies suggest that α-α crosslinks contribute more to an increase in clot rigidity than γ-γ crosslinks. For example, clots formed in the presence of antibodies which inhibit α chain crosslinking but permit γ chain crosslinking were found to exhibit a twofold decrease in elastic modulus [62]. In another study, clots formed from recombinant fibrinogen lacking C-terminal α chain FXIIIa crosslinking sites were found to be less stiff and have a higher loss modulus than control clots [63]. However, studies utilizing mutated γ chain crosslinking sites, demonstrate that γ-γ crosslinks also contribute to clot stiffness and may be required for the formation of higher-order crosslinked species (γ_n, α_n, α_m-γ_n) which further contribute to overall stiffness as the clot matures [62, 64]. γ-γ dimers have been the subject of much controversy and there is much debate as to whether the crosslinks between the chains occur longitudinally or transversely (Figure 3). Reviews by Weisel [65] and Mosesson [66] summarize evidence for both arguments, however recent stretching experiments by Guthold et. al on individual fibrin fibers support longitudinal crosslinking [67]. Regardless of the precise mechanism of crosslinking, we know that FXIIIa crosslinking significantly increases the mechanical stability of fibrin clots, which has far reaching implications in biomaterial design.

Figure 3. Potential FXIII mediated γ- γ crosslinks.

Longitudinal (A). Transverse (B). Crosslinks are shown in orange. Note that crosslinks are not to scale and are drawn to illustrate domain crosslinking. α chains are shown in blue, β chains are shown in green, and γ chains are shown in red. For simplicity, αC interactions are excluded in this illustration.

αC interactions

The role of αC interactions in fibrin polymerization has been extensively studied in the last ten years. The αC regions are comprised of a globular (Aα392–610) and an extended portion (Aα221–391) known as the αC domain and connector, respectively. The complete crystal structure of the αC domain has not been deduced, presumably because of its flexibility, however use of electron microscopy and NMR have provided insight into the structure of this region [68–70]. The αC domains interact with one another to form noncovalent inter- and intramolecular bonds. Prior to thrombin cleavage, αC domains interact intramolecularly with one another along with the central domain of the E region. Thrombin cleavage of fibrinopeptide B is thought to result in a large conformational change that allows the αC domains to swing out and interact with αC domains on adjacent molecules [13, 68]. These intermolecular interactions have been proposed to play a large role in lateral aggregation. However, lateral aggregation still occurs in clots formed from monomers lacking this region, suggesting that while the αC domains greatly enhance lateral aggregation, they are not required for it to occur [71–73]. Studies have demonstrated that protofibrils formed from recombinant fibrinogen lacking αC domains have thinner fibers, greater density of fibers, increased branch points and are lysed more rapidly than control clots [63]. These results differ from studies by Collet et al. which demonstrated that clots formed from native fibrinogen comprised of numerous thin fibers lysed slower than clots comprised of loose networks of thick fibers [74], indicating that additional factors besides fiber diameter and network conformations influence rates of fibrinolysis. In particular, these results demonstrate that the αC domains play a critical role in regulating fibrinolysis. Additional studies performed with fibrinogen fragment X, a proteolytic fragment which lacks the αC domain, or with dysfibrinogenemias with alterations of the αC domain, showed similar alterations in clot morphology and degradation as well as impaired fibrin polymerization [71].

Macroscopically, fibrin gels are known to exhibit nonlinear elasticity and strain hardening [75, 76], behaviors that are governed by interactions at the molecular level. Seminal work by John Ferry in the 1970’s first described the viscoelastic properties of fibrin clots [77–81]. Subsequent analysis of fibrin films through optical and x-ray scattering analysis suggested that stress could induce conformational changes within the individual fibrin monomers comprising the clot [82, 83]. While it is accepted that the macroscopic mechanical properties of fibrin networks depend on the mechanical properties of the individual fibrin monomers, the precise regions of fibrin involved in this strain-hardening phenomenon have been the topic of much investigation in recent years, and in addition to the coiled coil regions and the γ chains of fibrinogen, recent studies, including our own, have highlighted the importance of the αC domain in contributing to the elastic extensibility of fibrin fibers [63, 84–86]. AFM experiments demonstrate that fibrin fibers formed from fibrinogen variants containing exclusively α-α crosslinks resulted in fibers that were 2.5 times stiffer and had an elastic limit, defined as the strain to which a fiber can be stretched, without any discernible permanent deformation upon the release of the stress, 1.5 times higher than uncross-linked fibers, compared to fully cross-linked fibers which were 3.75 times stiffer, had an elastic limit 1.2 times higher and were 1.2 times less extensible than uncross-linked fibers [85], indicating that the αC domain contributes to fibrin fiber stiffness and elasticity. Our recent molecular dynamic simulations also suggest that αC region is largely responsible for the elasticity and strain hardening of fibrin fibers [86].

Degradation Mechanisms

Fibrin degradation, termed fibrinolysis, is mediated by cleavage at specific sites by the serine protease plasmin. Plasmin can act on either fibrinogen or cross-linked fibrin to yield a number of fibrin degradation products (FDP); some common FDPs are shown in Figure 4. Cleavage of the αC domains from fibrinogen results in a fragment known as Fragment X (Figure 4A). Fragment X is subsequently cleaved into Fragment Y, comprised of an E and D domain, and one fragment D. Fragment Y is then further cleaved resulting in the generation of one fragment D and one fragment E molecules, therefore complete plasmin-mediated degradation of one fibrinogen molecule leads to the formation of two fragment D and one fragment E molecules. Plasmin’s action on fibrin ultimately leads to the production of the previously mentioned products, but due to the crosslinking of the network, several additional degradation products can be produced (Figure 4B), including a larger number of high molecular weight intermediates (X-oligomers) and the D-dimer, which are produced primarily through cleavage of the αC domain and lysis in the coiled-coiled region between the D and E domains [87]. The D-dimer consists of two cross-linked D domains from adjacent fibrin molecules, which are commonly found non-covalently bound to an E domain. Levels of circulating D-dimer have been used clinically to evaluate a number of thrombosis related diseases as increased levels of D-dimer are typically observed in patients with pulmonary thromboembolism, deep vein thrombosis, atherosclerosis [88–90].

Figure 4. Fibrin degradation products.

Plasmin-mediated degradation products produced from fibrinogen (A) and cross-linked fibrin (B). αC chains are shown in green, D regions are shown in pink and E regions are shown in blue.

Initiation of coagulation also initiates fibrinolysis through the activation of the plasminogen by tissue plasminogen activator (tPA), yielding plasmin. Conversion of fibrinogen to fibrin results in a conformational change that allows plasminogen and tPA to bind to the molecule through attachment to lysine residues on the C-terminus of the α chain of fibrin (reviewed in [91]). This binding is required for activation of plasminogen by tPA. Upon activation, plasmin begins to specifically cleave peptide bonds formed by the carboxyl group of lysine, thereby exposing additional lysines residues for plasminogen and tPA binding. This process results in an amplification of the rate of degradation. It should be noted that fibrinolysis is a highly orchestrated event and the rate of fibrinolysis is fine tuned by the presence of various inhibitors including aprotinin, ε-aminocaproic acid (ACA), plasminogen activator inhibitors (PAI-1 and PAI-2), α2-macroglobulin and thrombin activatable fibrinolysis inhibitor (TAIF) [92]. Furthermore, in addition to tPA, plasminogen can also be activated through urokinase (uPA), streptokinase (SK) and staphylokinase (SAK).

Fibrin structure is known to affect rates of fibrinolysis but the relationships are not straightforward. Several studies that modulated fiber thickness through variation in thrombin concentration suggest clots comprised of thick fibrin fibers are lysed more quickly than thin fibrin fibers [58, 93–96]. However, this relationship is also related to clot density and examination of fibrinolysis through confocal microscopy demonstrates that while clots comprised of a loose network of thick fibers are lysed more quickly than a dense network of thin fibers, at the individual fiber level, thin fibers are actually lysed more quickly than thick fibers [74]. This phenomenon is related to the more rapid diffusion of tPA through a loose fibrin network compared to a dense fibrin network. Further complicating the relationship between clot structure and fibrinolysis, fiber thickness also affects cellular expression of proteins involved in fibrinolysis; cells cultured on thick fibers increase expression of plasminogen (specifically the PGI₂ isoform), tPA and PAI-1 while thinner fibers result in an increased production of von Willebrand Factor (vWF) by endothelial cells [97]. Degree of clot crosslinking has been suggested to affect fibrinolysis rates, however, like the role of clot structure and fiber thickness in clot degradation, this relationship does not appear to be straightforward [64] [98]. The presence of platelets also greatly affects the rate of fibrinolysis and platelet rich clots are lysed slower than platelet poor clots. Clots formed from platelet rich plasma are extremely heterogeneous, containing dense regions of fibrin associated with aggregates of platelets. These dense regions result in heterogeneous lysis within the clot, with lysis proceeding slower around the regions of platelet aggregation [99].

Fibrinolysis is extremely important in the consideration of biomaterial design, as it is the mechanism of biodegradability of fibrin materials. Fibrinolysis is also an important process in wound healing, mediating cell infiltration and tissue regeneration, and therefore modification of this process through or within fibrin-based biomaterials is an important design criteria. The fibrin degradation products, fragments E, fragments D and D-dimer have potent pro-healing activity. For example, fragment E has been shown to induce angiogenesis in a chick chorioallantoic membrane assay of angiogenesis, stimulate proliferation [10], migration and differentiation of human microvascular endothelial cells in vitro [100] and enhance the outgrowth of smooth muscle cells from rabbit aortic explants [101]. Furthermore, fragment D and D-dimer have been shown to induce neutrophil chemotaxis [102].

Fibrin binding species

Fibrin contains numerous binding sites for cells, ECM proteins and growth factors that contribute to its role in modulating a number of complex cellular responses. The majority of cell binding directly to fibrin occurs through integrin interactions with the molecule. Fibrin contains binding sequences for integrins α_IIBβ₃, α_vβ₃, α_Mβ₂ and α_xβ₂ [103]. Platelets interact with fibrin through α_IIBβ₃ and α_vβ₃ integrins, with α_IIBβ₃ promoting adhesion and activation and α_vβ₃ promoting aggregation. Furthermore, clots formed from platelet rich plasma (PRP) undergo platelet-mediated contraction which contributes to enhanced resistance to fibrinolysis [104]. Endothelial cells and fibroblasts interact with an Arg-Gly-Asp (RGD) sequence on the C-terminal Aα chain of fibrinogen through α_vβ₃ integrins. α_Mβ₂ and α_xβ₂ facilitate binding of leukocytes [105]. Early studies suggested that fibrinogen contained a binding site for integrin α₅β₁, however it seems that these interactions are actually facilitated by binding of fibronectin to fibrinogen and subsequent binding of α₅β₁ integrins to the attached fibronectin molecule. These interactions facilitate attachment of cell types that do not express fibrin-binding elements, such as keratinocytes [106–108]. Clots formed from fibrinogen depleted of fibronectin are not able to support keratinocyte attachment, while clots formed in the presence of fibronectin facilitates keratinocyte attachment and spreading [4]. Cells can also interact with fibrinogen through cell-surface proteoglycan, e.g. syndecan, binding to heparin binding domains on the molecule. This binding is facilitated by the syndecan’s heparan sulfate glycosaminoglycan chains. In addition to fibronectin, fibrinogen is known to bind albumin, thrombospondin, SPARC, von Willebrand Factor (vWF), fibulin, fibroblast growth factor 2 (FGF-2), vascular endothelial growth factor (VEGF) and interleukin-1 (IL-1).

Introduction to fibrin based biomaterials

Because of its role in hemostasis, directing tissue regeneration following injury, fast polymerization dynamics and ease of tunability, fibrin is a highly versatile biomaterial that has been used for a range of applications. This section will provide an overview of the range of applications in which fibrin gels have been utilized.

One of the first commercial applications of fibrin gels was in the development of fibrin sealants. The use of fibrin to augment clotting has been explored since the early 20^th century. In 1909, Bergerl described the utility of fibrin powder as a hemostatic agent [109]. Some years later, in 1944, Tidrick and others reported on the use of fibrin gels as skin grafts to treat soldiers with burn injuries [110, 111]. Over a half century after these studies, the US FDA approved the first fibrin tissue sealant, Tisseel, in 1998, however use of noncommercial forms of fibrin sealant, derived from cryopreciptate and plasma, were common in the US in the 1990’s prior to this approval [112]. Furthermore, commercial fibrin sealants had been available in Europe since 1972. Since the approval of Tisseel in 1998, a number of fibrin sealants have been approved by the FDA for use as hemostats in surgery, adhesives for skin graft attachment for treatment of burn victims and as sealants in colostomy closure [16]. Currently approved fibrin sealants include those sourced from human pooled plasma (Tisseel, Artiss (Baxter) and Evicel (Johnson & Johnson)), individual units of plasma (Cryoseal (Thermogenesis)) and individual units of plasma, bovine collagen and bovine thrombin (Vitagel (Orthovita)).

A number of off label uses of fibrin sealants have also been described including use in tissue engineering and drug delivery applications as well as wide spread use as sealants and adhesives in a wide variety of surgical procedures, including neural, vascular, urological and intestinal procedures [16], [113–116]. Furthermore, formulations of fibrin sealants mixed with hydroxyaptite have been utilized in reconstructive and maxillofacial and dental surgeries to enhance bone regeneration; hydroxyapatite is incorporated into these formulations due to its osteoconductive properties [117]. Additionally, fibrin-based sealants mixed with platelets to create platelet-fibrin gels have been utilized for many clinical fields including oral and maxillofacial surgery, orthopaedic surgery, sports medicine and ophthalmology. Anecdotally, these platelet enriched constructs appear to enhance wound healing compared or fibrin only sealants, however thorough evaluations of this claim remain under investigation [118]. In more recent years, fibrin sealants have also been used to facilitate tissue engineering, biologics delivery, wound healing, tissue regeneration, stem cell delivery and angiogenesis [119–127].

Limitations and design considerations

Despite the wide spread clinical use of fibrin sealants, these sealants lack several key features to truly meet the full potential of fibrin as a hemostat, sealant and matrix for tissue regeneration. Clinically, fibrin sealants can be applied utilizing multiple techniques including application through a double barrel syringe or by spray application [128]. Aerosolized fibrin glue is widely used, particularly in plastic surgery [129], however, spraying techniques can alter the properties of the final fibrin mesh [130]. In the double barrel syringe application system, fibrin sealants are administered in a two-component system in which high concentrations of fibrinogen (80–120 mg/mL) are immediately mixed with high concentrations of thrombin (300–600 NIH U/mL) of equal volume at the delivery site. These high concentrations ensure rapid polymerization at the surgical site but are problematic for subsequent wound healing and tissue regeneration. A number of studies have demonstrated that fibrin gels formed from such high concentrations of fibrinogen and thrombin produce extremely dense networks that are inhibitory to cell infiltration and migration [131, 132]. Fibrin gels formed with low fibrinogen concentrations are more permissive to cell infiltration and facilitate better tissue repair responses. MSCs cultured in fibrin gels were found to proliferate better in gels of low fibrinogen concentration (5 mg/mL) than higher concentrations (17 and 50 mg/mL) [133]. However, gels formed from low concentrations of fibrinogen are extremely soft and do not exhibit the mechanical strength that is necessary for these materials to perform as an adhesive or tissue sealant. Furthermore, these extremely soft gels are not practical for clinical use because of their lack of robustness when manipulated.

The ideal fibrin material would be both permissive to cell infiltration and regeneration while also exhibiting mechanical properties appropriate to the specific application. In the majority of clinical applications this would require the fibrin construct to be strong but also highly porous. This could be accomplished by changing the properties of the fibrin fibers themselves by potentially making fibers stiffer, more branched or by incorporation of polymeric elements into a forming fibrin gel. The specific approach for modifying mechanical properties of fibrin materials will be determined by design goals dictated by final application but modification of fibrin polymerization and/or network properties provides a rich design space for biomaterial scientists. The molecular scale organization of fibrin fibers largely dictates the macroscopic mechanical properties and can be modified in a number of ways. The next section will discuss methods for modifying macroscopic properties of fibrin gel by modification of molecular scale properties.

Modulation of fibrin properties

The role of fibrinogen, thrombin, CaCl₂, salt concentration

One of the simplest methods for altering fibrin polymerization dynamics and resulting fibrin gel properties is to simply alter the concentration(s) of fibrinogen and thrombin and/or modify Ca⁺⁺ or salt concentrations. Modulation of this set of parameters allows one to fine tune fibrin network structure and mechanical properties. The role of fibrinogen, thrombin, CaCl₂ and salt concentration in modulating clot physical properties is often assessed qualitatively by measuring opacity. In general, increasing fibrinogen concentration leads to formation of increasingly dense and more turbid gels [134]. If fibrin concentrations remain constant and thrombin concentrations are varied, low concentrations of thrombin lead to the formation of turbid fibrin gels consisting of thick fibers, few branch points and large pores, while higher concentrations of thrombin lead to the formation of less turbid gels consisting of tightly packed thin fibers. Increasing calcium concentration has been shown to result in the formation of more turbid gels with increased fiber size but lower rigidity [52, 135–137]. Increasing salt concentration, while utilizing low calcium concentrations (2mM) results in less turbid gels [135]. Furthermore, gel stiffness correlates with fibrinogen and thrombin concentration, with fibrinogen concentration affecting stiffness to a greater degree than thrombin concentration. Incorporation of cells into fibrin gels also influences the stiffness of the gel and fibroblasts incorporated into fibrin gels resulted in a small increase in gel stiffness [138].

Optimization of fibrinogen, thrombin, calcium and salt concentration has been used to produce fibrin gels for specific applications with the concentration of these components affecting the proliferation, migration and differentiation of cells embedded on top of (2D) or within (3D) the gels. Modulation of these components has been utilized to induce differentiation of murine embryonic stem cells into neural lineage cells [139], affect chick dorsal root ganglias (DRGs) migration [140], enhance fibroblast migration and proliferation [141] and enhance human mesenchymal stem cell proliferation [142]. Many studies utilize a combinatorial approach to determine the optimal concentrations of these components for a specific application, but modulation of a single factor has been described. Variation of fibrinogen concentration alone has been shown to affect hMSC proliferation, with proliferation being optimal on gels formed with lower concentrations (5 mg/mL) of fibrinogen [133]. Studies by Stegemann et al. describe the influence of thrombin concentration on vascular smooth muscle cell morphology; they found that cells cultured on top of fibrin gels formed from higher thrombin concentrations exhibited longer cellular protrusions than cells on gels formed from low thrombin concentrations [143]. Gugerell et al. demonstrated that high thrombin concentrations induce apoptosis of keratinocytes [144]. The effect of thrombin should be carefully considered when designing cellular constructs, because in addition to its role in fibrin network structure, thrombin itself directly influences cellular responses [145] through both proteolysis and activation of cell contractility. Studies in which NaCl concentration alone was varied from 0–4.4% (w/v) demonstrated that NaCl increased compressive modulus and fiber diameter and gels formed with 2.15% NaCl resulted in optimal alkaline phosphatase expression by hMSCs [146]. These studies demonstrate that optimization of reaction conditions is a simple yet effective method for controlling fibrin architecture and modulating cellular responses.

Modification of knob:hole interactions

The primary method we have used for engineering fibrin structure is through modulation of the knob:hole interactions, the inherent mechanism driving polymerization. As mentioned previously, the GPR sequence is the minimum knob A sequence required for engagement of hole a during polymerization, however slight modifications can be made to these sequences to enhance interactions between synthetic peptides and the polymerization holes. Numerous knob A variants have been identified from different species, including GPRP and GPRV, and these variants have been found to have various affinities for fibrinogen [39, 147, 148]. Likewise knob B mimics including GHRPY and AHRP derived from bovine and chicken fibrinogen, respectively, have been described [42, 149]. Our group has recently characterized binding kinetics of a number of synthetic knob A mimics to fibrinogen fragment D using surface plasmon resonance (SPR) to identify sequences with optimal affinity for fibrin [150]. In this study, we identify a novel knob A mimic (GPRPFPAC) with a 10-fold higher association rate (12.5 × 10⁻³ M⁻¹s⁻¹) than current mimics. These mimics can then be used alone, or in conjunction with carrier materials, to modify fibrin structure. We subsequently created a variety of knob mimic constructs including knob A-protein constructs, PEGylated knobs A and B and knob A modified elastin-like peptide (ELP) micelles and characterized their effect on fibrin properties such as polymerization, degradation and mechanical properties and network structure [31, 43, 151–153]. Knob A-protein constructs designed to display the tetrapeptide sequences GPRP, GPRV, GHRP or GSPE (non-binding control) at the N-termini of a protein fragment, specifically the cell binding 9th and 10th type III repeats of fibronectin, were found to stably bind fibrinogen via specific knob:hole interactions and were retained within fibrin matrices for 24 hours, therefore showing promise as a potential method for drug release from fibrin matrices [152]. Characterization of bivalent and tetravalent GPRP-PEG conjugates demonstrated that these constructs significantly alter network structure (Figure 5) without significantly affecting clotting time or degradation [31]. As shown in Figure 5, the addition of non-functionalized PEG at a 100:1 PEG:fibrinogen molar ratio resulted in thicker fibers compared to control clots (Figure 5B). The addition of free GPRP peptide at a 10:1 peptide:fibrinogen molar ratio (Figure 5C) produced a more fine-meshed network than clots formed in the presence of the non-binding GPSP peptide at identical ratios (Figure 5D). Clots formed in the presence of an 1:1 molar ratios of non-binding GPSP-PEG constructs:fibrinogen were appeared similar in structure to control clots (Figure 5F, H, J, L, N, P, R). However, clots formed in the presence of 1:1 molar ratios of GPRP-PEG constructs:fibrinogen altered clot structure, with the PEG conjugate size affecting the final clot structure. In particular, the addition of 2 kDa GPRP2 -PEG (Figure 5E) and 2 kDa GPRP4 -PEG (Figure 5M) resulted in thicker and straighter fibers compared to control clots, while clots formed in the presence of the presence of 3.5 kDa and 5 kDa GPRP2 -PEG (Figure 5G, I) were similar to GPSP2-PEG controls of the same size but were found to result in the formation of dense nodes of fibrin throughout the matrix. The inclusion of 7.5 kDa GPRP2 -PEG, 10 kDa and 20 kDa GPRP4-PEG resulted in networks comprised of highly branched, thin fibers (Figure 5K, O, Q), which also displayed dense nodes of fibrin throughout the matrix. These perturbations were not observed in matrices formed in the presence of 7.5 kDa GPRP2 -PEG, 10 kDa and 20 kDa GPRP4 -PEG (Figure 5L, P, R). In additional studies, our group has created PEG-knob constructs with higher affinity for fibrinogen through conjugation of a linear 5 kDa PEG to the GPRPFPAC sequence; addition of this construct to fibrin matrices was found to increase network porosity compared to the peptide sequence alone [151]. Likewise, studies utilizing a PEGylated knob B mimic, AHRPYAAC, found that this construct also enhanced the porosity of the fibrin network (Figure 6) and increased susceptibility to degradation, though interestingly, these constructs resulted in an increase in the complex modulus [43]. In recent additional studies, we have characterized the binding kinetics of this PEGylated AHRPYAAC to fragment D and have tested the effect of concentrations above and below the K_d on clot polymerization dynamics, clot structure and mechanics at both the fiber and bulk scales. The addition of increasing concentrations of PEGylated AHRPYAAC increases polymerization rates and significantly alters clot structure and mechanical properties at multiple length scales. These studies highlight the variety of applications for synthetic knob-conjugates in altering fibrin network properties.

Figure 5. Confocal images of clots formed in the presence of synthetic A knob mimic GPRP conjugated to 5 kDa PEG.

Hydrogels were formed within sealed chambers on glass slides and imaged using laser scanning confocal microscopy. Images were rendered from 45×45×10 μm slices. (A) Control clot (1 mg/mL fibrinogen, 0.25 NIH U/mL thrombin, 5 Loewy units/mL factor XIIIa). (B) Representative network structure formed in the presence of 100:1 molar ratio of PEG-to-fibrinogen. (C) Network structure in the presence of 10:1 molar ratio of GPRP-to-fibrinogen. (D) Network structure in the presence of 10:1 molar ratio of GPSP-to-fibrinogen. Subsequent paired images are of hydrogels formed in the presence of GPRPn-PEG or GPSPn-PEG conjugates at a 1:1 conjugate:fibrinogen molar ratio as follows: 2 kDa GPxP2-PEG (E, F), 3.5 kDa GPxP2-PEG (G, H), 5 kDa GPxP2-PEG (I, J), 7.5 kDa GPxP2-PEG (K, L), 2 kDa GPxP4-PEG (M, N), 10 kDa GPxP4-PEG (O, P), 20 kDa GPxP4-PEG (Q, R). From Soon AS, Lee CS, Barker TH: Modulation of fibrin matrix properties via knob:hole affinity interactions using peptide-PEG conjugates, Biomaterials, Copyright © 2011 by Elsevier. Reprinted by permission of Elsevier.

Figure 6. Confocal images of clots formed in the presence of synthetic B knob mimic AHRPYAAC conjugated to 5 kDa PEG.

Confocal and scanning electron microscopy: Representative confocal micrograph projections (10 μm z-stacks; A–D) and SEM images (E–H) from no peptide control (A and E), control peptide conjugate, GPSPFPAC-PEG (B and F), knob ‘A’ conjugate, GPRPFPAC-PEG (C and G), and knob ‘B’ conjugate, AHRPYAAC-PEG (D and H). Confocal: Objective = 63×, Scale bar = 10 μm. SEM: Magnification = 50,000×, Voltage = 3.0 kV, Scale bar = 500 nm. From Stabenfeldt SE, Gourley M, Krishnan L, Hoying JB, Barker TH: Engineering fibrin polymers through engagement of alternative polymerization mechanisms, Biomaterials, Copyright © 2012 by Elsevier. Reprinted by permission of Elsevier.

Perturbation of knob:hole interactions can also be achieved through the use of various snake venom enzymes which preferentially cleave either the A (batroxobin, ancrod) or B (venzyme) fibrinopeptides from fibrinogen. These thrombin-like serine proteases have been used to elucidate the roles of specific knob:hole interactions in fibrin network formation and resulting network structure. Batroxobin-catalyzed clots have been found to be similar to thrombin-catalyzed gels, while venzyme-catalyzed clots are highly fragile [41, 154]. Stegemann’s group has recently described the use of ancrod for creation of fibrin-collagen matrices. Ancrod-polymerized gels were found to have larger fiber diameters, decreased length between fiber nodes, increased number of fiber bundles per volume and increased maximum force to failure in uniaxial tensile testing compared to thrombin-polymerized gels [155]. These studies demonstrate the utility of modifying knob:hole interaction during clot formation to promote the desired micro and macroscale properties.

Polymeric modification: PEGylation

PEG is an extremely attractive biomaterial for modification of fibrin gels because it is biologically inert but can be easily functionalized to impart biospecificity. Furthermore, PEG can be cross-linked to form hydrogels, the properties of which can be easily modified by chain length and degree of crosslinking to obtain [156, 157]. Alteration of fibrin clot properties through some mode of PEGylation has been extensively described in the literature. As described in the previous section, our group has utilized PEGylated fibrin knobs to alter clot polymerization dynamics, network structure, degradation kinetics and mechanical properties of fibrin clots. Formation of hydrogels from PEGylated self-assembling peptides based on the coiled-coil region of fibrinogen have also been described [158]. Several groups, including ours, have PEGylated fibrinogen itself to alter network properties [159]. Sugg’s group has utilized amine-reactive PEG to conjugate fibrinogen to the PEG molecule and the resulting fibrinogen-PEG molecules [21]. They report that PEGylation of fibrinogen in this manner does not affect thrombin-catalyzed polymerization as long as the PEGylation ratio is less than 10 while our earlier study reported ~6 PEGs as a critical cut-off for maintaining full clottability of fibrinogen [21, 159]. These fibrin-PEG gels have been successfully loaded with a variety of soluble factors including antimicrobials [160], platelet derived growth factor-BB (PDGF-BB) and transforming growth factor-beta 1 (TGF-β1) [161] and have been shown to support neovascularization of adipose-derived stem cells [162], [163] and increase MSC viability [21]. In a very different approach, Seliktar’s group has created PEGylated-fibrin constructs by denaturing fibrinogen, reducing disulfide bonds and finally crosslinking free sulfhydryls using PEG-diacrylate. This approach is highly versatile and has been used to create gels with a wide range of elastic moduli (0.01 to 10 kPa), degradation rates and cellular invasion kinetics through simply modulating the size of PEG and fibrin to PEG ratios [21, 164]. Furthermore, a variety of cell types including smooth muscle, cardiac, cartilage, human foreskin fibroblasts, and human embryonic stem cell cultures have been successfully cultured in these PEG-fibrin gels [164–167]. Recently, these gels have been loaded with low doses of Bone Morphogenetic Protein-2 (BMP-2) and have shown efficacy in inducing bone formation for treatment of critical size calvarial defects in mice [168]. These studies demonstrate the utility of creating PEG-fibrin systems for tissue engineering and wound healing applications.

Fibrin-hybrid constructs

Composite fibrin gels are often constructed in order to obtain desired material, drug delivery or cell instructive properties. Fibrin gels copolymerized with polyurethane [169], polyethylene oxide [170], polycaprolactone [171], heparin [172], collagen [173], keratin [174], chitosan [175], carbon nanotubes [176], hyaluronic acid [177], poly-L-lactic acid (PLLA) and poly(lactic-co-glycolic acid) (PLGA) [178] and even “fibrin in fibrin” systems in which fibrin microspheres are impregnated in a fibrin scaffold for drug delivery [179] have been described. These composite fibrin gels can be created in a variety of ways. Recently Atala’s group described the development of a technique to create tissue constructs of 1mm thickness with viability after 1 week by alternating electrospun polycaprolactone fibers with inkjet printing of chondrocytes in fibrin-collagen hydrogels [180]. Lesman et al. demonstrated that PLLA/PLGA fibrin composites supported neovascularization in a mouse full defect injury model to a greater extent than fibrin gels alone [178] (Figure 7). Fibrin gels mixed with supplementary ECM molecules are often copolymerized with the second fibrillar molecule [173, 181, 182]. The precise mechanism by which a second fibrillar molecule changes network architecture and mechanical behavior are poorly understood, but recent studies in which each component of a fibrin-collagen hybrid gel were selectively degraded suggest that extensive end-to-end crosslinking of collagen and fibrin fibers does not occur, instead it seems that each component forms its own network that interpenetrates the second [181]. Additional studies have demonstrated that collagen-fibrin composite gels have mechanical properties that are distinct from gels constructed of only a single component, even when total protein concentrations are constant [155, 183].

Figure 7. Fibrin PLLA/PLGA composite gels facilitate graft neovascularization and perfusion.

(A) Low power images of FITC-Dextran (green) and bright-field images highlighting the graft area. (B) Zoom at the graft area revealed intense FITC-Dextran+ neovessels penetration following 10 days postimplantation (left). Histological H&E examination demonstrated neovessels occupied with red blood cells penetrating into the graft area. Quantification of FITC-Dextran+ vessel area penetrating into the graft area (C) as well as calculation of the graft area size (D) revealed that Fibrin + PLLA/PLGA (with or without cells) constructs were most supportive of graft neovascularization and perfusion (n = 3–6). From Lesman A, Koffler J, Atlas R, Blinder YJ, Kam Z, Levenberg S: Engineering vessel-like networks within multicellular fibrin-based constructs, Biomaterials, Copyright © 2011 by Elsevier. Reprinted by permission of Elsevier.

The majority of these composite structures have been shown to enhance mechanical properties compared to fibrin gels alone, however many of these composites are not simply structural but impart additional biological activity to the matrix. In particular, incorporation of heparin allows for the immobilization of heparin binding growth factors, such as BMP-2, Fibroblast Growth Factor 2 and Nerve Growth Factor, and controlled release of these molecules [184–186]. Heparin incorporation into fibrin matrices has been achieved through covalent coupling to the molecule [184] as well as through the use of a bi-domain heparin-binding peptide consisting of both a transglutaminase sequence to facilitate cross-linking by Factor XIIIa and a heparin binding domain [187–190]. Furthermore, ECM components have rich amounts of bioactivity and incorporation of additional natural components, such as collagen, fibronectin and keratin, can impart cell instructive cues not found in a fibrin gel alone. For example, Clark and co-workers recently described the differential but synergistic effects of fibrin and collagen in regulating sprouting angiogenesis of human dermal microvascular endothelial cells in 3-D matrices [191]. In particular they found that fibrin is essential for promoting sprouting angiogenesis of human dermal microvascular endothelial cells in the presence of angiogenic factors, while replacement of fibrin with collagen correlated with sprout regression. They also found that integrin β₃, a component of the angiogenesis-associated integrin α_vβ₃, was highly up regulated in vascular endothelial cells in fibrin-rich environments but not collagen-rich environments [192, 193]. Because the presence of additional matrix components, collagen, fibronectin etc., can greatly affect the phenotypes of cells embed within fibrin gels, the purity of fibrinogen utilized to construct these gels should be carefully considered.

Modulation of degradation

Fibrin gels are highly susceptible to degradation, predominantly mediated through proteolytic cleavage by plasmin and matrix metalloproteinases (MMPs). Though plasmin-mediated degradation is important for remodeling of fibrin constructs and subsequent tissue regeneration, if not controlled, fibrin gels can degrade before their desired effects can come to fruition. A number of protease inhibitors have been added to fibrin gels to slow proteolysis [194–196] including the serine protease inhibitor aprotinin and various pharmacological MMP inhibitors. Recently an engineered aprotinin has been shown to enhance longevity of fibrin matrices compared to wild type aprotinin [197]. Fibrin degradation has also been controlled by the addition of the fibrinolysis inhibitor ε-aminocaproic acid [198] [199]. Studies by Tranquillo’s group demonstrated a decrease in fibrin degradation products with increasing ε-aminocaproic acid concentration [198]. Interestingly, fibrin degradation products were found to stimulate vascular smooth muscle cell proliferation as well as collagen and elastin deposition, events that were all decreased in the presence of ε-aminocaproic acid. This study highlights the importance of fine-tuning fibrin degradation to obtain an optimal cellular response.

Additional considerations

In addition to controlling fibrin properties through modulation of gelation properties, knob:hole interactions and degradation kinetics, polymeric modification and creation of fibrin:hybrid constructs, a number of additional methods have been described to modify fibrin matrices to achieve desired properties including photo-crosslinking and conferring anisotropic mechanical properties by directing fibrin alignment. Ruthenium-catalyzed photo-crosslinking of fibrinogen can be utilized to create fibrin constructs that do not require thrombin for polymerization or to enhance the mechanical properties of fibrin constructs formed through thrombin-initiated polymerization [200, 201]. Werkmeister and colleagues utilized the later approach to create fibrin sealants, which polymerized rapidly (reaching maximum bond strength within 20 seconds) and were found to produce tissue bonds 5 times stronger than those produced by commercial tissue sealants [202]. Tranquillo and coworkers utilized ruthenium-catalyzed photochemical crosslinking to increase the mechanical stiffness of tissue-engineered scaffolds with the goal of minimizing cell-mediated fibrin gel contraction [201]. Fibroblasts were entrapped and cultured within thrombin-polymerized fibrin gels for 24 hours, at which point the constructs were photo-chemically cross-linked. This crosslinking method was found to increase the stiffness of the gels, while having minimal effects on cell viability, proliferation or collagen deposition. Furthermore, cross-linked constructs were twice as long as non-cross-linked gels after four weeks in culture. This crosslinking method is therefore useful for applications requiring rapid polymerization, such as tissue sealants, or for controlling fibrin gel size during long-term culture that is typically required for tissue engineering applications.

Controlling fibrin fiber alignment is an additional design parameter than can be utilized to impart anisotropic mechanical properties to fibrin gels and has been shown to affect cellular responses. Fiber alignment can be induced through a number of methods including cell-mediated gel contraction [203], magnetic alignment of fibrin molecules [204], flow-induced alignment [205] and electrospinning [206]. Induction of cell-mediated fiber alignment can be achieved by embedding cells within fibrin gels constrained to specific geometries [207, 208]; in this system, fiber alignment results from a combination of cell contraction forces, mechanical constraints due to the mold and gel geometry [207]. Sander et al. demonstrated that differing fiber alignment patterns can be induced by embedding fibroblasts within cruciform molds of either symmetric or asymmetric geometries. Furthermore, fiber alignment was found to correlate with increased predicted stress and strain and increased collagen synthesis [203]. In addition to increasing matrix synthesis, fibrin fiber alignment can be utilized to direct cell fate; for example, fibrin fiber alignment and has been shown to direct the alignment of internal actin stress fibers [209] and guide the direction of sprouting from endothelial spheroids in fibrin gels [210].

Summary and Conclusions

Fibrin is a rich biomaterial that can be applied in a multitude of ways to construct a wide variety of materials including hemostatic agents, bioglues, tissue engineering constructs and cell and drug delivery vehicles. Though fundamental questions remain about the polymerization, cross-linking and degradation mechanisms of fibrin, much of these processes are well characterized and offer an abundant design landscape for biomaterial scientists. Modulation of these events at the molecular level governs the resulting structural properties of the fibrin construct. These structural properties in turn govern the mechanical, degradation and cell infiltration properties of the gel. The mechanical properties of fibrin biomaterials can be supplemented by the construction of hybrid materials containing polymeric and/or additional ECM components, such as collagen. Furthermore, additional bioactivity can be easily incorporated into fibrin constructs through the additional ECM components, peptides or growth factors. The fast polymerization kinetics, biocompatibility and easy manipulation of fibrin properties make this coagulation protein an ideal candidate for an almost endless array of applications. By rationally designing interactions at the molecular level we can create fibrin materials with properties ideal for the desired application.