Structure of fibronectin isoforms

FN is a multi-domain glycoprotein composed of an array of multiple repeated modular structures: twelve FN type I repeats (FNI), two FN type II repeats (FNII), fifteen constitutively expressed and two alternatively spliced (in this paper, referred to as EIIIA and EIIIB) FN type III (FNIII) repeats, and a non-homologous variable (V) or type III connecting segment (IIICS) region. The multimodular structure and intermodular regions allow flexibility of the FN molecule, which is involved in regulating its function [3-8]. These modules are organized into functional domains, including the N-terminal 70-kDa domain (FNI_1-9), the 120-kDa central binding domain (CBD; FNIII_1-12) and the heparin-binding domain HepII (FNIII_12-14). The specific domains of FN can interact with multiple binding partners, including other ECM components and cell-surface receptors [9]. FN is secreted as a dimer maintained by two disulfide S-S bonds at its C-terminus [7,10-12] (Figure ​(Figure11).

Open in a separate window

Figure 1

Fibronectin (FN) and FN fragments. FN is composed of a series of FNI repeats (dark-gray boxes), FNII repeats (circles), conserved FNIII repeats (light-gray boxes) and alternatively spliced FNIII repeats (EDA).

Fibronectin in early wound-healing responses

Plasma FN is a major component of the fibrin clot. Multiple mechanisms allow FN to be incorporated into the fibrin matrix. FN can interact via non-covalent interactions with fibrin via its FNI_1-5 and FNI_10-12 domains [34]. FN is also covalently crosslinked to fibrin by activation of the blood coagulation cascade involving activated Factor XIIIa (plasma transglutaminase or coagulation factor XIII) [2,14,16,35]. This crosslinks FN via glutamine residues within its N-terminus [36] to the fibrin α chains via ε-(γ-glutamyl) lysine cross links [37]. Furthermore, plasma FN can also be bound and then assembled into a high-molecular-weight multimeric matrix on the platelet surface. Platelet activation by thrombin induces increased cell-surface expression of the major platelet integrin αIIbβ3, which binds and assembles FN via a fibrin-independent mechanism [38-44]. Platelets also express a lower number of α5β1 and αvβ3 receptors on their surface, which mediate platelet adhesion [45]. Aggregated platelets can also assemble FN via a fibrin-dependent pathway [44]. The polymerization of fibrin into a three-dimensional network has been shown to be essential for FN assembly, as the γ chain of unprocessed fibrinogen signals via αIIbβ3 to inhibit this process [38].

Platelet-producing megakaryocytes endocytose and pinocytose FN from the plasma, which becomes packaged into the α granules of platelets [46]. This has been shown to involve the αIIbβ3 integrin; mice with a mutation in the gene encoding the fibrinogen γ chain (which prevents fibrinogen interactions with αIIbβ3), von Willebrand factor null mice, and fibrinogen null mice show increased FN levels in their α granules [47-49]. Upon platelet activation, FN is released from α granules in a process called degranulation. The FN released from platelets can also assemble on the surface of platelets [42,46,50]. FN incorporation into the fibrin matrix is important for various platelet functions, including adhesion, migration and aggregation (Figure ​(Figure2)2) [16,36,38,45,46,48,49,51-98]. However, in vitro fibrin assembly was shown to be unaffected by a complete absence of plasma FN [51]. Furthermore, plasma FN conditional knockout mice, or mutant mice with reduced plasma FN levels of 50% or 70-80%, were shown to have normal clotting and bleeding times. Minimal effects on wound healing in these mice were seen in vivo [17,51,52]. Cellular FN isoforms possessing the alternatively spliced EIIIA and EIIIB domains, derived from platelets, are thought to compensate for the loss of plasma FN in these conditions [51]. Although it has been shown that cellular FN isoforms are not efficiently incorporated into the fibrin clots in vitro, it is thought that they may be sufficient to allow normal wound-healing events in vivo.

Open in a separate window

Figure 2

Functions of plasma and cellular fibronectin (FN) during wound healing. The different forms of FN play distinct roles during the different stages of wound healing.

The fibrin-FN provisional matrix allows FN to adopt extended conformations within the fibrin-FN matrix, which leads to the exposure of cryptic cell binding domains to facilitate cellular processes (Figure ​(Figure2)2) [36]. For example, fibroblast activation by various growth factors such as platelet-derived growth factor requires specific sequences within the heparin domain and IIICS region [99].

Fibronectin in the late wound-healing responses

Endothelial cells and fibroblasts repopulate the wound and deposit cellular FN, an important and abundant component of granulation tissue [1,2,66,100]. FN organizes into fibrillar structures within the stroma of granulation tissue, and forms a dense network around fibroblasts, which polarize along the FN fibrils, parallel to the epidermis [66,95]. FN assembly into a three-dimensional fibrillar network on the cell surface is vital for establishing and maintaining tissue architecture and for regulating cellular processes including adhesion [57-60], spreading [101], proliferation [58,101-103], migration [99,104-106] and apoptosis [107,108] (Figure ​(Figure2).2). The three-dimensional FN structural matrix plays an important role in regulating both ECM composition [61,67] and the deposition of other ECM molecules, including collagen types I and III [61,68-74], fibrinogen [75], fibrillins 1 and 2 [76-78], fibulin [79], laminin [61,73,80] and tenascin (TN)-C [81,82]. Reticulin has also been shown to colocalize with FN fibrils within the granulation tissue [66]. The FN matrix can also sequester growth factors and associated proteins, including bone morphogenetic protein-1 [83], vascular endothelial growth factor (VEGF) [84] and latent transforming growth factor (TGF)-β binding proteins (LTBP) 1, 3 and 4 [67,85-87] to regulate cell signaling events. While the FN matrix is continuously assembled, remodeled and turned over by cells, a more mature and stable ECM network assembles on this FN-matrix scaffold [61,85,88] (Figure ​(Figure22).

FN in the wound site is also vital for regulating the neovascularisation of granulation tissue during the resolution of tissue injury. Exposure of different ECM protein conformations in the vascular basement during wounding acts as an important cue to regulate angiogenesis [55,96]. FN expression, especially of isoforms possessing the alternatively spliced EIIIA+ and EIIIB+ modules are highly upregulated around neovessels and capillary sprouts within the highly vascularized granulation tissue [95,97,98]. An FN fibrillar network has been shown to be required for proliferation and migration of human umbilical-vein endothelial cells in a three-dimensional environment [109], and functions to promote endothelial-cell survival [24,110] (Figure ​(Figure22).

Myofibroblast differentiation is dependent on the presence of both TGF-β and EIIIA+FN and on the dramatic changes in the mechanical properties of the wound microenvironment [89,111,112]. Myofibroblasts form specialized actin-associated fibronexus adhesion complexes, which function in mechano-transduction to allow transmission of intracellular actin-generated contractile forces and the sampling of extracellular tension [111-114]. These specialized adhesion complexes may also function in myofibroblast-dependent contracture of the wound, which acts to 'shorten' and remodel the collagen-rich matrix, resulting in closure of the wound and recapitulation of normal tissue architecture and function.

The data highlighted here indicate distinct and discrete roles for the two different forms of FN during tissue injury and repair. Plasma FN and cellular FN are differentially expressed both temporally and spatially during wound healing: plasma FN circulates in the blood and functions during early wound-healing responses, whereas cellular FN is expressed and assembled locally and functions during later wound-healing responses. However, despite this non-overlapping expression and localization of the FN isoforms during wound healing, exogenous plasma FN can be assembled into pre-existing or newly assembling cellular FN matrices even if the plasma FN is isolated from a different species [115-117]. Although plasma FN shows slower initial kinetics of assembly than cellular FN [12], these data imply that plasma and cellular FN could potentially perform the same functions. Supporting this hypothesis, conditional plasma FN knockout mice were found to have normal wound healing and hemostasis [51], suggesting possible compensation by cellular isoforms of FN. However, in other physiological and pathological processes, these isoforms have been shown to be distinct and unique in their functions; plasma FN was found to be essential for protecting neuronal and non-neuronal cells from apoptosis after transient focal cerebral tissue ischaemia [51] and after traumatic brain injury [118], as cellular FN is not expressed in these damaged brain tissues. Furthermore, EIIIA-FN null mice were shown to have impaired abnormal skin wound-healing responses with reduced cell compaction and edematous-like areas within the granulation tissue and delayed re-epithelialization [90]. These results would suggest that EIIIA+FN plays an important role in the resolution of late wound-healing processes. Tan et al. reported that EIIIA-FN null mice on a different background strain showed no effect on wound healing, but did show reduced atherosclerosis, suggesting that cellular isoforms of FN contribute to pathological conditions [119]. These studies suggest independent roles for the different isoforms of FN, which cannot always be compensated for in their absence.

Fibronectin-matrix assembly

There is still much that we do not know about how FN is assembled or how the rate of deposition is controlled. Understanding the mechanisms that regulate FN-matrix assembly will allow us to control this process when it becomes misregulated.

Plasma FN in solution alone will not polymerize [120] and will not form a three-dimensional matrix in the absence of cells [116]. Both plasma and cellular FN are expressed and secreted in a soluble, compact form, which is maintained by intramolecular electrostatic interactions between the FNI_1-5, FNIII_1-2, FNIII_2-3 and FNIII_12-14 domains [7,10-12] (Figure ​(Figure3A).3A). In low-salt conditions, this compact quaternary structure can be seen by electron microscopy [6,8], and Förster resonance energy transfer studies have shown that the arms of soluble FN overlap each other [121]. FN mutants lacking FNIII_12-14 or having this region replaced by the alternatively spliced FN type III domains A1-A3 from tenascin-C (TN-C), had lower sedimentation coefficient values, reflecting adoption of a more open conformation and highlighting the importance of FNIII_12-14 in maintaining these intramolecular interactions [7].

Open in a separate window

Figure 3

Stages of fibronectin (FN)-matrix assembly: initiation, unfolding and fibrillar assembly. (A) FN initiation involves interactions with cell-surface receptors: (i) FNI_1-5 within the 70-kDa domain binds to cell-surface receptors possibly including integrins, (ii) FNIII_9-10 binds to integrin α5β1, (iii) integrin activation by outside-in or inside-out signaling induces integrins to adopt a high-affinity state and allow FN binding and (iv) FNIII_12-14 binds to heparan sulfate proteoglycans (HSPGs). (B) FN unfolding: (i) FN binding to cell-surface receptors induces cyctoskeletal reorganization of the actin cytoskeleton and myosin II-dependent contractility that results in (ii) receptor clustering and translation. This causes the tethered FN molecules to become unfolded. (C) Unfolding of FN results in the exposure of FN binding sites that allow FN-FN intermolecular interactions to occur. The domains important for each step are circled and denoted.

FN-matrix assembly is a stepwise, cell-mediated process [122,123]. The process seems to be rapid, as initial FN deposits appear at the cell surface within 10 minutes of plating, and in the conditioned medium within 30 minutes [122]. FN dimerisation is required for FN assembly, as mutation of C-terminal cysteines results in the loss of fibrillogenesis [124]. This process is briefly discussed below, but is discussed in more depth elsewhere [125].

Fibronectin and aberrant wound-healing conditions

FN-matrix assembly has to be tightly regulated. Cellular FN homeostasis is maintained by continual FN assembly and loss from the pericellular matrix [61]. Persistently high levels of FN promote cell proliferation, survival signals, and strong cell-ECM adhesions. Conversely, loss of FN-matrix assembly is observed in some transformed cells [168,196], and may be important during cell-migration events such as metastasis. In addition, loss of FN deposition [61,72] would accentuate the loss in the assembly of other ECM components within the wound bed, as seen in chronic non-healing wounds, even though FN expression persists and is actually increased in the surrounding dermis [197,198].

Fibronectin in fibrotic conditions

FN plays an important role in the development of fibrotic disease [27]. Fibrosis is characterized by an excessive deposition of connective tissue that leads to the impairment of organ structure or function, and is considered a chronic inflammatory tissue-repair response, similar in structure and composition to granulation tissue [199,200]. In fibrosis, there is a key interplay between the immune response, fibroblastic-cell response and ECM, which results in the formation of fibrotic lesions.

Although collagen is the most predominant ECM component of fibrotic tissue, excessive deposition of FN also occurs, and precedes the collagen deposition (Table ​(Table1)1) [201-208]. In glomerular and interstitial fibrosis, there is markedly increased expression of total FN levels, with increased levels of EIIIA+, EIIIB+ and oncofetal (IIICS+) isoforms detected in distinct areas of the kidney and in areas of fibrosis [26,202-204]. Fibrogenesis is driven by fibroblasts and myofibroblasts, which show increased migration, proliferation, ECM synthesis and assembly within affected tissues [114,199,209]. The presence of both alternatively spliced EIIIA+FN and pro-inflammatory cytokines such as TGF-β1 have been shown to be required in the differentiation or transdifferentiation of cells to myofibroblasts [210]. For example, experiments with EIIIA null mice suggest that EIIIA+ isoform of FN induces α-smooth-muscle actin myofibroblast differentiation in the presence of TGF-β1. In the absence of EIIIA+ FN, there is continuous interstitial fibrosis after bleomycin treatment, but no switch to a chronic fibrotic response mediated by myofibroblasts. Furthermore, these experiments also showed that EIIIA+FN is required for latent TGF-β1 activation, and plays a response in fibroblast response to TGF-β1 [201,211]. TGF-β1 and the TGF-β family have been shown to activate the Smad family of transcription factors (including Smad3), which are involved in the expression of profibrotic genes [212]. Furthermore, mouse models of atherosclerosis have shown the essential role of FN in initima-media thickening in vivo [68]. Using a 49-residue sequence from the FUD domain of the F1 adhesin protein produced by Streptococcus pyogenes (pUR4), which has been shown to inhibit FN-matrix assembly by binding to the N-terminal 70-kDa domain of FN [213], this study demonstrated that inhibiting FN-matrix assembly in vivo significantly reduced intimal, medial and adventitial thickening, collagen deposition, cell proliferation, and inflammatory-cell infiltration after induction of atherosclerosis [68]. The results confirm the essential role FN plays in mediating ECM deposition and inflammatory response, which are corollaries of fibrosis. This study also showed promise for possible in vivo applications for inhibiting FN assembly.

Recently clinical studies have shown that use of a 585 nm flashlamp-pumped pulsed-dye laser resulted in the regression or arrest of keloid development by reducing the expression of TGF-β1in keloid tissues and increasing the expression of matrix metalloproteinase (MMP)-13 (also called collagenase-3) [214]. As many fibrotic conditions are largely untreatable, it is imperative that the mechanisms involved in the development and maintenance of these diseases are understood so that effective therapies can be developed.

Regulation of fibronectin-matrix assembly

FN expression and assembly is stimulated in a cell-specific manner by a multitude of molecules (Table ​(Table3).3). Stimulation of FN expression, secretion and assembly by these agents emphasizes the complexity and requirement for tight regulation of this process. As mentioned earlier, TGF-β1 and connective tissue growth factor are key cytokines involved in upregulating FN expression during fibrogenesis [212,215,216]. FN deposition also requires RhoA-mediated cell contractility [4,121,165-167] (Table ​(Table3).3). FN can also be degraded by multiple proteases including MMP-9 [217]. FN fragments and modules can also inhibit FN-matrix assembly by competing for FN-assembly sites [187], which could act as a feedback system to regulate FN levels on the cell surface. The increased levels of proteases such as neutrophil elastase in chronic wound exudate, which can act to degrade FN [218-222], could further contribute to this condition. Aged clots in chronic wounds contain highly crosslinked fibrin, which is stripped of other functional proteins by the strongly proteolytic environment [223]. These examples illustrate the importance of FN and its assembly in regulating and resolving wound-healing processes to maintain tissue architecture.

Table 3

Regulators of fibronectin (FN) mRNA expression and assembly

	References
Positive regulators of FN mRNA
TGF-β1 and the TGF-β family	[212,253-255]
Platelet-derived growth factor-BB	[256]
Insulin-like growth factor-1	[256]
Hepatocyte growth factor	[257]
Glucose	[258,259]
Glucocorticoids	[260]
Negative regulators of FN mRNA
Cell contractility inhibitor	[3,121]
RhoA inhibitors	[165-167]
Positive regulators of FN assembly
Sphingosine-1-phosphate	[261]
Estrogen	[262]
Plasminogen activator inhibitor type I	[263,264]
Urokinase plasminogen receptor	[265]
Connective tissue growth factor	[266]
Lipoprotein A	[165,167]

Open in a separate window

TGF = transforming growth factor.

FN fibrils are continuously remodeled and turned over, which is mediated via a β1-dependent, caveolin-1-dependent and low-density lipoprotein receptor-related protein (LRP)-independent endocytic mechanism [85,224]. FN becomes targeted to the lysosomes and degraded intracellularly [61,88]. As FN is assembled into high-molecular-weight multimers by an endogenous disulfide isomerase activity [188], some reverse proteolytic activity must occur to allow FN endocytosis [61,88]. Furthermore, β1 integrin clustering can induce polarized expression of membrane type 1 (MT1)-MMP to invasive structures to cause localized ECM degradation [225]. Indeed, endocytosis of fibrillar FN from pre-assembled matrices was shown to be much slower than endocytosis of soluble FN [224]. Interestingly, mature FN matrices are as highly dynamic as immature FN matrices, but ECM maturation has been reported to assemble less dynamic ECM networks over time, which loses this dynamic remodeling [85].

Other ECM components can also influence FN-matrix assembly. Low levels of vitronectin (VN) have been shown to enhance FN-matrix assembly by increasing the expression of matrix-assembly sites on the cell surface [136,226]. However, high concentrations of VN are inhibitory for FN-matrix assembly [226-228]. The HepII domain of VN has been shown to interact with αvβ3 and αvβ5 integrins, preventing actin microfilament reorganization and causing loss of FN-matrix assembly sites [228]. Loss of collagen type VI also impairs complex FN-matrix assembly; FN fibrils become oriented parallel to the long axis of the cell [229]. As discussed earlier, the individual FN domains including 70-kDa can also inhibit FN-matrix assembly by interfering with fibril formation by the full-length molecules.

Work carried out in our laboratory has also shown that smaller domains of TN-C, but not the full-length protein, can inhibit FN-matrix assembly [230,231]. As TN-C is only coexpressed with FN in areas of physiological and pathological tissue remodeling and the presence of encrypted inhibitory activity within the individual TN-C domains suggests TN-C may also play an important role in regulating FN-matrix assembly. This highlights how ECM composition and the breakdown of ECM components can also act as a further level of control, which could be exploited to control pathological wound-healing events.