Vitronectin—Master controller or micromanager?

Authors


Address correspondence to: David I. Leavesley, Institute of Health and Biomedical Innovation, Queensland University of Technology, 60 Musk Avenue, Kelvin Grove, Brisbane, QLD 4059, Australia. E-mail: d.leavesley@qut.edu.au

Abstract

The concept that the mammalian glycoprotein vitronectin acts as a biological ‘glue’ and key controller of mammalian tissue repair and remodelling activity is emerging from nearly 50 years of experimental in vitro and in vivo data. Unexpectedly, the vitronectin-knockout (VN-KO) mouse was found to be viable and to have largely normal phenotype. However, diligent observation revealed that the VN-KO animal exhibits delayed coagulation and poor wound healing. This is interpreted to indicate that VN occupies a role in the earliest events of thrombogenesis and tissue repair. VN is the foundation upon which the thrombus grows in an organised structure. In addition to sealing the wound, the thrombus also serves to protect the underlying tissue from oxidation, is a reservoir of mitogens and tissue repair mediators, and provides a provisional scaffold for the repairing tissue. In the absence of VN (e.g., VN-KO animal), this cascade is disrupted before it begins. A wide variety of biologically active species associate with VN. Although initial studies were focused on mitogens, other classes of bioactives (e.g., glycosaminoglycans and metalloproteinases) are now also known to specifically interact with VN. Although some interactions are transient, others are long-lived and often result in multi-protein complexes. Multi-protein complexes provide several advantages: prolonging molecular interactions, sustaining local concentrations, facilitating co-stimulation of cell surface receptors and thereby enhancing cellular/biological responses. We contend that these, or equivalent, multi-protein complexes facilitate VN polyfunctionality in vivo. It is also likely that many of the species demonstrated to associate with VN in vitro, also associate with VN in vivo in similar multi-protein complexes. Thus, the predominant biological function of VN is that of a master controller of the extracellular environment; informing, and possibly instructing cells ‘where’ to behave, ‘when’ to behave and ‘how’ to behave (i.e., appropriately for the current circumstance). © 2013 IUBMB Life, 65(10):807-818, 2013

Introduction—What is Vitronectin?

An ability to repair damaged tissue(s) is essential to sustain life in all multi-cellular organisms. The biological, physiological and molecular events that underlie tissue repair and regeneration have been a focus of human intrigue and experimentation for centuries. However, only recently we are beginning to appreciate how complex this challenge is in higher order organisms, especially those with multiple and complex tissue types. It requires a cascade of temporal and spatial molecular and cellular events, which if absent or defective results in reduced individual fitness and life expectancy. In this brief review, we introduce and discuss the emerging concept of the glycoprotein vitronectin (VN) as a biological ‘superglue’ and key controller of mammalian tissue repair and remodelling.

VN was first described as an adhesive protein that supported the attachment and spreading on glass (Latin: vitreous, ‘of glass’) of ‘unadapted’ (primary) cells [1]. It is the property of serum that enables mammalian cells to adhere to culture vessels, supporting cell survival and propagation. Therein lies the first clue to its biological function. Since this first report, VN has been independently ‘rediscovered’ by several investigators and reported variously as ‘serum-spreading factor’ [2], ‘S protein’ (of complement, equivalent to ‘Protein X’, and not to be confused with ‘Protein S’) [3], plasminogen activator inhibitor-1-binding protein [4], somatomedin B (SMB, [5] and ‘epibolin’ [6]. It shares structural similarities with the membrane-attack-complex inhibitor (of complement) [7], plasminogen activator inhibitor-1 (PAI-1)-binding protein (PAIBP) [4] and with the matrix metalloproteinase (MMP)-21 [8]. Indeed, we have used this similarity to MMP and the well-described hemopexin-like repeats with the solution for the SMB domain to create a new model of VN (Fig. 1).

Figure 1.

Structural domains of VN (predicted tertiary structure). The NMR structure of the SMB domain (PDB accession ID 2jq8) was fused to a homology model of the remainder of the protein generated by the SWISS-MODEL repository by overlaying three contiguous residues found in both structures. This model is based on the crystal structure of pro-MMP1 (PDB accession ID 1su3), with which VN shares 12% of sequence identity. After extensive energy minimisation to resolve steric clashes, the structure was relaxed in a short 1-ns simulation in explicit water using NAMD. To generate the open conformation, an interactive in vacuo simulation was carried out in which only residues 1–60 (comprising the SMB domain and the flexible peptide linking it to the polyanionic region) were free to move. The SMB domain was pulled out of place using steering forces averaged over residues 1–40, with each atom experiencing a force scaled by its atomic mass. Both structures were then equilibrated in explicit water at 300 K for 10 ns. This model supports the contention that interactions with the SMB domain act to maintain the N- and C-terminal domains in a structured state, and that its disruption or removal triggers an order-to-disorder transition exposing previously cryptic RGD peptide and HBD recognition motifs. HBD: heparin-binding domain (blue); TG: transglutaminase motif (green); SMB: somatomedin B domain (yellow); RGD: arginine–glycine–aspartic acid cell adhesion motif (sticks); G: consensus glycosylation motif (red); P/S: consensus phosphorylation/sulphation motif (cyan).

VN can be detected, in most tissues, particularly after exposure to trauma or stress [9]. [The reader is directed to the COPE (www.cope.org) and UniProt (www.uniprot.org) databases for a comprehensive description of VN expression and distribution]. VN is present at high concentrations in the peripheral vasculature (200–400 μg/mL) [10], and is synthesised predominantly in the liver. Given the abundance of VN in blood plasma [c.f., fibronectin (FN), 300–500 μg/mL] the first decades of VN research were largely focused on how VN contributed to homeostasis and innate immunity [11, 12]. However, recent evidence indicates that VN has a wider role, consistent with its matricellular activity: modulating biological functions critical for the survival of the organism. In this review, we discuss the functional dependency many endogenous regulatory species have for VN, providing evidence that VN should be viewed as a molecular organiser or micromanager.

The synthesis of VN by somatic tissues is minimal in comparison to hepatocyte production [13]. VN is synthesised as a precursor species with limited activity and is transported into the interstitial space via receptor-mediated exo- and transcytosis [9, 14]. Messenger RNA encoding VN (gene VTN, on 17q11.2 [HGNC:12724]) can be detected in most cell and tissue types, and expression is frequently up-regulated in tissues experiencing stress (e.g., proinflammatory response) [15]. When tissue is subjected to stress, immunoreactive VN accumulates in the extravascular space. It also accumulates in the interstitial space around tumour islands (carcinomas and sarcomas). Deposits of VN are notably enriched in immune complexes associated with chronic inflammatory diseases (e.g., rheumatoid arthritis) and chronic fibrotic diseases (lung, liver and kidney fibrosis) [15]. In response to acute tissue injury, VN interacts directly with the elements of the complement [16] and coagulation cascades [17], contributes to thrombus formation, stability of vessel occlusion and subsequent inflammation and immune responses. This activity is largely a consequence of circulating in complex with PAI-1 [18]. Thus, VN participates in key physiological events that take place during the re-establishment of vascular homeostasis, wound repair and tissue regeneration. Confocal microscopy has localised VN to unidentified cytoplasmic structures within necrotic cells but it is not found associated with early apoptotic cells (i.e., cells with uncompromised membrane integrity) [19]. However, recent data have identified that VN is associated with late apoptosis in nematode worms (XX). VN accumulation, in sites of injury, has been interpreted as leakage from local vessels. It is conjecture whether VN acts as an alarm signal, phagocytic signal (opsonin), immunosuppressant or has some protective/sequestration function in this situation.

VN is also found to be associated with ‘solid’ human tumours, in particular aggressive and metastatic tumours [20-22]. As evident in traumatised tissue (above), VN is frequently enriched in the pericellular stroma, the interstitial boundary evident between tumour cell islands/clusters and adjacent ‘normal’ connective tissue [23, 24]. Less surprisingly, VN is also found in the walls of tumour-associated blood vessels [25]. These circumstantial observations suggest that VN fulfils a key function, as yet uncharacterised, at structure–function boundaries. In contrast, ‘normal’ or homeostatic tissues are at best ‘weakly reactive’ for VN. In uninjured tissues, immunoreactive VN is usually limited to ducts and small blood vessels as a homogeneous, weak staining reaction (22).

Structure and Function (Monomeric/Polymeric)

VN is present in vivo as monomeric (native) and multimeric (termed ‘denatured’) forms, depending on its association with other molecular species and the activity of proteolytic enzymes, and includes conserved structural motifs that confer VN with specialised functions [26, 27]. The 44-residue N-terminal SMB domain is frequently found in vivo cleaved from the native VN molecule retaining biological activity. The cleaved SMB fragment is reported to homodimerise and bind with the urokinase receptor (uPAR) and PAI-1 [28]. VN is commonly considered to be an incidental component of the extracellular matrix (ECM). Perhaps because it is a substantial component of plasma (∼300 μg/mL) [10], it is historically viewed as a component of the haemostatic system. Thus, it is invariably among the ‘first-on-scene’ responders during tissue trauma. In contrast to ‘classical’ ECM glycoproteins such as collagen (COL), FN and laminin, which have structural functions, VN is a ‘matricellular’ protein. Matricellular proteins are ‘modular, extracellular proteins whose functions are achieved by binding to matrix proteins as well as to cell-surface receptors, or to other molecules such as cytokines and proteases that, in turn, interact with the cell surface’ [29]. Thus, although not possessing bona fide structural credentials, nonetheless VN is a critical component of the extracellular space.

The mature VN polypeptide is stabilised by multiple disulphide bonds located at the carboxyl polycationic (heparin binding) domain [HBD] and amino SMB terminal domain (Fig. 1, yellow). It is extensively decorated with: (a) saccharides at three N-glycosylation motifs (Fig. 1, red); (b) at least two sulphates on tyrosine residues [30] (Fig. 1, cyan) and (c) multiple phosphate groups on threonine residues [25] (Fig. 1, cyan). VN exhibits a high degree of conformational flexibility [26] which presumably is determined by the local aqueous environment. Small-angle X-ray scattering experiments demonstrate that in aqueous phase, ‘inactive’ VN monomers assume a ‘peanut-shaped’ globular conformation, ∼11 nm long and ∼5 nm wide [31]. In this configuration, VN is largely unreactive and has little binding affinity towards heparin, presumably because the cell recognition RGD and polycationic (heparin-binding) motifs are structurally hindered (Fig. 1). In non-aqueous environments (e.g., associated with ‘insoluble’ ECM fibrils), functional studies suggest VN ‘denatures’ and self-assembles into large multimeric complexes [26]. In this configuration, previously cryptic functional motifs become available and confer new functional attributes to the VN molecule.

The human VN monomer is synthesised as a precursor polypeptide of 478 amino acids (aas) (75 kDa) including a 19-amino acid signal peptide [12] (Fig. 1). Oligosaccharides account for ∼30% of the mature mass. Additionally, VN is a substrate for the transglutaminases including factor XIIIa [32] (Fig. 1, green), allowing its covalent incorporation into fibrin clots and/or ECM [33]. In some tissues, the 44 aa N-terminal fragment is cleaved, releasing the cysteine-rich SMB peptide [34] (Fig. 1, yellow). VN fragments of 44–50 residues in length isolated from dialysis patient blood filtrates [35] suggest that in vivo plasmin may be responsible for cleavage of the SMB domain adjacent to the RGD cell adhesion (integrin-binding) motif (residues, 20–63; Fig. 1) in vivo; however, evidence supporting this mechanism remains insubstantial. Further activity by unknown enzymes cleaves the 75-kDa mature protein at aa 379 into daughter polypeptides of 65 and 10 kDa, linked by a disulphide bridge (data not shown). With a core polypeptide comprising multiple structural functions residing in four distinct domains (Fig. 1), interaction with partner molecules, especially glycosaminoglycans (GAGs), confer mature VN with dynamic structural flexibility.

Unsurprisingly, each conformation supports distinct biological activity. The function of VN is determined by, and dependent on, interactions with other species (e.g., divalent cations, GAGs) which induce conformational change and self-assembly into heterogeneous multimolecular complexes [26]. Protease activity and/or physical interactions with ‘target’ species (usually called ligands) that cause VN to ‘denature’ coincidently induces VN to self-assemble into fibrils, forming an interconnected network (ECM) with other extracellular glycoproteins [12]. The multimeric state further reveals previously cryptic sites, dramatically enhancing VN activity. In addition to the aforementioned SMB domain, VN includes the conserved consensus cell-recognition motif Arg-Gly-Asp (RGD), a flexible connecting region, and two hemopexin-like domains of two similar halves, approximately 200 aa residues large formed from a basic repetitive unit of 35 ≥ 45 residues connected by a histidine-rich hinge region [36] (data not shown). The hemopexin domains are speculated to prevent haem-mediated oxidative stress through sequestration [37]. A strongly basic region adjacent to the C-terminus (aa, 348–379) assembles as a groove [between HBD (blue) and G (red) in Model B], supporting interactions with molecular species containing polyanionic domains (e.g., GAGs, proteoglycans and metabolites) (reviewed in ref. [12]. The structural domains present in VN are largely responsible for the polyfunctional biological activity evident in experimental analyses of this multi-talented extracellular organiser.

In its ‘denatured’ conformation, VN interacts with multiple molecular species present in the extracellular and interstitial milieu. Complement intermediates C5b-C6, C5b-C7, and complement factors C8 and C9, interact directly with the HBD (Fig. 1, blue) of VN (Table 1) [7, 80]. PAI-1 and fibrinogen (γ-chain) directly associate with unreactive (native) circulating plasma VN [17, 81]. Perforins produced by cytolytic T-cells interact with the polycationic region (Fig. 1, blue). Thus, these antimicrobial polypeptides are effectively captured and stored within the provisional ECM and coincidently inhibit lytic activity. Proteases critical to the coagulation cascade, PAI-I [82], thrombin, TAT III complex [83], fibrinogen (and fibrin), collagen I and III, urokinase and proteolytic products, resulting from their activity (e.g., angiostatin/plasminogen, anastellin) (reviewed in refs. [9, 12] have all been identified to associate with multimeric, or denatured VN. Interestingly, and of potential significance, many of these interactions are mediated via the GAG intermediates which bind to VN through the basic consensus heparin binding groove (residues, 348–379; Fig. 1, blue) in the carboxyl terminus [84].

Table 1. Molecular species reported to interact with VN
Name of interactorExperimental evidenceInteractionReference
  1. Molecular species (human only) reported to interact with VN in vivo and/or in vitro. Species demonstrated to interact directly with VN are listed, and species that interact indirectly, via complex formation, are listed and shaded.

  2. Abbreviations EGF, epidermal growth factor; bFGF, fibroblast growth factor, basic; IGFBP, IGF-binding protein-2; -3; -4; -5; IGF-I, insulin-like growth factor I; IGF-II, insulin-like growth factor II; TNF-R IIB, tumour necrosis factor receptor superfamily, 11B; HGF, hepatocyte growth factor; POMC, proopiomelanocortin (β-endorphin); TGF, transforming growth factor-beta 1; -beta 2: VEGF, vascular endothelial growth factor; Shh: Sonic Hedgehog; PAI-1, plasminogen activator inhibitor 1 (serpine 1); uPAR, plasminogen activator receptor, urokinase type (CD87); MMP-2, gelatinase A/Type IV collagenase; MMP-7, matrilysin; MMP-26, matrix metalloproteinase-26; KLK6, kallikrein-6, neurosin; AT III, anti-thrombin III (serpine C1); PVR, Poliovirus receptor (nectin-5, CD155); SPARC, secreted protein acidic and rich in cysteine, osteonectin (BM40); HSPG, heparan sulphate proteoglycan; ADMATS, A disintegrin-like and metalloproteinase with thrombospondin type motif; TNF, tumour necrosis factor.

Epidermal growth factor (EGF)In vitroDirectSchoppet et al. [38]
Fibroblast growth factor; basic (bFGF)In vitroDirectSchoppet et al. [38]
IGF-binding protein 2 (IGFBP-2)In vitroDirectKricker et al. [39]
IGF-binding protein 3 (IGFBP-3)In vitroDirectKricker et al. [39]
IGF-binding protein 4 (IGFBP-4)In vitroDirectKricker et al. [39]
IGF-binding protein 5 (IGFBP-5)In vitroDirectNam et al. [40]
Insulin-like growth factor II (IGF-II)In vivoDirectUpton et al. [41]
Tumour necrosis factor receptor superfamily, 11B (TNF-R IIB)In vivoDirectZannettino et al. [42]
Hepatocyte growth factor (HGF)In vitroDirectRahman et al. [43]
Proopiomelanocortin (POMC, β-endorphin)In vitroDirectHildebrand et al [44]
Transforming growth factor (TGF) β1 (TGF-β1)In vivo; in vitroDirectIgnotz et al. [45]
TGF β2 (TGF-β2)In vitroDirectSchoppet et al. [38]
Vascular endothelial growth factor A (VEGF A)In vitroDirectSchoppet et al. [38]
Sonic Hedgehog (Shh)In vitroDirectPons and Marti [46]
Fibrinogen, γ-chainIn vivo; in vitroDirectPodor et al. [17]
Plasminogen activator inhibitor 1 (PAI-1, serpine 1)In vivo; in vitroDirectChain et al. [47]
Plasminogen activator receptor, urokinase type (uPAR, CD87)In vivo; in vitroDirectDeng et al. [48]
Gelatinase A/Type IV collagenase (MMP-2, EC 3.4.24.24)In vitroDirectBafetti et al. [49]
Matrilysin (MMP-7, EC 3.4.24.23)In vitroDirectImai et al. [50]
Matrix metalloproteinase-26 (MMP-26, EC 3.4.24.1)In vitroDirectMarchenko et al. [51]
Neurosin (KLK6, EC 3.4.21.B10)In vivo; in vitroDirectGhosh et al. [52]
Casein kinase II (EC 2.7.11.1)In vivo; in vitroDirectNiculescu et al. [24]
Protein kinase C, α (EC 2.7.11.13)In vitroDirectGechtman and Shaltiel [53]
Protein kinase C, β 1 (EC 2.7.11.13)In vitroDirectGechtman and Shaltiel [53]
Protein kinase C, γ (EC 2.7.11.13)In vitroDirectGechtman and Shaltiel [53]
Integrin β 1 (CD29)In vitroDirectClyman et al. [54]
Integrin β 3 (CD61)In vitroDirectClyman et al. [54]
Integrin, β 6In vitroDirectWeinacker et al. [55]
Integrin β 8In vitroDirectCambier et al. [56]
Integrin α 8In vitroDirectSchnapp et al. [57]
PlasminogenIn vitroDirectKost et al. [58]
Kininogen (Fitzgerald factor)In vivo; in vitroDirectChavakis et al. [59]
Amphiphysin1In vitroDirectOtsuka et al. [60]
Angiostatin (plasminogen fragment)In vitroDirectKost et al. [58]
Angiopoietin-1In vitroDirectCarlson et al. [61]
Angiopoietin-2In vitroDirectCarlson et al. [61]
Anti-thrombin III (AT III, serpine C1)In vivo; in vitroDirectIll and Ruoslahti [62]
Protein kinase A, cAMP-dependent (EC 2.7.11.11)In vitroDirectChain et al. [63]
Nectin-3, Poliovirus receptor-related protein (CD113)In vitroDirectMueller and Wimmer [64]
Nectin-5, Poliovirus receptor (PVR) (CD155)In vitroDirectLange et al. [65]
LacritinIn vitroDirectSanghi et al. [66]
Collagens, α1A, α2AIn vitroDirectGebb et al. [67]
SPARC (osteonectin, BM40)In vivo; in vitroDirectRosenblatt et al. [68]
β-EndorphinIn vitroDirectHildebrand et al. [44]
Syndecan-1 (CD138)In vitroDirectWilkins-Port et al. [69]
Syndecan-2 (HSPG-1, fibroglycan) (CD362)In vitroDirectWilkins-Port et al. [69]
Syndecan-4 (amphiglycan)In vitroDirectWilkins-Port et al. [69]
Heparin cofactor II, ProthrombinIn vivo; in vitroComplexLiu et al. [70]
Betaglycan (TGF-β-RIII)In vitroComplexLiu et al. [70]
A disintegrin-like and metalloproteinase with thrombospondin-type motif (ADMATS2; ADAMTS3)—heparan sulphate proteoglycan (HSPG)In vitroComplexXian et al. [71]
IGF-I—IGFBP−2, 3, −4, −5; (IGF-I:IGFBP−2/−3/−4/−5)In vitroComplexKricker et al. [39]
Fibroblast growth factor 1—HSPGIn vitroComplexHammes et al. [72]
Interleukins (2, 3, 4, 5, 6, 7, 12)—HSPGIn vitroComplexGarcía de Yébenes et al. [73]
Chemokines (C-C motif)—HSPGIn vitroComplexXian et al. [71]
Granulocyte-macrophage colony-stimulating factor (GM-CSF)—HSPGIn vitroComplexModrowski et al. [74]
Interferon-γ—HSPGIn vitroComplexDefilippi et al. [75]
Tumour necrosis factor alpha (TNF-α—HSPG)In vitroComplexDefilippi et al. [75]
Integrin alpha V—Integrin, beta 3 (αvβ3)In vivo; in vitroComplexDefilippi et al. [75]
Integrin alpha V—Integrin, beta 5 (αvβ5)In vivo; in vitroComplexSmith et al. [76]
Integrin alpha V—Integrin, beta 6 (αvβ6)In vivo; in vitroComplexWeinacker et al. [55]
Integrin alpha 8—Integrin, beta 1 (α8β1)In vitroComplexSchnapp et al. [57]
Thrombin—Antithrombin III—HSPGIn vivo; in vitroComplexZheng et al. [77]
Plasminogen activator inhibitor 1—thrombomodulinIn vitroComplexDekker et al. [78]
Galectin 1—thrombospondin I—Fibronectin 1-secreted phosphoprotein 1In vitroComplexMoiseeva et al. [79]

VN participates in the earliest response to tissue trauma. As micromolar concentrations are present in plasma, VN circulates as complexes with fibrinogen and with PAI-1. These interactions stabilise fibrinogen and PAI-1, preventing unintended coagulation and fibrinolysis, respectively [85, 86]. Upon activation of the clotting and coagulation cascades (triggered by exposure to air, debris from damaged cells, or endogenous alarm signals), plasma VN undergoes rapid conformational changes, self-assembling into fibrils and exposing cell-binding and GAG-binding motifs. Multimerisation reveals previously cryptic domains/sites dramatically increasing local concentrations of available binding sites and precipitating interactions that support the assembly of the thrombus: a complex interconnected network assembled from multiple molecular and cellular elements from the coagulation pathway (fibrin, FN, von Willebrand factor), complement pathway (e.g., factor XIII), and cell debris (e.g., necrotic cell fragments). Although the clot seals the wound (protecting underlying tissue from exposure to air) and re-establishes haemostasis, it coincidently supports cell-specific survival, cell infiltration and tissue repair as a provisional ECM. Thus, the clot is both a reservoir of biologically active species and metabolites [48] and a provisional ECM. Molecular crowding ensures bioactive species remain at high local concentrations, supporting beneficial and/or deleterious physiological processes. Tissue plasminogen activator (tPA) and urokinase plasminogen activator (uPA) attenuate further activation of plasmin sustaining clot integrity. Fibrin break-down (fibrinolysis) releases fibrin degradation products some of which are known to act upon thrombin and inhibit further clot formation; others act upon blood vessel growth and permeability [87].

VN interacts directly, and indirectly, with a repertoire of extracellular species which support a diversity of biological and physiological functions (Table 1). Although many such interactions are predicted from the presence of consensus sequences within the primary sequence of VN, a growing number of interactions are complex in nature, dependent on post-translational modifications (e.g., glycosylation, conformational rearrangements) and less well-characterised interactions with intermediate species (e.g., GAGs, polysulphated proteoglycans) [84].

In general, GAGs are classically thought to function as inert ‘space-filling’ molecules. Possessing a net negative charge, GAGs attract cations (e.g., Na+ and Ca2+), cationinc metabolites and glycoproteins. The high net charge also attracts and organises the surrounding water, resulting in a semi-viscous ‘hydrogel’ state. GAGs can be viewed as a hydrostatic ‘straight-jacket’, which restricts molecular flexibility and maintains order [88, 89]. Changes in the quantity and specificity of VN glycosylation and sialylation also affect interactions between VN and other molecular species in the extracellular space; for example, collagens [90, 91]. The importance of such interactions is illustrated by the appearance of disease associated with aberrant post-translational GAG-modifications. Aberrant glycation of Vn is linked with impaired endothelial cell function associated with hyperglycaemia, vascular disease and diabetes [92] A truncated hypo-sialylated form of VN has been recently associated with cells in late-stage apoptosis [93], suggesting specific isoforms of VN may support the removal of apoptotic bodies. Interestingly, denatured glycosylated VN inhibits the apoptosis of neutrophils in inflammed tissues [94]. GAGs may also function as intermediates, linking unrelated glycoproteins to multimeric VN via the polycation groove [95]. Thus, the pericellular microenvironment, also referred to as a ‘glycocalyx’, is a heterogeneous reservoir of well-ordered proteins, saccharides, lipids and metabolites. The composition of this external ECM is interrogated by specific cell-surface receptor systems and translated via outside-in ‘feedback’ signalling pathways.

VN is recognised by the VN-binding integrins αvβ1, αvβ3, αvβ5, αvβ6, αvβ8 and αIIbβ3 [96]. Why so many ‘VN receptors’? At the very least the presence of distinct receptors might suggest specialisation and/or divergent functions. Interactions of VN with integrins stimulate subsequent phosphorylation of intracellular targets and cause integrins to become clustered in the plane of the cell membrane. Aggregation is thought to precipitate the assembly of protein rafts comprising integrins, cytoskeletal proteins and signalling molecules, visible as punctate patterns under differential interference contrast microscopy and known as focal adhesion contacts [97]. Focal adhesion contacts are interpreted to represent intimate cell linkages between elements of the ECM and transmembrane outside-in signalling cascades. Ninety percent of cellular phosphoproteins are located within such multi-protein rafts at the cell membrane interface [98]. Signalling cascades initiated via these receptors ensure that cell responses are appropriate to the environment. In the context of tissue injury, VN provides a foundation scaffold for both the provisional ECM and coincidently facilitates inflammatory and immune cell infiltration, and concurrent healing of somatic tissue. It can be interpreted that extracellular VN moderated the intensity and duration of the inflammatory response to injury.

Given that the body of experimental evidence, demonstrating the earliest responses to tissue injury is orchestrated by VN, it came as a surprise to discover mice in which VN expression had been ‘knocked out’ (KO) were found to develop normally and had normal blood chemistry [99]. In a series of careful studies, Jang et al. [100] subsequently revealed that the VN-KO animals suffer from delayed healing. Thrombogenesis is decreased and fibrinolysis is increased in VN-KO animals, thereby delaying new blood vessel growth and thus wound healing [100]. New blood vessels that do form are ‘leaky’ due to the loss of vascular endothelial (VE)-cadherin, a molecule that mediates strong cell–cell attachment and forms the ‘water-proof’ seal between adjacent endothelial cells [101]. Macrophage infiltration is also notably reduced, whereas the levels of plasma tPA and uPA remained elevated in VN-KO mice. Thus, the resolution of inflammation is also indicated to be defective. The molecular basis of this pathology remains to be elucidated.

VN in Cancer Progression

Upon interacting with cell-surface receptors, VN triggers intracellular signalling cascades, which affect critical cellular events, including cell attachment, spreading, migration and survival [92]. Hence, VN occupies a prominent role in tissue repair and homeostatic processes. Considering the wealth and diversity of published experimental data, VN might be interpreted as the master controller, the organiser who intimately informs the cell of what is taking place in its immediate microenvironment. Evidence supporting such a key role is also implicated by the contribution of VN to the development and progression of cancer. These similarities were sufficient to prompt Dvorak [102] to describe cancer as ‘wounds that do not heal’. Mammalian development, wound repair and cancer share extensive common features of phenotype, physiology and biochemical systems.

Through interactions with cancer cell-surface integrins and uPAR, VN is reported to modulate cancer metastasis, possibly functioning as a scaffold, or structural support for tumour cell migration during invasion, and/or endothelial cell migration during angiogenesis [20, 22, 103]. Evidence supporting this concept is the migration of the highly metastatic MDA-MB-231 and MDA-MB-435 breast cancer cell lines. When assayed in vitro, migration of MDA-MB-231 is inhibited when αvβ3 and αvβ5 VN-receptors (VN-binding integrins) are blocked with antagonists [104]. The αvβ3 and αvβ5 integrins are also critical to angiogenesis; required to sustain tumour growth > 1 ≤ 2 mm, and subsequent tumour cell dissemination [97, 105]. Collectively, the weight of in vivo and in vitro evidence overwhelmingly supports a critical role for VN as a foundation ECM, or master organiser, orchestrating somatic tissue repair and cancer progression through direct interactions with binding partners which include PAI-I, uPAR and integrins. Subsequently, VN supports tumour cell survival and growth indirectly by supporting angiogenesis.

VN Supports Angiogenesis

In order to survive and thrive, all eukaryotic tissues require a blood supply. This is as true for tumour cells as it is for somatic tissues. Indeed, without vascularisation tumours cannot grow beyond 1–2 mm3, due to a lack of oxygen and other essential nutrients (167, [107]). Fundamental to new blood vessel growth and survival is the endothelial basement membrane. The basement membrane provides anchorage through which cell adhesive receptors, including integrins, interact and provide the mechanical tension required for blood vessel morphogenesis [108]. Dynamic remodelling of the endothelial ECM, particularly by membrane-type matrix metalloproteinases (MT-MMPs), co-ordinates de novo formation of new vessel tubes. Sprouting endothelium aligns and assembles into new tubes with a new vascular basement membrane, a specialised ECM that provides anchorage as well as a reservoir for cytokines and growth factors (GFs) that convey specific and specialised signalling functions to the developing vessels. Interestingly the VN receptor, αvβ3, is expressed at high levels in endothelial cells during angiogenesis [108].

Although angiogenesis was not found to be impaired in VN-KO mice, persuasive evidence indicates that VN is pivotal to this process. Upon addition of VN receptor antagonists, new blood vessel growth is effectively stopped during both physiological and tumour angiogenesis [72, 110]. Conversely, VN and FN also provide essential sources of endogenous angiogenic peptides (e.g., anastellin) and partner molecules for functional multi-protein complexes (e.g., anti-thrombin III) [111].

In addition to key angiogenic functions, VN is required to facilitate tumour cell invasion. As well as providing a provisional ECM for invading tumour cells, VN localises functional MMPs to the tumour–ECM boundary [112]. VN-immobilised MMPs cleave adjacent ECM species (including VN), allowing tumour cells to penetrate constraining interstitial tissue ECM. Further, it facilitates integrin-mediated tumour cell motility [112]. Recent evidence for the integrin dependence of ‘cancer stem cells’ (CSCs) and metastatic dissemination and/or residual disease has been described earlier [113]. Notwithstanding, CSC survival and subsequent progression remains dependent on interactions between the CSCs and the local microenvironment [114].

VN Confounds Serum-Free Tissue Culture Data

It is time to re-evaluate and re-interpret our past in vitro studies in terms of our current understanding of the informative in vivo pericellular environment, especially when it comes to cellular responses to GFs. Extensive evidence indicates that VN present within serum is the primary and the most effective species that supports cell attachment and spreading on tissue culture (TC) plastic [115]. This is attributed to the fact that at serum concentrations of 2%, or greater, VN is the most prominent ECM protein to adsorb onto TC plastic, when compared to other ECM proteins such as FN and laminin [116]. Interpretation of data gained from such in vitro assays invariably fails to account for the competitive activity of albumin, IgG, alpha-1-antitrypin and other non-adhesive factors present in the serum [117, 118]. On the contrary, VN outcompetes all of these plasma proteins to adsorb readily to TC plastic [119, 120] in the order of 42–55 ng/cm2 [118].

Until recently, in vitro studies investigating the effects of GFs on cell functions were performed with (a) serum-starved cells seeded in media containing the GF(s) of interest together with at least 0.5–2% serum or (b) on pre-plated cells starved from serum for 4–24 h. In either case, VN adsorbed from serum-containing media [121] remains present for the duration of the assay. Thus, VN can localise endogenous (serum) GFs more effectively and induce synergistic and/or confounding effects on the cells. Interpreting data from such experiments is ambiguous as it is unclear if the observed response(s) is a response to the GF, or a response to residual GF:VN/GF:ECM associations. Given what we now know, care should be taken when interpreting data investigating the effects of GFs on cell behaviour, especially when cells are exposed to even the smallest presence of serum. Hence, the contribution of VN and VN-binding integrins should be acknowledged and studied alongside the classical GF:receptor-mediated effects.

With this in mind, we re-evaluated our in vitro experimental design several years ago. We assembled (pre-bound) complexes of mitogen/GFs and ECM species for our in vitro experiments [122-126]. This approach acknowledged the evidence that cells and cell-surface ‘receptors’ do not encounter extracellular species in isolation, nor in fluid phase. In vivo the pericellular environment is a hydrogel of aqueous water organised by the glycocalyx of extracellular GAGs and metabolite polyions [127]. In such a constrained environment, molecular crowding ensures frequent and prolonged interactions are statistically favoured, can be sustained and effectively ensure ‘high’ local concentrations of the interacting partners [128].

Applying this strategy revealed that VN is able to assemble multi-protein complexes with a variety of biologically active species. Biochemical characterisation studies of insulin-like growth factor (IGF)-II were found to be reproducibly ‘contaminated’ with an unidentified 70-kDa serum species; later identified as VN [41]. At the time, the reason why IGF-II was ‘contaminated’ with VN when the closely related mitogen, IGF-I was not, could not be explained. Subsequent work discovered that IGF-I can also associate with VN, however, only via intermediates, select insulin-like growth factor binding proteins (IGFBP) [39]. The most surprising aspect of these studies was the observation that in the presence of the VN ‘contamination’, cellular responses to IGF-II where greater than uncontaminated IGF-II. The contaminating VN clearly contributed functional properties that were not induced when it was absent. We now recognise that VN acts as a scaffold for IGFs, and several other mitogens, and presents bioactive species to cognate cell-surface receptors in the context of simultaneous interactions with the VN receptor, and potentially other species in vivo (e.g., GAGs). This explains the observations of molecular rafts within the cell membrane [129] that effectively concentrate receptor and signalling pathway components in close proximity, dramatically increasing effective signal transduction and cell response. Co-activation of the VN receptor and IGF receptor is necessary for the enhanced response.

VN The Master Organiser

To date, most in vitro studies examining VN have employed the ‘solution in a test-tube’ approach. This strategy, while simple and convenient, is completely different from the in situ environment in which the experiment occurs. This approach denies processes that take place at the molecular scale, that do occur in vivo. For example, the constraints imposed by even transient interactions with large ‘insoluble’ structural polymers common in the ECM (e.g., collagens), change the dynamics of transit binding events and possibly also induce conformational reorganisation. Introduced above, VN responds to interactions with itself and/or unrelated molecular species, including other ubiquitous ECM polymers (FN, collagens, fibrin, poly-sulphated proteoglycans, etc.) and charged small metabolites (cations, glycopeptides, etc.) with conformational reorganisation. Dynamic responsiveness imparts VN with novel functional properties appropriate to the local needs. This property could explain, sometimes, the contradictory and confounding experimental observations present in the VN literature.

Evidence indicating VN primarily functions as a micromanager can be found in the original description of ‘alpha-one protein’ by Holmes [1]. Holmes used soda glass beads to both isolate VN from human serum, and later glass and polystyrene treated with VN to culture ‘unadapted’ HeLa, human conjunctiva and human heart cell lines for several years [1]. Today, we have replaced glass vessels with plasma-treated polystyrene but still require serum, containing substantial concentrations of VN, to support cell attachment and proliferation. As mentioned earlier, serum from which VN has been depleted fails to support mammalian cell attachment and survival [115, 121, 130]. One explanation for this observation is that in the absence of VN, the well-described structural components of coagulation (FN, fibrinogen, collagens and serine protease zymogens) are insufficient to self-assemble an appropriately organised provisional ECM capable of engaging cell attachment systems that, in turn, trigger classic ‘outside-in’ signalling cascades [131-133]. Intriguingly, the cell recognition systems that interact with the ECM microenvironment also have the capacity to propagate reverse ‘inside-out’ signalling events [133] and cross-talk with diverse unrelated transmembrane receptor systems [134-136]. It remains to be established if VN participates directly in the molecular events that take place in the ECM, or if it is merely a passive scaffold structure that organises and micromanages the local hydrogel milieu into appropriate functional structures recognised and bound by cell-surface receptor systems.

Throughout this brief review of the ubiquitous extracellular glycoprotein VN, we present and discuss experimental evidence, suggesting that VN is more than a structural molecule; it is a fundamental organiser or micromanager of the extracellular microenvironment. Assembling a functionally integrated, informative and responsive structure is essential to mammalian cell and tissue survival. We contend that VN is a principal conductor, orchestrating many of the essential processes that contribute to cell function. We have also highlighted the importance of confounding effects of ECM proteins present in conventional serum-free TC studies, and draw attention to the importance of re-evaluating these findings in light of the evidence outlined in this brief review.

Acknowledgement

DIL, BGH and ZU have purchased shares in Tissue Therapies Ltd., an enterprise spun-out from the Queensland University of Technology, Brisbane, to commercialise some of the technology described in this manuscript. ASK, TC, MS and AS have nothing to declare.

Ancillary