Integrins

2.7.1 Overview

Integrins enclose a family of highly conserved, ubiquitously distributed, heterodimeric transmembrane proteins. They are restricted to metazoans and found across diverse taxa, ranging from sponges to mammals [221, 507]. Meanwhile, they have become the best understood cell adhesion receptors [221].

Integrins are integral membrane receptor proteins that consist of two non-covalently bound glycoprotein subunits, designated α and β, respectively. Presently, in mammals at least 18 α and 8 β subunits combine with each other to form approximately 24 distinct integrin receptors [216, 219] (Figure 9). They are expressed in large numbers by virtually all cell types, and many different integrins can be expressed on the same cell simultaneously [161, 176].

Each subunit is composed of (i) a large extracellular domain of > 940 (α) and > 640 (β) amino acid residues (120 to 180 and 90 to 110 kDa), (ii) a single transmembrane domain and (iii) a short C-terminal cytoplasmic domain (10 to 50 residues) [8, 221, 456]. The relatively long cytoplasmic tails of β subunits are well conserved and are assumed to have similar functional properties [161]. α and β subunits are homologous to each other and both subunits bind their specific ligand [221, 516]. The substrate specificity of integrins depends on the cell type on which they are expressed [215, 291]. Integrins are able to recognize multiple ligands. This is achieved by some integrins in binding to the RGD motif, a common binding site of many ligands [154, 220, 373]. The binding of ligands to integrins is universally dependent on bivalent cations and usually stimulated by Mg²⁺ and Mn²⁺, and inhibited by Ca²⁺ [380].

Figure 9. The integrin receptor family. The scheme depicts the known αβ heterodimeric as-sociations: 8 β subunits combine with 18 α subunits to form 24 distinct integrins. These can be subdivided into several subfamilies based on, e.g. ligand specificity (RGD specificity is shown here as an example), or, in case of β2 and β7 integrins, based on restriction in expres-sion to leucocytes. Adapted and modified from Hynes et al. [221] and Gahmberg et al.

[161].

2.7.2 Integrin function

The primary function of integrins is to mediate cellular adhesion to the extracellular matrix by binding their specific ligand. Most integrins bind more than one ligand and most ligands bind to more than one integrin [380, 527]. It is of importance in this context that integrins are involved in the bidirectional transduction of signals [144]. This allows linkage of the extracel-lular matrix with the cytoskeleton across the plasma membrane and thereby the integrins

‘integrate’ the cell into its micro-environment [215]. Furthermore, integrins participate in a variety of physiological processes: cell-to-cell contacts, signal transduction, cytoskeleton organisation, cellular trafficking (migration), cell proliferation, apoptosis, morphology, differentiation, immune response (e.g. leukocyte trafficking), haemostasis, wound healing [25, 104, 171, 221, 431, 518], and they play a key role in tumour progression [3, 457] and several human diseases, e.g. Glanzmann’s thrombasthenia [236, 358], epidermolysis bullosa [388]

and leukocyte adhesion deficiency [143].

Integrins can only bind their ligands when they exceed a certain minimal number on a given cell surface area. Upon certain stimuli they initiate remodelling of the cytoskeleton and cluster in so-called focal contacts (i.e. focal adhesion sites), specialized structures in which integrins are connected to bunched actin filaments [222, 433, 527]. Adhesion to the extracellular matrix is achieved through multiple integrin bonds that increase the binding affinity as a result of their cooperative interaction, and thereby overcome the initial weak binding affinity [104, 161, 432].

Ligand binding by integrins triggers a variety of signal transduction events (outside-in signalling) that regulate many aspects of cell processes e.g. cell proliferation, gene expression, survival, differentiation, cell motility, and polarity [61, 171, 193]. The function of integrins itself is regulated by inside-out signalling [176, 203].

2.7.3 Integrin-mediated signalling

The cytoplasmic tails of integrins are devoid of any enzymatic activity or corresponding protein-binding motifs except for the NPXY motif in β subunits [362]. Thus, integrins rely on adaptor proteins that connect the cytoplasmic tail with the cytoskeleton and mediate signalling [439]. Many proteins have been reported to bind to integrin tails, most often to those of the β subunits [270] and the interaction chain of these submembrane linker proteins is complex. By downstream activation of several enzymes such as the focal adhesion kinase (FAK) and Src

family kinases, integrins control a variety of signalling pathways. FAK in turn activates Ras and Rho family GTPases (guanosine triphosphatases) [75, 183]. The Rho family GTPases Rac1 and Cdc42 regulate cell migration by actin cytoskeleton reformation, formation of filopodia and cell polarity [384]. Other integrin-mediated pathways include the mitogen-activated protein (MAP) kinase pathway involved in cell growth, the phosphatidylinositol-3 kinase (PI3-K) pathway which induces cytoskeleton changes, and the extra-cellular-signal-regulated kinase (Erk) pathway [232].

Cytoplasmic tails of α and β subunits interact with each other to control the activation state.

The separation of the cytoplasmic tails seems to be essential to initiate integrin activation (inside-out signalling) [241, 290]. Reciprocally, binding of extracellular ligands enhances separation of the cytoplasmic tails, and thereby allows the interaction with cytoskeleton and molecules involved in signal transducing (outside-in signalling).

The function of integrins and their adhesive properties are controlled by changing the activa-tion state which goes along with a conformaactiva-tional change. Most integrins are not constitutive-ly active [176, 220]. Crystallographic anaconstitutive-lysis and electron microscopy revealed three overall conformations that correspond to their activation state (see Figure 10): (i) integrins clamped in the inactive or low affinity state predominantly adopt a ‘closed’ or bent conformation, in which they do not bind to ligands and do not transduce signals, whereas (ii, iii) integrins activated e.g. by Mg²⁺ are predominantly in an ‘open’ or extended form with (ii) intermediate affinity and closed headpiece, or (iii) an active state with high affinity and an open headpiece induced by ligand binding [8, 161, 291, 465, 514, 515]. Another general principle is that integrins frequently intercommunicate to activate or inhibit each other [221, 433].

Figure 10. Integrin activation. A schematic view of integrins adopting different confor-mations due to different activation states. Integrins in their bent form are assumed to be in inactive (left). An intermediate affinity is achieved by straightening of the legs (middle).

The activated high affinity conformation requires opening the binding site and separation of the legs (right). The change in conformation is accompanied by a separation of the cyto-plasmic tails which allows binding of cytocyto-plasmic proteins and downstream signalling. The binding site for divalent cations is indicated. Adapted from Gahmberg et al. [161].

2.7.4 Integrin αvβ3

Integrin αvβ3 is expressed at low levels on many cell types but high level expression is limited to osteoclasts, activated (i.e. angiogenic) endothelial cells, platelets, fibroblasts, mid-menstrual cycle endometrium, placenta, inflammatory sites, migrating smooth muscle cells and invasive tumors [3, 90, 128, 136, 138, 162, 286]. Because of its participation in angiogen-esis, placentation and implantation, bone remodelling, rheumatoid arthritis, pathological neovascularisation, and tumor metastasis it has been in the focus of intense research [137, 151, 470, 509].

Originally, integrin αvβ3 was known to bind vitronectin. Meanwhile, it has been found to interact with many other ligands including fibrinogen, fibronectin, laminin, von Willebrand factor, thrombospondin, and collagen [153, 211]. Integrin αvβ3 also recognizes osteopontin, bone sialoprotein, tenascin, agrin [306], plasminogen activator inhibitor-1 [459], cell adhesion molecule L1 on neurites [329, 520], and a fragment of metalloproteinase 2 [52]. In addition, integrin αvβ3 has been described to synergise with other surface molecules by physical

contact and to modulate the expression levels of several cell surface receptors [204, 448, 484].

For bidirectional signalling, it interacts with a number of important intracellular signalling proteins namely paxillin, vinculin, α-actinin, talin, tensin, FAK, caspase 8, and others [270, 372, 463].

αv integrins comprise a subset of the integrin family sharing a common αv subunit combined with one of five β subunits (β1, β3, β5, β6 or β8). Most αv integrins recognize the RGD sequence in a variety of ligands that are mentioned above [211, 308, 490]. The β3 subfamily includes αIIbβ3 and αvβ3 and is involved in a broad array of important physiological and pathological functions [220]. αIIbβ3 is expressed on platelets and megakaryocytes only, and is essential for platelet aggregation, thrombosis and haemostasis [128, 238, 438].

2.7.5 Genetically modified integrin-deficient mice

The specific non-redundant function of each integrin becomes most obvious by the pheno-types of specific knock-out mice [221]. Modification of the gene that encodes the αv integrin subunit results in the lack of all five αv integrins. Remarkably, despite of the absence of αv integrins a considerable degree of development and organogenesis is seen [13]. All embryos start to develop normally, indicating that embryonic development is independent of αv integrins until embryonic day (E) 9.5. In mid-gestation (E9.5 to E11.5), 80 % of αv-null embryos die because of placental defects that lead to pericardial oedema, and delayed growth and development of the embryos. The remaining 20 % of αv-null embryos appear normal by E10.5 and are mostly born alive. These homozygote newborns consistently exhibit cerebral vascular defects with intracerebral haemorrhages probably due to neuroepithelial defects [312], intestinal bleedings, and malformation of the secondary palate. All these αv-null liveborns die perinatally. By contrast, heterozygous mice appear normal and do not show neither anatomical nor behavioural abnormalities [13].

β3-knock out mice were generated to get a model of the human disease Glanzmann’s throm-basthenia, and to facilitate further studies of haemostasis, thrombosis and angiogenesis [207, 223, 400]. Unlike homozygous αv deficient mice, mice with a full (homozygous) knock-out of the β3 gene are viable and fertile. They display defects in platelet aggregation, clot retrac-tion, prolonged bleeding times, cutaneous and gastrointestinal bleedings, which are the cardinal features of Glanzmann’s thrombasthenia, and evident osteosclerosis increased with age [313]. Mice of this phenotype have a reduced average life expectancy. Placental defects occur, affecting placental development and maintenance, which finally leads to increased

foetal mortality of approximately 10 % of embryos. However, implantation is not affected and numbers of embryos found per litter do not differ from those of wild type mice [207].