Erratum: Integrins in regulation of tissue development and function. J Pathol; 200: 471–480

Authors

  • Erik HJ Danen,

    Corresponding author
    1. Division of Cell Biology, The Netherlands Cancer Institute, Amsterdam, The Netherlands
    • Division of Cell Biology, The Netherlands Cancer Institute, Plesmanlaan 121, 1066 CX Amsterdam, The Netherlands.
    Search for more papers by this author
  • Arnoud Sonnenberg

    Corresponding author
    1. Division of Cell Biology, The Netherlands Cancer Institute, Amsterdam, The Netherlands
    • Division of Cell Biology, The Netherlands Cancer Institute, Plesmanlaan 121, 1066 CX Amsterdam, The Netherlands.
    Search for more papers by this author

Abstract

The original article to which this Erratum refers was published in Journal of Pathology; 200(4): 471–480.

It has been brought to the attention of the publishers that there were errors on pages 476, 477 and 480 of the orginally published manuscript. These errors have now been rectified, and to facilitate greater legibility for our readers, the corrected article has been reproduced in its entirety.

Cell adhesion is indispensable for embryonic development and for proper tissue function. In metazoans, integrins are the major adhesion receptors that connect cells to components of the extracellular matrix. Integrins are implicated in assembly of extracellular matrices, cell adhesion and migration on extracellular matrices, and in vertebrates (in which the integrin family has expanded) they can also mediate cell–cell adhesion. Furthermore, integrin-mediated adhesion can modulate many different signal transduction cascades and support cell survival, proliferation, and influence the expression of differentiation-related genes. In this review we briefly explain how integrins can affect so many different aspects of cell behavior and discuss evidence for roles of integrins in tissue development, function, and disease. Copyright © 2003 John Wiley & Sons, Ltd.

Introduction

The integrin family of adhesion receptors is found throughout metazoan evolution, and several of its members are indispensable for the development of flies, worms, and vertebrates 1. Integrins are heterodimeric transmembrane receptors that bind with their globular head domain to components of the extracellular matrix (ECM). Some integrins can also bind counter-receptors present on other cells, bacterial polysaccharides, or viral coat proteins. Intracellularly, integrins are connected via associated proteins to the actin cytoskeleton (Figure 1A). In the nematode Caenorhabditis elegans, one β subunit (termed βpat3) and two α subunits (termed αina1 and αpat2) form two integrins. In the fruit fly Drosophila melanogaster, five integrins are formed through combination of one β subunit (termed βPS) with five α subunits (termed αPS1 through 5). The integrin family has expanded in vertebrates and sequencing of the human genome has identified as many as 18 α and eight β subunits, from which 24 different functional integrins are currently known to be generated in humans 2, 3 (Figure 1B).

Figure 1.

(A) Schematic representation of a cell–matrix adhesion in which integrins connect the ECM with the actin cytoskeleton. (B) The integrin family. In blue, RGD-binding integrins; in red, laminin-binding integrins

Integrin clustering is one way to increase ligand binding but the activity of individual integrins can also be modulated through an as yet incompletely understood mechanism termed ‘inside-out signalling’ which involves the propagation of conformational changes from the cytoplasmic tails across the membrane towards the ligand-binding region 4. Integrins can also activate intracellular signal transduction cascades, a process referred to as ‘outside-in signalling’. Integrin-mediated cell adhesion can trigger calcium fluxes, activate tyrosine and serine/threonine protein kinases and inositol lipid metabolism, and regulate the activity of the Rho family of small GTPases 5–10. Rho GTPases (Rho, Rac, and Cdc42), in their active, GTP-bound form, bind and activate a variety of effector proteins, which modulate the organization of the actin cytoskeleton 11.

Over the past decade, fly, worm, and mouse genetics have provided insight into the roles of many ECM components, virtually every integrin, and several integrin-associated proteins in the development and maintenance of tissues and in the progression of diseases 12–14. There is evidence for a role of integrins in facilitating the stable anchorage of cells to the ECM but they are also implicated in cell migration and they act in concert with receptors for soluble factors to regulate cell fate decisions, including survival, proliferation, and differentiation. In this review, we briefly explain how integrins can affect such a multitude of signal transduction cascades and we discuss examples showing a role of integrins in tissue development, function, and disease.

Mechanisms of integrin signalling

The importance of integrins probably lies in their adhesive function as well as in their capacity to modulate signal transduction pathways downstream of other receptors. As integrins have very short cytoplasmic tails without enzymatic activity, they connect to these pathways by affecting other signalling and adaptor proteins. The connection of integrins to the actin cytoskeleton is probably important for their signalling functions but some conceptually different views on integrin signalling exist, placing the point of convergence between signals generated by integrins and by other receptors at different levels.

Firstly, integrins and growth factor receptors may activate parallel pathways that synergize at the level of phosphorylation of downstream signalling proteins (Figure 2A). For instance, growth factors and integrin-mediated cell adhesion can each independently trigger a weak transient activation of the Erk-type mitogen-activated protein kinase but the combined effect of cell adhesion and growth factors is a strong and sustained Erk activity. Signalling by integrins and growth factor receptors in this situation converges at the level of Raf or MEK 15, 16.

Figure 2.

Cross-talk between integrins and growth factor receptors leading to enhanced Erk signalling. (A) Activation of parallel pathways converges at the level of MEK. (B) Integrin-mediated clustering of signalling proteins enhances RTK signalling. (C) Integrin-mediated actin reorganization causes nuclear shape changes and cooperates with growth factor-stimulated Erk signalling at the transcriptional level. (D) Integrin- or ECM-mediated modification of growth factors or transactivation of RTKs augments RTK signalling

Secondly, integrins may increase signals generated by growth factor receptors by bringing kinases and substrates in close proximity (Figure 2B). ECM binding initiates clustering of integrins in the plane of the membrane and reorganization of the actin cytoskeleton, which further stimulates the organization of integrins and associated proteins into large multi-protein aggregates, termed cell–matrix adhesions 17, 18. As activated forms of signalling proteins such as focal adhesion kinase (FAK), Src, p130Cas, integrin-linked kinase (ILK), and Erk are clustered in cell–matrix adhesions 18, 19, signal transduction may be strongly amplified at these sites. Moreover, even mRNA and ribosomes 20 can be found in these adhesions, suggesting that they may locally connect signal transduction to protein synthesis.

Thirdly, cell–matrix adhesions act as anchoring sites for the actin cytoskeleton and as such they allow the generation of tension and shape changes (Figure 2C). Via actin cytoskeletal connections with the nucleus such changes can affect the nuclear shape and chromatin structure, which might explain the profound effect of integrin-mediated cell adhesion on the expression of genes 21, 22.

Fourthly, integrin-mediated adhesion can cluster and transactivate several receptor tyrosine kinases (RTKs) including PDGFR and EGFR 23, 24, Ron 25, Met 26, and VEGFR 27 (Figure 2D). As integrin-mediated transactivation of RTKs can occur in the absence of cognate growth factor, this phenomenon may in fact explain some of the signalling events previously attributed more directly to integrin signalling.

Finally, integrins regulate RTK signalling much further upstream through their ability to organize the ECM (Figure 2D). Proteoglycans in the ECM can bind and structurally modify various growth factors and integrin-mediated cell adhesion, then allows their subsequent presentation to growth factor receptors 28. There is also evidence for direct integrin-mediated modification of a growth factor; latent TGF-β is processed upon binding to integrin αvβ6, providing a mechanism for local activation 29. The in vivo relevance of αvβ6-mediated TGF-β activation was recently demonstrated; mice in which the gene encoding the integrin β6 subunit is deleted show increased levels of the metalloproteinase MMP12 in the lungs and develop age-related pulmonary emphysema that can be overcome by the loss of MMP12 or the transgenic expression of either β6 or activated TGF-β 30.

Regulation of proliferation, survival, and differentiation by integrin signalling

Integrin-mediated cell adhesion regulates the G1 phase of the cell cycle (Figure 3A). Integrins cooperate with RTKs to stimulate cyclin D1 expression and to suppress cyclin-dependent kinase inhibitor (cki) levels. In doing so, integrins support the cyclin E-cdk2 activity that drives S-phase entry 31, 32. The enhanced and sustained Erk activity in adherent cells (see above) explains in part the supportive role of integrins in cyclin D1 transcription 16, 33. Many pathways that connect integrins to Erk activation have been implicated in cell proliferation, such as the assembly of protein complexes including Src, FAK, and p130Cas 10 or including Fyn, caveolin, and Shc 34. Whichever pathway cells use, the organization of the actin cytoskeleton by integrins through modulation of the activity of Rho family GTPases is important. The combination of integrins expressed on a cell affects the activity of Rho GTPases 35 and both Rac and Rho have been implicated in integrin-mediated control of the levels of cyclin D1 and cki 36–38.

Figure 3.

Integrins regulate proliferation and survival. (A) Integrin-mediated cell adhesion supports growth factor-mediated cyclin D1 accumulation and suppresses cki levels—Rho, Rac, and Erk are all involved. (B) Integrin-mediated cell adhesion supports survival through suppression of caspases and p53

Most adherent cell types depend on integrin-mediated adhesion to the ECM for survival. Loss of adhesion causes cells to undergo apoptosis, a process referred to as anoikis (Figure 3B). This is most evident for endothelial and epithelial cells but has also been reported for fibroblasts 39. Integrin-mediated cell adhesion stimulates PI3K-mediated PKB/AKT activity 40, 41 and, in doing so, suppresses caspase levels, promotes a high bcl-2/Bax ratio, and suppresses p53 activity 42, 43. Integrins that are not ligand bound can also trigger apoptosis of fully adherent cells by recruitment and activation of caspase-8 44–46, suggesting that a given integrin expression profile renders a cell dependent on a specific ECM environment for its survival. Finally, in a three-dimensional culture system, the α6β4 integrin is required for the maintenance of a polarized morphology and NFκB-mediated survival of normal as well as malignant mammary epithelial cells 47.

Integrin-mediated cell adhesion also regulates the expression of genes related to differentiation. Adhesion to basement membrane components stimulates the synthesis of milk proteins by increasing the phosphorylation of the prolactin receptor in cultured mammary epithelial cells 48, 49. Another example of regulation of differentiation by integrins in vitro is the inhibition by blocking integrin antibodies of the formation of contracting myotubes and expression of meromyosin by embryonic myoblasts 50. Integrin-mediated adhesion also primes monocytes for inflammatory responses 51. Finally, the expression of involucrin, which is associated with terminal differentiation of cultured keratinocytes under semi-solid conditions, is inhibited by the integrin-ligand fibronectin 52.

Thus, many in vitro studies have shown that integrin-mediated cell adhesion can regulate signal transduction pathways involved in proliferation, survival, and differentiation. In the next paragraphs we review genetic evidence for the role of integrins in cell behaviour in vivo.

Integrin function in invertebrates

Integrins, ECM components, and many integrin-associated proteins are well conserved in vertebrates and invertebrates 1. Nematodes express two integrins, representing two classes of integrins, namely those binding laminin in basement membranes and those binding the RGD motif, which is present in several different ECM components (eg fibronectin) (Figure 1B). These two classes are also represented in flies (in which fibronectin is not present and the RGD-containing ECM component is tiggrin) and mammals, but additional integrins in these organisms allow integrin-mediated recognition of multiple other ligands including, for instance, collagens and immunoglobulin-type counter-receptors in mammals. Genetic studies in nematodes and fruit flies have demonstrated that integrin-mediated formation of ECM, adhesion to and migration on ECM, as well as integrin regulation of gene transcription are all important for development and maintenance of tissues.

The importance of integrin-mediated anchorage of cells to the ECM is particularly obvious in the developing muscles of flies and nematodes. In the fly, βPS-deficient muscle cells fail to connect to the ‘tendon’ matrix and in the worm βpat3-deficient muscle cells fail to attach to the ‘cuticle’ basement membrane 53–55. As a result, these muscles detach from the ECM upon their first contraction and round up. Cultured muscle cells from βPS-deficient Drosophila embryos can form multinucleated sarcomeric structures but their organization is severely impaired 56. Similar phenotypes have been found in flies and worms lacking components of integrin-containing cell–matrix adhesions such as talin, vinculin, ILK, and others, probably because these proteins mediate the connection with the actin cytoskeleton. Another good example of the critical role of integrins as mediators of stable anchorage is the fly wing. Here, αPS1βPS and αPS2βPS integrins connect the dorsal and ventral epithelial sheets, respectively, to a thin layer of ECM proteins. Hypomorphic integrin mutations or generation of clonal patches of epithelial cells lacking integrin subunits each result in the formation of liquid-filled blisters 57, 58. As a final example, there is evidence that integrin-mediated adhesion may play an important role in synapse plasticity in the central nervous system (CNS). Volado mutants in Drosophila display impaired olfactory memories and it was shown that these flies lack the αPS subunit that is expressed preferentially in the dendrites and mushroom body neurons implicated in olfactory learning 59.

Integrins are important for cell migration during development of invertebrates. In flies, the formation of the midgut and visceral branching of the tracheal system is controlled by integrins. Supposedly, the RGD-binding integrins in the visceral muscle cells are involved in the assembly of an ECM which, in turn, is a substrate for the laminin-binding integrins in the migrating gut and tracheal cells 14. In nematodes, integrins do not seem to play a role in the migratory events during early development but at later stages mutations in the βpat-3 subunit prevent migration of gonadal distal tip cells 60 and mutations in the αina-1 subunit prevent the migration of neuronal cell bodies on laminin and cause defective axon fasciculation 61. Intriguingly, a mutant βpat-3 subunit, which lacks the two conserved tyrosines in its cytoplasmic tail, cannot compensate for the absence of βpat-3, and a genetic interaction between αina-1, the βpat-3-binding kinase MIG-15, and the Rho family GTPase, Rac has recently been discovered 60, 62. MIG-15 may regulate the function of the laminin-binding integrin in response to Rac activity or, alternatively, αina-1/βpat3-binding to laminin may control Rac-mediated migration via MIG-15. Finally, in the developing nervous system of flies, integrins are required to turn or stop axon movement 63. A genetic interaction has been discovered between integrins, Src, and p190RhoGAP, a negative regulator of the Rho GTPase, which may inhibit axon branch stability 64. A model was proposed in which integrins, perhaps via Src, inhibit p190RhoGAP, and thus support a local increase in Rho-mediated actomyosin contractility and axon branch retraction.

Integrins also appear to control signal transduction processes during fly development. Gene transcription in the developing gut is deregulated in βPS knockout Drosophila embryos that lack all integrins. Interestingly, a transmembrane chimera of the βPS cytoplasmic domain rescues transcription, demonstrating that integrin-mediated cell adhesion is not required 65. Expression of the chimera may provide a scaffold that leads to enhanced growth factor signalling in these animals. Also, in the body wall of the Drosophila larva, where muscle cells attach to epidermal tendon cells, integrins are required for efficient ligand presentation to the EGFR on tendon cells, which promotes their differentiation 14.

Thus, the role of integrins in adhesion, migration, and control of differentiation is important in invertebrate development. As discussed below, genetic studies in mice have demonstrated that integrins play similar roles in vertebrates and have also implicated integrins in the control of survival and proliferation of cells in various tissues.

Integrins in mammals: mouse genetics and human diseases

In line with important roles of integrins in cell migration and the formation of basement membranes during early development of invertebrates, mutant mice that lack ECM components such as laminin or fibronectin, or those lacking the β1 family of integrins, die during the early steps of embryonic development. Nevertheless, the disruption of some integrin subunit genes leads to more subtle phenotypes and advanced genetic approaches have been developed that allow the generation of viable and fertile animals in which integrins are deleted only in certain organs. Some of these models mimic diseases in humans. Here, we give a few such examples and also discuss some potential anti-integrin therapeutic strategies. For a more detailed description of integrin knockout phenotypes in mice we refer to some excellent recent reviews 12, 13.

Skin

In the skin, a specialized type of cell–matrix adhesion structure, termed the hemidesmosome, connects basal keratinocytes to the laminin-rich basement membrane. Inactivating mutations in the genes encoding the α3, β3, or γ2 chains of laminin-5, the hemidesmosomal integrin α6β4, or other hemidesmosomal proteins such as plectin or BP180 have been identified in patients suffering from epidermolysis bullosa. In these patients, as well as in mice in which these genes are inactivated, the epithelium detaches from the underlying basement membrane upon application of mechanical stress 66, 67. Additionally, the laminin-binding integrin α3β1 seems to be critically involved in the formation of the basement membrane and mice lacking α3β1 also suffer from mild blistering of the skin.

Normally, the epidermis is renewed approximately every 14 days, epidermal stem cells giving rise to transit amplifying cells with limited proliferation potential, before they move upwards where they stop dividing and ultimately undergo terminal differentiation. Integrins of the β1 family are confined to the first few cell layers of the epidermis and keratinocyte differentiation is accompanied by a loss of expression of β1 integrins. It has been shown that epidermal stem cells can be isolated based on their strong expression of β1 integrins 68 and transgenic mice that express β1 integrins suprabasally display a sporadic psoriatic phenotype with increased levels of Erk activity in the hyperproliferative epidermal regions 69. However, in conditional mutant mice that lack the β1 subunit exclusively in keratinocytes, basement membrane organization, epidermal proliferation, and hair follicle morphogenesis were affected, demonstrating an important role for β1 integrins in these processes but arguing against an essential role of β1 integrins in the epidermal stem cell compartment 70, 71. Nevertheless, taken together, these studies show that integrins play important structural as well as signalling roles in the development and maintenance of the integrity of the skin.

Cardiovascular system

β1-Knockout embryos die during the very early steps of development but β1 chimeric mice have provided further insight into the role of β1 integrins during later stages. In these mice, no β1-deficient endothelial cells are found in the liver and spleen, demonstrating that blood vessel formation requires β1 integrins 72. Endothelial cells express multiple β1 integrins and mice lacking fibronectin or the α5β1 fibronectin receptor die during embryogenesis with a defective organization of heart and blood vessels 73, 74. α4 null mice display defects in the development of the epicardium and lack coronary vessels, leading to cardiac haemorrhage 75, 76. Epicardial precursor cells are released from the proepicardial serosa as cysts and migrate onto the myocardial surface of the heart. Firstly, α4β1 is involved in the budding off of the cysts and, secondly, epicardial precursor cells that do reach the heart in α4 null embryos fail to spread and migrate on the myocardial surface.

Endothelial cells express several αv integrins but αv null embryos develop normally to E9.5 with extensive vasculogenesis and angiogenesis 77. Expression of αvβ3 is prominent on proliferating endothelial cells but mice lacking αvβ3, αvβ5, or both show no defects in angiogenesis 78. It is especially surprising that tumour angiogenesis, as well as the angiogenic response to hypoxia or VEGF, is augmented in these mice, considering the fact that antagonists of αvβ3 or αvβ5 interfere with bFGF- and VEGF-induced neovascularization, respectively 79. Thus, while αv integrins are attractive targets for antiangiogenic therapy in oncology and perhaps other diseases such as rheumatoid arthritis (a humanized antibody against αvβ3 has entered clinical trials 80), the mode of action of such a strategy remains to be established.

CNS

Laminin and the laminin-binding integrins α3β1 and α6β1 are important for proper formation of the cerebral cortex. The cortical basement membrane is disrupted in α6-null mice and they show ectopia in the cerebral cortex and retina 81. Likewise, α3-null mice show a disorganized cortex and they have defective neuronal migration—a phenotype also seen in mice lacking ‘Reelin’, which is secreted by neurons of the developing brain and provides a stop signal for neuronal migration along the radial glia cells 82. It has been suggested that Reelin is a ligand for α3β1 but subsequent reports show that Reelin is in fact an extracellular serine protease that degrades fibronectin and laminin 83, 84. Therefore, its main function may be the local degradation of the ECM.

Yet another laminin-binding integrin, α7β1, is induced on regenerating motor and sensory neurons following axotomy and is required for axonal regeneration 85. During development of the CNS, survival of differentiating oligodendrocytes depends on axonal contact. Recently, increased cell death of myelin-forming oligodendrocytes was reported in the developing brain of α6-null mice 86. The authors showed that binding of α6β1 to axonal laminins switches the response to neuregulins from proliferation to enhanced survival and differentiation. Expression of a chimeric integrin containing the α5 extracellular and transmembrane domain fused to the α6 cytoplasmic tail in these cells led to enhanced survival on fibronectin, demonstrating that the survival signal is generated by the α6 cytoplasmic tail. Taken together, various integrins are implicated in the migratory behaviour of neurons and there is evidence for modulation of differentiation processes during the development of the CNS.

Urogenital tract

Two integrins have been implicated in kidney function. The branching of integrin α3β1-deficient glomerular capillaries is reduced and podocyte foot processes are immature 87. As α3β1 is a laminin receptor, it is not surprising that kidney development is also impaired in mice deficient in one of the laminin chains 88. In analogy to the apparent role of α3β1 in the formation of the basement membrane in the skin (see above), this integrin may have a similar function in basement membrane deposition in the kidney. Thus, the abnormal phenotype seen in the kidney of α3 null mice may be due to defective basement membrane development. Integrin α8β1-deficient mice also show a defect in kidney development with abnormal outgrowth and branching of the ureteric buds, leading to a severe dysfunctioning or even absence of the kidney 89. In these mice the interaction between the ureteric epithelium and the metanephric mesenchyme, which is essential for kidney morphogenesis, is disrupted. The relevant ligand for α8β1 in these structures is not known. Some can be ruled out as they are undetectable in kidney, while mice deficient in osteopontin, an α8β1 ligand which is expressed in the kidney, show no apparent abnormal kidney phenotype. Nephronectin is another ligand of α8β1 that is expressed in kidney 90, but it remains to be established which role nephronectin plays in the development and function of the kidney.

Skeleton

The roles of integrins in bone development remain incompletely understood but two phenotypes have been well characterized. First, a role of αvβ3 in osteoclast function has been reported. The β3-deficient osteoclasts fail to spread in vitro and to produce membrane ruffles in vivo, pointing to an impaired organization of actin structures, perhaps due to defective signalling to Rho family GTPases. This may explain the fact that such osteoclasts cannot efficiently resorb bone, leading to an increased bone mass in the β3 null mice 91. Second, mice lacking both the α3β1 and α6β1 laminin receptors, or mice lacking the laminin α5 chain, show severe skeletal abnormalities with shortened and abnormally shaped limbs 92, 93. Thus, integrins are involved in both aspects of bone development, ie the deposition and the resorption of bone.

Skeletal muscle

During development, vertebrate embryonic muscle cells express several different integrins but after birth most of them have disappeared and the main integrin is α7β1 in which the common β1A subunit has been switched to the muscle-specific splice variant, β1D 94. Expression of α7β1 is observed within myotendinous and neuromuscular junctions as well as outside those regions (the localization may be regulated by the α7 cytoplasmic domain which, like that of β1, is subject to alternative splicing). As in Drosophila (see above) the integrin in the myotendinous junctions is important in the organization and maintenance of the connection between muscle and tendon 95. In vertebrates, it is α7β1D that is required for the maintenance of adult muscles. The α7β1 integrin is a laminin receptor and loss of either the α7 subunit or its ligand laminin 2/4 leads to congenital muscular dystrophy in human patients as well as in mice 96–99.

In addition to a role of β1 integrins in regulating muscle fiber integrity, there is evidence that these integrins influence muscle development. Muscle defects have been observed in mice chimeric for α5-null cells and in conditional knockout mice lacking β1 in developing myoblasts 100, 101. However, in β1-chimeric mice, no such defects were reported 102. Although in the chimeric, wild type cells may have facilitated migration and cell fusion of the β1-deficient cells into the muscle fibers, their presence is not necessary, since β1-deficient ES cells induced to differentiate into myotubes in vitro can fuse 103. This finding, although arguing against a direct role of β1 integrins in cell fusion, does not exclude the possibility that β1 integrin influences the expression and function of other receptors important for cell fusion in vivo. Indeed, it was shown that the expression of CD9, the integrin-associated tetraspanin that regulates in vitro myoblast and sperm-egg fusion 104, was strongly reduced on the surface of β1-deficient myoblasts 101. Furthermore, analysis of knockin embryos in which β1D had been replaced by β1A 105 revealed a reduction in skeletal muscle mass, when the mice were bred on a predominantly FVB background 106. This reduction appeared to be due to impaired primary myogenesis, whereas there was no direct effect on secondary myogenesis.

In conclusion, integrins are thought to play important roles in muscle function, an idea that is corroborated by mutations in patients with muscular dystrophy. Furthermore, some evidence points to a role in myogenesis in vivo, but we are still far from a clear understanding of this process.

Haematopoietic cells

Integrins play fundamental roles in the development and function of the immune system. The use of cultured β1-null progenitor cells and the analysis of β1 chimeric mice have shown that β1 integrins are required for colonization of the fetal haematopoietic organs but not for the development of haematopoietic stem cells within those organs 107. Studies using mice chimeric for either the α4, α5, or αv subunit, which are the β1-associated α subunits on leucocyte progenitors, indicate that none of these integrins is required for colonization of the fetal haematopoietic organs but they do implicate α4 in the multilineage development of haematopoietic cells and in lymphocyte homing to lymph nodes and Peyer's patches 108, 109. Defective homing to Peyer's patches is also observed in mice deficient in β7, another β subunit with which α4 can combine to form the α4β7 integrin that recognizes MadCAM, an endothelial receptor expressed in the gut 110.

The β2 family of integrins is exclusively expressed on cells of the immune system 3. In contrast to β1 and β3 integrins, which bind predominantly to ECM proteins, β2 integrins bind mainly to immunoglobulin-type counter-receptors such as ICAMs on endothelial and other immune cells and to bacterial polysaccharides. The number of circulating leucocytes in mice lacking β2 integrins is increased. The leucocytes are unable to extravasate from the blood vessels and these mice suffer from bacterial infections and impaired wound healing, a phenotype that is also observed in patients suffering from type 1 leucocyte adhesion deficiency (LAD-1) due to a lack of β2 integrins 111, 112. A more subtle form of LAD has been reported recently. The amounts of β1, β2, and β3 integrins in a patient having clinical features of LAD-1 as well as Glanzmann thrombasthenia (see below) were normal, yet their activity state could not be properly regulated 113.

Platelets express multiple integrins but α2β1 and αIIbβ3 in particular are thought to be involved in haemostasis. Integrin αIIbβ3 is the ‘effector integrin’ on platelets. When this integrin is activated, it binds fibrinogen and thereby promotes the formation of platelet aggregates that close a wound. In line with such an important function, null mutations in the genes encoding αIIb or β3 cause a bleeding disorder termed Glanzmann thrombasthenia in mice as well as in human patients 114, 115. The fact that platelet numbers are not affected in the β3 null mice shows that β3 integrins are not critically involved in the development of the megakaryocytic lineage. The binding of αIIbβ3 to fibrinogen triggers several cellular responses, including actin cytoskeletal remodelling, which is important for platelet aggregation. Indeed, mice in which the two conserved tyrosines in the cytoplasmic tail of the β3 subunit were mutated to phenylalanine show defective platelet aggregation and clot retraction in vitro and suffer from a bleeding disorder in vivo, indicating that αIIbβ3-mediated ‘outside-in’ signalling is crucial to proper platelet function in vivo116. Because of the central role of this integrin in haemostasis, αIIbβ3 antagonists are currently being tested in clinical trials 117.

In contrast to that of αIIbβ3, the role of α2β1 is unclear. Obviously, the activity of αIIbβ3 must be tightly controlled to prevent inappropriate blood clotting. Besides thrombin binding to its receptor on platelets, the interaction of platelets with exposed collagen in damaged blood vessel walls can also shift αIIbβ3 from a low-affinity to a high-affinity state. It is in this process that the collagen-binding integrin α2β1 has been thought to act as a ‘regulatory integrin’. However, despite in vitro evidence for a role of α2β1 in collagen-induced platelet aggregation, mice lacking β1 in their megakaryocytes do not show increased bleeding (in contrast to the profound effect of deletion of the collagen receptor GPVI), which argues against an important role of α2β1 in platelet function in vivo118.

Thus, there is evidence that integrins are involved in the development of the immune system and that integrins are crucial for the functional integrity of immunity and haemostasis.

Concluding remarks

Integrins are adhesion receptors found in metazoans ranging from sponges to worms, flies, and mammals. They mediate stable adhesion as well as migration and modulate virtually every intracellular signalling cascade. How integrins fulfil these functions remains a topic of ongoing study. Regulation of the actin cytoskeleton appears to be their core function and may explain the anchorage dependence of proliferation, survival, and differentiation events. Progress in understanding the role of integrins will require more sophisticated models. On the one hand, the increasing use of three-dimensional culture systems will produce in vitro data that more closely resemble the in vivo situation. On the other hand, more subtle genetic models for integrins, associated proteins, and extracellular matrix proteins are being generated. Conditional mutants generating a spatially or temporally restricted null phenotype, as well as fine-tuned mutations which knock out only one specific functional property of the protein involved, will allow a much more detailed understanding of the highly complicated function of integrin-mediated adhesion in development and maintenance of tissues.

Acknowledgements

We thank Ed Roos (Division of Cell Biology, The Netherlands Cancer Institute, Amsterdam, The Netherlands) and Sólveig Thorsteinsdóttir (Department of Animal Biology, Faculty of Sciences, University of Lisbon, Lisbon, Portugal) for critically reading the manuscript. This work was supported by grants from the Dutch Cancer Society (NKI 1999–2117 and NKI 2001–2488).

Ancillary