The integrins are a superfamily of cell adhesion receptors that recognize mainly extracellular matrix ligands and cell-surface ligands, although some soluble ligands have been identified 1. They are transmembrane αβ heterodimers, and at least 18 α and eight β subunits are known in humans 2 (Figure 1; lists of the integrin subunits present in mouse, chicken, zebrafish, Caenorhabditis elegans, Xenopus laevis and Drosophila melanogaster are given in Additional data file 1). Integrin α and β subunits are totally distinct, with no detectable homology between them; sequence identity among α subunits is about 30% and among β subunits 45%, indicating that both the α and the β gene families evolved by gene duplication (Figure 2). In humans, genes for both α and β subunits are located on various chromosomes. However, genes for integrins expressed in leukocytes (subunits αL, αM, αD, and αX) are clustered at 16p11, while for those expressed in platelets and endothelial cells, the αIIb and β3 genes are at 17q21.32, and the α6, α4, and αV cluster at 2q31 (Table 1). Some integrin ⟨ subunits (⟨1, ⟨2, ⟨10, ⟨11, ⟨M, ⟨L, ⟨D, and ⟨X) contain a so-called I (insertion or interaction), or A, domain, while others do not. The I-domain integrin α subunits are closely related to each other (Figure 2a). Also closely related to each other are the family of non-I-domain α subunits that recognize the RGD motif (αV, α8, α5, and αIIb) and the family of laminin-binding α subunits (α3, α6, and α7). Studies on integrin genes from lower and higher eukaryotes clearly indicate that integrin genes (both α and β) derived from a common ancestral gene by gene duplications. A genomic analysis among 24 invertebrate and vertebrate species revealed that the α and β integrin structure, along with the inserted α I domain, has been highly conserved during the evolution of vertebrates 3.

The crystal structures of human integrins αVβ3 45 and αIIbβ3 6 show that the extracellular portion of an integrin heterodimer consists of multiple domains (Figure 3a). The headpiece of αVβ3, which contains the ligand-binding site, consists of the β-propeller domain and the plexin-semaphorin-integrin (PSI) domain of the αV subunit, and the β I-like, or βA, domain and the hybrid domain of the β subunit. The β-propeller domain contains seven repeats of about 60 amino acids each that fold into a seven-bladed β-propeller structure similar to the β subunit of a heterotrimeric G protein. I domains contain a metal-ion-dependent adhesive site (MIDAS) and I-like domains contain a structurally similar metal-binding motif. The RGD-binding site is located at the interface between the β-propeller domain and the β I-like domain and amino-acid residues from the two domains interact directly with the RGD peptide of a ligand 5. Mutagenesis studies have identified many other amino-acid residues that are critical for ligand binding 78. These residues are discontinuous in the primary structure but are exposed on the surface of the headpiece and generate the ligand-binding surface. The crystal structure of a complete integrin-ligand complex has not yet been published, but by comparing the crystal structures of RGD-bound and unbound forms it has been found that the disulfide-linked loop structure in the β I-like domain under goes conformational changes (a movement of 1.5 Å towards the RGD peptide), and the α helix 7 in the β I-like domain moves downward on ligand binding 45. Also, the hybrid domain swings outward upon integrin activation. In the I-domain integrins, the I domain can be present in either an open (active) or a closed (inactive) conformation. These are major conformational changes that affect ligand binding in the headpiece.

The cytoplasmic tails of human integrin subunits are less than 75 amino acids long (the β4 tail is an exception at a length of approximately 1,000 amino acids, which includes four fibronectin type III repeats). There is striking homology among the β-subunit cytoplasmic tails, but the α-subunit tails are highly divergent except for a conserved GFFKR motif next to the transmembrane region, which is important for association with the β tail. A large number of cytoskeletal and signaling proteins have been reported to bind to β cytoplasmic tails and some have been found to interact with specific α tails. Most integrin β tails contain one or two NPxY/F motifs (where x is any amino acid) that are part of a canonical recognition sequence for phosphotyrosine-binding (PTB) domains, which are protein modules present in a wide variety of signaling and cytoskeletal proteins. Phosphorylation of the tyrosine (Y) in the NPxY/F motif may represent a mode of regulating integrin interactions with other proteins at the cytoplasmic face of the plasma membrane. The integrin tails recruit several proteins, such as talin, that bind actin filaments, and thus form a connection to the cytoskeleton, a connection that is essential for most, if not all, integrin-mediated functions. The structural basis for talin's unique ability to activate integrins through PTBs has been defined 9. Structural data on integrins are mostly derived from mouse and human and the structural basis for the activation of integrins through their cytoplasmic domains in other species is not yet known.

Integrins function as traction receptors that can both transmit and detect changes in mechanical force acting on the extracellular matrix. In mammals, some integrins are limited to certain cell types or tissues: αIIbβ3 to platelets; α6β4 to keratinocytes; αEβ7 to T cells, dendritic cells and mast cells in mucosal tissues; α4β1 to leukocytes; α4β7 to a subset of memory T cells; and the β2 integrins to leukocytes. Other integrins are widely distributed, such as αVβ3, which is expressed on endothelium. The RGD sequence in fibronectin was originally identified as an integrin-binding motif 10 and this and related sequences in extracellular matrix molecules do act as integrin-binding motifs in vivo. However, integrins also recognize many non-RGD sequences in their ligands, such as the tripeptide LDV in the immunoglobulin superfamily member vascular cell adhesion molecule 1 (VCAM-1), which is expressed on inflamed endothelium and is bound by α4β1. This pattern of integrin recognition and activation appears to be conserved among most mammals studied.

In regard to ligand specificity, the mammalian integrins can be broadly grouped into laminin-binding integrins (α1β1, α2β1, α3β1, α6β1, α7β1, and α6β4), collagen-binding integrins (α1β1, α2β1, α3β1, α10β1, and α11β1), leukocyte integrins (αLβ2, αMβ2, αXβ2, and αDβ2), and RGD-recognizing integrins (α5β1, αVβ1, αVβ3, αVβ5, αVβ6, αVβ8, and αIIbβ3). Individual integrins have unique ligand specificities (Table 2). They are further defined by those α subunits that can contain the I domain: these are α1, α2, α10, α11, αL, αM, αX, αD, and αE. Non-I-domain subunits are α3, α4, α5, α6, α7, α8, α9, αV, and αIIb. In I-domain integrins, the I domains play a central role in ligand binding and intercellular adhesion. Integrin binding among invertebrate species is less well studied; RGD sequences have been found in species as diverse as sea urchins and amoebae, however, and integrins and their biochemical functions are likely to be highly conserved in metazoa, due to the essential nature of their function.

Upon binding an extracellular ligand, integrins generate an intracellular signal and, conversely, their functioning can be regulated by signals from within the cell 1. They serve as transmembrane links between extracellular contacts (other cells or the extracellular matrix) and the actin microfilaments of the cytoskeleton, whose behavior integrins also regulate and modulate. Many different proteins on the cytoplasmic side of the membrane, such as talin, vinculin, and ERM (ezrin, radixin, moesin) actin-binding proteins, act as linker proteins to connect the cytoplasmic domains of integrins to the cytoskeleton, resulting in complex interactions 1. Extracellular ligation of integrins triggers a large variety of signal transduction events that modulate cell behaviors such as adhesion, proliferation, survival or apoptosis, shape, polarity, motility, haptotaxis, gene expression, and differentiation, mostly through effects on the cytoskeleton.

The deletion of individual genes by gene knockout in mice shows that particular integrins play a critical role in development (the β1 integrins), vasculogenesis (αV integrins), lymphangiogenesis (α9β1), thrombus formation (αIIbβ3), the integrity of the skin (α6β4), and immune responses (the β2 integrins) (Table 3). Knockout of the gene for β3 enhanced tumorigenesis and angiogenesis 1112, enhanced wound healing 13, and enhanced inflammation and atherosclerosis 14, suggesting that αVβ3 normally suppresses these processes.

Integrins are transmembrane molecules that are essential for both embryonic development and immunological function by binding to a wide variety of ligands, including extra-cellular matrix molecules and members of the immunoglobulin superfamily. Their capacity to specifically recognize particular amino-acid motifs and regulate binding affinity to them lies in their heterodimeric structure. This molecular design incorporates a remarkable ability to direct conformational changes initiated at the cytoplasmic domain, and also to signal extracellular ligand binding back to the inside of the cell. Much of our current knowledge of the myriad of functions attributed to ligand binding of a particular αβ pair comes from gene knockout studies in mouse or from rare hereditary disorders in humans. Only a handful of crystal structures of integrins bound to their ligands have been solved. From these data it appears that small variations in the particular structure or charge of a ligand (that is, down to single atoms) can strongly influence the binding affinity and the capacity of the integrin to maintain a conformation that signals back into the cell. This implies that ligand binding can influence allosteric changes in the integrin, which in turn dictate how the integrin reports on the environment in which the cell finds itself. Thus, integrins serve as both sensors of their molecular surroundings and effectors that conduct motile forces exerted by the cell's cytoskeleton and from the dynamic environment (that is, shear forces within blood vessels). We are just beginning to understand the structural and chemical basis of this sensor-effector system. A particularly exciting development is the discovery of small molecules that bind tightly to the ligand-binding pocket or to other domains and allosterically stabilize integrin conformations that promote or antagonize binding. For instance, small molecules have been discovered that can allosterically tune conformations of α_Lβ₂ that favor low, intermediate, and high-affinity binding. In this manner, it is possible to steer the adhesive response of a leukocyte in a blood vessel to promote tethering and rolling, firm arrest, or no binding at all. It may be possible to apply such small molecules as therapeutics either to promote leukocyte recruitment at sites of infection or to block their accumulation in chronic inflammatory diseases such as in rheumatoid arthritis and psoriasis. As more knowledge accumulates relating amino-acid sequence to common structural motifs associated with the allosteric control of ligand recognition and outside-in signaling to the cytoplasm, it will become possible to design small molecules that target these critical domains.

Additional data are available online with this article: Additional data file 1 contains tables of the integrin subunits present in the mouse, chicken, zebrafish, nematodes, Xenopus laevis, and D. melanogaster.