Integrins are transmembrane receptors that help cell–cell and cell–extracellular matrix (ECM) adhesion. Upon ligand binding, integrins activate signal transduction pathways that mediate cellular signals such as regulation of the cell cycle, organization of the intracellular cytoskeleton, and movement of new receptors to the cell membrane. The presence of integrins allows rapid and flexible responses to events at the cell surface (e.g. signal platelets to initiate an interaction with coagulation factors).

Several types of integrins exist, and one cell generally has multiple different types on its surface. Integrins are found in all animal cells while integrin-like receptors are found in plant cells.

Integrins work alongside other proteins such as cadherins, the immunoglobulin superfamily cell adhesion molecules, selectins and syndecans, to mediate cell–cell and cell–matrix interaction. Ligands for integrins include fibronectin, vitronectin, collagen and laminin.

Structure Integrins are obligate heterodimers composed of α and β subunits. Several genes code for multiple isoforms of these subunits, which gives rise to an array of unique integrins with varied activity. In mammals, integrins are assembled from eighteen α and eight β subunits, in Drosophila five α and two β subunits, and in Caenorhabditis nematodes two α subunits and one β subunit. The α and β subunits are both class I transmembrane proteins, so each penetrates the plasma membrane once, and can possess several cytoplasmic domains.

+ alpha gene protein synonyms CD49a α1, VLA1 CD49b α2, VLA2 CD49c α3, VLA3 CD49d α4, VLA4 CD49e α5, VLA5 CD49f α6, VLA6 ITGA7 α7, FLJ25220 ITGA8 α8 ITGA9 α9, RLC ITGA10 α10, PRO827 ITGA11 α11, HsT18964 CD11d αD, FLJ39841 CD103 αE, HUMINAE CD11a αL, LFA1A CD11b αM, MAC-1 CD51 αV, VNRA, MSK8 CD41 αIIb, GPIIb CD11c αX

+ beta gene protein synonyms CD29 β1, FNRB, MSK12, MDF2 CD18 β2, LFA-1, MAC-1, MFI7 CD61 β3, GP3A, GPIIIa CD104 β4 ITGB5 β5, FLJ26658 ITGB6 β6 ITGB7 β7 ITGB8 β8

Variants of some subunits are formed by differential RNA splicing; for example, four variants of the beta-1 subunit exist. Through different combinations of the α and β subunits, 24 unique mammalian integrins are generated, excluding splice- and glycosylation variants.

Integrin subunits span the cell membrane and have short cytoplasmic domains of 40–70 amino acids. The exception is the beta-4 subunit, which has a cytoplasmic domain of 1,088 amino acids, one of the largest of any membrane protein. Outside the cell membrane, the α and β chains lie close together along a length of about 23 nm; the final 5 nm N-termini of each chain forms a ligand-binding region for the ECM. They have been compared to lobster claws, although they don't actually "pinch" their ligand, they chemically interact with it at the insides of the "tips" of their "pinchers".

The molecular mass of the integrin subunits can vary from to . Beta subunits have four cysteine-rich repeated sequences. Both α and β subunits bind several divalent cations. The role of divalent cations in the α subunit is unknown, but may stabilize the folds of the protein. The cations in the β subunits are more interesting: they are directly involved in coordinating at least some of the ligands that integrins bind.

Integrins can be categorized in multiple ways. For example, some α chains have an additional structural element (or "domain") inserted toward the N-terminal, the alpha-A domain (so called because it has a similar structure to the A-domains found in the protein von Willebrand factor; it is also termed the α-I domain). Integrins carrying this domain either bind to collagens (e.g. integrins α1 β1, and α2 β1), or act as cell-cell adhesion molecules (integrins of the β2 family). This α-I domain is the binding site for ligands of such integrins. Those integrins that don't carry this inserted domain also have an A-domain in their ligand binding site, but this A-domain is found on the β subunit.

In both cases, the A-domains carry up to three divalent cation binding sites. One is permanently occupied in physiological concentrations of divalent cations, and carries either a calcium or magnesium ion, the principal divalent cations in blood at median concentrations of 1.4 mM (calcium) and 0.8 mM (magnesium). The other two sites become occupied by cations when ligands bind—at least for those ligands involving an acidic amino acid in their interaction sites. An acidic amino acid features in the integrin-interaction site of many ECM proteins, for example as part of the amino acid sequence Arginine-Glycine-Aspartic acid ("RGD" in the one-letter amino acid code).

Structure Despite many years of effort, discovering the high-resolution structure of integrins proved to be challenging, as membrane proteins are classically difficult to purify, and as integrins are large, complex and highly glycosylated with many sugar 'trees' attached to them. Low-resolution images of detergent extracts of intact integrin GPIIbIIIa, obtained using electron microscopy, and even data from indirect techniques that investigate the solution properties of integrins using ultracentrifugation and light scattering, were combined with fragmentary high-resolution crystallographic or NMR data from single or paired domains of single integrin chains, and molecular models postulated for the rest of the chains.

The X-ray crystal structure obtained in 2001 for the complete extracellular region of one integrin, αvβ3, shows the molecule to be folded into an inverted V-shape that potentially brings the ligand-binding sites close to the cell membrane. Perhaps more importantly, the crystal structure was also obtained in 2003 for the same integrin bound to a small ligand containing the RGD-sequence, the drug cilengitide. As detailed above, this finally revealed why divalent cations (in the A-domains) are critical for RGD-ligand binding to integrins. The interaction of such sequences with integrins is believed to be a primary switch by which ECM exerts its effects on cell behaviour.

The structure poses many questions, especially regarding ligand binding and signal transduction. The ligand binding site is directed towards the C-terminal of the integrin, the region where the molecule emerges from the cell membrane. If it emerges orthogonally from the membrane, the ligand binding site would apparently be obstructed, especially as integrin ligands are typically massive and well cross-linked components of the ECM. In fact, little is known about the angle that membrane proteins subtend to the plane of the membrane; this is a problem difficult to address with available technologies. The default assumption is that they emerge rather like little lollipops, but there is little evidence for this. The integrin structure has drawn attention to this problem, which may have general implications for how membrane proteins work. It appears that the integrin transmembrane helices are tilted (see "Activation" below), which hints that the extracellular chains may also not be orthogonal with respect to the membrane surface.

Although the crystal structure changed surprisingly little after binding to cilengitide, the current hypothesis is that integrin function involves changes in shape to move the ligand-binding site into a more accessible position, away from the cell surface, and this shape change also triggers intracellular signaling. There is a wide body of cell-biological and biochemical literature that supports this view. Perhaps the most convincing evidence involves the use of antibodies that only recognize integrins when they have bound to their ligands, or are activated. As the "footprint" that an antibody makes on its binding target is roughly a circle about 3 nm in diameter, the resolution of this technique is low. Nevertheless, these so-called LIBS (Ligand-Induced-Binding-Sites) antibodies unequivocally show that dramatic changes in integrin shape routinely occur. However, how the changes detected with antibodies look on the structure is still unknown.

In 2018, X-ray crystal work revealed the complete structure of an extended β3 "headpiece" via the construction of a chimeric protein: the I domain of β3 was replaced with the I domain of αL. This work confirmed the swing-out motion of the hybrid domain (resulting in headpiece opening) and the switchblade-like extending motion of the headpiece from the "leg" domains previously observed and added more resolution and detail on how the two motions interact.

High-resolution structures (< 3.15 Å) for extracellular parts of whole integrin complexes were finally obtained in 2025 thanks to advancements in cryo-TEM and a micellar construction. 11 structures were obtained for αvβ3, including six apo and five ligand-bound. (The transmembrane part is visible in the low-resolution map but failed to refine to high resolution, indicating considerable flexibility.) An intermediate state was found with CWHM-12 as the ligand called "conformation 2". The classical inactive (1) and open (3) states were also observed in detail, with a notable resemblance to two states previously observed in low resolution using the LM609 antibody (known to not bind to the LIGS).

Activation

When released into the cell membrane, newly synthesized integrin dimers are speculated to be found in the same "bent" conformation revealed by the structural studies described above. One school of thought claims that this bent form prevents them from interacting with their ligands, although bent forms can predominate in high-resolution EM structures of integrin bound to an ECM ligand. Therefore, at least in biochemical experiments, integrin dimers must apparently not be 'unbent' in order to prime them and allow their binding to the ECM. In cells, the priming is accomplished by a protein talin, which binds to the β tail of the integrin dimer and changes its conformation. The α and β integrin chains are both class-I transmembrane proteins: they pass the plasma membrane as single transmembrane alpha-helices. Unfortunately, the helices are too long, and recent studies suggest that, for integrin gpIIbIIIa, they are tilted with respect both to one another and to the plane of the membrane. Talin binding alters the angle of tilt of the β3 chain transmembrane helix in model systems and this may reflect a stage in the process of inside-out signalling which primes integrins. Moreover, talin proteins are able to dimerize and thus are thought to intervene in the clustering of integrin dimers which leads to the formation of a focal adhesion. Recently, the Kindlin-1 and Kindlin-2 proteins have also been found to interact with integrin and activate it.

Function Integrins have two main functions: attachment of the cells to the ECM and signal transduction from the ECM to the cells. They are also involved in a wide range of other biological activities, including extravasation, cell-to-cell adhesion, cell migration, and as receptors for certain viruses, such as adenovirus, echovirus, hantavirus, foot-and-mouth disease, polio virus and other viruses. Recently, the importance of integrins in the progress of autoimmune disorders is also gaining attention of the scientists. These mechanoreceptors seem to regulate autoimmunity by dictating various intracellular pathways to control immune cell adhesion to endothelial cell layers followed by their trans-migration. This process might or might not be dependent on the sheer force faced by the extracellular parts of different integrins.