Leucocytes are highly motile cells. Their ability to migrate into tissues and organs is dependent on cell adhesion molecules. The integrins are a family of heterodimeric transmembrane cell adhesion molecules that are also signalling receptors. They are involved in many biological processes, including the development of metazoans, immunity, haemostasis, wound healing and cell survival, proliferation and differentiation. The leucocyte-restricted β2 integrins comprise four members, namely αLβ2, αMβ2, αXβ2 and αDβ2, which are required for a functional immune system. In this paper, the structure, functional regulation and signalling properties of these integrins are reviewed.
- cell adhesion
Integrins are heterodimeric cell adhesion molecules that are involved in immunity, wound healing, haemostasis and the development of metazoans . They mediate cell–cell and cell–ECM (extracellular matrix) interactions. In humans, there are 24 integrins formed by specific non-covalent associations of 18 α and eight β subunits (Figure 1A). In general, each integrin subunit has a large extracellular region, a single-pass TM domain (transmembrane domain) and a short cytoplasmic tail [1,2] (Figure 1B). The extracellular region of the α subunit is composed of a β-propeller fold with seven blades similar to the Gβ of G-proteins, a thigh domain resembling a C2-set Ig (immunoglobulin) fold, and two calf domains each having two anti-parallel β-sheets [3–5]. An I domain (inserted domain), which is also known as the αA domain because of its sequence homology with the A domains (A1–A3) of von Willebrand factor , is present in nine of the α subunits, and it adopts the Rossmann nucleotide-binding fold . The extracellular region of the β subunit is composed of a PSI (plexin-semaphorin-integrin) domain, an I-like (or βA) domain that shares structural similarity with the I-domain, a hybrid domain formed by two non-contiguous primary amino acid sequences, four I-EGF (integrin-epidermal growth factor) folds and a βTD (β tail domain) [3–5,8–11]. The two α-helical TM domains of a resting integrin adopt a ridge-in-groove packing [4, 12] and the association of the TM domains is specific . The cytoplasmic tails of the β subunits other than β4 and β8 share sequence homology, including two highly conserved NxxY/F motifs (x are other amino acids) (Figure 1C). Apart from having a highly conserved juxtamembrane GFFKR motif, the α cytoplasmic tails are divergent in their lengths and sequences. Integrin cytoplasmic tails recruit cytosolic proteins and many of these interactions are modulated by integrin tail phosphorylation [14,15].
Integrin-mediated cell adhesion is dependent on bivalent cations. In the α subunit, a conserved DxSxS motif and two non-contiguous amino acids formed a single metal-ion-binding site in the I-domain known as the MIDAS (metal-ion-dependent adhesion site) (Figure 2A). The MIDAS is the primary ligand-binding site in I-domain-containing integrins. An acidic residue from the ligand and two water molecules complete the metal ion co-ordination. For example, collagen and ICAM-1 (intercellular adhesion molecule 1) binding to the α2 and αL I domains respectively involve the acidic residue glutamate from each ligand (Figure 2A) [16–18]. In the β subunit, three metal-ion-binding sites are found in the I-like domain. The MIDAS serves to bind ligands and is regulated by the ADMIDAS (adjacent to MIDAS) and LIMBS (ligand-associated metal-binding site) or SyMBS (synergistic metal ion-binding site) (Figure 2B) [11,19–21]. Integrins that lack I-domain bind ligands via a binding pocket formed by these metal-binding sites in the β I-like domain and the α subunit's seven-bladed propeller . Four Ca2+-binding sites are found in the β-hairpin loops of blades 4–7 of the propeller (Figure 2B). A Ca2+-binding site is also found at the genu of the α subunit that may have a role in regulating integrin activation (Figure 2C) .
The ligand-binding affinity of an integrin is dependent on the transmission of allostery through its α and β subunits [23,24]. A breakthrough in integrin research came from a crystallography study that reported that the extracellular portion of integrin αVβ3 has a V-like conformation . This was followed by another report on the crystal structure of integrin αVβ3 in the presence of a cyclic RGD-containing peptide . In this structure, the RGD-containing peptide bound to the interface of the αV β-propeller and β3 I-like domain, and induced conformational changes in the β3 I-like domain, but the overall V-like conformation of integrin αVβ3 remains unchanged. However, another study using EM (electron microscopy) showed that the soluble extracellular fragment of integrin αVβ3 became fully extended in the presence of a cyclic RGD-containing peptide, and using a cell-based system, it was demonstrated that the ligand-binding affinity of integrin αVβ3 or αIIbβ3 locked in the bent conformation was low . These data suggest that for the β3 integrins the conversion from a bent into the highly extended conformation correlates with a transition from low to high ligand-binding affinity. The switchblade-like model of integrin activation is further supported by EM analyses of soluble extracellular fragments of αLβ2 and αXβ2 integrins with C-terminal clasps that constrained these integrins in a bent conformation . A population of extended integrin conformers was observed when the clasps were released, and for αXβ2 when an activating antibody CBR LFA-1/2 (leucocyte function-associated antigen 1/2) Fab fragment or the small-molecule allosteric antagonist XVA143 was added . Additionally, chemokine stimulation of leucocytes has been shown to induce integrin αLβ2 extension as reported by the conformation-sensitive antibody KIM127, which recognizes an epitope in the I-EGF2 of an extended β2 subunit [9,27,28]. Hence, a bent integrin on a cell may not be poised to bind large ligands because the ligand-binding globular head is turned towards the plasma membrane. Converting an integrin from a bent into a fully extended conformation could therefore promote ligand binding (Figure 3A). However, it has been shown by EM analyses that the soluble extracellular fragment of integrin α5β1 was not bent in an obtuse angle even in the presence of a C-terminal clasp . Transmission EM and single particle analyses of the fibronectin fragment bound to the extracellular portion of integrin αVβ3 have shown that a bent αVβ3 conformer can stably interact with a physiological ligand . In the same EM study of soluble extracellular fragments of αLβ2 and αXβ2 integrins aforementioned, images of bent αXβ2 fragments show various degrees of bending . Conformational fluctuations or ‘breathing’ of a bent integrin could allow its globular head to interact with ligand, but this is also dependent on the size, structure and presentation of the ligand. It has been shown that a number of constitutively activated integrin αLβ2 transmembrane mutants failed to react with the reporter antibody KIM127 . Furthermore, FRET (Förster resonance energy transfer) distance measurements of FITC-LDV-containing peptide bound to integrin α4β1 and the rhodamine B-labelled membrane of U937 cells have shown that chemokine stimulates different degrees of integrin α4β1 unbending, but not its full extension . These data suggest that not all inactive integrins are bent at an obtuse angle, and a fully extended conformation may not be required for all integrins to bind ligands.
Integrins undergo conformational changes induced by intracellular and extracellular stimuli, and inside-out and outside-in signalling describe the transmission of allostery along the entire length of the integrin from its cytoplasmic tails to its ligand-binding head and vice versa [1,33–35]. A hallmark of inside-out signalling that leads to integrin activation is the separation of its α and β cytoplasmic tails and TM domains [36–39]. It has been shown that integrin αIIbβ3 is constrained in an inactive state by an electrostatic interaction between Arg1026 and Asp749 of the αIIb and β3 cytoplasmic tails respectively . These residues are highly conserved in the α and β cytoplasmic tails of integrins (Figure 1C). Mutation that disrupts the salt-bridge has also been shown to induce integrin αLβ2 activation . NMR solution structures of the αIIb and β3 peptides containing the TM domain and part of the cytoplasmic tail sequences in phospholipids reveal an OMC(outer membrane association clasp) and an IMC (inner membrane association clasp; Figure 3B) . The integrin αIIbβ3 OMC is formed by the N-terminal halves of the TM domains centred on αIIbGly1003, αIIbGly1007 and β3Gly734 with a crossing angle of ~25° [4,12]. The association affinity of the α and β TM regions that constitute the OMC could vary amongst integrins because of sequence variation . Indeed, a polar interaction between αLSer1096 and β2Thr708 in the integrin αLβ2 OMC, which is absent from the integrin αIIbβ3 OMC, plays an important role in the packing of the αLβ2 TM helices . The TM–TM association free energy profiles of the αLβ2 and αIIbβ3 integrins are also different . The IMC of integrin αIIbβ3 is formed by αIIbPhe1023 and αIIbPhe1024 that are inserted into the plasma membrane and electrostatic interaction between αIIbArg1026 and β3Asp749 . It is now evident that inside-out activation of integrins can be induced by cytoplasmic proteins such as talin that bind directly to the integrin β cytoplasmic tail, leading to IMC disruption with the separation of the α and β tails and the unpacking of the α and β TM domains [36–38,44]. It is unclear how these events destabilize the assembly of the extracellular lower portions of the integrin. It has been shown that unbending of both subunits is co-ordinated in integrin αLβ2 . One interesting feature to note is the gimbal-like I-EGF1/I-EGF2 connection that forms the β knee because different angles of I-EGF1/IEGF2 have been reported (Figure 3C) [5,9,11]. Another notable feature is the crossing of the α and β legs observed in negative-stain EM images of certain populations of extended integrins [25,26,45]. However, more studies are needed to define the detailed mechanism(s) by which unbending of the α genu and β knee leads to switchblade-like opening of an integrin.
Apart from intracellular stimuli, studies have shown that tensile forces induce conformational changes that stabilize the interaction between integrins and their ligands [46–51]. In catch bonds, force prolongs bond lifetimes . It has been shown that shear force strengthens integrin αLβ2-mediated adhesive interactions of T-cells , which explains why physiological shear forces promote leucocyte transendothelial migration . SMD (steered molecular dynamics) simulations have also shown that tensile force applied on the I-like domain of integrin αVβ3 can induce its unbending . In the blood vascular system, force is generated by fluid flow. In the absence of fluid flow, traction force is generated in migrating cells, and it is dependent on the rigidity of the extracellular substrate and the cytoskeletal network that is linked to the integrins . To simulate an integrin engaged by extracellular ligand in the presence of a lateral pulling force exerted by the actin cytoskeleton, SMD simulations were performed by applying a force vertical to the membrane on the ligand-bound head of an extended integrin αIIbβ3 with or without the application of a perpendicular pulling force on the C-terminus of the β3 ectodomain . Different extended integrin αIIbβ3 conformers with either closed or open headpiece were observed . Three different integrin conformations could also be discerned from the negative-stain EM images of soluble recombinant integrins αLβ2 and αXβ2, and these are the bent, extended with open headpiece and extended with closed headpiece conformers . The major difference between the two extended conformers is the displacement of the hybrid domain relative to the I-like domain. Hybrid domain displacement or ‘swing-out’ has functional significance because hybrid domain-specific mAbs (monoclonal antibodies) have been shown to regulate integrin ligand-binding affinity [56–59].
For integrins without an I domain, the movement of the hybrid domain induces conformational changes in the I-like domain that directly affect its binding affinity to an extrinsic ligand (Figure 4A). From the crystal structure of the integrin αIIbβ3 headpiece bound to a peptide-ligand mimetic, the major structural changes observed in the β3 I-like domain include a downward movement of the last (α7) helix and lateral shift of the α1 helix, leading to rearrangements of the metal-ion-binding sites (Figures 4B and 4C) . For integrins containing an I domain, the I-like domain does not bind an extrinsic ligand; instead it binds to an intrinsic ligand, which is an invariant glutamate residue found in a loop C-terminal to the last helix of the I domains (Figure 4D) . This invariant glutamate residue can be seen in the crystal structure of a bent integrin αXβ2 (Figure 4E) . It is not evident from this structure that the integrin β2 I-like domain MIDAS is in the vicinity of αX Glu337 possibly because the integrin αXβ2 in this study is in an inactive conformation . However, it has been shown that an engineered disulfide bond that linked the integrin αL Glu335 to the β2 Ala232, of which the latter is located in the β2 I-like domain MIDAS, induced αLβ2 activation .
How does the α invariant glutamate residue and the β I-like domain MIDAS interaction regulate I domain ligand binding? The I domain can adopt two conformations. The I domain in the closed conformation has low ligand-binding affinity, whereas the one in the open conformation has high ligand-binding affinity [7,16,17,61–63]. The last helix in the open conformer is shifted downward compared with the closed conformer (Figure 4F). The positions of the co-ordinating residues in the α I domain MIDAS are also different between the two conformers (Figure 4G). It has been shown that locking the last helix of the I domain in the closed or open conformation using engineered disulfide bonds induces low or high ligand-binding affinity respectively in both the integrin αMβ2 and its isolated I domain . Thus the downward movement of the last helix in the α I domain regulates the ligand-binding affinity of its MIDAS. Apparently, this conformational change also involves disrupting the interaction between an isoleucine residue in the last helix and a hydrophobic pocket in the I domain known as the SILEN (socket for isoleucine; Figure 4F) [7,65–69]. However, SILEN may not serve as the primary allosteric switch in all I domains because a mutation that disrupted SILEN in the integrin αM I domain, but not the αL I domain, was activating .
A general model of integrin inside-out activation is shown (Figure 5A). In this model, a bent integrin undergoes extension as a result of cytosolic activator(s) that binds and disrupts the interactions between the integrin α/β TM domains and cytoplasmic tails. The extended closed headpiece integrin could bind ligand, but a high-affinity interaction requires further conformational changes such as headpiece opening. Tensile force exerted laterally on the β tail by acto–myosin contraction may induce the conversion of the extended integrin from a closed headpiece into an open headpiece conformation. This leads to hybrid domain swing-out that activates the β I-like domain (Figure 5B). For integrins without an I domain, the activation of β I-like domain directly modulates their extrinsic ligand-binding properties. For integrins with an I domain, the activated β I-like domain binds to an invariant glutamate residue that is located C-terminal to the last helix of the α I domain. This interaction displaces the last helix of the α I domain, leading to its transition from a closed (low ligand-binding affinity) to an open (high ligand-binding affinity) conformation.
Ligand binding induces structural changes or clustering of integrins, leading to intracellular signalling cascades that modulate cellular processes (Figure 6) [34,35]. One of the outcomes of integrin outside-in signalling is the formation of highly organized focal adhesions or adhesomes that contributes to adhesion strengthening [70–72]. Although discrete focal adhesions are lacking in migrating immune cells, different cellular zones containing populations of integrin αLβ2 having different ligand-binding affinities are present in T-cells [73,74]. An equally important process in migrating cells is the disassembly of the adhesome and de-activation of integrins to allow rear detachment. In T-cells, the expression of a constitutively activated integrin αLβ2 impedes rear retraction during migration [42,75]. Disassembly of the adhesome involves a panoply of molecular switches that release the integrins from the cytoskeleton . It has been proposed that with the loss of the cytoskeletal connectivity and the absence of a lateral pulling force on the integrin β cytoplasmic tail, the tensile force between the ligand and the integrin could lead to the collapse of the integrin from an extended open to an extended closed conformation and eventually its dissociation from the ligand . More studies are required to provide the detailed mechanism of cell de-adhesion with respect to integrin de-activation.
THE β2 INTEGRINS
The leucocyte-restricted β2 integrins comprise four members: αLβ2, αMβ2, αXβ2 and αDβ2 (Table 1) . They belong to the I-domain-containing integrins and are required for a functional immune system. The importance of the β2 integrins came to light from early reports of LAD I (leucocyte adhesion deficiency type I) patients having life-threatening recurrent microbial infections with associated neutrophilia [78–80]. The aetiology of this rare disease is the lack of expression or the expression of dysfunctional β2 integrins attributed to genetic aberrations in the ITGB2 gene (encoding β2 integrin) [81–83]. Consequently, the leucocytes of LAD I patients are defective in many adhesion-dependent processes, including chemotaxis, phagocytosis and homotypic aggregation [83–86]. These are phenocopied in integrin β2−/− mice [87,88].
Integrin αLβ2 (CD11aCD18, LFA-1)
Integrin αLβ2 is expressed on all leucocytes, but expression levels vary with cellular activation and differentiation state. The six ligands of integrin αLβ2 identified to date are members of the IgSF (Ig superfamily). They are ICAM-1, ICAM-2, ICAM-3, ICAM-4, ICAM-5 and JAM-1 (junctional adhesion molecule-1; also known as JAM-A; Table 1) [89–93]. Structural studies have shown that the integrin αL I-domain MIDAS binds to the acidic side chain of a conserved glutamate residue in the first IgSF domain of ICAM-1(Glu61), ICAM-3(Glu66) and ICAM-5(Glu68) [17,94,95]. Differential binding affinities of recombinant isolated αL I domain to the ICAMs in the order of ICAM-1>ICAM-2>ICAM-3 have been reported , which are attributed to a different number of hydrophilic residues surrounding the acidic glutamate residue in these ICAMs . It has been shown that different activation states of integrin αLβ2 are required for ICAM-1 compared with ICAM-3 binding . A subsequent study with engineered integrin αLβ2 mutants has shown that an extended αLβ2 binds ICAM-1, but an open headpiece is required for stable ICAM-3 binding .
One of the most described functions of integrin αLβ2 is its role in leucocyte extravasation. Leucocyte migration across the endothelium is a multiple-step process involving concerted interactions between adhesion molecules on leucocytes and endothelial cells [97,98]. Leucocytes tether and roll on the luminal surface of the endothelium mediated by the interactions of (i) leucocyte PSGL-1 (P-selectin glycoprotein ligand1) with endothelial E- or P-selectin, (ii) leucocyte integrin α4β1 with endothelial VCAM-1 (vascular cell adhesion molecule 1) and/or (iii) leucocyte integrin α4β7 with endothelial MAdCAM-1 (mucosal addressin cell adhesion molecule-1). The engagement of neutrophil PSGL-1 by E-selectin induces Syk (spleen tyrosine kinase)-dependent inside-out integrin αLβ2 activation that slows the rolling of neutrophils on ICAM-1 [99–101]. Using conformation-sensitive reporter mAbs, it has been shown that PSGL-1 engagement activates integrin αLβ2, but its ICAM-1-binding affinity is low . Slow-rolling of the leucocytes on endothelium also allows leucocytes to encounter chemoattractants and chemokines [examples include fMLP (N-formylmethionyl-leucyl-phenylalanine), PAF (platelet-activating factor), IL-8 (interleukin-8), RANTES (regulated upon activation, normal T-cell expressed and secreted) and SDF-1α (stromal-cell-derived factor-1α)] that are immobilized on the luminal surface of the endothelium, leading to leucocyte intracellular signalling that drives high-affinity integrin αLβ2 binding to its ICAM ligands on the endothelium . Following its arrest on the endothelium, the leucocyte migrates laterally over the endothelium, a process that is dependent on αLβ2 and other integrins [104,105]. This allows scouting for sites that are permissive to either para-cellular or trans-cellular diapedesis that are also dependent on the interactions between integrin αLβ2 and endothelial ICAM-1 and JAM-1 [93,106–109]. It has also been shown that endothelial microvilli projections enriched in ICAM-1, but not ICAM-2, form cup-like structures that embrace transmigrating leucocytes . Neutrophils from integrin αL−/− mice have defective adhesive and migratory properties in response to TNFα (tumour necrosis factor α), which lends support to the importance of αLβ2 in leucocyte trans-endothelial migration .
Integrin αLβ2 is involved in immune synapse formation. When a T-cell recognizes a MHC-bound antigen on an APC (antigen-presenting cell), activated integrin αLβ2 on the T-cell binds to ICAM-1 on the APC. Imaging studies of the immune synapse between CD4+ T-cells and antigen-loaded B-cells, thymic epithelium or lipid bilayers revealed discrete contact zones formed by segregated proteins in the synapse [111–113]. The cSMAC (central supramolecular activation complex) forms the centre of the synapse with a ring of pSMAC (peripheral supramolecular activation complex), which is surrounded by the dSMAC (distal supramolecular activation complex) [111,114]. Integrin αLβ2 is found in the pSMAC, whereas comparatively smaller molecules such as TCR (T-cell receptor), MHC, CD28 and CD4 are found in the cSMAC. The sustained contact of a naïve T-cell with an APC is required for inducing T-cell proliferation . Integrin αLβ2 outside-in signalling lowers the threshold of T-cell activation because ligation of CD3, CD28 and αLβ2, but not CD3 and CD28, induces a high level of ERK1/2 (extracellular-signal-regulated kinase 1/2) phosphorylation and IL-2 production in naïve CD4+ T-cells . Thus integrin αLβ2 determines the activation status of antigen-encountered naïve T-cells. Integrin αLβ2-mediated synapse formation is also required for cytolytic killing of target cells by cytotoxic T-cells and NK (natural killer) cells [117,118]. Interestingly, integrin αLβ2 outside-in signalling can induce polarization of cytotoxic granules in NK cells , but not in cytotoxic T-cells.
Integrin αMβ2 (CD11bCD18, Mac-1, Mo-1)
Early studies on the innate immune system identified a receptor for complement protein iC3b that is identical with the Mac-1 antigen expressed on mouse and human myeloid cells [120–123]. Mac-1 is also known as CR3 (complement receptor 3) or integrin αMβ2 . Integrin αMβ2 is expressed primarily on myeloid, NK and γδ T-cells [125,126]. Unlike integrin αLβ2, αMβ2 binds an array of ligands that do not share canonical binding sequences, including ICAM-1, fibrinogen, JAM-3, denatured proteins, neutrophil inhibitory factor from hookworm, microbial lipopolysaccharide and zymosan (Table 1) [127–133]. This may be attributed to ligand-binding sites on αMβ2 other than its I domain [132,134]. Integrin αMβ2 is a well-established receptor on phagocytes . Thioglycollate-elicited neutrophils obtained from the peritoneal cavities of αM−/− mice showed impaired oxidative burst and were defective in phagocytosis . It has been reported that integrin αMβ2-mediated phagocytosis is different from that mediated by Fc receptor because the former is dependent on Rho, whereas the latter is Cdc42 (cell division cycle 42)- and Rac-dependent . Phagocytosis of iC3b-opsonized sheep RBCs (red blood cells) by integrin αMβ2-transfected COS7 cells induced RhoA activation that is dependent on Leu732-Trp747 and Thr758-Thr760 in the β2 tail . Integrin αMβ2 is involved in leucocyte extravasation across the endothelium, but integrin αLβ2 plays a major role [110,136,139]. Recently, it has been shown that the endothelial-expressed pattern recognition receptor RAGE (receptor for advanced glycation end products), which is another ligand of integrin αMβ2, is involved in αMβ2-mediated neutrophils adhesion to the endothelium under acute inflammation .
Integrin αMβ2 is also involved in the degradation and remodelling of ECM. On the cell surface of monocytes and neutrophils, integrin αMβ2 forms a reversible complex with the GPI (glycosylphosphatidylinositol)-anchored uPAR (urokinase-type plasminogen activator receptor) and its ligand uPA that promotes fibrinolysis and clearance of fibrin clots [141–143]. uPAR also changes the conformation of integrin αMβ2 and modulates its ligand-binding properties [141,144–147]. Integrin αMβ2 has also been shown to interact with the low-density lipoprotein-like receptor, FcγRIIIB (the GPI-linked form of CD16) and FcγRIIA (CD32) [148–150]. The association between FcγRIIIB and integrin αMβ2 can be disrupted by saccharides [150,151], which lends support to an earlier observation that integrin αMβ2 has lectin-binding property . However, saccharides had no effect on the interaction between integrin αMβ2 and FcγRIIA . The lateral association of integrin αMβ2 with membrane receptor is also functionally important for αMβ2 outside-in signalling. It has been shown that integrin αMβ2 signalling involving Syk is abrogated in neutrophils and macrophages of FcγRI (CD64) and DAP12 (DNAX-activating protein of molecular mass 12 kDa)-knockout mice . Syk is essential for β2 integrin-mediated signalling in haematopoietic cells (Figure 6) . FcγRI and DAP12 contain the ITAM (immunoreceptor tyrosine-based activation motif) motif that recruits Syk when it undergoes tyrosine phosphorylation by SFKs (Src family kinases). The molecular basis of integrin αMβ2 interacting with FcγRI or DAP12 remains to be determined.
Integrin αMβ2-derived intracellular signalling affects neutrophil survival . Neutrophils undergo spontaneous caspase 3-mediated apoptosis with time . Antibody cross-linking of the β2 integrins, but not the adhesion molecule L-selectin, on resting neutrophils has been shown to delay their apoptosis, an effect that was reproduced in neutrophils that were allowed to adhere to the endothelium . Integrin αMβ2 binding to laminin 8, which is a major vascular endothelial basement membrane ECM molecule, also protects neutrophils from spontaneous apoptosis . Hence integrin αMβ2 acts as a pro-survival receptor on neutrophils and the extravasation of neutrophils across αMβ2 ligand-enriched endothelium delays their spontaneous apoptosis. Contrary to these observations, it has been reported that clustering of integrin αMβ2 augmented TNFα-induced neutrophil apoptosis . It appears that integrin αMβ2 serves as a double-edged sword that can either promote or delay neutrophil apoptosis depending on its activation state and valency, and the availability of pro-apoptotic stimuli, which regulate the ERK and PKB (protein kinase B) signalling pathways (Figure 6) .
Integrin αMβ2 is involved in immune tolerance. It has been shown that immune cells obtained from the inguinal lymph nodes of ovalbulmin-fed αM−/− mice, but not wild-type mice, when re-stimulated with ovalbulmin in vitro secreted high level of pro-inflammtory cytokines IL-6 and IL-17 . Using bone marrow-derived dendritic and T-cells from β2−/− or wild-type mice in mixed lymphocyte reaction assay and antibodies specific to different members of the β2 integrins, it has been shown that activated integrin αMβ2 on dendritic cells inhibits T-cell activation . Apart from integrin αVβ5 that is a major receptor on dendritic cells for the internalization of apoptotic cells, integrin αMβ2 is also involved in this process and it renders dendritic cells tolerogenic . It has been shown that integrin αMβ2 dampens TLR (Toll-like receptor)-induced inflammatory response in macrophages by promoting degradation of TLR effectors MyD88 (myeloid differentiation factor 88) and TRIF . The anti-inflammatory function of integrin αMβ2 is not restricted to dendritic cells and macrophages because it has been shown to reduce NK cell activation in the mouse model of poly(I:C)-induced acute hepatitis . The importance of integrin αMβ2 in inducing tolerance in humans may be extrapolated from genome-wide analyses of SLE (systemic lupus erythematosus) patients that identified many susceptibility genes, including genomic variations in the ITGAM gene (encoding integrin αM). ITGAM SNP rs1143679 (G→A) substitutes Arg77 with a histidine residue in the αM ectodomain and SNP rs1143678 (C→T) substitutes Pro1146 with a serine residue in the αM tail [165,166]. It has been shown that the Arg77His substitution in αM reduced the binding of integrin αMβ2 to ICAM-1 and iC3b, diminished αMβ2-mediated phagocytosis and IL-6 production in U937 monocytic cells .
Tangential to its role in tolerance, integrin αMβ2 has been implicated in the development of multiple sclerosis, a disease in which phagocytes infiltrate the CNS (central nervous system), leading to inflammation and demyelination . It has been shown that integrin αMβ2, but not other β2 integrins, binds MBP (myelin basic protein) that is a major component of the myelin sheath . Apparently, integrin αMβ2-mediated phagocytosis of MBP leads to its presentation to auto-reactive T-cells, triggering an autoimmune response.
Integrin αXβ2 (p150,95, CD11cCD18)
Integrin αXβ2 was first identified as a heterodimer comprising a subunit of Mr 150000 and another of Mr 95000; hence the given nomenclature p150,95 . Integrin αXβ2 binds iC3b and it is also known as CR4 . Integrin αXβ2 shares many ligands with αMβ2 (Table 1). Although integrin αX−/− mice do not exhibit gross abnormalities, these mice have a lesser propensity to develop adipose tissue inflammation induced by high-fat diet . It has been shown that the accumulation of monocytes or macrophages was reduced in atherosclerotic lesions in αX−/− apoE (apolipoprotein E)−/− mice compared with apoE−/− mice . To date, αXβ2 is the only I-domain-containing integrin that has its entire globular head structure determined . Integrin αXβ2 is more resistant to activation as compared with αLβ2 and αMβ2 integrins because of interactions between its β2 PSI and αX I-EGF . It is expressed in myeloid cells, dendritic cells, NK cells and populations of activated T- and B-cells. The expression of αXβ2 serves as a marker to distinguish mature from immature dendritic cells . Integrin αXβ2 binds with high affinity to carboxy groups in ligands possibly because of the positively charged residues surrounding its I domain MIDAS . It mediates phagocytosis of iC3b-opsonized particles and is involved in the adhesion of monocytes to endothelial cells and plastic surfaces [176,177]. Integrin αXβ2 has also been implicated in the development of experimental autoimmune encephalomyelitis, the animal model of multiple sclerosis .
Integrin αDβ2 (CD11dCD18)
The last member of the β2 integrins αDβ2 is expressed at moderate level on myelomonocytic cells and high levels on foamy macrophages in aortic fatty streaks . Integrin αDβ2 is also expressed on canine splenic red pulp macrophages and it may phagocytose effete RBCs [179,180]. Interestingly, αD−/− mice do not show gross abnormalities or leucocytosis, but exhibit defective thymocyte development . Integrin αDβ2 was initially reported to bind ICAM-3, but not ICAM-1 and VCAM-1 . It was later shown that integrin αDβ2 expressing eosinophils bind VCAM-1 in static adhesion assays (Table 1) . VCAM-1 expression on endothelial cells is up-regulated under inflammatory conditions . Thus the interaction of integrin αDβ2 with VCAM-1 may facilitate the tethering of eosinophils on activated endothelium and their subsequent egress across the endothelium. The binding interface between the integrin αD I-domain and VCAM-1 may be similar to that between integrin αL I-domain and ICAM-1, involving an acidic residue from the ligand that interacts with the MIDAS of the I domain .
INSIDE-OUT ACTIVATION OF THE β2 INTEGRINS
Rap1 (Ras-proximate-1) signalling regulates inside-out activation of the β2 integrins
The Rap1 signalling pathway is involved in the activation of β2 integrins on immune cells by chemokines or TCR–CD3 ligation (Figure 6) [185–188]. Rap1 is a Ras-related small GTPase and it cycles between the inactive GDP-bound and activated GTP-bound forms . There are two isoforms of Rap1 (Rap1A and Rap1B), which are encoded by different genes, but they share more than 95% sequence homology. Because they are not distinguished in many studies, Rap1 will be used when describing these studies in the following sections.
The chemokines SDF-1 [CXCL12 (CXC chemokine ligand 12)], BLC (B -lymphocyte chemokine; CXCL13) and SLC [secondary lymphoid tissue chemokine; CCL21 (CC chemokine ligand 21)] have been shown to activate Rap1 [190–193]. Conceivably, G-protein signalling is involved in chemokine activation of Rap1 (Figure 6). It has been shown that SLC treatment of mouse lymphocytes induced Gi-protein signalling that leads to Rap1 activation . The cycling of Rap1 between the activated GTP- and inactive GDP-bound forms is regulated by GEF (guanine-nucleotide-exchange factor) and GAP (GTPase-activating protein). CalDAG-GEF I (calcium and diacylglycerol GEF I) and SPA1 (signal-induced proliferation-associated protein 1) are Rap1 GEF and GAP respectively [187,192]. Signalling events that regulate the activities of CalDAG-GEF I and SPA1 modulate the activity of Rap1, which in turn regulates the ligand-binding affinity of the β2 integrins. It has been shown in human T-cells that SDF-1 induced PLC (phospholipase C)-mediated activation of CalDAG-GEF I, leading to the affinity up-regulation of integrin αLβ2 . By contrast, overexpression of SPA1 had the opposite effect .
TCR–CD3 engagement activates integrin αLβ2 on T-cells, and the signalling pathway involves Rap1 activation because a dominant-negative form of Rap1 has been shown to inhibit this process . How does TCR–CD3 engagement induce Rap1 activation? TCR signalosome comprises many cytoplasmic molecules, including Zap70 (ζ chain-associated protein of 70 kDa) and the SLP-76 (Src homology 2 domain-containing leucocyte protein of 76 kDa) . Zap70 and SLP-76 are involved in TCR-induced PLCγ-1 activation [195,196]. Activated PLCγ-1 catalyses the formation of DAG (diacylglycerol) and triggers cytosolic Ca2+ increase. DAG and Ca2+ are required for the activation of CalDAG-GEF I, which regulates Rap1 activation . T-cells that are deficient in PLCγ-1 or treated with PLC inhibitor had diminished level of TCR-induced Rap1 activation .
What are the effectors of Rap1? GTP-bound Rap1 has been shown to bind RAPL (regulator of cell adhesion and polarization enriched in lymphoid tissues) that interacts with and is dependent on Lys1122 and Lys1124 of the integrin αL cytoplasmic tail . RAPL is expressed preferentially in lymphocytes and is a splice variant of a Ras association domain family member Nore1, which is also known as Nore1B . RAPL regulates the localization and activity of Mst1 (mammalian Ste20-like kinase 1), which has a role in translocating integrin αLβ2 to the leading edge of migrating T-cells . Whether RAPL directly modulates the ligand-binding affinity of integrin αLβ2 remains to be clarified. Rap1 also interacts with the RIAM (Rap1 GTP-interacting adaptor molecule) that has been shown to be important in the activation of the platelet integrin αIIbβ3 . SDF-1 stimulation of the human T-cell line Jurkat was found to induce translocation of the RIAM–Rap1 complex to the plasma membrane, which activates integrin αLβ2 . In TCR signalling, SLP-76 recruits the ADAP (adhesion- and degranulation-promoting adaptor protein; also known as Fyb) and SKAP55 (55 kDa Src kinase-associated phosphoprotein; also known as SKAP1) complex to the plasma membrane. It has been shown that RIAM interacts with SKAP55, and this interaction facilitates the re-localization of Rap1 to the plasma membrane following TCR engagement on primary T-cells . Rap1–RAPL complex formation is attenuated in primary T-cells of SKAP55-knockout mice . Although these observations suggest that RIAM is an effector of Rap1, it has been reported that RIAM was not required for Rap1-mediated αMβ2 ligand binding in monocytes .
How does Rap1 signalling lead to conformational changes of the integrin? Overexpression of a constitutively activated Rap1 mutant Rap1V12 in the pro-B-cell line BAF induced integrin αLβ2 conformational changes and increased its ligand-binding affinity [185,186]. However, expression of the activated form of Rap1A in mouse thymocytes induced integrin αLβ2 clustering rather than an increase in its ligand-binding affinity . The difference in observations remains to be clarified. Nevertheless, Rap1A has been shown to up-regulate the ligand-binding affinity of platelet integrin αIIbβ3 by inducing complex formation between RIAM and talin . This finding provides a possible link between Rap1 signalling and integrin conformational changes because talin is a well-established integrin activator .
In vertebrates, two talin genes, talin 1 and talin 2, have been reported [208,209]. Talin is a large cytoskeletal protein that contains an N-terminal head region and a C-terminal rod region . Its head region contains the FERM (4.1/ezrin/radixin/moesin) domain with three sub-domains (F1, F2 and F3) and its rod region is formed by bundles of α-helices . Whereas the rod region regulates the activity of talin and has docking sites for F-actins and vinculin, its head region binds integrin β cytoplasmic tail [211,212]. FRET analyses of integrin αLβ2 with its αL and β2 cytoplasmic tails fused to cyan fluorescent protein and yellow fluorescent protein respectively have shown that the talin 1 head domain induces the separation of the integrin α/β tails . In chemokine-activated leucocytes, talin head domain also triggers the extension of integrin αLβ2, leading to an increase in its ligand-binding affinity [27,41].
Despite having little information on the structure of a talin-bound integrin β2 tail complex, the mechanism by which talin activates the β2 integrins could be extrapolated from studies of talin and other integrins. The F3 sub-domain of talin contains a PTB domain (phosphotyrosine-binding domain) fold, which binds to the membrane proximal canonical NxxY/F motif of the integrin β tails (Figure 7A) [37,38]. In turn, an ionic interaction between the F3 sub-domain and the juxtamembrane region of the integrin β cytoplasmic tail is established. This involves an acidic residue in the integrin β tail (Asp759 in β1D) and a basic residue in talin F3 sub-domain (Lys327 in talin 2) (Figure 7B, top panel). Consequently, it disrupts a salt-bridge interaction formed by the same aspartate residue in the integrin β tail and an arginine residue in the integrin α tail, which is one of the constraining factors that maintain the integrin in an inactive state. Docking of talin on to the integrin β tail is further stabilized by electrostatic interactions between basic patches on the F3, F2 and F1 sub-domains of an open conformation talin head domain and lipid head groups of the inner plasma membrane (Figure 7B, middle and bottom panels) [38,213]. These step-wise interactions lead to the separation of the integrin cytoplasmic tails followed by disruption of the TM–TM packing that clasps integrin in an inactive conformation [13,31,40–42,214]. In a seminal report using EM to image purified full-length integrin αIIbβ3 embedded in lipid nanodics, a population of extended integrin αIIbβ3 was observed in the presence of talin head domain, further reinforcing the role of talin in integrin affinity modulation . The interactions between the basic patches on talin head domain with phosphatidylinositol 4,5-bisphosphate-enriched membrane microdomains also regulate integrin clustering , a process that contributes to the overall avidity of integrin-mediated cell adhesion .
Other cytoplasmic molecules that regulate inside-out activation of the β2 integrins
Distal signalling events derived from chemokine receptors and TCRs modulate the ligand-binding affinity of the β2 integrins via the Rap1 pathway (Figure 6). Talin, which directly binds the integrin β2 tail, is essential for the regulatory effect of activated Rap1 on the β2 integrins. The following section discusses cytoplasmic proteins that have been shown to bind directly to the cytoplasmic tails of the β2 integrins (Figure 7A).
Dok1 (docking protein 1) , 14-3-3ζ , filamin A (ABP-280) , kindlin3 , α-actinin , cytohesin-1  and radixin  have been reported to bind the integrin β2 tail. Dok1 belongs to the Dok family of proteins, each of which has an N-terminal PH domain (pleckstrin homology domain), a middle PTB domain followed by a tyrosine/proline-rich sequence [223,224]. Dok1 and talin compete for the same membrane proximal NxxY-binding motif in the integrin β1A, β3 and β7 tails, and phosphorylation of the tyrosine in this motif increased their binding affinities to Dok1 [225,226]. Phosphorylation of the membrane proximal NxxY motif in these tails could serve as a phosphor-switch regulating integrin activation because Dok1 does not activate integrins . Based on the sequence Arg762–Glu775 of the integrin β3 tail peptide that binds Dok1 , the corresponding integrin β2 tail sequence is Lys744–Ala757. The integrin β2 tail contains a membrane proximal NxxF motif that is not amenable to phosphorylation, but it still interacts with Dok1 (Figure 7A) . This suggests that binding of Dok1 to the integrin β2 tail is regulated by an alternative mechanism. The integrin β2 tail contains two serine residues flanking its membrane proximal NxxF motif. Phosphorylations of these residues may regulate the binding of Dok1 to the integrin β2 tail. An in vitro phosphorylation study has shown that a number of serine and threonine residues in the integrin β2 tail can be phosphorylated by different PKC (protein kinase C) members (Figure 7A) [227,228]. The effect of these phosphorylations on Dok1 binding to the integrin β2 tail remains to be determined.
14-3-3 proteins are a family of seven phospho-serine/threonine binding proteins that can self-assemble into either homo- or hetero-dimers . The integrin β2 tail peptide with Thr758 phosphorylated binds with high affinity to a positive charged binding pocket in 14-3-3ζ (Figure 7C, bottom panel) . 14-3-3 proteins have been shown to associate with integrin αLβ2 in T-cells stimulated by TCR engagement or phorbol ester treatment . Mutations T758TT/AAA in the integrin β2 tail decreased ICAM-1 binding of COS cells expressing integrin αLβ2 . These data suggest that 14-3-3 regulates the ligand-binding affinity of the β2 integrins. Dimerization of 14-3-3 allows its interaction with two integrin β2 tails simultaneously, providing a mechanism for inducing integrin clustering (Figure 7C, top panel) . Proteomic and functional analyses of 14-3-3 have shown that it modulates RhoGTPases and therefore cytoskeletal reorganization . Indeed, a cell-permeant peptide that mimicked integrin β2 tail pThr758 of the 14-3-3ζ recognition sequence activated Rac-1 and Cdc42 in T-cells . 14-3-3 also competes with talin to bind the integrin β2 tail , which appears to run counter to the importance of 14-3-3 in integrin αLβ2-mediated adhesion . It was proposed that talin and 14-3-3 bind to the β2 tail at different stages of integrin αLβ2-mediated adhesion and they have different roles . At stage one, talin induces integrin αLβ2 activation required for ligand binding followed by PKC-mediated integrin β2 tail Thr758 phosphorylation. At stage two, talin is no longer needed after integrin αLβ2 ligand binding and is displaced by 14-3-3 that mediates cytoskeletal remodelling .
Filamins are actin filaments cross-linking proteins. There are three filamins (A, B and C) in vertebrates, each having an N-terminal F-actin-binding domain followed by 24 β-pleated sheet Ig-like domains (IgFLN) . The IgFLN24 mediates dimerization of the filamins [235,236]. Filamins not only confer mechanical rigidity to cells by cross-linking the F-actin, but also they have many binding partners, which include receptors, signalling molecules and transcription factors . Filamin A binds the integrin β1 [238,239], β2 , β3  and β7 tails . Filamin A IgFN21 binds to unmodified, but not phospho-Thr758, integrin β2 tail peptide (Figure 7D) . This suggests that Thr758 could serve as a phospho-switch in regulating the interactions of filamin and 14-3-3ζ with the integrin β2 tail. Filamin is also a negative regulator of integrin activation because it competes with talin to bind the integrin tail [240,241].
Another family of molecules that shares overlapping integrin β tail-binding sequence with 14-3-3ζ and filamin A is the kindlins. Kindlins are FERM-containing cytoplasmic proteins that regulate β1, β2 and β3 integrins-mediated cell adhesion . Three paralogues kindlin1, 2 and 3 have been reported in humans [219,243,244]. The F3 subdomain of kindlins binds to the membrane distal NxxY/F motif of the integrin tails [243–245] and the F0 domain has been shown to target kindlin1 to integrin αIIbβ3-containing focal adhesion sites . Kindlin contains a PH domain that has been reported to mediate kindlin2 binding to phosphatidylinositol 3,4,5-trisphosphate . Kindlin3 is expressed in platelets, haematopoietic cells and endothelial cells [248,249]. Its importance as a co-activator of β2 and β3 integrins is underscored by the disease LAD III in which patients are susceptible to microbial infections and have bleeding disorders because of functionally defective β2 and β3 integrins respectively [250–253]. Kindlins are not direct activators of integrins . Kindlin2 has been shown to bind the scaffold protein migfilin , which strengthens integrin-mediated cell–ECM interaction (Figure 6) . Migfilin interacts with filamin, which competes with talin for binding to integrins . Thus kindlin may recruit migfilin that disrupts the interaction of filamin with integrin β tail, allowing talin to bind and activate the integrin .
The integrin β2 tail binds α-actinin . α-Actinin binds actin and it forms anti-parallel dimers that cross-link F-actin . It has been shown that the activation of PMNs (polymorphonuclear cells) by the chemoattractant fMLP triggers cleavage of talin into two fragments by the Ca2+-dependent protease calpain followed by the association of α-actinin with the β2 integrins . It was proposed that talin cleavage on cellular activation facilitates the release of β2 integrins from the cytoskeletal network, allowing lateral diffusion and clustering of the β2 integrins. Subsequent binding to α-actinin stabilizes the β2 integrin clusters by re-establishing the cytoskeletal connection. Calpain was further shown to mediate lateral diffusion and clustering of integrin αLβ2 in T-cells stimulated by TCR–CD3 cross-linking, a Ca2+ mobilizer or a phorbol ester . Shear flow experiments have shown that the arrest of chemokine-stimulated leucocytes on low (500 sites/μm2), but not high (6000 sites/μm2), density of ICAM-1 is dependent on calpain . However, another study reported no detectable effect of calpeptin treatment on chemokine-stimulated arrest of leucocytes even at an ICAM-1 coating density of 60 sites/μm2 . It is not clear what accounts for this difference in observations.
Adding to the complexity, talin has been shown to localize to mid-cell zone containing high-affinity integrin αLβ2 , whereas α-actinin-1 localizes to the leading edge containing intermediate-affinity integrin αLβ2 in migrating T-cells . Cleavage of the last three residues of the integrin β2 tail by the protease cathepsin X, a cysteine carboxypeptidase, has also been shown to promote α-actinin  and talin  binding to the β2 tail. The overall strength of cell adhesion or avidity is contributed by the affinity of individual adhesion receptors and the number of these receptors at the site of cell–cell or cell–substrate interaction . The apparent conundrum from these observations may be attributed to the different methods used in measuring cell adhesion and interpretations made based either on receptor affinity or receptor valency.
The integrin β2 tail interacts with cytohesin-1, a GEF for ADP-ribosylation factor GTPases . Cytohesin-1 contains SEC7 and PH domains that bind the membrane proximal sequence Trp723–Asp731 in the integrin β2 tail  and phosphatidylinositol 3,4,5-trisphosphate  respectively. Overexpression of cytohesin-1 has been shown to up-regulate integrin αLβ2-mediated T-cell adhesion to ICAM-1, which is dependent on PI3K (phosphoinositide 3-kinase) activity for the recruitment of cytohesin-1 to the membrane . Cytohesin-1 has been shown to directly compete with radixin to bind the integrin β2 tail . Radixin, a FERM protein, has been shown to activate integrin αMβ2 and its ERMAD (ERM-associated domain) binds to the membrane proximal β2 tail sequence Trp723–Phe738, which overlaps with the binding site of cytohesin-1 . The mechanism by which cytohesin-1 and radixin compete for binding to the integrin β2 tail remains to be determined.
OUTSIDE-IN SIGNALLING OF THE β2 INTEGRINS
Integrins transduce intracellular signals (Figure 6). SFKs are well-established components of integrin signalosome. Hck, Lyn and c-Yes, but not c-Src, Fyn and c-Fgr, have been shown to associate with the integrin β2 tail based on pull-down assays performed on platelet lysates using integrin β2 tail peptide, and these interactions appear to be dependent on the SH3 (Src homology 3) domain of the SFKs . The association of Hck with integrin αMβ2 has been shown to be dependent on the β2 tail sequence Asp750–Ser756 . This sequence is required for Hck- or Lyn-mediated integrin αMβ2 outside-in signalling . The integrin β2 tail lacks the polyproline sequence or the PxxP consensus motif found in SH3 domain ligands [269,270]. It is unclear how SFKs dock on to the integrin β tails. Inactive SFK has its SH3 domain interacting with a linker that contains an invariant proline residue between its SH2 and catalytic domains . This linker adopts a structure similar to the left-handed polyproline type II helix found in canonical SH3 targets. It was hypothesized that the SH3 domain of the SFKs binds to similar helical structures in the β tails , but NMR studies have shown that the integrin β2 and β3 tails are largely disordered except for their helical membrane proximal regions (Figures 8A and 8C) [38,272,273].
The mechanism by which SFK is activated by ligand-bound integrin has been proposed for integrin αIIbβ3 and c-Src. c-Src binds selectively to the C-terminal sequence of the integrin β3 tail . This interaction primes c-Src, and clustering of integrin αIIbβ3 upon ligand engagement increases the local concentration of c-Src, leading to its full activation via trans-autophosphorylation . Whether the same mechanism is involved in SFK activation by the β2 integrins requires further investigation. Nonetheless, SFKs have been shown to be important in β2 integrins outside-in signalling. Both human neutrophils pretreated with SFK inhibitor and neutrophils isolated from Hck/Fgr/Lyn triple knockout mice showed reduced β2 integrin-mediated adhesion, spreading and transmigration across endothelium . Intravital microscopy analysis revealed that the arrest of neutrophils from Hck/Fgr double knockout mice in inflamed muscle venules was reduced compared with wild-type cells because of defective β2 integrin-mediated cell spreading . Neutrophils isolated from Hck/Fgr double knockout mice also showed impaired respiratory burst induced by mAb cross-linking of the β2 integrins , and failure to undergo β2 integrin-mediated homotypic adhesion . Activation of NK cell cytotoxicity is dependent on the interaction of integrin αLβ2 on the NK cells and ICAM-1 on the target cells . It has been shown that engagement of integrin αLβ2 by ICAM-1 on NK cells triggers Vav1 (a Rac GEF) phosphorylation, leading to SFK-dependent cytoskeletal remodelling . Despite the observations that the SFKs are important for neutrophil functions aforementioned, a surprising finding came from a study that showed normal migratory properties of neutrophils obtained from Hck/Fgr/Lyn triple knockout mice . A possible explanation for this observation is that SFKs are required for β2 integrin-mediated neutrophil spreading, respiratory burst and degranulation, but they are less important for neutrophil migration .
In T-lymphoblasts, cross-linking of the integrin αLβ2 induces Pyk2 (proline-rich tyrosine kinase 2) phosphorylation . Pyk2 is a non-receptor tyrosine kinase that belongs to the same family as FAK (focal adhesion kinase) . Whereas FAK is widely expressed in different cell types, Pyk2 expression is restricted to haematopoietic cells, neural cells and epithelial cells . Pyk2 and FAK amplify SFK-mediated responses. For example, in osteoclasts, Pyk2 couples with c-Src to activate the Cbl proteins that affect integrin αVβ3 cell spreading and migration . Macrophages from c-Cbl-knockout mice are also defective in adhesion and migration . These observations can be attributed but not limited to the interaction between Cbl and PI3K that is dependent on SFK phosphorylation of Cbl . Apart from regulating cell spreading and migration, the same upstream signalling conduit could be involved in immune cell survival induced by the β2 integrins. In neutrophils, cross-linking of activated integrin αMβ2 activates the PI3K/PKB and the MEK [MAPK (mitogen-activated protein kinase)/ERK kinase]/ERK pathways, which delay caspase-mediated spontaneous apoptosis .
Ligand-bound β2 integrins also trigger activation of the Syk family of molecules. The family consists of two non-receptor tyrosine kinases, Syk and Zap70, which are involved in various immune responses . Each member has two N-terminal SH2 domains followed by a C-terminal kinase domain. Recombinant Syk (residues 6–370), which contains the two SH2 domains but not the kinase domain, bound with high affinity to integrin β2 tail peptide . An early study showed that Syk and SFKs formed a complex with the β2 integrins in neutrophils that spread on fibrinogen . It has been shown that during neutrophil spreading, Syk co-localizes with the β2 integrins and it initiates signalling events leading to cytoskeletal reorganization . The mechanism by which Syk is activated by the β2 integrins on neutrophils and macrophages overlaps with that of ITAM-containing immunoreceptors. Ligand engagement of the β2 integrins induces SFK-mediated tyrosine phosphorylation of the ITAM of the adaptor molecule DAP12 . Because Syk SH2 domains bind phosphorylated ITAM, Syk is recruited to the membrane, leading to its phosphorylation and activation by the SFKs . It has been reported that Vav, Pyk2 and Cbl phosphorylations were diminished in neutrophils of Syk-knockout mice compared with wild-type mice when the cells were plated on poly-RGD . Later studies using the same adhesion system also showed a lack of Vav, Pyk2, ERK and p38 MAPK phosphorylations in neutrophils obtained from DAP12 and ITAM-containing FcRγ double knockout mice compared with wild-type mice . These studies provide evidence that Syk plays an important role in early integrin signalling events in immune cells. A study on superantigen staphylococcal enterotoxin induced conjugation of Jurkat T- and EBV (Epstein–Barr virus)-B-cells has shown that the SFK Lck is critical for integrin αLβ2-dependent adhesion between these cells . Lck was later shown to be important for the activation of integrin αLβ2 in Jurkat T-cells . Recently, it has been reported that Lck and Zap70 are constitutively associated with integrin αLβ2 in T-lymphocytes, but it is not known if Lck directly associates with the integrin β2 tail . The same study also demonstrated that active Zap70 has an important role in converting integrin αLβ2 from an intermediate into a high-affinity ICAM-1-binding state necessary for T-cell spreading and migration . However, more studies are needed to provide a better understanding of the mechanisms involved.
Signalling events derived from the β2 integrins regulate cytoskeletal reorganization. In this regard, integrin αMβ2-mediated phagocytosis has been extensively studied. An early study showed that binding of integrin αMβ2 on murine macrophage cell line J774 to iC3b-opsonized RBC induced selective Rho recruitment to the phagosomes and its activation, whereas FcR-mediated phagocytosis involved Cdc42 and Rac . There are three isoforms of Rho (Rho A, B, C) and two of their effectors are ROCK1 and ROCK2 (Rho-associated kinases 1 and 2) and mDia (mammalian diaphanous homologue) . ROCK controls contraction of the acto-myosin filaments and mDia increases the rate of polymerization of actin [295,296]. Using integrin αMβ2-transfected COS7 cells that are able to phagocytose iC3b-opsonized RBCs, it has been shown that the sequences Leu732–Trp747 and Thr758–Thr760 in the integrin β2 tail are required for RhoA activation and recruitment to phagosomes respectively . However, there are reports on integrin αMβ2-mediated phagocytosis involving Cdc42 and Rac activation. It has been shown that phagocytosis of non-opsonized zymosan by integrin αMβ2-transfected CHO cells (Chinese-hamster ovary cells) involves Rac and Cdc42, but not Rho . Activated Rac and Cdc42 were also detected in neutrophils undergoing integrin αMβ2-mediated phagocytosis . It appears that integrin αMβ2 signalling activates different RhoGTPases. Furthermore, Vav1/3 (GEFs of RhoGTPases) and Rac1/2 have been shown to be important in integrin αMβ2-mediated phagocytosis [299,300]. It remains to be clarified whether there is a direct link between integrin αMβ2-induced Syk activation and Vav-mediated Rac activation. In migrating leucocytes, RhoA and ROCK are also important for rear detachment [301–303].
Cytohesin-1 and 14-3-3ζ are also involved in cytoskeletal remodelling induced by the β2 integrins. As mentioned above, 14-3-3ζ has been shown to bind with high affinity to integrin β2 tail peptide with phospho-Thr758 . The recruitment of 14-3-3ζ to the integrin β2 tail is an important event for integrin αLβ2 outside-in signalling because it leads to Rac-1 and Cdc42, but not Rho, activation that modulates actin cytoskeletal reorganization . Cytohesin-1, which plays a role in chemokine-induced integrin αLβ2 activation, has an additional function in that it lies upstream of RhoA that regulates chemotaxis of the dendritic cells .
The integrin β2 tail has been shown to bind RACK1 (receptor for activated C-kinase 1) [305,306]. RACK1 is a small WD-repeat scaffold protein that folds into a seven-blade propeller, and it binds to many cytoplasmic molecules, including c-Src and activated PKCs . RACK1 localized to leading edges of chemoattractant-induced polarized Jurkat and HL-60 cells . It has been shown that RACK1 mediates PKCβI recruitment to αIIbβ3 adhesion site in platelets . The biological function of the RACK1 interaction with β2 integrins remains to be determined.
Integrin ligand binding modulates the transcription of genes . It has been shown in T-cells that the ERK1/2 pathway is activated by ICAM-2-bound integrin αLβ2 and this is dependent on cytohesin-1 . Concomitantly, the integrin β2 tail Ser745 is phosphorylated by PKCδ that leads to the dissociation of constitutively bound JAB-1 (Jun-activating binding protein 1) from its binding sequence Lys740–Asp750 in the β2 tail. JAB-1 translocates into the nucleus, leading to the activation of AP-1 (activator protein-1) transcriptional response required for IL-2 production . It has been shown that cross-linking of integrin αMβ2 on neutrophils up-regulates IL-8 transcript expression . In monocytes, aggregation of integrin αMβ2 reduces the expression of the transcription repressor Foxp1 (forkhead box p1), leading to an increase in transcript expression of the M-CSF (macrophage colony-stimulating factor) receptor [311,312]. The early integrin αMβ2 signalling events in this case could involve IRAK-1 (IL-1-receptor-associated kinase) or PKCδ-mediated signalling pathway [268,313].
β2 INTEGRINS AND SIGNALLING SPECIFICITY
Very little is known of molecules interacting with the integrin αL, αM, αX and αD cytoplasmic tails. The integrin α tails have different lengths and sequences, and they could contribute to the signalling specificities of the β2 integrins. It has been shown that the kinetics of activation for αLβ2 and αMβ2 integrins were different between αL and αM reconstituted αL-deficient Jurkat T-cells treated with SDF-1α . Apoptosis was delayed in K562 cells expressing integrin αMβ2 that was ligand-bound, but the effect was attenuated by replacing the αM tail with the αL tail . Selective recruitment of the SFK Hck to integrin αMβ2 in CHO cell transfectants was abrogated when the αM tail was replaced with that of αL or αX . In addition, cross-linking of integrin αMβ2, but not αLβ2, on monocytes triggers SFK-dependent phosphorylation and activation of PKCδ . In fact interchanging the α tails of α2β1, α4β1 and α5β1 integrins has different effects on cellular functions [316,317]. It follows that specific signalling derived from integrin α tail is conferred by distinct cytoplasmic molecule(s) interacting with individual α tails. The scaffold protein paxillin, for example, has been shown to bind integrin α4 but not αIIb, α3, α5 or α6 tail .
NMR solution structures of the integrin αL tail reveal a tri-helical fold that has an electronegative solvent-accessible surface formed by acidic residues (Figure 8B) . This is different from the integrin αIIb tail that adopts a hairpin fold formed by electrostatic interactions between its positively charged N-terminal sequence and C-terminal residues glutamate or aspartate (Figure 8C) . The integrin αL tail is constitutively phosphorylated in leucocytes [319,320]. Approx. 40% of T-cell-surface integrin αLβ2 has been shown to have its αL tail constitutively phosphorylated at Ser1165 although the kinase involved remains to be clarified . Phosphorylation of Ser1165 in the integrin αL tail could enhance the electronegativity of the acidic surface (Figure 8B). This could serve as a docking site for cytoplasmic protein that regulates integrin αLβ2 function. Indeed, it has been reported that Rap1 activation of integrin αLβ2 requires Ser1165 phosphorylation . The integrin αL tail has been shown to interact with the first PTP (protein tyrosine phosphatase) domain of CD45 . CD45, also known as leucocyte-common antigen, is a type I transmembrane receptor-like PTP with multiple splice variants expressed in leucocytes . The first PTP domain of CD45 dephosphorylates a negative regulatory phosphotyrosine in the SFKs that are required for the development and proliferation of T- and B-cells . The physiological function of CD45 interacting with integrin αL tail remains to be determined. The integrin αL tail does not contain a tyrosine residue; hence CD45 could regulate tyrosine phosphorylation of molecules associating with or in the vicinity of the integrin αLβ2 tails. It has been shown that a small population of SFK Lck phosphorylated on Tyr394, which represents the active form of Lck, is constitutively associated with unbound integrin αLβ2 in T lymphoblasts . It is tempting to speculate that the integrin αLβ2-associated Lck could be dynamically regulated by CD45.
The integrin αM tail has a membrane helical segment followed by a C-terminal loop that folds back as determined by NMR studies . The integrin αM tail contains two phosphorylation sites Tyr1137 and Ser1142. Treatment of human monocytes with PMA or fMLP induced the phosphorylation of integrin αM tail Ser1142 . Integrin αM Ser1142 has also been shown to be important for αMβ2-mediated leucocyte extravasation because a low number of the αL-deficient T-cell line Jβ2.7 transfected with αM S1142A as compared with wild-type αM was detected in the lungs and spleens of mice injected with these cells . In the same study, the integrin αMS1142Aβ2 mutant bound poorly to ICAM-1, and it reacted minimally with the mAb KIM127 . Phosphorylation of Ser1142 does not induce marked changes to the conformation of the integrin αM tail based on NMR studies , but it could serve as a docking site for cytoplasmic protein that modulates the function of integrin αMβ2. To date, the structures and functions of the integrin αX and αD tails have not been well characterized.
The cytoplasmic tails of the β2 integrins are binding sites for many cytosolic proteins. These associations modulate the ligand-binding affinities and/or the outside-in signalling properties of the β2 integrins. Many of these cytosolic proteins have overlapping binding sites on the integrin β2 tail; hence there will be competition for occupancy. The kinetics of association of these cytosolic proteins with the integrin β2 tail and how these binding events are regulated will require further investigation. The α tails of the β2 integrins have different lengths and sequences, but little is known about their functions. This will be another interesting area for future studies.
This work was supported by Nanyang Technological University [grant number RG34/08]; the Singapore Ministry of Education [grant numbers MOE2008-T2-1-044, MOE2010-T2-2-014]; and the Singapore A*STAR Biomedical Research Council (BMRC) [grant number 10/1/22/19/654].
I apologize for not being able to cite all the related publications because of space limitations.
Abbreviations: AP-1, activator protein-1; APC, antigen-presenting cell; apoE, apolipoprotein E; CalDAG-GEF I, calcium and diacylglycerol GEF I; Cdc42, cell division cycle 42; CHO cell, Chinese-hamster cell; cSMAC, central supramolecular activation complex; CR, complement receptor; CXCL, CXC chemokine ligand; DAG, diacylglycerol; DAP12, DNAX-activating protein of molecular mass 12 kDa; Dok1, Docking protein 1; ECM, extracellular matrix; ERK, extracellular-signal-regulated kinase; ERM, ezrin/radixin/moesin; EM, electron microscopy; FAK, focal adhesion kinase; FERM, 4.1/ezrin/radixin/moesin; fMLP, N-formylmethionyl-leucyl-phenylalanine; FRET, Förster resonance energy transfer; GAP, GTPase-activating protein; GEF, guanine-nucleotide-exchange factor; GPI, glycosylphosphatidylinositol; I domain, inserted domain; I-EGF, integrin-epidermal growth factor; ICAM, intercellular adhesion molecule; Ig, immunoglobulin; IgSF, Ig superfamily; IL, interleukin; IMC, inner membrane association clasp; ITAM, immunoreceptor tyrosine-based activation motif; JAB-1, Jun-activating binding protein 1; JAM, junctional adhesion molecule; LAD I, leucocyte adhesion deficiency type I; LFA-1/2, leucocyte function-associated antigen 1/2; LIMBS, ligand-associated metal-binding site; mAb, monoclonal antibody; MBP, myelin basic protein; mDia, mammalian diaphanous homologue; MIDAS, metal-ion-dependent adhesion site; ADMIDAS, adjacent to MIDAS; Mst1, mammalian Ste20-like kinase 1; NK, natural killer; OMC, outer membrane association clasp; PH domain, pleckstrin homology domain; PI3K, phosphoinositide 3-kinase; PKB, protein kinase B; PKC, protein kinase C; PLC, phospholipase C; PSGL-1, P-selectin glycoprotein ligand 1; pSMAC, peripheral supramolecular activation complex; PSI, plexin-semaphorinintegrin; PTB domain, phosphotyrosine-binding domain; PTP, protein tyrosine phosphatase; Pyk2, proline-rich tyrosine kinase 2; RACK, receptor for activated C-kinase; RAGE, receptor for advanced glycation end products; Rap1, Ras-proximate-1; RAPL, regulator of cell adhesion and polarization enriched in lymphoid tissue; RBC, red blood cell; RIAM, Rap1 GTP-interacting adaptor molecule; ROCK, Rho-associated kinase; SDF, stromal-cell-derived factor; SFK, Src family kinase; SH3 domain, Src homology 3 domain; SILEN, socket for isoleucine; SKAP55, 55 kDa Src kinase-associated phosphoprotein; SLC, secondary lymphoid tissue chemokine; SLP-76, Src homology 2 domain-containing leucocyte protein of 76 kDa; SMD, steered molecular dynamics; SPA1, signal-induced proliferation-associated protein 1; Syk, spleen tyrosine kinase; SyMBS, synergistic metal ion-binding site; TCR, T-cell receptor; TM, transmembrane; TNFα, tumour necrosis factor α; uPAR, urokinase-type plasminogen activator receptor; VCAM, vascular cell adhesion molecule; Zap70, ζ chain-associated protein of 70 kDa
- © The Authors Journal compilation © 2012 Biochemical Society