The Leukocyte Integrins*

Integrins on Leukocytes Leukocytes are marrow-derived cells of diverse form and function that circulate in the blood in a quiescent state of low adhesiveness before migrating into tissues to defend against invading microbes, participate in immune functions and wound repair, or become fixed extracellular residents. Some, such as T-lymphocytes, recirculate and traverse blood, organ, and lymphatic compartments during long cycles of immune surveillance. Others, notably polymorphonuclear leukocytes (PMNs, neutrophils), are rapid response cells specialized for acute spatially targeted defensive actions that can be mounted in minutes. Leukocytes are also effectors of pathologic inflammation when their accumulation and actions are disregulated. Integrins on their surfaces, together with other plasma membrane adhesion molecules, are required for interactions of leukocytes with endothelial cells and other cell types and with matrix structures (1). The functional state, density, and topography of integrins on leukocytes are regulated by lipid, cytokine, and chemokine signaling molecules and by “cross-talk” from other surface adhesion molecules (1–6). Each class of leukocytes displays a particular pattern of integrins that can change in a signaland time-dependent fashion. For example, resting human T lymphocytes (T cells) express b1, b2, and b7 integrins, but this varies with the subclass and is altered by immune stimulation (5). Freshly isolated human monocytes express b1 and b2 integrins, but their culture and/or differentiation into macrophages changes the pattern and induces avb3 (5, 7). Human PMNs, once thought to express only b2 integrins, display b1 and b3 heterodimers and use them in motility and migration (8– 12). A common feature, however, is that each leukocyte subtype expresses one or more members of the b2 integrin family. Further, the b2 heterodimers are restricted to cells of the leukocyte lineage. The remainder of this minireview will focus on the structure and function of the b2 or “leukocyte” integrins, which were among the first adhesion molecules to be studied at the molecular level. The most recently identified member of the subfamily, aDb2, 3 is still being characterized (13–16).

␣ D ␤ 2 also recognize proteins of other classes and (in the case of ␣ M ␤ 2 and ␣ X ␤ 2 ) polysaccharides. The unifying feature of protein ligands may be the presence of acidic residues (aspartate or glutamate) positioned in flexible loops that allow them to coordinate Mg 2ϩ or Mn 2ϩ and form a bridge to the I and/or I-like domains on the integrin heterodimer (17). The RGD motif, which is a critical feature in many integrin ligands, is not a required feature in ligands for ␤ 2 heterodimers.
Ligand Recognition The crystal structures of the I domains of ␣ M and ␣ L have been solved (17,23), providing rosetta stones for general understanding of integrin structure-function relationships. The crystal structure of the I domain of ␣ M reveals a coordination locus for Mg 2ϩ and Mn 2ϩ that is proposed as a general "metal ion-dependent adhesion site" (MIDAS) consisting of the sequence DXSXS together with downstream non-contiguous Asp and Thr residues (17,24) (Fig. 1). Blocking and mutagenesis of the metal coordinating sites in the MIDAS motifs of ␣ subunits alter or abolish ligand binding, as does mutagenesis of key residues in the flanking regions. 2 A DXSXS motif is also found in the I-like domain of the ␤ 2 subunit. Its mutation eliminates ligand recognition (17,18,25).
Changes in tertiary conformation of the ␣ I domains may be a general mechanism that dictates active and inactive states of the ␤ 2 integrins (1). The crystal I domain of ␣ M assumes two conformations, an open or "active" and a closed or "inactive" structure, that differentially recognize ligands although conformational alterations were not seen when crystals of the I domain of ␣ L were grown under various ionic conditions (23,26,27). Quaternary structural alterations likely also occur. One model proposes that the ␤ subunit I-like domain folds over the ␣ subunit propeller in the low affinity unactivated state, blocking the ␣ I domain and its central MIDAS motif (Fig. 1). Quaternary changes triggered by inside-out signals (see below) shift the ␤ chain I-like domain, exposing the ␣ subunit I domain and other ligand recognition sites in the ␣ propeller in concert with simultaneous conversion of the ␣ subunit I domain to an active conformation via tertiary changes (28). Thus, dynamic structural alterations in ␤ 2 heterodimers are involved in ligand recognition, as with other classes of integrins (see the first minireview in this series by Plow et al. (84)). Modulation of avidity is also involved (see below).

Inside-out Signaling
Activation of leukocytes by agonists that bind to diverse classes of receptors triggers ligand recognition by ␤ 2 integrins. This process is termed "inside-out signaling," integrin "activation," and "functional up-regulation" (1). Rapid, regulated modulation of ligand recognition is critical for leukocytes, because they must circulate in a non-adhesive state before targeting and arrest at specific sites. The molecular mechanisms that mediate inside-out signaling of integrins have been elusive (see the second minireview in this series by Ginsburg and co-workers (85)). More than one pathway may trigger inside-out signaling of an individual ␤ 2 integrin heterodimer (29,30). Transfected and mutated cell systems are now popular for analysis of integrin signaling (30 -32). Such models suggest that inside-out signaling of ␤ 2 integrins occurs via mechanisms dependent on the small GTPase Rho (33,34) and that there is differential intracellular regulation of the activity of ␤ 2 versus ␤ 1 , ␤ 3 , and ␤ 7 integrins in the same transfected cell type (35)(36)(37). There are cell-specific aspects of integrin regulation, however, that may operate in model cell systems but not in primary leukocytes and vice versa (38).
Analysis of inside-out signaling of ␤ 2 integrins is most detailed for ␣ L ␤ 2 . Changes in both affinity and avidity are involved, depending in part on the cellular system and the stimulus chosen to alter its adhesive function. Low and high affinity states of ␣ L ␤ 2 occur on lymphocytes (39) and other leukocytes (1). Manipulation of the "extracellular face" of ␣ L ␤ 2 using divalent cations and functionperturbing antibodies induces rapid dynamic changes in recognition of ICAM-1 and other ligands (1,40). Recent experiments with soluble recombinant peptides based on the I domain of ␣ L indicate that this region is required for ICAM-1 recognition when ␣ L ␤ 2 shifts to the high affinity state on cultured T cells treated with Mg 2ϩ (41). In contrast, treatment of T cells with phorbol esters or agents that increase intracellular Ca 2ϩ causes clustering of ␣ L ␤ 2 and increased avidity of binding without detectable change in affinity (41)(42)(43)(44). Lateral motion of ␣ L ␤ 2 in the plasma membranes of Epstein-Barr virus-transformed lymphoblasts is enhanced by phorbol esters and by cytochalasins, suggesting that conversion from the non-adhesive to the adhesive state involves signals that alter cytoskeletal interactions and allow the integrin to segregate into clusters or adhesive patches (45). Clustering of membrane "rafts" containing ␣ L ␤ 2 occurs in murine thymocytes and activated T cells and confers adhesion to ICAM-1 (46). There is additional evidence for avidity modulation of ␣ L ␤ 2 in surrogate cell models (47,48). Lateral segregation of ␣ L ␤ 2 in the plane of contact occurs in T cells adherent to cellular targets or purified proteins in model membranes (49,50). Differential regulation of affinity versus avidity of ␣ L ␤ 2 is triggered by specific signals and may involve a two-step mechanism in which increased affinity, which can be induced by ICAM-1 and potentially by other binding partners, and altered avidity occur in sequence (41,48).
Truncations, mutations, and deletions of the GFFKR motif of ␣ L (Fig. 1) in a Jurkat T cell line and in K562 transfectants cause constitutive ICAM-1 recognition; there is also additional evidence that the membrane-proximal region of the ␣ cytoplasmic tail exerts a general inhibitory effect on ligand recognition that is independent of the ␤ chain (1,21,30,51). The ␤ 2 cytoplasmic tail is, however, critical for modulation of ␣ L ␤ 2 adhesiveness (1,38,47,51). The ␤ 2 chain associates with a variety of cytoskeletal and regulatory proteins, including ␣-actinin, talin, filamin, vinculin, and Rack 1. Certain of these interactions have been demonstrated to be altered when leukocytes of different classes are activated by physiologically relevant stimuli and to regulate ␣ L ␤ 2 function (1, 52). Insideout signaling of ␣ L ␤ 2 on B lymphoblastoid cells is regulated by Rho acting downstream of protein kinase C (PKC) (53), consistent with studies in transfected cells mentioned above. The GFFKR sequence (green) in the cytoplasmic tail of the ␣ subunit is involved in heterodimer assembly and regulation of ligand recognition. The heterodimer is illustrated in the "closed" or inactive state that undergoes tertiary and quaternary changes in response to inside-out signals. See "Structure and Distribution," "Ligand Recognition," and "Inside-out Signaling" for details.
A member of the Sec7 family, cytohesin-1, modulates ␣ L ␤ 2 in lymphocytic and monocytic cell lines (35, 54 -56). Cytohesin-1 was identified using the intracellular domain of ␤ 2 in a yeast two-hybrid screen; it has a C-terminal pleckstrin homology (PH) domain and an N-terminal motif similar to the yeast Sec7 domain, conferring membrane association and guanine nucleotide exchange factor activity (35,54). Cytohesin-1 weakly coimmunoprecipitated with ␣ L ␤ 2 in lysates of a Jurkat T cell line, but not with ␣ 4 ␤ 1 (35). Overexpression of cytohesin-1 or its Sec7 motif in Jurkat cells induced ␣ L ␤ 2 -dependent adhesion to immobilized ICAM-1 without increasing ␣ L ␤ 2 heterodimers on the surface, whereas overexpression of the PH domain inhibited adhesion stimulated by T cell receptor engagement (35). Expression of a constitutively active chimeric phosphatidylinositol 3-kinase (PI 3-kinase) induced adhesion of Jurkat cells to ICAM-1 and increased membrane association of cytohesin-1, both of which were inhibited by overexpression of the PH domain but not by a control PH motif (54). This suggests that PI 3-kinase regulates recruitment of cytohesin-1 to the plasma membrane utilizing the PH domain where it mediates inside-out signaling of ␣ L ␤ 2 , potentially via the Sec7 domain (54,56). Additional evidence in other systems indicates that signals delivered via PI 3-kinases activate integrins, potentially converging with signals from PKC. 2,5 It is unknown if inside-out signaling of ␣ L ␤ 2 on primary leukocytes is mediated by PI 3-kinase-dependent cytohesin-1 translocation. A recently identified factor that is structurally similar to cytohesin-1, GRP1, also regulates ␣ L ␤ 2 in leukocyte cell lines (57).
Inside-out signaling and ligand recognition of ␣ M ␤ 2 follows paradigms outlined for ␣ L ␤ 2 (1, 26 -29). Two functional states of the crystallized ␣ M I domain indicate a basis for changes in conformation (17,26,27). Avidity modulation involving clustering of ␣ M ␤ 2 or cell spreading also occurs (1,29,30,58). L-plastin, an actin-organizing protein, regulates ␣ M ␤ 2 on human monocytes and neutrophils based on experiments with cell-permeant peptides (59). In human neutrophils, inside-out signaling of ␣ M ␤ 2 was dependent on PI 3-kinase and actin cytoskeletal reorganization when triggered by engagement of Fc␥ receptors but not when the leukocytes were activated through the receptors for inflammatory peptides (29). An inhibitor of PI 3-kinase also failed to block ␣ M ␤ 2 -dependent adhesive interactions of neutrophils in earlier experiments (60). These findings are consistent with differential regulation of ␣ M ␤ 2 and ␣ L ␤ 2 via the ␣ cytoplasmic tails (30) and also suggest alternative mechanisms for signaling of cytohesin-1 interaction with the cytoplasmic tail of ␤ 2 if it is central to activation of ␣ M ␤ 2 , as proposed for ␣ L ␤ 2 (see above). Pak1, a serine/threonine kinase, may be an intermediary in a pathway leading to inside-out signaling of ␣ M ␤ 2 in neutrophils (29).

Outside-in Signaling
Engagement of integrins delivers outside-in signals that trigger intracellular transduction cascades, in addition to mediating adhesion. These outside-in signals can also be integrated with signals delivered through receptors for signaling molecules to yield coordinated functional responses. Cross-linking of leukocyte integrins with antibodies against ␤ 2 or specific ␣ subunits, engagement of ␤ 2 heterodimers with specific ligands, or coengagement of ␤ 2 integrins together with other surface structures or receptors delivers outside-in signals that lead to diverse cellular responses (1). 2 Signaling to gene regulatory pathways involves transcriptional events and mRNA stabilization and is differentially affected by experimental conditions when ␤ 1 versus ␤ 2 integrins are engaged (61)(62)(63)(64). Outside-in signaling by specific ␤ 2 integrins may vary in leukocyte subtypes (65) and is impaired in leukocytes from subjects with leukocyte adhesion deficiency type I (LAD I) (see below) (66,67).
Several intracellular signaling pathways are triggered by engagement of ␤ 2 integrins (1). Activation of focal adhesion kinase (FAK) is central to many paradigms of outside-in signaling by integrins, and FAK binds peptides based on the sequence of the ␤ 2 cytoplasmic domain in addition to ␤ 1 and ␤ 3 cytoplasmic peptides (68). Nevertheless, FAK does not appear to be critical for outside-in signaling by ␤ 2 integrins, and it is inconsistently detected in human monocytes and neutrophils (63, 69) even though phosphorylation on tyrosine occurs in these cells in response to ␤ 2 integrin engagement (1,67). In contrast, the human THP-1 monocytic leukemia cell does express FAK (70), illustrating the potential differ-ences in integrin signaling in primary leukocytes versus transformed cell lines.
Integrin-linked kinase, a serine/threonine kinase that associates with ␤ 1 cytoplasmic domains, and calreticulin, which associates with the GFFKR sequence of ␣ subunits, mediate signaling functions of integrins (71,72), but it is not known if they regulate ␤ 2 heterodimers.

Genetically Altered Animal Models
Mice with partial (73) and complete (74) deficiency of ␤ 2 integrins and specific deletions of ␣ L ␤ 2 (75), ␣ M ␤ 2 (76,77), and ␣ D ␤ 2 6 have been produced. Animals deficient in all ␤ 2 integrins display phenotypic features of humans with LAD I (see below) including neutrophilia and a defect in accumulation of PMNs in inflamed skin. PMN emigration into lung alveoli in response to a bacterial challenge and into the peritoneum in response to a sterile irritant are preserved (74). Blocking antibodies against ␤ 2 integrins inhibit neutrophil accumulation under the same or similar conditions (1,74), however, suggesting alternative adhesion mechanisms that compensate for deficiency of ␤ 2 integrins in mice. In contrast, absence of PMNs from inflamed or injured extravascular sites is the usual outcome in most humans with ␤ 2 integrin deficiency (1).
Individual knockouts of ␣ L ␤ 2 and ␣ M ␤ 2 yielded surprises, including the fact that PMNs efficiently use ␣ L ␤ 2 for emigration in mice null for ␣ M ␤ 2 (77). This was unexpected, because the diversity of ligands recognized by ␣ M ␤ 2 relative to ␣ L ␤ 2 (1) suggested that it would be required for traversing complex tissue compartments. Integrins ␣ X ␤ 2 and/or ␣ D ␤ 2 may substitute for ␣ M ␤ 2 , together with ␤ 1 integrins (1, 8 -16). In inflammatory disease models neutrophil effector functions are impaired in mice lacking ␣ M ␤ 2 (77, 78), potentially because of absent outside-in signaling or deficient signal integration (see above). One of these models indicated interaction between ␣ M ␤ 2 and Fc receptors (78), which was predicted by earlier in vitro observations (1). The same strain of mice was deficient in tissue mast cells, indicating that ␣ M ␤ 2 is important in targeting and/or development of this extravascular leukocyte subtype (79).

LAD I and LAD I Variant Syndromes
Humans with LAD I have absent or a greatly reduced display of all ␤ 2 integrin heterodimers on the surfaces of their leukocytes, absent or dramatically reduced accumulation of PMNs and monocytes at extravascular sites, recurrent life-threatening bacterial infections, and impaired tissue remodeling and wound healing. Cells from these subjects have defective adhesive and signaling functions when studied in vitro (1,2,66,67). The search for the molecular basis of LAD I led to identification and initial characterization of ␤ 2 integrins (1). The syndrome results from a variety of mutations that prevent normal heterodimer formation and surface display. Recently, variant LAD I syndromes have been identified (80,81). 7 Each of these subjects had clinical features consistent with LAD I but, in contrast to the phenotype outlined above, had normal or only moderately reduced levels of ␤ 2 integrins (40 -60% of control) on the surfaces of circulating leukocytes at the time of diagnosis. In one of the subjects there were two new mutations of the ␤ chain; one was located in the MIDAS motif of the I-like domain (81). Phenotypic and sequence characterization of leukocytes from the other two subjects indicates a defect in inside-out signaling. These rare LAD variants, like variations in structure and signaling of integrin ␣ IIb ␤ 3 in the syndrome of Glanzmann thrombasthenia (82), may yield unique insights that are relevant both to ␤ 2 heterodimers on leukocytes and to the biology of integrins in general.