Cytoskeletal Interactions with the Leukocyte Integrin β2 Cytoplasmic Tail

Abstract

Circulating leukocytes are nonadherent but bind tightly to endothelial cells following activation. The increased avidity of leukocyte integrins for endothelial ligands following activation is regulated, in part, by interaction of the β2 subunit cytoplasmic tail with the actin cytoskeleton. We propose a mechanism to explain how tethering of the actin cytoskeleton to leukocyte integrins is regulated. In resting leukocytes, β2 integrins are constitutively linked to the actin cytoskeleton via the protein talin. Activation of cells induces proteolysis of talin and dissociation from the β2 tail. This phase is transient, however, and is followed by reattachment of actin filaments to integrins that is mediated by the protein α-actinin. The association of α-actinin with integrins may stabilize the cytoskeleton and promote firm adhesion to and migration across the endothelium. Glutathione S-transferase-β2 tail fusion protein/mutagenesis experiments suggest that the affinity of α-actinin binding to the β2 tail is regulated by a change in the conformation of the tail that unmasks a cryptic α-actinin binding domain. Positive and inhibitory domains within the β2 tail regulate α-actinin binding: a single 11-amino acid region (residues 736–746) is necessary and sufficient for α-actinin binding, and a regulatory domain between residues 748–762 inhibits constitutive association of the β2 tail with α-actinin.

Integrins are heterodimeric, transmembrane adhesion molecules composed of noncovalently associated α and β subunits that physically link extracellular ligands to the cytoskeleton (1). The cytoplasmic domain of integrin β subunits links these receptors to the actin cytoskeleton (2, 3). However, actin filaments cannot bind directly to integrins. Instead, integrins are linked indirectly to actin filaments via several actin-binding proteins, including α-actinin, talin, and filamin. (4–7). The importance of integrin-cytoskeletal linkage is demonstrated by the observation that deletion of the β subunit tail prevents association of integrins with the cytoskeleton and disrupts normal integrin-ligand interactions (reviewed in Ref. 8).

Several integrins, including LFA-1¹ and Mac-1, share a common β2 subunit and are present exclusively on leukocytes. These leukocyte integrins mediate cell adhesion to endothelial cell ligands such as intracellular adhesion molecules (reviewed in Ref. 9). Unactivated leukocytes in the circulation are non-adherent, and LFA-1 and Mac-1, both of which are expressed constitutively on resting neutrophils, show very little, if any, binding to their physiologic ligands (10). Activation of neutrophils with cytokines (leukotriene B₄ and tumor necrosis factor-α), chemoattractants (FMLP), or phorbol 12-myristate 13-acetate (PMA) results in increased binding of neutrophils to the endothelium (11, 12). This increase results, in part, from insertion of integrins onto the cell surface, but both LFA-1 and Mac-1 also undergo a rapid change in ligand avidity as a result of activation, which involves conformational changes in their extracellular domain and a clustering of the receptors in the membrane. The association of cytoskeletal proteins with integrin cytoplasmic tails appears to play an important role in regulating intracellular signaling events that affect the conformation of integrin extracellular domains and promote integrin clustering. Several recent studies suggest that LFA-1 on lymphocytes is constrained by the actin cytoskeleton in resting cells (13–15) and that Ca²⁺-mediated activation of the enzyme calpain releases LFA-1 from this constraint to allow integrin clustering upon activation (14). Kucik et al. (15) also demonstrated that the mobility of LFA-1 in lymphocytes is increased by activation with phorbol ester. In this report, we examine the association of two cytoskeletal proteins that directly interact with β1 and β2 integrin cytoplasmic domains to learn more about how these associations may be regulated. Based on our results, we propose a model in which β2 integrins on unactivated PMNs are tethered to the actin cytoskeleton via the protein talin. Following activation, proteolysis of talin causes release of integrins from the actin cytoskeleton. This would facilitate increased integrin mobility in the membrane, increased ligand binding, and clustering of ligand occupied receptors. Following clustering, integrins reengage the cytoskeleton by binding to the actin cross-linking protein α-actinin as we have previously described (6). Results of the glutathione S-transferase (GST) fusion protein affinity binding experiments presented here suggest that the association of α-actinin with β2 integrins is regulated by conformational changes in the cytoplasmic domain that unmask the α-actinin binding site between residues 736 and 746 in the membrane proximal half of the tail that is cryptic resting cells. Our model helps explain how the cytoskeleton plays an apparent dual role in the regulation of integrin-mediated leukocyte adhesion: inhibition of adhesion in resting cells but stimulation of adhesion and motility in activated cells.

EXPERIMENTAL PROCEDURES

PMN Isolation, Activation, and Extraction

Human PMNs were isolated from fresh citrate phosphate dextrose-anticoagulated blood obtained from volunteer donors using 6% dextran to sediment erythrocytes, followed by separation from lymphocytes by centrifugation in Histopaque 1077 (Sigma). The PMN pellet, which also contained monocytes, was washed in Dulbecco’s phosphate buffered saline (PBS) containing 0.2% glucose and resuspended in Dulbecco’s PBS with glucose at a concentration of 10⁷ cells/ml. PMNs were activated with either 10 nM formyl-methionyl-leucyl-phenylalanine (FMLP) peptide (Sigma) or 50 nM PMA (Sigma). In some experiments, PMNs were preincubated in the calpain inhibitor calpeptin (Calbiochem-Novabiochem), which was prepared in Me₂SO at 100 mM and diluted to a final concentration of 100 μM (36 μg/ml) in Dulbecco’s PBS. Control cells were incubated in Dulbecco’s PBS containing 1 μl of Me₂SO/ml. PMN extracts for immunoblot analysis with anti-talin antibody 8d4 (Sigma) and for affinity binding studies were prepared by treating 10⁷ cells with 1 ml of lysis buffer consisting of Tris-buffered saline (TBS) (50 mM Tris, pH 7.4, 150 mM NaCl) containing 1% Triton X-100, 1% sodium deoxycholate, 1 mM EDTA, and the protease inhibitors aprotinin (25 μg/ml), leupeptin (10 μg/ml), and phenylmethylsulfonyl fluoride (1 mM). Following extraction for 10 min on ice, insoluble material was removed by centrifugation at 14,000 × g, and the supernatant was saved. Protein concentration was determined using BCA reagent (Pierce); approximately 20 μg was loaded per lane for immunoblot analysis, and 500 μg was loaded onto GST fusion protein affinity columns. For co-immunoprecipitation experiments, 5 × 10⁶ cells were lysed in 1 ml of TBS/CHAPS buffer containing 1% CHAPS, 10 mM Tris-HCl, pH 7.6, 150 mM NaCl, 1 mM CaCl₂, 1 mM MgCl₂, 0.01% NaN₃, and 20 mM DNase I and protease inhibitors aprotinin (25 μg/ml), leupeptin (10 μg/ml), and phenylmethylsulfonyl fluoride (1 mM) as described previously (6).

Cell Culture

Rat embryo fibroblasts (REF-52) were grown in Dulbecco’s modified Eagle’s medium containing 10% fetal calf serum at 37 °C in 5% CO²/95% air in a humidified incubator. For co-immunoprecipitation experiments, cells were plated on vitronectin-coated (10 μg/ml) or fibronectin-coated (50 μg/ml) plastic in Dulbecco’s modified Eagle’s medium without serum for 6 h. Suspension cells were obtained by trypsinization of substrate adherent cells, which were then washed in Dulbecco’s modified Eagle’s medium containing 10% fetal calf serum and maintained in suspension at 37 °C for 2 h to allow the cells time to recover.

Co-immunoprecipitation Assay

Extracts in TBS/CHAPS buffer were clarified by centrifugation for 10 min at 15,000 × g, and 1 ml aliquots of supernatant were transferred to a fresh 1.5-ml microcentrifuge tube. This extract was incubated with 50 μl of a 10% w/v solution of protein A-positive Staphylococcus aureus cells in 10% CHAPS lysis buffer for 30 min at 4 °C to remove cellular proteins binding to protein A. The S. aureus cells were sedimented, and an excess of either anti-β2 mAb (KIM127 or MHM23, Dako, Inc.) or anti-β1 integrin antibody (1938, Chemicon) was added to the supernatant and then incubated for 1 h at 4 °C. Then, 100 μl of a 10% suspension of protein A-Sepharose conjugated to rabbit anti-mouse Ig was added to the mixture, which was incubated for an additional 1 h at 4 °C. The complexes containing antibody-bound integrin and co-precipitating proteins were pelleted and washed five times with TBS/CHAPS buffer containing 0.1% CHAPS. Then, 15 μl of gel sample buffer was added to the co-immunoprecipitated complexes, which were boiled to release bound proteins. Samples were electrophoresed on 10% polyacrylamide gels and transferred to nitrocellulose. Talin or α-actinin co-precipitating with integrins was visualized using monoclonal antibodies 8d4 or BM75.1 (obtained from Sigma), horseradish peroxidase-conjugated secondary antibodies, and chemiluminescence detection (Pierce).

GST Fusion Protein Affinity Columns

Plasmids harboring the cDNAs encoding the human β2 or β1 integrin cytoplasmic tails were generously provided by Drs. William Roeder and A. F. Horwitz, respectively. The region encompassing the cytoplasmic tails of these integrins was amplified by PCR and subcloned into the GST fusion vector, pGEX-2TK (Amersham Pharmacia Biotech). GST-β2 cytoplasmic domain mutations were generated by oligonucleotide-based site-directed mutagenesis using the method of Kunkel (16). The resultant mutations were verified by nucleotide sequencing of the region of interest. The fusion protein constructs were introduced into the Escherichia coli strain BL21, and expression of the GST-β2 or GST-β1 proteins was induced with isopropyl-1-thio-β-D -galactopyranoside. GST fusion proteins were purified using glutathione immobilized on Sepharose beads (Amersham Pharmacia Biotech) and then covalently coupled to glutathione-Sepharose beads using dimethylpimelimidate cross-linker for 1 h at room temperature, prior to use in affinity chromatography-like assays as described by the manufacturer. 500 μg of PMN lysate or 5 μg of purified α-actinin or talin was applied to the column and allowed to bind for 30 min (α-actinin and talin were purified from chicken gizzard as described previously (5). Unbound proteins were removed by extensive washing with cell lysis buffer, and specifically bound proteins were eluted with 1 M KCl containing 10% glycerol, 0.5 mM EDTA, and 10 mM dithiothreitol. Following trichloroacetic acid precipitation and acetone washes, samples were subjected to SDS-PAGE and transferred to nitrocellulose membranes. Immunoblot analysis were performed by blocking for 1 h in buffer containing 2.5% bovine serum albumin, 0.2% gelatin, and 0.05% Tween 20 in TBS and incubation with anti-α-actinin antibody (BM75.1) or anti-talin antibody (8d4) for 1 h. After washing two times in wash buffer (0.2% gelatin, 0.05% Tween 20 in TBS), blots were incubated for 1 h with horseradish peroxidase-labeled secondary antibody, washed again, and exposed to ECL color development reagents (Amersham Pharmacia Biotech).

RESULTS

Talin Constitutively Associates with the β2 Subunit in Un-activated PMNs

We previously reported that the protein α-actinin became associated with the cytoplasmic domain of the β2 subunit in PMNs following activation but that α-actinin did not interact with β2 integrins in unactivated cells (6). Therefore, we searched for cytoskeletal proteins that constitutively bound to β2 integrins and might function to link integrins to actin filaments prior to activation. For these experiments, we first used an integrin co-immunoprecipitation assay. We found that the protein talin, previously reported to bind to the cytoplasmic domain of β1 (4) and β3 integrins (17), co-immunoprecipitated with β2 integrins from unactivated PMNs (Fig. 1A ,−FMLP). Significantly, only the higher molecular mass (approximately 225 kDa) form of talin was seen in the β2 immunoprecipitates from unactivated (−FMLP) cells; the 190-kDa proteolytic fragment of talin, clearly present in PMN extracts, was not present in the integrin co-immunoprecipitates. Somewhat surprisingly, talin failed to co-immunoprecipitate with β2 integrins after activation (Fig. 1A ,+FMLP). Thus, the 225-kDa talin, corresponding to the intact protein, appeared to constitutively bind to the cytoplasmic tail. The bands in the first two lanes that migrate below the 190-kDa talin band and above the band corresponding to the Ig heavy chain appear to be an additional proteolytic fragment of talin. In addition to co-immunoprecipitation, we have also confirmed that purified talin (purified from chicken gizzard) can bind directly to the β2 cytoplasmic tail. In this experiment, the purified 225-kDa talin (but not the 190-kDa proteolytic fragment) bound to a GST fusion protein corresponding to the full-length β2 integrin cytoplasmic tail but did not bind to GST alone (Fig. 1B).

Fig. 1. Interaction between the cytoplasmic domain of β2 and talin.

A, co-immunoprecipitation of talin with β2 integrins. β2 integrins were immunoprecipitated from extracts of unactivated PMNs (−FMLP) or PMNs activated with 10 nM FMLP for 10 min, and these were immunoblotted with anti-talin. Intact (225 kDa) talin co-immunoprecipitates with β2 from unactivated cells (−FMLP) but not from activated cells (+FMLP). B, binding of purified talin to a GST-β2 cytoplasmic domain fusion protein. Purified talin was applied to GST fusion protein column expressing the entire β2 cytoplasmic domain (residues 724–769) or to a control column expressing GST only. Both intact, 225-kDa, talin and the 190-kDa proteolytic fragment of talin are present in the purified talin preparation. The intact, 225-kDa, talin band binds to the GST-β2 cytoplasmic domain fusion protein but does not bind to the GST alone. C, proteolysis of talin in PMNs following activation. Extracts of unactivated PMNs or PMNs activated with FMLP or PMA were immunoblotted with anti-talin. Talin from unactivated cells (−) was primarily (>90%) in the intact, 225-kDa form, with less than 10% in the 190-kDa form. Following activation with either FMLP or PMA, approximately 90% of the talin was in the 190-kDa form. D, calpeptin inhibits talin proteolysis in FMLP-activated PMNs. Pretreatment of PMNs with calpeptin (100 μM for 30 min) inhibited talin proteolysis.

Next, we further investigated the significance of the observation that only intact, 225-kDa talin co-immunoprecipitated with β2 integrins from resting PMNs. Evidence that talin undergoes proteolysis in PMNs following activation was provided by immunoblot analysis. 20 μg of total protein from freshly isolated, unactivated PMNs or 40 μg of total protein from PMNs activated with either the chemotactic peptide FMLP (10 nM) or with the phorbol ester PMA (100 nM) was separated by SDS-PAGE, transferred to nitrocellulose, and subjected to immunoblot analysis using a monoclonal antibody against talin (the greater amount of protein from FMLP- and PMA-activated cells was necessary to allow us to detect the small amount of the intact, 225-kDa talin band that was present following activation). Fig. 1C shows that in resting PMNs, most of the talin is in the high molecular mass (approximately 225 kDa) form, corresponding to the intact protein. The ratios of the 225- to the 190-kDa bands were determined by densitometry prior to and following activation of PMNs with FMLP or PMA. In unactivated cells, the ratio of 225-kDa talin to 190-kDa proteolytic fragment was 0.9; following activation, the ratio of 225-kDa talin to 190-kDa proteolytic fragment was 0.05. Fig. 1D shows that pretreatment of PMNs for 30 min with the membrane permeant calpain inhibitor calpeptin (100 μM) significantly reduced talin proteolysis following activation so that only 20% of the talin corresponded to the 190-kDa proteolytic fragment (ratio of 225-kDa:190-kDa, 0.8). Together, these results suggest that only full-length 225-kDa talin associates with the β2 cytoplasmic tail as determined by co-immunoprecipitation and in vitro binding. Following activation, the majority of the talin is cleaved by a calpain-like protease to a 190-kDa form that does not associate with the β2 tail.

Association of α-Actinin with β1 Integrins Is Constitutive

Consistent with our previous report (6), we found that association of α-actinin with β2 integrins in PMNs requires activation of the cells. Fig. 2 shows that following activation, α-actinin rapidly associates with the cytoplasmic tail and can be detected in immunoprecipitates of β2 integrin. In contrast, α-actinin in fibroblasts appears to be constitutively associated with the β1 integrins. As shown in Fig. 2, α-actinin can be detected by Western blotting in co-immunoprecipitates with anti-β1 antibody from fibroblasts that have either been grown on fibronectin or vitronectin or maintained in suspension for 2 h.

Fig. 2. Association of α-actinin with β2 in PMNs requires activation of cells, whereas in fibroblasts, α-actinin associates constitutively with β1 integrins.

An antibody against β2 was used to immunoprecipitate from extracts of PMNs under nondenaturing conditions that were unactivated (0 min) or were activated with FMLP (10 nM) for 5 or 10 min prior to extraction (left panel). Immunoprecipitates were immunoblotted with anti-α-actinin. Anti-β1 integrin was used to immunoprecipitate from extracts of rat embryo fibroblasts (REF-52) under nondenaturing conditions grown on either fibronectin or vitronectin or maintained in suspension for 2 h prior to extraction (right panel). α-Actinin co-precipitated with β1 integrins from adherent or suspension fibroblasts but co-precipitated with β2 integrins from PMNs only after activation.

α-Actinin Binding to β1 and β2 Cytoplasmic Domain GST Fusion Proteins

We postulated that sequence differences between the β1 and β2 cytoplasmic domains may be responsible for regulating this differential association of α-actinin. To investigate this question, we constructed GST fusion proteins coupled to Sepharose that correspond to the entire human β1 or β2 cytoplasmic domains sequences (β1, residues 752–798; β2, residues 724–769) and used them as affinity chromatography matrices to assess the ability of α-actinin from cell extracts to bind to each sequence. Fig. 3A and Table I show that α-actinin bound to the full-length β1 tail (β1 w.t.) but interacted only weakly with the full-length β2 tail (β2 w.t.) fusion protein. α-Actinin binding to the full-length β1 tail was 9.7-fold higher than binding to the full-length β2 tail (Table I). Similar results were obtained when purified α-actinin, instead of cell extract, was applied to the fusion protein affinity columns (not shown). Together with the requirement for activation of PMNs to induce α-actinin-β2 binding, these results suggest that additional events, such as conformational changes in the β2 tail, are required to allow enhanced binding of α-actinin to the cytoplasmic domain.

Fig. 3. Analysis of α-actinin binding to GST-integrin cytoplasmic domain fusion proteins.

A, comparison of α-actinin binding to β1 and β2 integrin cytoplasmic domain-GST fusion proteins. Cell extracts were applied to GST fusion protein affinity columns corresponding to integrin cytoplasmic tail constructs (illustrated in Table I), and eluates were immunoblotted with anti-α-actinin. α-Actinin bound relatively weakly to the full-length β2 cytoplasmic tail (β2 w.t.) but with relatively higher affinity to the full-length β1 cytoplasmic domain (β1 w.t.). B, comparison of α-actinin binding to distinct regions of the β₂ cytoplasmic domain. α-Actinin bound to the NH₂-terminal half (Δ746–769) but not to COOH-terminal half (Δ727–746) of the β2 tail. Deletion of residues 727–736 (Δ727–736) bound α-actinin weakly, at a level comparable to the full-length tail (β2-w.t.). C, confirmation of a binding site for α-actinin between residues 736 and 746. α-Actinin bound to a fusion protein containing only residues 736–746 (736–746). Control lanes confirm that α-actinin does not bind to the deletion mutant lacking residues 727–746 from the NH₂-terminal half of the tail (Δ727–746); α-actinin binds weakly to the wild-type β2 tail but binds with relatively higher affinity to the truncation mutant lacking the COOH-terminal half of the tail.

Table I. Summary of GST-fusion protein-α-actinin binding studies.

Deleted segments are shown by dashed line. Single base pair substitutions are shown underlined.

Construct designation	Amino acid sequence	Fold increase in α-actinin relative to β2 w.t.a
β2 w.t. tail	KALIHLSDLREYRRFEKEKLKSQWNNDNPLFAKSATTTVMNPKFAES	1.0
Δ727–746	KAL-------------------QWNNDNPLFAKSATTTVMNPKFAES	Binding not detectable
Δ727–736	KAL---------RRFEKEKLKSQWNNDNPLFAKSATTTVMNPKFAES	2.2
Δ746–769	KALIHLSDLREYRRFEKEKLKS-------------------------	12.2
736–746	-----------RRFEKEKLKS-------------------------	9.8
Δ758–769	KALIHLSDLREYRRFEKEKLKSQWNNDNPLFAKSA------------	6.0
β1 w.t. tail	KLLMIIHDRREFAKFEKEKMNAKWDTGENPIYKSAVTTVVNPKYEGK	9.7
N748A	KALIHLSDLREYRRFEKEKLKSQWANDNPLFAKSATTTVMNPKFAES	19.7
N749A	KALIHLSDLREYRRFEKEKLKSQWNADNPLFAKSATTTVMNPKFAES	20.3
D750A	KALIHLSDLREYRRFEKEKLKSQWNNANPLFAKSATTTVMNPKFAES	18.2
P752A	KALIHLSDLREYRRFEKEKLKSQWNNDNALFAKSATTTVMNPKFAES	16.4
T758E	KALIHLSDLREYRRFEKEKLKSQWNNDNPLFAKSAETTVMNPKFAES	15.7
T758A	KALIHLSDLREYRRFEKEKLKSQWNNDNPLFAKSAATTVMNPKFAES	14.5
T759A	KALIHLSDLREYRRFEKEKLKSQWNNDNPLFAKSATATVMNPKFAES	17.8
T760A	KALIHLSDLREYRRFEKEKLKSQWNNDNPLFAKSATTAVMNPKFAES	18.9
M762V	KALIHLSDLREYRRFEKEKLKSQWNNDNPLFAKSATTTVVNPKFAES	13.5
A767E	KALIHLSDLREYRRFEKEKLKSQWNNDNPLFAKSATTTVMNPKFEES	17.4
K755A	KALIHLSDLREYRRFEKEKLKSQWNNDNPLFAASATTTVMNPKFAES	1.9
F766A	KALIHLSDLREYRRFEKEKLKSQWNNDNPLFAKSATTTVMNPKAAES	0.7
E768A	KALIHLSDLREYRRFEKEKLKSQWNNDNPLFAKSATTTVMNPKFAAS	0.8

^aβ2 w.t., full-length β2 cytoplasmic domain.

Consistent with this hypothesis, truncation to remove the COOH-terminal half of the β2 cytoplasmic tail at residue 746 (Δ746–769) resulted in enhanced binding of α-actinin compared with the wild-type β2 tail (Fig. 3B and Table I). Results from five separate experiments showed that α-actinin binding to Δ746–769 averaged 12.2-fold higher compared with α-actinin binding to wild-type β2. The membrane proximal (NH₂-terminal) half of the cytoplasmic domain of β2 expressed in the Δ746–769 truncation mutant contains several conserved residues, including the sequence RFEKEKEK that was previously implicated in α-actinin-β2 binding (6). Consistent with the predicted importance of residues 727–746 for binding, Fig. 3B and Table I also show that α-actinin could not be detected binding to a deletion mutant in which residues 727–746 (Δ727–746) corresponding to the membrane proximal half of the tail were deleted. A third deletion mutant lacking residues 727–736 (Δ727–736) bound α-actinin weakly, with an increase of only 2.2-fold above wild-type β2 binding levels (Fig. 3B and Table I). Fig. 3C and Table I show that the 11-amino acid region within residues 736–746 is sufficient for binding α-actinin at a level 9.8-fold above wild-type β2 and was comparable both to α-actinin binding to GST-β1 and to the β2 truncation mutant representing the membrane proximal sequence (Δ746–769). These results indicate that a binding site for α-actinin resides within residues 736–746 and that binding of α-actinin to this region may be masked or sterically impaired by an inhibitory domain between residues 746–769.

Mutational Analysis of the COOH-terminal Half of the β2 Cytoplasmic Tail

To further investigate whether the putative α-actinin inhibitory domain in the membrane distal half of the β2 cytoplasmic tail might regulate integrin association with α-actinin, a series of point mutations were created in this region. The results of experiments assessing the ability of α-actinin to bind to these fusion proteins are illustrated in Fig. 4 and summarized in Table I.

Fig. 4. Analysis of α-actinin binding to GST-β2 cytoplasmic domain fusion proteins with point mutations in the COOH-terminal half of the tail.

Point mutations at residues Asn⁷⁴⁸, Asn⁷⁴⁹, Asp⁷⁵⁰, Pro⁷⁵², Thr⁷⁵⁸, Thr⁷⁵⁹, Thr⁷⁶⁰, Met⁷⁶² (not shown), and Ala⁷⁶⁷ each confer on the β2 tail the ability to bind α-actinin at levels comparable to that seen when the entire COOH-terminal half of the tail is deleted. Other point mutations of residues at the very end of the COOH terminus, Phe⁷⁶⁶ and Glu⁷⁶⁸, or the upstream lysine residue, Lys⁷⁵⁵, did not alter the wild-type (weak α-actinin binding) phenotype.

Point mutations at nine residues in the COOH terminus of the tail, N748A, N749A, D750A, P752A, T758(A/E), T759A, T760A, A767E (Fig. 4 and Table I), and M762V (Table I) resulted in binding of α-actinin at levels comparable to or higher than α-actinin binding to Δ746–769. Point mutations at other, apparently less critical residues, including Lys⁷⁵⁵ (K755A), Phe⁷⁶⁶ (F766A), and Glu⁷⁶⁸ (E768A), shown in Fig. 4 and Table I, however, did not alter the wild-type binding phenotype; i.e. weak binding of α-actinin. Thus, altering the conformation of the distal half of the β2 cytoplasmic domain by introduction of point mutations at any of several apparently critical residues relieves the inhibitory effect on α-actinin binding to residues 736–746 in the membrane proximal half of the tail. Although there does not appear to be a critical role for the extreme COOH-terminal end of the tail, between residues 763 and 769, in regulating cytoskeletal interactions, the significance of the A767E point mutation that induced α-actinin binding is not clear.

DISCUSSION

The cytoplasmic tails of integrin β subunits play an important role in mediating interactions with the cytoskeleton and are important in controlling cell-cell and cell-substrate interactions. Integrins that are clustered in the membrane co-localize and physically associate with cytoskeletal proteins in vitro and in vivo (3, 18). In this study, we have examined the role of the β2 integrin cytoplasmic tail in mediating interactions with two cytoskeletal proteins, talin and α-actinin, using human neutrophils and GST-β2 cytoplasmic domain fusion protein affinity columns.

Based on our results, we propose a mechanism (illustrated in Fig. 5) through which β2 integrins in unactivated neutrophils are tethered to the actin cytoskeleton through a linkage mediated by talin. The activation-induced proteolysis of talin, which can be inhibited by the calpain protease inhibitor calpeptin, may regulate binding of talin to β2 integrins. Such a role for talin has recently been proposed in lymphocytes to explain the regulation of LFA-1-mediated adhesion by calpain (13). Lub et al. (14) showed that the cytoskeleton has different effects on LFA-1-mediated adhesion depending on whether or not LFA-1 was initially clustered or was diffuse on the membrane. They suggested that release of cytoskeletal constraints on LFA-1 in unactivated lymphocytes was necessary to allow lateral movement of LFA-1, and clustering on the membrane was necessary to stabilize cell adhesion. Following activation, β2 integrins may be released from cytoskeletal constraint by breaking the link between talin and integrin upon proteolysis of talin. Once β2 integrins have engaged ligand and clustered in the membrane, reengagement of the actin cytoskeleton, mediated by binding of α-actinin to the membrane proximal half of the tail, may stabilize cytoskeletal-integrin interactions necessary for firm adhesion and transendothelial migration.

Fig. 5. A model to explain the role of talin and α-actinin in linking actin filaments to β2 integrins in resting and activated leukocytes.

A, in unactivated cells, actin filaments are linked to integrins via their association with talin, which binds to the cytoplasmic domain of β2. Thus, in unactivated cells, mobility of β2 integrins is constrained by the actin cytoskeleton. Activation of cells results in proteolysis of talin, which causes talin to no longer bind to the β2 cytoplasmic domain, resulting in free mobility of the integrins in the membrane. This phase is transient, so that β2 integrins are rapidly reengaged by the actin cytoskeleton as a result of α-actinin binding to a previously cryptic binding site in the membrane proximal half of the cytoplasmic domain. B, the α-actinin binding domain in the β2 cytoplasmic tail (shown highlighted in green) is located between residues 736 and 746. The α-actinin inhibitory domain in the COOH-terminal half of the β2 tail (highlighted in red) prevents α-actinin binding in unactivated cells. Point mutations at several residues (residues shown in yellow) result in a mutant fusion protein that binds α-actinin with relatively higher affinity, suggesting that the conformation of the COOH-terminal half of the tail is critical for efficient inhibition of α-actinin binding to the membrane proximal half of the tail.

Distinct domains within the β2 cytoplasmic tail that regulate binding to the protein α-actinin have also been identified. GST-β2 cytoplasmic tail fusion protein binding experiments identified an 11-amino acid region (RRFEKEKLKSQ) between residues 736 and 746 in the membrane proximal half of the tail that are necessary and sufficient for α-actinin binding. A second distinct region in the membrane distal half of the tail was shown to have a unique role in regulating α-actinin binding. An inhibitory region resides between residues 748–769 and prevents association of α-actinin to the membrane proximal binding region. The ability of point mutations at any of several residues in this inhibitory domain (residues Asn⁷⁴⁸, Asn⁷⁴⁹, Asp⁷⁵⁰, Pro⁷⁵², Thr⁷⁵⁸, Thr⁷⁵⁹, Thr⁷⁶⁰, Met⁷⁶², and Ala⁷⁶⁷) to apparently unmask the previously cryptic α-actinin binding site in the β2 tail suggest that the conformation of the tail is important in regulating α-actinin binding.

Analysis of the membrane distal region (residues 747–769) revealed the presence of several motifs within this domain that could have significant influence on the structural conformations of the β2 cytoplasmic tail. One of these is the NPXF/Y motif (NPLF in β2), which is structurally important, is conserved among the integrins β subunits (19), and has been shown to be important for postligand recognition activities, such as cytoskeletal interactions, firm adhesion, growth control, and shape change (20, 21). Melanoma cells transfected with β3 integrin in vitro and in vivo failed to migrate on vitronectin-coated surfaces when the NPXY domain was altered by mutation, whereas cells transfected with wild-type β3 attached, spread, and migrated normally (22). This deletion did not appear to affect ligand recognition because the cells were still able to bind soluble vitronectin. Hibbs et al. (23, 24) have also shown that mutations within the NPXY motif of β2 did not prevent binding of LFA-1-transfected COS cells to intracellular adhesion molecule-1-coated surface, whereas they prevented downstream phosphorylation of a serine residue. This suggests that this motif could have a potential regulatory role both in influencing protein conformation and in providing a site for kinase binding.

NPXY also imparts a tight β turn to the cytoplasmic domain. Our model predicts a structural configuration of the β2 tail in unactivated neutrophils in which the α-actinin binding site is masked as a result of this turn. This model is supported by the findings of Reszka et al. (25), who reported that all four of the amino acids in the region upstream of the NPXY domain in β1 integrins that were essential for cytoskeletal binding were in fact on one face of an α-helix. The exact mechanism of conformational change in vivo, however, is not clear. The membrane distal half of the β2 tail has several potential phosphorylation sites, including a serine (Ser⁷⁵⁶) and the threonine triplet (Thr⁷⁵⁸-Thr⁷⁵⁹-Thr⁷⁶⁰). Hibbs et al. (23, 24) have shown that whereas in vivo phosphorylation of the β2 tail was predominantly at Ser⁷⁵⁶, alteration of this residue by mutagenesis has little effect on ligand binding. Mutation of the threonines to alanine, however, led to a complete loss of ligand binding in their study, suggesting that the region between Ser⁷⁵⁶ and Met⁷⁶² could be important for ligand binding and association of cytoskeletal proteins. In our studies, point mutations T758E, which may mimic phosphorylation by introducing a negative charge, and T758A both enhanced α-actinin binding compared with wild-type, suggesting that phosphorylation of threonine 758 is not necessary for inducing binding of α-actinin but may be important for ligand binding.

Comparison of the regions previously identified to be important for extracellular ligand binding versus the cytoskeletal protein binding regions described here (summarized in Table II) suggest that distinct domains in the β2 cytoplasmic tail regulate these two events. Hibbs et al. (23, 24) showed that the region between 726 and 755 in the β2 tail was dispensable for intracellular adhesion molecule-1 binding, whereas we have mapped α-actinin binding to this region (727–746). Furthermore, they and others have demonstrated the need for presence of intact threonine triplets for ligand binding, whereas we have shown that single alanine substitution at any of those sites induces α-actinin binding. Additionally, the carboxyl-terminal five amino acid residues, KFAES, have been implicated in ligand binding, in particular the residue Phe⁷⁶⁶. In contrast, our studies showed that F766A mutation maintained the β2 tail in a wild-type phenotype, i.e. it was unable to bind α-actinin.

Table II. Comparison of residues important for cytoskeletal association versus adhesion to ICAM-1.

Based on our findings and those of Hibbs et al. (24), it is clear that distinct regions within the β2 tail are involved in regulating its intra-cellular association with α-actinin and extracellular association with ICAM-1.

β2 sequence	Role in regulating the association with α-actinin	Role in regulating ICAM-1 binding (24)
736RRFEKEKLKSQ746	Essential	Dispensable
755KSATTTVMNPKFAES769	Inhibits association	Essential
756SATTTVM762	Inhibits association	Essential
Asn748, Asn749, Asp750, Pro752, Thr758, Thr759, Thr760, Met762	These point mutations relieve the inhibitory phenotype	No data
Lys765, Glu768, Ser769	No role	Essential
Phe766	Mutation relieves the inhibitory phenotype	Essential

Taken together, these studies suggest involvement of some as yet unknown cytoplasmic factors, which could directly regulate both the above events. Recent evidence has shown that novel proteins, such as cytohesin (26), specifically interact with β2 integrin cytoplasmic tail and regulate its function. Similar proteins that are specific for β1 and β3 have also been described (27, 28) that do not appear to interact with other β integrins despite the high degree of homology in their cytoplasmic tails. The interaction of these proteins with specific integrins could lead to alterations in their properties that might explain differential regulation of the various family members. Lub et al. (14) have demonstrated presence of distinct signaling mechanisms for activation of β1 and β2, acting via their cytoplasmic tail, and have suggested the need for involvement of lymphocyte specific factors in β2 integrin activation. Studies aimed at identifying novel proteins that interact with the β2 tail in the context of our findings here could reveal other important mediators of LFA-1/Mac-1 activation.