A single amino acid in the cytoplasmic domain of the beta 2 integrin lymphocyte function-associated antigen-1 regulates avidity-dependent inside-out signaling.

The leukocyte-specific beta(2) integrin lymphocyte function-associated antigen-1 (LFA-1) (alpha(L)/beta(2)) mediates activation-dependent adhesion to intercellular adhesion molecule (ICAM)-1. In leukocytes, LFA-1 requires activation by intracellular messengers to bind ICAM-1. We observed malfunctioning of LFA-1 activation in leukemic T cells and K562-transfected cells. This defective inside-out integrin activation is only restricted to beta(2) integrins, since beta(1) integrins expressed in K562 readily respond to activation signals, such as phorbol 12-myristate 13-acetate. To unravel these differences in inside-out signaling between beta(1) and beta(2) integrins, we searched for amino acids in the beta(2) cytoplasmic domain that are critical in the activation of LFA-1. We provide evidence that substitution of a single amino acid (L732R) in the beta(2) cytoplasmic DLRE motif, creating the DRRE motif, is sufficient to completely restore PMA responsiveness of LFA-1 expressed in K562. In addition, an intact TTT motif in the C-terminal domain is necessary for the acquired PMA responsiveness. We observed that restoration of the PMA response altered neither LFA-1 affinity nor the phosphorylation status of LFA-1. In contrast, strong differences were observed in the capacity of LFA-1 to form clusters, which indicates that inside-out activation of LFA-1 strongly depends on cytoskeletal induced receptor reorganization that was induced by activation of the Ca(2+)-dependent protease calpain.

LFA-1 has to be activated via outside-in or inside-out signals to efficiently bind ICAM-1. Outside-in signaling has been identified by LFA-1 activating antibodies (6) or immobilized ligands, resulting in cell spreading, rise in intracellular Ca 2ϩ and pH, phosphorylation of proteins, and costimulatory signals (7,8). Inside-out signals are initiated upon triggering of specific cell surface molecules, generating intracellular signals that induce a high affinity and/or avidity state of LFA-1 (9). Both conformational changes (affinity) in the presence of Mg 2ϩ and altered surface distribution of LFA-1 into clusters (avidity) upon Ca 2ϩ binding result in strong ligand binding (8,10,11).
Although the ␣ L and ␤ 2 cytoplasmic domains of LFA-1 are relatively short and do not contain any intrinsic kinase activity, they are important for affinity and avidity regulation. Previous studies have shown that LFA-1 adhesiveness is controlled by the cytoplasmic domain of the ␤ 2 subunit, since truncation of the cytoplasmic ␤ 2 domain, but not the ␣ L domain, eliminates LFA-1 binding to ICAM-1 (12). Complete deletion of the ␤ 2 cytoplasmic domain results in clustering and spontaneous activation of LFA-1. This constitutively active LFA-1 deletion mutant strongly binds to ICAM-1. The phorbol ester PMA that activates PKC cannot further increase the adhesion to ICAM-1 of this constitutive active LFA-1, in contrast to wild type LFA-1 (13). It has been proposed that the ␣ L cytoplasmic domain of LFA-1 is involved in post-ligand binding events, since deletion of the cytoplasmic ␣ L domain does not affect binding to ICAM-1 (12). Also, cytoskeleton restraints play a crucial role in regulating LFA-1 avidity, since clustering of LFA-1 is induced on resting PBLs after treatment with cytochalasin D (14). Interestingly, this is not the case for ␤ 1 integrins, indicating that ␤ 2 and ␤ 1 integrins differ in their ability to cluster into specialized lipidic membrane microdomains, also termed rafts (15). Replacement of the ␤ 2 cytoplasmic domain for that of ␤ 1 (␣ L /␤ 2 / ␤ 1 ), creating a chimeric LFA-1 molecule containing a ␤ 1 cytoplasmic domain, provided us with additional evidence. The chimeric LFA-1 (␣ L /␤ 2 /␤ 1 ) showed a clustered cell surface distribution when expressed in the erythroleukemic cell line K562. Furthermore, PMA activation of the chimeric LFA-1 molecule increased the adhesion to ICAM-1. This was in contrast to wild type LFA-1 that, when expressed in K562, is not clustered and is defective for PMA-induced activation (13).
Several regions within the ␤ 2 cytoplasmic domain are thought to be important in regulating LFA-1. Alanine substitutions of conserved threonines (TTT) in the ␤ 2 cytoplasmic domain reduce ICAM-1 binding, and a serine residue is phosphorylated upon PKC activation by PMA (16). There are different consensus sequences known in the ␤ 2 cytoplasmic do-* The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
The observation that, despite their homology, ␤ 2 and ␤ 1 integrins are differently regulated by inside-out signals, prompted us to identify residues within the ␤ 2 cytoplasmic domain that are involved in the PMA-induced LFA-1-mediated ligand binding. To this end, we substituted ␤ 2 amino acids for those of the ␤ 1 cytoplasmic domain that are critical for PMAinduced adhesion in K562 cells. We observed that a single ␤ 2 /␤ 1 amino acid substitution is sufficient to completely restore the PMA responsiveness by enhancing LFA-1 avidity but not the affinity. In addition, we observed that activation of LFA-1 by PMA is dependent on cytoskeletal rearrangements that seem to be mediated by the Ca 2ϩ -dependent protease calpain.
Soluble ICAM-1 Fc Binding-Transfectants were resuspended in TSA (5 ϫ 10 6 cells/ml). 50,000 cells were preincubated with or without LFA-1-blocking mAb (20 g/ml) for 10 min at room temperature in a 96-well V-shaped bottom plate. Different concentrations of purified soluble ICAM-1Fc were added together with medium or the LFA-1activating mAb KIM185 (15 g/ml), and the suspension was incubated for 30 min at 37°C. The cells were washed with TSA and incubated for 30 min at room temperature with FITC-conjugated goat anti-human Fc-specific antibody (Jackson Immunoresearch Laboratories, West Grove, PA). The cells were washed with TSA and resuspended in 100 l of TSA. The percentage of positive cells was measured by flow cytometry using the FACScan. Values are depicted as the percentage of positive cells (i.e. cell adhesion percentage minus cell adhesion percentage in the presence of an integrin-blocking mAb). Alternatively, the concentration of soluble ICAM-1Fc that gives half-maximal adhesion (ED 50 ) is depicted.
Confocal Microscopy-Cells were fixed with 0.5% paraformaldehyde. Fixed cells were stained with TS2/4 mAb (10 g/ml) for 30 min at 37°C, followed by incubation with FITC-labeled goat F(abЈ) 2 anti-mouse IgG mAb (Zymed Laboratories Inc., San Francisco, CA) for 30 min at room temperature. Cells were attached to poly-L-lysine-coated glass slides, after which cell surface distribution of integrins was determined by Confocal laser-scanning microscopy (CLSM) at 488 nm with a krypton/ argon laser (Bio-Rad; model 1000). The CLSM settings were as follows: lens, ϫ 60; gain, 1300; pinhole, 1.5 m; and magnification, ϫ 2.0. The same instrument settings of the CLSM were used throughout the distinct experiments.

Substitution of the ␤ 2 by the ␤ 1 Cytoplasmic Domain Restores PMA Responsiveness in K562 Cells Transfected with LFA-1-
LFA-1 is a cell adhesion receptor that is exclusively expressed on leukocytes. Activation of LFA-1 is required for efficient binding to its ligand ICAM-1. The addition of the phorbol ester PMA has been shown to activate LFA-1 on leukocytes (10). When wild type LFA-1 is transfected into the erythroleukemic cell line K562 (K562-␣ L /␤ 2 ) (Figs. 1 and 2), the ␤ 2 -activating antibody KIM185 can activate LFA-1 and induce LFA-1-mediated adhesion to ICAM-1; however, an inside-out activator of LFA-1, such as PMA, cannot activate LFA-1 ( Fig. 3A) (13). This lack of PMA responsiveness of LFA-1 is not caused by a general defect of intracellular signal molecules, since other endogenous expressed integrins like VLA-5 can be activated by PMA to bind its ligand fibronectin. Our finding that expression of chimeric LFA-1 containing the ␤ 1 cytoplasmic domain (␣ L /␤ 2 /␤ 1 ) can completely restore the PMA responsiveness of LFA-1 in K562 cells prompted us to search for single amino acids that differ between the ␤ 1 and ␤ 2 cytoplasmic domain.
Substitution of Leucine for the ␤ 1 Arginine in the DLRE Motif of the ␤ 2 Cytoplasmic Domain Restores PMA Responsiveness of LFA-1 in K562 Cells-To analyze in detail the regions in the ␤ 1 cytoplasmic domain that are responsible for the PMA responsiveness of ␤ 2 /␤ 1 chimeric LFA-1 molecule transfected in K562, several point mutations were created in the ␤ 2 cytoplasmic domain. Amino acids of the ␤ 2 cytoplasmic domain were substituted for the residues present in the ␤ 1 cytoplasmic domain ( Fig. 1). K562 cells were transfected with ␣ L -chain together with the ␤ 2 -chain containing cytoplasmic domains of ␤ 1 or ␤ 2 as described under "Experimental Procedures." The expression levels of K562 cells transfected with the LFA-1 chimeras and point mutants were determined by fluorescence-activated cell sorting analysis using anti-CD11a and anti-CD18 antibodies (Fig. 2). The mutations in the ␤ 2 cytoplasmic domain did not affect the ␣/␤ heterodimerization based on the expression of ␣/␤ heterodimer dependent MHM23 epitope, and immunoprecipitation of LFA-1 from all mutants confirmed that mutant LFA-1 was expressed as ␣/␤ heterodimers (data not shown). The capacity of the LFA-1 mutants to adhere to ICAM-1 was determined using an ICAM-1-coated fluorescent bead adhesion assay (24) in the absence or presence of PMA or the LFA-1activating antibody KIM185 (Fig. 3B). This adhesion assay allows analysis of only those cells that have similar expression levels of LFA-1. Only cells with a mean fluorescence of 70 -80 were analyzed for ICAM-1-coated fluorescent bead binding. All LFA-1 mutants were able to adhere to ICAM-1 for at least 40% when activated by KIM185, indicating that the LFA-1 molecules are functionally expressed on the K562 cells. The level of LFA-1-mediated adhesion to ICAM-1 without any activation is low (Ͻ8%) except for the mutant ␣ L /L732R-␤ 2 (18%). Mutation of potential tyrosine and serine phosphorylation sites within the ␤ 2 -chain to ␤ 1 residues (␣ L /Y735F-␤ 2 and ␣ L /S756C-␤ 2 ) do not restore the PMA responsiveness (Fig. 3B). The same holds true for the threonine at position 758, which has been reported to be involved in cell spreading, and the phenylalanine at position 766, which affects ligand binding (16). Of all mutants, only substitution of leucine for the ␤ 1 amino acid arginine (␣ L /L732R-␤ 2 ) in the DLRE motif of the ␤ 2 cytoplasmic domain results in a significant increase in adhesion to ICAM-1 upon PMA stimulation. Since the adhesion of unstimulated ␣ L / L732R-␤ 2 cells is already high (18%), which might facilitate PMA activation, we investigated adhesion to decreasing ICAM-1 concentrations to a level in which the default adhesion of ␣ L /L732R-␤ 2 was similar to that of wild type LFA-1 (Fig. 3B,  inset). At low ICAM-1 concentration, PMA could still enhance the adhesion of mutant ␣ L /L732R-␤ 2 , indicating that the PMA responsiveness is truly induced by the point mutation and not by inherent stronger adhesion. In addition, similar as ␣ L / L732R-␤ 2 , mutant ␣ L /S756C-␤ 2 has also a default adhesion of 10% but does not respond to PMA. To investigate whether this single mutation is crucial for PMA responsiveness, we mutated in the ␤ 2 /␤ 1 chimera the arginine present in the DRRE motif of the ␤ 1 cytoplasmic domain to a leucine (␣ L /R732L-␤ 1 ). However, in mutant ␣ L /R732L-␤ 1 , the PMA responsiveness was not abolished, suggesting that this arginine residue within the full ␤ 1 cytoplasmic domain is not essential for PMA responsiveness and thus that the PMA activation of ␤ 1 integrins is differently regulated than ␤ 2 integrins. The PMA-induced adhesion of the latter mutant seems somewhat lower than in mutant ␤ 2 /␤ 1 chimera. However, comparing the relative PMA inducibility between the PMA-responsive mutants demonstrates that there are no significant differences (Table I). Taken together, these results indicate that the created DRRE motif in the ␤ 2 cytoplasmic domain is essential for PMA-mediated activation of ␤ 2 integrin LFA-1.
Avidity but Not Affinity Changes Correlate with PMA Induction of ␤ 2 Cytoplasmic Mutants-To investigate whether the restored PMA responsiveness of the ␣ L /L732R-␤ 2 mutant is due to a change in avidity and/or affinity, we performed confocal microscopy to detect avidity alterations and soluble ICAM-1 binding studies as affinity measurements. Analysis of the LFA-1 cell surface distribution by confocal microscopy shows that substitution of the ␤ 2 cytoplasmic domain for the ␤ 1 domain leads to clustering of LFA-1 on the cell surface (13), whereas wild type LFA-1 expressed in K562 cells shows homogeneous distribution of LFA-1 (Fig. 4A). All point mutants in the ␤ 2 cytoplasmic domain have a homogeneous LFA-1 distribution (Fig. 4, C and D, and data not shown), with the exception of the PMA-responsive LFA-1 mutant ␣ L /L732R-␤ 2 that exhibits clusters of LFA-1 (Fig. 4B). Surprisingly, LFA-1 is clustered in all of the ␤ 1 cytoplasmic domain point mutants investigated, whereas the reversed mutation ␣ L /R732L-␤ 1 has a homogeneous LFA-1 distribution (data not shown) despite the fact that the mutant is able to bind to ICAM-1 upon stimulation by PMA. These results suggest that the clustered status of the ␤ 2 integrin LFA-1 in ␣ L /␤ 2 /␤ 1 and ␣ L /L732R-␤ 2 may facilitate the PMA responsiveness of LFA-1.
Whether also the affinity of LFA-1 for ICAM-1 is altered in the ␣L/L732R-␤2 mutant that responds to PMA, we determined the concentration of soluble ligand (ICAM-1Fc) that yielded half-maximal direct ligand binding activity (ED 50 ). High affinity of LFA-1 for ICAM-1 results in a low concentration of ICAM-1Fc needed to obtain 50% of maximal binding. Strong binding of ICAM-1Fc was observed after stimulation of c Student's t test: K-␣ L /␤ 2 p value Ͻ 0.0001. No significant differences between ␣ L /␤ 2 /␤ 1 , ␣ L /L732R-␤ 2 , and ␣ L /L732R-␤ 1 .
Importance of the ␤ 2 Cytoplasmic Domain for PMA Responsiveness of LFA-1 in K562 Cells-To determine whether only L732R in the ␤ 2 domain was enough to generate PMA-induced adhesion, we deleted the ␤ 2 cytoplasmic domain directly after the wild type leucine at position 732 in the DLRE motif, at the similar position in the mutant ␣ L /L732R-␤ 2 , or after the aspartic acid at position 731 (Figs. 1A and 2). Deleting the cytoplasmic domain immediately after position 731 or 732 completely abolished the PMA-induced adhesion to ICAM-1 (Fig. 5A), whereas the LFA-1-activating antibody KIM185 induced ICAM-1 binding equally well (50%). These results demonstrate that next to the ␤ 1 residue L732R in the ␤ 2 cytoplasmic domain also other residues within the C-terminal part of the ␤ 2 cytoplasmic domain are necessary for the acquired PMA responsiveness of mutant L732R-␤ 2 .
Threonine 758 Is Important for PMA Responsiveness of L732R-␤ 2 -To investigate which amino acids C-terminal of position 732 in the ␤ 2 cytoplasmic domain are necessary together with L732R for PMA-induced LFA-1 activation, double mutants were created that contained both L732R and serine and threonine mutations located C-terminal of L732R (Fig. 1). Serine and threonine residues have been shown to be important in LFA-1 phosphorylation and function (16,25). These two double mutants, designated ␣ L /L732R,S756C-␤ 2 and ␣ L / L732R,T758V-␤ 2 , were stained for functional expression of LFA-1 (data not shown), and the LFA-1-mediated adhesion to ICAM-1 was studied using the ICAM-1-coated fluorescent bead adhesion assay (Fig. 5B). To our surprise, the double mutant ␣ L /L732R,T758V-␤ 2 disrupted the L732R-induced PMA response from 35 to 5%. Mutation of the potential serine phosphorylation site (S756C) did not alter the PMA responsiveness of double mutant ␣ L /L732R,S756C-␤ 2 (37%). These data suggest that the acquired PMA responsiveness of mutant ␣ L / L732R-␤ 2 depends on the threonine residue at position 758 but not the serine residue at position 756.
The ␤ 2 -cytoplasmic domain contains several phosphorylation-sensitive serine and threonine residues that are phosphorylated upon phorbol ester stimulation (25). Since the ␣ L / L732R-␤ 2 mutant could restore the PMA response, whereas the double mutant ␣ L /L732R,T758V-␤ 2 blocked this responsiveness, we investigated the importance of serine or threonine phosphorylation of the ␤ 2 -cytoplasmic domain due to PMA activation. Both mutants and wild type LFA-1 in K562 cells are serine-and threonine-phosphorylated on the ␣ L -and ␤ 2 -cytoplasmic domain with or without PMA stimulation (data not shown), suggesting that the lack of PMA responsiveness of mutant ␣ L /L732R,T758V-␤ 2 is not caused by an impaired phosphorylation on serine or threonine residues in the LFA-1 molecule.
To further investigate whether affinity and/or avidity changes regulate the PMA response of the double mutants, we determined the affinity of ICAM-1 by measuring the soluble ICAM-1Fc concentration needed to yield half-maximal binding activity (Table II). Although the double mutants ␣ L / L732R,S756C-␤ 2 and ␣ L /L732R,T758V-␤ 2 differed in PMA responsiveness, no significant changes could be observed for the ED 50 (1.9 Ϯ 0.70 g/ml ICAM-1 and 3.3 Ϯ 1.79 g/ml ICAM-1, respectively). However, analysis of the cell surface distribution of LFA-1 revealed that LFA-1 was homogeneously distributed on double mutant ␣ L /L732R,T758V-␤ 2 (Fig. 4E), whereas the PMA responsiveness of mutant ␣ L /L732R,S756C-␤ 2 (Fig. 4F) showed a clustered LFA-1 surface distribution. This demonstrates a strong correlation between a clustered LFA-1 cell surface distribution and the capacity to respond to PMA.
PMA-mediated Activation of L732R-␤ 2 Involves Cytoskeletal Rearrangements-It has been reported that activation of LFA-1 in leukocytes is tightly regulated by the organization of the cytoskeleton. Since LFA-1 clustering strongly correlates with PMA responsiveness, we studied which cytoskeleton associated proteins are involved in reorganization of LFA-1. We therefore investigated by inhibiting actin assembly whether low concentrations of cytochalasin D could allow integrin clustering and adhesion of non-PMA-responsive LFA-1 transfectants. To our surprise, cytochalasin D could not restore the PMA response (Fig. 6) or enhance clustering of LFA-1 (data not shown) as was observed in leukocytes (14). In contrast, mutant ␣ L /L732R-␤ 2 exposed to cytochalasin D resulted in a constitutively active LFA-1 molecule, and cell adhesion was not further increased by PMA. This indicates that the clustered ␣ L /L732R-␤ 2 might still be associated with the cytoskeleton.
Recently, increased levels of intracellular Ca 2ϩ could enhance LFA-1-mediated adhesion through induction of avidity changes in LFA-1 due to the activation of a Ca 2ϩ -dependent protease calpain that disrupts the cytoskeletal association with LFA-1 (26). Activation of calpain can be blocked by reagent calpeptin. To investigate whether our LFA-1-transfected K562 cells regulate their cell surface distribution by the activation of calpain, we inhibited adhesion to ICAM-1 with calpeptin at concentrations previously demonstrated to block calpain activity (26). Surprisingly, calpeptin completely abrogated the PMA responsiveness of mutant ␣L/L732R-␤2, whereas a lower con-centration of calpeptin (20 g/ml) could not fully block the PMA-induced adhesion (Fig. 6). These results indicate that PMA acts via calpain to promote activation of LFA-1 through partial dissociation from the cytoskeleton facilitating clustering of LFA-1 molecules. DISCUSSION Using a cell transfection system in which inside-out signaling of the ␤ 2 integrin LFA-1 could be modified by substitution of the ␤ 2 for the ␤ 1 cytoplasmic domain, we searched for single amino acids in the ␤ 2 cytoplasmic domain that regulate insideout signaling. We identified one amino acid at position 732, in which a leucine is substituted for an arginine that could restore the PMA responsiveness of LFA-1 completely. PMA inside-out signaling also depends on a threonine located more C-terminally at position 758. Of all LFA-1 mutants that respond to  Wild type LFA-1 and point mutant ␣ L /L732R-␤ 2 transfected into K562 cells were preincubated with 5 g/ml cytochalasin D or 100 g/ml calpeptin. Cells were subsequently stimulated with medium, PMA (100 nM), or the activating anti-␤ 2 mAb KIM185 (10 g/ml) for 30 min at 37°C in the absence or presence of the LFA-1-blocking mAb NKI-L15. ␣ L /L732R-␤ 2 cells were also preincubated with 20 g/ml (dotted bars), 60 g/ml (hatched bars), or 100 g/ml (black bars) calpeptin before stimulation with PMA. Depicted is the mean percentage Ϯ S.D. of LFA-1-specific adhesion to ICAM-1 of the gated cells that expressed equal amounts of LFA-1 (mean fluorescent intensity 70 -80) as determined by staining with the FITC-conjugated nonblocking anti-LFA-1 antibody (TS2/4). Integrin-specific adhesion is calculated as explained in Fig. 5. Data are representative for three experiments. PMA, avidity alterations and not affinity changes or ␤ 2 phosphorylation seemed important for proper function. We propose a model for this PMA responsiveness regulated by position 732 and 758 and avidity changes regulated by activation of a Ca 2ϩdependent protease calpain that releases LFA-1 from the cytoskeleton, thereby allowing the formation of a signaling complex leading to active LFA-1.
Integrin-dependent adhesion is strongly induced upon inside-out signaling when PKC is activated through the addition of PMA or via T cell receptor triggering. It remains still obscure how "inside-out" signaling by PMA results in LFA-1 activation. The newly identified single amino acid mutation (L732R) responsible for PMA activation of LFA-1 is situated in the ␤ 2 DLRE motif, which is conserved throughout the other integrins but distinct in one amino acid (DRRE) in ␤ 1 and ␤ 7 integrins. PMA can activate ␤ 1 integrins, but not ␤ 2 and ␤ 7 in K562 cells, suggesting that besides the DRRE motif also lymphocyte specific elements are involved (13). In addition, creating the DLRE motif in the ␤ 1 cytoplasmic domain (␣ L /R732L-␤ 1 ) did not abolish the PMA responsiveness, indicating that the PMA activation of ␤ 1 integrins is differently regulated compared with ␤ 2 integrins. The DLRE motif has been proposed to bind the GFFKR motif in the ␣-chain, and both of these cytoplasmic domains serve to constrain LFA-1 into a default low affinity state (27). Mutations in the TTT region (positions 758 -760) into alanines residues of the ␤ 2 cytoplasmic domain have been shown to reduce the default adhesion to ICAM-1 and the phorbol ester-mediated LFA-1 phosphorylation when expressed in COS cells or B lymphoblastoid cells, but they do not abrogate the binding to ICAM-1 in response to PMA (16). In contrast, we observed with the double mutant ␣ L /L732R,T758V-␤ 2 that substitution of the threonine into the ␤ 1 residue valine in K562 cells completely decreased the PMA-induced adhesion restored by the L732R mutation and altered the LFA-1 surface distribution. The phosphorylation level of these threonines after stimulation with PMA is strongly increased upon pretreatment with okadaic acid, which inhibits serine and threonine phosphatases (25). Threonine-phosphorylated CD18 molecules have been shown to associate with the cytoskeleton (28) and play an important role in the formation of stress fibers and specialized microdomains, such as rafts (15,29,30). However, in our system we could not identify any role of threonine phosphorylation and the PMA responsiveness of LFA-1, although the threonines themselves are a prerequisite for the PMA response together with the mutation L732R. In line with the threonine phosphorylation, we observed also no differences in serine phosphorylation, even in double mutant ␣ L /L732R,S756C-␤ 2 that was still able to adhere to ICAM-1 upon stimulation with PMA. This further questions the relevance of this serine residue in phosphorylation and ICAM-1 binding as shown in other studies (16). The tyrosine-based NPKY motif in the ␤ 1 cytoplasmic domain has been implicated in regulating integrin function (31). However, substitution of the phenylalanine for the ␤ 1 tyrosine (F766Y-␤ 2 ) in this motif did not restore PMA sensitivity. Thus, phosphorylation of the ␤ 2 cytoplasmic domain is not a prerequisite for the acquired PMA responsiveness.
Cytohesin, a member of the guanine nucleotide exchange factors for ADP-ribosylation factor G-proteins, specifically interacts with the ␤ 2 cytoplasmic domain directly after the transmembrane region (positions 723-725), thereby controlling T cell receptor or phorbol ester-induced activation of LFA-1 (17). Cytohesin expression is involved in maintaining LFA-1 in a high avidity state. Since cytohesin is expressed in K562 cells and associates with LFA-1 (32), it is likely that the high avidity state of our DRRE mutant is a direct result of cytohesin binding. Double staining of LFA-1 and cytohesin in the ␤ 2 point mutants did not demonstrate differences in colocalization of LFA and cytohesin (data not shown). Upon PMA activation, many proteins are phosphorylated and activated via PKC such as the ␤ 2 -linked proteins Rack1, MacMARCKS, and L-plastin. Phosphorylated Rack1 binds PKC, allowing subsequent recruitment of Rack1 to the KALI region in the ␤ 2 cytoplasmic domain, which is the same binding region for cytohesin (17,18). The WD repeats 5-7 of Rack1 interact with ␤ integrins, leaving the other repeats free for binding to PKC and possible cytohesin. Thus, Rack1 merely functions as a scaffold protein to recruit PKC and other ␤ 2 regulators to the site of action. Mac-MARCKS is a PKC substrate phosphorylated upon PMA activation, which leads to an increase in the lateral diffusion of ␤ 2 integrins and enhanced LFA-1-dependent cell clustering (33). We could not demonstrate by confocal microscopy any differences in cytosolic localization of MacMARCKS after activation with PMA (data not shown). Hence, it remains unclear whether MacMARCKS directly binds the ␤ 2 cytoplasmic domain or Rack1. The leukocyte-specific actin-bundling protein L-plastin proved to be important in enhanced integrin avidity through PKC and PI-3 kinase (34). Upon PMA activation, calcium is released from intracellular stores and binds to the EF-hand type calcium-binding domain of L-plastin, thereby inhibiting actin bundling activity. Thus, MacMARCKS as well as L-plastin play a crucial role in the association of the integrin with the cytoskeleton and subsequently integrin activation. Besides PKC, phosphatidylinositol 3-kinase and the small GTPase Rap1 modulate LFA-1 avidity in leukocytes (35,36). Both Rap1 and the phosphatidylinositol 3-kinase but not PKC mediated activation up-regulated the NKI-L16 epitope, indicating increased LFA-1 avidity (35).
The actin cytoskeleton plays a critical role in integrin activation and signaling by acting as a platform to bring different components close together, leading to a signaling complex. The ␤ cytoplasmic domain has been demonstrated to be associated with the cytoskeletal component ␣-actinin (19), vinculin (37), filamin (38), or talin (39). Treatment of cells with cytochalasin D, which disrupts the cytoskeleton network, results in activation of LFA-1 that coincides with clustering of LFA-1, indicating that the cytoskeleton restraints keep integrins inactive (14). However, cytochalasin D had no effect on activation or clustering of wild type LFA-1 expressed in K562 cells, indicating a different mechanism of cytoskeletal organization. In contrast, the PMA-responsive mutant ␣ L /L732R-␤ 2 could be spontaneously activated by cytochalasin D. The release of LFA-1 from the cytoskeleton in lymphocytes is also thought to be regulated by the cysteine protease calpain that is activated by local Ca 2ϩ fluxes (26). Indeed, we have evidence that the PMAinduced activation of LFA-1 is mediated by calpain, since inhibition with calpeptin abrogated the PMA responsiveness of mutant ␣ L /L732R-␤ 2 . Calpeptin also has been reported to induce stress fiber formation in fibroblasts due to its inhibitory action on protein-tyrosine phosphatases upstream of the small GTPase Rho (40). It is rather unlikely that this additional effect of calpeptin on protein-tyrosine phosphatases also occurs during the integrin-mediated adhesion of the nonadherent K562 cells that do not induce stress fiber formation. Proteins identified as potential calpain targets include talin, filamin, and ␣-actinin. Talin forms the bridge between the ␤ 2 integrin and the actin filaments. Upon activation, LFA-1 is released from the cytoskeleton as a result of proteolysis of talin, probably by calpain, leading to freely mobile integrin as postulated by Sampath et al. (46). Next, ␣-actinin binds the ␤ 2 cytoplasmic domain between residues 736 and 746 directly C-terminal of the DLRE motif, thereby stabilizing the cytoskeleton-integrin interaction necessary for strong adhesion. We speculate that this ␤ 2 -specific event is impaired in K562, whereas the DRRE mutant partly restores the lateral mobility of LFA-1, resulting in increased LFA-1 avidity. The ␣-actinin binding motif is also important for endoplasmic reticulum retention, assembly, and transport to the cell surface of LFA-1 (37). Peptides from the ␤ 1 cytoplasmic domain reveal that pp125 FAK and paxillin bind the membrane-proximal KLLMIIHDRREFA motif, which includes the DRRE motif (41). In ␤ 2 integrins, paxillin is tyrosine-phosphorylated by MacMARCKS, and both colocalize in the membrane ruffles of spreading macrophages (42).
The acquired PMA signaling of our LFA-1 mutants in K562 cells coincides with a change in a clustered LFA-1 cell surface distribution. Much attention has been recently given to specialized lipidic membrane microdomains, also termed "rafts." They function as platforms for signaling molecules and are involved in the regulation of LFA-1 function and adhesion through avidity changes (15,30). Whether the PMA-responsive LFA-1 mutant is differently organized in rafts compared with wild type LFA-1 remains so far unsolved. Our results suggest that increased avidity facilitates PMA induced adhesion to ICAM-1 rather than affinity changes. Clustering of LFA-1 molecules probably leads to a higher concentration of signaling components involved in LFA-1 signaling such as cytohesin, Rack1, paxillin, or MacMARCKS as shown in a model in Fig. 7. Furthermore, activation of LFA-1 by PMA is dependent on calpain, which cleaves cytoskeletal components. This is probably a crucial event, because one can imagine that dislodgment from the cytoskeleton facilitates binding of signaling molecules, leading to a reorganization of the cytoskeleton and activation of LFA-1. We have shown that this process can be overruled by adding cytochalasin D. However, PMA seems ineffective in cells with a homogeneous distribution of LFA-1. In the case of the mutants with a clustered LFA-1 distribution, the threshold for triggering the PMA signaling cascade is lower, since the concentrations of signaling molecules directly or indirectly connected to the cytoplasmic domain of LFA-1 are higher. Therefore, we cannot exclude the possibility that PMA indeed has an effect on LFA-1 in K562 cells, albeit small and not detectable with our assays. By using the soluble ICAM-1 binding assay, we can detect relatively small changes in LFA-1 affinity. However, all of our mutants showed an equal ability to bind soluble ICAM-1, indicating no affinity differences. Previous reports suggest that affinity changes play an important role in regulating integrinmediated adhesion, although we showed earlier that this is not true for ␤ 2 integrins (43,44). Activation of LFA-1 by EGTA or Mg 2ϩ leads to enhanced expression of the M24 epitope, indicating that Mg 2ϩ binding induces conformational changes in LFA-1, leading to enhanced ICAM-1 binding (45). Since we observed no changes in affinity and M24 expression of the mutants, we conclude that the acquired PMA responsiveness has no effect on the extracellular conformational changes of LFA-1.
In summary, we present in this study evidence that substitution of a single amino acid in the ␤ 2 DLRE motif together with an intact C-terminal TTT sequence is sufficient to restore PMA induced LFA-1 adhesion to ICAM-1. The gained PMA is nonclustered and cannot be activated by PMA to bind ligand. The threshold for PMA to activate LFA-1 is insufficient because an efficient signaling complex or microdomain cannot be formed due to the absence of clusters of LFA-1. However, substitution of the leucine at position 732 for the ␤ 1 amino acid arginine together with an intact TTT motif induces LFA-1 to reorganize into clusters and thereby restores the PMA responsiveness of LFA-1. PMA triggers PKC and subsequently activates, phosphorylates, and recruits a cascade of substrates, such as cytohesin (32), Rack1 (18), MacMARCKS (33), paxillin (42), L-plastin (34), and phosphatidylinositol 3-kinase (35), and increases intracellular Ca 2ϩ levels (9). Calcium can activate specific proteases such as calpain that releases LFA-1 from the cytoskeleton (26). Presumably, talin is cleaved from the cytoskeleton (46), resulting in mobile LFA-1 and reorganization of the cytoskeleton network. Cytochalasin D disrupts the cytoskeleton and bypasses the PMA-induced activation. This allows LFA-1 to recruit signaling components that re-establish the contact with the cytoskeleton, leading to the formation of a signaling complex able to efficiently bind ligand. Because LFA-1 is in a default clustered status in the L732R-␤ 2 mutant, these signaling components are concentrated near the LFA-1 cytoplasmic domain, leading to a signaling complex able to activate the LFA-1 threshold for PMA induced inside-out signaling. B, amino acid sequence of the ␤ 2 cytoplasmic domain with the binding sites (underlined) for cytohesin, Rack1, and ␣-actinin. The threonine residues at positions 758 -760 together with the phenylalanine (position 766) are required for ligand binding. The mutated lysine creating the DRRE motif and the essential threonine at position 758 restores the PMA responsiveness of LFA-1 expressed in K562 cells. signaling is probably due to the presence of a dense LFA-1 intracellular signaling complex, since these mutants have a clustered surface distribution of LFA-1. The activation of LFA-1 is dependent on rearrangements of the cytoskeleton through a mechanism involving a Ca 2ϩ -dependent protease calpain. This work clearly demonstrates that the function of LFA-1 is strictly regulated and involves a leukocyte-specific signaling element.