Actin cytoskeletal association of cytohesin-1 is regulated by specific phosphorylation of its carboxyl-terminal polybasic domain.

Cell adhesion mediated by integrin receptors is controlled by intracellular signal transduction cascades. Cytohesin-1 is an integrin-binding protein and guanine nucleotide exchange factor that activates binding of the leukocyte integrin leukocyte function antigen-1 to its ligand, intercellular adhesion molecule 1. Cytohesin-1 bears a carboxyl-terminal pleckstrin homology domain that aids in reversible membrane recruitment and functional regulation of the protein. Although phosphoinositide-dependent membrane attachment of cytohesin-1 is mediated primarily by the pleckstrin homology domain, this function is further strengthened by a short carboxyl-terminal polybasic amino acid sequence. We show here that a serine/threonine motif within the short polybasic stretch of cytohesin-1 is phosphorylated by purified protein kinase C delta in vitro. Furthermore, the respective residues are also found to be phosphorylated after phorbol ester stimulation in vivo. Biochemical and functional analyses show that phosphorylated cytohesin-1 is able to tightly associate with the actin cytoskeleton, and we further demonstrate that phosphorylation of the protein is required for maximal leukocyte function antigen-1-mediated adhesion of Jurkat cells to intercellular adhesion molecule 1. These data suggest that both phosphatidylinositol 3-kinase and protein kinase C-dependent intracellular pathways that stimulate beta(2)-integrin-mediated adhesion of T lymphocytes converge on cytohesin-1 as functional integrator.


Cell adhesion mediated by integrin receptors is controlled by intracellular signal transduction cascades.
Cytohesin-1 is an integrin-binding protein and guanine nucleotide exchange factor that activates binding of the leukocyte integrin leukocyte function antigen-1 to its ligand, intercellular adhesion molecule 1. Cytohesin-1 bears a carboxyl-terminal pleckstrin homology domain that aids in reversible membrane recruitment and functional regulation of the protein. Although phosphoinositide-dependent membrane attachment of cytohesin-1 is mediated primarily by the pleckstrin homology domain, this function is further strengthened by a short carboxyl-terminal polybasic amino acid sequence. We show here that a serine/threonine motif within the short polybasic stretch of cytohesin-1 is phosphorylated by purified protein kinase C␦ in vitro. Furthermore, the respective residues are also found to be phosphorylated after phorbol ester stimulation in vivo. Biochemical and functional analyses show that phosphorylated cytohesin-1 is able to tightly associate with the actin cytoskeleton, and we further demonstrate that phosphorylation of the protein is required for maximal leukocyte function antigen-1-mediated adhesion of Jurkat cells to intercellular adhesion molecule 1. These data suggest that both phosphatidylinositol 3-kinase and protein kinase C-dependent intracellular pathways that stimulate ␤ 2integrin-mediated adhesion of T lymphocytes converge on cytohesin-1 as functional integrator.
Integrins are a diverse family of heterodimeric transmembrane adhesion receptors that are present on most vertebrate cell types. They are known to play important roles in either development or somatic functions such as wound healing and the regulation of complex cell-cell or cell-matrix interactions within the immune system (1).
The avidity of integrins for their ligands is dependent on the activation state of the cell they are expressed on (1)(2)(3). This type of regulation of cell adhesion has been termed inside-out signaling because intracellular signaling pathways triggered by protein tyrosine kinase or G-protein-coupled receptors have been shown to contribute to integrin-mediated adhesiveness. The mechanisms by which cytoplasmic signals are transmitted across the plasma membrane through integrin receptors are just emerging, but compelling evidence suggests that the intracellular domains of both ␣ and ␤ chains participate in this process (4 -8).
Previous studies have attempted to elucidate these signaling pathways. Recently, candidate cytoplasmic regulatory factors of integrin activation have been identified, either by biochemical methods or with the help of the two-hybrid system (7). One of them, cytohesin-1, is a 47-kDa intracellular protein that interacts specifically in several systems with the cytoplasmic domain of the leukocyte integrin ␣ L ␤ 2 (CD11a/18, LFA-1) (9). Cytohesin-1 bears a short amino-terminal domain that may aid in oligomerization, an extended central homology region that is similar to that of the yeast Sec7 protein, and a carboxyl-terminal pleckstrin homology (PH) 1 domain as well as a short polybasic region. Overexpression of cytohesin-1 or subdomain constructs in the Jurkat T-cell line was shown to have pronounced in vitro effects on the binding of ␣ L ␤ 2 to its ligand, intercellular adhesion molecule 1 (ICAM-1). Whereas the overexpression of full-length cytohesin-1 resulted in a constitutive adhesion of ␣ L ␤ 2 , expression of the PH domain construct specifically inhibited the activation of LFA-1 in a dominant negative fashion (9).
PH domains are structural modules present in more than a hundred proteins that play known or postulated roles in signal transduction. It is commonly found that PH domains aid in membrane recruitment of proteins through their interactions with phosphorylated ligands present at the inner leaflet of cellular membranes (10 -14). A number of PH domains, including the PH domain of cytohesin-1, were shown to bind phosphatidylinositol 3,4,5-trisphosphate (PIP 3 ) in vitro with high affinity (12,(15)(16)(17)(18). These findings implicated the involvement of phosphatidylinositol 3-kinase in the LFA-1 activation pathway, and we have subsequently shown that a constitutively active version of phosphatidylinositol 3-kinase suffices to activate the ␣ L ␤ 2 adhesion pathway in a T-cell line (17). Functional, biochemical, and cell biological evidence furthermore suggested that cytohesin-1 is located downstream of phosphatidylinositol 3-kinase and that it is regulated by the recruitment of its PH domain to the plasma membrane (17,20,21). The PH domain was supported in this function by the carboxyl-terminal polybasic region of the protein, which proved to be necessary for membrane association and stabilized PIP 3 binding in vitro (22). These data supported a model in which membrane recruitment of a regulatory molecule provided an effective means for activating ␤ 2 integrin function.
Cytohesin-1 belongs to a family of molecules that bear specific guanine nucleotide exchange activity for so-called ARF-GTPases (23,24); these molecular switches belong to the Ras superfamily. ARFs are involved in intracellular vesicle budding and have also been implicated in remodeling of the actin cytoskeleton. We have very recently shown that a mutation in the Sec7 domain of cytohesin-1 that blocks GTP binding to ARF proteins in vitro also interferes with cell spreading and adhesion in Jurkat cells and peripheral blood mononuclear cells (25).
In this study, we show that the regulatory molecule cytohesin-1 is a likely target for other important elements of the signaling machinery that regulates ␤ 2 integrin-mediated cell adhesion (4). We found that phorbol ester stimulation of Jurkat cells results in phosphorylation of cytohesin-1. The protein is specifically modified at a site within the carboxyl-terminal polybasic region, the presence of which had been shown to play a crucial role in membrane recruitment of the protein; mutational and functional analysis showed that phosphorylated cytohesin-1 is able to associate with the actin cytoskeleton.
In Vivo [ 32 P]Orthophosphate Labeling of Cytohesin-1-Jurkat E6 cells (1.6 ϫ 10 7 ) were infected with recombinant vaccinia viruses as described previously (28,29). 1 h after infection, cells were washed with PBS, resuspended in phosphate-free RPMI 1640 medium, and incubated with 1000 Ci [ 32 P]orthophosphate/ml for 5 h at 37°C with 5% CO 2 in a humidified incubator. Subsequently, 40 ng/ml PMA was added. After an incubation period of 30 min, the cells were collected and lysed with lysis buffer (100 mM Tris-Cl, pH 8.0, 150 mM NaCl, 2 mM EDTA, 1% Triton X-100, 10 g/ml leupeptin, 10 g/ml aprotinin, and 1 mM phenylmethylsulfonyl fluoride). Samples were centrifuged, and the resulting supernatants were incubated with protein A-Sepharose 6MB beads (Amersham Pharmacia Biotech). Beads were washed multiple times with lysis buffer, and the resulting samples were loaded onto SDS-polyacrylamide gels. Immunoblots from these gels were exposed to x-ray films.
PMA-induced Phosphorylation in Vivo-Jurkat E6 cells or transfected COS-7 cells were stimulated for 1 h with 40 ng/ml PMA. Cells were collected and processed as described under "Cellular Fractionation." Subsequently, the cytosolic fractions were subjected to nondenaturing gel electrophoresis.
In Vitro Dephosphorylation of Cytohesin-1-COS-7 cells were treated as described under "PMA-induced in Vivo Phosphorylation." 5 units of alkaline phosphatase were added to 50-l aliquots of cell extract where indicated. Samples were incubated at 37°C and subjected to nondenaturing PAGE.
In Vitro Phosphorylation of Cytohesin-1-Cytohesin-1 protein was expressed in COS-7 cells by DEAE-dextran transfection. Cells that had been transfected with cDNA coding for cytohesin-1 were collected by centrifugation and resuspended on ice in 0.5 ml of ice-cold hypotonic solution (10 mM HEPES, pH 7.5, 10 mM KCl, 10 mM MgCl 2 , and 0.5 mM dithiothreitol) containing 10 g/ml leupeptin, 10 g/ml aprotinin, 2 mM benzamidine, and 1 mM phenylmethylsulfonyl fluoride. The cells were then sheared mechanically and centrifuged at 21,000 ϫ g for 15 min to remove nuclei and the insoluble fraction, and the resulting supernatant cytosol was collected. This cytosolic fraction containing cytohesin-1 was used for in vitro phosphorylation under the following reaction conditions: 50 mM K 2 HPO 4 /KH 2 PO 4 , 10 mM HEPES, 10 mM MgCl 2 , 0.1 mM EGTA, 200 g/ml phosphatidylserine, 20 g/ml PMA, 2 mM ATP, 0.05% Nonidet P-40, and 20 g/ml PKC␦ at pH 7.4 and 30°C for 120 min.
Nondenaturing Gel Electrophoresis-Nondenaturing gel electrophoresis was essentially performed as described previously (30). The following, slightly modified buffers and solutions were used in the present study: (a) separation gel, 3 ml of 1 M Tris-Cl (pH 9.0), 1. Cellular Fractionation-Jurkat E6 cells were pretreated with cytochalasin D (50 M, 60 min) where indicated to disrupt actin filaments or with nocodazole (2.5 g/ml, 60 min) to disrupt microtubules. These concentrations of cytochalasin D or nocodazole were maintained throughout the whole fractionation procedure. Cells were collected by centrifugation and resuspended on ice in 0.5 ml of ice-cold hypotonic solution (10 mM HEPES, pH 7.5, 10 mM KCl, 10 mM MgCl 2 , and 0.5 mM dithiothreitol) containing 10 g/ml leupeptin, 10 g/ml aprotinin, and 1 mM phenylmethylsulfonyl fluoride. Fractionation of cells was performed as described previously (31). Briefly, cells were sheared, nuclei were removed by centrifugation at 1000 ϫ g for 10 min, and the supernatant cytosol was collected. The cytosolic fraction was brought to a final concentration of 1% (v/v) Nonidet P-40 and 150 mM NaCl and centrifuged at 15,000 ϫ g, and the resulting pellet was resuspended in hypotonic buffer. A typical sample, containing mainly cytoskeletal components from ϳ2 ϫ 10 7 cells was incubated with 50 ng of GST-cytohesin-1 for 10 min, as indicated. The samples were centrifuged at 21,000 ϫ g for 15 min, and the pellet (insoluble fraction containing cytoskeletal components) and supernatant (soluble fraction) were subjected to gel electrophoresis.
Adhesion Assay-Jurkat E6 cells were infected with recombinant vaccinia viruses. 6 h after infection, cells were labeled with 12 g/ml bisbenzimide H33342 fluorochrome trihydrochloride (Calbiochem) for 30 min at 37°C, collected by centrifugation, resuspended in Hank's buffered saline solution, and delivered to 96-well plates (NUNC; Maxisorp) at 1.5 ϫ 10 5 cells/well. Before adhesion, plates were coated with goat anti-human IgG (Fc␥-specific) antibody at 0.85 g/well for 90 min at 25°C, blocked with 1% (w/v) BSA in PBS, incubated with culture supernatants from COS-7 cells expressing ICAM-1-Ig fusion protein, and subsequently used in the assay. Where indicated in the figures, cells were incubated with 40 ng/ml PMA or 10 g/ml cytochalasin D for 0.5 h before the adhesion assay. Cells were then allowed to adhere for 1 h at 37°C, and unbound cells were carefully washed off with 3 ϫ 300 l of Hank's buffered saline solution. Bound cells were assayed in 100 l of 2% (v/v) formaldehyde in PBS using a fluorescence plate reader (Cytofluor II; PerSeptive). The signal of 1.5 ϫ 10 5 cells/well at 490 nm corresponds to 100% adhesion.
Indirect Immunofluorescence-6 h after infection of Jurkat E6 cells with recombinant vaccinia viruses and pretreatment for 1 h with 20 ng/ml PMA or 10 g/ml cytochalasin D, where indicated, cells were placed on poly-L-lysine-covered microscope slides for 1 h in a humidified chamber at 37°C. The nonadherent cells were then washed off with Hank's buffered saline solution, and adherent cells were fixed and immobilized with freshly prepared 2% (w/v) paraformaldehyde in PBS for 1 h at 4°C. Subsequently, cells were permeabilized for 15 min with 0.2% (v/v) Triton X-100 in PBS, blocked with 2% (w/v) glycine in PBS, and incubated with a fluorescein isothiocyanate-labeled goat anti-human IgG (Fc␥-specific) antibody (Dianova) in PBS for 2 h at room temperature. For the actin stain, slides were incubated with an antiactin antibody for 1 h. Slides were subsequently washed with PBS and incubated with Texas Red-labeled donkey anti-rabbit IgG antibody. After the final wash with PBS, slides were mounted on a 9:1 mixture of glycerol and 100 mM Tris-HCl, pH 9.0, containing n-propyl-gallate (20 mg/ml) as an anti-fading reagent. The samples were then examined on a confocal laser scanning apparatus (Leica TCS-NT system; Leica) attached to a Leica DM IRB inverted microscope with a PLAPO 63 ϫ 1.32 oil immersion objective.
Measurement of Phosphatidylinositol Binding of GST-PH Domain Constructs by IAsys Biosensor Technology-PH domains of cytohesin-1 were expressed as GST fusion proteins as described previously (17). An optical evanescence resonant mirror cuvette system (IAsys; Affinity Sensors) was used to measure interaction of GST fusion proteins with PIP 3 . A lipid monolayer containing 70% (w/w) ␤-palmitoyl-␥-oleoyl-L-␣phosphatidylcholine and 30% (w/w) dioleoyl-L-␣-phosphatidyl-DL-glycerol or a lipid mixture of 60% (w/w) ␤-palmitoyl-␥-oleoyl-L-␣-phosphatidylcholine and 30% (w/w) and 10% (w/w) PIP 3 (Matreya, Inc.) was mounted on a hydrophobic sensor surface (FCH-0601) at 0.1 mg/ml lipid. The cuvette was subsequently washed with 0.1 M HCl, PBS, and 10 mM NaOH. After the final wash with PBS, the cuvette was equilibrated in PDI binding buffer (PBS, 2 mM dithiothreitol, and 0.001% (v/v) Igepal CA-630), and affinity-purified GST fusion proteins dissolved in binding buffer were added to a final concentration of 150 nM each. The binding of the GST fusion proteins was monitored for 500 s. Dissociation was initiated by adding PDI binding buffer to the cuvette. Determination of the association equilibrium constant was done by equilibrium titration. The interaction profiles for each protein were analyzed using FASTfit kinetics analysis software supplied with the instrument.

Treatment of Jurkat Cells with Phorbol Ester Induces Phosphorylation of Cytohesin-1 in Vivo-With the help of in vivo
labeling experiments, we discovered that a cytoplasmic Ig (cIg) fusion protein of cytohesin-1 was strongly phosphorylated in Jurkat cells after incubation of the cells with PMA. COS-7 cells were transfected with a cDNA coding for cIg-cytohesin-1 fusion protein or control protein (Fig. 1); the cells were then labeled with 2 mCi/ml [ 32 P]orthophosphate and subsequently activated for 1 h with PMA. After this treatment, cells were lysed, and the fusion proteins from the samples were immunoprecipitated on protein A and subjected to SDS-PAGE and autoradiography. Fig. 1 shows that the 32 P content of the cIg-cytohesin-1 fusion protein increases dramatically and specifically after PMA activation, whereas background labeling of the control protein remained constant. Loading of the fusion proteins was identical, as detected by anti-cytohesin-1 or anti-Ig immunoblotting ( Fig. 1; data not shown).
We then discovered that the phosphorylated form of endogenous cytohesin-1 in Jurkat cells could be visualized as a fastmigrating species in nondenaturing Tris-glycine gels. Because no difference in the migratory behavior of endogenous or overexpressed cytohesin-1 from unstimulated or PMA-induced cells was observed when SDS-PAGE was employed ( Fig. 1; data not shown), we subsequently used nondenaturing Tris-glycine gels as well as an adapted immunotransfer protocol. Using this methodology, we found that PMA stimulation of the cells resulted in a second fast-migrating band that also reacted specifically with an anti-cytohesin-1 monoclonal antibody ( Fig. 2; data not shown). This band was hypothesized to be the phosphorylated form of the protein because in nondenaturing gels, modification of a protein with charged compounds should normally lead to an increase in its electrophoretic mobility (30). This was confirmed by adding alkaline phosphatase to the samples before electrophoresis, transfer, and detection: calf intestine phosphatase treatment was found to remove the lower band quantitatively from the samples. This was consist-ent with the notion that the lower band corresponded to in vivo phosphorylated cytohesin-1 and not to the PMA-dependent induction of a different isoform of the cytohesin family or to an unknown modification (Fig. 2).  Mapping the Site of PMA-dependent Phosphorylation in Cytohesin-1-Using the anti-phosphotyrosine antibody 4G10 as a detecting agent, we found that cytohesin-1 was likely not tyrosine-phosphorylated after PMA treatment (data not shown). It was therefore assumed that the modification occurred on serine(s) or threonine(s). Because cytohesin-1 belongs to a family of at least four extremely conserved proteins in mammals, we tested whether other family members became phosphorylated under the conditions used. Previously, we had produced a rat antibody that reacted exclusively with cytohesin-1 (25), but there are no reagents available to specifically detect other isoforms (e.g. cytohesin-2/ARNO or cytohesin-3/GRP1). We therefore employed expression of the respective proteins containing amino-terminally fused small FLAG tags because larger fusion elements were found to obscure the differences in migratory behavior of the phosphorylated versus the unphosphorylated forms (data not shown). Fig. 3 shows FLAG-tagged cytohesin-1, ARNO, and GRP1, which were expressed in Jurkat cells, subjected to Tris-glycine electrophoresis, and detected on immunoblots with the help of an anti-FLAG antibody. Comparing the samples, we observed that phosphorylation of cytohesin-1 was most robust and reproducible. Usually about 40 -50% of the protein was phosphorylated after 1 h of PMA treatment. In the ARNO sample, a fast-migrating but normally somewhat weaker band was detected in the majority of the experiments performed. However, we never found any indication for phosphorylated GRP1 in this system.
An important clue to the mapping of the phosphorylation site was therefore obtained by comparison of the peptide sequences of cytohesin-1, ARNO, and GRP1. There are several PKC consensus phosphorylation sites in these proteins, but most of these are conserved among them. However, the carboxyl-terminal polybasic stretch is an exception, inasmuch as this portion has among the most divergent sequence elements in the otherwise highly conserved primary peptide structures of the protein family. Furthermore, the polybasic stretch was shown to be important for the function of cytohesin-1 because it collaborated with the PH domain in membrane recruitment and phospholipid binding. A close look revealed that cytohesin-1 contained two potential phosphorylation sites in this region, namely, serine 394 and threonine 395. Only one of these sites was conserved in ARNO (serine 392), whereas GRP1 did not contain a potential phosphorylation site in the homologous carboxyl-terminal sequence stretch at all.
Assuming that the phosphorylation site may be located in the carboxyl-terminal polybasic stretch (Fig. 4A), site-directed mutagenesis was used to specifically map the phosphorylated residues in cytohesin-1. Using the nondenaturing gel assay, we found that mutagenesis of both serine 394 and threonine 395 affected the phosphorylation status of cytohesin-1 (Fig. 4B). However, complete inhibition of phosphorylation was observed when both residues were simultaneously replaced by alanines or glycines, respectively. This result led us to conclude that serine 394 and threonine 395 are the phosphorylated residues in cytohesin-1. Because it was formally possible that by mutagenesis we had obtained an indirect effect, e.g. by targeting regulatory sites that remained unphosphorylated themselves, we mutated all consensus PKC phosphorylation sites in the protein and consistently found that none of these residues affected the phosphorylation of the protein at all (Fig. 4C). These results are therefore interpreted as a strong indication that cytohesin-1 is exclusively phosphorylated at residues 394 and 395 after PMA treatment of Jurkat cells.
Functional Relevance of the Phosphorylation of Cytohesin-1-We then addressed the functional significance of cytohesin-1 phosphorylation. Because the phosphorylation sites were mapped to a region of the protein that had been shown to be FIG. 3. Cytohesin-1 and ARNO, but not GRP1, are phosphorylated upon PMA stimulation. Cell extracts of COS-7 cells transfected with either FLAG-cytohesin-1, FLAG-ARNO, or FLAG-GRP1 were subjected to nondenaturing PAGE. Cells had been stimulated with PMA where indicated. Anti-FLAG immunoblots are shown. In the case of cytohesin-1 and ARNO, a faster migrating species is found in PMAtreated cells, whereas the electrophoretic motility of GRP1 remained unchanged.

FIG. 4. Identification of PMA-dependent phosphorylation sites in cytohesin-1.
A, sequence comparison of the carboxyl-terminal polybasic regions of cytohesin-1, ARNO, and GRP1. B, a cytohesin-1 mutant that lacks serine 394 and threonine 395 is not phosphorylated upon PMA stimulation in COS-7 cells. FLAG-cytohesin-1 (wild-type or mutant constructs) had been transfected into COS-7 cells and stimulated with PMA as indicated. All constructs were subjected to nondenaturing PAGE and subsequent immunoblot analysis. C, the remaining putative PKC sites of cytohesin-1 were disabled by the introduction of respective point mutations and used as specificity controls. All of these mutants still responded to PMA stimulation. important for membrane recruitment and function (22), we speculated that the addition of negative charge through phosphorylation events might affect the association of cytohesin-1 with phospholipids. We therefore generated another set of mutants in which serine 394 and threonine 395 were replaced by "constitutively active" residues, aspartate or glutamate, respectively. The rationale of this approach was based on the frequent observation that negatively charged amino acids can functionally substitute for phosphorylated serines or threonines, respectively.
The resultant cytohesin-1 mutants, S394D and T395E, or the double mutant ST394/95DE were expressed in Jurkat cells using recombinant vaccinia viruses, and subcellular localization was subsequently analyzed with the help of laser scanning confocal microscopy. The result of this experiment is shown in Fig. 5. We observed that the subcellular localization of cytohesin-1 or of the proposed loss of phosphorylation mutants S394A or T395G was very similar and corresponded to that published previously for a wild-type cytohesin-1 (22): the mutant cytohesin-1 fusion proteins were expressed in the cytoplasm, and a large proportion of the detectable material was constitutively associated with the cell cortex. To our surprise, gain of charge mutants S394D, T395E, and ST394/95DE showed an identical staining pattern. Furthermore, PMA treatment of an aliquot of the sample in which wild-type cytohesin-1 was expressed had no effect (Fig. 5).
This result was unexpected, and therefore it was possible that the in vivo effects of cytohesin-1 phosphorylation were somehow perturbed by the experimental system employed. We therefore analyzed whether specific phosphorylation of cytohesin-1 could be obtained by in vitro methods. To this end, COS-7 cell lysates that contained overexpressed FLAG-cytohesin-1 or control protein were incubated with purified PKC␦ protein in the presence of ATP. As shown in Fig. 6, PKC␦ induced a fast-migrating species of FLAG-cytohesin-1 in native gels, very similar to what we observed with PMA-treated cells. Consequently, the migration pattern of the FLAG-tagged mutant cytohesin-1 394/5AG was not altered by PKC in this assay (Fig.   6). Therefore, phosphorylation of cytohesin-1 could be reliably obtained by using PKC␦ in vitro.
GST fusion proteins harboring either the intact PH domain of cytohesin-1 (GST-PH) or the PH domain and the polybasic region (GST-PHc) were then purified from an Escherichia coli overexpression system (Fig. 7A). The proteins were subsequently phosphorylated in vitro using purified PKC␦ and [␥-32 P]ATP. The degree of 32 P incorporation into the protein was monitored using scintillation counting (data not shown). GST-PHc, either phosphorylated or unphosphorylated, was then employed in phospholipid binding assays, using IAsys biosensor technology (22). However, as shown in Fig. 7B, phosphorylation of the carboxyl-terminal segment of cytohesin-1 FIG. 6. Phosphorylation of cytohesin-1 by PKC␦ in vitro. FLAGtagged cytohesin-1 or cytohesin-1 394/5AG were expressed in COS-7 cells by transfection. Cell extracts were subsequently incubated with PKC␦ in vitro as indicated and subjected to nondenaturing PAGE. Before the addition of PKC, the nonphosphorylated isoform of cytohesin-1 corresponding to the upper band was detectable (left lane). In the presence of PKC␦, the lower, phosphorylated cytohesin-1 band was strongly induced in vitro. However, band-shift was only observed for the wild-type protein (left panel) and was not observed for the phosphorylation site mutant 394/5AG (right panel).

FIG. 5. Subcellular localization of cytohesin-1 or cytohesin-1 mutants as determined by confocal microscopy.
All constructs were expressed as cIg fusion proteins in Jurkat E6 cells. Overexpressed cytohesin-1 wild-type (cIg-cyh1) and mutants were mainly associated with the cell cortex in both unstimulated and phorbol ester-stimulated cells. Gain of charge mutants S394D (394D) and T395E (395E) and double mutant S394D/T395E (394/5DE) displayed a similar localization as loss of phosphorylation mutants S394A (394A) and T395A (395A) and double mutant S394A/T395G (394/5AG). Cytohesin-1 and all mutant isoforms thereof localized to the cortical area of the cell; the proteins were also found to be concentrated in regions of cell-cell attachment. had no effect on binding to PIP 3 in vitro. The observed binding of the carboxyl-terminal protein modules of cytohesin-1 to the phospholipid appeared to be specific because GST control protein had no affinity for the PIP 3 -derivatized interaction matrix, and deletion of the polybasic domain strongly reduced the affinity of the fusion protein for PIP 3 (GST-PH), as described previously (22). We therefore concluded that the phosphorylation of cytohesin-1 had no effect on its association with phospholipid membranes.
Cytoskeletal Association of Cytohesin-1 Is Regulated by Phosphorylation-Assessed by microscopy, the interaction of cyto-hesin-1 with phospholipids of the plasma membrane might be indistinguishable from its cortical association with the cytoskeleton. Interestingly, it had been described previously that expression of a dominant negative mutant of cytohesin-1 in peripheral blood mononuclear cells or Jurkat cells had resulted in abrogation of cell spreading (25). Furthermore, introduction of the same mutant in Jurkat cells led to characteristic alterations of the actin cytoskeleton in adherent cells. 2 We therefore 2 W. Nagel and W. Kolanus, unpublished observations. speculated that the phosphorylation of cytohesin-1 might regulate its association with the actin cytoskeleton.
The association of cytohesin-1 with the actin cytoskeleton in native cell particulate fractions was assessed biochemically through crude separation of Jurkat cell extracts. Recombinant GST-cytohesin-1 was preincubated with PKC␦ in the presence or absence of ATP, as indicated (Fig. 8). A suspension of the particular fraction that had been depleted from membranes by detergent was subsequently added to the samples. After a coincubation period, samples were centrifuged, and both precipitate and supernatant materials were subjected to SDS-PAGE. The presence of GST-cytohesin-1 in the samples was detected by Western blotting. Fig. 8B shows that GST-cytohesin-1 preferentially associates with the detergent-insoluble fraction in vitro, if PKC␦ and ATP are both present in the sample (left panels). However, this particulate association is mostly abrogated in samples that had been preincubated with cytochalasin D, an agent known to inhibit the polymerization of F-actin (39). Furthermore, the same pattern of particulate as-sociation is observed with endogenous cytohesin-1 in phorbol ester-and/or cytochalasin D-treated Jurkat cells (Fig. 8C). As a control, Fig. 8D shows that the addition of nocodazole, an agent known to disrupt microtubules, did not alter the distribution of phosphorylated or unphosphorylated cytohesin-1. Taken together, we conclude that association of cytohesin-1 with the cellular actin cytoskeleton, but not with membrane phospholipids, is regulated through phosphorylation of its polybasic carboxyl terminus. Fig. 8E shows that both phosphorylation sites apparently play an important role in these associations. Whereas single mutagenesis of either S394 or T395 had only partial effects on the association of GST-cytohesin-1 with the cytoskeleton (right panel), double mutants lost any ability to bind to the insoluble material. It is interesting that the "gain of negative charge" mutants (ST394/95DE) also exerted loss of function in this experiment. This result indicates that the role of phosphorylation in the association of cytohesin-1 with the actin cytoskeleton is not likely the result of a simple electrostatic repulsion effect. GST-cytohesin-1 was incubated with cellular lysates from which nuclei and membranes had been depleted and with additional reagents, as indicated. Note that the normally soluble fusion protein (far left lane) associates with the insoluble fraction in the presence of PKC and ATP (lane 4 from the left); however, this association is abrogated when cells were incubated with cytochalasin D (far right lane). C, an experiment similar to that shown in B; however, association of endogenous cytohesin-1 with the insoluble Jurkat cell fraction is shown. Note that PMA stimulation induces association of cytohesin-1 with the insoluble fraction (lane 4 from the left); this association is abrogated in the presence of cytochalasin D. D, the microtubule-disrupting agent nocodazole has no effect on the PKC-dependent association of GST-cytohesin-1 with the insoluble cell fraction. E, analysis of the association of cytohesin-1 phosphorylation site mutants with the insoluble cell fraction. GST fusion proteins were employed; PKC and ATP were present throughout. Note that the double mutation of serine/threonine 394/95 is required to completely abrogate the association of the fusion protein with the insoluble cell fraction. The charge of the substituted amino acid does not play a role.

Regulation of Cell Adhesion to ICAM-1 by Phosphorylation
Site Mutants of Cytohesin-1-We subsequently compared the ability of the cytohesin-1 mutants to induce LFA-1-dependent adhesion to ICAM-1 in a T-cell overexpression system. To this end, all described mutants were expressed in Jurkat cells or in SKW3 cells by recombinant vaccinia viruses, and adhesion of the infected cells to an ICAM-1-Fc chimera was measured as described previously (9). The results of these experiments are shown in Fig. 9, and they corresponded well to all of the data above. Overexpression of cytohesin-1 induced Jurkat adhesion to ICAM-1 5-10 fold (Fig. 9A; data not shown), as described previously, whereas the expression of the PH domain construct or the E157K mutant suppressed basal adhesion significantly. The phosphorylation site mutants had no measurable effect on basal adhesion, but we detected a significant reduction of PMAstimulated adhesion when the mutants were employed, as opposed to wild-type cytohesin-1. In accordance with this finding, pretreatment of cells with cytochalasin D at a concentration that strongly interferes with actin polymerization significantly inhibited the effect of cytohesin-1 on cell adhesion in this system (Fig. 9B). Although it is possible that the effects of cytochalasin D are of a complex nature in this assay, our results hint at a contribution of cytoskeletal association of cytohesin-1 to its regulatory role in ␤ 2 integrin activation. We also tested whether actin-cytoskeletal association and membrane localization of cytohesin-1 are interdependent, which would complicate interpretation. To this end, we determined the subcellular localization of cytohesin-1 or of the 394/5AG mutant in cytochalasin D-treated Jurkat cells. Fig. 10 shows that the actindepolymerizing agent did not change the distribution of cytohesin-1 constructs in these cells, which is consistent with the notion that the cortical association of the protein is primarily phospholipid-mediated. Taken together, our results furthermore suggest that membrane recruitment of the molecule and cytoskeletal association are cooperative functions. DISCUSSION In this study, we show that cytohesin-1 is phosphorylated in vivo after stimulation of Jurkat cells with phorbol ester. We mapped the phosphorylation sites of the protein by use of nondenaturing gel electrophoresis. This technique allowed us to detect both the phosphorylated and the unphosphorylated species, yielding two respective bands of differential electrophoretic mobility that were visualized by immunoblot analysis. Under the nondenaturing electrophoresis conditions employed, the phosphorylated isoform bears an additional negative charge due to the phosphate group, which overcompensates for the gain of mass and therefore allows faster migration. In consequence, the upper band observed corresponds to the unphosphorylated protein, whereas the lower band corresponds to the phosphorylated isoform. This is unlike the sometimes observed slow migration of phosphorylated proteins in SDS gels, which is not due to the addition of negative charge but may rather be attributed to indirect effects (32).
We show that the phorbol ester PMA induces phosphorylation of cytohesin-1 in vivo at both serine 394 and threonine 395. We furthermore observed that cytohesin-1 may be phosphorylated in vitro by the PKC␦ isoform. Both the respective serine and threonine residues are embedded in a favorable consensus motif of PKC-dependent phosphorylation. These findings indicate that cytohesin-1 may be phosphorylated in vivo by a member of the PKC family because phorbol esters are known to stimulate PKC activity (33). However, we cannot exclude the possibility that a kinase located downstream of PKC may phosphorylate cytohesin-1 in vivo.
In the next series of experiments, we attempted to unravel the functional relevance of the observed phosphorylation events. Previous analyses had implicated the carboxyl terminus of cytohesin-1 in membrane recruitment and phospholipid binding (22). It therefore appeared likely that this function might be affected by phosphorylation events. Interestingly, however, extensive analyses in these directions, employing in vitro assays as well as cell-based experimental systems, yielded no clues. Previous studies (34,35) had described an interesting mode of regulation of the homologous protein ARNO (cytohesin-2). Either phosphorylation of a carboxyl-terminal serine residue of ARNO or constitutive addition of negative charge to the ARNO carboxyl terminus by substituting the serine residue with an acidic amino acid (Glu) had resulted in an electrostatic repulsion effect and yielded membrane detachment and abrogation of phospholipid binding (35). It is quite surprising that this type of regulation apparently does not hold true for cytohesin-1, although it contains two phosphorylated residues at that site; that is, electrostatic repulsion should be even more efficient. Recently, a study has been published in which the consequences of ARNO phosphorylation were analyzed in greater detail (36). The authors concluded that although nonspecific binding of ARNO to phosphatidylserine was diminished by the phosphorylation, there was no effect of the modi- fication on ARNO binding to PIP 3 . From these data and from our studies, it may be concluded that nonspecific membrane association of the proteins, but not PIP 3 binding, might be abrogated by the phosphorylation events.
We have recently shown that the GDP/GTP exchange function of cytohesin-1 is involved in cell spreading (25). Our data and data from several other studies implicated the actin cytoskeleton in the spreading phenomenon. We therefore analyzed whether cytohesin-1 was regulated in its ability to associate with F-actin in cells. In marked contrast to our attempts to relate the phosphorylation of cytohesin-1 to membrane and phospholipid binding, we found a very clear correlation of cytohesin-1 phosphorylation and its association with the nucleidepleted, detergent-insoluble cell fraction. Furthermore, pretreatment of the cells with an agent known to prevent the formation of F-actin completely abrogated this association. It therefore appears highly likely that the phosphorylation events mediate direct or indirect association of cytohesin-1 with cellular F-actin. Notably, the gain of negative charge mutants were not capable of mimicking the effects of phosphorylation in vitro and in cells. The electrostatic switch mechanism, which had been proposed for ARNO, appears not to play a role in this system. Intriguingly, an actin cytoskeletal remodeling activity had previously been attributed to ARNO (34).
What is the mechanistic, functional role of the regulated association of phosphorylated cytohesin-1 with the actin cytoskeleton? In previous studies, we have shown that cytohesin-1 interacts specifically with the cytoplasmic domain of CD18, the ␤-subunit of the leukocyte integrin LFA-1 (9,25). Furthermore, phorbol ester is widely known to induce LFA-1mediated cell adhesion to ICAM-1 (38). These results and the observation that cytohesin-1 recruitment to the actin cytoskeleton is enhanced by phorbol ester stimulation suggest that the phosphorylation may aid in targeting cytohesin-1 to a favorable position for its interaction with LFA-1. We employed adhesion assays to approach this question at the functional level. Significantly, we found that the ability of cytohesin-1 to induce adhesion was altered upon introduction of the phosphorylation site mutations. The reducing effect of the mutations was not observed when basal adhesion levels were analyzed. However, there was a clear reduction of maximal adhesion obtained by the synergistic actions of cytohesin-1 overexpression and phorbol ester co-stimulation. This appeared plausible because there is a very good correlation with cytohesin-1 phosphorylation in this situation. Consequently, the effect of the mutants was only observable when PMA was employed.
We do not currently know how association of cytohesin-1 with elements of the actin cytoskeleton regulates ␤ 2 integrindependent adhesion. It is also unclear which elements of the actin cytoskeleton interact with the phosphorylated protein.
The polybasic domain might not necessarily be directly dependent for this interaction, but its phosphorylation could also induce conformational changes of the whole protein, which is then enabled to form novel associations. However, the importance of F-actin remodeling for adhesion regulation has been stressed by a number of recent studies (reviewed in Ref. 39). A model has been proposed that predicts a highly dynamic interaction of the cytoplasmic domains of LFA-1 with the actin cytoskeleton (4). According to this model, ligand binding by LFA-1 is initiated by detachment of the relatively immobile LFA-1 molecule from the cytoskeleton. This, in turn, might enable lateral movement of the protein in the plane of the membrane and might thus favor oligomerization or clustering of the adhesion molecules, which would result in ligand capture. After ligand binding, the association of the receptor with the cytoskeleton could be reinforced, and the molecular assembly could thus be stabilized over longer periods of time. Cytohesin-1 could certainly play a role in these events, either by regulating calpain proteases (likely mediating the initial detachment of the integrins from the F-actin network (41)) or by later stabilizing a clustered state through the association of the ligand-bound receptor with actin fibers. We have considered the possibility that cytohesin-1 might be a direct adaptor between the cytoskeleton and LFA-1 and that this function may be regulated by the observed phosphorylation. However, preliminary binding studies did not reveal phosphorylation-induced regulation of the cytohesin-1-LFA-1 interaction to date. On the other hand, it might well be that the phosphorylation of cytohesin-1 affects LFA-1 function indirectly through regulated association of the integrin with other cytoskeletal linker proteins, which in turn bind LFA-1, such as ␣-actinin (42), talin (43), or filamin (44). Furthermore, a number of recent studies point to a specific involvement of the ARF-GTPase-activating protein p95-APP1, a PKL family member, in the coordination of membrane transport and actin polymerization during cell migration (19,37,40). Thus, there is an emerging theme of functional cross-talk between ARF-and Rho-regulatory proteins in cytoskeletal regulation pertaining to cell adhesion and migra- FIG. 10. Subcellular localization of cytohesin-1 or cytohesin-1 394/5 AG double mutant in cytochalasin D-treated cells. Constructs were expressed as cIg fusion proteins in Jurkat E6 cells, which were incubated with 40 ng/ml phorbol ester (PMA) or 10 g/ml cytochalasin D as indicated. In the presence of cytochalasin D, the cortical F-actin is depolymerized, resulting in a diffuse localization of actin (red panels), whereas cytohesin-1 proteins are still localized to the membrane (green panels). The overlay shows partial colocalization of cytohesin-1 and cortical actin in cells that were not treated with cytochalasin D.
tion. Future biochemical and functional studies employing direct microinjection or transfection of phosphorylated or unphosphorylated protein into cells may aid in further elucidation of the specific role of the phosphorylation event.
Taken together, we provide compelling evidence for a specific and functional phosphorylation of cytohesin-1 at its carboxyl terminus. Surprisingly, we observe different consequences of the phosphorylation event in cytohesin-1 as compared with data shown previously for the homologous protein ARNO. However, these differences may contribute to the functional specificities of the highly similar proteins in physiological contexts.