Introduction
During lymphocyte activation, lipid rafts may function as platforms for the formation of multicomponent transduction complexes. These microdomains are constitutively enriched in proteins involved in the early phases of T-cell receptor (TCR) signalling, such as the Src-family kinases Lck and Fyn, the adapter protein LAT (linker of activation in T cells), phosphoprotein associated with glycosphingolipid-enriched domains (PAG) or Csk-activating protein (Cbp) and Lck-interacting molecule (LIME). The composition of raft-associated proteins changes after T-cell stimulation, suggesting that rafts are dynamic platforms for T-cell signalling (reviewed in ref. 6). Based on these data, and on the use of cholesterol-depleting reagents, a direct role for lipid rafts in controlling TCR triggering was initially proposed7, 8. However, the recent literature indicates that TCR triggering, at least in its initial phases, is independent of cholesterol-based microdomains9, 10, 11. On the other hand, during physiological T-cell stimulation, rafts may be required to organize a complete immunological synapse or for signal transduction associated with T-cell costimulatory molecules5, 10. Furthermore, raft mobilization towards the immunological synapse requires the simultaneous engagement and signalling of TCR and CD28 (Refs 4,5,12). It was proposed that, by reorganizing lipid rafts, CD28 recruits signalling molecules into the immunological synapse, thus explaining how CD28 triggering leads to higher and more stable tyrosine phosphorylation of several substrates4, 5. Moreover, recruitment and clustering of rafts within the immunological synapse may segregate negative and positive players of T-cell activation and protect TCR signalling13, 14, 15.
Lipid raft dynamics, in addition to T-cell activation, depend largely on F-actin rearrangements16. Different reports have identified molecules that participate in tethering lipid rafts to the actin cytoskeleton. These include actin binding proteins (such as ERM (ezrin, radixin, moesin) proteins, talin and vinculin, among others) that have been found associated with lipid microdomains, and the lipid PtdIns(4,5)P2, which is also enriched in rafts (reviewed in ref. 17).
The signalling pathway that connects TCR–CD28 triggering to actin reorganization depends on Vav-1 (a guanine nucleotide exchange factor (GEF) for Rho GTPases), the small Rho GTPase Cdc42 and the Wiskott-Aldrich syndrome protein (WASP)18, 19, 20. It is likely that TCR–CD28 triggering induces lipid-raft mobilization through a signalling pathway involving Vav-1, Cdc42 and WASP5, 21, 22. However, TCR triggering alone, which also induces WASP activation, does not result in lipid-raft accumulation at the immunological synapse5.
WASP induces actin polymerization through the Arp2/3 complex, which is one of the most important actin filament-crosslinking factors in cortical cytoplasm. It has been suggested that the Arp2/3 complex cooperates with filamins to establish cortical actin architecture23. Filamins are large cytoplasmic proteins that crosslink cortical actin into a dynamic three-dimensional structure and, by interacting with proteins of great functional diversity, may represent versatile signalling scaffolds23. FLNa is expressed in T lymphocytes and participates in T-cell activation24. We speculated that this molecule may function in CD28-induced raft dynamics.
FLNa–CD28 interaction was analysed by affinity immunoisolation of CD28-containing membranes using anti-CD28 antibody-coated magnetic beads after nitrogen cavitation of human peripheral blood CD4+ T lymphocytes. In the resting state, neither Vav-1 nor FLNa were coisolated with CD28 beads; CD28 ligation notably induced FLNa and Vav-1 copurification with CD28 (Fig. 1a). To verify that CD28 triggering by the physiological ligand B7.1 induces association of the T-cell costimulatory receptor with the actin-binding protein FLNa, human peripheral blood CD4+ T lymphocytes were incubated with L-cell transfectants expressing the human B7.1 molecule (5-3.1/B7). CD28 triggering resulted in enhanced association of CD28 with FLNa, as demonstrated by coimmunoprecipitation experiments (Fig. 1b).
Figure 1: CD28 associates with FLNa and induces its recruitment into the immunological synapse.
(a) Coimmunoisolation of CD28 and FLNa. Peripheral blood CD4+ T cells were incubated with anti-CD28-coated magnetic beads. After nitrogen cavitation, affinity-isolated membranes and 1/10 of the respective pelleted homogenates were blotted with anti-Vav-1 and anti-FLNa antibodies. An uncropped image of the blot is shown in the Supplementary Information, Fig. S2. (b) Peripheral blood CD4+ T cells were incubated or not with unpulsed 5-3.1/B7 cells for 5 min and CD28 was immunoprecipitated with anti-CD28 antibody. Anti-human c-Rel was used as isotype control for immunoprecipitation (data not shown). Immunoprecipitated proteins were sequentially analysed by western blotting with anti-FLNa and anti-CD28 antibodies. Results are representative of three independent experiments. An uncropped image of the blot is shown in the Supplementary Information, Fig. S2. (c, d) Jurkat (c) or peripheral blood CD4+ T cells (PBT; d) were stimulated as indicated and FLNa localization was analysed by confocal microscopy. The scale bars represent 10 
m. (e) The relative recruitment index (RRI), calculated as described in the Methods, of FLNa, CFP–Epac(
DEP–CD)–YFP (irrelevant cytoplasmic marker) and ezrin–GFP (actin cytoskeleton marker) are compared. Data represent mean 
s.e.m. of thirty cells from three independent experiments. Asterisk indicates P < 0.001 compared with controls. In c and d, more than 70% of T cell-superantigens (SAGs)-pulsed 5-3.1/B7 conjugates showed FLNa RRI >2.
To characterize the FLNa domains responsible for interaction with CD28, we examined the ability of the CD28 intracellular domain (CD28-ICD) to interact with FLNa deletion mutants by one-on-one transformation in a two-hybrid screening. In transformed colonies, interaction between CD28-ICD and FLNa fragments conferred the ability to grow in medium lacking adenosine, histidine, tryptophan and leucine, and blue colonies in a 
-galactosidase assay. Using this approach, we identified repeat 10 (FLNaD10, residues 1158–1246) as the minimum domain that retained interaction with CD28-ICD (see Supplementary Information, Table S1).
T-cell stimulation by APCs induces cytoskeletal rearrangements and accumulation of F-actin and FLNa at the immunological synapse24. We have previously shown that the Jurkat T-cell line is a valuable model for investigating CD28 costimulation under physiological activation conditions (for example, staphylococcus enterotoxin E (SEE)-pulsed APCs, bearing or lacking B7.1)5. We found that FLNa was recruited into the immunological synapse only when T cells were stimulated through TCR–CD3 and CD28 (Fig. 1c). Similar results were obtained with peripheral blood CD4+ resting T cells (Fig. 1d). To exclude the possibility that the observed CD28-induced FLNa recruitment into the immunological synapse was due to accumulation of cytoplasm or increased actin polymerization, we analysed the distribution of an irrelevant cytosolic marker (CFP–Epac(DEP–CD) –YFP) and of ezrin–GFP in Jurkat T cells stimulated by SEE-pulsed APCs (Fig. 1e). CD28 costimulation did not significantly enhance accumulation of cytoplasmic or cytoskeletal markers at the immunological synapse, indicating that FLNa is specifically recruited by CD28 into the immunological synapse.
In resting T lymphocytes, stimulation of T cells with anti-CD3 plus anti-CD28 antibody-coated beads induces recruitment of the ganglioside GM1 to the TCR triggering site4. Several studies confirmed and expanded this initial observation, and suggested that raft recruitment to the immunological synapse occurs only in a subset of T cells that are characterized by high activation stringency, and also requires CD28 signalling (reviewed in ref. 6). In contrast, another study suggested that lipid rafts are randomly distributed during cell stimulation by antibody-coated beads and that the apparent enrichment in GM1 observed at the TCR contact site is the sole consequence of nonspecific membrane ruffling25. Antibody-induced T-cell activation is a highly useful experimental tool, but bears little resemblance to physiological situations. To understand whether and how lipid-based compartmentalization of plasma membrane occurs during T-cell stimulation by APCs, the distribution of endogenous or transfected proteins showing affinity for either rafts or conventional membranes was analysed. Lipid rafts were visualized using either a cyan fluorescent protein carrying consensus sequences for myristoylation plus palmitoylation (MyrPalm–mCFP) or by staining of the Src kinase Lck. Non-raft membranes were visualized through expression of the transmembrane protein LGFPGT46 or CD71 staining. Accumulation of MyrPalm–mCFP and Lck was observed at the immunological synapse of T cells stimulated by B7+APCs, indicating that rafts are mobilized in response to physiological CD28 costimulation and participate in T-cell synapse organization (Fig. 2). Raft recruitment can not be the sole consequence of membrane ruffling as it is highly specific — fluorescent markers of non-raft membranes did not accumulate at the immunological synapse (Fig. 2). These data, together with many previous results, demonstrate that lipid-based membrane asymmetries are generated during physiological processes in lymphocytes, suggesting a pivotal role for lipid membrane microdomains in controlling the immunological synapse formation.
Figure 2: CD28 costimulation induces specific accumulation of lipid rafts at the immunological synapse.
(a–d) Jurkat T cells expressing MyrPalm–mCFP (b), GT46-GFP (c) or not transfected (a, d) were conjugated with SEE-pulsed 5-3.1 (data not shown) or 5-3.1/B7 cells for 15 min. Fixed conjugates were either stained for Lck (a) or CD71 (d) or directly analysed by confocal microscopy (b, c). The scale bars represent 10 m. FITC, mCFP and GFP were colour-coded green. The RRI was calculated as described in the Methods and represent mean s.e.m. of thirty cells from three independent experiments. Asterisk indicates P < 0.001 compared with both GT46–GFP and CD71 RRI.
Clearly, CD28 signalling is required to mobilize lipid rafts towards the immunological synapse during T-cell stimulation, even if the mechanism responsible for costimulation-induced membrane rearrangement is not known. We asked whether the capacity of CD28 to concentrate rafts at the T-cell immunological synapse might depend on its capacity to recruit FLNa. We investigated this hypothesis by decreasing FLNa expression in T cells using siRNA. In Jurkat T cells in which 50% knockdown of FLNa was achieved (Fig. 3a), accumulation at the immunological synapse of lipid raft markers (such as the endogenous Src-kinase Lck or MyrPalm–mCFP) was impaired (Fig. 3b, c), indicating that FLNa is required for raft recruitment into the T-cell immunological synapse. TCR recruitment into the immunological synapse was not affected by FLNa knockdown (data not shown).
Figure 3: FLNa is required for CD28-induced raft recruitment into the immunological synapse.
(a) Jurkat T cells were transfected with FLNa-specific or control siRNAs. After 48 or 72 h, cells were harvested and samples blotted with anti-FLNa antibody. Blotting with the anti-actin antibody resolved in the same gel is shown as control. Densitometric analysis is shown in the graph. Results are representative of at least three independent experiments. An uncropped image of the blot is shown in the Supplementary Information, Fig. S2. (b, c) Jurkat T cells expressing (c) or not (b) the raft marker MyrPalm–mCFP were transfected with FLNa-specific or control siRNAs. After 72 h, cells were conjugated with SEE-pulsed 5-3.1/B7 cells for 15 min. Fixed conjugates were either stained for Lck (b) or directly analysed by confocal microscopy (c). (d) Jurkat T cells expressing the raft marker MyrPalm–mCFP were transfected with Myc-tagged FLNaD10–12. After 12 h, cells were conjugated with SEE-pulsed 5-3.1/B7 cells for 15 min. Fixed conjugates were directly analysed by confocal microscopy. In this figure, FITC was colour-coded green and mCFP red. (e) Localization of endogenous FLNa or ezrin–GFP in Jurkat T cells expressing Myc-tagged FLNaD10–12. The scale bars represent 10 m. The RRI was calculated as described in the Methods and represent mean s.e.m. of thirty cells from three independent experiments. The double asterisk indicates P <0.001 and the single asterisk indicates P <0.05 compared with controls.
To specifically block the interaction between CD28 and FLNa, the FLNa fragment containing repeats 10–12 (FLNaD10–12) was overexpressed. This fragment interacted with CD28 in a two-hybrid screen (see Supplementary Information, Table S1) and thus can function as a dominant-negative mutant by preventing CD28 interaction with endogenous FLNa. Overexpression of FLNaD10–12 in T cells inhibited CD28-mediated raft recruitment into the immunological synapse (Fig. 3d), indicating that CD28-mediated raft accumulation at the immunological synapse depends on the ability of CD28 to interact with FLNa. Overexpression of FLNaD10–12 in T cells inhibited endogenous FLNa recruitment into the immunological synapse, whereas ezrin localization remained unaffected (Fig. 3e).
We have previously shown that mutations in the carboxy-terminal proline motif of CD28 impair raft mobilization during physiological activation of T cells5. Therefore, we asked whether the same CD28 motif is required for CD28–FLNa association. To test this hypothesis, we used a Jurkat cell line (CH7C17) deficient for CD28 expression reconstituted with human wild-type CD28 or with a CD28 mutant in the C-terminal PxxPP (CD283A), generated by substituting the prolines at positions 208, 211 and 212 with alanine5. We found that after stimulation with 5-3.1/B7 cells, CD283A impairs the association with FLNa, indicating that the C-terminal proline motif of CD28 required to mobilize rafts is the same used by CD28 to interact with FLNa (Fig. 4a). Moreover, Jurkat cells expressing the CD283A mutant (J–CD283A) were unable to recruit FLNa into the immunological synapse during stimulation with SEE-pulsed 5-3.1/B7 cells (Fig. 4b). TCR and ezrin recruitment was not impaired in J–CD283A T cells (data not shown).
Figure 4: FLNa recruitment into the immunological synapse requires the CD28 C-terminal PxxPP motif.
(a) The C-terminal proline motif of CD28 is required for CD28–FLNa interaction. Jurkat cells lacking CD28 and reconstituted with either wild-type CD28 (J–CD28) or with the mutant CD283A(J–CD283A) were stimulated or not with unpulsed 5-3.1/B7 cells for 5 min. CD28 was immunoprecipitated with an anti-CD28 antibody and immunoprecipitated proteins were sequentially blotted with anti-FLNa and anti-CD28 antibodies. Results are representative of three independent experiments. An uncropped image of the blot is shown in the Supplementary Information, Fig. S2. (b) J–CD28 or J–CD283A were conjugated with SEE-pulsed 5-3.1/B7 cells for 15 min. Fixed conjugates were stained for FLNa and analysed by confocal microscopy. The scale bars represent 10 m. The RRI was calculated as described in Methods and represent mean s.e.m. of thirty cells out of three independent experiments. Asterisk indicates P < 0.001 compared with controls.
In addition to filamentous actin, filamins bind a variety of signalling molecules. It has been proposed that, by bringing together membrane receptors, the actin cytoskeleton and intracellular signalling molecules, filamins facilitate the activation of local cellular processes23. In particular, FLNa binds to Rho family GTPases and to some of their regulatory cofactors, indicating that FLNa may be crucial in regulating local processes involving actin polymerization26.
In T cells, CD28 activation promotes actin polymerization through the activation of the small Rho GTPase Cdc42 (ref. 27). CD28 triggering induces tyrosine phosphorylation of Vav-1, which, in turn, activates Cdc42 and Rac-1 (ref. 28). To investigate the role of CD28–FLNa association in CD28 signalling, Vav-1 phosphorylation and Cdc42 activation was analysed in J–CD283A cells that were stimulated by 5-3.1/B7 cells. In T cells expressing the CD283A mutant, Vav-1 was phosphorylated in response to B7.1 stimulation at levels comparable to those observed in T cells expressing wild-type CD28 (see Supplementary Information, Fig. S1a) whereas Cdc42 activation was inhibited (see Supplementary Information, Fig. S1b). Activation of Cdc42 in J–CD283A cells stimulated through TCR using SEE-pulsed 5-3.1 cells was unaffected and similar to that observed in wild-type J–CD28 cells (see Supplementary Information, Fig. S1b). Cdc42 activation was unaffected in J–CD283A cells receiving both TCR and CD28 stimulation (data not shown).
Taken together, these data indicate that the C-terminal PxxPP motif of CD28 is dispensable for Vav-1 phosphorylation, but it is required for optimal Cdc42 activation. The observation that the same motif of CD28 is required for CD28–FLNa association, and that FLNa binds Cdc42 and Rac-1, suggests that CD28 uses FLNa to recruit Rho GTPases at the site of Vav-1 activation, thus organizing a complete signalling compartment for actin remodelling. Cdc42 activation by CD28 was analysed in peripheral blood T lymphocytes in which 80% knockdown of FLNa was achieved by siRNA (Fig. 5a). In T cells, FLNa knockdown did not affect CD28, CD4 or TCR-CD3 expression (data not shown). Activation of Cdc42 in FLNa-knockdown T cells stimulated through CD28 was inhibited (Fig. 5b), thus confirming that CD28 uses FLNa to activate Cdc42 in response to physiological stimulation.
Figure 5: FLNa is required for CD28-induced T-cell costimulation.
(a) Peripheral blood CD4+ T cells were transfected with FLNa-specific or control siRNAs. After 24, 48 or 72 h, cells were harvested and samples blotted with anti-FLNa antibody. Blotting with the anti-actin antibody resolved in the same gel is shown as control. Densitometric analysis is shown in the graph. An uncropped image of the blot is shown in the Supplementary Information, Fig. S2. (b) Peripheral blood CD4+ T cells were transfected with FLNa-specific or control siRNAs. After 72 h, cells were stimulated or not with unpulsed 5-3.1/B7 cells (CD28 triggering) for 5 min. GTP–Cdc42 was detected as described in Methods. Densitometric analysis is shown in the graph. (c) Jurkat T cells were transfected with NF-AT luciferase reporter construct together with empty vector (pcDNA) or FLNaD10–12 and then stimulated for 8 h with medium alone (control) or 5-3.1/B7 cells pulsed with SEE. The results are expressed as the mean of relative luciferase units s.d. from triplicate samples. Asterisk indicates P < 0.001 compared with cells expressing FLNaD10–12. (d) Peripheral blood CD4+ T cells were transfected with FLNa-specific or control siRNAs. After 72 h, T cells were mixed with beads coated with anti-CD3 or anti-CD3 plus anti-CD28 antibodies. After 12 h, supernatants were collected and IFN
production measured by ELISA. Results are representative of four experiments. Asterisk indicates P < 0.001 compared with controls.
It has been proposed that CD28 amplifies TCR signalling by recruiting lipid rafts into the immunological synapse29. If FLNa is required for CD28-induced cytoskeletal rearrangements leading to raft mobilization, we reasoned that expression of FLNaD10–12, which prevents CD28 interaction with endogenous FLNa, or FLNa knockdown, would result in defective CD28 costimulation. Expression of FLNaD10–12 in Jurkat cells strongly inhibited NF-AT (nuclear factor of activated T cells) activity induced by TCR plus CD28 signalling, thus indicating that CD28-induced FLNa recruitment into the immunological synapse regulates T-cell costimulation (Fig. 5c). In addition, peripheral blood CD4+ T cells, showing 80% knockdown of FLNa achieved by siRNA (Fig. 5a), were stimulated by beads coated with anti-CD3 antibodies alone or together with anti-CD28 antibody and interferon- production was analysed. As expected, stimulation of resting T cells with anti-CD3 plus anti-CD28 antibodies induced stronger T-cell stimulation as compared to anti-CD3 stimulation alone (Fig. 5d). Interestingly, FLNa knockdown resulted in loss of costimulation and left anti-CD3 response unperturbed (Fig. 5d). These data indicate that when FLNa expression levels are reduced in T cells, CD28-induced costimulation is selectively inhibited, whereas TCR–CD3 signalling remains unaffected.
This study shows that CD28 binds to FLNa and recruits this actin-binding protein into the immunological synapse. The molecular details of the interaction between CD28 and FLNa are still unclear. Our data indicate that FLNa can directly bind CD28, but this association is stabilized on CD28 triggering. We also found that FLNa has a direct role in CD28 signalling by recruiting Cdc42 at the site of Vav-1 activation.
Interestingly, mutations in the CD28 cytoplasmic proline motif resulting in impaired recruitment of lipid rafts into the immunological synapse5, also resulted in loss of CD28–FLNa interaction. FLNa knockdown by siRNA induces defects in CD28-dependent raft mobilization, indicating that CD28 uses FLNa to integrate signalling pathways, resulting in actin crosslinking and lipid raft mobilization. Considering these data, we speculate that FLNa stabilizes and crosslinks actin filaments at the immunological synapse, resulting in accumulation of lipid membrane microdomains. It has been suggested that the links between actin filaments that are formed by the Arp2/3 complex are metastable and dissociate, whereas interactions between FLNa and actin filaments are robust on a longer timescale23.
In the absence of tools to manipulate rafts without altering T-cell physiology, the hypothesis that costimulatory molecules use rafts to amplify TCR signalling and to facilitate T-cell activation, could not be definitively exploited. However, the finding that CD28 requires FLNa to recruit lipid rafts into the immunological synapse allowed us to investigate the role of membrane microdomains during T-cell activation. We found that in T cells in which FLNa expression is reduced, lipid raft mobilization and CD28 costimulation are lost. Importantly, TCR–CD3 responses are not altered in these cells, indicating that FLNa knockdown specifically affects T-cell costimulation.
Our data suggest that during T cell activation by APCs, CD28 coligation with the TCR alters the local environment around the TCR, thereby providing a spatially distinct domain for signal amplification.
Methods
Cell culture.
The Jurkat T cell line J.E6-1 (Jurkat cell clone purchased by the American Type Culture Collection, Manassa, VA) and EBV-B cells were maintained in RPMI 1640 supplemented with 10% FCS, L-glutamine, penicillin and streptomycin (Gibco, Grand Island, NY). The wild-type CD28 and CD283A Jurkat T cell line were obtained as previously described5 and maintained as above, with the addition of 400 g ml-1 hygromycin B, 4 g ml-1 puromycin and 2 mg ml-1 G418 (Sigma, St Louis, MO). The L-cell transfectants expressing HLA-DR (
, 1*0101; 5-3.1) or cotransfected with HLA-DR (, 1*0101) and human B7.1 (5-3.1/B7) were previously described30. Human peripheral blood CD4+ T cells were purified by negative selection by using a RosetteSep kit (StemCell Technologies, Vancouver, Canada).
Antibodies and reagents.
Mouse anti-Lck (3A5), goat anti-CD28 (N20), mouse anti-CD3-
(6B10.2) and mouse anti-human c-Rel (B-6) antibodies were purchased from Santa Cruz Biotechnology (Santa Cruz, CA). Anti-human CD28 (CD28.2) and anti-human CD71 antibodies were purchased from BD Pharmingen (San Diego, CA). Anti-human filamin-A (FLMN01) was purchased from NeoMarkers (Fremont, CA). Anti-human Cdc42 and anti-human Vav-1 were purchased from Upstate Biotechnology (Lake Placid, NY). Anti-mouse HRP-conjugated secondary antibody was purchased from Biorad (Hercules, CA), anti-goat HRP-conjugated secondary antibody was purchased from Santa Cruz. Anti-mouse fluorescein-conjugated secondary antibody was purchased from Calbiochem (San Diego, CA), and anti-mouse rhodamine-conjugated secondary antibody was purchased from Chemicon (Temecula, CA). The immunoblots were analysed by enhanced chemiluminescence (Amersham Biosciences, Uppsala, Sweden). Staphylococcus enterotoxin A (SEA), B (SEB) and E (SEE) were purchased from Toxin Technology (Sarasota, FL).
Yeast two hybrids.
CD28 bait was constructed by amplifying the CD28-ICD (amino acids 180–221) by PCR with specific primers (5'-GCGAATTCAGGAGTAAGAGGAGCAGG-3' and 5'-GCGGATCCTCAGGAGCGATAGGCTGC-3'), and the product subcloned into Gal4-binding domain vector (pGBKT7; Clontech, Mountain View, CA). FLNa D8–9, D10–12, D13–16, D16–20 and D20–24 fragments were amplified by PCR with specific primers (FLNa repeat 10: 5'-GCGAATTCTTTGACG CATCCAAA-3' and 5'-GCGGATCCGCCTTGCTGGGGAAGTT-3'; FLNa D8–9: 5'-GCGAATTCCTGGACCTCAGCAAG-3' and 5'-GCGGATCCGCGGGAACCACGTGGGC-3'; FLNa D10–12: 5'-GCGAATTCTTTGACGCATCCAAA-3' and 5'-GCGGATCCCTTGAAAGGACTGCCTGG-3'; FLNa D13–16: 5'-CGGAATTCACAGATGCGTCCAAG-3' and 5'-GCCTCGAGCTGCAAGGGGCTTCC-3'; FLNa D16–20: 5'-CGGAATTCGCCCCGGAGAGGCCC-3' and 5'-GCGGATTCCCCTAGGGGCCCCAC-3'; FLNa D20–24: 5'-CGGGATCCCGAAAGAGAGCATCACC-3' and 5'-GCCTCGAGTCAGGGCACCACAAC-3') and cloned in the pGADT7 vector in EcoRI–BamHI (D8–9, D10–12, D16–20), EcoRI–;XhoI (D13–16), and BamHI–;XhoI (D20–24), respectively. Yeasts were cotransformed with pGBKT7–CD28-ICD and each of the pGADT7–;FLNa fragments by one-on-one transformation. Positive interactions were verified by colony growth on agar plates lacking adenine, histidine, tryptophan and leucine, and using a -galactosidase assay with X--Gal as a substrate (Clontech).
Plasmids and transfections.
MyrPalm–mCFP in pcDNA3 plasmid, coding for lipid-modified fluorescent proteins, was the kind gift from R. Tsien (Howard Hughes Medical Institute, San Diego, CA). pGFP–GT46 was a gift from K. Simons (Max Planck Institute, Dresden, Germany). FLNa D10–12 was amplified with the primers 5'-GCGAATTCTTTGACGCATCCAAA-3' and 5'-GCGGATCCCTTGAAAGGACTGCCTGG-3', and subcloned in pcDNA3. Ezrin–GFP, in EGFP–C2 plasmid, was a kind gift from S. Mayor (National Centre for Biological Science, Bangalore, India). CFP–Epac(DEP–CD) –YFP, in pcDNA3 plasmid, was a kind gift from K. Jalink (The Netherlands Cancer Institute, Amsterdam, The Netherlands)31. cDNA constructs were transfected by electroporation as previously described5.
RNA interference.
Jurkat or peripheral blood CD4+ T cells were transfected with 5 g of the pooled SMART selection-designed siRNAs (CCAACAAGGUCAAAGUAUAUU; GCAGGAGGCUGGCGAGUAUUU; GUAUGGAGAUGAAGAGGUAUU; UCACAGAAAUUGACCAAGAUU; Dharmacon, Lafayette, CO) for FLNa or with 1.5 g of control siRNAs, using electroporation or Amaxa Nucleofector kit, respectively. Cells were then incubated in complete medium for 24, 48 or 72 h before harvesting, depending on the type of experiment. Both FLNa bands with relative molecular masses of 280,000 and 250,000 (Mr , 280K and 250K) were reduced, but the 280K band was more stable. Recently, p56lck has been described to bind to and phosphorylate FLNa, thus altering its susceptibility to calpain32. Thus, the 280K form of FLNa may be more stable than the 250K, especially in Jurkat cell lines expressing high levels of p56lck.
Immunoisolation of plasma membrane fragments.
M-450 goat anti-mouse magnetic beads (Dynal A.S., Oslo, Norway) were coated with anti-CD28 monoclonal antibody (BD Pharmingen) following manufacturer's instructions. Peripheral blood CD4+ T cells (1.2 
107) were incubated (4 °C, 2 min) with anti-CD28 beads (bead:cell ratio, 1:2) in RPMI with 1% FCS. Bead–cell conjugates were incubated (37 °C) for 0, 3 and 7 min, washed once with ice-cold H buffer (10 mM sodium Hepes at pH 7.2, 250 mM sucrose, 2 mM MgCl2, 10 mM NaF and 1 mM vanadate) and resuspended in 1 ml H buffer containing 1 g BIS(sulfosuccinimidyl) suberate (BS3; Pierce, Rockford, IL), 0.2 mM pervanadate and CLAP protease inhibitor (chymostatin, leupeptin, antipain and pepstatin; 100 M each, Sigma). Cells were nitrogen-cavitated (4 °C, 50 bar, 7 min) using a nitrogen cavitation bomb (Parr Instrument Company, Moline, IL). Beads were retrieved with a magnet (Dynal) and washed three times with 10 ml H buffer (1 min each at 4 °C); homogenates were pelleted by ultracentrifugation (100,000g, 4 °C, 20 min), in an Optima XL-100K ultracentrifuge (Beckman Coulter, Fullerton, CA). Beads and cell pellets were analysed by immunoblotting.
Confocal microscopy.
5-3.1 and 5-3.1/B7 cells were suspended at 107 cells ml-1 and incubated (or not) with 1 g ml-1 of the bacterial superantigen SEE or with a cocktail of superantigens (SEA, SEB, SEE, 1 g ml-1 each) for 2 h at 37 °C, with mixing every 20 min. Pulsed cells were washed and incubated for 15 min at 37 °C with equal amount of T cells. Cells were then adhered to microscope slides coated with 50 g ml-1 poly-L-lysine, fixed with 4% paraformaldehyde and either permeabilized with 0.1% Triton in PBS and stained for filamin-A (scarcely expressed in 5-3.1 cells) or for Lck, CD71 and CD3- (not expressed in 5-3.1 cells) or directly mounted in 2.5% 1,4-diazobicyclo[2.2.2]octane (DABCO, Fluka, St Louis, MO), 90% glycerol, 10% PBS.
Confocal microscopy was performed with a Biorad confocal microscope (BioRad, Hercules, CA) with 60 objective lenses (Nikon, Tokyo, Japan), using laser excitation at 488 nm, 550 nm or 430 nm. Images were analysed with the Adobe Photoshop 7.0 programme. Surface plot analysis was performed with the Image J programme.
Fluorescence quantitation
To quantitate the recruitment of FLNa, TCR, Ezrin–GFP, CFP–Epac(DEP–CD) –YFP and MyrPalm–mCFP to the immunological synapse, boxes were drawn around the immunological synapse, the regions of the T cell not in contact with APCs, and a background area. The relative recruitment index (RRI) was calculated as indicated: (mean fluorescence intensity (MFI) at synapse – background) / (MFI at all the cell membrane not in contact with APCs – background). Quantitative analysis of MFI was performed with the Image J programme. At least thirty conjugates were examined quantitatively for each experiment. Statistical significance was calculated using a Student's t-test.
Densitometry.
Densitometric analyses were performed on a Image Master VDS-CL densitometer using volume analysis of Image Master Total Lab software (Amersham Biosciences, Amersham, UK). All densitometric values obtained were calculated from nonsaturated signals. For siRNA experiments, FLNa expression values were normalized for actin; for cdc42 assay, values were normalized for the controls.
Cdc42-activation assay.
Wild-type J–CD28 or J–CD283A cells (1 107) were stimulated or not with 1 107 SEE-pulsed 5.3-1 or unpulsed 5.3-1/B7 cells, for 5 min at 37 °C. In some experiments, human peripheral blood CD4+ T lymphocytes, transfected with FLNa-specific or control siRNAs, were used. Cells were then lysed in 500 l of MLB buffer (125 mM HEPES at pH 7.5, 750 mM NaCl, 5% Igepal CA-630, 50 mM MgCl2, 5 mM EDTA and 10% glycerol) containing protease and phosphatase inhibitors, and the activation of Cdc42 was determined by affinity precipitation using a GST-fusion-protein based kit, according to the manufacturer's protocol (Rac/cdc42 assay reagent, Upstate Biotechnology). Proteins then were separated by 12% SDS–PAGE, transferred to nitrocellulose membrane, and probed with anti-Cdc42.
Immunoprecipitation.
Wild-type J–CD28 or J–CD283A cells (1 107), or peripheral blood CD4+ T cells (2 107) were stimulated or not with 1 107 5.3-1/B7 cells, for 5 min at 37 °C; cells were then lysed in 1% Nonidet P-40 buffer (20 mM Tris–HCl at pH 7.5, 150 mM NaCl, 1 mM MgCl2, 1 mM EGTA), containing 10 g ml-1 aprotinin, 10 g ml-1 leupeptin, 20 mM NaF, 1 mM Pefabloc-SC, 1 mM Na3VO4 and 10 mM Na4P2O7. Postnuclear lysates were precleared for 1 h at 4 °C with protein G–Sepharose (Amersham Biosciences) and then incubated for 2–3 h with anti-CD28 antibodies pre-adsorbed to protein G. Immunoprecipitates were washed twice in 1% Nonidet P-40, twice in 0.05% Nonidet P-40 lysis buffer, and boiled in SDS buffer before electrophoresis on 8% SDS–polyacrylamide gels. After protein transfer, nitrocellulose membranes were blotted with anti-filamin A and anti-CD28 antibodies. Anti-human c-Rel was used as isotype control for immunoprecipitation.
NF-AT luciferase assay.
J–E6-1 cells (1 107)were transfected with 10 g NF-AT luciferase reporter construct, together with 30 g of empty vector (pcDNA) or FLNaD10–12. After 24 h, 1 105 T cells were stimulated for 8 hours with medium alone (control) or with 5-3.1/B7 cells pulsed with SEE; the cells were then washed in PBS and lysed in 1 lysis buffer (Reporter lysis Buffer 5; Promega, Madison, WI). After 30 min, 20 l of each sample were subjected to the luciferase assay, according to the manufacturer's instructions (Luciferase Assay System, Promega).
ELISA.
After siRNA transfection (72 h), peripheral blood CD4+ T cells were plated at a 1:1 ratio with beads coated with 10 g ml-1 anti-CD3 (OKT3), or with 10 g ml-1 anti-CD3 plus 10 g ml-1 anti-CD28 antibodies (CD28.2). After 24 h, supernatants were collected and IFN production was measured using standard commercially available ELISA kits (Pierce Endogen, Rockford, IL), according to the manufacturer's instructions.
Statistical analysis.
All the data are representative of at least three different experiments. Values are expressed as mean s.e.m. Statistical analysis was performed using the Student's t-test.
Note: Supplementary Information is available on the Nature Cell Biology website.
