AF-6 Controls Integrin-mediated Cell Adhesion by Regulating Rap1 Activation through the Specific Recruitment of Rap1GTP and SPA-1*

In the present study, we showed that SPA-1, a Rap1 GTPase-activating protein (GAP), was bound to a cy-toskeleton-anchoring protein AF-6. SPA-1 and AF-6 were co-immunoprecipitated in the 293T cells transfected with both cDNAs as well as in normal thymocytes. In vitro binding studies using truncated fragments and their mutants suggested that SPA-1 was bound to the PDZ domain of AF-6 via probable internal PDZ ligand motif within the GAP-related domain. The motif was conserved among Rap1 GAPs, and it was shown that rapGAP I was bound to AF-6 comparably with SPA-1. RapV12 was also bound to AF-6 via the N-terminal domain, and SPA-1 and RapV12 were co-immunoprecipi-tated only in the presence of AF-6, indicating that they could be brought into close proximity via AF-6 in cells. Immunostaining analysis revealed that SPA-1 and RapV12 were co-localized with AF-6 at the cell attachment sites. In HeLa cells expressing SPA-1 in a tetracy-cline-regulatory manner, expression of AF-6 inhibited endogenous Rap1GTP and (cid:1)

Rap1 is a member of Ras family GTPases and is suggested to play roles in the regulation of cell proliferation, differentiation, and cell adhesion (reviewed in Ref. 1). Rap1 is activated by a wide variety of extracellular stimuli through different kinds of specific guanine nucleotide exchange factors, which are coupled with various receptors or the second messengers via distinct interaction motifs (1). The amounts and duration of intracellular Rap1GTP, on the other hand, are controlled by specific GTPase-activating proteins (GAPs). 1 So far, two families of Rap1 GAPs are identified including rapGAPs (I and II) and SPA-1 family proteins (SPA-1, E6TP1/SPAR/SPA-Ls), which may have different tissue distribution profiles (2)(3)(4)(5)(6)(7). Whereas all the Rap1 GAPs share a highly conserved domain, called GAP-related domain (GRD), responsible for GAP catalytic activity, they additionally bear unique functional domains. For instance, rapGAP II has a G␣ i -binding domain and is recruited to plasma membrane, inducing Rap1 inactivation there and concomitant increase in the basal Ras/extracellular signal-regulated kinase-signaling in certain cells (7). SPAR bearing actin-binding domains is located at the dendritic spines of neurons in association with actin-cytoskeleton and controls the spine morphology (5). E6TP1 is shown to bind human papilloma virus E6 oncoprotein via the C-terminal region and is targeted for protein degradation (8).
We reported previously (9 -11) that activated Rap1 induced cell adhesion mediated by ␤ 1 as well as ␤ 2 integrins, and SPA-1 could negatively regulate the integrin-mediated cell adhesion via Rap1 GAP activity. Thus, overexpression of SPA-1 in T cells almost completely suppressed the immunological synapse formation with the specific antigen-presenting cells by inhibiting T cell-receptor-induced activation of LFA-1 (12). In the present study, we have attempted to identify the molecules that are associated with SPA-1 in order to understand how the intracellular localization and the functions of SPA-1 are regulated. Here we report that SPA-1 is bound to AF-6, which has been isolated originally as a fusion partner of ALL-1 in human acute myeloid leukemia (13). AF-6 (also called afadin) is an actinbinding multidomain protein and is reported to bind a number of proteins such as Ras family GTPases including Rap1 (14 -16), a tight-junction protein ZO-1 (17), actin-regulatory profilin (16), and subsets of Eph-related receptor protein tyrosine kinases (18). Although these features imply that AF-6 may function as a molecular scaffold integrating the signals related to cell adhesion and cytoskeletal reorganization, its exact functions remain to be seen. We provide evidence that AF-6 binds both SPA-1 and its specific substrate Rap1GTP via distinct domains and can control integrin-mediated cell adhesion by regulating Rap1 activation.

EXPERIMENTAL PROCEDURES
Cells and Antibodies-Thymocytes, splenic T and B cells were obtained from normal BALB/c mice. 293T, HeLa, and Caki-2 (human kidney cancer) cells were maintained in Dulbecco's modified Eagle's medium supplemented with 10% fetal calf serum. HeLa cells stably transfected with SPA-1 cDNA (HeLa/Tet-SPA-1) in a pTRE vector * This work was supported by a grant-in-aid for scientific research from the Ministry of Education, Science, Culture, Sports, and Technology of Japan. 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.
Immunoprecipitation and Immunoblotting-Cells were lysed with lysis buffer (0.5% Triton X-100, 10 mM Tris-HCl, pH 7.6, 150 mM NaCl, protease inhibitor mixture), incubated with specific antibodies overnight at 4°C with gentle rotation, and then precipitated with protein A-Sepharose beads for 30 min at 4°C. After extensive washing, the beads were eluted with SDS sample buffer, boiled, and electrophoresed in SDS-PAGE followed by immunoblotting and ECL detection (Amersham Biosciences). To detect intracellular Rap1GTP, cell lysates (0.8 -1 mg of proteins) were incubated with a GST fusion protein of RalGDS-RBD coupled with glutathione-Sepharose 4B (Amersham Biosciences) on ice for 1 h, washed, and eluted with SDS sample buffer followed by immunoblotting with anti-Rap1 antibody.
Immunostaining-HeLa/Tet-SPA-1 cells were transfected with pEF-BOS AF-6 and selected with puromycin (Sigma) to establish a stable cell line (HeLa/Tet-SPA-1/AF-6). The latter cells were transfected transiently with T7-tagged RapV12 cDNA in a pSR␣ vector by using Effectene Transfection Reagent (Qiagen). These cells were cultured on coverslips in the presence of Dox at an either inductive (0.1 ng/ml) or non-inductive (1 ng/ml) dose for 24 h. The cells were rinsed with Trisbuffered saline, fixed with 3% paraformaldehyde, permeated with 0.5% Triton X-100/Tris-buffered saline, blocked with 2% bovine serum albumin/Tris-buffered saline, and incubated with anti-AF-6 or anti-Rap1 antibody followed by AlexaFluor 546 anti-rabbit IgG (Molecular Probes). After washing, the cells were incubated further with biotinylated anti-SPA-1 or anti-T7 antibody followed by AlexaFluor 488 streptavidin or anti-mouse IgG. Normal rabbit or mouse IgG was used as a control for the corresponding primary antibody. The stained cells were analyzed using a confocal microscopy (Olympus).
Cell Adhesion Assay-96-Well flat-bottom plates were coated with 5 g/ml fibronectin (FN) (Sigma) overnight at 4°C followed by blocking with 3% bovine serum albumin/phosphate-buffered saline for 1 h at 37°C. 293T or HeLa/Tet-SPA-1 cells were treated with trypsin/EDTA, washed, resuspended in serum-free Dulbecco's modified Eagle's medium containing 0.02% bovine serum albumin and 10 mM Hepes, rotated in suspension for 2 h, and plated in triplicate onto FN-coated wells at 0.5 ϫ 10 5 cells per well. After incubation for 30 min at 37°C, nonadherent cells were removed gently by aspiration, and the remaining adherent cells were fixed with 3.7% paraformaldehyde followed by staining with 0.5% crystal violet in 20% methanol. After extensive washing with distilled water, dye was extracted with extract solution (50% ethanol in 50 mM sodium citrate, pH 4.5) and measured using an enzyme-linked immunosorbent assay reader (Molecular Devices). Absorbance in uncoated wells was subtracted from that in the FN-coated wells to show specific adhesion.
Statistics Analysis-Statistics analysis was done by Student's t test.

RESULTS
Association of SPA-1 and AF-6 in Normal Cells-By yeast two-hybrid screening of a mouse spleen cell cDNA library using a full-length SPA-1 as bait, AF-6 was identified as a potential SPA-1-binding protein (data not shown). We therefore examined the association of SPA-1 and AF-6 by transient gene expression in 293T cells, which expressed only marginal SPA-1 and AF-6 if any. In the 293T cells co-transfected with SPA-1 and AF-6 cDNAs, SPA-1 (130 kDa) was co-immunoprecipitated by anti-AF-6, and reciprocally AF-6 (ϳ200 kDa) was co-immunoprecipitated by anti-SPA-1 antibody, although AF-6 tended to be degraded in overexpression system (Fig. 1A). Neither protein was immunoprecipitated by control preimmune IgG. We then investigated whether the association of SPA-1 and AF-6 occurred physiologically in normal cells. As shown in Fig.  1B, thymocytes abundantly expressed both SPA-1 and AF-6 among normal lymphoid cells, and the endogenous SPA-1 and AF-6 were co-immunoprecipitated with anti-AF-6 and anti-SPA-1 antibodies, respectively, indicating that SPA-1 and AF-6 were associated physiologically in normal cells. It was estimated that around 40% of SPA-1 and 15% of AF-6 were associated with AF-6 and SPA-1, respectively, in normal thymocytes.
Involvement of GRD of SPA-1 and PDZ Domain of AF-6 for Binding-To identify the domains involved in the association of SPA-1 to AF-6, binding of truncated SPA-1 fragments to AF-6 was examined in vitro. Among the fragments, a GRD fragment of SPA-1 (fragment 2, residues 212-532) was bound to the AF-6-coated beads, whereas none of the other fragments were bound ( Fig. 2A). Although not shown, fragment 2 was not bound to the control beads. Similar analysis was performed using truncated fragments of AF-6. As also indicated in Fig. 2A, only a fragment containing the PDZ domain of AF-6 (fragment 3, residues 914 -1129) was bound specifically to the SPA-1coated beads. To confirm these results, we generated a GRDdeletion mutant of SPA-1 (⌬GRD SPA-1) and a PDZ-deletion mutant of AF-6 (⌬PDZ AF-6) by IVTT. As shown in Fig. 2B, ⌬GRD SPA-1 and ⌬PDZ AF-6 failed to be bound to the beads coated with AF-6 and SPA-1 respectively, whereas full-length proteins were bound to the partner proteins. These results strongly suggested that SPA-1 and AF-6 were associated via interaction between GRD of SPA-1 and PDZ domain of AF-6.
Binding of SPA-1 to PDZ Domain of AF-6 via Probable Internal PDZ Ligand Motif, a Common Property of Rap1 GAPs-We next intended to investigate whether the binding of SPA-1 and AF-6 occurred by PDZ-mediated protein interaction. By further truncation of a SPA-1 GRD fragment (fragment 2), it was shown that smaller fragments, G1 (residues 338 -532) and G2 (residues 398 -459), were bound to AF-6-coated beads, whereas another overlapping fragment G3 (residues 435-489) was not (Fig. 3B), suggesting the binding site was in the region between the residues 398 and 434. By using peptide libraries, it was reported that the AF-6 PDZ preferred a class 2 ligand motif (-X-, where is a hydrophobic residue), and a hydrophobic residue was also preferred at Ϫ1 position (20). In the G2 fragment, it was noticed that a stretch of residues at 432-434, IVF, fitted the predicted PDZ ligand motif, which was in a probable ␤-sheet immediately followed by a turn and a ␤-sheet as predicted by Chou-Fasman secondary structure prediction (Fig. 3A). To investigate possible involvement of this motif, we generated mutant proteins of G2, M1 (V433A), and M2 (F434A), by IVTT, and examined their binding to AF-6-coated beads. As shown in Fig. 3B, M1 was bound barely to AF-6-coated beads and M2 with markedly reduced efficiency as compared with G1, suggesting that both Val-433 and Phe-434 were required for the binding to AF-6 PDZ domain. Among known Rap1 GAPs, the motif was conserved (IVF for SPA-1 and E6TP1 and VVF for rapGAP), and a fragment of rapGAP I (residues 263-322) corresponding to G2 of SPA-1 was bound specifically to the AF-6-coated beads comparably (Fig. 3B). Furthermore, as shown in Fig. 3C, rapGAP I and AF-6 were co-immunoprecipitated in the 293T cells co-transfected with full-length rapGAP I and AF-6 cDNAs as well as in a renal cancer cell line endogenously expressing both rapGAP I and AF-6, suggesting that the binding to AF-6 was a shared feature of Rap1 GAPs.
Co-localization of SPA-1 and Rap1 with AF-6 at Cell Attachment Sites-We next examined the intracellular localization of SPA-1 in relation to AF-6 by using HeLa/Tet-SPA-1 cells and those stably transfected with AF-6 (HeLa/Tet-SPA-1/AF-6), in which SPA-1 expression was repressed at the undetectable level in the presence of 1.0 ng/ml Dox while induced strongly in the presence of 0.1 ng/ml Dox within 24 h. In HeLa/Tet-SPA-1 cells cultured with 0.1 ng/ml Dox, SPA-1 was expressed diffusely at the cortical area as well as in the cytosol with little expression of AF-6 ( Fig. 4A). On the other hand, AF-6 was localized predominantly at the cell attachment sites with undetectable SPA-1 expression in the HeLa/Tet-SPA-1/AF-6 cells cultured with 1.0 ng/ml Dox (Fig. 4B). In HeLa/Tet-SPA-1/AF-6 cells additionally transfected with T7-tagged RapV12 and cultured with 1.0 ng/ml Dox, a significant proportion of RapV12 was co-localized with AF-6 ( Fig. 4C). Strong nuclear staining by anti-T7 antibody was nonspecific, because it was detected in untransfected cells as well. When both SPA-1 and AF-6 were induced to express in HeLa/Tet-SPA-1/AF-6 cells in the presence of 0.1 ng/ml Dox, on the other hand, SPA-1 was co-localized with AF-6 at the cell attachment regions (Fig. 4D). It was noted that those cells expressing both SPA-1 and AF-6 tended to become slender, which was followed by the cell detachment from a dish in 2 days, much earlier than HeLa/Tet-SPA-1 cells . Lysate of 293T cells transfected with AF-6 or SPA-1 cDNA was immunoprecipitated with specific antibody and protein A-Sepharose beads. Each IVTT fragment was incubated with the AF-6-or SPA-1coated Sepharose beads, and the specific binding was determined by a pull-down assay followed by autoradiography (right panel of each). Although not shown, none of the fragments were precipitated by the control beads incubated with the cell lysates and preimmune IgG. The experiments were repeated twice with identical results. B, left, wild type (wt) and ⌬GRD SPA-1 proteins generated by IVTT (left panel) were incubated with AF-6-coated beads, and the washed beads were eluted with SDS sample buffer followed by autoradiography (right panel). Right, wild type and ⌬PDZ AF-6 proteins generated by IVTT (left panel) were incubated with SPA-1-coated beads, and the washed beads were eluted with SDS sample buffer followed by autoradiography (right panel).
(see below). As shown in Fig. 4E, SPA-1 was co-localized also with endogenous Rap1 under this condition. These results strongly suggested that both SPA-1 and Rap1GTP were recruited to the cell attachment sites by AF-6.
AF-6 Induces the Association of SPA-1 and Rap1GTP and Enhances Rap1 Inactivation-A possibility that AF-6 recruited both SPA-1 and its substrate Rap1GTP was investigated more directly. 293T cells were transfected with SPA-1 and T7-tagged RapV12 with or without AF-6 cDNA, and the lysate was immunoprecipitated by anti-T7 antibody. In the absence of AF-6, SPA-1 was co-immunoprecipitated barely with RapV12 (Fig. 5A), conforming to a general consensus that the interaction of GAPs with the substrate was only transient (21). In the presence of AF-6, however, a signifi-cant proportion of SPA-1 was co-immunoprecipitated with RapV12 along with AF-6 ( Fig. 4A). On the other hand, expression of ⌬RBD AF-6, which failed to be co-immunoprecipitated with RapV12, hardly induced the co-immunoprecipitation of SPA-1 with RapV12 either (Fig. 4A). The results indicated that AF-6 could bind both SPA-1 and Rap1GTP simultaneously via distinct domains, leading to the efficient association of SPA-1 and Rap1GTP. We then examined the effect of AF-6 expression on the efficiency of Rap1 inactivation by SPA-1 in the cells. Expression of SPA-1 in 293T cells suppressed the amount of endogenous Rap1GTP generated by the transfection of C3G-F cDNA in a dose-dependent manner as reported previously (9). Additional expression of AF-6 significantly enhanced the Rap1 inactivation by SPA-1 (Fig.  4B). Thus, transfection of 0.3 g of SPA-1 cDNA resulted in almost complete inhibition of Rap1GTP generation in the presence of AF-6, whereas 40% of Rap1GTP still remained in the absence of AF-6 ( Fig. 4B). As also shown in Fig. 4B, the enhancing effect of AF-6 on the Rap1 inactivation by SPA-1 was reduced markedly by the deletion of the PDZ domain.
AF-6 Enhances the Inhibitory Effect of SPA-1 on ␤ 1 -Integrinmediated Cell Adhesion-We finally investigated functional effects of the association of SPA-1 and AF-6 on cell adhesion. 293T cells in suspension adhered to fibronectin (FN)-coated plates in a manner dependent on VLA4 andVLA5 (Fig. 6A). Although expression of SPA-1 in 293T cells significantly inhibited the cell adhesion as reported before (9), that of AF-6, ⌬RBD AF-6, or ⌬PDZ AF-6 alone hardly affected the cell adhesion (Fig. 6A). However, co-transfection of AF-6 with SPA-1 cDNA resulted in the greater inhibition of cell adhesion than that of SPA-1 alone (Fig. 6A). Neither ⌬RBD AF-6 nor ⌬PDZ AF-6 FIG. 3. Probable involvement of internal PDZ ligand motifs within GRD in the binding of SPA-1 to AF-6. A, schematic illustration of G1, G2, and G3 fragments of SPA-1 GRD, and amino acid sequence of G3 fragment. Aligned sequences of rapGAP and E6TP1 are also shown. A potential ligand motif for AF-6 PDZ is underlined, and mutated residues (V for M1 fragment and F for M2 fragment) are indicated in boldface. A ␤-finger-like structure predicted by Chou-Fasman secondary structure prediction is also indicated. B, IVTT products of G1, G2, G3, M1 (V432A), and M2 (F433A) of SPA-1 and a fragment of rapGAP corresponding to G2 were incubated with AF-6-coated or control (IgG) beads, and the binding were analyzed as in Fig. 2A using Tricine-buffered SDS-PAGE. Input of each IVTT product is shown in the left panel of each group. C, left, 293T cells were transfected with AF-6 with or without rapGAP I cDNA (1 g), and the lysates were immunoprecipitated with anti-AF-6 followed by immunoblotting with anti-rapGAP. Expression of transfected cDNAs was confirmed by straight immunoblotting. Right, Caki-2 cells were lysed, immunoprecipitated with anti-AF-6, anti-rapGAP, or control IgG, and immunoblotted with anti-AF-6 antibody. These experiments were repeated twice with reproducible results. affected the SPA-1-induced inhibition of cell adhesion, suggesting that the effect of AF-6 was dependent on the association with SPA-1 and Rap1. Similar experiments were performed using HeLa/Tet-SPA-1 cells. HeLa/Tet-SPA-1 cells were transfected with a control vector, AF-6, or ⌬PDZ AF-6 cDNA and cultured for 1 day in the presence of Dox at 1.0 or 0.1 ng/ml. At 1.0 ng/ml Dox, SPA-1 expression was almost completely repressed (Fig. 6B, left). Under this condition, expression of AF-6 or ⌬PDZ AF-6 induced a slight increase in Rap1GTP, and the cell adhesion tended to be enhanced marginally although with statistical insignificance (Fig. 6B, right). When the cells were cultured in the presence of 0.1 ng/ml Dox, on the other hand, SPA-1 expression was induced, and concomitantly Rap1GTP was reduced, leading to the significant decrease in the cell adhesion (Fig. 6B). Expression of AF-6 under this condition resulted in the much greater decrease in both cell adhesion and Rap1GTP level, whereas that of ⌬PDZ AF-6 was without effect at all (Fig. 6B). These results suggested that AF-6 could contribute to the inhibition of ␤ 1 integrin-mediated cell adhesion by enhancing the efficiency of SPA-1-mediated Rap1GTP inactivation.

DISCUSSION
In the present study, we demonstrated that a Rap1 GTPaseactivating protein SPA-1 was bound to AF-6. SPA-1 was coimmunoprecipitated specifically with AF-6 and vice versa not only in the 293T cells co-transfected with SPA-1 and AF-6 cDNAs but also in normal thymocytes, indicating that the association was physiological. In vitro binding studies revealed that the binding was mediated by the interaction between GRD of SPA-1 and the PDZ domain of AF-6. Most known PDZmediated protein interactions occur through recognition of short C-terminal PDZ ligand motifs of partner proteins (22,23). However, exceptional interactions of PDZ domain with internal motifs of partner proteins have been also reported (24 -27). It was reported that neuronal nitric-oxide synthase was bound to a PDZ domain of syntrophin via the internal region, in which a PDZ ligand motif in a sharp ␤-finger structure of neuronal nitric-oxide synthase mimicked a C-terminal ligand motif as a pseudo-peptide (27). It was reported that a class 2 ligand motif (-X-) with hydrophobic residues also at the Ϫ1 position was preferred by AF-6 PDZ domain by using peptide libraries (20). SPA-1 GRD contained a stretch of residues IVF (residues 432-434), which fitted the predicted ligand motif for AF-6 PDZ, and the mutations of Val-433 to Ala or Phe-434 to Ala markedly reduced the binding of a minimal fragment of SPA-1 (G2) to AF-6 in vitro. Marginally retained binding activity of a F434A mutant fragment may be due to the residual activity of alanine as an anchoring residue. Chou-Fasman secondary structure prediction suggested that the motif was located in a ␤-fingerlike structure. These results suggested, but did not prove, that the interaction of internal PDZ ligand motif in SPA-1 GRD with the PDZ domain of AF-6 mediated the specific binding of SPA-1 to AF-6. The motif was conserved in rapGAP and E6TP1, and present results indicated that rapGAP was bound to AF-6 comparably with SPA-1, suggesting that the binding to AF-6 was a shared feature of Rap1 GAPs. On the other hand, mutations of RKR at positions 421-423 to LIG, which attenuated the GAP catalytic activity in vivo, did not affect the binding of SPA-1 to AF-6 at all. 2 In polarized epithelial cells, it was reported that AF-6 was localized at the cadherin-based cell-cell adhesion sites such as tight and adherens junctions, in which association with a tight junction protein ZO-1 might play a role (19). The present results indicated that AF-6 expressed in non-polarized HeLa cells was located focally at the cell attachment sites to the matrix. On the other hand, SPA-1 expressed in HeLa/Tet-SPA-1 cells, which barely expressed endogenous AF-6, was detected diffusely at the cortical area as well as in the cytosol. In the HeLa/Tet-SPA-1 cells transfected with AF-6, however, SPA-1 was co-localized significantly with AF-6 at the cell adhesion sites. It was shown also that RapV12 was co-localized with AF-6 irrespective of the presence of SPA-1, conforming to the previous report (16) that RapV12 was bound to AF-6. The present results further indicated that SPA-1 and Rap1V12 could be co-immunoprecipitated significantly in the presence of AF-6 but not of ⌬RBD AF-6. Collectively these results strongly suggested that both SPA-1 and Rap1GTP (RapV12) could be recruited to AF-6 via independent binding sites, a central PDZ domain and a N-terminal RBD, respectively.
It was reported that Ras GAP activity of a catalytic fragment of p120 RasGAP was inhibited in vitro by an RBD fragment of AF-6 due to competitive binding of the two fragments to overlapping sites of RasGTP (15). Independent binding of SPA-1 and Rap1GTP to the distinct domains of AF-6 may prevent such a possible interference with the GAP catalytic activity of SPA-1 by AF-6. Rather, it was shown that the GAP activity of SPA-1 in vivo was enhanced significantly in the presence of AF-6, indicating that AF-6 recruited SPA-1 and its substrate Rap1GTP into close proximity in the cells and facilitated the efficient catalytic interaction between them. We reported previously (9 -11) that Rap1GTP activated ␤ 1 and ␤ 2 integrins and induced the integrin-mediated cell-cell or cell-matrix adhesion, whereas SPA-1 could negatively regulate the cell adhesion by inactivating Rap1. The present results using HeLa/Tet-SPA-1 cells indicated that expression of AF-6, but not ⌬PDZ AF-6, induced the decrease in Rap1GTP levels and significant inhibition of ␤ 1 integrin-mediated cell adhesion to FN in an SPA-1-induced condition, whereas AF-6 did not affect either of them in an SPA-1-repressed condition. These results have suggested strongly that AF-6 plays a role in the control of integrinmediated cell adhesion by enhancing the efficiency of Rap1 inactivation by SPA-1 at the cell adhesion sites.
Integrin-mediated cell adhesions are regulated dynamically during various cellular functions. In the immune system, for instance, leukocyte movement and migration involve a spatiotemporally organized regulation of cell adhesion (28,29). Also, T cell activation in immune responses depends on the integrinmediated intimate cell-cell interactions between T cells and antigen-presenting cells called immunological synapse, the extent and duration of which profoundly affect the modes and intensity of immune responses (30). We reported previously (12) that Rap1 activation and its regulation by SPA-1 played crucial roles in the immunological synapse formation through LFA-1/ICAM-1 interaction. The present results indicated that AF-6 was expressed abundantly in the normal thymocytes partly in association with SPA-1, and our results 3 using SPA-1 transgenic mice showed that Rap1 activation was essential for the differentiation and expansion of thymocytes in vivo. It thus seems possible that AF-6 plays a significant role in the fine control of integrin-mediated cellular functions via regulation of Rap1 activation in the immune as well as other systems.