The Slp homology domain of synaptotagmin-like proteins 1-4 and Slac2 functions as a novel Rab27A binding domain.

rab27A, which encodes a small GTP-binding protein, was recently identified as a gene in which mutations caused human hemophagocytic syndrome (Griscelli syndrome) and ashen mice, which exhibit defects in melanosome transport as well as in regulated granule exocytosis in cytotoxic T lymphocytes. However, little is known about the molecular mechanism of Rab27A-dependent membrane trafficking or the specific effector molecules of Rab27A. In this study, we discovered that the Slp (synaptotagmin-like protein) homology domain (SHD) of Slp1--3 and Slac2-a/b specifically and directly binds the GTP-bound form of Rab27A both in vitro and in intact cells but not of the other Rabs tested (Rab1, Rab2, Rab3A, Rab4, Rab5A, Rab6A, Rab7, Rab8, Rab9, Rab10, Rab11A, Rab17, Rab18, Rab20, Rab22, Rab23, Rab25, Rab28, and Rab37). Immunocytochemical analysis revealed that Slp2 (or Slp1) colocalized with Rab27A in the melanosomes of melanoma cells. Slp2 and Rab27A were distributed to the periphery of the cells (especially at the dendritic tips) in the wild-type melanoma cells, whereas they accumulated in the perinuclear region in the melanosome transport-defective cells (S91/Cloudman). These results strongly indicated that the SHD of Slp1--3 and Slac2 functions as an in vivo Rab27A binding domain.

The C2 domain is a protein module of ϳ130 amino acids that is often found in various signaling molecules and proteins involved in vesicular trafficking in all eukaryotes (reviewed in Refs. 1 and 2). The prototypical C2 domains (e.g. protein kinase C C2 domain) are known to bind both several Ca 2ϩ ions and phospholipid membranes and to regulate various Ca 2ϩ -dependent signaling processes (2). Recently, however, Ca 2ϩ -independent type C2 domains, which fail to bind Ca 2ϩ ions as a result of amino acid variations in residues critical for Ca 2ϩ binding, have been reported, and some of them function as a protein interaction site (3)(4)(5) or a nuclear localization signal (6,7).
The members of the Slp family are coded by at least four distinct genes in mouse and human and have several alternatively splicing isoforms (15): Slp1/JFC1 (8,16), Slp2-a-d, Slp3a/b (8,15), and Slp4/granuphilin-a (17). The Slp family contains two conserved domains at the N terminus, referred to as Slp homology domain 1 (SHD1) and Slp homology domain 2 (SHD2) (15) (see Fig. 1, A and C). The SHD1 and SHD2 of Slp3-a and Slp4 are separated by a sequence containing two zinc-finger motifs, whereas Slp1 and Slp2-a lack such zinc-finger motifs, and their SHD1 and SHD2 are linked together. The SHD has also been found in other proteins, including Slac2-a (Slp homologue lacking C2 domains-a) (15), suggesting a more general role of the SHD in cellular signaling. However, no research on the function of the SHD domains has ever been attempted. The only available information is that the SHD shows weak homology with Rab3 binding domains (RBD) of rabphilin-3 (18), RIM (19), and Noc2 (20) (see also Fig. 1A).
In this study, we first discovered that the SHD of Slp1, Slp2-a, Slp3-a, and Slac2 is a novel binding domain for the GTP-bound form of Rab27A (21)(22)(23), one of the small GTPbinding proteins that are believed to be essential components of the membrane-trafficking mechanism of eukaryotic cells (reviewed in Ref. 24). Mutation in the rab27A gene causes defects in melanosome transport as well as defects in granule exocytosis in cytotoxic T lymphocytes in human hemophagocytic syndrome (Griscelli syndrome) and ashen mice (22,23,25,26). Immunocytochemical analysis of melanoma cells showed that Slp2 (or Slp1) and Rab27A colocalized in the melanosome of the wild-type and the melanosome transport-defective melanoma cells. Based on our findings, we discuss the possible role of the SHD of Slps and Slac2s in Rab27A-dependent membrane trafficking.
Preparation of Recombinant SHD of Slps or Slac2s and GST Pulldown Assay-cDNAs encoding the SHD or RBD were amplified by conventional PCR and subcloned into the pGEX-4T-3 (or -2T) vector (Amersham Biosciences, Inc.) as described previously (29). GST fusion proteins were expressed and purified on glutathione-Sepharose (Amersham Biosciences, Inc.) by the standard method (37). The protein concentrations were determined with a protein assay kit (Bio-Rad Laboratories) using bovine serum albumin for reference. GST-Slp1-SHD encoded amino acids 1-117 of mouse Slp1, GST-Slp2-a-SHD encoded amino acids 1-87 of mouse Slp2-a, GST-Slp3-a-SHD encoded amino acids 1-152 of mouse Slp3-a, GST-Slp4-SHD encoded amino acids 1-150 of mouse Slp4, GST-Slac2-a-SHD encoded amino acids 1-153 of mouse Slac2-a, GST-Slac2-b-SHD encoded amino acids 1-87 of human Slac2-b, and GST-rabphilin-3-RBD encoded amino acids 1-186 of mouse rabphilin-3. COS-7 cells transiently expressing FLAG-tagged Rab proteins were homogenized in 50 mM HEPES-KOH, pH 7.2, 150 mM NaCl, 0.1 mM phenylmethylsulfonyl fluoride, 10 M leupeptin, and 10 M pepstatin A, and the proteins were solubilized with 1% Triton X-100 at 4°C for 1 h. After centrifugation at 17,360 ϫ g for 10 min at 4°C, each of the cell lysates was appropriately diluted beforehand so that the same amount of Rab proteins would be contained in the reaction mixture (see Fig. 2A). The amount of Rab proteins in the reaction mixtures was confirmed by immunoblotting with horseradish peroxidase (HRP)-conjugated anti-FLAG tag antibody (Sigma Chemical Co.). A 20-l volume of glutathione-Sepharose beads (wet volume) coupled with 1 g of GST fusion proteins (or GST alone) was incubated with 1 ml of COS-7 cell lysate containing Rab proteins for 2 h at 4°C. After washing the beads five times with 1 ml of the buffer 1 (50 mM HEPES-KOH, pH 7.2, 150 mM NaCl, 0.2% Triton X-100, and protease inhibitors), proteins trapped with the beads were analyzed by 12.5% SDS-PAGE, transferred to a polyvinylidene difluoride (PVDF) membrane (Millipore Corp., Bedford, MA), and then immunoblotted with HRP-conjugated anti-FLAG tag antibody and HRP-conjugated anti-GST antibody (Santa Cruz Biotechnology, Inc.) as described previously (27,34).
Direct Interaction between Rab27A and the SHD of the Slp Family or Slac2-The FLAG-Rab27A protein in the cell lysates described above was affinity-purified by incubation with 10 l of anti-FLAG M2 Affinity Gel (Sigma Chemical Co.) for 1 h at 4°C. The beads were then washed three times with 1 ml of buffer 1 containing 1 M NaCl to remove proteins that nonspecifically bound to the beads or Rab27A. After washing twice with 1 ml of the buffer 1, the beads were incubated for 2 h at 4°C with 0.5 g each of GST-Slp1-SHD, -Slp3-a-SHD, or -Slac2-a-SHD in 1 ml of buffer 2 (50 mM HEPES-KOH, pH 7.2, 150 mM NaCl, 1% Triton X-100, and protease inhibitors) containing 0.5 mM GTP␥S. After washing the beads with 1 ml of buffer 1 five times, GST fusion proteins trapped with the beads were analyzed by 12.5% SDS-PAGE, transferred to a PVDF membrane, and immunoblotted with HRP-conjugated anti-GST antibody and HRP-conjugated anti-FLAG tag antibody as described previously (27,34).
Guanine Nucleotide-selective Interaction of Rab27A with the Slp Family or Slac2-The recombinant FLAG-Rab27A was affinity-purified with the anti-FLAG M2 Affinity Gel as described above. A guanine nucleotide-exchange reaction was performed by incubating of the FLAG-Rab27A-bound beads for 1 h at 4°C with buffer 1 containing either 0.5 mM GTP␥S or 1 mM GDP. The beads were then incubated with COS-7 cell lysates containing T7-tagged Slp1, Slp3-a, or Slac2-a in 1 ml of buffer 2 containing either 0.5 mM GTP␥S or 1 mM GDP for 2 h at 4°C. After washing the beads with 1 ml of buffer 1 five times, proteins trapped with the beads were analyzed by 10% SDS-PAGE, transferred to a PVDF membrane, and immunoblotted with HRP-con-jugated anti-T7 tag antibody and HRP-conjugated anti-FLAG tag antibody as described previously (27,34).

RESULTS AND DISCUSSION
Identification of the SHD as a Novel Rab27A Binding Domain-In our previous study, we found a novel protein motif, the Slp homology domain (SHD) in the N-terminal domain of Slp1-4 and Slac2-a (15). A data base search revealed the existence of an additional SHD-containing protein lacking two C2 domains, and we therefore named it Slac2-b (Fig. 1A). The SHD consists of SHD1 and SHD2, which are separated by two zincfinger motifs in some molecules (asterisks in Fig. 1A), and homology search analysis indicated that the SHD of Slp1-4 and Slac2 shows striking homology to the Rab3 binding domain (RBD) of RIM1, RIM2, rabphilin-3, and Noc2 (18 -20) (Fig. 1, B and C). It should be noted that all the SHD2 included a sequence similar to the Rab3-binding site of rabphilin-3 (SGAWFF; # in Fig. 1C) (38). However, because the SHD of the Slp family and Slac2 forms a distinct branch from the RBD of RIM1, RIM2, rabphilin-3, and Noc2 in the phylogenetic tree, we hypothesized that the SHD functions as a certain Rab (other than Rab3) binding domain.
Rab23, Rab25, Rab27A, Rab28, and Rab37), which are known to be present in distinct membrane structures and involved in membrane trafficking (reviewed in Ref. 24). Each of the Rab proteins with a FLAG tag was expressed in COS-7 cells (Fig.  2A), and a GST pull-down assay was performed by using the SHD of Slp1, Slp2-a, Slp3-a, Slp4, Slac2-a, Slac2-b, and the RBD of rabphilin-3 (see "Experimental Procedures" for details). To our surprise, all the SHDs interacted with Rab27A. The SHD of Slp1, Slp2-a, and Slac2-b specifically interacted with Rab27A but not with other Rabs (Fig. 2, C, D, and H). The SHD of Slp3-a and Slac2-a preferentially interacted with Rab27A and weakly with Rab10 (Fig. 2, E and G). The SHD of Slp4 preferentially interacted with Rab8 and Rab27A and weakly with Rab3A (Fig. 2F). By contrast, the RBD of rabphilin-3 interacted with Rab3A, Rab8, and Rab27A with almost similar affinity (Fig. 2I). Under our experimental conditions, none of the Rab proteins were trapped with the beads coupled with GST alone (Fig. 2B), indicating that the interaction of the SHD with Rab27A is not a nonspecific interaction with GST or beads alone. It should be noted that Rab3A, Rab8, and Rab27A are phylogenetically similar and form a small branch on the phylogenetic tree (Fig. 3I).
To further investigate whether the Slp family and Slac2s interact with specific Rab proteins in intact cells, we performed a co-transfection assay in COS-7 cells. Briefly, T7-tagged Slp or T7-Slac2 and FLAG-tagged Rabs were co-transfected into COS-7 cells, and associations of T7-and FLAG-tagged proteins were evaluated by the immunoprecipitation method (27,30). As shown in Fig. 3, T7-Slp1, Slp2-a, Slp3-a, Slac2-a, and Slac2-b specifically co-immunoprecipitated with Rab27A but not with Rab3A, Rab8, or Rab10, whereas T7-Slp4 and T7-rabphilin-3 specifically co-immunoprecipitated with Rab3A, Rab8, and Rab27A, but not with Rab10, indicating that the Slp family and Slac2 interact with specific Rab proteins in intact cells as well. The association of the full-length Slps and Slac2 with Rab in intact cells was more specific than observed in the in vitro GST pull-down assay. For instance, the GST-SHD of Slp3-a and Slac2-a preferentially interacted with Rab27A, and weakly with Rab10 (Fig. 2, E and G), whereas the full-length Slp3-a and Slac2-a specifically interacted with Rab27A, but not with Rab10 (Fig. 3, C and E). This discrepancy was probably attributable to the nonspecific interaction of the degradation product of GST-Slp3-a-SHD and -Slac2-a-SHD with Rab10 (Fig. 2, E  and G, asterisks).
The Slp Family and Slac2 Directly Interact with the GTPbound Form of Rab27A-Next we investigated whether the interaction between Slp or Slac2 and Rab proteins occurs directly to rule out the possibility that additional proteins in the COS-7 cell lysates form a link between Slp (or Slac2) and Rab27A, and purified recombinant proteins (FLAG-Rab27A and GST-SHDs) were used for binding assay (see "Experimental Procedures" for details). Briefly, recombinant FLAG-Rab27A protein coupled with the anti-FLAG antibody-conjugated agarose was incubated with the GST-fused SHD of Slp1, Slp3-a, or Slac2-a, and proteins bound to Rab27A-beads (or beads alone, as a control) were detected by immunoblotting. As expected, the SHD of Slp1, Slp3-a, and Slac2-a interacted with the Rab27A-beads but not with the beads alone (Fig. 4A), thereby demonstrating direct interaction of the Slp family or Slac2 with Rab27A.
The small G protein superfamily, including the Rab proteins, is generally activated in a GTP-bound form and inactivated in a GDP-bound form, and exchange between the GTP/GDP forms is an essential step in expression of the function of small G proteins. A variety of Rab effector molecules (e.g. Rab-GAP,

FIG. 3. Specific interaction of the Slp family and Slac2 with Rab proteins in intact cells.
A-G, pEF-T7-Slps (or -T7-Slac2s) and pEF-FLAG-Rab3A (-Rab8, -Rab10, or -Rab27A) were co-transfected into COS-7 cells. The proteins expressed were solubilized with 1% Triton X-100 and immunoprecipitated with anti-T7 tag antibody-conjugated agarose as described previously (27). Co-immunoprecipitated FLAG-Rabs were first detected by HRP-conjugated anti-FLAG tag antibody (1/10,000 dilution, upper panels in B-H). The same blots were stripped and reprobed with HRP-conjugated anti-T7 tag antibody to ensure loading of the same amounts of T7-Slp (or T7-Slca2) proteins (1/10,000 dilution, lower panels in B-H). Total expressed FLAG-Rab proteins (1/80 volumes of reaction mixtures; input) used for immunoprecipitation are shown in A. Note that the full-length Slp, Slac2-a, or Slac2-b-SHD interact with specific Rab even in the intact cells. The results shown are representative of at least two or three independent experiments. Molecular weight markers (ϫ 10 Ϫ3 ) are shown at the right. I, phylogenetic tree of the Rab proteins. This tree was drawn with the ClustalW program. Note that Rab3A, Rab8, and Rab27A are phylogenetically similar (boxed).

FIG. 4. Direct interaction of the SHD in the Slp family and
Slac2 with GTP-bound form of Rab27A. A, direct interaction of Rab27A and the SHD of Slp1, Slp3-a, or Slac2-a. The FLAG-Rab27A protein expressed in COS-7 cells was affinity-purified as described under "Experimental Procedures." GST-fused SHD (0.5 g) was incubated with the beads for 2 h at 4°C. After washing the beads five times, co-precipitated GST-SHD was detected by immunoblotting with HRPconjugated anti-GST antibody. Note that the GST-Slp and -Slac2 directly bound Rab27A, but not the beads alone. B, guanine nucleotideselective interaction of Rab27A with the Slp family and Slac2. The GTPor GDP-bound form of FLAG-Rab27A coupled with the agarose beads was prepared as described under "Experimental Procedures," and T7-Slp or T7-Slac2 expressed in COS-7 cells was incubated with the beads for 2 h at 4°C. After washing the beads five times, co-precipitated T7-Slp or -Slac2 was detected by immunoblotting with HRP-conjugated anti-T7 antibody. Note that the Slp family and Slac2 preferentially interacted with GTP-bound form of Rab27A beads but not the beads alone. The results shown are representative of at least two or three independent experiments.
Rab-GEF, and Rab-GDI) specific for the GTP or GDP forms of Rab proteins have been reported (24). Because it is important to determine whether the SHD functions as an effector domain for GTP-or GDP-Rab27A to understand the physiological function of Rab27A binding to the SHD, we investigated the guanine nucleotide-selective interaction of Rab27A with the Slp family or Slac2. Recombinant FLAG-Rab27A coupled with agarose beads was first incubated with 0.5 mM GTP␥S or 1 mM GDP and then with T7-Slp1, -Slp3-a, or -Slac2-a in the presence of GTP␥S or GDP. The proteins bound to the beads were analyzed by immunoblotting with HRP-conjugated anti-T7 tag antibody. As shown in Fig. 4B, Slp1, Slp3-a, and Slac2-a preferentially associated with GTP-Rab27A rather than GDP-Rab27A (Fig. 4B). These results support the idea that the Slp family and Slac2 function as a Rab27A effector by directly interacting with the GTP-bound active form of Rab27A via the SHD.
Slp1 and Slp2 Localize in the Melanosome Transport Machinery in Melanoma Cells-Genetic mapping in the coat colormutated mouse and in human disease characterized by an immune deficiency and a partial albinism (Griscelli syndrome) indicated that Rab27A is a key player in melanosome transport (22, 23, 39 -42). To investigate the possible involvement of the Slp family in melanosome transport in concert with Rab27A, immunofluorescence analyses of Slp1 and Slp2 in the B16-F1 melanoma cell line derived from the wild-type mouse (35) were performed using rabbit polyclonal antibodies against the C2B domains of Slp1 and Slp2. These antibodies were highly specific for each antigen and did not cross-react with other Slp proteins (data not shown). Slp1 and Slp2 proteins were highly expressed in the melanoma cells and closely colocalized with Rab27A (Fig.  5, A-C, and data not shown). Note that both Slp2 and Rab27A were distributed to the peripheral region of the wild-type melanoma cells (Fig. 5C, shown in yellow). The localization of Slp2 in melanosomes was further confirmed by co-immunostaining with anti-TRP-1 (a melanosome-resident protein) (Fig. 5E, shown in white). High magnification images revealed that Slp2 and Rab27A were distributed to melanosomes especially at the dendrite tips (Fig. 5, A-E, insets).
Myosin Va, a product of MyoVa (dilute), is known to function in the actin-based transport of melanosomes to the cell periphery, and recent cellular and molecular analyses revealed the synergistic role of Rab27A and myosin Va in the melanosome transport mechanism (39 -43). The S91/Cloudman melanoma cell line was derived from the dilute mouse (36), which shows a defect in melanosome transport as a result of mutation in MyoVa, and the mislocalization of TRP-1 (the perinuclear clustering of melanosomes) was observed in this cell line (Fig. 5I). If Slp1 and Slp2 were an in vivo rab27A effector, localization of Slp1 and Slp2 should be different in the S91/Cloudman melanoma cells from the wild-type cells (Fig. 5, A-E). As expected, Slp2 also accumulated around the nucleus and did not show any peripheral or dendritic localization in contrast to the wildtype mouse-derived B16-F1 cells (Fig. 5, F-J). Similar contrast distribution between wild-type mouse-derived and dilute mouse-derived melanoma cells was observed in Slp1 (data not shown). These observations, together with biochemical evidences, strongly indicate that the SHD of the Slp family is indeed an in vivo Rab27A binding domain and suggest that at least Slp1 and Slp2 are involved in the melanosome transport mechanism probably by functioning as effector molecules of Rab27A.
Possible Roles of the Slp Family and Slac2s in Rab27A-dependent Melanosome Transport-If the Slp family and Slac2 function as Rab27A effector molecules, how do they function in the transport of melanosome? While this manuscript was being prepared, Slac2-a was identified as a melanophilin in which a mutation causes a defect in melanosome transport in leaden mice (44). It is noteworthy that ashen mice carrying a Rab27A mutation and dilute mice carrying a myosin Va mutation showed the same defects in pigment granule transport (i.e., clumping of melanosomes in the perinuclear region), and as a result ashen, leaden, and dilute mice all exhibit a similar lighter coat color (22, 23, 39 -45). In addition, genetic analysis has shown that these three proteins function in the same or overlapping transport pathways (44), although the functional relationships between the three molecules (Rab27A, Slac2-a/ melanophilin, and myosin Va) remain to be clarified. Thus, our finding of a specific interaction between Rab27A and the SHD of Slac2-a (Fig. 2G) should be physiologically relevant and implies that their interaction is an essential step in melanosome transport.
Although recent studies indicated that Rab27A enables myosin Va to capture melanosomes by recruiting myosin Va to melanosomes (40 -42), Rab27A did not directly interact with myosin Va in our preliminary experiment, 2 suggesting that one 2 M. Fukuda and K. Mikoshiba, unpublished data.  C and H). TRP-1, a melanosome-resident protein, was also co-localized with Slp2 and Rab27A (white in E and J, merged images of Slp2, Rab27A, and TRP-1), indicating that Slp2 was present in the melanosome along with Rab27A. In B16-F1 wild-type melanoma cells, Slp2 was distributed in the periphery of cells (A), especially at the dendritic tips (A, inset), whereas in melanosome transport-defective S91/Cloudman cells, Slp2 accumulated in the perinuclear region of the cells. Similar results were observed in Slp1 (data not shown). Scale bars in E and J indicate 25 m. or more additional "linker proteins" must be required for functional interaction between Rab27A and myosin Va. Thus, it is highly possible that Slac2-a and the Slp family function as an unidentified linker: The N-terminal SHD binds Rab27A, and the large C-terminal domain may interact with myosin Va. Indeed, some C2 domains are now known to function as a protein interaction site (3)(4)(5). Another possible function of the Slp family is that it functions as the "anchoring" protein of Rab27A to melanosomes through the SHD domain and the phospholipid binding C2 domains. These two hypotheses can explain why the subcellular localization of Slp2 (or Slp1) and Rab27A was different between wild-type and melanosome transport-defective melanoma cells and why Slp2 (or Slp1) colocalized with Rab27A independent of myosin Va (Fig. 5). Because Slp2 (or Slp1) was localized at the dendritic tip in B16-F1 melanoma cells (Fig. 5, A-E, insets), it is also possible that the Slp family is involved in the melanosome transfer from melanocytes into keratinocytes, although the mechanism of melanosome transfer itself remains unclear. Further work is necessary to determine the specific step or steps of melanosome transport in which the Slp family and Slac2 are involved.
In summary, we have demonstrated by GST pull-down assays that the SHD of the Slp family and Slac2 function as a novel GTP-Rab27A binding domain. The interaction of the Slp family and Slac2 with Rab27A was confirmed in intact cells. Co-localization of Slp2 (or Slp1), Rab27A, and the melanosome was further demonstrated in the peripheral region of the B16-F1 (wild-type) cell line or in the perinuclear region of S91/Cloudman (defect in melanosome transport) cell line. Our discovery should greatly accelerate understanding of the molecular mechanisms of Rab27A-dependent melanosome transport.