Involvement of Rabphilin3 in Endocytosis through Interaction with Rabaptin5*

Rabphilin3 and rabaptin5 are downstream target molecules of the Rab3 and -5 subfamily small G proteins that are implicated in exocytosis and endocytosis, respectively. We examined here the physical and functional relationship between the Rab3-rabphilin3 and Rab5-rabaptin5 systems. Rabphilin3 interacted with rabaptin5 at the N-terminal region (amino acids 1–280), which GTP-Rab3A interacted with. The interaction of rabphilin3 with rabaptin5 was inhibited by guanosine 5′-(3-O-thio)triphosphate-Rab3A. Overexpression of the N-terminal fragment of rabphilin3 (amino acids 1–280) inhibited the receptor-mediated endocytosis of transferrin, and this inhibition was overcome by co-transfection with a dominant active mutant of Rab3A or rabaptin5 in PC12 and HeLa cells. These results suggest that rabphilin3, free of GTP-Rab3A, regulates endocytosis through interaction with rabaptin5 after rabphilin3 complexed with GTP-Rab3A regulates exocytosis.

ecules of the Rab3 and -5 subfamily small G proteins that are implicated in exocytosis and endocytosis, respectively. We examined here the physical and functional relationship between the Rab3-rabphilin3 and Rab5-rabaptin5 systems. Rabphilin3 interacted with rabap-tin5 at the N-terminal region (amino acids 1-280), which GTP-Rab3A interacted with. The interaction of rabphi-lin3 with rabaptin5 was inhibited by guanosine 5-(3-Othio)triphosphate-Rab3A. Overexpression of the N-terminal fragment of rabphilin3 (amino acids 1-280) inhibited the receptor-mediated endocytosis of transferrin, and this inhibition was overcome by co-transfection with a dominant active mutant of Rab3A or rabap-tin5 in PC12 and HeLa cells. These results suggest that rabphilin3, free of GTP-Rab3A, regulates endocytosis through interaction with rabaptin5 after rabphilin3 complexed with GTP-Rab3A regulates exocytosis.
Endocytosis is often coupled with exocytosis, particularly at nerve terminals (for reviews, see Refs. 1 and 2). Molecular mechanisms of exocytosis and endocytosis have extensively been investigated, but the coupling mechanism between these two events is unknown. The Rab family small G proteins are implicated in intracellular vesicle trafficking, such as exocytosis and endocytosis (for reviews, see Refs. [3][4][5]. The Rab3 subfamily consists of four members, Rab3A, -B, -C, and -D, and Rab3A and -C are implicated in Ca 2ϩ -dependent exocytosis, particularly in neurotransmitter release from nerve terminals (for a review, see Ref. 6). A current model for the mode of action of Rab3A and -C in neurotransmitter release is as follows: GDP-Rab3 complexed with Rab GDI 1 stays in the presynaptic cytosol and is converted to GTP-Rab3 by the action of Rab3 GEP. GTP-Rab3 interacts with rabphilin3 associated with synaptic vesicles. This interaction causes the docking of the vesicles with the presynaptic plasma membrane by an unknown mechanism. Ca 2ϩ influx into the cytosol causes the fusion of the vesicles with the presynaptic plasma membrane, finally resulting in neurotransmitter release. In one of these processes, GTP-Rab3 is converted to GDP-Rab3 by the action of Rab3 GAP. GDP-Rab3 is complexed with Rab GDI and returns to the cytosol. In this Rab3 system, we have isolated and characterized Rab3A, -B, -C, Rab GDI, rabphilin3, Rab3 GEP, and Rab3 GAP (7)(8)(9)(10)(11)(12)(13)(14). The Rab5 subfamily consists of three members, Rab5A, -B, and -C, and is implicated in endocytosis (3,(15)(16)(17)(18). A model for the mode of action of Rab5 in endocytosis is as follows: GDP-Rab5 complexed with Rab GDI stays in the cytosol and is converted to GTP-Rab5 by the action of Rab5 GEP after its association with early endosomes. GTP-Rab5 interacts with rabaptin5 and recruits it to early endosomes. This interaction causes both the fusion of the plasma membrane-derived endocytotic vesicles with early endosomes and the lateral fusion of early endosomes. In one of these processes, GTP-Rab5 is converted to GDP-Rab5 by the action of Rab5 GAP. GDP-Rab5 is complexed with Rab GDI and returns to the cytosol. In this Rab5 system, Rab5, rabaptin5, and Rab5 GAP have been isolated and characterized (15,18,19), but Rab5 GEP has not yet been isolated or characterized.
In the Rab3-regulated exocytosis, it still remains unknown how the interaction of Rab3 with rabphilin3 causes the docking of the vesicles with the presynaptic plasma membrane. To clarify this mechanism, we have attempted to isolate molecules interacting with rabphilin3. Rabphilin3 consists of at least two functionally different domains: the N-terminal C1-like domain responsible for interaction with GTP-Rab3, and the C-terminal two C2-like domains responsible for interaction with Ca 2ϩ and phospholipid (20,21). We have found that ␤-adducin, known to be involved in the assembly of spectrin-actin complexes (for a review, see Ref. 22), interacts with the C-terminal region of rabphilin3 in the presence of Ca 2ϩ and phospholipid (23,24) and that ␣-actinin, known to cross-link actin filaments into a bundle (for reviews, see Refs. 25 and 26), interacts with the N-terminal region of rabphilin3 in a manner competitive with GTP-Rab3 (27). Here, we have found that rabphilin3 interacts with rabaptin5 at the N-terminal region and that this interaction is inhibited by GTP-Rab3A. We have furthermore shown that rabphilin3 is involved in endocytosis.

EXPERIMENTAL PROCEDURES
Materials and Chemicals-Bovine Rab3A and rabphilin3 were purified from the membrane fraction of Spodoptera frugiperda cells (Sf9 cells) overexpressing each protein (28,29). GDP-Rab3A and GTP␥S-Rab3A were prepared as described (30). Truncated rat rabaptin5, which corresponds to an amino acid position from 385 to 753 of human rabap-tin5 (GenBank accession number X91141), was expressed as a GSTfusion protein and purified from overexpressing Escherichia coli. Anti-HA and anti-myc mouse monoclonal antibodies and an anti-rabphilin3 rabbit polyclonal antibody were prepared as described (31)(32)(33).
Construction of Expression Vectors-Mammalian expression plasmids, pEFBOS-HA and pEFBOS-myc, were generated to express fusion proteins with the N-terminal HA and myc epitopes, respectively (33, * This work was supported by grants-in-aid for Scientific Research and for Cancer Research from the Ministry of Education, Science, Sports, and Culture, Japan (1996,1997), by grants-in-aid for Abnormalities in Hormone Receptor Mechanisms and for Aging and Health from the Ministry of Health and Welfare, Japan (1996,1997), and by grants from the Human Frontier Science Program (1996, 1997) and the Uehara Memorial Foundation (1996). 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.
‡ Present address: The Third Department of Internal Medicine, Faculty of Medicine, Kyushu University, Fukuoka 812-82, Japan.
Screening for Rabphilin3-interacting Molecules by the 2-hybrid Method-A strain L40 was transformed with a derivative of pBTM116 bearing the N-terminal fragment (1-280 aa) of rabphilin3 fused to the LexA DNA-binding domain: pBTM116-rabphilin3-N. A strain L40 carrying pBTM116-rabphilin3-N was transformed with the library DNA (MATCHMAKER rat brain oligo(dT)-primed library in pGAD10, CLONTECH). Approximately 1.1 ϫ 10 6 transformants were screened for the growth on SD plate media lacking tryptophan, leucine, and histidine, but containing 0.5 mM 3-amino-1,2,4-triazole, which is a specific inhibitor of the HIS3 gene product. His ϩ colonies were then placed on the nitrocellulose filter and stained with 5-bromo-4-chloro-3indolyl-␤-D-galactopyranoside for ␤-galactosidase activity as described (38). From the 50 His ϩ and lacZ ϩ positive clones obtained with this screening, library plasmids were recovered through E. coli transformation. The recovered plasmids were transformed again into L40 containing pBTM116-rabphilin3-N, and it was found that 14 clones conferred the His ϩ and lacZ ϩ phenotypes on L40 containing pBTM116-rabphi-lin3-N. The nucleotide sequences of the insert DNA of these 14 clones were determined.
Assay for Rabphilin3-Rabaptin5 Interactions in an Intact Cell System-PC12 cells were plated at a density of 1.5 ϫ 10 6 cells/35-mm dish and then incubated for 18 h at 37°C. The cells were infected with T7 RNA polymerase recombinant vaccinia virus (LO-T7) for 30 min and then transfected with 2 g of pGEM-HA-rabaptin5, 2 g of pGEM-HA-rabphilin3, or both the plasmids, by use of LipofectAMINE reagent (Life Technologies, Inc.) (35). Immunoprecipitation was performed at 4 h after the transfection. The cells were washed with phosphate-buffered saline and lysed in 0.5 ml per dish of a lysis buffer containing 10 mM Tris/HCl at pH 7.5, 150 mM NaCl, 1.5% Nonidet P-40, 91 M CaCl 2 , and 1 mM EGTA. The cell lysate was subjected to immunoprecipitation with 10 pmol of the anti-rabphilin3 antibody bound to 25 l of protein A-Sepharose (Pharmacia Biotech Inc.). Comparable amounts of the pellets were subjected to SDS-PAGE, followed by immunoblot analysis with the anti-HA antibody.
Assay for Rabphilin3-rabaptin5 Interactions in a Cell-free System-Full-length rabphilin3 (50 pmol) was incubated with GST or GST-rabaptin5 (100 pmol) bound to glutathione-Sepharose in a buffer containing 20 mM Tris/HCl at pH 7.5, 1 mM dithiothreitol, 5 mM MgCl 2 , 0.5% Nonidet P-40, and 20 mM NaCl for 90 min at 4°C in the presence or absence of 0.4 M GDP-Rab3A or GTP␥S-Rab3A. After the beads were washed four times with the same buffer, the bound proteins were eluted by addition of 100 l of a buffer containing 20 mM Tris/HCl at pH 8.0, 1 mM dithiothreitol, 5 mM MgCl 2 , 0.5% Nonidet P-40, 20 mM NaCl, and 10 mM reduced glutathione. The eluates were subjected to SDS-PAGE, followed by protein staining with Coomassie Brilliant Blue.
Assay for the Receptor-mediated Endocytosis in PC12 and HeLa Cells-PC12 or HeLa cells were transfected with the plasmids encoding the indicated proteins by use of Lipofectin reagent (Life Technologies, Inc.). After 48 h, the cells were incubated for 60 min at 37°C with serum-free Dulbecco's modified Eagle's medium containing 0.1 mg/ml FITC-transferrin (Molecular Probes, Inc.). The cells were then fixed with 4% formaldehyde and permeabilized with 0.2% Triton X-100. The HA and myc epitopes were detected with the anti-HA and anti-myc antibodies as the first antibodies, respectively, and rhodamine-conjugated anti-mouse donkey antibody (Chemicon International) as the second antibody. To examine the co-expression of the HA-and myctagged proteins, rhodamine-conjugated anti-HA antibody and FITCconjugated anti-myc antibodies were used. The cells were analyzed with a LSM 410 confocal laser scanning microscope (Carl Zeiss, Oberkochen, Germany). To examine whether the expression level of the indicated protein was affected by co-expression of the other proteins, transfected HeLa cells were lysed, and the expression levels of the HA-tagged and myc-tagged proteins were examined by immunoblot analysis of whole cell lysates using the anti-HA and anti-myc antibodies, respectively.
Other Procedures-SDS-PAGE was performed as described (39). Rabphilin3 and rabaptin5 transferred to nitrocellulose sheets were detected using the ECL immunoblotting detection system (Amersham Corp.). Protein concentrations were determined with bovine serum albumin as a standard protein (40).

Interaction of Rabphilin3 with Rabaptin5 Estimated by the
Yeast two-Hybrid System-We first attempted to isolate rab-philin3-interacting proteins by use of the yeast two-hybrid system with the N-terminal fragment or the C-terminal fragment of rabphilin3 as a bait from a rat brain cDNA library. We did not isolate any clones with the C-terminal fragment as a bait. But with the N-terminal fragment as a bait, we isolated several clones and selected one clone (N22) encoding a protein, which corresponds to an amino acid position from 385 to 753 of human rabaptin5, because rabaptin5 is known to be the downstream target molecule of the Rab5 subfamily implicated in endocytosis (3,(15)(16)(17)(18). The isolated clone indeed interacted with the N-terminal fragment of rabphilin3 by the yeast twohybrid method, and this interaction was comparable to the interaction of Rab3A with rabphilin3 (Fig. 1A). This clone also interacted with full-length rabphilin3 but not with its Cterminal fragment (Fig. 1B).

Rabphilin3-Rabaptin5 Interactions in an Intact
Cell System-The rabphilin3-rabaptin5 interactions were confirmed by co-immunoprecipitation of HA-tagged full-length rabphilin3 and HA-tagged truncated rabaptin5 from cultured PC12 cells. PC12 cells were transiently transfected with the plasmid encoding HA-tagged rabphilin3 or HA-tagged rabaptin5, or with both the plasmids. The cells were lysed in the lysis buffer, and the cell lysates were incubated with the anti-rabphilin3 antibody immobilized on protein A-Sepharose. Rabphilin3 was immunoprecipitated by the anti-rabphilin3 antibody in both the cells expressing HA-tagged rabphilin3 alone and the cells expressing HA-tagged rabphilin3 and HA-tagged rabaptin5 (Fig.  2, lanes 2 and 3). Rabaptin5 was not co-immunoprecipitated by the anti-rabphilin3 antibody in the cells expressing HA-tagged rabaptin5 alone or the cells expressing HA-tagged rabphilin3 alone (Fig. 2, lanes 1 and 2) but co-immunoprecipitated with rabphilin3 in the cells expressing both HA-tagged rabphilin3 and HA-tagged rabaptin5 (Fig. 2, lane 3).
Inhibition by GTP␥S-Rab3A of the Rabphilin3-Rabaptin5 Interactions-The rabphilin3-rabaptin5 interactions were further confirmed by use of the recombinant proteins. GST-rabap-tin5 was prebound to glutathione-Sepharose, and the binding of rabphilin3 to this prebound protein was analyzed. Rabaptin5 indeed interacted with rabphilin3 (Fig. 3, lanes 1 and 2).
Inhibition of the Receptor-mediated Endocytosis of Transferrin by Rabphilin3-Finally, effect of rabphilin3 on the receptor-mediated endocytosis of transferrin was examined. PC12 cells were transiently transfected with the plasmid encoding myc-tagged human transferrin receptor alone or cotransfected with the plasmid encoding the HA-tagged N-termi-nal fragment of rabphilin3 and then incubated with FITCtransferrin. The expressed myc-tagged human transferrin receptor and HA-tagged N-terminal fragment of rabphilin3 were detected by confocal immunofluorescence microscopy using the anti-myc and anti-HA antibodies, respectively. The FITC-transferrin incorporated into the cells by endocytosis was examined by confocal immunofluorescence microscopy. In the cells expressing myc-tagged human transferrin receptor alone, FITC-transferrin was markedly incorporated into the cells (Fig. 4A, 1 and 2). However, in the cells co-expressing both myc-tagged human transferrin receptor and the HA-tagged N-terminal fragment of rabphilin3, the incorporation of FITCtransferrin into the cells was not detected (Fig. 4A, 3 and 4). This inhibitory action of the N-terminal fragment of rabphilin3 was overcome by co-expression with a dominant active mutant of Rab3A (Rab3AQ81L) (Fig. 4A, 5 and 6) or rabaptin5 (Fig. 4A,  7 and 8). These data were obtained in about 90% of the cells expressing the HA-tagged N-terminal fragment of rabphilin3. The co-expression of the HA-tagged N-terminal fragment of rabphilin3 and myc-tagged human transferrin receptor was confirmed by the anti-HA and anti-myc antibodies, respectively (Fig. 4A, 9 and 10). The expression of the Rab3A mutant and rabaptin5 was confirmed by confocal immunofluorescence microscopy using the anti-myc antibody (data not shown).
Similar results were obtained in HeLa cells. Because this cell line expresses endogenous human transferrin receptor, FITCtransferrin was incorporated into the cells by this endogenous receptor. In the cells expressing the HA-tagged N-terminal fragment of rabphilin3, the incorporation of FITC-transferrin into the cells was not detected, whereas FITC-transferrin was incorporated into the non-expressing cells (Fig. 4B, 1 and 2). This inhibitory action was overcome by co-expression with a dominant active mutant of Rab3A (Rab3AQ81L) (Fig. 4B, 3 and 4) or rabaptin5 (Fig. 4B, 5 and 6). The expression level of the HA-tagged N-terminal fragment of rabphilin3 was not impaired by co-expression of the Rab3A mutant or rabaptin5 (Fig.  4C). Overexpression of the HA-tagged C-terminal fragment of rabphilin3 alone did not inhibit the incorporation of FITCtransferrin into the cells (Fig. 4B, 7 and 8). DISCUSSION We have shown here by use of three different methods, the yeast two-hybrid, affinity column chromatography, and immunoprecipitation, that rabphilin3 directly interacts with rabap-tin5. Rabaptin5 interacts with the N-terminal region of rabphi-lin3, and this interaction is inhibited by GTP-Rab3A. The precise mode of action of the Rab3-rabphilin3 system in Ca 2ϩ - FIG. 2. Rabphilin3-rabaptin5 interactions in an intact cell system. PC12 cells were transiently transfected with the plasmid encoding HA-tagged rabphilin3 alone or HA-tagged rabaptin5 alone, or with both the plasmids. The cells were lysed, and the lysates were incubated with the anti-rabphilin3 antibody, which was immobilized on protein A-Sepharose. The immunoprecipitates were subjected to SDS-PAGE, followed by immunoblotting with the anti-HA antibody. Lane 1, with rabaptin5 alone; lane 2, with rabphilin3 alone; lane 3, with rabaptin5 and rabphilin3; arrowhead, HA-tagged rabaptin5; arrow, HA-tagged rabphilin3. The results shown are representative of three independent experiments. FIG. 3. Inhibition by GTP␥S-Rab3A of the rabphilin3-rabap-tin5 interactions. Full-length rabphilin3 was incubated with GST-or GST-rabaptin5-bound glutathione-Sepharose beads in the presence or absence of GDP-Rab3A or GTP␥S-Rab3A. After the beads were washed, the bound proteins were eluted, and the eluates were subjected to SDS-PAGE, followed by protein staining with Coomassie Brilliant Blue. dependent neurotransmitter release has not been clarified, but GTP-Rab3 is converted to GDP-Rab3 by the action of Rab3 GAP before, during, or after the fusion of synaptic vesicles with the presynaptic plasma membrane (6). Our results suggest that rabphilin3 interacts with rabaptin5 after GTP-Rab3 is converted to GDP-Rab3 and released from rabphilin3.
Moreover, we have shown here that overexpression of the N-terminal fragment of rabphilin3 in PC12 and HeLa cells inhibits the receptor-mediated endocytosis as estimated by the incorporation of FITC-transferrin and that this inhibition is overcome by co-expression with the dominant active mutant of Rab3A or rabaptin5. These results suggest that the N-terminal fragment interacts with endogenous rabaptin5 and inhibits its function in the endocytotic processes. The dominant active mutant of Rab3A or rabaptin5 may overcome the inhibitory action of the N-terminal fragment of rabphilin3 by inhibiting the interaction of the N-terminal fragment with endogenous rabaptin5. These results are consistent with the fact that rab-philin3 free of GTP-Rab3A, not that complexed with GTP-Rab3A, interacts with rabaptin5.
Endocytosis is often coupled with exocytosis, especially at nerve terminals (1,2). The mechanism of this type of endocytosis remains a matter of debate, but several lines of evidence implicate that endocytosis of synaptic vesicles appears to resemble receptor-mediated endocytosis mediated by clathrincoated vesicles in non-neuronal cells (for a review, see Ref. 41). However, since endocytosis of synaptic vesicles is faster than receptor-mediated endocytosis in non-neuronal cells and is not inhibited by decrease in the pH or the K ϩ concentration which disrupt receptor-mediated endocytosis (42), endocytosis of synaptic vesicles is considered a specialization of receptor-mediated endocytosis. Therefore, our present results used here indicate that the Rab3-rabphilin3 system is involved in receptor-mediated endocytosis, but it is not known whether rabphilin3 is implicated in endocytosis coupled with Ca 2ϩ -dependent exocytosis from nerve terminals. Further study is necessary for the precise role of rabphilin3 in endocytosis.