Rabphilin and Noc2 Are Recruited to Dense-core Vesicles through Specific Interaction with Rab27A in PC12 Cells*

Rabphilin and Noc2 were originally described as Rab3A effector proteins involved in the regulation of secretory vesicle exocytosis, however, recently both proteins have been shown to bind Rab27A in vitro in preference to Rab3A (Fukuda, M. (2003) J. Biol. Chem. 278, 15373-15380), suggesting that Rab3A is not their major ligand in vivo. In the present study we showed by means of deletion and mutation analyses that rabphilin and Noc2 are recruited to dense-core vesicles through specific interaction with Rab27A, not with Rab3A, in PC12 cells. Rab3A binding-defective mutants of rabphilin(E50A) and Noc2(E51A) were still localized in the distal portion of the neurites (where dense-core vesicles had accumulated) in nerve growth factor-differentiated PC12 cells, the same as the wild-type proteins, whereas Rab27A binding-defective mutants of rabphilin(E50A/I54A) and Noc2(E51A/I55A) were present throughout the cytosol. We further showed that expression of the wild-type or the E50A mutant of rabphilin-RBD, but not the E50A/I54A mutant of rabphilin-RBD, significantly inhibited high KCl-dependent neuropeptide Y secretion by PC12 cells. We also found that rabphilin and its binding partner, Rab27 have been highly conserved during evolution (from nematoda to humans) and that Caenorhabditis elegans and Drosophila rabphilin (ce/dm-rabphilin) specifically interact with ce/dm-Rab27, but not with ce/dm-Rab3 or ce/dm-Rab8, suggesting that rabphilin functions as a Rab27 effector across phylogeny. Based on these findings, we propose that the N-terminal Rab binding domain of rabphilin and Noc2 be referred to as “RBD27 (Rab binding domain for Rab27)”, the same as the synaptotagmin-like protein homology domain (SHD) of Slac2-a/melanophilin.

membrane trafficking within cells (1)(2)(3). The Rab3 subfamily is evolutionarily conserved from nematoda to humans, and mammalian Rab3A protein is present on synaptic vesicles in neurons and secretory granules in endocrine cells, and it appears to control Ca 2ϩ -regulated exocytosis (reviewed in Refs. 6 -8). The appropriate GDP/GTP exchange cycle of Rab3A is required for Ca 2ϩ -regulated exocytosis to occur, and interaction of the GTP-bound form of Rab3A with effector molecule(s) is widely believed to be essential for this process.
In the present study, we report finding that rabphilin and Noc2 are recruited to dense-core vesicles through specific interaction with Rab27A, not with Rab3A, and regulate densecore vesicle exocytosis in PC12 cells. We also show that only rabphilin⅐Rab27 interaction, not rapbhilin⅐Rab3 interaction, is phylogenetically retained in Caenorhabditis elegans and Drosophila. Based on our findings, we propose that rabphilin functions as a Rab27 effector across phylogeny.
Molecular Cloning of Mouse Noc2 and Rabphilin and C. elegans and Drosophila Rab3, Rab8, Rab27, and Rabphilin cDNAs-cDNAs encoding the full open reading frame of mouse Noc2 were amplified by reverse transcriptase-PCR from Marathon-Ready adult mouse brain cDNA (Clontech Laboratories, Inc.) with specific primers designed on the basis of the mouse sequence in the public data base (GenBank TM accession number, XM_110922) as described previously (71): 5Ј-CGGATCCATG-GCTGACACCATCTTCAG-3Ј (Noc2-Met primer with a BamHI linker (underlined); sense) and 3Ј-TGTCAGCTGGCCACTTCAGTA-3Ј (Noc2stop primer with a stop codon (in boldface type); antisense). The PCR products were purified from an agarose gel, directly inserted into the pGEM-T Easy vector (Promega), and verified by DNA sequencing as described previously (71).
cDNAs encoding the full open reading frame of the mouse rabphilin were screened from a mouse cerebellum oligo-dT-primed cDNA library constructed in gt11 (screening of 4 ϫ 10 5 plaques) as described previously (16). The cDNA inserts were subcloned into pBluscript KS(-) (Stratagene), and both strands were completely sequenced using a BcaBEST Dideoxy Sequencing Kit (Takara Bio Inc.; Shiga, Japan).
Unless otherwise noted, rabphilin means mouse rabphilin throughout the text.
Three days after transfection, PC12 cells (one 6-cm dish) expressing T7-GST-rabphilin-RBD, T7-GST-Noc2-RBD, or T7-GST, as a control, were harvested, and homogenized in 1 ml of the homogenization buffer described above. After solubilization with 1% Triton X-100, insoluble material was removed by centrifugation, and the GST fusion proteins expressed were affinity-purified on glutathione-Sepharose beads (wet volume 20 l) according to the manufacturer's notes. After washing the beads five times with 10 mM HEPES-KOH, pH 7.2, 150 mM NaCl, and 0.2% Triton X-100, proteins bound to the beads were analyzed by 10% SDS-PAGE, followed by immunoblotting with anti-Rab antibodies or HRP-conjugated anti-T7 tag antibody (1/10,000 dilution).
Competition Experiments-T7-tagged rabphilin-RBD (or Noc2-RBD) proteins were expressed in COS-7 cells and affinity-purified by anti-T7 tag antibody-conjugated agarose (Novagen) as described previously (28). The T7-rabphilin-RBD (or T7-Noc2-RBD) beads (wet volume 10 l) were then incubated with 400 l of solution containing FLAG-Rab3A and FLAG-Rab27A in various proportions (indicated in Fig. 3A) for 1 h at 4°C in 50 mM HEPES-KOH, pH 7.2, 150 mM NaCl, 1 mM MgCl 2 , 1% Triton X-100, and protease inhibitors. The amount of FLAG-Rab proteins in the solution was much greater than that of the T7-tagged FIG. 1. Distinct Rab27A recognition by rabphilin and Noc2. A, schematic representation of deletion mutants of the RBD of rabphilin and Noc2. The Rab binding activity of each mutant (Ϫ, Ϯ, ϩ, or ϩϩ) is indicated after its name. The RBD of rabphilin and Noc2 consists of two ␣-helical regions (RBD1(␣ 1 ) and RBD2(␣ 2 )) (black boxes and shaded boxes, respectively) that are separated by two zinc finger motifs (indicated by C) (12)(13)(14). The conserved SGAWF(Y/F) motif in the RBD2 is essential for Rab3A binding by rabphilin (14). The amino acid positions are indicated on both sides. B, Rab binding activity of RBD1 of rabphilin (Rph) and Noc2. Note that the RBD1 of Noc2 alone was sufficient for high affinity GTP-dependent Rab27A binding (lanes 6 and 10 in the middle panel in B), but not for Rab3A or Rab8A binding, the same as the SHD of Slp4-a and Slac2-a (34,36). By contrast, the RBD1 of rabphilin marginally interacted with Rab27A (lane 3 in the middle panel in B), indicating that the RBD1 alone was insufficient for recognition of all three Rabs (Rab3A, Rab8A, and Rab27A) by rabphilin. C, RBD1 of rabphilin and Noc2 is essential for Rab binding. Both the ⌬RBD1 of rabphilin and Noc2 mutants failed to bind all three Rabs (middle panel in C). D, differential contribution of RBD2 of rabphilin and Noc2 to Rab27A binding. The Noc2-⌬RBD2 mutant strongly interacted with Rab27A, but not with Rab3A or Rab8A (lanes 4 -6 in the middle panel in D), whereas the rabphilin-⌬RBD2 mutant marginally interacted only with Rab3A (lane 3 in the middle panel in D).
Miscellaneous Procedures-NPY-T7-GST secretion assay in PC12 cells was performed as described previously (34,75). NPY cDNA was provided by Dr. Wolfhard Almers (76). PC12 cell culture, transfection with pEGFP-C1-Noc2 or pEGFP-C1-rabphilin-RBD into PC12 cells, and immunocytochemical analysis of nerve growth factor (NGF)-differentiated PC12 cells were also performed as described previously (32,69,77). Immunoelectron microscopic analysis was essentially performed as described previously (78). T7-tagged rabphilin or Noc2 mutants and FLAG-tagged Rab (Rab3A, Rab8A, and Rab27A) were co-expressed in COS-7 cells, and the association of the two proteins was evaluated by immunoprecipitation as described previously (28,71). The blots shown in this report are representative of at least two independent experiments. Multiple sequence alignment and depiction of the phylogenetic tree of the Rab family proteins were performed by using the ClustalW program (available at www.hypernig.nig.ac.jp/homology/clustalw. shtml) set at the default parameters as described previously (11).

Different Contributions of the RBD1, Zinc Finger Motifs, and RBD2 of Rabphilin and Noc2 to Rab27A
Recognition-In our previous study, we showed by in vitro binding experiments that the RBD of rabphilin and Noc2 preferentially interacts with Rab27A/B, rather than with Rab3 isoforms (40). Since the RBD of rabphilin and Noc2 share the same subdomain structures as the type I SHD (two putative ␣-helical regions separated by zinc finger motifs; see Fig. 1A) (52), systematic deletion analysis was performed to investigate whether the mechanism of Rab27A recognition by the RBD of rabphilin or Noc2 is similar to the mechanism of recognition by the SHD of Slac2-a or Slp4-a (34,36). Consistent with our previous findings, the Noc2-RBD1 and Noc2-⌬RBD2 mutants, but not the Noc2-⌬RBD1 mutant, specifically interacted with Rab27A, but not with Rab3A or Rab8A (compare lanes 4 -6 in the middle panels of Fig. 1, B-D). In addition, the Noc2-RBD1 mutant preferentially recognized the Rab27A(Q78L) mutant (mimics GTP-bound form) rather than the Rab27A(T23N) mutant (mimics GDP-bound form) (Fig. 1B,  lanes 9 and 10 in the middle panel). These results indicate that FIG. 2. Identification of critical residues responsible for Rab27A and Rab3A binding in RBD1 of rabphilin and Noc2 by sitedirected mutagenesis. A, sequence alignment of RBD1 of mouse rabphilin and Noc2 and SHD1 of mouse Slp4-a. The positions of four conserved amino acids between the RBD1 and SHD1 are indicated (#), and mutation of one of the conserved residues of Slp4-a or Slac2-a completely abrogated GTP-Rab27A binding activity (32,36). Arrowheads indicate the amino acid substitutions in the RBD1 of rabphilin and Noc2. B, effect of single and double mutations in the RBD1 of Noc2 on Rab3A and Rab27A binding activity. C, effect of single and double mutations in the RBD1 of rabphilin on Rab3A and Rab27A binding activity. Note that the single mutation (E50A of rabphilin and E51A of Noc2) differently affected Rab3A and Rab27A binding activity (compare lanes 2 and 5 in B and C). pEF-T7-rabphilin (or -Noc2) point mutants and pEF-FLAG-Rab were cotransfected into COS-7 cells, and associations between the T7-rabphilin-RBD (or T7-Noc2) mutant and FLAG-Rab were evaluated by co-immunoprecipitation assay as described previously (35,71). Co-immunoprecipitated FLAG-Rab and the immunoprecipitated (IP) T7-rabphilin (or T7-Noc2) mutant were visualized with HRP-conjugated anti-FLAG tag antibody (Blot: anti-FLAG, IP: anti-T7; middle panels) and HRP-conjugated anti-T7 tag antibody (Blot: anti-T7, IP: anti-T7; bottom panels), respectively. Input means 1/80 volume of the reaction mixture used for immunoprecipitation (top panels). The positions of the molecular mass markers (ϫ10 Ϫ3 ) are shown on the left. the RBD1 of Noc2 alone is necessary and sufficient for GTP-dependent Rab27A binding activity, whereas the whole RBD of Noc2 is required for Rab3A and Rab8A recognition (40). Unexpectedly, however, none of the rabphilin deletion mutants (GST-RBD1, ⌬RBD1, and ⌬RBD2) strongly interacted with all three Rabs (lanes 1-3 in the middle panels of Fig. 1, B-D), although the RBD1 of rabphilin alone interacted very weakly with Rab27A (Fig. 1B, lane 3 in the middle panel). Therefore, the RBD1 of rabphilin may be a central Rab27A binding site, the same as the Slac2-a (or Slp4-a) SHD (34, 36), but the zinc finger motifs and the RBD2 are also indispensable for high affinity Rab27A recognition by rabphilin.
Next, we attempted to create a loss-of-function-type rabphilin and Noc2 (Rab3A binding-defective mutant or Rab27A binding-defective mutant) to investigate their Rab27 effector function in vivo. Ala-based site-directed mutagenesis was performed, especially focusing on the amino acids conserved among rabphilin, Noc2, Slp4-a, and Slac2-a ( Fig. 2A, # and  arrowheads) (34). The single mutation (E51A of Noc2 or E50A of rabphilin) was found to result in loss of Rab3A binding activity, but had no effect on Rab27A binding activity at all (compare lanes 2 and 5 in the middle panels of Fig. 2, B and C). By contrast, the double mutations (E51A/I55A of Noc2 and E50A/I54A of rabphilin) dramatically reduced Rab27A binding activity (lane 6 in the middle panels of Fig. 2, B and  C). Therefore, the same amino acid residues in the RBD1 of rabphilin and Noc2 are involved in Rab3A and Rab27A rec-ognition, although the RBD2 and zinc finger motifs of rabphilin and Noc2 contribute differently to Rab3A and Rab27A recognition and high affinity Rab27A recognition.
Recruitment of Rabphilin and Noc2 to Rab27A on Dense-core Vesicles in PC12 Cells-To determine whether rabphilin and Noc2 interact with Rab27A under physiological conditions, we selected PC12 cells for further analysis because they express all three Rabs (32), rabphilin (19), and Noc2 (41) and are often utilized for studies on secretion. We previously determined the ratio of the expression levels of the three Rabs in PC12 cells: Rab3A:Rab8A:Rab27A ϭ 1:2.5:0.25 (32). In vitro competition experiments (FLAG-Rab3A versus FLAG-Rab27A) clearly showed that Noc2 interacted only with Rab27A, even when a ten times greater amount of Rab3A than Rab27A was present in the reaction mixtures (Fig. 3A, closed arrowhead in the right middle panel). Since the ratio of expression levels of Rab3A and Rab27A in PC12 cells was about 0.8:0.2 (ϭ4:1) (Rab3A/ Rab27A), Noc2 should function as a specific Rab27A effector in intact PC12 cells (arrows in Fig. 3A). Rabphilin was also shown to prefer Rab27A to Rab3A in competition experiments, but rabphilin interacted with two molecules at the 0.75:0.25 ratio (Rab3A/Rab27A), almost corresponding to the ratio observed in PC12 cells (Fig. 3A, open and closed arrowheads in the left middle panel), suggesting that rabphilin interacts with both Rab3A and Rab27A in intact PC12 cells.
To investigate whether rabphilin⅐Rab27A (or Noc2⅐Rab27A) complexes actually form in vivo, an immunoprecipitation experiment was performed using anti-rabphilin or anti-Noc2 antibody (Fig. 3C). Consistent with the results of the in vitro binding experiments described above, all three Rabs were coimmunoprecipitated by the anti-rabphilin antibody (lane 6, arrowheads), whereas only Rab27A was co-immunoprecipitated with the anti -Noc2 antibody (lane 3 in the bottom panel).  E). Note that the wildtype rabphilin (or Noc2) colocalized well with Rab27A, a dense-core vesicle associated protein in PC12 cells (32,65) (arrowheads in C and F), indicating that Noc2 and rabphilin are recruited to dense-core vesicles through interaction with Rab27A in PC12 cells. Panels C and F are superpositions of A and B, and D and E, respectively. Scale bar in F indicates 20 m. G and H, localization of GFP-rabphilin-RBD and -Noc2 on dense-core vesicles as revealed by immunoelectron microscopy. Panels show representative views of the distal portion of the neurites stained with anti-GFP antibody followed by silver enhancement. Note that rabphilin (or Noc2) signals are often found on the dense-core vesicles (arrows) in the neurites. Scale bar in H indicates 0.5 m.
By contrast, no Rabs were detected in the immunoprecipitates of the control antibody (lanes 2 and 5). Since the antibodies used in this study were not appropriate for immunoprecipitation experiments (i.e. immunoprecipitation efficiency was low), and the signals of the co-immunoprecipitated Rabs were always weak, binding of exogenously expressed GST-rabphilin-RBD (or GST-Noc2-RBD) to endogenous Rabs in PC12 cells was also investigated. As expected, GST-Noc2-RBD specifically interacted with Rab27A, but not Rab3A or Rab8A (lane 1 in the middle panels), whereas GST-rabphilin-RBD interacted with all three Rabs (lane 2 in the middle panels). By contrast, the negative control, GST alone, did not trap any Rabs (lane 3 in the middle panels).
Rabphilin used to be thought to be present on dense-core vesicles as a result of a specific interaction with Rab3A in endocrine cells (19), but our in vitro binding experiments and co-immunoprecipitation assay strongly refute this notion, and rabphilin and Noc2 are most likely to be recruited to dense-core vesicles via Rab27A in endocrine cells. When the wild-type rabphilin-RBD or Noc2 tagged with GFP was expressed in NGF-differentiated PC12 cells, both molecules were predominantly localized in the distal portion of the neurites, where dense-core vesicles are known to be accumulated (Fig. 4, A and  G), and they colocalized with a dense-core vesicle protein, Rab27A (32, 65) (Fig. 4, B and H and Fig. 5, A-F), indicating that Noc2 and rabphilin are present on dense-core vesicles in PC12 cells. Dense-core vesicle localization of the wild-type GFP-rabphilin-RBD and GFP-Noc2 was further confirmed by immunoelectron microscopic analysis (arrows in Fig. 5, G and  H). Although the immunostaining pattern of the single mutant GFP-Noc2(E51A) was indistinguishable from that of the wildtype protein (Fig. 4C), the double mutant GFP-Noc2(E51A/ I55A) was localized throughout the cytoplasm (Fig. 4E). Similar results were obtained for the single and double mutants of rabphilin (Fig. 4, I and K), although some populations of PC12 cells expressing the GFP-rabphilin-RBD(E50A) exhibited weak cytoplasmic localization of this protein (compare G and I in Fig.  4). The results of these immunocytochemical analyses together with those of the binding experiments described above indicated that Noc2 and the majority of the rabphilin proteins are recruited to dense-core vesicles through specific interaction with Rab27A, not with Rab3A, in PC12 cells.
Rabphilin and Noc2 Modulate Dense-core Vesicle Exocytosis through Specific Interaction with Rab27A in PC12 Cells-Although Slp4-a has been shown to negatively control dense-core vesicle exocytosis through specific interaction with Rab27A in some endocrine cells (31,32,34,67), no endogenous positive regulator(s) of Rab27A in PC12 cells has yet been identified. Since overexpression of Slp3-a or Slp5 significantly promoted high KCl-dependent NPY secretion by PC12 cells, we investigated the expression of Slp3-a and Slp5 in PC12 cells by immunoblotting with specific antibodies. However, neither Slp3-a nor Slp5 was endogenously expressed in PC12 cells (data not shown), suggesting the presence of an additional Rab27 effector(s) other than the Slp family in PC12 cells. Since previous studies had demonstrated that overexpression of either fulllength rabphilin or Noc2 in PC12 cells cotransfected with growth hormone enhances high KCl-induced (but not low KClinduced) growth hormone secretion (19,41), whereas overexpression of the N-terminal domain of rabphilin inhibits high KCl-induced growth hormone secretion (18,19), we hypothesized that rabphilin and Noc2 function as Rab27A effectors that promote dense-core vesicle exocytosis in PC12 cells. To test this hypothesis, we used the Rab binding-defective mutants of the rabphilin-RBD described above for a NPY secretion assay. Consistent with the previous reports, expression of the wild-type RBD in PC12 cells significantly inhibited high KCldependent NPY secretion (Fig. 6A). The inhibitory effect of expression of the rabphilin-RBD on NPY secretion was completely abolished when the E50A/I54A mutation (i.e. Rab27A binding-defective mutant), but not the E50A mutation (i.e. Rab3A binding-defective mutant), was introduced (compare shaded and hatched bars in Fig. 6A). Since the expression levels of the recombinant proteins did not differ very much (inset in Fig. 6A), the effects observed in Fig. 6A were unlikely to be attributable to differences in protein levels. Although the single (E51A) and double (E51A/I55A) mutants of Noc2 were not applied to the NPY secretion assay in PC12 cells because of their very low protein expression levels compared with the wild-type protein (data not shown), expression of full-length Noc2 significantly promoted high KCl-dependent NPY secretion by PC12 cells (shaded bar in Fig. 6B), whereas the Rab27A binding-defective mutant of Noc2 (full-⌬RBD1) had no effect at all (hatched bar in Fig. 6B). Similar results were obtained when cells were pretreated with 10 g/ml brefeldin A for 30 min to eliminate constitutive NPY secretion (data not shown). Since similar expression levels of NPY were observed both in rabphilin (or Noc2)-transfected cells and nontransfected cells (open arrowhead in inset of Fig. 6B and data not shown) and the Noc2 and rabphilin expression had no significant effect on low KCldependent NPY secretion (11.3 Ϯ 2.2% (control), 10.7 Ϯ 1.1% (Rph-RBD), 10.4 Ϯ 1.0% (E50A), 8.0 Ϯ 5.1% (E50A/I54A), 14.4 Ϯ 0.8% (Noc2), and 14.8 Ϯ 5.1% (Noc2-⌬RBD1)), both rabphilin and Noc2 are likely to be involved in a late step of post-Golgi step in the dense-core vesicle exocytic pathway, either in vesicle translocation, docking or fusion. These results clearly indicated that rabphilin and Noc2 modulate dense-core vesicle exocytosis through specific interaction with Rab27A in PC12 cells.
Rabphilin Functions as a Rab27 Effector, but Not Rab3 Effector, during Evolution-Rab3, Rab8, and Rab27 (putative rabphilin binding partners in mammals), and rabphilin are known to be conserved in invertebrates, including C. elegans and Drosophila (4, 24, 79 -81). If the rabphilin⅐Rab interaction plays an important role in membrane trafficking across phylogeny, the relationship between Rab and its effector should be conserved during evolution. Interestingly, the Rab27 subfamily (black boxes) from mice, Drosophila, and C. elegans formed FIG. 7. Conservation of Rab27 proteins during evolution (from C. elegans to humans). A, sequence alignment of Rab27 proteins from various species: human (h), mouse (m), white leghorn (w) (GenBank TM accession number, BU105886), Xenopus (x) (CA986516), rainbow trout (t) (BX082375), Japanese medaka (md) (BJ007272), C. elegans (ce) (AB112930), and Drosophila (dm) (AB112931). Residues in the sequences that are conserved or similar are shown against a black background and against a shaded background, respectively. Parentheses indicate the switch I and switch II regions, putative Rab27A contact sites in SHD or RBD (14,36). The critical residues in the switch I and switch II regions of mouse Rab27A for recognition by the Slac2-a SHD are indicated by number signs and asterisks, respectively (36,40). It should be noted that these residues are invariably retained during evolution, whereas the C-terminal region is highly diversified. Putative C-terminal geranylgeranylation sites are boxed. B, phylogenetic tree of mouse Rabs1-40 and C. elegans and Drosophila Rab3, Rab8, and Rab27. This tree was drawn with the ClustalW program as described previously (11). Note that the Rab3 subfamily (shaded boxes), the Rab8 subfamily (open boxes), and the Rab27 subfamily (black boxes) of mice, Drosophila, and C. elegans form distinct branches in the phylogenetic tree. branches in the phylogenetic tree of the Rab family that were completely different from the branches formed by the Rab3 subfamily (shaded boxes) and the Rab8 subfamily (open boxes) (Fig. 7B). Data base searching for Rab27 proteins from different species and sequence alignment also revealed very high sequence conservation of Rab27 proteins during evolution. Vertebrates possess two Rab27 isoforms, Rab27A and Rab27B, whereas C. elegans and Drosophila contain only a single isoform (Fig. 7A). It should be noted that the switch I and switch II regions (especially the three conserved amino acids; # and * in Fig. 7A) (36,40), the putative Rab27 contact sites of Slac2-a, were invariably conserved in all Rab27 proteins, suggesting that Rab27 effector protein(s) may also be retained during evolution. By contrast, the C-terminal region of Rab27 was diversified during evolution, and the putative geranylgeranylation motif also differed between vertebrates and invertebrates (CXC versus CXNC) (open boxes in Fig. 7A).
What is the Rab27 effector(s) in invertebrates? We previously searched for putative Rab27 effector proteins in C. elegans and Drosophila but discovered that neither mammalian homologue of Slp nor Slac2 proteins (52) was present in these animals (79, 80) (note that the Drosophila Slp homologue bite size lacks a SHD; see Ref. 82). Since Noc2 was also not found in either species, the only possible Rab27 effector is rabphilin. To confirm this, we prepared C. elegans and Drosophila rabphilin (named ce-rabphilin and dm-rabphilin, respectively), ce/dm-Rab3, ce/dm-Rab8, and ce/dm-Rab27, and tested their interaction in COS-7 cells. To my surprise, ce-rabphilin specifically interacted with ce-Rab27, but neither the ce-Rab3 band nor the ce-Rab8 band was detected even after prolonged exposure of the x-ray film (lane 3 in the middle panel of Fig. 8A). The same result was obtained for the dm-rabphilin: it specifically interacted with dm-Rab27, not with dm-Rab3 or dm-Rab8 (lane 3 in the middle panel of Fig. 8B). This was in strong contrast to mouse rabphilin, because it interacted with Rab3A, Rab8A, and Rab27A (Fig. 8C). These results strongly suggest that rabphilin functions as a Rab27 effector rather than a Rab3 effector during evolution. DISCUSSION In a previous study, we showed that rabphilin and Noc2 interacted with three different Rab proteins, Rab3A, Rab8A, and Rab27A, in vitro (40). In the present study, we have demonstrated several lines of evidence that rabphilin and Noc2 function as Rab27A effectors in dense-core vesicle exocytosis in PC12 cells. First, in vitro binding experiments clearly showed that the RBD of rabphilin and Noc2 preferentially interacted with Rab27A rather than with Rab3A (Fig. 3A), although the mechanism of Rab27A recognition by rabphilin was different from the mechanism of recognition by Noc2 (Fig. 1). Interaction of rabphilin with Rab27A requires all three subdomains of the RBD (i.e. RBD1, zinc finger motifs, and RBD2), whereas the RBD1 of Noc2 is necessary and sufficient for Rab27A binding, the same as the Slac2-a SHD (Fig. 1) (36). In addition, the Rab3A binding activity of Noc2 seemed to be much weaker than that of rabphilin (about 5-10 times lower affinity; see Refs. 12,26,and 42), and thus Noc2 behaves as a specific Rab27-binding protein in intact PC12 cells (Fig. 3, B and C). Second, the wild-type or Rab3A binding-defective mutant of rabphilin and Noc2 colocalized with Rab27A in the distal portion of the neurites in NGF-differentiated PC12 cells, whereas the Rab27A binding-defective mutants of rabphilin and Noc2 were present throughout the cytosol (Figs. 2 and 4). This finding explained the previous observations that rabphilin (or Rab3A binding-defective mutants) is present on secretory vesicles and promotes stimulated secretion independently of Rab3A (25,26,83): rabphilin is recruited to secretory granules via Rab27A/B (not via Rab3A) and promotes stimulated secretion.
Third, expression of the wild-type rabphilin-RBD and Noc2 in PC12 cells modulated high KCl-dependent NPY secretion, whereas the Rab27A binding-defective mutants (E50A/I54A mutant of rabphilin and Noc2-⌬RBD1) had no effect on NPY secretion (Fig. 6). Last, we found that Rab27 proteins are highly conserved across phylogeny (Fig. 7) and that ce/dm-rabphilin specifically interacted with ce/dm-Rab27, not with ce/dm-Rab3 or ce/dm-Rab8, although mouse rabphilin interacted with all three Rabs (Fig. 8). This finding is quite consistent with the fact that rabphilin knock-out animals (C. elegans and mice) do not exhibit obvious genetic interaction with Rab3 mutant animals (23,24). Therefore, invertebrate rabphilin is most likely to function as a specific Rab27 effector, whereas vertebrate rabphilin functions as an effector for both Rab3 isoforms and Rab27 isoforms. Since the expression levels of Rab27A in the brain are very low, and Rab27B, a closely related isoform of Rab27A, is present in the brain instead (31,44,58), Rab27B may be present on synaptic vesicles and regulate neurotransmitter release. In our prelimi- FIG. 8. Specific interaction of invertebrate rabphilin with invertebrate Rab27. A, specific interaction of C. elegans rabphilin (cerabphilin) with ce-Rab27. B, specific interaction of Drosophila rabphilin (dm-rabphilin) with dm-Rab27. C, interaction of m-rabphilin with m-Rab3A, m-Rab8A, and m-Rab27A. pEF-T7-rabphilin-RBD and pEF-FLAG-Rab were cotransfected into COS-7 cells, and associations between the T7-rabphilin-RBD and FLAG-Rab were evaluated by co-immunoprecipitation assay as described previously (35,71). Coimmunoprecipitated FLAG-Rab and the immunoprecipitated (IP) T7-rabphilin-RBD were visualized with HRP-conjugated anti-FLAG tag antibody (Blot: anti-FLAG, IP: anti-T7; middle panels in A-C) and HRP-conjugated anti-T7 tag antibody (Blot: anti-T7, IP: anti-T7; bottom panels in A-C), respectively. Input means 1/80 volume of the reaction mixture used for immunoprecipitation (top panels in A-C). The positions of the molecular mass markers (ϫ 10 Ϫ3 ) are shown on the left. nary experiments, rabphilin⅐Rab27B complex was detected in mouse brain by co-immunoprecipitation assay. 2 Further work is needed to determine whether the complex is actually present on synaptic vesicles and involved in the regulation of neurotransmitter release.
In conclusion, we have demonstrated that rabphilin and Noc2, previously characterized as Rab3A effector proteins, are recruited to dense-core vesicles through specific interaction with Rab27A in PC12 cells and that these two proteins positively regulate dense-core vesicle exocytosis. We have also shown that invertebrate rabphilin acts as a Rab27-binding protein, not as a Rab3-binding protein, suggesting that a rabphilin⅐Rab27 complex has an important role(s) in the regulation of fundamental membrane trafficking that is conserved from C. elegans to humans. Based on our results, we propose that the N-terminal RBD of rabphilin and Noc2 be referred to as RBD27 (Rab binding domain for Rab27), in the same manner as the SHD of the Slp and the Slac2 families (52).