The Rab5 effector Rabaptin-5 and its isoform Rabaptin-5d differ in their ability to interact with the small GTPase Rab4 Elena Korobko 1,2 , Sergey Kiselev 1 , Sjur Olsnes 3 , Harald Stenmark 3 and Igor Korobko 1 1 Institute of Gene Biology, Russian Academy of Sciences, Moscow, Russia 2 University of Oslo, Centre for Medical Studies at Moscow, Moscow, Russia 3 Institute for Cancer Research, The Norwegian Radium Hospital, Oslo, Norway Intracellular membrane transport is an important pro- cess for eukaryotic cells. Intracellular membranes are organized in compartments, and transport between compartments requires high specificity and tight regu- lation. The intercompartmental transport typically occurs through transport vesicles budding from a donor compartment and fusing to an acceptor com- partment. In this process, Rab GTPases were demon- strated to play a central role by regulating vesicle budding, motility and fusion [1]. Another group of molecules, v- and t-SNAREs, were suggested to con- tribute to specificity of vesicle targeting [2–4] although the specificity of the transport is also maintained by Rab GTPases [5]. A large number of Rab GTPases have been identified in mammalian cells, and each of them seems to regulate a specific step in membrane transport through their downstream effectors, which are recruited by the GTPases in their active GTP- bound conformation [1]. The Rab5 GTPase is a regulator of early endocytic events in the cell [6]. It plays a role in the formation of clathrin-coated vesicles at the plasma membrane [7], in their heterotypic fusion with early endosomes and homotypic fusion of early endosomes [8,9], and in microtubule-dependent motility of endocytic vesicles [10]. By now, several Rab5 effector molecules have been identified that, in a coordinated way, act in a chain of molecular events (reviewed in [11]). One of Keywords endocytosis; Rab5; Rab4; Rabaptin-5 isoform; effector molecule Correspondence E. Korobko, Institute of Gene Biology, Russian Academy of Sciences, 34 ⁄ 5 Vavilov Street, Moscow 119334, Russia Fax: +7 095 1354105 Tel: +7 095 1359970 E-mail: alenavk@igb.ac.ru (Received 19 July 2004, accepted 16 August 2004) doi:10.1111/j.1432-1033.2004.04399.x Rabaptin-5 is an effector for the small GTPase Rab5, a regulator of the early steps in endocytosis. In addition, Rabaptin-5 interacts with the small GTPase Rab4 that has been implicated in recycling from early endosomes to the cell surface. Recently we have identified a ubiquitous transcript encoding the Rabaptin-5 isoform, Rabaptin-5d. To evaluate the interaction properties of Rabaptin-5 d with the small GTPases Rab4 and Rab5, we have applied protein interaction assays using the yeast two-hybrid system and a glutathione S-transferase pull-down assay. We found that unlike Rabaptin-5, that interacts with both GTPases in GTP-bound conforma- tions, Rabaptin-5d interacts only with GTP-bound Rab5, and does not interact with Rab4, presumably due to a disrupted Rab4 binding site. Immunofluorescence microscopy analysis carried out to address the locali- zation of Rabaptin-5d relative to GTP-bound Rab4 and Rab5 in BHK-21 cells supported these data. Our data suggests that while Rabaptin-5 was proposed to act as a molecular linker between Rab5 and Rab4, to coordi- nate endocytic and recycling traffic, Rabaptin-5d is involved only in the Rab5-driven events. Abbreviations 3AT, 3-amino-(1,2,4)-triazole; EGFP, enhanced green fluorescent protein; GEF, guanine nucleotide exchange factor; GAL4AD, GAL4 transcriptional activating domain; GAL4BD, GAL4 DNA-binding domain; GST, glutathione S-transferase; SD, synthetic dextrose; TGN, trans-Golgi network. FEBS Journal 272 (2005) 37–46 ª 2004 FEBS 37 these is Rabaptin-5, which is an essential and rate- limiting factor for homotypic early endosome fusion and heterotypic fusion between early endosomes and clatrin-coated vesicles [12–14]. Cytosolic Rabaptin-5 is complexed with Rabex-5, a Rab5 guanine nucleotide exchange factor (GEF), and both proteins act syner- gistically to activate Rab5 in early endosome fusion events [13,15]. As well as being a Rab5 effector, Rabaptin-5 has been suggested to play a role in connecting different steps of the membrane transport process. First, Rabaptin-5 was demonstrated to interact through a distinct binding domain with another small GTPase, Rab4 [14], which was implicated in regulation of rapid recycling from early endosomes [16,17]. Second, Rabaptin-5 can participate in balancing and coordina- tion of endo- and exocytosis through interaction with Rabphilin, a regulator of the exocytic pathway [18,19]. Finally, Rabaptin-5 interacts with the ear domains of c1-adaptin, a subunit of the AP-1 adaptor complex of clathrin-coated vesicles derived from trans-Golgi network (TGN) [20], and GGAs, a family of Arf-dependent clathrin adaptors involved in selec- tion of TGN cargo [21]. This reveals a functional link between proteins regulating TGN cargo export and endosomal tethering ⁄ fusion events through Rabaptin- 5. Thus, Rabaptin-5, along with several other pro- teins, emerges to be a multivalent effector molecule with a possible role in the regulation of subcompart- mental organization and sorting of membrane vesicles [22]. Rabaptin-5 is a 100-kDa protein encoded by the RAB5EP gene. Recent studies revealed that in addition to the main transcript, a number of minor transcripts exist that bear small deletions in the coding region [23–25]. Evidence was provided that these transcripts are probably generated by alternative splicing from a single pre-mRNA [23]. One of the recently identified Rabaptin-5 isoforms is Rabaptin-5d [23]. Compared to Rabaptin-5, this isoform has a short deletion of 40 amino acid in the N-terminus. The deleted region is partially inside of the second N-terminal coiled-coil domain of Rabaptin-5 [14,23] (Fig. 1). The ubiquitous occurrence of the Rabaptin-5d transcript suggests that this protein could play a significant role in the cell, probably through modulation of the Rabaptin-5 func- tions. In an attempt to clarify this suggestion, and to better understand the functional role of Rabaptin-5d, we have characterized this molecule with respect to its ability to interact with the known Rabaptin-5 interaction partners, Rab4, Rab5 and Rabex-5 as well as its ability to be recruited to specific endosomal compartments. Results Rabaptin-5d interacts differentially with the small GTPases Rab4 and Rab5 The interaction of Rabaptin-5 with the small GTPases Rab4 and Rab5 can be assessed readily in the yeast two-hybrid system [12,14]. We therefore used this assay to analyze the interaction of Rabaptin-5d with Rab4 and Rab5. Similarly to Rabaptin-5, Rabaptin-5d was not able to bind Rab4 or Rab5 in the inactive, GDP-bound form as no HIS3 reporter gene trans-activation was observed upon coexpression of GAL4AD–Rabaptin-5 or -Rabaptin-5d and GAL4BD fused to Rab5S34N or Rab4S22N, dominant negative mutants of Rab5 and Rab4 with decreased affinity for guanine nucleotides (Figs 2 and 3). Likewise, no HIS3 reporter gene trans- activation was revealed upon coexpression of the wild- type Rab5 bait and the Rabaptin-5d isoform prey (Fig. 2), which is consistent with no detectable reporter gene activation upon Rab5 bait and full-length Rabap- tin-5 prey coexpression [14]. Therefore, the GTPase- deficient mutant of Rab5, Rab5Q79L, was used as a bait to assay the interaction with Rabaptin-5d. Similarly to what has been reported for Rabaptin-5 [12,14], expression of the Rabaptin-5d prey together with the Rab5Q79L bait resulted in the HIS3 reporter gene trans-activation (Fig. 2). Unlike Rab5, both wild-type Rab4 and its GTPase- deficient mutant bait coexpression with full-length Rabaptin-5 prey have been shown to result in readily detectable reporter gene trans-activations [14]. We R4BD CC1-1 CC1-2 CC2-1 CC2-2 R4BD CC1-1 CC1-2 CC2-1 CC2-2 Rabaptin-5 Rabaptin-5 12 93 191 262 aa 5 135 187-226 R5B D R5B D aa R4BD 140 295 R4BD Fig. 1. Schematic representation of Rabaptin-5 and its d isoform. Coiled-coil regions (CC1–1, CC1–2, CC2–1, CC2–2) are shown in black. Positions of the Rab4 binding domain (R4BD) are marked as white and black hatched boxes as determined in [14] and [26], respectively. The Rab5 binding domain (R5BD) is marked as a dot- filled box. The region deleted in Rabaptin-5d is shown in white. Positions of coiled-coil regions CC1-1 and CC1-2, Rab4 binding domain, and the region deleted in Rabaptin-5d are numbered according to the amino acid sequence of mouse Rabaptin-5 (Gene- Bank Accession No. D86066). Rabaptin-5 isoform d is not a Rab4 effector E. Korobko et al. 38 FEBS Journal 272 (2005) 37–46 ª 2004 FEBS observed a similar pattern of the reporter gene trans- activations for Rabaptin-5 (Fig. 3). However, when Rabaptin-5d was assayed as a prey in the yeast two- hybrid system, no HIS3 reporter gene trans-activation was observed with neither bait, wild-type Rab4 nor its GTPase-deficient mutant (Fig. 3) suggesting the lack of interaction between Rab4 and Rabaptin-5d. The interaction properties of the Rabaptin-5 iso- forms with Rab4 and Rab5 were further assayed in glutathione S-transferase (GST) pull-down assays. GST–Rab4 preloaded with either GDP or the unhy- drolyzable GTP analogue, GTPcS, was unable to pull down enhanced green fluorescent protein (EGFP)- tagged Rabaptin-5d from cytosol of transiently trans- fected BHK-21 cells (Fig. 4A). At the same time, consistent with the data from yeast two-hybrid system, EGFP–Rabaptin-5d was pulled down by GTPcS-loa- ded but not with GDP-loaded GST-Rab5; similar to Rabaptin-5 (Fig. 4B). To exclude the possibility that the EGFP-tag located at the N-terminus of Rabaptin- 5d, which is close to the mapped Rab4 binding site, might interfere with Rab4 binding, similar experiments were performed with cytosols prepared from BHK-21 cells transiently expressing Rabaptin-5 or Rabaptin-5d with a C-terminal FLAGÒ tag (DYKDDDDK). Simi- larly to experiments with EGFP-tagged proteins, GST–Rab4 was unable to pull down Rabaptin-5d– FLAGÒ in either GDP or GTPcS-bound form. At the same time, Rabaptin-5 was specifically pulled down from cytosol by GTPcS-loaded GST-Rab4 (Fig. 4C). In summary, both protein interaction analyses in the yeast two-hybrid system and GST pull-down assays suggest that Rabaptin-5d can specifically interact only with Rab5 but not with Rab4. Cytosolic Rabaptin-5d is complexed with Rabex-5 The Rab5 effector properties of Rabaptin-5 are mani- fested upon complexing of Rabaptin-5 with Rabex-5 [13,15]. Although Rabaptin-5 can interact with Rab5– Fig. 2. Interaction specificity of Rabaptin-5d with Rab5 in the yeast two-hybrid system. Y153 reporter yeast cells were cotransformed with plasmids encoding GAL4BD alone or fused to Rab5 or its mutants, and GAL4AD alone or fused to Rabaptin-5 (Rn5) or Rabaptin-5d (Rn5d). HIS3 reporter gene activation caused by inter- action between GAL4BD- and GAL4AD-fused proteins in yeast was assessed by spotting onto synthetic minimal medium supplemen- ted (bottom) or not supplemented (top) with 25 m M 3-amino- (1,2,4)-triazole (3AT). Fig. 3. Interaction specificity of Rabaptin-5d with Rab4 in the yeast two-hybrid system. Y153 reporter yeast cells were cotransformed with plasmids encoding GAL4BD alone or fused to Rab4 or its mutants, and GAL4AD alone or fused to Rabaptin-5 (Rn5) or Rabap- tin-5d (Rn5d). HIS3 reporter gene activation caused by interaction between GAL4BD- and GAL4AD-fused proteins in yeast was assessed by spotting yeast onto synthetic minimal medium supple- mented (bottom) or not supplemented (top) with 25 m M 3AT. E. Korobko et al. Rabaptin-5 isoform d is not a Rab4 effector FEBS Journal 272 (2005) 37–46 ª 2004 FEBS 39 GTP in vitro, Rabaptin-5 alone could not support bio- logical activity of Rab5 in early endosome fusion [12]. We therefore asked if the Rabaptin-5d isoform is complexed with Rabex-5 in vivo. As shown in Fig. 5, Rabex-5 coimmunoprecipitates with EGFP-tagged Rabaptin-5d from cytosol of transfected BHK-21 cells, similarly to Rabaptin-5. Thus Rabaptin-5d is associ- ated with the Rab5 GEF, Rabex-5, in vivo. Subcellular localization of Rabaptin-5d upon coexpression with GTPase-deficient mutants of Rab4 or Rab5 GTPases The findings based on the yeast two-hybrid system and GST pull-down analyses suggest that, whereas Rabap- tin-5 interacts with both Rab4 and Rab5 in their GTP- bound forms, Rabaptin-5d can interact only with GTP-bound Rab5 and not with GTP-bound Rab4. To address the question of how relevant these findings are to the protein interaction properties in vivo, we next examined the subcellular localization of Rab4, Rab5 and Rabaptin-5d by confocal immunofluorescence microscopy. As the available antibodies against Rab4 and Rab5 failed to detect the endogenous proteins, we coexpressed myc-tagged Rab4Q67L or Rab5Q79L with Rabaptin-5 isoforms as a C-terminal fusion partner of EGFP. Expression of the GTPase-deficient Rab5 mutant results in the appearance of enlarged Rab5-positive early endosomes that also recruit Rabaptin-5 [12]. Consistent with the results of protein interaction assays in vitro and in the yeast two-hybrid system, EGFP- Rabaptin-5d was localized on enlarged myc-positive vesicular structures when coexpressed with myc- Rab5Q79L in BHK-21 cells (Fig. 6B,B¢). In this case, localization of EGFP and the myc-epitope in cotransfected cells were similar to those observed upon A B C Fig. 4. Interaction specificity of Rabaptin-5d with Rab4 and Rab5 in a GST pull-down assay. GST–Rab4 (A, C) or GST-Rab5 (B) were immobilized on Gluthatione Sepharose 4B beads and preloaded with either GDP or GTPcS as indicated. EGFP-tagged, or EGFP alone (E) (A, B) or C-terminally FLAGâ-tagged Rabaptin-5 (5), Ra- baptin-5d (5d) (C) were pulled down from cytosols prepared from BHK-21 cells transiently expressing respective proteins. Pulled down proteins (upper panels) and aliquots of cytosols (bottom pan- els) were separated in SDS ⁄ PAGE and immunoblotted with poly- clonal antiserum against Rabaptin-5 (A, B) or with anti-FLAGâ-M2 monoclonal Igs (C). Arrows indicate endogenous Rabaptin-5 (Rn5) and EGFP-fused Rabaptin-5 and its isoform (E-Rn5’s) (A,B), and FLAGÒ-tagged Rabaptin-5 and its isoform (Rn5-FLAGâ)(C). AB DC Fig. 5. Cytosolic Rabaptin-5d expressed in BHK-21 cells is com- plexed with Rabex-5. Cytosols were prepared from BHK-21 cells transiently expressing EGFP alone (E) or EGFP C-terminally fused with Rabaptin-5 (5) or Rabaptin-5d (5d). Cytosols (A, C) or proteins immunoprecipitated from cytosols with anti-EGFP Igs (B, D) were separated by SDS ⁄ PAGE and immunoblotted with polyclonal anti- sera against Rabaptin-5 (A, B) or against Rabex-5 (C, D). Arrows indicate endogenous Rabaptin-5 (Rn5) and EGFP-fused Rabaptin-5 and its isoform (E-Rn5’s), Rabex-5 (Rabex-5), and antibody used for immunoprecipitation (Ab). Rabaptin-5 isoform d is not a Rab4 effector E. Korobko et al. 40 FEBS Journal 272 (2005) 37–46 ª 2004 FEBS coexpression of EGFP–Rabaptin-5 and myc-tagged Rab5Q79L (Fig. 6A,A¢). When a myc-tagged GTPase-deficient mutant of Rab4, Rab4Q67L, was coexpressed in BHK-21 cells with EGFP–Rabaptin-5, we found that the two pro- teins colocalize on vesicular structures in the perinu- clear area (Fig. 7A,A¢ and insets), which is consistent with previous reports [14,24]. Whereas EGFP–Rabap- tin-5d also concentrated in the perinuclear region when coexpressed with myc-tagged Rab4Q67L, the two pro- teins did not show any significant colocalization (Fig. 7B,B¢ and insets). To exclude the possibility that N-terminally fused EGFP protein influences Rab4-binding properties of Rabaptin-5d, similar colocalization experiments were performed with N-terminally FLAGÒ-tagged Rabap- tin-5 and Rabaptin-5d. Figure 8 shows that FLAGÒ– Rabaptin-5d does not colocalization with Rab4Q67L- positive structures while FLAGÒ–Rabaptin-5 inten- sively colocalizes with Rab4Q67L. Taken together, the data from confocal immunofluo- rescence microscopy provide further support that, unlike Rabaptin-5, Rabaptin-5d exclusively interacts with Rab5 but not with Rab4. Discussion We have identified previously a cDNA encoding a novel protein, Rabaptin-5d. This protein is similar to Rabaptin-5 but has small deletions in the N-terminal Fig. 6. Confocal immunofluorescence analysis of BHK-21 cells coexpressing myc- tagged Rab5Q79L and EGFP-tagged Rabap- tin-5 or EGFP-tagged Rabaptin-5d. Cells cotransfected with myc-tagged Rab5Q79L and EGFP-tagged Rabaptin-5 (A–A¢)or Rabaptin-5d (B–B¢) were stained with mouse monoclonal anti-(myc 9E10 Ig) followed by Alexa546-conjugated anti-mouse secondary Igs (A¢,B¢), or EGFP fluorescence was recorded (A, B). Overlays are shown in A and B with yellow color indicating colocalization, red color indicating myc– Rab5Q79L, and green color indicating EGFP-fused proteins. Panel dimensions are 80 lm  80 lm. Fig. 7. Confocal immunofluorescence analysis of BHK-21 cells coexpressing myc- tagged Rab4Q67L and EGFP-tagged Rabap- tin-5 or EGFP-tagged Rabaptin-5d. Cells cotransfected with myc-tagged Rab4Q67L and EGFP-tagged Rabaptin-5 (A–A¢)or Rabaptin-5d (B–B¢) were stained with mouse monoclonal anti-(myc 9E10 Ig) followed by Alexa546-conjugated anti-mouse secondary antibodies (A¢,B¢), or EGFP fluorescence was recorded (A, B). Overlays are shown in A and B with yellow color indicating colo- calization, red color indicating myc- Rab4Q67L, and green color indicating EGFP- fused proteins. The black-and white insets in panels A¢ and B¢ shows higher magnifica- tion of indicated regions for green (left )and red (right) channels. Panel dimensions are 100 lm  100 lm. E. Korobko et al. Rabaptin-5 isoform d is not a Rab4 effector FEBS Journal 272 (2005) 37–46 ª 2004 FEBS 41 portion of the polypeptide chain [23]. Rabaptin-5 was identified initially as a Rab5 effector protein that specif- ically interacted with Rab5 in the GTP-bound form and was an essential component for homotypic early endo- some fusion and heterotypic fusion between clathrin- coated vesicles and early endosomes [12–14]. It was pro- posed that Rabaptin-5 functioned by coupling Rab5 to its GEF, Rabex-5, which is bound to Rabaptin-5 [13,15]. Besides a Rab5 effector function, the specific interaction between Rabaptin-5 and another small GTPase in the GTP-bound conformation, Rab4, was demonstrated [11,14] thus suggesting the possibility that Rabaptin-5 could function as a linker between two sequential steps in membrane transport; early endosome fusion and recycling [22]. Here we report the characteri- zation of Rabaptin-5d isoform interaction properties with the small GTPases, Rab4 and Rab5. The analyses in the yeast two-hybrid system and pull-down experiments suggest that Rabaptin-5d spe- cifically interacts with Rab5 in its GTP-bound confor- mation. However unlike Rabaptin-5, Rabaptin-5d prey failed to trans-activate reporter genes when either Rab4 or its GTPase-deficient mutant were used as a bait. This suggests a lack of interaction between Rab4 and Rabaptin-5d, and this was further supported by results of GST pull-down experiments. Interestingly, both the deletion found in the Rabaptin-5d polypep- tide chain and the Rab4 binding site of Rabaptin-5 are located in the N-terminal portion of protein. However, some discrepancy exists in the literature concerning the position of the Rab4-binding site. While Vitale et al. mapped it between amino acids 5 and 135, by analyz- ing protein–protein interactions in yeast two-hybrid system [14], the recent findings of Deneka et al. suggest that the Rab4-binding motif of Rabaptin-5 resides between amino acids 140 and 295 as demonstrated in GST pull-down experiments [26] (Fig. 1). To clarify this point, we assayed the interaction properties of dif- ferent N-terminal fragments of Rabaptin-5 with Rab4 in the yeast two-hybrid system. Consistent with the data of Deneka et al. [26], amino acids 140–294 of Rabaptin-5 were sufficient to interact with Rab4 as judged by trans-activation of the LacZ reporter gene, while a polypeptide consisting of the first 149 amino acids was not (Fig. 9). At the same time, a fragment of Rabaptin-5d corresponding to amino acids 140–294 of Rabaptin-5, which contains the deleted region, was unable to bind Rab4. In addition, the first 251 amino acids of the Rabaptin-5c isoform, bearing a natural deletion of amino acids 22–64 of Rabaptin-5 [23], was able to interact with Rab4, thus demonstrating the dis- pensability of these amino acids for the interaction (Fig. 9). Taken together, the conclusion can be made that Rab4 binding site is located between amino acids 140 and 294 of Rabaptin-5. Rabaptin-5d has amino acids 187–226 deleted from the polypeptide chain, which is inside the minimal Rab4 binding fragment. This suggests that the deletion disrupts the Rab4 bind- ing site of Rabaptin-5, which is further supported by the demonstrated lack of interaction between Rab4 and Rabaptin-5d. The colocalization studies in BHK-21 cells support the assumption that Rabaptin-5 d can interact with Rab5 but not with Rab4. Accordingly, the recruitment of EGFP–Rabaptin-5d on the enlarged Rab5Q79L- positive endosomes was observed whereas no apparent colocalization with Rab4Q67L was seen. Finally, we observed association of Rabaptin-5d with Rabex-5 Fig. 8. Confocal immunofluorescence anal- ysis of BHK-21 cells coexpressing myc- tagged Rab4Q67L and FLAGÒ-tagged Rabaptin-5 or FLAGÒ-tagged Rabaptin-5d. Cells cotransfected with myc-tagged Rab4Q67L and FLAGÒ-tagged Rabaptin- 5(A–A¢) or Rabaptin-5d (B–B¢) were stained with mouse monoclonal anti-(myc 9E10 Ig) and anti-FLAGÒ rabbit polyclonal Igs fol- lowed by Alexa546-conjugated anti-mouse and Alexa488-conjugated anti-rabbit secon- dary Igs. Red (A¢,B¢) or green (A, B) fluores- cence was recorded. Overlays are shown in A and B with yellow color indicating colo- calization, red color indicating myc- Rab4Q67L, and green color indicating FLAGÒ-tagged proteins. Panel dimensions are 100 lm  100 lm. Rabaptin-5 isoform d is not a Rab4 effector E. Korobko et al. 42 FEBS Journal 272 (2005) 37–46 ª 2004 FEBS in vivo. Taking into account that association of Rabap- tin-5 with Rabex-5 is essential for exerting Rab5 effec- tor functions, this finding suggests that Rabaptin-5d not merely binds Rab5, but similarly to Rabaptin-5, can function as its effector. In summary, our data suggest that whereas Rabaptin- 5 is a bifunctional protein interacting with two GTP- ases, Rab5 and Rab4, the Rabaptin-5d isoform lacks this bifunctionality. Our results indicate that Rabaptin- 5d is a Rab5 effector but unlike Rabaptin-5, does not interact with the Rab4, presumably due to a disrupted Rab4 binding site, and is thus unlikely to provide a link to a Rab4-positive domain. It is tempting to hypothesize that selective recruitment of the d Rabaptin-5 isoform by an early endosome would disfavor it moving along the early recycling pathway. This model should be veri- fied experimentally, and if confirmed, it would provide an example of membrane traffic regulation through selective recruitment of a minor Rab effector variant. Experimental procedures Plasmids To construct pPC97-Rab5, pPC97-Rab5Q79L and pPC97- Rab5S34N, the respective cDNAs were excised from pLexA-Rab5, pLexA-Rab5Q79L and pLexA-Rab5S34N [12] with EcoRI and NheI. The cDNAs with filled-in EcoRI site were subcloned between XbaI and filled-in BamHI sites in pBK-CMV vector (Stratagene, La Jolla, CA, USA), and subsequently excised and cloned between SalI and NotI sites in pPC97 vector [27] to obtained in-frame fusion with GAL4 DNA-binding domain (GAL4BD). To construct pPC97-Rab4, pPC97-Rab4Q67L and pPC97-Rab4S22N, the respective cDNAs were excised from pLexA-Rab4, pLexA-Rab4Q67L and pLexA-Rab4S22N [14] with EcoRI and SalI and cloned between EcoRI and XhoI sites in pBK-CMV vector. After digestion with SpeI, filling-in and self-ligation, the cDNAs were excised and cloned between the SalI and NotI sites in pPC97 vector to produce in-frame fusion with GAL4BD. pPC86-Rabaptin-5 plasmid with a pPC86 backbone [27] for expression of the GAL4 transcription activating domain (GAL4AD) linked to the full-length mouse Rabaptin-5 was described previ- ously [23]. pPC86-Rabaptin-5d was constructed in a similar way. To obtain pPC97-Rabaptin-5 and pPC97-Rabaptin-5d , the respective cDNAs from pPC86-Rabaptin-5 and pPC86- Rabaptin-5d were subcloned into the pPC97 vector between SalI and NotI sites. To construct deletion mutants of Rabaptin-5 in pPC86 vector, the Rabaptin-5 cDNA was subcloned from pPC86-Rabaptin-5 vector to pBlue- script SK II(+)plasmid (Stratagene) between SalI and NotI sites (plasmid pRn5). The resulting plasmid was used as a template to amplify cDNA fragments encoding amino acids 1–294 and 140–294 with the antisense primer 5¢-GTCTCA CATCAGCAAACGCT-3¢. As a sense primer, plasmid T7 primer or the primer 5 ¢-AGCAGGTCGACAGCACAGT GGGCACAGTAT-3¢ containing SalI site were used, respectively. Amplified sequences were cloned between SalI and SmaI sites in the pBluescript SK II(+) vector, se- quenced, and subcloned into the pPC86 vector between the SalI and NotI sites. pPC86 plasmid expressing amino acids 1–149 of Rabaptin-5 fused to GAL4AD was obtained by subcloning of the SalI-PstI5¢-end fragment of Rabaptin-5 cDNA from pRn5 between the SalI and PstI sites of pBlue- script SK II(+) vector with sequential subcloning of the SalI-NotI fragment into the pPC86 vector. Deletion mutants of Rabaptin-5d and Rabaptin-5c in the pPC86 vector were constructed in a similar way. To construct plasmids for expression of enhanced green fluor- escent protein (EGFP)-tagged Rabaptin-5 and Rabaptin-5d, the respective cDNAs were excised from the pRn5 and pRn5d plasmids with XhoI and NotI (the NotI site was blunted after digestion) and cloned between the XhoI and SmaI sites in pEGFP-C2 vector (Clontech. Palo Alto, CA, USA). Plasmids for expression of N-terminally FLAGÒ- tagged Rabaptin-5 and Rabaptin-5d were constructed in a following way. The 5¢-end of the Rabaptin-5 coding region from amino acid 2 was amplified from the pRn5 template with the antisense primer used to construct deletion mutants of Rabaptin-5 in the pPC86 vector and the Fig. 9. Mapping of the Rab4 binding site in Rabaptin-5 using the yeast two-hybrid protein–protein interaction assay. Indicated frag- ments of Rabaptin-5 (Rn5), Rabaptin-5d (Rn5d) or Rabaptin-5c (Rn5c) fused to GAL4AD or GAL4AD alone (none) were coex- pressed with GAL4BD fused to Rab4Q67L in yeast Y153. Amino aids of Rabaptin-5 or its isoforms fused to GAL4AD are shown in brackets, and deleted regions in Rabaptin-5 isoforms are presented as thin lines. LacZ reporter gene activation caused by interaction between GAL4BD- and GAL4AD-fused proteins (b-gal) in yeast was assessed by filter a b-galactosidase assay. A similar assay with GAL4BD alone as a bait revealed no activation of the LacZ reporter gene (data not shown). E. Korobko et al. Rabaptin-5 isoform d is not a Rab4 effector FEBS Journal 272 (2005) 37–46 ª 2004 FEBS 43 sense primer 5¢-ATGGTACCACTAGATCTGGCGCAG CC-3¢ containing KpnI and BglII sites, and subcloned between the KpnI and HindIII sites of pBluescript SK II(+) vector using the internal HindIII site (plasmid p5Rn5). The 3¢-end of the Rabaptin-5 coding region inclu- ding the translation termination codon was amplified with the sense primer 5¢-ACAAGGAATTCAGATTCAGGAA- 3¢ and the antisense primer 5¢-GCAGTCTAGAATCCTGC TATC-3¢ containing an XbaI site, and cloned between the EcoRI and XbaI sites of the p5Rn5 multiple cloning site using the internal EcoRI site in the amplified fragment (plasmid p53Rn5). All the amplified fragments of coding regions were sequenced to confirm their authenticities. The complete coding regions of Rabaptin-5 and Rabaptin-5d were obtained by cloning of ApaI-EcoRV internal fragments from pRn5 and pRn5d, respectively, into the p53Rn5 plas- mid (plasmids pRn5NF and pRn5dNF). Finally, the entire coding regions were excised from the pRn5NF and pRn5dNF plasmids with Bgl II and XbaI and cloned into the pFLAGÒ-CMV2 vector (Sigma, St Louis, MO, USA) to obtain plasmids for expression of N-termin- ally tagged Rabaptin-5 and Rabaptin-5d (plasmids pFLAGÒ-Rn5 and pFLAGÒ-Rn5d). To construct plasmids for expression of C-terminally FLAGÒ-tagged Rabaptin-5 and Rabaptin-5d, the 3¢-end of the open reading frame of Rabaptin-5 cDNA was amplified with a sense primer upstream of the unique EcoRV internal site and an anti- sense primer that was designed to substitute the Rabaptin-5 translation termination codon for a coding triplet followed by the FLAGÒ epitope sequence, a translation termin- ation codon and a NotI site. The amplified fragment was sequenced to confirm its authenticity. The amplified frag- ment was used to substitute the 3¢-ends in Rabaptin-5 and Rabaptin-5d cDNAs in the pRn5 and pRn5d plasmids. The resulting cDNAs encoding C-terminally FLAGÒ-tagged proteins were then excised with SalI and NotI and cloned between the XhoI and NotI sites of the pEGFP-N1 vector (Clontech) to produce expression plasmids for FLAGÒ- tagged Rabaptin-5 and Rabaptin-5d (the EGFP coding sequence was removed by digestion with XhoI and NotI). The plasmid for expression of myc-tagged Rab5Q79L was described previously [28]. For construction of the myc- Rab4Q67L expression vector, the Rab4Q67L cDNA from pPC97-Rab4Q67L was excised and cloned between the SalI and NotI sites into a pBK-CMV vector engineered to con- tain after the CMV promoter the fragment of the human preproinsulin cDNA 5¢-untranslated region and ATG codon (GeneBank Accession No. NM_000207, nucleo- tides 10–47) linked by an NcoI site with the NcoI-EcoRI fragment from the pECT plasmid encoding His 6 and myc tags. To construct plasmids for expression of glutathione S-transferase (GST)-tagged Rab4 and Rab5, pGEX-Rab4 and pGEX-Rab5, the respective cDNAs were excised with EcoRI and SalI from pLexA-Rab4 and pLexA-Rab5 and cloned into the pGEX-4T-1 vector (Amersham Biosciences, Chalfont St. Giles, UK). Yeast two-hybrid methods The protein interaction assay was performed as described [29]. In brief, the yeast strain Y153 was used to cotrans- form bait and prey plasmids, and transformants were selec- ted by plating onto synthetic dextrose (SD) leucine and tryptophan dropout plate. To evaluate HIS3 reporter gene trans-activation, yeast from a single colony was spotted onto SD medium lacking leucine and tryptophan, and sup- plemented or not with 25 mm 3-amino-(1,2,4)-triazole (3AT), and grown at 30 °C. The HIS3 reporter gene trans- activation was monitored by the ability of yeast to grow on 3AT-containing medium or LacZ reporter gene trans-acti- vation was assessed in filter b-galactosidase assay. Cells and transfection BHK-21 cells were maintained in Dulbecco’s modified Eagle’s medium (Invitrogen, Carlsbad, CA, USA) supple- mented with 10% fetal bovine serum, 110 mgÆL )1 sodium pyruvate and 100 unitsÆmL )1 penicillin ⁄ 100 lgÆ mL )1 strep- tomycin. For transfections, cells were plated at 100 mm cell culture dish at a density 1.2  10 6 cells per dish. For immu- nofluorescence microscopy, cells were plated on 10 mm cov- erslips in six-well plates at a density 200 000 cell per well. The next day, cells were transfected with 0.25 lg of plasmid DNA per 0.75 lL Unifectin-56 liposome transfection rea- gent (kindly provided by A. Surovoy, Shemyakin’s and Ovchinnikov’s Institute of Bioorganic Chemistry, Moscow, Russia) for a six-well plate, or with 1 lg of plasmid DNA per 3 lL Unifectin-56 liposome transfection reagent for a 100 mm dish. Antibodies The anti-Rabex-5 rabbit polyclonal antiserum [13] was a gift from M. Zerial (Max-Plank Institute of Molecular Cell Biology and Genetics, Dresden, Germany). Mouse anti- (EGFP 2G7) monoclonal Igs were kindly provided by A. Surovoy (Moscow, Russia). Mouse anti-(myc 9E10) mono- clonal Igs were used as supernatant from the respective hybridoma, and Alexa546-conjugated anti-mouse Igs (Molecular Probes, . To detect the FLAGÒ-epitope, rabbit anti-FLAGÒ polyclonal antibodies (Sigma) were used following Alexa-488-conjugated antirabbit antibodies (Molecular Probes, Eugene, OR, USA). Rabbit polyclonal anti-(Rabaptin-5 Ig) antiserum was raised using a His 6 - tagged mouse Rabaptin-5 fragment (amino acids 407–663) as the immunogen. To produce and purify recombinant protein, QIAexpressionist expression and purification system (Qiagen, Valencia, CA, USA) was used. Rabaptin-5 isoform d is not a Rab4 effector E. Korobko et al. 44 FEBS Journal 272 (2005) 37–46 ª 2004 FEBS Confocal immunofluorescence microscopy Twenty-four hours after transfection, cells on coverslips were washed in phosphate-buffered saline (PBS) and fixed with 3% (w ⁄ v) paraformaldehyde. Free aldehyde groups were quenched with 50 mm ammonium chloride, and cells were permeabilized with 0.05% (w ⁄ v) saponin (Sigma). After permeabilization, coverslips were washed with PBS and incubated with primary antibodies diluted in PBS con- taining 5% (w ⁄ v) nonfat dry milk and 0.1% (w ⁄ v) Tween 20 for 1 h. After washing with PBS coverslips were incubated as above with secondary antibody solution, washed and mounted in Mowiol (EMD Biosciences, Inc., San Diego, CA, USA). Coverslips were examined with Leica (Wetzlar, Germany) or Radiance 2100 (Bio-Rad, Hemmel Hempstead, UK) confocal microscopes and images were taken at Â100 magnification and captured at 1024  1024 pixels or at Â60 magnification and captured at 512  512 pixels, respectively. Montages of images were prepared with use of photoshop 5.0 (Adobe, Mountain View, CA, USA). Preparation of cytosol Cytosol for GST pull-down experiments was prepared as described previously [12] with minor modifications. Thirty- six hours after transfection, cells from four 100 mm (diam.) cell culture dishes were scraped into PBS, then pelleted and homogenized in 400 lL of 250 mm sucrose, 10 mm sodium phosphate, pH 7.2 by passages through a 27 gauge needle attached to 1 mL syringe. Cell breakage and nucleus integ- rity were monitored by phase-contrast microscopy. Nuclei and debris were pelleted by centrifugation at 4000 r.p.m. for 10 min in an Eppendorf microcentrifuge. The postnu- clear supernatant was centrifuged at 60 000 r.p.m. for 1 h at 4 °C in a Beckman TLA-100 rotor to obtain cytosol. Cytosol for coimmunoprecipitation with Rabex-5 was obtained in a similar way but cells were homogenized in a buffer known to preserve the Rabaptin-5–Rabex-5 com- plex [13] [20 mm Hepes ⁄ KOH, pH 7.2, 5 mm MgCl 2 , 1mm dithiothreitol, 100 mm NaCl, 1 mm EDTA contain- ing Protease Inhibitor Cocktail (Sigma)]. Recombinant proteins GST-Rab4 and GST-Rab5 were produced in Escheri- chia coli BL21 carrying pGEX-Rab4 and pGEX-Rab5 plas- mids. Production and purification of GST-tagged proteins were carried out on Gluthatione Sepharose 4B (APBiotech) according to the manufacturer’s instructions. Eluted pro- teins were extensively dialyzed against 50 mm tris ⁄ HCl, pH 8.0, 135 mm NaCl, 1 mm EDTA followed by dialysis against the same buffer containing 50% (v ⁄ v) glycerol, and stored at )20 °C. The final protein concentration was about 2mgÆmL )1 with purity over 95%. GST pull-down assay GST pull-down assays with GDP- or GTPcS-loaded GST– Rab4 and GST–Rab5 were performed essentially as des- cribed [11]. Binding and washing were performed in batch, and cytosols prepared from two 100 mm (diam.) dishes of BHK-21 transiently transfected with plasmids for expression of EGFP- or FLAGÒ-tagged proteins were used to pull down EGFP- or FLAGÒ-tagged Rabaptin-5 or Rabaptin- 5d. Before addition of cytosols to immobilized and nucleo- tide-preloaded GST-Rab4 or GST-Rab5, cytosols were diluted two-fold to adjust a final binding buffer composition. After washing, SDS ⁄ PAGE loading buffer was added to Sepharose beads, samples were boiled and loaded onto a 7.5% SDS ⁄ polyacrylamide gel. After separation, proteins were transferred onto Immobilon P poly(vinylidene difluo- ride) (PVDF) membrane (Millipore, Billerica, MA, USA), and membrane was immunoblotted with rabbit polyclonal anti-(Rabaptin-5 Ig) antiserum for EGFP-tagged proteins or with monoclonal anti-(FLAGÒ-M2) Igs (Sigma) for FLAGÒ-tagged proteins. Aliquots of cytosols were also immunoblotted with respective antibodies to monitor input protein contents. Coimmunoprecipitation and immunoblotting For coimmunoprecipitation of EGFP-tagged Rabaptin-5 or Rabaptin-5d with Rabex-5, cytosol was prepared from four 100 mm dishes. Cytosol was incubated for 1 h with Protein G Sepharose beads (Amersham Biosciences) precoated with 2 lg of mouse monoclonal anti-(EGFP 2G7) Igs, and washed three times with buffer used for cytosol preparation. After addition of SDS ⁄ PAGE loading buffer, samples were boiled and loaded onto a 7.5% SDS ⁄ polyacrylamide gel. After separation, proteins were transferred onto Immobilon P PVDF membrane (Millipore), and membrane was immuno- blotted with rabbit polyclonal anti-(Rabex-5 Ig) serum or rabbit polyclonal anti-(Rabaptin-5 Ig) serum. Aliquots of cytosols were also immunoblotted with anti-Rabex-5 serum or anti-Rabaptin-5 serum to monitor input protein contents. Acknowledgements We thank Dr Marino Zerial (Max-Plank Institute of Molecular Cell Biology and Genetics, Dresden, Ger- many) for the anti-Rabex5 antiserum and Dr Andrey Surovoy (Shemyakin’s and Ovchinnikov’s Institute of Bioorganic Chemistry, Moscow, Russia) for anti- EGFP Igs and transfection reagents. This work was supported by grants from the Russian Foundation for Basic Research and the Physical and Chemical Biology Program of the Russian Academy of Sciences. E.K. was partially supported by a FEBS Short-Term Travel Fellowship. E. Korobko et al. Rabaptin-5 isoform d is not a Rab4 effector FEBS Journal 272 (2005) 37–46 ª 2004 FEBS 45 References 1 Stenmark H & Olkkonen VM (2001) The Rab GTPase family. Genome Biol 2, 3007.1–3007.7. 2 Sollner T, Whiteheart SW, Brunner M, Erdjument- Bromage H, Geromanos S, Tempst P & Rothman JE (1993) SNAP receptors implicated in vesicle targeting and fusion. Nature 362, 318–324. 3 Rothman JE (1994) Mechanisms of intracellular protein transport. Nature 372, 55–63. 4 Weber T, Zemelman BV, McNew JA, Westermann B, Gmachl M, Parlati F, Sollner TH & Rothman JE (1998) SNAREpins: minimal machinery for membrane fusion. Cell 92, 759–772. 5 Rubino M, Miaczynskz M, Lippe R & Zerial M (2000) Selective membrane recruitment of EEA1 suggests a role in directional transport of clathrin-coated vesicles to early endosomes. J Biol Chem 275, 3745–3748. 6 Rybin V, Ullrich O, Rubino M, Alexandrov K, Simon I, Seabra MC, Goody R & Zerial M (1996) GTPase activ- ity of Rab5 acts as a timer for endocytic membrane fusion. Nature 383, 266–269. 7 McLauchlan H, Newell J, Morrice N, Osborne A, West M & Smythe E (1998) A novel role for Rab5-GDI in ligand sequestration into clathrin-coated pits. Curr Biol 8, 34–45. 8 Gorvel J-P, Chavrier P, Zerial M & Gruenberg J (1991) rab5 controls early endosome fusion in vitro. Cell 64, 915–925. 9 Bicci C, Parton RG, Mather IH, Stunnenberg H, Simons K, Hoflack B & Zerial M (1992) The small GTPase rab5 functions as a regulatory factor in the early endocytic pathway. Cell 70, 715–728. 10 Nielsen E, Severin F, Backer JM, Hyman AA & Zerial M (1999) Rab5 regulates motility of early endosomes on microtubules. Nat Cell Biol 1, 376–382. 11 De Renzis S, Soennichsen B & Zerial M (2002) Divalent Rab effectors regulate the sub-compartmental organiza- tion and sorting of early endosomes. Nat Cell Biol 4, 124–132. 12 Stenmark H, Vitale G, Ullrich O & Zerial M (1995) Rabaptin-5 is a direct effector of the small GTPase Rab5 in endocytic membrane fusion. Cell 83, 423–432. 13 Horiuchi H, Lippe R, McBride H, Rubino M, Wood- man P, Stenmark H, Rybin V, Wilm M, Ashman K, Mann M & Zerial M (1997) A novel Rab5 GDP ⁄ GTP exchange factor complexed to Rabaptin-5 links nucleo- tide exchange to effector recruitment and function. Cell 90, 1149–1159. 14 Vitale, G, Rybin V, Christoforidis S, Thornqvist P-O, McCaffrey Stenmark H & Zerial M (1998) Distinct Rab- binding domains mediate the interaction of Rabaptin-5 with GTP-bound rab4 and rab5. EMBO J 17, 1941–1951. 15 Lippe R., Miaczynska M, Rybin V, Runge A & Zerial M (2001) Functional synergy between Rab5 effectors Rabaptin-5 and exchange factor Rabex-5 when physically associated in a complex. Mol Biol Cell 12, 2219–2228. 16 Van der Sluijs P, Hull M, Webster P, Male P, Goud B & Mellman I (1992) The small GTP-binding protein rab4 controls an early sorting event on the endocytic pathway. Cell 70, 729–740. 17 Sheff DE, Daro EA, Hull M & Mellman I (1999) The recycling pathway contains two distinct populations of endosomes with different sorting functions. J Cell Biol 145, 123–139. 18 Ohya T, Sasaki T, Kato M & Takai Y (1998) Involve- ment of Rabphilin3 in endocytosis through interaction with Rabaptin5. J Biol Chem 273, 613–617. 19 Coppola T, Hirling H, Perret-Menoud V, Gattesco S, Catsicas S, Joberty G, Macara IG & Regazzi R. (2001) Rabphilin dissociated from Rab3 promotes endocytosis through interaction with Rabaptin-5. J Cell Sci 114, 1757–1764. 20 Shiba Y, Takatsu H, Shin H-W & Nakayama K (2002) g-Adaptin interacts directly with Rabaptin-5 through its ear-domain. J Biochem 131, 327–336. 21 Mattera R, Arighi CN, Lodge R, Zerial M & Bonifa- cino JS (2003) Divalent interaction of the GGAs with the Rabaptin-5-Rabex-5 complex. EMBO J 22, 78–88. 22 Deneka M & van der Sluijs P (2002) ‘Rab’ing up endo- somal membrane transport. Nat Cell Biol 4, E33–E35. 23 Korobko EV, Kiselev SL & Korobko IV (2002) Multi- ple Rabaptin-5-like transcripts. Gene 292, 191–197. 24 Nagelkerken B, van Anken E, van Raak M, Gerez L, Mohrmann K, van Uden N, Holthuizen J, Pelkmans L & van der Sluijs P (2000) Rabaptin4, a novel effector of the small GTPase rab4a, is recruited to perinuclear recy- cling vesicles. Biochem J 346, 593–601. 25 Kawasaki T, Kunisato A, Hazama K, Uyeda A & Taguchi T (2001) Identification of active regions for neurite outgrowth activity of neurocrescin. Biochem Bio- phys Res Commun 281, 761–765. 26 Deneka M, Neeft M, Popa I, van Oort M, Sprong H, Oorschot V, Klumperman J, Schu P & van der Sluijs P (2003) Rabaptin-5alpha ⁄ rabaptin-4 serves as a linker between rab4 and gamma (1) -adaptin in membrane recycling from endosomes. EMBO J 22, 2645–2657. 27 Chevray PM & Nathans D (1992) Protein intaraction cloning in yeast: identification of mammalian proteins that interact with the leucine zipper of Jun. Proc Natl Acad Sci USA 89, 5789–5793. 28 Stenmark H, Parton RG, Steele-Mortimer O, Lutcke A, Gruenberg J & Zerial M (1994) Inhibition of rab5 GTPase activity stimulates membrane fusion in endo- cytosis. EMBO J 13, 1287–1296. 29 Dufree T, Becherer K, Chen P-L, Yeh S-H, Yang Y, Kilburn AE, Lee W-H & Elledge SJ (1993) The retinoblastoma protein associates with the protein phosphatase type I catalytic domain. Genes Dev 7 , 555–569. Rabaptin-5 isoform d is not a Rab4 effector E. Korobko et al. 46 FEBS Journal 272 (2005) 37–46 ª 2004 FEBS . The Rab5 effector Rabaptin-5 and its isoform Rabaptin-5d differ in their ability to interact with the small GTPase Rab4 Elena Korobko 1,2 ,. assay to analyze the interaction of Rabaptin-5d with Rab4 and Rab5. Similarly to Rabaptin-5, Rabaptin-5d was not able to bind Rab4 or Rab5 in the inactive, GDP-bound form