Identification of a Novel Rab11/25 Binding Domain Present in Eferin and Rip Proteins*

Rab11, a low molecular weight GTP-binding protein, has been shown to play a key role in a variety of cellular processes, including endosomal recycling, phagocytosis, and transport of secretory proteins from the trans-Golgi network. In this study we have described a novel Rab11 effector,EF-hands-containingRab11-interacting protein (Eferin). In addition, we have identified a 20-amino acid domain that is present at the C terminus of Eferin and other Rab11/25-interacting proteins, such as Rip11 and nRip11. Using biochemical techniques we have demonstrated that this domain is necessary and sufficient for Rab11 binding in vitro and that it is required for localization of Rab11 effector proteins in vivo. The data suggest that various Rab effectors compete with each other for binding to Rab11/25 possibly accounting for the diversity of Rab11 functions.

Members of the Rab/Ypt GTPase family have emerged as important regulators of vesicular trafficking (1). Rab proteins have been proposed to mediate a variety of functions, including vesicle translocation and docking at specific fusion sites. Like all small GTPases, Rabs cycle between active (GTP-bound) and inactive (GDP-bound) conformations (2). In the GTP-bound state, Rab proteins can bind a variety of downstream effector proteins, while GTP hydrolysis leads to a conformational change in the switch region that renders the Rab GTPase unrecognizable to its effector proteins (3,4). A key question in understanding the interactions between Rabs and their effectors concerns the mechanisms by which Rab GTPases specifically bind a diverse spectrum of effectors and how this is regulated by the common structural motif used as a GTP switch. Biochemical and genetic studies have identified several hypervariable regions that might be involved in determining Rab specificity, including N and C termini, as well as the ␣3/␤5 loop (5,6). Indeed, the recently reported structure of Rab3a bound to a putative effector, rabphilin-3a, revealed that the Rab3a-rabphilin-3a complex interacts through two main regions (7). The first consists of conformationally sensitive switch regions of Rab3a bound to the a1 helix and the C-terminal part of rabphilin-3a. The second involves the SGAWFF domain of rabphilin-3a, which fits into a pocket formed by the three hypervariable complementary determining regions (CDRs) 1 of Rab3a, corresponding to the N and C termini and the ␣3/␤5 loop. Thus, it appears that the hypervariable RabCDR is involved in determining the specificity of effector binding, while the conserved switch regions impart GTP dependence and binding. It remains to be determined, however, whether this paradigm also applies to other Rab-effector complexes.
Rab11a, 11b, and 25 are closely related members of Rab GTPase family that have been implicated in regulating a variety of different post-Golgi trafficking pathways, such as protein recycling (8), phagocytosis (9), insulin-stimulated Glut4 insertion in the plasma membrane (10), and membrane trafficking from early endosomes to the trans-Golgi network (11). During the last few years several Rab11/25-interacting proteins have been identified, including Rab11BP/rabphilin-11, Rip11, nRip11, and myosin Vb (12)(13)(14)(15). However, the mechanisms of their function, as well as molecular aspects of their interactions with Rab11, remain to be fully understood. In the present study, we report the identification of EF-hands-containing Rab11/25-interacting protein (Eferin). Furthermore, we characterized a Rab binding domain (RBD11) that is present at the C terminus of Eferin as well as other Rab11/25-binding proteins, such as Rip11 and nRip11. Using biochemical techniques, we demonstrated that RBD11 is the region that encodes the specificity for Rab11/25 but is distinct from the region interacting with the Rab switch domain, since its interactions with the Rab11/25 are not GTP-dependent.

EXPERIMENTAL PROCEDURES
Materials and Antibodies-Cell culture reagents were obtained from Life Technologies, Inc. unless otherwise specified. Miscellaneous chemicals were obtained from Sigma. Actin-rhodamine conjugates were purchased from Molecular Probes (Eugene, OR). Mouse monoclonal anti-Myc antibody (9E10) was obtained from Santa Cruz Biotechnology Inc. (Santa Cruz, CA). Fluorescein isothiocyanate-labeled anti-rabbit IgG and Texas Red-labeled anti-mouse IgG antibodies were obtained from Jackson ImmunoResearch Laboratories (West Grove, PA).
Cell Culture and Immunofluorescence Microscopy-Madin-Darby canine kidney (MDCK II) cells were cultured, and immunofluorescence microscopy was performed as described previously (16). For immunofluorescence microscopy, cells were fixed with 4% paraformaldehyde for 15 min. Cells were then permeabilized in 0.4% saponin and nonspecific sites blocked with phosphate-buffered saline containing 0.2% BSA, 0.4% saponin, and 1% bovine serum. Following incubation with antibodies, samples were extensively washed and mounted in VectaShield (Vector Laboratories, Burlingame, CA). Immunofluorescence localization was performed using a Molecular Dynamics laser confocal imaging system (Beckman Center Imaging Facility, Stanford University, Stanford, CA). * 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.
The nucleotide sequence(s) reported in this paper has been submitted to the GenBank TM  MDCK cells were transfected using the LipofectAMINE 2000 system (Life Technologies Inc.). At 24-h post-transfection, cells were trypsinized and resuspended in low Ca 2ϩ (5 M) minimum Eagle's medium with Earle's salts. 1.6 ϫ 10 5 cells were seeded into one well of a 24-well collagen-coated Transwell filter (6.5-mm diameter, 0.4-mm pore size; Corning-Costar Corp., Corning, NY). After 5 h, the low Ca 2ϩ medium was changed to Dulbecco's modified Eagle's medium with 10% fetal bovine serum. Cells were cultivated for 2-3 days to grow a confluent and fully polarized cell layer before imaging them with a Bio-Rad laser confocal imaging system (Beckman Center Imaging Facility, Stanford University). X-Z views were obtained by averaging sections over a line at each z position in 0.5 mm steps.
GST and BSA Pull-down Assays-For pull-down assays, Rab11a or RBD11 expressed as GST fusion proteins were bound to glutathioneagarose beads by incubating at room temperature for 1 h. Where indicated, BSA-RBD11 conjugates were bound to Affi-Gel beads (Bio-Rad) and used for the pull-down assays. Beads were extensively washed, resuspended in reaction buffer (50 mM Hepes, pH 7.4, 100 mM NaCl, 1 mM MgCl 2 , and 1 mM dithiothreitol), and incubated either with recombinant proteins or with HeLa cell Triton X-100 extracts (13). Beads were then extensively washed and binding proteins eluted with 1% SDS. Samples were then separated by SDS-polyacrylamide gel electrophoresis and analyzed either by Western blotting or Coomassie Blue staining.
Yeast Two-hybrid Screens-The bait construct was prepared by subcloning full-length Rab25 into the pGBKT7 plasmid. The construct was used to transform AH109 yeast cells according to the manufacturer's protocol (CLONTECH, Palo Alto, CA). A human kidney cDNA library was screened by mating Y187 yeast cells with AH109 cells expressing Rab25. After incubation for at least 48 h at 30°C, prominent colonies were spotted onto fresh double (synthetic dropout media-Trp/-Leu), triple (synthetic dropout media-Trp/-Leu/-His), and quadruple (synthetic dropout media-Trp/-Leu/-His/-Ade) dropout media and checked for ␤-galactosidase expression by the X-gal filter assay. DNA from each positive clone was extracted and sequenced. As a further test to eliminate false positives, isolated prey vectors were co-transformed with bait plasmids and checked again for reporter gene expression. For protein interaction studies, a series of bait constructs was prepared using the following cDNA: Rab11 wild-type, Rab11-S25N, Rab17, Rab25, and Sec8. For prey constructs, the following cDNA were inserted into pAct2: Rip11-(566 -653), nRip11-(425-511), and Eferin-(665-756). AH109 yeast was pairwise co-transformed with bait and prey constructs and grown on double dropout media. Clones were transferred to increasingly stringent selection media as described above. Images of colony growth were obtained using a Linotype-Hell transparency scanner in conjunction with Adobe Photoshop.

Eferin Is a Novel Rab11 and Rab25-binding Protein-
Despite the accumulating evidence implicating Rab11/25 GTPases in regulating multiple membrane trafficking pathways, we still know very little about the mechanism of their function. In an attempt to isolate additional Rab11/25 effectors we screened a human kidney cDNA library using Rab25 as bait in the yeast two-hybrid system. This screen resulted in the isolation of 25 cDNA clones that specifically interacted with Rab11 and Rab25. 15 out of 25 clones were identified as KIAA0665, a 759-amino acid open reading frame of unknown function (17). Amino acid sequence analysis using PROSITE and PFAM revealed that KIAA0665 contains an N-terminal proline-rich region, as well as two EF-hand motifs (Fig. 1B). Thus, we will be referring to KIAA0665 as Eferin (AF395731), an EF-hands-containing Rab11/25-interacting protein.
To determine whether Eferin co-localizes with Rab11 in living cells, we transiently co-transfected normal rat kidney (NRK) cells with Eferin-GFP and Myc-Rab11a cDNAs. As shown in Fig. 1, C and D, Eferin-GFP shows a predominately perinuclear staining, reminiscent of recycling endosomes and the trans-Golgi network (Fig. 1C) and largely overlapping with Myc-Rab11a (Fig. 1D). Furthermore, immunofluorescence studies using transiently transfected MDCK cells showed that, similarly to Rab11, in epithelial cells Eferin-GFP is localized to the apical pole (Fig. 1E).
Identification of a Novel Rab11 Binding Domain (RBD11) Present in Rip and Eferin Proteins-In addition to Eferin, the yeast two-hybrid screen yielded 10 other clones that specifically interacted with Rab11 and Rab25. Seven of these clones encode fragments of Rip11 (AF334812), while the other three were identified as nRip11 (AY037299). Rip11 (13) and nRip11 2 are previously identified related Rab11/25-binding proteins. The data from the yeast two-hybrid screens (Fig. 1, A and B) demonstrate that the information necessary to ensure Rab specificity and GTP dependence is encoded within the last C-terminal 86 amino acids of Rip proteins. This result was somewhat unexpected, since we have previously reported that Rip11 also binds to Rab11a through the middle region of the protein (Rip11-(200 -507)) (13). To test whether both regions of Rip11 are involved in association with Rab11, we performed GST-Rab11a pull-down assays using in vitro translated fragments of Rip11 and nRip11. As shown in Fig. 2C, the binding of Rip11-(200 -507) is much weaker compared with that of Rip11-(507-653). Furthermore, the region in nRip11 equivalent to Rip11-(200 -507) failed to bind to GST-Rab11a (Fig. 2C)  teins, the alignment of Rab11/25-interacting domains revealed that Eferin, Rip11, and nRip11 contain a highly conserved domain of 20 amino acids at the very C terminus of the protein (Fig. 3A). Deletion of these residues in Rip11 and Eferin resulted in loss of binding to Rab11 (Figs. 2C and 3B), suggesting that these amino acid residues might directly participate in Rab11/25 binding. To address that possibility we synthesized the peptide corresponding to Rip11-(628 -645) (Rab11/25 binding domain, RBD11) and used it in competition assays. As shown in Fig. 3C, while the Rip11 N-terminal peptide (amino acids 1-17) had no effect on Rip11 binding to GST-Rab11a, the addition of RBD11 partially inhibited Rip11/Rab11 interaction. Furthermore, RBD11 also plays a role in Eferin binding to Rab11/25 since it inhibits Myc-Eferin co-precipitation with Rab25-GFP (Fig. 3D).
While our data suggest that RBD11 is involved in Rab11/25 binding in vitro, it remains to be determined whether it is involved in Rip11 trafficking in living cells. To address that issue we transiently transfected MDCK cells with GFP fused to either full-length Rip11 (Rip11-GFP), or Rip11 lacking the RBD11 domain (Rip11⌬RBD11-GFP) (Fig. 4). As previously reported (13), in non-polarized MDCK cells Rip11-GFP was localized to perinuclear endosomes, which also contained Rab11a (Fig. 4, A-C). Furthermore, in polarized MDCK cells, Rip11-GFP was present at the apical pole of the cell, suggesting that GFP tagging did not affect the Rip11 subcellular distribution (Fig. 4, G-I). Deletion of RBD11, on the other hand, dramatically affected the subcellular distribution of Rip11-GFP. As shown in Fig. 4, D-F, in non-polarized cells Rip11⌬RBD11-GFP is present predominately on the plasma membrane and shows no co-localization with Rab11a-containing endosomes. Furthermore, in polarized cells Rip11⌬RBD11-GFP has lost its polarized distribution and is present throughout the entire cell ( Fig. 4, J--L). As in non-polarized cells, a large portion of Rip11⌬RBD11-GFP is present on the plasma membrane, although some Rip11⌬RBD11-GFP also shows cytosolic staining (Fig. 4J).
RBD11 Is Necessary and Sufficient to Mediate Specific Interactions between Rip11 and Rab11-The data presented above demonstrate that RBD11 is necessary for Rab11/25 binding to Eferin and Rip proteins. To determine whether RBD11 is sufficient for Rab-specific and GTP-dependent Rab11/25 binding, the peptides corresponding to either Rip11-RBD11 or Rip11-(1-17) (Fig. 5A, N pept) were conjugated to BSA-coated agarose beads and used in Rab11 pull-down assays. As shown in Fig.  5A, while Rab11 did not co-sediment with BSA alone or BSA-N pept, it bound to BSA-RBD11. The binding was specific since it was observed only with Rab11a (Fig. 5B) and Rab11b (data not shown) but not Rab1a and Rab3a. Furthermore, the BSA-RBD11/Rab11a interactions could be inhibited with soluble RBD11 in a concentration-dependent manner (Fig. 5C). To test whether full-length RBD11 is necessary for Rab11/25 interactions, we expressed RBD11 as a GST fusion protein. In agreement with data reported above, Rab11a also interacted with GST-RBD11. Truncations of the RBD11 motif resulted in decreased Rab11a binding, suggesting that intact RBD11 is necessary for binding to Rab GTPases (Fig. 5E). Surprisingly, while BSA-RBD11 specifically interacted with Rab11/25 proteins, it showed no GTP dependence (Fig. 5D). Perhaps, RBD11 determines Rab specificity by interacting with RabCDR, while additional motifs interacting with Rab switch I/II domains are responsible for the GTP-dependent component.

DISCUSSION
Rab11 is a small GTPase that was implicated in regulating a variety of distinct membrane trafficking steps (8 -10). How- ever, the molecular mechanisms involved in regulation and determining the specificity of Rab11 functions remain to be understood. The ever growing number of Rab11-binding proteins suggests that sequential or competitive interactions of Rab11 with different effector proteins might account for the diversity of Rab11 functions. In this study, we have identified a novel Rab11/25-interacting protein and named it Eferin. Several lines of evidence suggest that Eferin is an effector for Rab11/25 GTPases. First, Eferin binds specifically to Rab11/25 but not other Rabs. Second, Eferin preferentially associates with the GTP-bound, thus active, form of Rab11. Third, Eferin co-localizes with Rab11 and Rab25 (data not shown). Thus, our data suggest that Eferin is an effector protein for Rab11/25 GTPases, although its role in membrane trafficking remains to be elucidated.
Immunofluorescence studies using transiently transfected MDCK cells showed that Eferin-GFP is localized exclusively to the apical plasma membrane, suggesting that Eferin might be involved in regulation of apical targeting. Interestingly, besides EF-hands, Eferin also contains a region resembling the Cterminal domain of the ezrin/radixin/moesin protein family, also known as ERM proteins, which also localize near the apical plasma membrane in actin-rich cytoskeletal structures (18 -20). ERM proteins can form homo-and heterodimers via C-terminal interactions with the N-terminal FERM domain (21). The FERM/C-terminal interaction masks the binding sites for other molecules, in this way regulating ERM protein association with the cytoskeleton (22). Thus, it is tempting to speculate that the Rab11-Eferin complex might interact with the FERM domain of the ERM proteins, in this way regulating ERM protein activity and cross-linking membranes to the cytoskeleton.
The ever growing Rab11 binding protein family already includes five members. The ability to interact with several effector proteins seems to be a common feature of many Rab GTPases (23). The main challenge of future studies will be to determine the functions of all of these proteins, as well as to understand the mechanisms of their interactions with Rab proteins. Indeed, it remains unclear whether the effector proteins compete with each other for binding to Rabs or work in a consecutive fashion. The work presented here suggests that Rips and Eferin compete for interactions with Rab11/25. Indeed, identification of a common Rab11/25 binding domain indicates that Eferin and Rips use the same binding site on Rab11/25. Interestingly, RBD11 is not present in the other known Rab11/25 effector proteins, such as myosin Vb and . Bound proteins were analyzed as described above. D, to analyze the GTP dependence of Rab11 binding to RBD11, varying concentrations of Rab11a were incubated with Affi-Beads conjugated to BSA-RBD11 in the presence of either GTP␥S (top row) or GDP␤S (bottom row). E, to further map the RBD11 binding domain, DNA coding for wild-type Rip11-RBD11 and several RBD11 truncation/deletion mutants were fused to GST. Fusion proteins were purified and used in GST pull-down assays to determine their ability to bind recombinant Rab11a. Recombinant Rab11a was bound for 1 h at 4°C to glutathione beads coated with equal amounts of various GST-RBD11 constructs. The beads were then washed and samples eluted with 1% SDS, followed by separation on SDS-polyacrylamide gels and staining with Coomassie Blue. Gels were then scanned, and the amount of Rab11a bound was determined using IQMac v1.2 software. The values were expressed as a percentage of Rab11a bound to wild type GST-RBD11 fusion protein.
Rab11BP/rabphilin-11 (12,14,15). It will be interesting to see whether these proteins can actually co-bind to Rab11/25 with either Eferin or Rips, perhaps forming signaling complexes that regulate transport vesicle trafficking.
While RBD11 appears to be involved in determining the specificity of interactions with Rab GTPases, its binding is independent of GTP. Perhaps, RBD11 is an equivalent of a SGAWFF motif in rabphilin-3a, which interacts with Rab3a CDRs (7). If this is the case additional motifs will be required to interact with Rab switch I/II domains to provide the GTP-dependent component of binding. Thus, it is likely that Eferin and Rip proteins interact with Rab11/25 via several motifs. At least some of the GTP-sensing motifs are encoded within the 60amino acid domain located upstream of RBD11. Interestingly, while the putative GTP-sensing regions in Rip11 and nRip11 are very highly conserved, they have no apparent homology to Eferin. Thus Rip and Eferin proteins share the common motif involved in Rab11/25 recognition but may use different mechanisms for mediating the GTP-dependent component of Rab11/25 binding.
The functional significance of the differences in Rip and Eferin interactions with Rab11/25 remains to be determined. One possibility is that additional cellular factors can regulate the affinity of Rab11/25 binding to its effectors. Indeed, the recombinant full-length Rip11 binds poorly to Rab11a in pulldown and yeast two-hybrid assays as compared with full-length endogenous Rip11 from cellular TX-100 extracts (data not shown). Furthermore, it has been previously shown that Rip11 can also interact with ␥SNAP and cytoskeleton (13,24). Thus, the interactions of Rips and Eferin with different factors could be used as a means of differentially regulating Rab11/25 binding. Alternatively, the Rab11/25 binding motif in Eferin and Rip11 might be conformationally hidden and require activation before binding to Rab11/25. We have previously demonstrated that phosphorylation of Rip11 plays an important role in its trafficking (13). Thus, differential phosphorylation on Rab11/25 binding motifs could also play a role in regulating the binding of Rip11 and Eferin to Rab GTPases.
Despite the recent progress in understanding the roles of Rabs and their effectors in regulating membrane trafficking, we are only beginning to unravel the structural determinants of their function. Identification and characterization of the Rab11/25 binding regions in Rip and Eferin proteins will be of crucial importance in understanding the molecular mechanisms involved in differential regulation of the variety of Rab11dependent trafficking pathways.