Rab Coupling Protein (RCP), a Novel Rab4 and Rab11 Effector Protein*

Rab4 and Rab11 are small GTPases belonging to the Ras superfamily. They both function as regulators along the receptor recycling pathway. We have identified a novel 80-kDa protein that interacts specifically with the GTP-bound conformation of Rab4, and subsequent work has shown that it also interacts strongly with Rab11. We name this protein Rab coupling protein (RCP). RCP is predominantly membrane-bound and is expressed in all cell lines and tissues tested. It colocalizes with early endosomal markers including Rab4 and Rab11 as well as with the transferrin receptor. Overexpression of the carboxyl-terminal region of RCP, which contains the Rab4- and Rab11-interacting domain, results in a dramatic tubulation of the transferrin compartment. Furthermore, expression of this mutant causes a significant reduction in endosomal recycling without affecting ligand uptake or degradation in quantitative assays. RCP is a homologue of Rip11 and therefore belongs to the recently described Rab11-FIP family.

To further understand the function of Rab4, a yeast twohybrid screen was performed to identify proteins that interact with the GTP-bound form of Rab4. Here, we report the identification of a novel, ϳ80-kDa protein that interacts with Rab4 and Rab11. We localize the endogenous protein to the early endosomal recycling compartment (ERC) and show that overexpression of a deletion mutant of this protein perturbs the morphology of this compartment and strongly inhibits endosomal recycling. Because of its interaction with two Rabs that function in endosomal recycling, its localization to the early endosomal recycling compartment, and its functional effects on the recycling process, we propose to name this the Rab coupling protein (RCP). Interestingly, this protein is related to the Rab11-interacting protein (Rip11) (27) and, therefore, belongs to the recently reported Rab11-FIP protein family (29,30).

MATERIALS AND METHODS
cDNA Cloning and Plasmid Construction-The Rab4 two-hybrid constructs have been described previously (24). pLex-Rab5, pLex-* This project was supported by Health Research Board of Ireland Grant 19/96 (to M. M.) and European Community Grants CHRX-CT94-0592 (to M. M. and B. G.) and FMRX-CT96-0020 (to M. M., C. B., and B. G.). 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 /EBI Data Bank with accession number(s) AF368294.
Two-hybrid Screening-The yeast two-hybrid screen was performed as previously described (40). Briefly, the yeast reporter strain L40 (37) (MATa trp1 leu2 his3::lexA-His3 URA3::lexA-lacZ) was transformed with pVJL10-Rab4Q67L-CXC and grown on synthetic medium lacking tryptophan. This transformant was then transformed with an oligo-dTprimed HeLa cDNA library (MATCHMAKER, CLONTECH) (37) in pGADGH. The transformants were grown for 8 h in medium lacking tryptophan and leucine and then plated onto medium lacking tryptophan, leucine, and histidine. Colonies were picked 4.5 days after plating and tested for ␤-galactosidase activity. Library plasmids were rescued into Escherichia coli HB101 cells plated on leucine-free medium. To quantify protein-protein interactions, ␤-galactosidase was assayed using o-nitrophenyl-␤-D-galactosidase as substrate.
Gene Expression Panel-1 l of first strand cDNA from each tissue of a Rapid Scan Gene Expression Panel (Origene) was used as template in a 40-cycle PCR reaction with the RCP gene-specific primers GSP-1 (5Ј-GAAAATTGGAGGTCTCGGTTCA-3Ј) and GSP-2 (5Ј-GCTTCCACC-AAGGACTCCTTGA-3Ј). The primers were designed to amplify the region of RCP corresponding to amino acids 384 -529, yielding a product of 435 base pairs. An annealing temperature of 60°C and an elongation time of 1 min were used. Clone H13 cDNA was used as template in the positive control reaction, and the negative control contained 1 l of distilled H 2 O instead of template.
Recombinant Proteins-The H13 polypeptide and Rab4Q67L protein were purified as follows. BL21 (DE3) E. coli cells were transformed with either pTrcHisA-Rab4Q67L or pTrcHisC-H13. Transformants were grown to an OD 600 ϭ 0.4 and induced with 0.3 mM isopropyl-1-thio-␤-D-galactopyranoside for 5 h. The proteins were purified with Ni 2ϩagarose (Qiagen) according to a native purification protocol (QIAexpressionist, Qiagen). Purified His-H13 polypeptide was injected into rabbits for antibody generation.
Cell Lines and Transfection-HeLa and HeLa Rab4GFP cells were maintained in culture as previously described (24). BHK cells were cultured in Glasgow's minimal essential medium supplemented with 10% tryptose phosphate broth, 5% fetal calf serum, 100 units/ml penicillin, 100 g/ml streptomycin, and 2 mM L-glutamine. For overexpression studies, HeLa cells were infected for 45 min with either the attenuated MVA T7 RNA polymerase recombinant vaccinia virus (a kind gift from G. Sutter) or the vT7 vaccinia virus and then transfected with the appropriate plasmids using either DOSPER or DOTAP (Roche Molecular Biochemicals) according to the manufacturer's instructions. For colocalization with Rab11, ␦RCP was transfected into HeLa cells using the Effectene transfection reagent (Qiagen) according to manufacturer's instructions. Six hours after the addition of the DNA/Effectene complex, the cells were fixed and processed for immunofluorescence microscopy.
Immunofluorescence Microscopy-Cells on 11-mm round glass coverslips were fixed with 3% paraformaldehyde. Free aldehyde groups were quenched with 50 mM NH 4 Cl. Cells were then permeabilized with 0.05% saponin and incubated with the primary antibodies diluted in 5% fetal bovine serum, phosphate-buffered saline. The secondary antibodies FIG. 1. RCP interacts with Rab4 and Rab11 in the yeast twohybrid system. A, clone H13 interacts with Rab4 and Rab11. Clone H13 was co-transformed into the L40 Saccharomyces cerevisiae yeast strain with the indicated Rab construct. An interaction was determined by growth on medium lacking histidine. B, quantitative assay of lacZ reporter gene activation. L40 cells were co-transformed with H13 and the indicated GAL4 DNA binding domain fusion plasmids. The ␤-galactosidase activity of the co-transformants was assayed using O-nitrophenyl ␤-D-galactopyranoside as substrate. The Ras/Raf interaction was used as a positive control (37). The plotted results are the average from four independent transformants for each plasmid combination. used were donkey anti-mouse conjugated to fluorescein isothiocyanate (FITC) and donkey anti-rabbit conjugated to Texas Red or donkey anti-rabbit conjugated to 9-amino-6-chloro-2-methoxyacridine (Jackson Immunoresearch). The coverslips were mounted onto slides with Mowiol (Sigma). Immunofluorescence images were recorded on a Lei-caTCS confocal microscope as previously described (42) or by conventional epifluorescence microscopy using a NIKON E-600 microscope fitted with a Hamamatsu chilled CCD C5985 camera. All images were processed using Adobe Photoshop and Adobe Illustrator software.
Ligand Internalization-Six hours post-transfection, serum-starved cells grown on glass coverslips were allowed to internalize FITC-coupled iron-saturated holotransferrin (33 g/ml; Sigma) for 1 h at 37°C. The cells were then briefly rinsed with ice-cold phosphate-buffered saline and fixed as previously described. For the uptake/chase experiments FITC-Tfn was internalized at 18°C for 45 min and then chased at 37°C for 10 min. The cells were washed with ice-cold phosphatebuffered saline and fixed.
Subcellular Fractionation and Membrane Washing-HeLa cells were resuspended in 250 mM sucrose, 90 mM KCl plus protease inhibitors, lysed by freeze-thawing twice and passing through a 26-gauge needle, and centrifuged at 10,000 ϫ g for 5 min. Membrane (pellet) and cytosol (supernatant) fractions were obtained by centrifuging the lysate at 100,000 ϫ g for 1 h at 4°C in a Beckman L8 -55 ultracentrifuge. Membrane extractions were performed as described (43); briefly, the postnuclear supernatant was incubated with 1% Triton X-100, 1 M NaCl, or 0.1 M Na 2 CO 3 (pH 11) for 30 min on ice and then centrifuged for 1 h at 100,000 ϫ g at 4°C. The pelleted membranes were resuspended in the same volume as supernatant, and equal volumes of either fraction were analyzed by SDS-PAGE and immunoblotting.
Estimation of 125 I-Tfn Recycling-The recycling assays were performed as previously described (8). Briefly, BHK cells infected with the MVA T7 recombinant vaccinia virus were co-transfected with the indicated construct and the human TfnR. 125 I-Tfn was bound to the cell surface on ice for 1 h. Unbound ligand was washed off, and the cells were allowed to internalize and recycle the 125 I-Tfn at 37°C. At the indicated time points the amount of recycled Tfn in the culture medium was quantitated. To analyze the rate of recycling from the recycling compartment, the transfected BHK cells were pulsed with 125 I-Tfn for 5 min, excess ligand was washed off, and the cells were chased at 37°C for 30 min. After the chase, the amount of 125 I-Tfn in the culture medium was measured. In all cases the amount of 125 I-Tfn is indicated as a percent of the total Tfn, i.e. Tfn associated with the cell plus Tfn in the culture medium. Control cells were transfected with the human TfnR alone.

Isolation of Rab4-interacting Clones
Using the Yeast Twohybrid System-We utilized the yeast two-hybrid technique to identify proteins that interact with Rab4. A GTPase-deficient, therefore constitutively active, mutant of Rab4 (Rab4Q67L) was used as "bait" to screen an oligo-dT-primed HeLa cDNA library. Approximately 3 ϫ 10 6 transformants were screened, of which 44 grew on selective medium lacking histidine and activated the ␤-galactosidase reporter gene. Fourteen of these displayed a specific, nucleotide-dependent interaction with Rab4. By hybridization and sequencing analysis, five of the clones were determined to be rabaptin 5, a Rab5 effector that also interacts with Rab4 (20). Four clones, represented by clone H13, expressed an identical 1.9-kilobase cDNA. The remaining five clones from the screen have yet to be characterized. Clone H13 was chosen for this study, and we demonstrate here that it interacts strongly with Rab4Q67L and Rab4WT, but weakly with the GDP-locked mutant (Rab4S22N) (Fig. 1). No interaction was observed with a mutant that is GTP-locked but has an amino acid change in the putative effector domain (Rab4Q67L/ I41E), suggesting that the interaction is mediated through this domain. No reporter gene activation was observed with the GTPase-deficient forms of Rab3, Rab5, Rab6, or Rab7 (Fig. 1B). Interestingly, there was also a strong interaction with all Rab11 constructs (Fig. 1). These results demonstrate that clone H13 encodes a polypeptide that interacts with two early endosomal Rab GTPases that have been shown to overlap in their localization (11,12) and function sequentially (5)(6)(7)(8) in the regulation of receptor recycling.
Furthermore, in a previously reported independent Rab4 two-hybrid screen (24), we isolated a mouse clone named M25 that comprises the extreme carboxy-terminal 65 amino acids of H13 and retains the ability to interact with Rab4 and Rab11. 2 Thus, we have been able to localize the Rab4/Rab11-interacting motif of this polypeptide to 65 amino acids.
Identification of Full-length Rab Coupling Protein-Clone H13 contains an open reading frame of 270 amino acids. While our work was in progress, the full-length mouse homologue of clone H13 was submitted to the data bases (GenBank TM accession number AK014696) (31). Using the mouse cDNA to search the human genome sequence, we identified the genomic sequence of clone H13, consisting of five exons on the short arm of chromosome 8. We identified an EST (IMAGE 3956619) that contained the full-length human RCP, and we report here its predicted 649 primary amino acid sequence ( Fig. 2A). Primary sequence analysis indicates that RCP is a predominantly hydrophilic protein with a putative C2-phospholipid binding domain at its amino terminus and two putative coiled-coil domains (Fig. 2, C and D). Homology searches of the GenBank TM data base identified KIAA0857 (Rip11/pp75), KIAA0941 (Rab11-FIP2/nRip11), and KIAA0665 (Rab11-FIP3/Eferin) as homologues of RCP. While our work was in progress, descriptions of these three homologues as well as a further family member (Rab11-FIP1) appeared in press (27,29,30). To avoid confusion in the nomenclature, the published names are indicated in parentheses after the GenBank TM reference name in each case above. An alignment of RCP with its four related proteins is shown (Fig. 2B). Over their entire lengths, RCP shares 36, 28, 35, and 21% identity with Rip11/pp75, Rab11-FIP1, Rab11-FIP2/nRip11, and Rab11-FIP3/Eferin, respectively. However, the greatest degree of homology between these proteins exists toward their termini (Fig. 2B). Four of these proteins (all except Rab11-FIP1) contain either a C2 or an EF domain in their amino-terminal half, and all five contain a Rab binding domain/predicted coiled-coil domain near their carboxy termini.
Rip11 has been reported as a regulator of apical membrane  (27). The remaining three members (FIP1, FIP2/nRip11, and FIP3/Eferin) of the family have been reported as Rab11/Rab25-interacting proteins localizing to the endosomal recycling compartment in polarized and non-polarized cells. However, while these proteins share a number of similar features, RCP is the only member reported to interact with Rab4. RCP is therefore a new member of a family of proteins that interact with Rab11/Rab25 and/or Rab4 and function along the endosomal recycling pathway.
RCP Is a Widely Expressed, Predominantly Membranebound Protein-We investigated the expression pattern of RCP by a PCR approach utilizing, as template, the first strand cDNA generated from a panel of human tissues and RCP genespecific primers to amplify a product of known size (435 base pairs) corresponding to RCP amino acids 384 -529. Similar amounts of the expected 435-base pair product were generated from all eight human tissues tested (Fig. 3A), suggesting the presence of equivalent amounts of RCP mRNA in these tissues. To confirm PCR specificity, we included reactions for a positive control (clone H13 cDNA as template) in addition to a negative (no template) control.
To study the biological function and intracellular localization of RCP, antisera was raised against bacterially expressed and purified H13 polypeptide (H13). Affinity-purified antisera identified a protein migrating with an apparent molecular mass of ϳ80 kDa (Fig. 3B) in all cell lines tested. The RCP full-length open reading frame was subcloned into a mammalian expression vector and expressed in HeLa cells. The molecular weight of the recombinant protein corresponded to that of endogenous RCP (data not shown).
Fractionation of HeLa cells into total cellular lysate, postnuclear supernatant, high speed pellet (membrane fraction), and high speed supernatant (cytosol fraction) demonstrated that RCP is a predominantly membrane-associated protein (Fig. 4A). We investigated the stability of RCP membrane association by detergent, high salt, and high pH treatments. RCP was totally membrane-extracted by 1% Triton X-100, whereas it was partially extracted by either 1 M NaCl or 0.1 M Na 2 CO 3 (pH 11) (Fig. 4B). The localization of the TfnR and ␤-actin was monitored as fractionation controls.
Taken together, the data described above indicate that the expression levels of RCP, when tested both by reverse transcriptase-PCR and immunoblotting, appeared to be relatively similar in all cell types tested. Furthermore, RCP encodes an 80-kDa protein that is predominantly membrane-associated and is partially resistant to high salt and high pH treatments but is completely solubilized from membranes by Triton X-100.

FIG. 4. RCP is a predominantly membrane-bound protein.
A, total cell extract, postnuclear supernatant (PNS), high speed supernatant (Cytosol), and high speed pellet (Membranes) were prepared from HeLa cells and analyzed by SDS-PAGE and immunoblotting with anti-H13 and anti-␤ Actin. B, HeLa postnuclear supernatant was treated on ice for 30 min under the conditions indicated before centrifugation. RCP was identified in the 100,000 ϫ g pellet (P) or supernatant (S). The accuracy of the fractionation was controlled by probing with anti-TfnR and anti-␤ actin. C, H13 polypeptide expressed in E. coli BL21 cells was resolved on a polyacrylamide gel and transferred to nitrocellulose (in triplicate). After renaturation, the blots were incubated with 10 g of [␣-32 P]GTP-labeled Rab5WT or Rab4WT or Rab11WT. The blots were washed several times and exposed on a PhosphorImager. PAGE and transferred to nitrocellulose (in triplicate). The blots were then incubated with either [␣-32 P]GTP-Rab4 or [␣-32 P] GTP-Rab11 or, as a control, [␣-32 P]GTP-Rab5. After several washes, the blots were analyzed by phosphorimaging. Both Rab4p and Rab11p associated with a band corresponding to the position of H13 (Fig. 4C). Furthermore, as expected, Rab5p did not bind H13. Thus, the carboxyl-terminal region of RCP can interact directly and independently with Rab4 and Rab11 in vitro.
RCP Colocalizes with Markers of the Recycling Pathway but Not with Markers of the Degradative Pathway-To further understand the biological function of RCP, we examined the localization of endogenous RCP in HeLa cells by immunofluorescence. It displays a punctate vesicular pattern in the cytoplasm that colocalizes partially with the TfnR in the perinuclear region of the cell (Fig. 5A). Since the available Rab4 antibodies do not detect endogenous Rab4 by immunofluorescence, we compared endogenous RCP localization with Rab4, utilizing a previously described stable GFP-Rab4 HeLa cell line (24). Endogenous RCP again partially colocalized with GFP-Rab4 in the perinuclear area (Fig. 5B). To investigate colocalization with Rab11, endogenous Rab11 was visualized with an affinity-purified rabbit antibody. Since the RCP antibody is also rabbit, HeLa cells were transfected with an epitope-tagged RCP construct (␦RCP) (see Fig. 2D). ␦RCP was visualized with a mouse monoclonal antibody directed against the epitope. The recombinant protein was expressed at a low level such that its pattern closely resembled the endogenous RCP pattern. We observed extensive colocalization with Rab11. Furthermore, when ␦RCP is expressed, Rab11 is more distinctly membraneassociated, appearing to be either recruited or retained on the ␦RCP-positive structures (Fig. 5C). RCP does not colocalize with lysobisphosphatidic acid, a phospholipid found on late endosomes and, thus, a marker of the degradative pathway (32, 33) (Fig. 5D). We also compared the localization of transiently overexpressed ␦RCP with endogenous early endosome-associated antigen 1 (EEA1) by immunofluorescence and find that these two markers display minor co-localization (data not shown). EEA1 is a Rab5 effector protein that appears to poorly localize to the perinuclear recycling compartment (11,12,34). Since RCP and EEA1 display very little co-localization, it is likely that RCP distribution is to a post-EEA1/Rab5, ERC location.
Overexpression of H13 Tubulates the Transferrin Compartment-To determine any morphological effects of H13 on the early endosomal compartment, we transiently transfected HeLa cells with a plasmid that expresses this truncated RCP polypeptide. After transfection, the cells were allowed to continuously internalize FITC-labeled Tfn at 37°C. As expected, in the mock transfected (control) cells, the FITC-Tfn accumulated in distinct endosomal structures in the perinuclear area of the cytoplasm as well as to some more peripheral vesicles (Fig. 6A). In cells expressing H13, the morphology of the Tfn compartment changed dramatically, displaying FITC-Tfn-labeled membrane tubular structures that extend throughout the cytoplasm (Fig. 6B). In transfected cells, which display this morphology but express H13 at a lower level, it is possible to visualize H13 on these transferrin-positive membranes (Fig.  6C). Since H13 is likely to be nonfunctional, it is probably acting in a dominant-negative manner with respect to Rab4 and/or Rab11 by sequestering them in a nonfunctional complex, thus preventing their interaction with endogenous RCP. It is interesting to note that the abnormal Tfn compartment generated on H13 expression is similar to that observed when the dominant-negative Rab11S25N mutant is expressed (10).
To investigate the identity of the tubulated Tfn compartment, we examined the structures labeled by FITC-Tfn after internalization at 18°C for 45 min, or uptake at 18°C followed by a chase at 37°C for 10 min. A temperature block of 18°C permits endosomal internalization but inhibits passage from early/sorting endosomes. When H13-expressing cells were incubated at 18°C, only vesicular structures were labeled by FITC-Tfn (Fig. 7A). However, when the temperature block was lifted, we observed FITC-Tfn labeling a tubulated compartment (Fig. 7B). Since the H13-induced tubules are inaccessible to FITC-Tfn at 18°C, it is likely that these tubules represent an abnormal endosomal recycling compartment.
Since Rab11 also regulates transport to the TGN, we investigated the localization of TGN46 in cells expressing either H13 or Rab11S25N. TGN46 is a glycoprotein that cycles between the plasma membrane and the TGN. As previously reported (10), we observe a proportion of TGN46 localizing to the Rab11S25N-generated tubules. H13-induced tubules similarly display TGN46 on the transferrin receptor positive tubules (data not shown).

FIG. 5. RCP colocalizes with the TfnR, Rab4, and Rab11 in HeLa cells.
A, HeLa cells fixed with paraformaldehyde and permeablized with saponin were labeled with anti-H13 antibody (red) and anti-TfnR (green) to detect endogenous RCP and the TfnR, respectively. The two proteins displayed partial colocalization in the perinuclear region of the cell. B, a stable cell line expressing Rab4WTGFP (green) was fixed and labeled with the affinity-purified anti-H13 antibody (red). Colocalization between the endogenous RCP and the recombinant Rab4WTGFP was seen in the perinuclear region. C, because the antibodies that detect endogenous RCP and endogenous Rab11 were both raised in rabbits, it was necessary to express epitope-tagged ␦RCP at a level similar to that of the endogenous protein. The ␦RCP construct was transfected into HeLa cells for 6 h, after which the cells were fixed and permeablized and labeled with anti-Rab11 (red) and anti-Xpress (green). There was almost total colocalization between ␦RCP and Rab11. It appeared that Rab11 was recruited from the cytosol to the membranes in transfected cells. D, paraformaldehyde-fixed and saponin-permeabilized HeLa cells were labeled with anti-H13 and antilysobisphosphatidic acid. No colocalization was observed between endogenous RCP and the late endosomal marker.
Increased Rab11 Function Reverses H13 Tubulation of the Tfn Compartment-As already indicated, H13 expression produces an early endosomal phenotype reminiscent of that generated by Rab11GDP (10). With this in mind, we were interested in determining whether the H13-induced tubulation of the Tfn compartment could be reversed by excess wild-type or dominant-positive Rab11. We examined this by co-transfecting clone H13 with either Rab11WT or Rab11Q70L and allowing cells to internalize FITC-Tfn. In cells that co-overexpressed H13 and active Rab11, the Tfn compartment phenotype was reversed (Fig. 8). This is consistent with the proposal that H13 sequesters endogenous Rab11 in a nonfunctional complex and the addition of active Rab11 bypasses this block. We also investigated the effect of H13 co-expression with Rab4 WT or Rab4Q67L. As previously reported, increased Rab4 function leads to tubulation of the Tfn compartment (7). Thus, as expected, co-expression of H13 with Rab4 WT or Q67L also resulted in tubulation of the Tfn compartment (data not shown).
Overexpression of H13 Inhibits Recycling-To investigate the effect that H13 or full-length RCP has on the kinetics of the receptor recycling pathway, the trafficking of 125 I-Tfn was analyzed. BHK cells cotransfected with H13 or full-length RCP and the human TfnR were incubated with 125 I-Tfn on ice. Excess 125 I-Tfn was washed off, and the cells were allowed to internalize and recycle the ligand at 37°C. The amount of 125 I-Tfn in the culture medium, and therefore recycled, was measured at different time points (Fig. 9A) and compared in each case to the amount of recycled Tfn in control cells (cells transfected with the human TfnR alone). Overexpression of H13 resulted in a dramatic inhibition in the rate of recycling after ϳ10 min. In contrast, full-length RCP slightly stimulated recycling when compared with the control. Dominant-negative Six hours after transfection FITC-Tfn was internalized at 18°C for 45 min. The cells were either washed with ice-cold phosphate-buffered saline and fixed immediately or chased at 37°C for 10 min before fixation. A, in cells fixed after the 18°C uptake, FITC-Tfn displayed a vesicular pattern, with no tubules labeled. B, in cells that were chased at 37°C for 10 min, the FITC-Tfn was able to access the H13-induced tubules. Rab4, Rab4S22N, also inhibited recycling but to a lesser extent than H13.
To specifically examine the effects on recycling from the ERC, the assay was modified to allow the kinetics of the "slow/ long" pathway to be analyzed (8). BHK cells transfected with the human transferrin receptor and the indicated constructs were pulsed with 125 I-Tfn for 5 min, excess ligand was washed off, and the cells were chased for a further 30 min at 37°C. This allows the Tfn following the "fast/short" pathway to cycle out of the cell and the Tfn trafficking along the slow pathway to load the ERC. After the chase, the amount of 125 I-Tfn in the culture medium was measured after various time points. Overexpression of H13 resulted in a strong inhibition of recycling from the ERC, whereas full-length RCP showed a slight stimulation of recycling (Fig. 9B). Rab11S25N also inhibited recycling, as previously reported (8), but to a lesser extent than H13.
Furthermore, because Rab4 affects late endosomal trafficking (7) as well as recycling (5), we investigated whether RCP may also play a role in regulating transport along the degradative pathway. Overexpression of H13 did not affect the uptake or degradation of 125 I-EGF but inhibited EGF recycling (data not shown). Taken together, the data from the quantitative assays demonstrate that RCP functions in endosomal recycling, most likely through the ERC.

DISCUSSION
Rab proteins serve a critical role in the regulation of intracellular transport events. With the deposition of the human genome sequence, it is now known that there are 60 different human Rab proteins (35), and it is expected that each member controls a specific event in a transport pathway. The receptor recycling pathway is under the control of several Rabs, including Rab5, Rab4, and Rab11, which function in a sequential fashion. The current view proposes that Rab5 controls transport from the plasma membrane and early endosomal fusion, Rab4 regulates transport events from the sorting endosomes, and Rab11 regulates transport through the perinuclear recycling compartment.
Here, we report the identity and characterization of Rab coupling protein, a novel ϳ80-kDA membrane-associated protein that interacts with both Rab4 and Rab11. We demonstrate that the carboxy-terminal 65 amino acids of RCP are sufficient for Rab4 and Rab11 binding. The RCP/Rab4 interaction is strongly nucleotide-dependent, with RCP interacting preferentially with the GTP bound form of Rab4. Furthermore, the interaction with Rab4 is abolished when the GTP-locked form carries a mutation in its putative effector domain. The situation with Rab11 appears somewhat different, in that the RCP/ Rab11 interaction appears stronger (two-hybrid and transient expression/immunofluorescence data) than that of RCP with Rab4. However, the RCP/Rab11 interaction is less dependent on the Rab11 nucleotide-bound state. RCP is homologous to pp75, a phosphoprotein that has been implicated in neonatal and systemic lupus erythematosus (36). While our work was in progress, a study from Scheller and co-workers (27) reported pp75 as a Rab11-interacting protein (Rip11) that functions in the regulation of apical membrane trafficking. More recently, work from two groups has described other members of this family of Rab-interacting proteins under various names, Eferin or Rip proteins (30) and Rab11-family interacting proteins (Rab11-FIPs 1-3) (29). RCP is distinct from, but related to, these recently described Rab11/Rab25-interacting proteins and is therefore a new member of this novel Rab-interacting protein family.
Our immunofluorescence data indicates that RCP colocalizes with a small pool of Rab4, whereas there is almost complete colocalization with Rab11. Additionally, there is only minor colocalization with EEA1 (data not shown). Altogether, this suggests that RCP is predominantly an ERC protein and, therefore, is likely to interact with the pool of Rab4 that is destined for transport to this compartment. Furthermore, overexpression of H13 (a truncated form of RCP containing the Rab binding domain) results in the formation of an extensive tubular network that can be labeled with FITC-Tfn. These tubules appear to be derived from the recycling compartment since they are inaccessible to Tfn internalized at 18°C. This phenotype is very similar to that seen upon overexpression of the dominantnegative mutant of Rab11, Rab11S25N (10). Because, as indicated earlier, the interaction between RCP and Rab11 appears stronger than the RCP/Rab4 interaction, it is likely that excess H13 will generate a Rab11 dominant-negative phenotype by preferentially sequestering endogenous Rab11. Consistent with this view, expression of active Rab11 rescues the H13 phenotype. These data supports our view that the H13-induced tubulation is generated as a result of Rab11 being sequestered in a non-functional complex.
RCP is the second example of a putative effector protein that interacts with more than one RabGTPase subfamily. Rabaptin 5 interacts with both Rab5 and Rab4 (20). Since Rab4 functions after Rab5 in early endosomal trafficking, it is suspected that rabaptin 5 acts as an intermediate between these two GTPase proteins. Likewise, it is probable that RCP is an intermediate between Rab4 and Rab11, since Rab11 acts after Rab4 in the recycling pathway. Rab11 appears to function exclusively on trafficking through the perinuclear ERC. Rab4, on the other hand, functions primarily at the level of the sorting endosome and affects both recycling and degradation (7), probably by interacting with different sets of effectors. A number of putative Rab4 effectors have already been reported such as rabaptin 5 (20), rabaptin 4 (21), syntaxin 4 (22), Rab4-interacting protein (Rab4ip) (23), and the dynein light intermediate chain-1 (dynein LIC-1) (24).
Our quantitative data demonstrates that RCP functions in the endosomal recycling process and is not involved in the late endosomal pathway. Overexpression of the carboxy-terminal region of RCP, which contains the Rab4 and Rab11 binding sites, results in a dominant-negative effect on Tfn-recycling kinetics. Analysis reveals that this inhibition begins after 10 min of ligand internalization. Further analysis that examined the rate of recycling specifically from the ERC revealed that the RCP dominant-negative mutant inhibited recycling from this compartment. This inhibition was stronger than that observed with the dominant-negative Rab mutants. This may be due to the cumulative effect of the sequestration of Rab4 and Rab11 in an inactive complex and/or the probability that the RCP mutant also sequesters other members of the Rab11 family.
RCP, as a predominantly membrane-associated protein, may be involved in the targeting of a Rab4-or Rab11-mediated complex to the correct compartment. Considering the localization of RCP, which appears predominantly on the Rab11-positive perinuclear ERC, we suggest that it may serve as a target for Rab4 donor vesicles arriving at the acceptor ERC membrane. Another possibility is that RCP interacts with Rab4 at a sorting endosome location and participates in Rab4-mediated trafficking to, or generation of, the ERC. RCP may then serve to recruit Rab11 onto the ERC membrane or, alternatively, interact with Rab11 already associated with the ERC membrane. In fact, when H13 or ␦RCP are overexpressed, the membrane localization of Rab11 increases consistent with either of these possibilities. Further studies will be necessary to clarify these issues. It will also be important to investigate the precise events leading to RCP membrane localization after its biosynthesis as well as whether it subsequently cycles on/off mem-branes. Pertinent to this will be an investigation of whether Rabs are involved in this process and the role of the RCP C2 domain in lipid binding.
The identification of RCP as an interacting partner of Rab4 and Rab11, with many analogies to its related proteins, pp75/ Rip11 and the Rab-FIPs (27,29,30), contributes further to our understanding of the eukaryotic membrane-trafficking machinery. Furthermore, the similarity between the Rab5/rabaptin 5/Rab4 molecular network and the Rab4/RCP/Rab11 network serve to link-up or couple the endosomal-trafficking machinery, involving several RabGTPases sequentially linking endosomal sub-domains/sub-compartments, both in terms of their overlapping distribution as well as by their interaction with common sub-groups of effectors. Future studies to decipher the molecular details of such interactions and to understand the precise molecular functions of this novel Rab-interacting protein family will be crucial toward our better understanding of the fundamental cellular process of endosomal sorting, recycling, and transcytosis with respect to the events controlled by the Rab GTPases.