Copyright C Blackwell Munksgaard 2003 Traffic 2003; 4: 97–112 Blackwell Munksgaard ISSN 1398-9219 CD2AP/CMS Regulates Endosome Morphology and Traffic to the Degradative Pathway Through its Interaction with Rab4 and c-Cbl

Mireille Cormont1,*, Isidoro Meto´ n1, in T cells (1). CD2AP, an ubiquitously expressed , was Muriel Mari1, Pascale Monzo1,Fre´ de´ rique subsequently characterized as a p130Cas partner and named Keslair1, Chantell Gaskin2, Timothy E. McGraw2 CMS for Cas ligand with multiple SH3 domains (2). CD2AP/ and Yannick Le Marchand-Brustel1 CMS was also identified as a up-regulated during the differentiation of metanephrogenic mesenchyme cells into 1 Inserm U568, IFR 50, Faculty of Medicine, University of epithelial kidney cells (3). CD2AP/CMS is a 80-kDa adaptor Nice, 06107 Nice cedex 02, France protein containing three SH3 domains in its N-terminus, a 2 Department of Biochemistry, Weill Medical College of proline-rich region and a C-terminal coiled-coil domain with Cornell University 1300 York Ave. New York, NY 10021. USA putative b thymosin-like actin binding sites (Figure 1A). * Corresponding author: Mireille Cormont, CD2AP/CMS forms a new family of adaptor together [email protected] with the RUK isoforms (regulator of ubiquitous kinase), also named CIN85 (c-Cbl interacting protein of 85kDa), or SETA (SH3 domain-containing gene expressed in tumorigenic as- The small GTPase Rab4 is involved in endocytosis through sorting and recycling early endosomes. To trocytes) (4–7). better understand the role of Rab4 in regulation of ves- icular trafficking, we searched for effectors that Following ligand attachment of the adhesion molecule CD2, specifically interact with Rab4-Q67L, the GTP-bound CD2AP induces the clustering of CD2 and actin cytoskeletal form of Rab4. We cloned an ubiquitous 80-kDa protein, reorganization in T cells. It was proposed to facilitate receptor identical to CD2-associated protein/Cas ligand with patterning in the T-cell-contact areas by linking specific re- multiple SH3 domains (CD2AP/CMS), that interacts ceptors to the cytoskeleton (1). The function of CD2AP is not with Rab4-Q67L in the yeast two-hybrid system and in limited to T cells. Mice lacking CD2AP died at a young age vitro. CD2AP/CMS expressed in mammalian cells was due to the progressive development of glomerulosclerosis, localized to punctate structures and along actin filaments. None of the known markers of early although they presented minor immunological alterations (8). endosomes [Early Endosomes Antigen 1 (EEA1), Rab5 In the knockout mice the major defect resides in the podocy- and Rab11] colocalized with the CD2AP/CMS-positive te, a cell type that composes the glomerular slit diaphragm vesicles. However, coexpression of Rab4-Q67L with (9). Podocytes, which normally express high levels of CD2AP CD2AP/CMS induces a significant enlargement of (9), were unable to develop foot processes needed to main- EEA1-positive early endosomes. Rab4, CD2AP/CMS tain the integrity of the slit diaphragm in the knockout mice. and Rab7 colocalized in these modified endosomes. This phenotype is likely due to the fact that CD2AP interacts Coexpression of c-Cbl and CD2AP/CMS also resulted with (9), (9,10), and polycystin 2 (3), pro- in an enlargement of early endosomes. Using various teins that play a major role in podocytes. In agreement with truncated forms of CD2AP/CMS, we demonstrate that a crucial role of this complex in kidney, mutations in NPHS1, early endosomes enlargement requires that CD2AP/ CMS interacts with both Rab4 and c-Cbl. The expres- NPHS2 or PDK2, encoding for nephrin, podocin and sion of a truncated form of CD2AP/CMS that retains polycystin 2, respectively, are associated with severe kidney the ability to interact with Rab4 but not c-Cbl inhibits diseases in humans (11–13). ligand-induced PDGF receptor degradation. We pro- pose that CD2AP/CMS, through interactions with CD2AP/CMS also appears to be involved in various intracellu- Rab4 and c-Cbl, controls early endosome morphology lar signaling pathways. Indeed, it interacts in vitro and in and may play a role in traffic between early and late over-expressing systems with a variety of signaling mole- endosomes, and thus in the degradative pathway. cules that contain SH3-domains. In addition to the p130Cas, the Src kinase family, the p85 regulatory subunit of PI3 kinase Key words: actin, effector, endosomes, Rab 4, Rab 7, transferrin recycling and Grb2 (2), CD2AP/CMS interacts with the proto-onco- gene c-Cbl. Importantly, this interaction is regulated by extra- Received 6 May 2002, revised and accepted for publi- cellular factors, since it is dependent on the tyrosine phos- cation 14 November 2002 phorylation of c-Cbl (14). The recently described role of c- Cbl in actin cytoskeleton organization (15) might involve the adapter CD2AP/CMS. Along the same line, the interaction of CD2-associated protein (CD2AP) was originally discovered CIN85 with c-Cbl has recently been described to play a key by its ability to interact with the transmembrane protein CD2 role in receptor-mediated signaling and endocytosis (16,17).

97 Cormont et al.

Figure1: CD2AP/CMS interacts with active Rab4 in the yeast two-hybrid system and in vitro. (A) Schematic structure of CD2AP/ CMS. Numbers indicate the limits of the different domains. SH3: Src homology 3 domain. PR: proline-rich region. CC: Coil-coiled domain. The region corresponding to the A3 clone, which contains the Rab4 binding domain, is noted, as well as the c-Cbl interacting region (14). (B) Characterization of the interactions between the initial clone A3 or CD2AP/CMS and various Rab4 proteins in the yeast two-hybrid system. L40 yeasts were cotransformed with the pLex constructs encoding for LexA-fusion proteins, and the rescued VP16-A3 or pACT2- CD2AP/CMS. b-galactosidase activities (arbitrary units) are presented as means∫SEM obtained with 3–6 independent transformants. The expression of the LexA-Rab4 fusion proteins, checked by immunodetection, was similar in all conditions. (C) Rab4-GTP interacts with myc- CD2AP/CMS in vitro. 293 cells were transfected with 10mg of pcDNA3-myc-CD2AP/CMS and submitted to a hypotonic lysis. Following ultracentrifugation, cytosol was incubated with GST or GST-Rab4 (loaded with GDP or GTPgS) immobilized on glutathione-sepharose beads. Bound proteins were eluted, analyzed by SDS-PAGE and immunodetection was performed using anti-myc antibody, as detailed in Materials and Methods. A portion of the cytosol (input) is shown. (D) In vitro transcribed and translated CD2AP/CMS interacts with Rab4-GTP. GST, GST-Rab4-GDP or GST-Rab4-GTPgS bound to glutathione-sepharose beads were incubated with 35S-methionine-labeled in vitro translated CD2AP/CMS or Rabip4. Bound proteins were eluted, analyzed by SDS-PAGE and autoradiographed.

By screening a cDNA library using an active form of Rab4 as Results bait in the yeast two-hybrid system, we identified a clone that corresponds to the C terminus of CD2AP/CMS. The co- CD2AP/CMS is an effector of the small GTPase Rab4 expression of active Rab4 with CD2AP/CMS induces a sig- To search for Rab4 effectors, L40 yeast was transformed with nificant enlargement of early endosomes, indicating a func- a plasmid encoding a fusion protein between Rab4 Q67L, a tional interaction between these proteins in vivo. An identical GTPase-deficient mutant (18), and the DNA binding domain enlargement of early endosomes was observed when c-Cbl of LexA, which recognizes specific DNA sequences up- was coexpressed with CD2AP/CMS. Using various truncated stream of the two reporter genes HIS3 and LacZ. The estab- forms of CD2AP/CMS, we show that the change in early en- lished strain was transformed with a yeast two-hybrid mouse dosome morphology induced by expression of CD2AP/CMS TA adipocyte cDNA library fused to the VP16 transcriptional requires that it interacts with both Rab4 and c-Cbl. Our data activation domain. Screening of 106 transformants yielded indicate that CD2AP/CMS, via interactions with Rab4 and c- three clones that strongly interacted with Rab4 Q67L but not Cbl, plays a role in determining early endosome morphology. with lamin. The clone named A3, constituted by a 402-bp Further, since CD2AP/CMS seems to link Rab4 and Rab7 insert in frame with the transactivation domain of VP16, was positive compartments, it might be involved in some degrad- further characterized in the yeast two-hybrid system (Figure ative pathways. 1B, open bars). VP16-A3 induced b-galactosidase activity

98 Traffic 2002; 4: 97–112 CD2AP/CMS: a Novel Rab4 Effector

Figure2: CD2AP/CMS is a membrane-associated protein, which labels vesicular structures and actin filaments. (A) CD2AP/ CMS is associated to membrane. Control CHO cells (left panel) or cells transiently transfected with pcDNA3-myc-CD2AP/CMS (right panel) were homogenized. Following ultracentrifugation, equal volumes of homogenate (H), cytosol (CY) and pellet (P) corresponding to total crude membranes were resolved by SDS-PAGE. Endogenous CD2AP/CMS was detected using a polyclonal antibody (left panel) and over-ex- pressed protein was revealed by an anti-myc mAb (right panel). (B) Indirect immunofluorescence analysis of CD2AP/CMS subcellular distribution. CHO cells, transiently transfected with pcDNA3-myc-CD2AP/CMS, were serum starved for 16h and treated or not for 1h with insulin (1mM). Myc-CD2AP/CMS was detected using 9E10 anti-myc mAb followed by anti-mouse antibodies coupled to Texas Red (top panels) or coupled to FITC (other panels). Actin filaments were labeled with phalloidin coupled to Texas Red. The upper lane shows three confocal sections of the same cell, at the top, in the middle and at the bottom close to the coverslip. The two other lanes show the picture obtained at the bottom of the cell with green labeling of myc-CD2AP/CMS, red labeling of phalloidin and the merge image in which colocalization appears yellow. The bar corresponds to 1mm.

only when coexpressed with LexA-Rab4 Q67L or LexA-Q67L RACE-PCR approaches. We isolated a 5427-bp cDNA con- D CT (an active form of Rab4 deleted of its Cys-Gly-Cys C- taining a 1914-bp open reading frame, which predicted for a terminal motif needed for its geranylgeranylation), but not 636 AA protein (accession number AJ459109). We first with active Ras or Rab5. We did not detect any interaction named it Rabef4 for Rab4 effector (19). This protein was between VP16-A3 and LexA-Rab4 WT, or N121I or S22N, cloned at the same time and published as CD2AP (1) and two mutated Rab4 proteins that do not bind GTP. Further- CMS (2). more, two mutations in the Rab4 effector domain (T40A and Gly42D) abolished the interaction with VP16-A3. Thus the The primary structure of CD2AP/CMS is represented in sequence encoded by the A3 clone preferentially recognizes Figure 1(A). It contains three SH3 domains, a central region the GTP-bound Rab4. rich in prolines, and a C-terminal part predicted to form coiled-coil structures (20). A3 corresponds to the amino When this screen was performed, the A3 amino-acid se- acids 539–637 of CD2AP/CMS. Similar to the A3 clone, quence did not correspond to any known sequence. We ob- CD2AP/CMS specifically interacted with active Rab4 in the tained the full-length cDNA sequence by successively yeast two-hybrid system (Figure 1B, bars), whether or not screening a lGT11 3T3-L1 adipocyte library and 3ƒ and 5ƒ the C-terminal Cys-Gly-Cys motif was present. Importantly,

Traffic 2002; 4: 97–112 99 Cormont et al.

Figure3: CD2AP/CMS self-associates through its C terminal region. (A) Ex- pression of CD2AP/CMS tar- gets the association of CD2AP/CMS [539–637] to the membrane. CHO cells were transiently transfected with pcDNA3-myc-CD2AP/ CMS [539–637] or [1–328] alone or together with GFP- CD2AP/CMS. Forty-eight hours after, the cells were serum starved for 16h before homogenization. The homo- genate (H) was fractionated into cytosol (CY) and pellet (P). Equal volumes of each fraction were analyzed by SDS-PAGE and the myc-con- structs detected using the 9E10 mAb. (B) Expression of GFP-CD2AP/CMS redistrib- utes Myc-CD2AP/CMS [539–637] into the GFP- CD2AP/CMS labeled struc- tures. CHO cells were transi- ently transfected and treated as in A. Myc-CD2AP/CMS [539–637] were immunode- tected with 9E10 anti-myc antibody followed by anti- mouse antibodies coupled with Texas Red. The green label corresponds to GFP- CD2AP/CMS, the red image corresponds to CD2AP/CMS [539–637] and in the merge image, the yellow color results from the overlay of green and red. The bar corresponds to 1mm.

there was no interaction between CD2AP/CMS and active firm that CD2AP/CMS interacts with Rab4 in vitro, CD2AP/ Rab5, Rab6 and Rab11. We also noted that the site of CMS, produced by over-expression in 293 cells (Figure 1C) interaction between Rab4 Q67L and CD2AP/CMS is or in vitro translated in the presence of 35S-methionine located between amino acids 539 and 637 on the C- (Figure 1D), was incubated with recombinant GST-Rab4. terminal end of CD2AP/CMS, since the full-length CD2AP/ The results of the pull-down experiments (Figure 1C,D) CMS and CD2AP/CMS [539–637] interacted with Rab4 indicate that CD2AP/CMS interacts preferentially with Q67L while CD2AP/CMS [1–539] did not (Table1). To con- Rab4-GTP-g-S, as does Rabip4 another effector of Rab4.

100 Traffic 2002; 4: 97–112 CD2AP/CMS: a Novel Rab4 Effector

Figure4: Localization of CD2AP/CMS along the en- dosomal pathway: The co- expression of CD2AP/CMS and Rab4 induces the en- largement of early endo- somes. CHO cells were tran- siently transfected with CD2AP/CMS constructs (in fusion with GFP or with a myc epitope tag as indicated) alone (a) or together with GFP-Rabip4 (b), GFP-Rab4 WT (c), myc-Rab4 WT (d), GFP-Rab4N121I (e), GFP- Rab5 (f) GFP-Rab7 (g), or myc-Rab11 (h). The myc-con- structs were detected with anti-myc antibodies followed by anti-mouse coupled to Texas Red. Endogenous EEA1 was detected using anti- EEA1 mAb and anti-mouse antibody coupled to Texas Red. Panels show the merge images of the green and red labeling corresponding to the proteins written in the same color, except in e where only the labeling of CD2AP/CMS is shown. In panel d, the label- ing corresponding to Rab4 (detected using polyclonal antibodies against Rab4 fol- lowed by anti-rabbit anti- bodies coupled to Cy5) is not shown. The bar represents 1mm.

Traffic 2002; 4: 97–112 101 Cormont et al.

Table1: Characterization of the interactions of CD2AP/CMS with itself, Rab4 or c-Cbl in the yeast two-hybrid system pACT CD2AP/CMS CD2AP/CMS CD2AP/CMS c-Cbl c-Cbl [1–539] [539–637] [1–436] pLexA Rab4 Q67L Yes No Yes No ND pLexA Rab4N121I No No No No ND pLexA CD2AP/CMS Yes No Yes Yes No L40 yeasts were cotransformed with the pLex constructs encoding for LexA-fusion proteins, and the indicated pACT2-CD2AP/CMS or Cbl. The interaction of the proteins was determined by the measurement of b-galactosidase activity. Yes indicated that the two proteins interacted while No is for no interaction. ND not determined. The expression of the proteins was checked by immunodetection.

CD2AP/CMS is localized along actin filaments CMS was detected using a polyclonal antibody directed We examined the subcellular distribution of CD2AP/CMS in against the amino acids 539–637. CD2AP/CMS migrated CHO cells using biochemical and confocal microscopy ap- with an apparent molecular weight of 80k and was predomi- proaches. We first determined whether CD2AP/CMS be- nantly in the membrane pellet. Over-expressed myc-CD2AP/ haves as a cytosolic protein as predicted by its amino acid CMS, detected by the myc epitope, was identically distrib- sequence. Serum-starved CHO cells were fractionated into uted to the endogenous protein (Figure 2A). total membranes and cytosol. Equal amounts of both frac- tions were analyzed by SDS-PAGE and endogenous CD2AP/ The anti-CD2AP/CMS antiserum did not detect endogenous

Figure5: CD2AP/CMS [539–637] is not sufficient to induce the enlargement of early endosomes when coexpressed with active Rab4. CHO cells were transiently transfected with pEGFP-Rab4Q67L alone (a) or together with Myc-CD2AP/CMS WT (b) or [539– 637] (c and d). The myc constructs were detected as in Figure 3. In panels a–c, only the labeling corresponding to GFP-Rab4 Q67L is shown, while panel d shows the labeling of Myc-CD2AP/CMS [539–637]. Arrowheads point to the presence of GFP-Rab4 Q67L or Myc- CD2AP/CMS [539–637] at the level of membrane ruffles. The bar corresponds to 1mm.

102 Traffic 2002; 4: 97–112 CD2AP/CMS: a Novel Rab4 Effector

CD2AP/CMS in immunofluorescent staining experiments; colocalize with Rab11b (Figure 4h), a marker of early recyc- therefore, we used epitope-tagged CD2AP/CMS to study its ling endosomes, or Rab7, a marker of late endosomes (Figure cellular localization. In serum-starved CHO cells, myc- 4g). Since CD2AP/CMS interacts with active Rab4, we deter- CD2AP/CMS labeled punctate structures, visualized by in- mined whether it was present in the same structures as direct immunofluorescence and confocal microscopy, that Rab4. When CD2AP/CMS and GFP-Rab4 were coexpressed, were dispersed throughout the cytoplasm (Figure 2B, upper the distribution of CD2AP/CMS was dramatically altered, panels). At the bottom of the cells, myc-CD2AP/CMS decor- with both proteins localized to large vesicles (Figure 4c), ated long, straight filaments, reminiscent of actin stress reminiscent of the enlarged endosomal structures induced fibers. The F-actin nature of these fibers was confirmed by by the coexpression of Rab4 with Rabip4 (23,24). Similar the codistribution of myc-CD2AP/CMS with phalloidin- alterations were observed when an active form of Rab4 is labeled actin filaments at the bottom of cells (Figure 2B, expressed with CD2AP/CMS (see Figure 5). In cells express- middle panels). The myc-CD2AP/CMS-labeled punctate ing CD2AP/CMS and GFP-Rab4, the distribution of EEA1 structures dispersed throughout the cytoplasm were not was profoundly changed. In these cells, EEA1 was mainly positive for phalloidin. When CHO cells were treated with redistributed to the CD2AP/CMS and Rab4-positive large 1mM insulin (Figure 2B, lower panels), F-actin and myc- vesicles, although some diffuse EEA1 cytosolic labeling re- CD2AP/CMS were redistributed in a similar fashion to mem- mained (Figure 4d). The coexpression of CD2AP/CMS with brane ruffles at the periphery of the cells. These studies show a form of Rab4 unable to bind GTP (Rab4N121I, Figure 4e) that CD2AP/CMS is localized to vesicular structures present or with other Rab proteins such as Rab5 (Figure 4f), Rab7 throughout the cytoplasm as well as partially aligned along (Figure 4g) and Rab11 (Figure 4h) did not result in a change actin fibers or ruffles. in CD2AP/CMS labeling. Thus, only Rab4, wild-type or constitutively active, was able to modify the localization of CD2AP/CMS self-associates through its C terminal CD2AP/CMS. These findings along with the yeast two-hybrid portion and the in vitro data suggest that CD2AP/CMS interacts only Coiled-coil motifs are frequently involved in dimerization. with active Rab4. Using the yeast two-hybrid system, we found that lexA- CD2AP/CMS interacts with Gal4-CD2AP/CMS (Table1), indi- The results of the yeast two-hybrid studies indicate that the cating that CD2AP/CMS can self-associate. Deletion of interaction between Rab4 and CD2AP/CMS requires the C- amino acids 539–637 prevented this interaction (Table1), terminal portion of CD2AP/CMS. Indeed, CD2AP/CMS demonstrating that amino acids 539–637, which include the [539–637] interacted with Rab4, while CD2AP/CMS [1–539] coiled-coil motif, are required for CD2AP/CMS self-associ- did not (Table1). We next tested whether coexpression of ation in yeast. CD2AP/CMS self-association is also sug- CD2AP/CMS [539–637] with Rab4 was sufficient to induce gested by the behavior of the fragment of CD2AP/CMS the change in endosomal morphology. GFP-Rab4 Q67L ex- [539–637] expressed in mammalian cells (Figure 3). Myc- pressed alone labeled punctate structures enriched in the CD2AP/CMS [539–637] expressed alone was predominantly perinuclear region as well as some regions of the plasma cytosolic (Figure 3A, left panel), although there may be some membrane (Figure 5a). Coexpression of active Rab4 with full- enrichment in the perinuclear region (Figure 3B). However, length CD2AP/CMS led to the modification of the structures when Myc-CD2AP/CMS [539–637] was coexpressed with labeled by active Rab4 (Figure 5b). In this case the Rab4 GFP-CD2AP/CMS, its distribution shifted from cytoplasmic to Q67L labeling was predominantly concentrated in larger, membrane (Figure 3A, right panel) and it completely colocal- more peripheral, punctate structures. By contrast, the localiz- ized with GFP-CD2AP/CMS, labeling punctate structures and ation of active Rab4 was not modified by the expression of the actin stress fibers and ruffles (Figure 3B). By contrast, the myc-CD2AP/CMS [539–637], although the two proteins co- distribution of Myc-CD2AP/CMS [1–328], which lacks the C- localized in the perinuclear region and in membrane ruffles terminus and is also mainly cytosolic, was not changed by (Figure 5c,d, arrowheads). Thus, interaction between CD2AP/ coexpression of GFP-CD2AP/CMS (Figure 3B) or immuno- CMS [539–637] and active Rab4 is not sufficient to change fluorescence (not shown). Taken together, these results indi- endosomes morphology. cate that CD2AP/CMS likely forms a complex with other CD2AP/CMS molecules. Furthermore, these data show that CD2AP/CMS induces the enlargement of early the C-terminus of CD2AP/CMS, which contains the Rab4 endosomes when coexpressed with c-Cbl binding domain and the b thymosin-like sequence, is not suf- Since the interaction of Rab4 with the C-terminus of CD2AP/ ficient to target the molecule to its correct subcellular localiz- CMS is not sufficient to induce the enlargement of early en- ation. dosomes, other domains and/or other interactions are necessary. As described in the introduction, c-Cbl was a likely CD2AP/CMS induces enlargement of early candidate to study in this context. We found in the yeast two- endosomes when coexpressed with Rab4 hybrid analysis that CD2AP/CMS interacted with c-Cbl and CD2AP/CMS was not colocalized with the early endosomal that c-Cbl did not interact with Rab4 active (Q67L) or inactive markers EEA1 (Figure 4a) (21,22) nor with Rabip4 (Figure (N121I) (Table1). As proposed by Kirsch et al. (14), the inter- 4b), an effector of Rab4 we previously characterized as an action between c-Cbl and CD2AP/CMS probably requires the early sorting endosomal protein (23). CD2AP/CMS did not proline-rich sequences in the C-terminus of c-Cbl and the

Traffic 2002; 4: 97–112 103 Cormont et al.

Figure6: The interaction of CD2AP/CMS with c-Cbl in- duces changes in early en- dosomal morphology. The coexpression of CD2AP/CMS with c-Cbl induces enlarge- ment of early endosomes. CHO cells were transiently co- transfected with pcDNA3- HA-c-Cbl and pEGFP-CD2AP. HA-c-Cbl was revealed using anti-HA antibodies followed by Texas Red coupled anti- rabbit antibodies, endoge- nous EEA1 by monoclonal antibodies followed by Texas Red coupled anti-mouse anti- bodies. Panels a–c represent the labeling of each protein and the merge image. Panel d shows the overlay of the im- ages corresponding to GFP- CD2AP/CMS and EEA1 (c- Cbl labeling is not shown). (B) Characterization of c-Cbl sub- cellular localization. The local- ization of HA-c-Cbl was studied with regard to EEA1 (a–c) and actin (d–f). HA-c- Cbl was detected using anti- HA antibodies followed by FITC coupled anti-rabbit anti- bodies, while EEA1 was de- tected as described in A. F- actin was detected using Tex- as-Red coupled phalloidin. The labeling corresponding to c-Cbl is shown in a and d. Ar- rows point to a perinuclear re- gion and arrowheads point to filamentous structures where c-Cbl is enriched. The insets a–c show the partial colocaliz- ation of EEA1 and c-Cbl in vesicular structures. Bars cor- respond to 1mm.

104 Traffic 2002; 4: 97–112 CD2AP/CMS: a Novel Rab4 Effector

first SH3 domains of CD2AP/CMS (schematized in Figure tor (entry or exit). This is in accordance with the work of 1A). Indeed, no interaction was observed between CD2AP/ Bucci’s group (22), in which the expression of Rab4 (WT or CMS and c-Cbl [1–436] that does not contain the proline- active forms) results in only negligible changes in the rate of rich region (Table1). transferrin receptors recycling back to the plasma membrane.

In cells that coexpressed CD2AP/CMS and c-Cbl, both pro- Since the coexpression of CD2AP/CMS and Rab4 results in teins were mostly colocalized to large vesicular structures significant alterations in the morphology of early endosomes, (Figure 6A, a–c), which were also positive for EEA1 (Figure a compartment through which transferrin receptors recycle, 6A,d). The structures containing EEA1 and CD2AP/CMS we tested whether coexpression of CD2AP/CMS and Rab4 were markedly enlarged compared to control cells, indicating affected the distribution of transferrin receptors. For these that the coexpression of CD2AP/CMS and c-Cbl profoundly studies we coexpressed CD2AP/CMS and Rab4Q67L in modified the morphology of early endosomes. Reciprocally, A549 cells, a human cell line from non-small-cell lung can- we noted that c-Cbl labeling was also altered by the expres- cer. The amount of cell surface transferrin receptors was esti- sion of CD2AP/CMS. Indeed, when expressed alone, c-Cbl mated in indirect immunofluorescence of fixed non-permea- was located in a well-defined region near the nucleus prob- bilized cells, using a monoclonal antibody directed against ably representing Golgi (25) (arrows in Figure 6B, panels a the extracellular domain of the receptor. As shown in Figure and d), in vesicles (Figure 6B, panels a–c) and at the level of 7(C) (panels a and b), the density of cell surface transferrin actin stress fibers and membrane ruffles (Figure 6B, d–f). receptors was in the same range in the doubly transfected Some of the vesicles were positive for EEA1 (see insets of cells as in the surrounding cells (the doubly transfected cells Figure 6B,C). Although c-Cbl might associate some Golgi are identified by the large vesicular structures induced by structures, the Golgi morphology was not affected in cells the coexpression of Rab4 Q67L and CD2AP/CMS). However, over-expressing c-Cbl and CD2AP/CMS (data not shown). when the intracellular distribution of the transferrin receptors CD2AP/CMS [1–328], which interacts with c-Cbl but not was examined in permeabilized cells (Figure 7,c,d), it was Rab4 (Table1), did not lead to the formation of enlarged ves- apparent that the transferrin receptors were enriched in the icular structures (data not shown). Thus, these data indicate enlarged structures caused by the over-expression of that CD2AP/CMS must interact with c-Cbl and Rab4 to in- CD2AP/CMS and active Rab4 (Inset Figure 7d). Thus, based duce modification of the morphology of endosomes. on the qualitative assessment that there was no significant change in surface expression of the transferrin receptors, we CD2AP/CMS expression alters fluid phase conclude that, although coexpression of CD2AP/CMS and endocytosis but not transferrin receptor recycling Rab4 induced morphologic changes in early endosomes, We next looked at some of the endocytic processes that have these changes do not have a significant effect on the traf- been reported to require Rab4. To look at fluid phase endo- ficking of the transferrin receptor. Over-expression of CD2AP/ cytosis, control CHO cells or cells stably expressing low CMS [539–637] (Figure 7C,e,f) and CD2AP/CMS [1–328] (clone 6) or high levels (clone 4) of CD2AP/CMS were incu- (Figure 7C,g,h) did not alter the amount of cell surface trans- bated with Horse Radish Peroxidase (HRP). After increasing ferrin receptors. By contrast, Rab4N121I expression led to an period of times, cells were washed and HRP associated with obvious decrease in the amount of transferrin receptors at the cells was quantified. The accumulation of HRP by the the plasma membrane (Figure 7,i,j), most likely as a result of cells was decreased by over-expression of CD2AP/CMS, and an inhibition of transferrin receptor recycling rate induced by the effect was most pronounced for the clone expressing Rab4N121I (22). These results suggest that CD2AP/CMS is high amounts of CD2AP/CMS. There was no difference dur- probably not involved in the Rab4-dependent process that ing the first 15min of incubation with HRP. This phenomenon controls transferrin recycling. was also observed when Rab4 was over-expressed in CHO cells (26) and is thus in agreement with the fact that CD2AP/ The expression of CD2AP/CMS with Rab4 alters the CMS acts as an effector of Rab4. morphology of late endosomes and interferes with the degradation of PDGF receptors We next studied the effect of CD2AP/CMS over-expression c-Cbl acts as an E3 ubiquitin ligase, bringing together tyro- on transferrin receptor recycling in the TRVb-1 cells, a CHO sine phosphorylated receptors and E2 ubiquitin-conjugating cell line that over-expresses the human transferrin receptor. enzymes. It thus induces the polyubiquitination of activated No change in the rate constant for transferrin receptor recyc- tyrosine kinase receptors (including Epidermal Growth Fac- ling was observed in cells over-expressing CD2AP/CMS tor Receptors and PDGF receptors), that would enhance compared with control cells. The recycling rate constant, cal- their transport into late endosomes/lysosomal compart- culated from two independent clones of TRVb-1-CD2AP/ ments in which the receptors are degraded (27,28). There- CMS (Figure 7B, inset), was similar in both cell lines. Further, fore, we next investigated whether CD2AP/CMS regulates the expression of CD2AP/CMS did not change the steady- communication between sorting endosomes, where Rab4 state distribution of transferrin receptors between the interior resides, and late endosomes. We found that GFP-Rab7 (a and the surface of cells. Taken together, these observations marker for late endosomes) was present in enlarged ves- indicate that CD2AP/CMS does not induce major modifi- icles produced by the over-expression of CD2AP/CMS with cations in any endocytic parameters of the transferrin recep- Rab4 Q67L (Figure 8A). Thus the morphology of late endo-

Traffic 2002; 4: 97–112 105 Cormont et al.

Figure7: Effect of CD2AP/CMS expression on endocytotic pathways. (A) CD2AP/CMS modifies HRP uptake in CHO cells. HRP uptake was measured, as described in Materials and Methods, in CHO cells stably over-expressing GFP-CD2AP/CMS or control cells (WT). Two clones were used with low (cl 6) or high (cl 4) levels of GFP-CD2AP/CMS. Results are expressed as the percentage of the HRP uptake measured at the end of the 4-h incubation period in control cells. Results are the means∫SEM of 4 independent experiments. *p Ͻ0.01 using Student’s t-test for paired analysis. (B) CD2AP/CMS does not modify the recycling of transferrin receptors. Control TRVb-1 cells (open symbols) or cells stably expressing CD2AP/CMS were treated as indicated in Materials and Methods to measure the recycling of iodinated transferrin from the endosomal recycling compartment. The graph represents the percentage of 125I-transferrin remaining associated with the cells at increasing periods of times after induction of the recycling. Inset reports the calculated recycling rate of 125I-transferrin obtained for two independent clones over-expressing CD2AP/CMS compared to control cells. Results are the means∫SEM of 3 experiments. (C) CD2AP/CMS does not affect the membrane distribution of transferrin receptors. A549 cells were cultured until 70% of confluence before transfection with pEGFP-Rab4Q67L and myc-CD2AP/CMS (a,b,c,d), pcDNA3-myc-CD2AP/CMS [539–637] (e,f), pcDNA3-myc-CD2AP/ CMS [1–328] (g,h), or pEGFP-Rab4N121I (i,j). Cells were treated for immunofluorescence as described in Materials and Methods. In the left panels, red labeling corresponds to plasma membrane transferrin receptors (a,b,e,f,g,h,i,j) or total receptors cells (c and d). The right panels show the green labeling corresponding to transfected cells in a merge image.

106 Traffic 2002; 4: 97–112 CD2AP/CMS: a Novel Rab4 Effector

Figure8: CD2AP/CMS modifies late endosomes morphology and PDGF-R b degradation. (A) GFP-Rab7 localizes into enlarged ves- icles induced by CD2AP/CMS and Rab4 coexpression. CHO cells were transiently trans- fected with pcDNA3-myc- CD2AP/CMS, pCis-Rab4 Q67L and pEGFP-Rab7 WT. GFP-Rab7 is shown in a. CD2AP/CMS is detected using anti-myc antibodies fol- lowed by anti-mouse coupled to Texas Red (b), while Rab4 is detected with anti-Rab4 antibodies followed by anti- rabbit coupled to Cy5 (c). Panel d shows the overlay of the images corresponding to GFP-Rab7 and myc-CD2AP/ CMS. (B) Effect of CD2AP and CD2AP/CMS [539–637] on ligand-induced PDGF-R b degradation. 293 cells grown in a 35-mm well are transfect- ed with 0.2mg of pcDNA3- PDGF-R b,6mg of pcDNA3- myc-CD2AP/CMS or myc- CD2AP/CMS [539–637]. Two days after transfection, cells were serum starved overnight and then incubated for 5, 30 or 210min with PDGF-BB (50ng/ml). Lysates were pre- pared, analyzed by SDS- PAGE and immunodetection was performed using anti- PDGF-R b antibodies. The figure shows the quantifi- cation∫SEM of three inde- pendent experiments and a representative immunode- tection.

Traffic 2002; 4: 97–112 107 Cormont et al. somes positive for GFP-Rab7 is highly modified by the These might explain why we do not find CD2AP/CMS and coexpression of active Rab4 with CD2AP/CMS, while it is Rabip4 colocated in the cell. unaffected by the coexpression of Rab4 with Rabip4, an- other effector of Rab4 (23). We propose that CD2AP/CMS could act as a Rab4 effector involved in sorting from early to late endosomes. Indeed, To test whether CD2AP/CMS might affect the down-regula- HRP uptake was modified by CD2AP/CMS over-expression tion of activated tyrosine kinase receptors, 293 cells were at late time points and the morphology of Rab7 positive en- transfected with pcDNA3-PDGF-R b alone or together with dosomes was affected by the coexpression of Rab4 and pcDNA3-myc-CD2AP/CMS or pcDNA3-myc CD2AP/CMS CD2AP/CMS. Finally, the ligand-induced degradation of [539–637] and then incubated with PDGF for increasing PDGF receptors was inhibited by the expression of the C- periods of time (Figure 8B). The total amount of PDGF-R b terminal end of CD2AP/CMS, which probably acts as a domi- was then measured. We observed that CD2AP/CMS slightly nant negative protein by complexing active Rab4. A role of increased the degradation of PDGF-R b, to a degree similar Rab4 in such a pathway (early to late endosomes) is sup- to that obtained with over-expression of c-Cbl in our experi- ported by the fact that EGF degradation was inhibited by a mental conditions (data not shown). More evident was the form of Rab4 unable to bind GTP (22). Moreover, CD2AP/ observation that CD2AP/CMS [539–637] abolished the CMS does not appear to be involved in the recycling of trans- degradation of PDGF-R b normally induced by PDGF. This ferrin receptors. The steady-state distribution of the transfer- result suggests that CD2AP/CMS [539–637] might block the rin receptors between the plasma membrane and its intra- degradation of stimulated PDGF-R b, perhaps by affecting cellular locations was not changed by the over-expression of the sorting of the receptors into the late endosomes. CD2AP/CMS, suggesting that the route of transferrin recep- tors is unmodified. Consistent with this observation, the re- cycling of transferrin from the recycling endosomes to the Discussion plasma membrane was not markedly affected by the over- expression of CD2AP/CMS as it is the case for active or WT In this report we present data indicating that the adaptor pro- Rab4 (22). However, while Rab4N121I decreased the tein, CD2AP/CMS, is a Rab4 effector. CD2AP/CMS interacts amount of cell surface transferrin receptors, CD2AP/CMS with Rab4-GTP in the yeast two-hybrid system but not with [539–637], that could act as a dominant negative protein by wt Rab4, Rab4-GDP or several other Rab proteins tested. The complexing active Rab4, had no significant effect. CD2AP/ interaction between the two proteins (active Rab4 and CMS is thus probably not involved in the Rab4-dependent CD2AP/CMS) was also evidenced in vitro. It should be noted step involved in the traffic to recycling endosomes. that CD2AP/CMS and Rabip4 interact only with Rab4 and not with the other Rab proteins of early endosomes, i.e. Rab5 The entire CD2AP/CMS molecule was required to induce, in or Rab11. In this respect, they differ from the Rab4 dual effec- a Rab4-dependent way, the morphological alteration of the tors such as Rabaptin-5 (29), Rabenosyn-5 (30) and the Rab early endosomes, an indication that partners of CD2AP/CMS coupling protein RCP (31,32), which interact both with Rab4 interacting with the SH3 domains and/or the proline-rich se- and Rab5 or Rab4 and Rab11, respectively. This suggests that quences are required. Among these factors, the proto-onco- the function of CD2AP/CMS as an effector of Rab4 is not gene c-Cbl appears to play a key role. First, as confirmed in coupled to the activity of Rab5 and Rab11. We also noted this paper, CD2AP/CMS effectively interacts with c-Cbl (14). that a putative consensus sequence for Rab4 binding does Second, the coexpression of CD2AP/CMS with c-Cbl induces not appear through the comparison of all the known Rab4 the same alterations of endosomes as does its coexpression effectors (23,30,31,33–35). with Rab4. The SH3 domains of CD2AP/CMS, which interact with c-Cbl, are not sufficient to induce these alterations. A functional interaction between Rab4 and CD2AP/CMS is These observations suggest that CD2AP/CMS, through its in- implied by the finding that coexpression of CD2AP/CMS and teractions with both c-Cbl and Rab4, controls early endosom- Rab4 resulted in an enlargement of EEA1-containing ves- es morphology. This probably requires signals that activate icles. The mechanism involved in the change in early endo- tyrosine kinase, since the interaction between CD2AP/CMS some morphology induced by the coexpression of Rab4 with and c-Cbl in intact cells occurs only when c-Cbl become tyro- its effectors, i.e. Rabip4 (23,24) or CD2AP/CMS (this paper), sine phosphorylated (14). is not understood, but it probably results from two indepen- dent processes. Indeed, Rab11 is found in the structures c-Cbl is involved in the desensitization of activated tyrosine formed by the coexpression of Rabip4 and Rab4, while Rab7 kinase receptors. Two independent mechanisms might ex- is excluded (23). At the opposite, Rab7 but not endogenous plain this ligand-induced desensitization process. First, c-Cbl Rab11 (data not shown) is included in the enlarged endo- is involved in the ubiquitination of tyrosine phosphorylated somes formed by the coexpression of CD2AP/CMS and receptors (27,36), a modification that constitutes a signal to Rab4. These results might indicate that Rab4 controls two enter the endocytic degradative pathway (37). Second, c-Cbl independent sorting steps from sorting endosomes, one to recruits endophilin (16,17), which through its interaction with the recycling endosomes and another to the multivesicular phospholipidic membrane bilayer, generates high curvature body/late endosomes, by interacting with distinct effectors. membranes that might promote the invagination of clathrin-

108 Traffic 2002; 4: 97–112 CD2AP/CMS: a Novel Rab4 Effector coated vesicles (38,39). The recruitment of endophilin to c- ends (RACE), using a library of 3T3-L1 adipocyte cDNAs Cbl requires CIN85 (16,17), a c-Cbl interacting protein that made with the Marathon cDNA Amplification Kit (BD Clon- possesses the same molecular organization as CD2AP/CMS tech, Palo Alto, CA, USA), according to the manufacturer pro- (5,7). Our data suggest that the interaction between c-Cbl tocols. To amplify the complete coding sequence of the pro- and CD2AP/CMS, a Rab4 effector, might constitute another tein, oligonucleotides corresponding to the sequences before way to control ligand-induced receptor desensitization. the ATG initiation codon and after the 3ƒ stop codon were Further investigations will be required to determine the mole- selected. The obtained cDNA was automatically sequenced cular mechanisms involved in these processes. Further, it is on the double strands. It corresponds to the cDNA which also possible that CD2AP/CMS might link extracellular sig- encodes for CD2AP/CMS (1,2). nals to Rab4-regulated functions, such as insulin-induced Glut 4 translocation (40), induced a-granules secretion in CDNA constructs platelets (41), regulated amylase secretion by pancreatic acini pLexA-Rab4 wild-type (WT) or mutated forms were de- (42) or PDGF-stimulated avb3 integrin recycling (43). scribed in (23). The CD2AP/CMS and CD2AP/CMS [1–539] sequences were subcloned into the pACT2 vector (BD Clon- tech, Palo Alto, CA, USA). cDNA sequences encoding for the entire CD2AP/CMS or the fragments 539–637 or 1–328 Materials and Methods were subcloned into pcDNA3.1 (Invitrogen, Breda, the Netherlands) containing an N-terminal myc epitope (23). Antibodies and ligands CD2AP/CMS was subcloned in pEGFP (BD Clontech, Palo Monoclonal antibodies (mAb) against the myc epitope Alto, CA, USA) and was used for expression of CD2AP/CMS (9E10), EEA1, human transferrin receptor and PDGF-R b were as a C-terminal fusion with the green fluorescent protein from Santa Cruz Biotechnology, Inc (Santa Cruz, CA, USA), (GFP). The sequence encoding for myc-CD2AP/CMS were BD Pharmingen and Transduction Laboratories (Erembodeg- also subcloned into pIREShyg2 (BD Clontech, Palo Alto, CA, em, Belgium) and Mononsan (Uden, the Netherlands), Upst- USA). Truncated forms of CD2AP/CMS were obtained by in- ate Biotechnogy (Lake Placid, NY, USA), respectively. Rabbit troduction of stop mutations at the indicated position. polyclonal anti-Rab4 has been described in (44). Polyclonal anti-CD2AP/CMS was obtained by immunizing a rabbit with Cells and transfection the glutathione-S-transferase-CD2AP/CMS-[539–637] fu- Chinese hamster ovary and TRVb-1 cells were grown in sion protein. Secondary antibodies coupled to Horse Radish Ham’s F12 medium with 10% FCS at 37æC in a humidified Peroxidase (HRP) or with fluorochromes were from Jackson atmosphere of 5% carbon dioxide in air. CHO cells were tran- ImmunoResearch Laboratories, Inc (West Grove, PA, USA). siently transfected by electroporation. Briefly, cells (1– Texas Red coupled Phalloidin was from Molecular Probes 2¿10 6/400ml of Ham’s F12) were placed in a 0.4-cm gap (Eugene, OR, USA). Horse Radish Peroxidase and human cuvette along with 10–50mg of plasmids and electroporated transferrin (Tf) were obtained from Sigma-Aldrich (Saint (260V-1050 mF) with an Easyject electroporator system. Vari- Louis, Missouri, USA). 125I-Tf was prepared as described ous CHO cell lines stably expressing GFP-CD2AP/CMS were (45). PDGF-BB was from Pepro Tech. Inc. (Rocky Hill, NJ, obtained by selection with G418 (500mg/ml) and limit dilution USA). of the transfected cells. TRVb-1 cell lines [CHO cells derived that do not express endogenous hamster transferrin recep- Two-hybrid screening and interaction measurements tors but over-express human transferrin receptors (45)] The yeast reporter strain L40 was transformed with pLexA- stably over-expressing Myc-CD2AP/CMS were obtained Rab4 Q67L using a lithium acetate-base method and grown after transfection with pIREShyg2-Myc-CD2AP/CMS and in synthetic medium lacking tryptophan (46). It was trans- selection with Hygromycin. HEK 293 cells were grown in Dul- formed with a TA adipocyte library made in pVP16 plasmid. becco’s modified Eagle’s medium with 10% FCS at 37æCin Cells were plated on synthetic medium lacking leucine, tryp- a humidified atmosphere of 5% carbon dioxide in air. Cells tophan and histidine. Growing colonies were tested for b- were transfected when they achieved 70–80% confluence galactosidase activity. Library plasmids from positive clones using the calcium/phosphate precipitation procedure. The were rescued into Escherichia coli HB101 cells plated on leu- amount of plasmid DNA used will be mentioned in the figure cine-free medium and analyzed by transformation tests and legends. Cells were used 2days after the transfection follow- DNA sequencing. For interaction measurement, cotransform- ing an overnight serum starvation. ed L40 yeasts were selected on medium lacking leucine and tryptophan, and induction of the reporter gene LacZ was Interaction measurements by a pull-down assay quantified (23). The pull-down assay was performed as described (30,47). Briefly GST-Rab4 was purified (48) and loaded with GDP or Cloning of the full-length CD2AP/CMS cDNA GTPgS. One hundred micrograms of GST-Rab4 immobilized One positive clone of 402bp (named A3) was selected. The on glutathione-sepharose beads were incubated with 45mlof full-length cDNA was obtained by successively screening a myc-CD2AP/CMS or myc-Rabip4 transcribed and translated phage lGT11 library of 3T3-L1 adipocytes (BD Clontech, Palo in vitro in the presence of 35S-methionine using the TnT Alto, CA, USA) and 5ƒ and 3ƒ rapid amplifications of cDNA coupled transcription-translation kit (Promega, Madison, WI,

Traffic 2002; 4: 97–112 109 Cormont et al.

USA) or 1mg of cytosol prepared from 293 cells transfected Binding of 125I-Tf at the cell surface and recycling with pcDNA3-myc CD2AP/CMS. After 2h of incubation at assay of Tf 4æC in buffer containing 20mM Hepes, pH 7.8, 100mM NaCl, For each biochemical assay, TRVb-1 or TRVb-1CD2AP/CMS 5mM MgCl2,1mM DTT and 1mM guanine nucleotide (GDP cells were grown in 6-well clusters. Assays to measure TR or GTPgS), beads were washed twice with 20mM Hepes, pH kinetics have been described previously (49). Cells were in-

7.8, 150mM NaCl, 5mM MgCl2,1mM DTT and 10mM of the cubated in Ham’s F12 medium supplemented with 20mM adequate guanine nucleotide, and twice with 20mM Hepes, Hepes, pH7.4, for 1h. To measure surface TfR, cells were pH 7.8, 250mM NaCl, 1mM DTT. Beads were then incubated then placed on ice and incubated for 2h with 125I-Tf (3mg/ with 100mlof20mM Hepes pH, 7.8, 1 M NaCl, 20mM EDTA, ml) in the previous medium supplemented with 2mg/ml ov- 1mM DTT and 5mM GDP to elute the proteins which were albumin. Cells were washed with 20mM Hepes, pH7.4 con- associated in a GTP-dependent fashion. The eluate were ana- taining 150mM NaCl, 1mM CaCl2,5mM KCl, 1 mM MgCl2 lyzed by SDS-PAGE and either autoradiography(35S) or (neutral wash buffer) to remove unbound 125I-Tf, prior to Western blot using an anti-myc antibody. measuring the radioactivity associated with the cells in a g- counter. Non-specific binding is determined in the presence Confocal immunofluorescence microscopy of a 100-fold excess of unlabeled Tf and represents typically Cells grown on glass coverslips were washed and fixed in less than 10% of the total. To measure total TfR, cells were 4% paraformaldehyde. Cells expressing GFP-fusion proteins incubated with 3mg/ml 125I-Tf for 1.5 at 37æC, washed with were directly mounted in Mowiol (Hoechst) and examined in neutral wash buffer, solubilized and cell-associated radioac- scanning confocal fluorescence microscopy (TCS SP, Leica, tivity was determined. Deerfield, IL, USA) with a PL APO 63¿1.40 oil objective (Le- ica, Deerfield, IL, USA). To detect non-GFP-fusion proteins, The rate at which internalized Tf is released from cells is used cells were permeabilized in PBS pH7.4, 0.1% Triton X-100, as a measure of externalization rate constant. Cells were incu- 1% fetal bovine serum, incubated with the appropriate pri- bated for 1h in Ham’s F12 medium supplemented with mary antibodies, washed and incubated with secondary anti- 20mM Hepes, pH7.4, and then 3mg/ml 125I-Tf together with bodies coupled with Texas Red or Cy5 fluorochromes. To de- 2mg/ml ovalbumin for 2h. The cells were washed in neutral tect plasma membrane transferrin receptors, incubation with wash buffer and incubated for 2min in mild acid wash buffer anti-transferrin receptors was performed prior to permeabil- (50mM MES, pH5.0; 0.5 M NaCl). Cells were rapidly re- ization and adding the appropriate other antibodies. The cells washed with the neutral wash buffer and incubated in Ham’s were examined by sequential excitation at 488nm (GFP), F12 medium containing 20mM Hepes, pH7.4, 2mg/ml oval- 568nm (Texas-Red) and 647nm (Cy5). The images were bumin, 100mM desferrioxamine and 3mg/ml of unlabeled Tf then combined and merged by using PhotoShop (Adobe sys- at 37æC for increasing periods of time. At the end of the time- tem, Mountain View, Ca, USA). period, the efflux media were collected and the cells were solubilized in 2ml of 0.1% Triton X100 containing 0.1% Subcellular fractionation NaOH. The radioactivity of both the efflux media and the sol- Chinese hamster ovary cells (2 preconfluent 100-mm dish- ubilized cells were counted separately in a g counter. The rate es) were homogenized with a Thomas potter (type AA) in constant of externalization, ke, was determined by the slope 300ml of Tris-HCl, pH7.4, containing 1mM EDTA, 250mM of the natural logarithm of the percent Tf remaining cell-as- sucrose, and a protease inhibitor cocktail (CompleteTM, sociated vs. time. Boehringer Manheim, Germany). Homogenates were centri- fuged at 100000¿g for 1h (Optima X100 ultracentrifuge, Beckman, Palo Alto, CA, USA) to separate the cytosol and Acknowledgments the total membrane pellet. The pellet was then resus- pended in 300ml of Tris-HCl, pH7.4, 1mM EDTA, 250mM We thank A. Tavitian and A. Zahraoui (Institut Curie, Paris) for the gift of sucrose and protease inhibitors (24). Proteins from equal Rab4 cDNA, B. Goud (Institut Curie, Paris) for pLex-Rab5 Q79L, pLexRab6 volumes of homogenate, cytosol or membrane-associated Q72L and Ras G12V, S. Me´resse (CIML, Marseille) for Rab5 and Rab7 fractions were resolved by SDS-PAGE and analyzed by cDNA, C-H. Helding for the gift of PDGF-R b cDNA (Ludwig Institute for immunodetection. Cancer Research, Uppsala). c-Cbl constructs were a gift of Y.C. Liu (La Jolla Institute for Allergy and Immunology, San Diego). The TA adipocyte Measurement of HRP-uptake cDNA library was a gift of Dr Hollenberg (Fred Hutchinson Cancer Re- Wild-type CHO and CHO/GFP-CD2AP/CMS were serum search Center, Seattle). S. Mari is acknowledged for technical assistance. starved for 2h before the experiment. Cells were incubated This work was supported by the Institut National de la Sante´ et de la Re- during increasing periods of time with Horse Radish Peroxi- cherche Me´dicale (Special grant APEX 97–01), the Juvenile Diabetes Fed- eration International (JDFI Grant 198302), the Association pour la Recher- dase (0.8mg of HRP in Ham’s F12/ 10mM Hepes pH7.4/ che contre le Cancer (ARC Grants 5634 and 7449), the French Association 0.5% BSA). Cells were successively washed with PBS 2% against the Myopathy (grant 7989), the Re´gion Provence Alpes Coˆte d’A- BSA and with PBS. They were lysed in PBS containing 0.1% zur and the Conseil Ge´ne´ral des Alpes Maritimes. The support of Fondation Triton X-100. Cellular HRP uptake was quantified by measur- Bettencourt-Schueller is gratefully acknowledged. I. Meton received a fel- ing HRP activity in the lysate using o-phenylenediamine dihy- lowship from INSERM (Poste vert). M. Mari and P. Monzo were supported drochloride as a substrate. by Fondation Pour la Recherche Me´dicale (Paris, France), and La Ligue

110 Traffic 2002; 4: 97–112 CD2AP/CMS: a Novel Rab4 Effector contre le Cancer, respectively. C. Gaskin and T.E. MacGraw were supported CIN85-endophilin complex mediates ligand-induced downregulation in part by NIH DK57689 (TEM). of EGF receptors. Nature 2002;416:183–187. 18. Cormont M, Bortoluzzi M-N, Gautier N, Mari M, Van Obberghen E, Le Marchand-Brustel Y. Potential role of Rab4 in the regulation of subcel- References lular localization of Glut4 in adipocytes. Mol Cell Biol 1996;16:6879– 6886. 1. Dustin ML, Olszowy MW, Holdorf AD, Li J, Bromley S, Desai N, Widd- 19. Cormont M, Mari M, Meton I, Le Marchand-Brustel Y. Cloning and er P, Rosenberger F, van der Merwe PA, Allen PM, Shaw A. A novel characterization of two effectors of Rab4, a small GTPase involved in adaptor protein orchestrates receptor patterning and cytoskeletal po- Glut4 subcellular distribution. Diabetes 2000;49 (Suppl. 1):A29. larity in T-cell contacts. Cell 1998;94:667–677. 20. Lupas A. Coiled coils: new structures and new functions. Trends Bio- 2. Kirsch KH, Georgescu M-M, Ishimaru S, Hanafusa HCMS. An adapter chem Sci 1996;21:375–382. molecule involved in cytoskeletal rearrangements. Proc Natl Acad Sci 21. Lawe DC, Patki V, Heller-Harrison R, Lambright D, Corvera S. The FYVE 1999;96:6211–6216. domain of early endosome antigen 1 is required for both phosphati- 3. Lehtonen S, Ora A, Olkkonen VM, Geng L, Zerial M, Somlo S, Lehton- dylinositol 3-phosphate and Rab5 binding. J Biol Chem 2000;275: en E. In vivo interaction of the adapter protein CD2-associated protein 3699–3705. with the type 2 polycystic disease protein, Polycystin-2. J Biol Chem 22. McCaffrey MW, Bielli A, Cantulapo G, Mora S, Roberti V, Santillo M, 2000;275:32888–32893. Drummond F, Bucci C. Rab4 affects both recycling and degradative 4. Gout I, Middleton G, Adu J, Ninkina NN, Drobot LB, Filonenko V, Mat- endosomal trafficking. FEBS Lett 2001;495:21–30. suka G, Davies AM, Waterfield M, Buchman VL. Negative regulation 23. Cormont M, Mari M, Galmiche A, Hofman P, Le Marchand-Brustel Y. of PI3-kinase by Ruk, a novel adaptor protein. EMBO J 2000;19: A FYVE-finger-containing protein, Rabip4, is a Rab4 effector involved 4015–4025. in early endosomal traffic. Proc Natl Acad Sci USA 2001;98:1637– 5. Take H, Watanabe S, Takeda KYuZ-X, Iwata N, Kajigaya S. Cloning 1642. and characterization of a novel adaptor protein, CIN85, that interacts 24. Mari M, Macia E, Le Marchand-Brustel Y, Cormont M. Role of the with c-Cbl. Biochem Biophys Res Commun 2000;268:321–328. FYVE-finger and the RUN domain for the subcellular localization of Rabip4. J Biol Chem 2001;276:42501–42508. 6. Bogler O, Furnari FK, Kindler-Roehrborn A, Sykes VW, Yung R, Huang 25. Bard F, Patel U, Levy JB, Horne WC, Baron R. Molecular complexes HG, Cavenee WK. SETA: a novel SH3 domain-containing adapter mol- that contain both c-Cbl and c-Src associated with Golgi membranes. ecule associated with malignancy in astrocytes. Neuro-Oncol 2000;2: Eur J Cell Biol 2002;81:26–35. 6–15. 26. van der Sluijs P, Hull M, Webster P, Maˆle P, Goud B, Mellman I. The 7. Dikic I. CIN85/CMS family of adaptor molecules. FEBS Lett 2002; small GTP-binding protein rab4 controls an early sorting event on the 529:110–115. endocytic pathway. Cell 1992;70:729–740. 8. Shih N-Y, Li J, Karpitskii V, Nguyen A, Dustin ML, Kanagawa O, Miner 27. Tsygankov AY, Teckchandani AM, Feshchenko EA, Swaminathan G. JH, Shaw AS. Congenital nephrotic syndrome in mice lacking-CD2- Beyond the RING. CBL proteins as multivalent adapters. Oncogene associated protein. Science 1999;286:312–315. 2001;20:6382–6402. 9. Shih NY, Li J, Cotran R, Mundel P, Miner JH, Shaw AS. CD2AP local- 28. Clague MJ, Urbe´ S. The interface of receptor trafficking and signalling. izes to the slit diaphragm and binds to nephrin via a novel C-terminal J Cell Sci 2001;114:3075–3081. domain. Am J Pathol 2001;159:2303–2308. 29. Vitale G, Rybin V, Christoforidis S, Thornqvist P, McCaffrey M, Sten- 10. Schwarz K, Simons M, Reiser J, Saleem MA, Faul C, Kriz W, Shaw mark H, Zerial M. Distinct Rab-binding domains mediate the interac- AS, Holzman LB, Mundel P. Podocin, a raft-associated component of tion of Rabaptin-5 with GTP-bound Rab4 and Rab5. EMBO J the glomerular slit diaphragm, interacts with CD2AP and nephrin. J 1998;17:1941–1951. Clin Invest 2001;108:1621–1629. 30. De Renzis S, Sönnichsen B, Zerial M. Divalent Rab effectors regulate 11. Kestila M, Lenkkeri U, Manniko M, Lamerdin J, McCready P, Putaala H, the sub-compartmental organization and sorting of early endosomes. Ruotsalainen V, Morita T, Nissinen M, Herva R, Kashtan CE, Peltonen L, Nat Cell Biol 2002;4:124–133. Holmberg C, Olsen A, Tryggvason K. Positionally cloned gene for a 31. Lindsay A, Hendrick A, Cantalupo G, Senic-Matuglia F, Goud B, Bucci novel glomerular protein – nephrin – is mutated in congenital neph- C, McCaffrey MW. Rab coupling protein (RCP), a novel Rab4 and rotic syndrome. Mol Cell 1998;1:575–582. Rab11 effector protein. J Biol Chem 2002;277:1219 0–1219 9. 12. Boute N, Gribouval O, Roselli S, Benessy F, Lee H, Fuchshuber A, 32. Wallace DM, Lindsay AJ, Hendrick AG, McCaffrey MW. The novel Dahan K, Gubler MC, Niaudet P, Antignac C. NPHS2, encoding the Rab11-FIP/Rip/RCP family of proteins displays extensive homo- and glomerular protein podocin, is mutated in autosomal recessive ster- hetero-interacting abilities. Biochem Biophys Res Commun 2002; oid-resistant nephrotic syndrome. Nat Genet 2000;24:349–354. 292:909–915. 13. Schneider MC, Rodrigez AM, Nomura H, Zhou J, Morton CC, Reeders 33. Bielli A, Thornqvist PO, Hendrick AG, Finn R, Fitzgerald K, McCaffrey ST, Weremowicz S. A gene similar to PKD1 maps to M. The small GTPase Rab4A interacts with the central region of cyto- 4q22: a candidate gene for PKD2. Genomics 1996;38:1–4. plasmic dynein light intermediate chain-1. Biochem Biophys Res Com- 14. Kirsch KH, Georgescu M-M, Shishido T, Langdon WY, Birge RB, Han- mun 2001;281:1141–1153. afusa H. The adapter type protein CMS/CD2AP binds to the proto- 34. Nagelkerken B, van Anken E, van Raak M, Gerez L, Mohrmann K, van oncogenic protein c-Cbl through a tyrosine phosphorylation-regulated Uden N, Holthuizen J, Pelkmans L, van der Sluijs P. Rabaptin4, a novel Src homology 3 domain interaction. J Biol Chem 2001;276:4957– effector of the small GTPase , is recruited to perinuclear recyc- 4963. ling vesicles. Biochem J 2000;346:593–601. 15. Scaife RM, Langdon WY. c-Cbl localizes to actin lamellae and regu- 35. Stenmark H, Vitale G, Ullrich O, Zerial M. Rabaptin-5 is a direct effec- lates lamellipodia formation and cell morphology. J Cell Sci 2000; tor of the small GTPase Rab5 in endocytic membrane fusion. Cell 113:215–226. 1995;83:423–432. 16. Petrelli A, Gilestro GF, Lanzardo S, Comoglio PM, Migone N, Giordano 36. Thien CB, Langdon WY. Cbl: many adaptations to regulate protein S. The endophilin-CIN85-Cbl complex mediates ligand-dependent tyrosine kinases. Nature Rev Mol Cell Biol 2001;2:294–307. down regulation of c-Met. Nature 2002;416:187–190. 37. Riezman H. The ubiquitin connection. Nature 2002;416: 17. Soubeyran P, Kowanetz K, Szymkiewicz I, Langdon WY, Dikic I. Cbl- 381–383.

Traffic 2002; 4: 97–112 111 Cormont et al.

38. Farsad K, Ringstad N, Takei K, Floyd SR, Rose K, De Camilli P. Gener- 45. McGraw TE, Dunn KW, Maxfield FR. Phorbol ester treatment in- ation of high curvature membranes mediated by direct endophilin bi- creases the exocytic rate of the transferrin receptor recycling pathway layer interactions. J Cell Biol 2001;155:193–200. independent of serine-24 phosphorylation. J Cell Biol 1988;106: 39. Oved S, Yarden Y. Molecular ticket to enter cells. Nature 1061–1066. 2002;416:133–136. 46. Fields S, Song O-K. A novel genetic system to detect protein–protein 40. Cormont M, Le Marchand-Brustel Y. The role of small G-proteins in interactions. Nature 1989;340:245–246. the regulation of glucose transport. Mol Membr Biol 2001;18:213– 47. Christoforidis S, Miaczynska M, Ashman K, Wilm M, Zhao L, Yip SC, 220. Waterfield MD, Backer JM, Zerial M. Phosphatidylinositol-3-OH ki- 41. Shirakawa R, Yioshioka A, Horiuchi H, Nishioka H, Tabuchi A, Kita T. nases are Rab5 effectors. Nat Cell Biol 1999;4:249–252. Small GTPase Rab4 regulates Ca2π-induced alpha-granule secretion 48. Bortoluzzi M-N, Cormont M, Gautier N, Van Obberghen E, Le Marc- in platelets. J Biol Chem 2000;275:33844–33849. hand-Brustel Y. GTPase activating protein activity for Rab4 is enriched 42. Ohnishi H, Mine T, Shibata H, Ueda N, Tsuchida T, Fujita T. Involve- in the plasma membrane of 3T3-L1 adipocytes. Possible involvement ment of Rab4 in regulated exocytosis of rat pancreatic acini. Gastro- in the regulation of Rab4 subcellular localization. Diabetologia enterology 1999;116:943–952. 1996;39:899–906. 43. Roberts M, Barry S, Woods A, van der Sluijs P, Norman J. PDGF- 49. Garippa RJ, Judge TW, James DE, McGraw TE. The amino terminus regulated rab4-dependent recycling of avb3 integrin from early endo- of GLUT4 functions as an internalization motif but not an intracellular somes is necessary for cell adhesion and spreading. Curr Biol retention signal when substituted for the transferrin receptor cyto- 2001;11:1392–1402. plasmic domain. J Cell Biol 1994;124:705–715. 44. Cormont M, Tanti J-F, Zahraoui A, Van Obberghen E, Tavitian A, Le Marchand-Brustel Y. Insulin and okadaic acid induce Rab4 redistri- bution in adipocytes. J Biol Chem 1993;268:19491–19497.

112 Traffic 2002; 4: 97–112