CLIPR-59, a new trans-Golgi/TGN cytoplasmic linker protein belonging to the CLIP-170 family

The cytoskeleton plays a fundamental role in cell organization. The microtubule network is particularly important in higher eukaryotes, determining organelle localization and regulating the exchange between membranous compartments (Cole and Lippincott-Schwartz, 1995; Schroer, 2000). In particular, it has long been known that microtubules are required for the acquisition and maintenance of Golgi complex localization and structure, which in fibroblastic cells normally consists of long stacks of cisternae juxtaposed to the microtubule organizing center (Thyberg et al., 1980; Ho et al., 1989; Cole et al., 1996; Tian et al., 1996; Kreis et al., 1997). The intrinsic asymmetry of microtubules is used by the cells to generate cytosolic polarity. The rapidly polymerizing microtubule plus ends generally point toward the cell periphery, whereas the more slowly polymerizing minus ends are associated with the microtubule organizing center. Steady-state organelle localizations are thought to result mainly from a “balance of power” between the activity of plus end–directed kinesin family motors and minus end–directed dynein-related motors (Goodson et al., 1997), and this “equilibrium model” has been particularly well studied in the case of Golgi complex localization (Burkhardt, 1998).

Although these data might suggest that the steady-state localization of organelles like the Golgi complex relies only on the function of these microtubule-based motors, it has been suggested that membranous organelles can also interact with microtubules via nonmotor microtubule-binding (MTB)^* proteins known as cytoplasmic linker proteins (CLIPs). One hypothesis suggests that these CLIPs promote the initial interaction of membranous organelles with microtubules and, once docking is achieved, may participate in the regulation of motor activity (Rickard and Kreis, 1996). CLIP-170 was the first characterized CLIP. In vitro experiments have shown that CLIP-170 is essential for efficient binding of endocytic carrier vesicles to microtubules (Rickard and Kreis, 1990; Pierre et al., 1992), whereas experiments in vivo have suggested that CLIP-170 interacts, directly or indirectly, with the dynein regulator dynactin complex (Valetti et al., 1999; Vaughan et al., 1999).

CLIP-170 is an elongated homodimeric protein bearing an amino-terminal MTB domain. The MTB domain allows CLIP-170 to interact in a particular way with microtubules because it is specifically and dynamically localized at the plus extremity of growing microtubules (Pierre et al., 1992; Perez et al., 1999), probably through a rapid association with polymerizing tubulin subunits (Diamantopoulos et al., 1999). The MTB domain of CLIP-170 contains two repeats of an MTB motif (referred to as the CAP-GLY motif in the PROSITE database) found in other tubulin-interacting proteins (Pierre et al., 1992). The family of proteins containing this motif includes other proteins that may be functionally considered to be CLIPs. Notably, one of the subunits of the dynactin complex, p150^glued, contains a CLIP-170–related MTB domain (Swaroop et al., 1987; Holzbaur et al., 1991). A close relative of CLIP-170, CLIP-115, has been localized in neurons on dendritic lamellar bodies (De Zeeuw et al., 1997), and a CLIP-170 homologue in Drosophila, CLIP-190, has been localized on Golgi membranes (Sisson et al., 2000). Another yet unidentified CLIP-170–related protein is involved in the interaction of peroxysomes with microtubules in vitro (Thiemann et al., 2000).

Not all CLIPs are related to CLIP-170, as exemplified by CLIMP63 (Klopfenstein et al., 1998), an integral membrane protein that links ER membranes to microtubules. A number of other proteins, including the Golgi microtubule–associated protein GMAP-210 (Infante et al., 1999), have also been implicated in membrane–microtubule interactions. However, it is attractive to speculate that the CLIP-170 family contains additional CLIPs. We thus systematically analyzed sequence databases to identify new members of the CLIP family (HVG, FP, and TK in preparation). We report here the cloning and characterization of a 59.5-kD member of this family, CLIPR-59 (CLIP-170–related protein of 59 kD). Like CLIP-170, CLIPR-59 contains two MTB motifs separated by a serine-rich region. Three ankyrin-like repeats and an acidic domain are present amino-terminal to the CLIP domain, whereas the carboxy-terminal region is not strongly related to any protein in the databases. We show that in HeLa cells, CLIPR-59 is localized to Golgi-like structures, most likely related to the TGN, and the carboxy-terminal 60 amino acids are necessary and sufficient for this localization. We further show that the activity of the MTB motifs of CLIPR-59 is regulated in vivo by the adjacent domains, preventing cytoskeletal rearrangement upon overexpression. In contrast, major reorganization and perturbation of TGN/endosomal compartments, together with alteration of retrograde transport, can be observed in these overexpression conditions. Altogether, these data suggest that CLIPR-59 is a TGN CLIP involved in early endosome–TGN dynamics.

We used the CLIP-170 MTB motifs to search the sequence databases for the family of proteins related to CLIP-170. Two sequences (ESTH2 and ESTH4) grouped with each other at a moderate confidence level, suggesting that they might encode related proteins, but did not group with any other known CLIP-170–related protein (Fig. 1 A). The ESTH4 sequence was chosen for further analysis. Corresponding cDNA clones were obtained from the IMAGE consortium and, with the help of RT-PCR, we reconstructed a full-length clone of the 3.4-kb cDNA (Fig. 1 B). The corresponding gene, localized on chromosome 19q13, has been sequenced by the human sequencing project and found to be spliced in 14 exons spanning 17 kb. The open reading frame codes for a 59.5-kD protein that we call CLIPR-59. Because the nomenclature is very confusing in the CLIP field, we propose that the name “CLIPR” may only be given to proteins possessing at least one CLIP-170–related MTB motif. CLIPR-59 is expressed in many different tissues as indicated by EST database analysis (see Unigene Cluster hs.7357). Northern blot analysis confirmed the presence of CLIPR-59 mRNA in various tissues, although we observed that expression in the brain is particularly high (unpublished data).

Schematically, CLIPR-59 possesses a four-domain structure (Fig. 1, B and C). The amino terminus contains an acidic region rich in glutamic acid and proline (E/P region) followed by three ankyrin-like repeats (ANK domain). The second half of the protein contains the two CLIP-170–related MTB motifs separated, as is the case in CLIP-170, by a serine-rich region. Finally, the carboxy-terminal region following the two MTB motifs shows no significant similarity to any protein in the databases.

We raised anti–CLIPR-59 antibodies in rabbits, and Western blot analysis using affinity-purified anti-48 antibody (Fig. 2 A) showed that this antibody detects a faint band at around 60 kD in HeLa cell extracts. An additional stronger band is detected in the extract of cells transfected with HA•CLIPR-59 (also detected by an anti-HA antibody; unpublished data). Immunofluorescence experiments showed that anti-48 stains a Golgi-like structure (Fig. 2 B), although careful analysis suggested that it only partially colocalizes with the cis/medial marker GM-130. This staining was obtained in nontransfected HeLa cells only upon long antibody incubation time and was much stronger after transfection of tagged or untagged CLIPR-59 (unpublished data), further suggesting that CLIPR-59 is only weakly expressed in HeLa cells. Immunofluorescence analysis of HeLa cells transiently expressing HA•CLIPR-59 (Fig. 3, a–c) or stably expressing green fluorescent protein (GFP)•CLIPR-59 (Fig. 3, d–f) showed that the recombinant proteins localized to Golgi-like structures very similar to the structures stained by the anti-48 antibodies in untransfected HeLa cells. Similarly, the colocalization was only partial with the different Golgi markers tested so far (CTR433, GM-130, mannosidase II, galactosyltransferase, TGN46; Figs. 3 and 5, and unpublished data).

We further analyzed CLIPR-59 localization by immunogold labeling of cryosections of HeLa cells transiently or stably expressing GFP•CLIPR-59 (Fig. 4). In agreement with our immunofluorescence data, GFP•CLIPR-59 was detected on Golgi stacks as well as on tubulovesicular elements juxtaposed to Golgi cisternae. Colabeling experiments indicated that GFP•CLIPR-59 was localized on the same side of Golgi stacks as galactosyltransferase. This suggests that CLIPR-59 is localized to membranes of the trans-Golgi/TGN.

Thus far, no membrane-targeting domains have been characterized in proteins of the CLIP-170 family. Therefore, we were particularly interested in studying the regions of CLIPR-59 responsible for membrane targeting. Deletion of the ankyrin repeat–containing amino terminus (CLIPR-59–ΔANK) had no effect on the subcellular localization of the expressed protein (Fig. 5 b). In contrast, the deletion of the last 60 amino acids of CLIPR-59 was sufficient to completely abolish Golgi targeting (Fig. 5 c). This 60–amino acid domain is not only necessary but sufficient for Golgi targeting because the HA-tagged (and GFP-tagged, unpublished data) carboxy-terminal domain of CLIPR-59 efficiently localized to the Golgi (Fig. 5 d). This indicates that the carboxy terminus of CLIPR-59 is a Golgi localization domain (GoLD), but its sequence is different from previously described GoLDs (see Discussion).

Although CLIPR-59 possesses two MTB motifs highly related to those of CLIP-170, neither endogenous nor recombinant CLIPR-59 was observed to localize to microtubules, and overexpression of CLIPR-59 had no discernible effect on the microtubule cytoskeleton. This was unexpected because constructs containing the MTB motifs of CLIP-170 localize to microtubules and provoke microtubule bundling upon overexpression (Pierre et al., 1994). A possible explanation was that the GoLD signal of CLIPR-59 was dominant to the MTB domain, precluding quantitative microtubule localization of CLIPR-59. We thus tried to reveal microtubule binding of the cytosolic mutant CLIPR-59–ΔC60 by preextracting the cells with Triton X-100 before fixation (Fig. 6). Under these conditions, full-length GFP-tagged CLIPR-59 stained central structures, and no significant microtubule localization could be observed (Fig. 6, a–c). As expected, GFP•CLIPR-59–ΔC60 was dispersed in the cytosol but still did not show extensive microtubule localization (Fig. 6, d–f). However, upon closer examination, it was clear that the GFP•CLIPR-59–ΔC60 faintly stained some filamentous structures that were colocalized with microtubules (Fig. 6, arrowheads). This observation suggested that CLIPR-59 devoid of GoLD can bind microtubules in vivo. However, the apparent weakness of this association suggested that the amino terminus containing the E/P domain and the ankyrin repeats might also be inhibitory. To test this hypothesis, we expressed in HeLa cells the GFP-tagged CLIPR-59 MTB domain and observed under the same conditions as above. GFP•CLIPR-59–MTB (or HA•CLIPR-59–MTB, unpublished data) showed comparatively strong microtubule binding (Fig. 6, g–i). Indeed, this construct was even able to induce microtubule bundling upon overexpression, in a dose-dependent manner. Careful examination revealed that the colocalization between GFP•CLIPR-59–MTB and microtubules was only partial. This suggests that only a subset of microtubules are recognized by CLIPR-59. Similar behavior had already been described for CLIP-170 (Pierre et al., 1992; Diamantopoulos et al., 1999; Perez et al., 1999).

We noticed that high overexpression of full-length CLIPR-59 often resulted in loss of colocalization between the CLIPR-59 proteins and Golgi markers. In as many as 35% of transfected cells (depending on length and efficiency of transfection), overexpressed CLIPR-59 accumulated in one or more discrete locations, usually juxtaposed to Golgi membranes (Fig. 7, a–c). Because immunoelectron microscopy indicated that CLIPR-59 localizes to the trans/TGN part of the Golgi, we tested the localization of the TGN marker TGN46 under these conditions. We observed that CLIPR-59 overexpression leads to reduced TGN46 staining in the Golgi region (Fig. 7, d–f; see also Fig. 10 A). The steady-state localization of TGN46 results from active recycling between the plasma membrane, early/recycling endosomes, and the TGN. It was thus interesting to check the localization of other markers in this pathway. We found that the localization of both transferrin receptor (TfR, a marker of early/recycling endosomes) and Rab11 (a marker of the recycling endosome) was strongly perturbed. Moreover, we observed extensive colocalization of TfR- and Rab11-positive membranes with juxtanuclear CLIPR-59 aggregates (Fig. 7, g–l). Similarly, we found that the punctiforme early endosomes positive for EEA1 were depleted from the cell periphery and coaccumulated with CLIPR-59, although EEA1 staining seemed more diffuse than that of Rab11 or TfR (Fig. 7, m–o). In the same conditions, we observed no significant effect on the localization of lyso-bis-phosphatidic acid (LBPA)–containing late endomes or lysosomes (Fig. 7, p–r), or on the localization of cation-independent mannose-6-phosphate receptor (unpublished data). In agreement with our previous experiments (Fig. 6), no effect on microtubule organization could be observed under these conditions. It is also worth noting that CLIPR-59 lacking the GoLD does not affect TGN/endosome localization (unpublished data).

Immunoelectron microscopy confirmed the presence of accumulated, CLIPR-59–positive, vesicular membranes in a juxta-Golgi location (Fig. 8 a). These tubulo-vesicular membranes were densely packed and not homogenous in size (Fig. 8, b and c). It is worth noting that some of the vesicular structures present in these membrane clusters appeared to be coated (Fig. 8 e), although more work is necessary to address the nature of this coat and the frequency of such a coating. We noticed that overexpression sometimes led to plasma membrane staining (Fig. 8 d), which was also observed by immunofluorescence (unpublished data). Finally, in agreement with our immunofluorescence data, immunoelectron microscopic analysis showed that TfR-positive vesicular membranes were present in these aggregated membranes (Fig. 8 f).

Because CLIPR-59 overexpression altered the localization of early and recycling endosomes, we tested whether these aggregated endosomes were still functional in internalizing rhodamine-labeled transferrin (Rh–Tf). We found that transferrin could be actively endocytosed by GFP•CLIPR-59–transfected cells. As observed before for its receptor, the localization of endocytosed transferrin was altered in overexpressing cells. Under the same conditions, endocytosed α2-macroglobulin was still transported to late endosomes/lysosomes and did not reach the CLIPR-59–positive aggregated membranes (unpublished data). Internalized transferrin was transported to the center of the cells where it extensively, although not perfectly, colocalized with GFP•CLIPR-59 (Fig. 9 A). In agreement with the observed depletion of the punctiforme EEA1-positive structures from the cell periphery, fewer Rh–Tf-positive peripheral endosomes were present. In comparison, no effect of GFP•CLIPR-59–ΔC60 on internalized transferrin localization could be detected (Fig. 9 A).

We then quantified the effect of CLIPR•59 overexpression on transferrin uptake and recycling. Because only cells strongly overexpressing CLIPR-59 showed the aggregation phenotype, we had to devise a way to measure the kinetics of transferrin endocytosis in the population of moderate and strong overexpressing cells, respectively. We used FACS^® analysis to this end, measuring in parallel the green fluorescence of transfected GFP-tagged protein and the red fluorescence of internalized transferrin. Three windows were defined, corresponding to control, low, or high levels of green fluorescence (Fig. 9 B, a), and the kinetics of transferrin endocytosis and recycling were measured (Fig. 9 B, b and c). In agreement with immunofluorescence data, we observed that cells strongly overexpressing CLIPR-59 were still able to internalize transferrin. However, we also observed, in these high overexpressers, a reproducible reduction of transferrin uptake (80% of control) whereas the kinetics of transferrin release were not affected. This could indicate a reduction in the motility of early endosomes or in the pool or endosomes participating in transferrin endocytosis. No such reduction was observed for low level GFP•CLIPR-59 expression. Indeed, both low levels of GFP•CLIPR-59 expression and any level of GFP•CLIPR-59–ΔC60 expression led to a slight increase in transferrin internalization. This observation may indicate that the normal function of CLIPR-59 is to accelerate the rate of internalization, and the CLIPR-59–ΔC60 mutant titrates out a negative regulator of endogenous CLIPR-59.

Immunofluorescence data (Fig. 6) suggested that not only early endosomes, but also the TGN was affected by CLIPR-59 overexpression. Triple staining immunofluorescence analysis of CLIPR-59–overexpressing cells (Fig. 10 A) indicated that cells with accumulated TfR-positive membranes also had reduced TGN46 labeling. In comparison, β1,4-galactosyltransferase (GalT) staining was much less affected. We thus conducted uptake experiments to test whether the TGN46 pathway, from plasma membrane to early endosomes and the TGN, still occurs in cells overexpressing CLIPR-59. We used a well-known and easy to follow marker of this pathway, the B subunit of the Shiga toxin (STxB). STxB is transported from the plasma membrane to the TGN via early/recycling endosomes (Wilcke et al., 2000). GFP•CLIPR-59–transfected cells were incubated with fluorescently labeled STxB for 90 min at 37°C, and the localizations of endocytosed STxB, GFP•CLIPR-59, and the Golgi complex were analyzed (Fig. 10 B). Cells transfected with GFP•CLIPR-59–ΔC60 transported STxB to the Golgi complex normally. In contrast, although mild GFP•CLIPR-59 overexpression did not perturb STxB transport, cells strongly overexpressing GFP•CLIPR-59 showed a reduced ability to transport the endocytosed STxB to the Golgi complex. Interestingly, the colocalization of STxB with CLIPR-59 was clear at the cell periphery but far less at the cell center, suggesting that STxB might be blocked in the prerecycling endosomal compartment.

Because only strong overexpression led to perturbation of STxB transport, we were unable to directly quantify this effect. To further document the interaction between the CLIPR-59–sensitive pathway and the pathways followed by STxB and transferrin, we followed GFP•CLIPR-59 and fluorescent endocytosis markers in living cells (unpublished data). These experiments revealed that although GFP•CLIPR-59 does not colocalize with either endocytosed transferrin or STxB, it is “dynamically juxtaposed” with these markers, moving beside them in the cytoplasm without apparent mixing. This observation suggests that the two target compartments somehow interact during transportation and/or sorting steps.

We have identified a new member of the CLIP-170 family that behaves as a cytoplasmic linker protein involved in the TGN–endosome dynamics. The structure of CLIPR-59 differs notably from that of the other CLIPs from the CLIP-170 family because it has its MTB motifs near the carboxy terminus, has no identifiable coiled-coil region, and possesses three ankyrin-like repeats. The function of the CLIPR-59 ankyrin repeats is unknown, but it has generally been observed that ankyrin repeats form exposed domains involved in protein–protein interactions (Michaely and Bennett, 1992; Gorina and Pavletich, 1996). It is thus attractive to propose that the CLIPR-59 ankyrin domain mediates interaction with other proteins, although it could also be a regulatory domain (see below).

The localization of CLIPR-59 is also unusual for the CLIP-170 family. Both endogenous and recombinant CLIPR-59 are localized on Golgi membranes in vivo, and not on microtubules. In contrast, although association of CLIP-170, CLIP-115, or p150^glued (as part of the dynactin complex) has been documented on various cellular organelles (Pierre et al., 1992; De Zeeuw et al., 1997; Habermann et al., 2001), exogenous expression of these proteins systematically leads to microtubule targeting and not to membrane association. CLIPR-59 is the first member of this linker family for which a membrane localization domain has been identified.

Whereas other ankyrin repeat–containing proteins are targeted to intracellular membranes (for review see Michaely and Bennett, 1993; De Matteis and Morrow, 2000), the CLIPR-59 ankyrin repeats are not involved in targeting to the Golgi complex. We directly demonstrate that the carboxy-terminal 60 amino acids of CLIPR-59 encode the GoLD of the protein. This domain, which is both necessary and sufficient for addressing a cytosolic protein to the Golgi complex (most likely to the trans-Golgi/TGN), does not show any strong conservation with already described GoLDs. In particular, no similarity could be detected to the GRIP domains present in certain golgins (Barr, 1999; Kjer-Nielsen et al., 1999; Munro and Nichols, 1999). The CLIPR-59 GoLD thus represents a new GoLD, and we will now mutagenize it to dissect the molecular basis of its targeting activity.

Also particular to CLIPR-59 is its apparent lack of interaction with microtubules. Neither endogenous nor transfected CLIPR-59 localized to microtubules in tissue culture cells. Moreover, overexpression of CLIPR-59 failed to obviously alter the microtubule network. This failure is in marked contrast to CLIP-170 and many other MTB proteins, which induce microtubule bundling upon overexpression. It should be noted that we did observe cosedimentation of in vitro–translated CLIPR-59 with taxol-stabilized microtubules in vitro (unpublished data). However, this type of experiment is prone to artifacts, and sedimentation was still observed when the CLIP domain was removed.

Finally, deletion experiments demonstrated that both the amino-terminal region and the membrane-targeting domain of CLIPR-59 inhibit microtubule association. The isolated CLIP domain did indeed behave as a MTB domain, eventually bundling microtubules upon overexpression. Careful analysis of immunofluorescence data suggested that the MTB domain of CLIPR-59 differentially recognized a subset of microtubules, but more experiments are necessary to establish both the nature of this subset as well as the function of this discrimination. It would not be unexpected that the MTB domain of CLIPR-59 could recognize a subset of microtubules, because the CLIP domain seems to confer conformation- (or structure-) sensitive tubulin binding to at least some members of the family. For example, CLIP-170 interacts with growing microtubule plus ends probably through a copolymerization mechanism (Diamantopoulos et al., 1999; Perez et al., 1999), a property that may be shared by p150^glued. Another group of CLIPRs also seems to be sensitive to tubulin conformation, as these proteins bind tubulin in a prefolded form (Lewis and Cowan, 1997).

The function of CLIPR-59 is still uncertain. It was proposed by Rickard and Kreis (1996) that specific CLIPs, phylogenetically related to CLIP-170 or not, play a role at the interface between membranous organelles and microtubules. Some described CLIPs are indeed members of the CLIP-170 family: CLIP-170 for endocytic carrier vesicles (Pierre et al., 1992); CLIP-115 for dendritic lamellar bodies (De Zeeuw et al., 1997); and a yet unidentified CLIPR for peroxysomes (Thiemann et al., 2000). But so far, p150^glued, as part of the dynactin complex, is the only CLIP-170–related protein for which clear involvement in membrane dynamics has been established (Burkhardt et al., 1997; Presley et al., 1997).

Although early observations indicated the existence of Golgi CLIPs, the nature of the proteins encoding this activity has largely remained unknown (Karecla and Kreis, 1992; Rickard and Kreis, 1996). GMAP-210 has recently been characterized as a CLIP, mediating interactions between Golgi membranes and stable microtubules (Infante et al., 1999). Hook3 appears to be a CLIP located in the cis-Golgi (Walenta et al., 2001). It is also worth mentioning that the Drosophila CLIP-170 homologue, CLIP-190, colocalizes with Golgi markers during cellularization of the embryo, although no direct involvement of CLIP-190 in Golgi localization has yet been obtained. In this context, CLIPR-59 may be one of the elusive Golgi CLIPs, more particularly involved in trans-Golgi/TGN interaction with microtubules.

CLIPR-59 overexpression strongly perturbs membrane dynamics in the early endosome–TGN pathway, leading to the accumulation of membranes positive for both Rab11 and TfR, hence likely to be recycling endosomes (Ullrich et al., 1996). EEA1-positive structures coaccumulate with overexpressed CLIPR-59, indicating that early endosomes are also affected. Immunofluorescence experiments together with quantification of transferrin uptake and recycling by FACS^® analysis showed that transferrin can be internalized by cells overexpressing CLIPR-59, although less efficiently. It will be interesting to test whether these defects are due to a general inhibition of endosome motility or a failure of aggregated endosomes to participate in transferrin recycling.

Finally, a reduction in TGN46 staining was observed in CLIPR-59–overexpressing cells. CLIPR-59 overexpression also correlated with perturbation of STxB retrograde transport, which also transits through the recycling endosome to the TGN pathway (Mallard et al., 1998; Wilcke et al., 2000). Interestingly, although most of the TfR accumulated at the center of CLIPR-59–overexpressing cells, the STxB appeared to be blocked in peripheral structures partially positive for overexpressed CLIPR-59. However, it is difficult to use overexpression experiments to precisely pinpoint the site of CLIPR-59 action. In addition, we were not able to quantify this effect because cells overexpressing CLIPR-59 represent only a minority of transfected cells. We will now try to reconstitute this effect in semi-intact cells to gather more quantitative data and identify the perturbed stage.

The central accumulation of TfR-positive early/recycling endosomes observed upon CLIPR-59 overexpression may result from either direct or indirect effects on endosome function. Indirect effects could be due to the titration of essential components necessary for function of early endosome–TGN pathways, for example components involved in sorting (receptors), budding (coats), or delivery (fusion machinery and motors). One such component could be the clathrin adaptor PACS-1 (Wan et al., 1998), because the CLIPR-59 E/P domain closely resembles the PACS-1 interaction domain present in furin and HIV-Nef (Wan et al., 1998; Piguet et al., 2000). Moreover, electron microscopy suggested that CLIPR-59 overexpression induced some accumulation of coated membranes. However immunofluorescence analysis showed no major effect on clathrin-coated membranes, or on AP-1, -2, and -3–positive membranes (unpublished data).

According to the proposed model for CLIP function (Rickard and Kreis, 1996; Schroer, 2000), CLIPR-59 may stably link its target membranes to microtubules until they are mature enough to be translocated by molecular motors. It could also help the mature membrane to select a particular subset of microtubules for movement. Because we observed a slight acceleration of endocytosis upon moderate expression of CLIPR-59, this step may represent a necessary checkpoint along the TGN46–STxB pathway. It will be important to determine whether specific kinesins are involved in this pathway, as was shown for the M6PR pathway (Nakagawa et al., 2000). When overexpressed, CLIPR-59 may attach membranes to microtubules too strongly, thus perturbing their motility. It is worth mentioning that CLIPR-59 mutants missing their amino-terminal domain can still efficiently perturb the endosome–TGN membrane dynamics, whereas the additional deletion of the MTB domain prevents this effect. Overexpressed CLIPR-59 would thus behave like the rotavirus protein NSP4 that stably attaches ER-derived membranes to microtubules and inhibits secretory transport (Xu et al., 2000), thus behaving like a nonregulatable CLIP. It is, however, worth noting that the FACS^® experiments also suggested that moderate expression of GFP•CLIPR-59, as well as expression of GFP•CLIPR-59–ΔC60, had a weak stimulatory effect on transferrin endocytosis. This may indicate that low levels of endogenous CLIPR-59 may be an activator of this pathway.

Our domain mapping experiments suggest a model where the membrane-targeting domain of CLIPR-59 is dominant over the MTB domain, and the ankyrin repeat–containing amino-terminal half of CLIPR-59 interferes with microtubule binding. We thus propose that newly synthesized CLIPR-59 is unable to bind to microtubules and first associates with membranes. This membrane association may then allow microtubule binding of CLIPR-59, possibly by displacing the ANK domain or after posttranslational modifications. According to this model, CLIPR-59 would be the first CLIP that binds microtubules only when already localized to its target membrane. In contrast to the other previously characterized CLIPs, its overexpression would thus only affect its target compartment without affecting the microtubule network. This could be of some importance because we observed that CLIPR-59 is strongly expressed in some neurons during development (Bloch-Gallego, E., and C. Sotelo, personal communication; unpublished data), and may thus be used during neuronal maturation to regulate the function of the TGN and recycling endosomes.

In conclusion, we have identified a new CLIP-170–related cytoplasmic linker protein that is involved in the early/recycling endosome–TGN transport pathway. Its unusual characteristics, including its membrane interactions and fine-tuned microtubule interactions, suggest that it plays an important role in membrane–microtubule interactions.

Antibodies against CLIPR-59 were raised in rabbits using a GST fusion protein produced in bacteria. Extensive carboxy-terminal degradation of CLIPR-59 in bacteria resulted in the production of antibodies primarily directed against the amino-terminal domain. Sera were depleted of anti-GST antibodies before being affinity purified on GST–CLIPR-59 resin. Other polyclonal antibodies used in this study were: anti-GFP (Molecular Probes), anti-EEA1 (Santa Cruz Biotechnology, Inc.), anti-galactosyltransferase (provided by E.G. Berger, Institute of Physiology, University of Zurich, Zurich, Switzerland), and anti-Rab11 (Wilcke et al., 2000). Monoclonal antibodies used were anti–GM-130 (Transduction Laboratories), anti–α-tubulin (Sigma-Aldrich), anti–Golgi CTR433 (provided by M. Bornens, Institut Curie, Paris, France), anti-HA, anti-LBPA (6C4; provided by J. Gruenberg, University of Geneva, Geneva, Switzerland), and anti–TfR OKT9, or H68.4 (for immunoelectron microscopy; Zymed Laboratories). Fluorescent secondary antibodies were from Jackson ImmunoResearch Laboratories. Rhodamine- and Alexa633-labeled transferrin were from Molecular Probes.

Culture medium, sodium pyruvate, and glutamine were from GIBCO BRL, restriction enzymes and T4 DNA ligase were from New England Biolabs, Inc., and oligonucleotides were obtained from Sigma-Genosys. DNA was purified using Jetstar columns (Genomed).

Phylogenetic analysis as well as cloning and tagging of CLIPR-59 are described in the online supplemental material (available).

HeLa cells were grown as previously described (Mallard et al., 1998) and transfected using the calcium phosphate precipitate method. 24 or 36 h after transfection, cells were fixed with 3% paraformaldehyde and permeabilized with 0.05% saponin (ICN Biomedicals). Alternatively, when indicated, cells were prepermeabilized with 0.5% Triton X-100 as described by Kreis (1987) and fixed in methanol (4 min, −20°C). Fixed cells were incubated with antibodies for 30 min (except anti-48, which was incubated overnight). Images were then acquired using a Leica Microsystem confocal microscope (TCS4D or SP2) or, in the case of Fig. 3 (a–c), with a cooled CCD camera (CH250; Photometrics) installed on an Axiovert TV135 microscope (ZEISS). Figures were prepared using Adobe Photoshop^® 6.0 running on a Power Macintosh (Apple Computer, Inc.).

HeLa cells were plated on tissue culture dishes 24 h before the experiment to obtain 80% confluency at the time of infection. For transient expression, cells were then infected with the vT7 recombinant vaccinia virus (Fuerst et al., 1986) and cotransfected using DOTAP (Roche) with GFP–CLIPR-59 (subcloned in pSP72 under the T7 promoter). Cells were fixed 4 h after the transfection with 2% paraformaldehyde and 0.125% glutaraldehyde and processed for cryosectioning. The cryosections were made at −120°C using a cryo-ultramicrotome (Leica-Reichert) and retrieved with a 1:1 solution of 2.3 M sucrose and 2% methyl cellulose. Cryosections were then incubated with primary antibodies and revealed with protein A gold (purchased from J.W. Slot, Utrecht Medical School, Utrecht, Netherlands). Labeled cryosections were analyzed with a CM120 electron microscope (Philips Electronic Instrument).

Rh–Tf was obtained from Molecular Probes and Cy3-labeled STxB (STxB–Cy3) was provided by L. Johannes (Institut Curie, Paris, France; Mallard et al., 1998). Transfected cells were incubated with 25 μg/ml of Rh–Tf or 10 μg/ml of STxB–Cy3 in DME for 90 min at 37°C to achieve steady-state labeling of their respective target compartment. Cells were then washed in medium and fixed in 3% paraformaldehyde before being processed for immunofluorescence.

Cells transfected by GFP•CLIPR-59 or GFP•CLIPR-59–ΔC60 for 24 h were detached in PBS-EDTA, pelleted, resuspended in endocytosis medium (DME, 10 mM Hepes, pH 7.4, 0.1% BSA, 5 μg/ml Alexa633-labeled transferrin), and incubated at 37°C. After 60 min of internalization, cells were diluted in cold PBS, pelleted, resuspended in recycling medium (DME, 10% FCS, 10 mM Hepes, pH 7.4, 100 μg/ml unlabeled transferrin), and incubated for another 60 min at 37°C. At the indicated time during the endocytosis and recycling periods, aliquots of incubated cells were taken, diluted five times in PBS and left for 10 min at 4°C in the presence of 100 μg/ml unlabeled transferrin (Sigma-Aldrich). Cells were then pelleted, resuspended in PBS-EDTA, and fixed in 1% paraformaldehyde. FACS^® analysis was then performed using a FACScalibur^® (Becton Dickinson), measuring GFP fluorescence in FL1 and Alexa633 in FL4. The mean Alexa633 fluorescence was then calculated in three separate windows chosen according to relative green fluorescence (control, low, and high). In separate experiments, we checked that cells from the lowest fluorescence window behaved as mock-transfected cells, and could thus be taken as an internal control. At least 7.5 × 10² and up to 10⁴ cells were counted in the high window.

Additional Materials and methods concerning the cloning and phylogenetic analysis of CLIPR-59, as well as additional references related to the monoclonal antibodies used in this study are available online (available).

This article is dedicated to the memory of Thomas Kreis.

We thank E.G. Berger, M. Bornens, and J. Gruenberg for the gift of antibodies, L. Johannes for the gift of fluorescent STxB, A. El Marjou for the purification of the anti-CLIPR-59 antibodies, V. Braun and R. Stalder for their help with some experiments, and P. Benaroch for advice concerning FACS^® analysis. We also thank M. Rojo and V. Lallemand for careful reading of the manuscript.

This work was supported by the Fondation pour la Recherche Médicale (RA00064-01) and the Association pour la Recherche sur le Cancer (ARC-5747) (F. Perez and B. Goud), the American Heart Association (H. Goodson), and the Fond National Suisse (T. Kreis).