Advertisement

A role for the Rab6A′ GTPase in the inactivation of the Mad2‐spindle checkpoint

Stéphanie Miserey‐Lenkei, Anne Couëdel‐Courteille, Elaine Del Nery, Sabine Bardin, Matthieu Piel, Victor Racine, Jean‐Baptiste Sibarita, Franck Perez, Michel Bornens, Bruno Goud

Author Affiliations

  1. Stéphanie Miserey‐Lenkei,
  2. Anne Couëdel‐Courteille,
  3. Elaine Del Nery,
  4. Sabine Bardin,
  5. Matthieu Piel,
  6. Victor Racine,
  7. Jean‐Baptiste Sibarita,
  8. Franck Perez,
  9. Michel Bornens and
  10. Bruno Goud*,1
  1. 1 UMR 144 CNRS/IC, Institut Curie, Paris, France
  1. *Corresponding author. UMR 144 CNRS/Institut Curie, 26 rue d'Ulm, 75248 Paris Cedex 05, France. Tel.: +33 1 4234 6398; Fax: +33 1 4234 6382; E-mail: bruno.goud{at}curie.fr

Abstract

The two isoforms of the Rab6 GTPase, Rab6A and Rab6A′, regulate a retrograde transport route connecting early endosomes and the endoplasmic reticulum via the Golgi complex in interphasic cells. Here we report that when Rab6A′ function is altered cells are unable to progress normally through mitosis. Such cells are blocked in metaphase, despite displaying a normal Golgi fragmentation and with the Mad2‐spindle checkpoint activated. Furthermore, the Rab6 effector p150Glued, a subunit of the dynein/dynactin complex, remains associated with some kinetochores. A similar phenotype was observed when GAPCenA, a GTPase‐activating protein of Rab6, was depleted from cells. Our results suggest that Rab6A′ likely regulates the dynamics of the dynein/dynactin complex at the kinetochores and consequently the inactivation of the Mad2‐spindle checkpoint. Rab6A′, through its interaction with p150Glued and GAPCenA, may thus participate in a pathway involved in the metaphase/anaphase transition.

Introduction

Small GTPases of the Rab family play an essential role in membrane transport and dynamics in interphase (Zerial and McBride, 2001). When cells enter mitosis, transport is arrested and intracellular compartments are disassembled and/or fragmented, ensuring an equal partitioning of organelles between mother and daughter cells. The function and the regulation of Rab GTPases during mitosis is poorly understood. The best characterized are those of Rab4 and Rab11. The phosphorylation of Rab4 by the mitotic kinase p34cdc2 (Bailly et al, 1991) increases the amount of Rab4:GTP in the cytosol, likely resulting in a less efficient recruitment of Rab4 effectors onto endosomal membranes during mitosis and an arrest of the endocytic process (Bailly et al, 1991; Gerez et al, 2000). Recently, Rab11 has been implicated in the regulation of final steps of cytokinesis (Wilson et al, 2005).

Rab6 is associated with Golgi and trans‐Golgi (TGN) membranes in interphase (Goud et al, 1990; Antony et al, 1992) and regulates a retrograde transport route connecting early endosomes to the endoplasmic reticulum (ER) (Martinez et al, 1997; Girod et al, 1999; White et al, 1999; Mallard et al, 2002). Two Rab6 isoforms, termed Rab6A and Rab6A′, are expressed in mammalian cells. They result from alternative splicing of a duplicated exon and differ in only three amino acids (Echard et al, 2000).

Several previously identified Rab6‐interacting proteins suggest a specific role of Rab6 in mitosis. The first one is the Rab6 GTPase‐activating protein (GAP), termed GAPCenA, partially localized to the centrosome in interphase (Cuif et al, 1999). GAP domains of Rab/Ypt GAPs share similarities with GAP domains of yeast checkpoint proteins Bub2p (Saccharomyces cerevisiae) and Cdc16p (Schizosaccharomyces pombe) (Neuwald, 1997). Bub2p in association with Bfa1p and Cdc16p in association with Byr4p display GAP activity towards the Ras‐like small GTPases Tem1p and Spg1p, respectively (Furge et al, 1998; Pereira et al, 2000). Tem1p activates the mitotic exit network (MEN) and Spg1p regulates the septation initiation network (SIN). These two signalling cascades control key events during exit from mitosis and cytokinesis (Bardin and Amon, 2001).

The second is the kinesin‐like protein Rabkinesin‐6 (RK6, also named Rab6KIFL or MKlp2), with which Rab6A directly interacts and whose expression is upregulated at the onset of mitosis (Echard et al, 1998; Hill et al, 2000; Fontijn et al, 2001). It has been shown that RK6/MKlp2 is involved in the localization of Polo‐like kinase 1 (Plk1), Aurora B and Cdc14A at the central spindle (Neef et al, 2003; Gruneberg et al, 2004). These proteins are required for successful cytokinesis.

Finally, the dynein/dynactin complex has been shown to be an effector of Rab6. Rab6 serves as a receptor for the human orthologs of Bicaudal on Golgi membranes. Bicaudal, through an interaction with p50/dynamitin, recruits the dynein/dynactin complex onto transport intermediates moving between Golgi and ER (Matanis et al, 2002). Rab6A and Rab6A′ also interact directly with p150Glued, a subunit of the dynactin complex (Short et al, 2002; B Goud/Hybrigenics, unpublished results). In addition to its role in membrane trafficking and organelle dynamics in interphasic cells, the dynein/dynactin complex is involved in many aspects of mitosis and specifically in checkpoint protein transport, such as Mad2, off kinetochores at the metaphase/anaphase transition (Echeverri et al, 1996; Merdes et al, 2000; Howell et al, 2001; Wojcik et al, 2001; Basto et al, 2004; Siller et al, 2005). The Mad2‐spindle checkpoint senses an absence of tension of mono‐oriented chromosomes and defects in kinetochore attachment (Biggins and Murray, 2001; Musacchio and Hardwick, 2002; Tanaka et al, 2002).

Here, we show that alteration of Rab6A′ function leads to a block in metaphase. Our results indicate that, in addition to its role in interphase, Rab6A′ functions in a pathway involved in the metaphase/anaphase transition implicating GAPCenA and p150Glued.

Results

Alteration of Rab6A′ function leads to a mitotic arrest

To investigate Rab6 function throughout the cell cycle, time‐lapse videomicroscopy was performed on HeLa cells depleted of Rab6A and Rab6A′ isoforms (Rab6A/A′ small interfering RNA (siRNA)) (Figure 1Aa). We calculated (see Supplementary data) that about 40% of cells of the starting population were blocked in mitosis (Figure 1Ab). We next examined the respective contribution of each Rab6 isoform to this phenotype. siRNA that allow selective silencing of each Rab6 isoform were designed (Del Nery et al, 2005). The same sequences (small‐hairpin RNA (shRNA); see Materials and methods) were introduced in pSUPER plasmids and used for transfection. Western blotting indicated a significant depletion of each Rab6 isoform after transfection with the corresponding shRNA (Figure 1Ba). Depletion of Rab6A only slightly impaired progression through mitosis (11% as compared to 7% in control cells) (Figure 1Bb). In contrast, 25% of cells were found blocked in mitosis after treatment with Rab6A′ shRNA (Figure 1B and C). Arrested cells, which exhibited a normal alignment of their chromosomes at the metaphase plate, were unable to exit mitosis and died after 5–7 h (Figure 1C). The specificity of the Rab6A′ effect was confirmed by rescuing the function of Rab6A′ (Supplementary Figure S1), or by using another Rab6A′ siRNA targeted against a different region of Rab6A′ mRNA (Young et al, 2005).

Figure 1.

Alteration of Rab6A′ function leads to a metaphase arrest. (A) (a) At 72 h after transfection with siRNA that target Rab6A and Rab6A′ isoforms (Rab6A/A′) or control siRNA, cell lysates were subjected to Western blotting analysis and probed with an anti‐Rab6 (top) or anti‐β‐tubulin antibody (bottom). (b) Percentage of cells blocked in metaphase after depletion of Rab6A/A′. (B) (a) At 96 h after transfection with Rab6A or Rab6A′ shRNAs or control shRNA, cell lysates were subjected to Western blotting analysis and probed with an anti‐Rab6 (top) or anti‐β‐tubulin antibody (bottom). Since the anti‐Rab6 antibody recognizes both isoforms, the remaining band after depletion of Rab6A or Rab6A′ corresponds to the isoform that has not been depleted. Its intensity is consistent with Rab6A and Rab6A′ being expressed at about the same levels in HeLa cells (Echard et al, 2000). (b) Percentage of cells blocked in metaphase after depletion of Rab6A or Rab6A′. (C) HeLa cells were transfected with control (top) or Rab6A′ shRNA (middle) and imaged for 72 h using time‐lapse phase‐contrast videomicroscopy. Arrowheads point to Rab6A′‐depleted cells arrested in metaphase. Bottom panel: enlarged view of a Rab6A′‐depleted cell arrested in metaphase; note the alignment of chromosomes on the metaphase plate (arrowhead). The numbers correspond to the time in hours after the beginning of the recording. The corresponding movies (movie 1 and movie 2) are presented in Supplementary data. (D) Percentage of cells blocked in metaphase after overexpression of Rab6A′ constructs. In (A, B and D), the number of cells analyzed ranged from 63 to 175. Results are representative of two to four independent experiments and are presented as means±s.e.m.

The role of Rab6A′ during mitosis was further investigated by transient overexpression of wild‐type (wt), dominant‐negative (T27N) and GTPase‐deficient (Q72L) mutants. The phenotype was again analyzed by videomicroscopy. The overexpression of Rab6A′ T27N, but not that of Rab6A′ wt or Rab6A′ Q72L, led to a metaphase block (Figure 1D). About 20% of cells were arrested in mitosis 48 h after transfection. The percentage of transfected cells was estimated on fixed samples to be around 50%.

Altogether, these results indicate that alteration of the function of Rab6A′ leads to an arrest in metaphase.

In metaphase‐blocked cells, Golgi fragmentation appears normal and the active pool of Rab6 is cytosolic

Golgi fragmentation has been shown to be critical for mitotic progression (Sutterlin et al, 2002; Altan‐Bonnet et al, 2003; Hidalgo Carcedo et al, 2004; Preisinger et al, 2005). We therefore investigated whether Golgi fragmentation was altered in metaphasic cells depleted in Rab6A′ or overexpressing Rab6A′ T27N. As shown in Figure 2A, the distribution of two Golgi markers, GM130 (Golgi matrix protein) (a) and CTR433 (medial Golgi) (b), was similar to that observed in control metaphasic cells.

Figure 2.

In metaphase‐arrested cell, Golgi fragmentation appears normal and the active pool of Rab6 is cytosolic. (A) Images of metaphasic cells cotransfected for 72 h with mtGFP (used here as a reporter gene) and either control, Rab6A′ T27N or Rab6A′ shRNA. Cells were costained with DAPI (middle, blue) and GM130 (a) or CTR433 (b). Bar, 10 μm. (B) (a) Images of control metaphasic cells permeabilized or not with saponin prior to fixation and stained with AA2 antibody. For these cells, the mean gray values inside the outlined area are 84 prior and 56 after saponin permeabilization. Bar, 10 μm. (b) Mitotic (lanes 1–6) cells transfected (lanes 3 and 4) or not (lanes 1, 2 and 5–8) with Rab6A/A′ siRNA and nontransfected interphasic cells (lanes 7 and 8) were permeabilized (lanes 2, 4, 6 and 8) or not (lanes 1, 3, 5 and 7) with saponin prior to fixation and then stained with AA2 (recognizing the GTP‐bound conformation of Rab6) (lanes 1–4, 7 and 8) or Rab1 antibodies (lanes 5 and 6). Relative fluorescence intensity (expressed as a ratio for each nontreated staining intensity) was then quantified using Image J software. Number of cells analyzed ranged from 20 to 56. (C) Percentage of cells blocked in metaphase after cotransfection with control, Rab6A′ Q72LΔC, Rab6A′ T27N with or without Rab6A′ Q72LΔC. The number of cells analyzed ranged from 236 to 318. Results are representative of two to three independent experiments and are presented as means±s.e.m.

We then tested whether Rab6 function during mitosis is dependent on its membrane association. To do so, we took advantage of a recombinant antibody (AA2) that recognizes specifically Rab6:GTP (Nizak et al, 2003b). Metaphasic cells were permeabilized with saponin to remove cytosolic pool prior to fixation and the fluorescence intensity obtained with the AA2 antibody was quantified (Figure 2B). The background level of fluorescence intensity for the AA2 antibody was estimated after Rab6 depletion using Rab6A/A′ siRNAs (Figure 2Bb, lanes 3 and 4). After saponin treatment, in nontransfected metaphasic cells, the AA2 staining intensity was decreased by about 50% and lowered to background level (Figure 2Ba and b, lanes 1 and 2). These results suggest that the majority of Rab6:GTP was in fact cytosolic in metaphasic cells. As control, we used Rab1, which is predominantly membrane‐bound during mitosis (Bailly et al, 1991). Rab1 staining did not vary before and after permeabilization with saponin (Figure 2B, lanes 5 and 6). Of note, in interphasic cells, the AA2 staining increased after saponin treatment (Figure 2B, lanes 7 and 8), maybe due to a better accessibility of Golgi‐associated Rab6:GTP.

The AA2 antibody does not discriminate between Rab6A:GTP and Rab6A′:GTP (Nizak et al, 2003b). To directly address whether the active form of Rab6A′ is cytosolic in mitosis, we took advantage of the observation that overexpression of Rab6A′ Q72LΔC (lacking the hypervariable region and the two terminal prenylated cysteines required for membrane association) had no effect on mitosis, as in the case of Rab6A′ Q72L (Figure 1D). As shown in Figure 2C, coexpression of this construct with Rab6A′ T27N led to a suppression of the metaphase block. This experiment also further supports the fact that GTP hydrolysis is likely not required for Rab6A′ function in mitosis.

Altogether, the above results suggest that the function of Rab6A′ in mitosis may not be directly linked to its role in traffic at the Golgi level. In addition, the active pool of Rab6A′ is likely cytosolic in metaphase.

Alteration of Rab6A′ function leads to a Mad2‐dependent metaphase arrest

To further characterize the mitotic arrest obtained after an alteration of Rab6A′ function, cells were stained for cyclin B1 and β‐tubulin. Metaphasic cells expressing Rab6A′ shRNA or overexpressing Rab6A′ T27N mutant were visualized by cotransfection with a plasmid encoding GFP targeted to mitochondria (mtGFP (Rojo et al, 2002)). The efficiency of cotransfection was estimated to be over 95%. Cyclin B1, which is a target of the anaphase‐promoting complex/cyclosome and is degraded at the metaphase/anaphase transition (Peters, 2002), was still detectable concentrated at the spindle poles and at the metaphase plate (Figure 3Aa). In addition, the mitotic spindle, visualized with β‐tubulin staining, appeared normal (Figure 3Ab).

Figure 3.

Alteration of Rab6A′ function leads to a Mad2‐dependent metaphase arrest. (A) Images of metaphasic cells cotransfected for 72 h with mtGFP and control, Rab6A′ T27N or Rab6A’ shRNA. Cells were costained with DAPI (middle, blue) and either (a) cyclin B1 or (b) β‐tubulin. In (b), maximal intensity projection through the z‐axis of deconvolved images stacks is shown. Bar, 10 μm. (B) Maximal intensity projection through the z‐axis of deconvolved image stacks of metaphasic cells 48 h after cotransfection with mtGFP and either control or Rab6A′ shRNA. Cells were costained with a CREST serum (left), anti‐Mad2 antibody (middle) and DAPI (right). Higher magnification (corresponding to the box area of the first and second columns) for details of localization of CREST (green, left) and Mad2 (red, right) staining are shown on the right. Bar, 10 μm. (C) Measurement by the Spot Detection Software of the number of structures present at kinetochores where Mad2 and the CREST serum labelling are colocalized in Rab6A′ shRNA‐transfected cells. Results are presented as means±s.e.m. (D) At 72 h after cotransfection with control, Rab6A′ T27N or Rab6A′ siRNA with or without Mad2 shRNA, cell lysates were subjected to Western blotting analysis and probed with either an anti‐p150Glued antibody (top) as a control of equal loading of material, anti‐Rab6 antibody (middle) or anti‐Mad2 antibody (bottom). (E) Percentage of cells blocked in metaphase after cotransfection of control, Rab6A′ T27N or Rab6A′ shRNA with or without Mad2 shRNA. Number of cells analyzed ranged from 72 to 115. (F) Measurement of the interkinetochore distance (shown in μm) of either control cells treated or not for 10 h with 10−7 M nocodazole or of Rab6A′‐depleted cells displaying or not Mad2‐labelled kinetochores. Number of pairs of kinetochores analyzed ranged from 15 to 28. Results are representative of two to three independent experiments and are presented as means±s.e.m.

The localization of the spindle assembly checkpoint protein Mad2 during metaphase was then investigated in Rab6A′‐depleted cells and in Rab6A′ T27N‐overexpressing cells. In the majority of metaphasic cells depleted of Rab6A′ or overexpressing Rab6A′ T27N (80 and 53%, respectively, as compared to 22% in control cells, number of cells analyzed ranged from 56 to 115), Mad2 was present on at least two kinetochores identified with a CREST serum (Figure 3B). Similar results were observed using another spindle‐checkpoint protein, BubR1 (Sudakin et al, 2001) (Supplementary Figure S2). The presence of Mad2 at kinetochores was confirmed using the Spot Detection Software (Supplementary Figure S3 for details on the method). This measurement also revealed a five‐fold increase in the number of structures present at kinetochores where Mad2 and the CREST serum are colocalized in Rab6A′‐depleted cells as compared to control cells (Figure 3C).

Depletion of Mad2 has been shown to overcome the mitotic spindle checkpoint (Chen et al, 1998; Gorbsky et al, 1998; Canman et al, 2002; Jones et al, 2004). Accordingly, about 50% of cells blocked in metaphase by the microtubule‐depolymerizing agent nocodazole were able to exit mitosis after transfection with Mad2 shRNA (data not shown). Cells were cotransfected either with Rab6A′ shRNA or Rab6A′ T27N and with Mad2 shRNA, resulting in an efficient depletion of Mad2 (Figure 3D, lanes 4 and 6). As shown in Figure 3E, the block in mitosis caused by either Rab6A′ depletion or Rab6A′ T27N overexpression could be reduced to control levels with silencing Mad2.

To further characterize the Mad2‐dependent metaphase arrest, we calculated the distance between pairs of kinetochores in Rab6A′‐depleted cells. The interkinetochore distance of Mad2‐labelled kinetochores was similar to that of unlabelled kinetochores (Figure 3F). In addition, this distance was comparable to that found in control cells, but different from that calculated in nocodazole‐treated cells in which tension between pairs of kinetochores is abolished (Waters et al, 1996).

The above results thus indicate that in cells in which Rab6A′ function is altered the Mad2‐spindle checkpoint is activated. However, metaphase‐arrested cells display a normal spindle and kinetochores are likely under normal tension.

Alteration of Rab6A′ function affects p150Glued localization at kinetochores in metaphase

We next determined which proteins act in the Rab6A′ pathway. Rab6A′ directly interacts with p150Glued, a subunit of the dynein/dynactin complex (Short et al, 2002). The p150Glued Rab6‐binding domain (p150Rab6‐BD) is located between amino acids 738 and 916 (B Goud/Hybrigenics, unpublished results). This domain lies between the two coiled‐coil domains, CC1 (corresponding to the dynein intermediate chain‐binding domain) and CC2 (King et al, 2003). Interestingly, the majority of p150Rab6‐BD‐microinjected cells were arrested in metaphase (Figure 4A), suggesting that Rab6A′/p150Glued interaction is important at the metaphase/anaphase transition. A low proportion of cells also displayed a prometaphase‐like arrest and a cytokinesis defect (Figure 4A). These two phenotypes likely result from an impairment of the activity of the dynein/dynactin complex, since it has been implicated in prometaphase and cytokinesis (Echeverri et al, 1996; Aumais et al, 2003; Zhou et al, 2003).

Figure 4.

Alteration of Rab6A′ function affects p150Glued localization at kinetochores in metaphase. (A) Percentage of cells blocked in various stages of mitosis after microinjection of the p150Rab6‐BD. Number of cells analyzed ranged from 58 to 64. Results are representative of two experiments and are presented as means±s.e.m. (B) Maximal intensity projection through the z‐axis of deconvolved image stacks of metaphasic cells after 48 h cotransfection with mtGFP and control, Rab6A′ T27N or Rab6A′ shRNA. Cells were costained with a CREST serum (left), p150Glued (middle) and DAPI (right). Bar, 10 μm. Higher magnifications (corresponding to the box area of the second column) for details of localization of CREST (green, left) and p150Glued (red, right) staining are shown on the right. (C) Measurement by the Spot Detection Software of the number of structures present at kinetochores where p150Glued and the CREST serum labelling are colocalized in Rab6A′ shRNA‐transfected cells. Results are presented as means±s.e.m. (D) Maximal intensity projection of deconvolved image stacks of metaphasic cells after 48 h cotransfection with mtGFP and Rab6A′ shRNA through the z‐axis. Cells were costained with a CREST serum (left), Mad2 (middle) and p150Glued (right). Bar, 10 μm. Higher magnifications (corresponding to the box area of the first and second columns) for details of localization of CREST (blue, left), Mad2 (middle, green) and p150Glued (red, right) staining are shown on the right.

p150Glued is associated with kinetochores in prophase and prometaphase, from where it dissociates when chromosomes are properly aligned. In metaphasic control cells (Figure 4Ba), the bulk of p150Glued was detectable on the mitotic spindle and at the spindle poles, with only occasional weak kinetochore labelling. In contrast, the majority of metaphase cells either overexpressing Rab6A′ T27N (Figure 4Bb), depleted of Rab6A′ (Figure 4Bc) or microinjected with p150Rab6‐BD (data not shown) showed conspicuous p150Glued still associated with some kinetochores. The presence of p150Glued to kinetochores was confirmed using the Spot Detection Software (Supplementary Figure S3). This measurement showed also an increase in the number of structures present at kinetochores where p150Glued and the CREST serum are colocalized as compared to control cells (Figure 4C).

Evidence exist that the dynein/dynactin complex participates in the inactivation of the Mad2‐spindle checkpoint by removing several checkpoint proteins, including Mad2, from the kinetochores (Howell et al, 2001; Wojcik et al, 2001; Siller et al, 2005). As shown in Figure 4D, the presence of Mad2 on kinetochores was correlated with that of p150Glued in Rab6A′‐depleted cells. Likely, Rab6A′ depletion or Rab6A′ T27N overexpression leads to an alteration of p150Glued dynamics and thus that of the dynein/dynactin complex at the kinetochores. Together with the observation that the interkinetochore distance appears normal in these cells, the above results thus suggest that cells are blocked in metaphase as a consequence of an impairment of the inactivation of the Mad2‐spindle checkpoint.

Role of GAPCenA in the Rab6A′ pathway

To further explore the Rab6A′ pathway in mitosis, we investigated the role of GAPCenA. GAPCenA has been identified as a GAP for Rab6A, but it interacts in vitro with both Rab6A and Rab6A′ (data not shown). We have previously shown that a pool of GAPCenA is associated with centrosomes in interphasic cells (Cuif et al, 1999). The centrosomal localization of endogenous GAPCenA is maintained throughout mitosis (Figure 5Aa–d). During cytokinesis, a pool of GAPCenA is visible adjacent to the midbody, often being more abundant on one side of the bridge (Figure 5Ae, arrowhead). GFP‐tagged GAPCenA has also recently been shown to localize to the midbody ring during cytokinesis (Gromley et al, 2005). GAPCenA expression was silenced using siRNAs. By Western blotting, we verified that the expression of endogenous GAPCenA was decreased after 96 h treatment with two different siRNAs, siRNA‐GAPCenA‐A and siRNA‐GAPCenA‐B (Figure 5B). By videomicroscopy, we observed that both GAPCenA siRNAs resulted in a block in metaphase (Figure 5C). A similar phenotype was observed after microinjection of affinity‐purified antibodies directed against the C‐terminus (Cter) or the Rab6‐binding domain of GAPCenA (GAPCenARab6‐BD). In this case, about 50% of microinjected cells were arrested in metaphase (Figure 5D).

Figure 5.

Role of GAPCenA in the Rab6A′ pathway. (A) Cells were labelled with anti‐GAPCenA antibody (bottom, red), antiacetylated tubulin (green, top) and DAPI (blue, top). (B) At 96 h after transfection with control or two different GAPCenA siRNAs (A or B), cell lysates were subjected to Western blotting analysis and probed with the anti‐GAPCenA ‘c74’ (top) or antiβ‐tubulin antibodies (bottom). (C) Percentage of cells blocked in metaphase after transfection with control and GAPCenA siRNAs (A or B). (D) Percentage of cells blocked in metaphase after microinjection of antibodies directed against either the Cter domain or the Rab6‐binding domain (Rab6BD) of GAPCenA. (E) Percentage of cells blocked in metaphase after cotransfection of GAPCenA siRNA or the Cter domain of GAPCenA with or without Mad2 shRNA. (F) Maximal intensity projection through the z‐axis of deconvolved images stacks of metaphasic cells transfected with GAPCenA siRNA. Cells were costained with a CREST serum (left), p150Glued (middle) and DAPI (right). Bar, 10 μm. Higher magnifications (corresponding to the box area of the second column) for details of localization of CREST (green, left) and p150Glued (red, right) staining are shown on the right. (G) Measurement of the number of structures present at kinetochores where p150Glued and the CREST serum labelling are colocalized in GAPCenA siRNA transfected cells by the Spot Detection Software. In (C–E), the number of cells analyzed ranged from 36 to 91. Results are representative of two to four independent experiments and are presented as means±s.e.m. (H) At 48 h after comicroinjection with either control, Rab6A siRNA with or without GAPCenA siRNA, cells were stained with AA2 (recognizing the GTP‐bound conformation of Rab6). We verified by immunofluorescence that in cells comicroinjected with Rab6 and GAPCenA siRNAs, Rab6 and GAPCenA were efficiently depleted (data not shown). The relative fluorescence intensity (expressed as a ratio of nontreated cells) of metaphase cells was then quantified using Image J software. The number of cells analyzed ranged from 20 to 56. Results are representative of two to three independent experiments and are presented as means±s.e.m.

The block in metaphase caused by GAPCenA depletion could be overcome with silencing Mad2 (Figure 5E). In addition, overexpression of the Cter domain of GAPCenA also led to a Mad2‐dependent metaphase block (Figure 5E). To further explore the contribution of GAPCenA function in the Rab6A′ pathway, we looked at the localization of p150Glued in GAPCenA‐depleted cells. In the majority of metaphase‐arrested cells, p150Glued was still associated with some kinetochores (Figure 5F). Consistently, the Spot Detection Software showed an increase in the number of structures present at kinetochores, where p150Glued and the CREST serum are colocalized as compared to control cells (Figure 5G). Depletion of GAPCenA thus leads to a metaphase arrest similar to the one observed after alteration of Rab6A′ function, involving mislocalization of p150Glued.

We then investigated whether the depletion of GAPCenA caused a change in the behaviour of Rab6A′. To test this, GAPCenA siRNA were microinjected into cells depleted of Rab6A. If GAPCenA was able to stimulate GTP hydrolysis on Rab6A′, one would expect that its downregulation leads to an increase in the amount of Rab6A′:GTP. However, the amount of Rab6A′:GTP measured with the AA2 antibody was only slightly increased in cells depleted of both GAPCenA and Rab6A (Figure 5H). This observation suggests that GAPCenA might not act as a GAP for Rab6A′ during mitosis. It is also consistent with the other experiments indicating that GTP hydrolysis is not required for Rab6A′ function (Figure 1D).

Altogether, the above results indicate that GAPCenA acts in the Rab6A′ pathway, likely as an effector protein.

Discussion

Here, we report that when Rab6A′ function is altered the Mad2‐spindle checkpoint is activated and cells are blocked in metaphase. We show that Rab6A′ functions in a pathway involved at the metaphase/anaphase transition implicating GAPCenA and p150Glued.

An important question is whether Rab6A′ functions in interphase and in mitosis are linked. At the onset of mitosis, the Golgi complex undergoes extensive fragmentation that results in a dispersal of membranes into small vesicles. Whether such vesicles merge with the ER or remain distinct Golgi fragments remains controversial (Zaal et al, 1999; Jesch et al, 2001; Jokitalo et al, 2001; Seemann et al, 2002; Shorter and Warren, 2002; Nizak et al, 2004). There is growing evidence that interfering with Golgi dynamics impairs mitosis. For instance, microinjection of antibodies against the Golgi matrix protein GRASP‐65 (Sutterlin et al, 2002) or the inhibition of the membrane‐fissioning protein CtBT3/BARS (Hidalgo Carcedo et al, 2004) prevents Golgi‐membrane dispersal and block cells mainly in G2. The persistence of Golgi structures in mitotic cells as a result of the inhibition of Arf1 GTPase inactivation was also shown to induce defects in chromosome segregation and cytokinesis (Altan‐Bonnet et al, 2003). One current interpretation of these data is that Golgi membranes serve as scaffold for various mitotic proteins. If the scaffold is mislocalized, mitotic events are disrupted. In our experiments where Rab6A′ function was impaired (Figure 2), we found no obvious differences in the distribution of various Golgi resident and matrix proteins in metaphase‐blocked as compared to control cells. This suggests that the metaphase arrest induced by the alteration of Rab6A′ function is not caused by an alteration of Golgi fragmentation.

A surprising result is that the cytosolic pool of Rab6 appears to be in its GTP‐bound conformation in mitosis. This was shown by the use of an antibody only recognizing the GTP‐bound conformation of Rab6. The observation that the overexpression of a cytosolic Rab6A′ GTPase mutant was able to suppress the Rab6A′‐dependent block in metaphase also suggests that the active pool of Rab6, and specifically that of Rab6A′, is cytosolic in mitosis. The currently accepted model of Rab function indicates that active forms of Rabs are membrane‐bound (Zerial and McBride, 2001; Pfeffer and Aivazian, 2004). However, Rab4 is also found without membrane association and in its GTP‐bound form during mitosis (Gerez et al, 2000). Rab4:GTP is maintained in the cytosol through an association with the peptidyl–prolyl isomerase Pin1 (Gerez et al, 2000). It remains to be established whether Rab6 also interacts with Pin1 or to another protein that fulfills a similar function.

Why does the depletion of Rab6A′ or the overexpression of Rab6A′ T27N lead to a block in mitosis? In cells arrested in metaphase, we observed that a fraction of the dynactin p150Glued subunit is still associated with kinetochores. p150Glued binds to the dynein intermediate chain and plays an important role in the association of the dynactin complex with dynein (Karki and Holzbaur, 1995; Vaughan and Vallee, 1995; King et al, 2003). The dynein/dynactin complex is involved in many aspects of mitosis from prometaphase to cytokinesis (Echeverri et al, 1996; Merdes et al, 2000; Howell et al, 2001; Wojcik et al, 2001). In addition, a direct role of the dynein/dynactin complex in the transport of several kinetochore proteins to spindle poles and inactivation of the spindle checkpoint has been found (Howell et al, 2001; Wojcik et al, 2001; Siller et al, 2005). Thus, a plausible interpretation of our results is that p150Glued is the downstream effector of Rab6A′ and that Rab6A′/p150Glued interaction is important for the dynamics of the dynein/dynactin complex at the kinetochores. It was recently shown that Drosophila Lis1/dynactin mutants display at least two phenotypes, a reduced interkinetochore distance and a failure to transport checkpoint proteins off kinetochores (Siller et al, 2005). We did not observe an obvious reduction of the interkinetochore distance in Rab6A′‐depleted cells arrested in metaphase, suggesting that the main cause of the metaphase block observed after alteration of Rab6A′ function is due to an impairment of the inactivation of the Mad2‐spindle checkpoint rather than in a defect in tension generation in metaphase. However, we do not rule out the possibility that enough p150Glued remains active in Rab6A′‐depleted cells to warrant a normal tension between kinetochores. Future experiments aimed to further characterize the interaction between Rab6A′ and p150Glued will help to clarify this issue. As a colocalization between Rab6:GTP and Mad2 can be detected in the mitotic cytosol (data not shown), Rab6A′ may also play a role in other mechanisms required to inactivate the Mad2‐spindle checkpoint that involve cytosolic checkpoints proteins away from kinetochores (Meraldi et al, 2004; De Antoni et al, 2005).

Another protein involved in the Rab6A′ pathway is GAPCenA. This is supported by our observations that depletion of GAPCenA leads to a phenotype similar to the one observed after alteration of Rab6A′ function, including mislocalization of p150Glued. GAPCenA has been described as a Rab6 GAP based on its in vitro activity (Cuif et al, 1999). However, the overexpression of the GTPase‐deficient mutant Rab6A′ Q72L has no effect on mitosis, suggesting that GTP hydrolysis may not be required for Rab6A′ function during mitosis. In addition, depletion of GAPCenA only slightly increased the amount of Rab6A′:GTP in metaphase. One possibility to reconcile this apparent contradiction is that GAPCenA does not act as a GAP for Rab6A′ in mitosis, but rather as an effector protein. This possibility is not without precedent, since RN‐tre, a Rab5 GAP, can also function as an effector protein for F‐actin and actinin‐4 (Lanzetti et al, 2004). GAPCenA possesses a phospho‐tyrosine‐binding domain at its N‐terminus, which suggests that it may interact with other regulatory proteins involved in Rab6A′ function.

In summary, our work uncovered an unexpected role for a Rab GTPase during mitosis. Growing evidence indicates that membrane traffic events are very important for completion of cytokinesis (Guertin et al, 2002; Echard et al, 2004; Schweitzer and D'Souza‐Schorey, 2004; Gromley et al, 2005). Rab proteins, such as Rab11, likely play an important role in this process (Wilson et al, 2005). That alteration of Rab6 function arrests cells in metaphase was more surprising. Interestingly, the other Rab6 isoform, Rab6A, is also required for the metaphase/anaphase transition, although in interaction with other effectors, including RK6/MKlp2 (Miserey‐Lenkei et al, unpublished observations). It will be interesting to address in future studies whether other members of the Rab family play a dual role in interphase and in mitosis.

Materials and methods

RNA interference

The pSUPER vector (Brummelkamp et al, 2002) was used to express Rab6A, Rab6A′ and control shRNAs. siRNAs were used for silencing GAPCenA. The siRNA sequences for targeting human Rab6 or both isoforms separately are described in Del Nery et al (2005) and were inserted in the pSUPER vector (corresponding to shRNA). For targeting human GAPCenA, two siRNA sequences were used: (1) 758–781 (GAPCenA siRNA‐A) and (2) 961–984 (GAPCenA siRNA‐B). Each sequence is expressed relatively to the first nucleotide of the coding sequence. The control vector was made using the sequence of a single‐stranded siRNA against giantin (Nizak et al, 2003a). Mad2 shRNA cloned in the pSUPER vector was a gift from Dr Andrea Musacchio (Milan, Italy). Oligos were obtained from Proligo (Paris, France) for GAPCenA siRNA‐B and from Dharmacon for GAPCenA siRNA‐A. A plasmid encoding GFP with a targeting signal to mitochondria (mtGFP) cloned in the pCB6 vector (Rojo et al, 2002) was used as a reporter gene.

GAPCenA and p150Glued Rab6‐BD construct

The GST‐tagged Cter domain of GAPCenA was obtained by inserting the 759–1069 PCR fragment of the pGFPN1‐GAPCenA vector (Cuif et al, 1999) into the EcoRI/XhoI sites of the pGEX‐4T‐3 vector (Pharmacia Biotech). Myc‐tagged Cter domain of GAPCenA was obtained by inserting the same 759–1069 PCR fragment into EcoRI/XhoI sites of the pcDNA3‐myc vector. This vector was derived from pcDNA3 (Invitrogen) by inserting the myc‐tag of pGem‐4Z‐myc vector described in Cuif et al (1999). The His‐tagged p150‐Rab6BD was obtained by inserting the 2208–2748 PCR fragment of the pP6‐A6 clone (obtained from Hybrigenics) into the NdeI/XhoI sites of the pET‐15b vector (Novagen).

Cell culture, transfection and microinjection

HeLa cells were grown in DMEM medium (Gibco BRL) supplemented with 10% fetal bovine serum, 100 U/ml penicillin/streptomycin and 2 mM glutamine. For expression of Rab6A′ T27N construct and Rab6A′ shRNA, HeLa cells grown on 24‐well plates were transfected using Lipofectamine (Invitrogen) following the manufacturer's instructions. For silencing Rab6A and Rab6A′ or both isoforms, HeLa cells were transfected with the corresponding siRNA using Oligofectamine (Invitrogen) following the manufacturer's instructions. For silencing GAPCenA, HeLa cells were transfected twice at 24 h interval using Oligofectamine. For cotransfection experiments using Mad2 shRNA and single transfection using Rab6A and Rab6A′ shRNA or Rab6A′ T27N and Rab6A′ Q72LΔC, HeLa cells were transfected by electroporation with a Gene Pulser II apparatus (Bio‐Rad). For cotransfection experiments with Mad2 and Rab6A′ shRNAs or GAPCenA siRNA, cells were transfected with Lipofectamine. Microinjection of HeLa cells grown on 35‐mm glass dishes (Iwaki) or coverslips was performed using an automatic microinjector (Eppendorf) or a manual microinjector (TransferMan NK2 Eppendorf). For siRNA microinjection, 5 μM of the corresponding siRNA was diluted in the microinjection buffer (5 mM phosphate sodium buffer, pH 7.2, 100 mM KCl). For each experiment, the efficiency of transfection was estimated in parallel by immunofluorescence on fixed cells to be around 80% using Oligofectamine, 50% using Lipofectamine and 50–80% by electroporation.

Anti‐GAPCenA antibodies

Rabbit antiserum was raised against purified His‐tagged full length GAPCenA expressed in Escherichia coli. Three different regions can be delineated in GAPCenA: an N‐terminal domain, a central GAP domain and a C‐terminal (Cter) domain containing the Rab6‐binding domain (Rab6‐BD) (Cuif et al, 1999). Antibodies present in the serum directed against the Cter domain of GAPCenA were affinity‐purified using GST–Cter fusion protein transferred onto nitrocellulose sheet as described previously (Goud et al, 1990). We confirmed by Western blotting that affinity‐purified antibodies were specific for the Cter domain of GAPCenA (data not shown). The antibody directed against the Rab6‐binding domain of GAPCenA (‘c74’) has been described previously (Cuif et al, 1999).

Immunofluorescence microscopy and Western blotting

HeLa cells grown on coverslips were fixed in methanol at −20°C for 4 min or in 3% paraformaldehyde (PFA) for 15 min at RT. For saponin‐prepermeabilization experiments, cells were treated for 5 min at RT in a buffer containing 0.05% saponin, 80 mM PIPES, pH 6.8, 5 mM EGTA and 1 mM MgCl2 prior to fixation with PFA. Cells were then processed for immunofluorescence as described previously (Martinez et al, 1994, 1997). The following primary antibodies were used: mouse antiacetylated tubulin (1:200; Sigma), mouse CTR433 (1:10; Jasmin et al, 1989), mouse GM130 (1:1000; Transduction Laboratories), mouse anti‐β‐tubulin (1:1000; Sigma), rabbit anti‐GAPCenA (1:500, c74; Cuif et al, 1999), rabbit anti‐Rab1 (1:20; Bailly et al, 1991), purified AA2 (1:200; Nizak et al, 2003b) coincubated with mouse anti‐His (1:1000; Sigma), rabbit anti‐Mad2 (1:2000; a gift from Dr ED Salmon, Chapel Hill, USA), human CREST serum (1:1000; a gift from Dr J‐C Courvalin, Institut J Monod, Paris, France), mouse anti‐p150Glued (1:500; Transduction Laboratories) and mouse anti‐BubR1 (1:400; Santa Cruz). Secondary antibodies were from Jackson and Molecular Probes. DAPI was purchased from Sigma.

For Western blotting experiments, cells were processed as in Miserey‐Lenkei et al (2001). The following primary antibodies were used: rabbit anti‐GAPCenA (1:500, c74; Cuif et al, 1999), mouse anti‐β‐tubulin (1:1000; Sigma), rabbit anti‐Rab6 (1:1000; Santa Cruz), mouse anti‐Mad2 (1:25; a gift from Dr Andrea Musacchio, Milan, Italy) and mouse anti‐p150Glued (1:2000; Transduction Laboratories). Secondary HRP‐coupled antibodies were from Jackson.

Live‐cell imaging

Transfected cells were plated on 35‐mm glass dishes (Iwaki) and put in an open chamber (Life imaging) equilibrated in 5% CO2 and maintained at 37°C. Time‐lapse sequences were recorded at 10‐min intervals for 72 h on a Leica DMIRBE microscope using an × 20 objective controlled by the Metamorph software (Universal Imaging). This microscope was equipped with a cooled CCD camera (Micro Max 5 Mhz; Ropper Scientific).

Supplementary data

Supplementary data are available at The EMBO Journal Online.

Supplementary Information

Supplementary data [emboj7600929-sup-0001.pdf]

Supplementary Figure 1 [emboj7600929-sup-0002.mov]

Supplementary Figure 2 [emboj7600929-sup-0003.mov]

Acknowledgements

We thank A Echard, Y Bellaïche, V Doye, J‐P Kleman, Y Barral, O Vielemeyer, J Young and M Morgan for helpful comments and critical reading of the manuscript; V Fraisier for videomicroscopy equipment; F Jollivet, S Moutel and A Boulet for methodological assistance; ED Salmon, A Musacchio, M Rojo and F Barr for their generous gift of antibodies, shRNAs and mtGFP constructs. This work was supported by the Centre National de la Recherche Scientifique (CNRS), the Institut Curie and the Association de la Recherche contre le Cancer (ARC #5294). SM‐L was a recipient of a fellowship from the Association de la Recherche contre le Cancer and from the European Union (Project number LSHM‐CT‐2004‐503228, Project Acronym, SIGNALLING & TRAFFIC, http://www.signallingtraffic.com).

References