The Centrosomal Protein RAS Association Domain Family Protein 1A (RASSF1A)-binding Protein 1 Regulates Mitotic Progression by Recruiting RASSF1A to Spindle Poles*

The protein RAS association domain family protein 1A (RASSF1A), which is encoded by a gene that is frequently silenced in many types of sporadic tumor, functions in mitosis as a regulator of the anaphase-promot-ing complex (APC). With the use of a yeast two-hybrid screen, we identified a human protein, previously designated C19ORF5, that interacts with RASSF1A. This protein, here redesignated RASSF1A-binding protein 1 (RABP1), contains two microtubule-associated protein domains, and its association with RASSF1A was confirmed in mammalian cells by immunoprecipitation and immunofluorescence analyses. RABP1 was found to be localized to the centrosome throughout the cell cycle in a manner dependent on its microtubule-associated protein domains. Ectopic expression of RABP1 induced both stabilization of mitotic cyclins and mitotic arrest at prometaphase in a RASSF1A-dependent manner. It also increased the extent of association between RASSF1A and Cdc20. Conversely depletion of RABP1 by RNA interference prevented both the localization of RASSF1A to the spindle poles as well as its binding to Cdc20, resulting in premature destruction of mitotic cyclins and acceleration of mitotic progression. These findings indicate that RABP1 is required for the recruitment of RASSF1A

The RAS association domain family protein 1 (RASSF1) 1 gene is implicated as a tumor suppressor by the observations that it is frequently silenced in a wide range of tumors as a result of hypermethylation of a CpG island within its promoter (1)(2)(3)(4)(5) and that restoration of RASSF1A expression inhibits tumorigenic cell growth both in vitro and in vivo (1,3,4,6). Two major isoforms of RASSF1, A and C, are produced from the human RASSF1 gene on chromosome 3p21.3 (1,3). RASSF1A contains a C1 (cysteine-rich, diacylglycerol binding) domain in its NH 2 -terminal region and an RAS association domain in its COOH-terminal region, whereas RASSF1C also contains the RAS association domain but possesses a distinct 50-amino acid sequence in place of the C1 domain. RASSF1 proteins are thought to participate in regulation of the cell cycle and cell death given that their ectopic expression promotes cell cycle arrest and apoptosis (7)(8)(9). Coexpression of RASSF1C thus enhances the induction of apoptosis by activated RAS, and association of RASSF1A with mammalian STE20-like kinase 1 (MST1) appears to contribute to apoptotic signaling (7,10). RASSF1A blocks cell cycle progression at the G 1 -S transition by inhibiting the accumulation of cyclin D and promotes arrest at G 1 by binding to p120 E4F (11,12). RASSF1A also associates with and stabilizes microtubules, suggesting that it plays a role in microtubule dynamics (13,14).
The anaphase-promoting complex (APC) is a ubiquitin ligase that specifically targets mitotic cyclins A and B for degradation and thereby allows mitotic progression (15)(16)(17). Cdc20 and Cdh1 activate the APC in mitosis and in G 1 phase, respectively (18). Several proteins, including Mad2, BubR1, and Emi1, inhibit APC-Cdc20 activity to prevent destruction of mitotic cyclins in mitosis (19 -24). We have recently shown that RASSF1A is required for normal mitotic progression as a result of its inhibition of APC activity during mitosis (25,26). RASSF1A thus binds to Cdc20 that is localized on the mitotic spindle and at the spindle poles during prometaphase and thereby inhibits its ability to promote the destruction of mitotic cyclins and subsequent mitotic progression. This role of RASSF1A in APC regulation is independent both of the Mad2-dependent spindle checkpoint machinery and of the Emi1-Cdc20 inhibitory complex. Depletion of RASSF1A by RNA interference (RNAi) accelerates mitotic progression as a result of premature APC activation.
We have now examined further the function of RASSF1A in mitotic regulation and identified, with the use of a yeast twohybrid screen, a centrosomal protein, designated RASSF1A binding protein 1 (RABP1), that interacts with RASSF1A. We further demonstrated that RABP1 plays an important role in the recruitment of RASSF1A to spindle poles and in the subsequent RASSF1A-mediated regulation of the APC in mitosis.
Yeast Two-hybrid Screen-Yeast two-hybrid screening was performed as described previously (27). In brief, yeast strain AH109 transformed with pGBKT7-RASSF1A was mated with strain Y187 that had been transformed with a Matchmaker human testis cDNA library (Clontech). A total of 5 ϫ 10 6 yeast colonies was screened of which 30 survived in selective medium and were positive for ␤-galactosidase expression. Clones were rescued and reintroduced into yeast to confirm positive interactions. For deletion analysis, double transformants of HF7c cells were grown in selective medium, and ␤-galactosidase activity was determined with o-nitrophenyl ␤-D-galactoside as substrate.
Cell Culture and Transfection-293T, NCI-H23, and HeLa cells were cultured in Dulbecco's modified Eagle's medium supplemented with 10% fetal bovine serum. For thymidine block and release experiments, cells were incubated for 16 h with 2 mM thymidine (Sigma), washed, and then incubated in medium without thymidine. Unless indicated otherwise, 293T cells were transiently transfected with plasmids by CaCl 2 precipitation, and NCI-H23 and HeLa cells were transfected with the use of the Effectene transfection reagent (Qiagen).
Antibodies-Polyclonal antibodies to human RABP1 were generated in rabbits by injection with a recombinant glutathione S-transferase fusion protein containing amino acids 528 -828 of RABP1. Mouse and rabbit polyclonal antibodies to RASSF1A were generated as described previously (25). Other antibodies used in the present study included mouse monoclonal antibodies to HA (Roche Applied Science); to FLAG, to ␤-tubulin, or to ␥-tubulin (Sigma); or to cyclin B, to cyclin D, or to Cdc20 (Santa Cruz Biotechnology). Rabbit polyclonal antibodies to HA, to Chk2, to GFP, to Cdc20, or to cyclin A were from Santa Cruz Biotechnology.
Immunofluorescence Analysis-Cells grown on chamber slides (LabTakII, Nunc) were washed twice with phosphate-buffered saline, fixed in ice-cold methanol, washed again with phosphate-buffered saline, and then incubated for 30 min at room temperature with 5% normal goat serum (Sigma) in phosphate-buffered saline. Cells were then exposed consecutively to primary antibodies and to rhodamine-or fluorescein isothiocyanate-conjugated goat antibodies to rabbit or mouse IgG (Santa Cruz Biotechnology). Slides were mounted in medium containing 4Ј,6-diamidino-2-phenylindole (DAPI) and imaged either with an Olympus microscope equipped with a Hamamatsu Orca charge-coupled device camera or with a confocal laser scanning microscope (Bio-Rad CLSM 1024). Data were processed with Adobe Photoshop 5.5 software.
Cell Cycle Analysis-Cells were cotransfected with pcDNA-CD4 (25) and Flag-RABP1 vectors at a mass ratio of 1:10, collected at various times after transfection, and incubated for 1 h at 4°C with fluorescein isothiocyanate-conjugated mouse antibodies to CD4 (Calbiochem). They were then washed three times with phosphate-buffered saline, fixed in 70% ethanol, stained with propidium iodide (25 g/ml) (Sigma), and incubated for 30 min at 37°C with RNase A (20 g/ml) (Roche Applied Science). The DNA content of the cells was then evaluated by flow cytometry with a FACScan instrument (BD Biosciences). Linear red fluorescence (FL2) in green (CD4-expressing) cells was analyzed. The mitotic index and mitotic stage of cells expressing Flag-RABP1 were determined by immunostaining with antibodies to ␤-tubulin and to FLAG.
Immunoprecipitation-Cells were harvested 48 h after transfection and lysed in a solution containing 50 mM Tris-HCl (pH 7.5), 150 mM NaCl, 1 mM EDTA, 1 mM MgCl 2 , and 0.5% Triton X-100. Immunoprecipitation was performed as described previously (25) with antibodies coupled either to Sepharose (Amersham Biosciences) or to protein Gagarose (Oncogene). The immunoprecipitates were washed four times with lysis buffer and then subjected to immunoblot analysis.
Centrosome Isolation-A centrosome fraction was prepared as described previously (28). HeLa cells in the exponential phase of growth were incubated for 1 h at 37°C with cytochalasin D (1 g/ml) and 0.2 M nocodazole before lysis in a solution containing 1 mM Hepes-NaOH (pH 7.2), 0.5% Nonidet P-40, 0.5 mM MgCl 2 , 0.1% 2-mercaptoethanol, protease inhibitors, and phosphatase inhibitors. Swollen nuclei and chromatin aggregates were removed by centrifugation of the lysate at 2500 ϫ g for 10 min. The Hepes concentration of the supernatant was adjusted to 10 mM, DNase I (Promega) was added to a concentration of 2 units/ml, and the mixture was then incubated for 30 min on ice before underlaying with a solution containing 60% (w/w) sucrose, 10 mM Pipes-NaOH (pH 7.2), 0.1% Triton X-100, and 0.1% 2-mercaptoethanol. Centrosomes were sedimented into the sucrose cushion by centrifugation at 10,000 ϫ g for 30 min and then subjected to immunoblot analysis.

Identification of RABP1 as a Protein That Binds
RASSF1A-To explore the mechanism by which RASSF1A regulates the cell cycle, we performed a yeast two-hybrid screen to FIG. 1. Interaction of RASSF1A with RABP1. A, delineation of the region of RABP1 responsible for binding to RASSF1A. The indicated RABP1 deletion mutants fused to the DNA binding domain of Gal4 were examined for their ability to bind to full-length RASSF1A fused to the transactivation domain of Gal4. Pairs of plasmids were introduced into HF7c yeast cells, and ␤-galactosidase activity was quantitated with the o-nitrophenyl ␤-D-galactoside assay. B, delineation of the region of RASSF1A responsible for binding to RABP1 in yeast cells. The indicated RASSF1A deletion mutants were examined for their ability to bind to full-length RABP1 as in A. C, interaction of RASSF1A with RABP1 in mammalian cells. Lysates of 293T cells expressing HA-RASSF1A or Flag-RABP1, as indicated, were subjected to immunoprecipitation (IP) with antibodies to FLAG, and the resulting precipitates as well as the cell lysates were subjected to immunoblot analysis (IB) with antibodies to HA or to FLAG. The asterisk indicates Ig heavy chain. D, determination of the region of RASSF1A responsible for binding to RABP1 in mammalian cells. Lysates of 293T cells expressing Flag-RABP1 and GFP fusion proteins of RASSF1A, RASSF1A-NT, or RASSF1A-CT were subjected to immunoprecipitation with antibodies to FLAG, and the resulting precipitates as well as the cell lysates were subjected to immunoblot analysis with antibodies to GFP and to FLAG. search for proteins that interact with RASSF1A. With fulllength RASSF1A as the bait, we screened a human testis cDNA library and identified 30 positive clones, seven of which encoded COOH-terminal fragments of the hypothetical protein C19ORF5. The full-length protein, which we now designate RABP1, has a calculated molecular size of ϳ130 kDa and manifests substantial sequence similarity to microtubule-associated protein (MAP) 1A and MAP1B, the MAP domains of which are thought to regulate microtubule dynamics (29).
To define the region of RABP1 responsible for interaction with RASSF1A, we constructed four deletion mutants of RABP1 fused with the DNA binding domain of Gal4 and tested them for their ability to bind to RASSF1A in the yeast twohybrid assay. The regions containing either of the two MAP domains (amino acids 209 -500 or 798 -1059) of RABP1 were sufficient for interaction with RASSF1A (Fig. 1A). Similarly a COOH-terminal region of RASSF1A containing the RAS association domain (amino acids 120 -340) was sufficient for interaction with RABP1 (Fig. 1B). To examine the interaction between RASSF1A and RABP1 in mammalian cells, we transiently cotransfected 293T cells with expression vectors for HA-tagged RASSF1A and FLAG-tagged RABP1 and subjected Flag-RABP1 or transfected with empty vector were subjected to immunoblot analysis with preimmune serum, antiserum to RABP1, or antiserum to RABP1 that had been pretreated with antigen (left, middle, and right pairs of lanes, respectively). Right panel, association of endogenous RABP1 with RASSF1A in HeLa cells. Cell lysates as well as immunoprecipitates prepared from HeLa cells with antibodies to RABP1 or with control rabbit IgG were subjected to immunoblot analysis with antibodies to RASSF1 or to RABP1. B, localization of RABP1 to the centrosome. Left panel, HeLa cells were stained with DAPI (blue), antibodies to RABP1 (red), and antibodies to ␥-tubulin (green). The merged images show localization of RABP1 to centrosomes in both interphase and mitotic cells. Right panel, total HeLa cell lysate (T) and the prepared centrosome fraction (C) were subjected to immunoblot analysis of the indicated proteins. C, colocalization of RASSF1A and RABP1 at the centrosome in interphase and mitotic cells. HeLa cells were stained with DAPI (blue), antibodies to RASSF1A (red), and antibodies to RABP1 (green). The lower merged image shows colocalization of RASSF1A and RABP1 at spindle poles during mitosis. D, delineation of the region of RABP1 responsible for localization to the centrosome. HeLa cells expressing the indicated FLAG-tagged RABP1 mutants were stained with DAPI (blue), antibodies to FLAG (red), and antibodies to ␥-tubulin (green). All scale bars, 10 m. IB, immunoblot; IP, immunoprecipitation.
lysates of the transfected cells to co-immunoprecipitation and immunoblot analysis. HA-RASSF1A coprecipitated with Flag-RABP1 from the cell lysates (Fig. 1C). We also confirmed that the COOH-terminal region of RASSF1A containing the RAS association domain was sufficient for interaction with Flag-RABP1 in these cells (Fig. 1D). Immunofluorescence analysis also revealed that Flag-RABP1 colocalized with HA-RASSF1A at microtubules during interphase and at spindle poles during mitosis in HeLa cells (data not shown).
RABP1 Is a Centrosomal Protein-We next generated polyclonal antibodies to a fragment of RABP1 comprising amino acids 528 -828. These antibodies specifically recognized both the endogenous protein (ϳ130 kDa) and exogenous Flag-RABP1 in NCI-H23 cells ( Fig. 2A). With the use of these antibodies, we investigated whether the expression of RABP1 is cell cycle-dependent. The abundance of endogenous RABP1 in HeLa cells was relatively constant throughout cell cycle progression (data not shown). We next assessed the interaction between endogenous RABP1 and RASSF1A in HeLa cells. Endogenous RASSF1A was indeed precipitated by the antibodies to RABP1 but not by normal rabbit IgG ( Fig. 2A). However, given that none of the available antibodies to RASSF1A are able to immunoprecipitate RASSF1A, we could not perform the inverse co-immunoprecipitation experiment.
We then examined the subcellular distribution of endogenous RABP1 in HeLa cells by immunofluorescence analysis with the antibodies to RABP1. A substantial proportion of endogenous RABP1 was localized to the centrosome, as revealed by staining with antibodies to ␥-tubulin, during both interphase and mitosis (Fig. 2B). To confirm that RABP1 is a centrosome-associated protein, we isolated a centrosome-enriched fraction from HeLa cells (28) and subjected it to immunoblot analysis with antibodies to RABP1. A substantial proportion of cellular RABP1 was present in the centrosome fraction, whereas the nuclear protein Chk2 was not detected in this fraction (Fig. 2B). HeLa cells transiently transfected with empty vector or with 1, 2.5, or 5 g of Flag-RABP1 vector were subjected to immunoblot analysis of the indicated proteins. Immunoprecipitates prepared from cell lysates with antibodies to Cdc2 were also assayed for kinase activity with [␥-32 P]ATP and histone H1 as a substrate. C, RABP1-induced stabilization of cyclins A and B. HeLa cells expressing Flag-RABP1 or transfected with empty vector were released from thymidine block for the indicated times after which cell lysates were subjected to immunoblot analysis of the indicated proteins. D, RASSF1A dependence of RABP1-induced mitotic arrest and accumulation of mitotic cyclins. HeLa cells cotransfected with vectors for Flag-RABP1 (or empty vector) and for RASSF1A siRNA were subjected to mitotic analysis (left panel), to immunoblot analysis of the indicated proteins (middle panel), and to immunoprecipitation with antibodies to Cdc20 followed by immunoblot analysis with antibodies to RASSF1A and to Cdc20 (right panel). The asterisk indicates Ig heavy chain. IB, immunoblot; IP, immunoprecipitation.
We further examined the association of RABP1 with RASSF1A in HeLa cells by indirect immunostaining. Endogenous RABP1 colocalized with endogenous RASSF1A at the centrosome during both interphase and mitosis (Fig. 2C), indicating that the two proteins interact throughout the cell cycle. In addition, both GFP-tagged RABP1 and red fluorescent protein-tagged RASSF1A in transiently transfected cells localized to the centrosome throughout the cell cycle (data not shown). These results thus demonstrate that the newly identified centrosomal protein RABP1 associates with RASSF1A in vivo.
Given that the MAP domains of RABP1 were shown to be responsible for binding to RASSF1A, we tested whether these domains are also required for the localization of RABP1 to the centrosome throughout the cell cycle. Immunostaining of transfected HeLa cells with antibodies to FLAG and to ␥-tubulin revealed that FLAG-tagged deletion mutants of RABP1 that contain either of the two MAP domains (RABP1-N2 and RABP1-N4) localized to the centrosome, whereas those that did not contain these domains (RABP1-N1 and RABP1-N3) did not (Fig. 2D). Either of the MAP domains of RABP1 thus appears sufficient for localization to the centrosome as well as for interaction with RASSF1A.

RASSF1A-dependent Mitotic Arrest and Stabilization of Mitotic Cyclins Induced by Overexpression of RABP1-Given that
we had previously shown that overexpression of RASSF1A induces stabilization of mitotic cyclins and mitotic arrest at prometaphase by inhibiting APC-Cdc20 activity, we next examined the effects of RABP1 overexpression on the cell cycle. Flow cytometric analysis of HeLa cells transiently cotransfected with vectors for Flag-RABP1 and CD4 (surface marker) revealed that overexpression of RABP1 induced G 2 -M arrest (Fig. 3A). Immunostaining with antibodies to ␤-tubulin and to FLAG revealed that overexpression of RABP1 markedly increased the mitotic index of cells and that most RABP1-overexpressing cells were in prometaphase (Fig. 3A). Overexpression of RABP1 in HeLa cells also increased the abundance of cyclins A and B in a concentration-dependent manner (Fig. 3B), consistent with cell cycle arrest at prometaphase. In contrast, RABP1 overexpression did not increase the level of cyclin D and actually reduced its abundance, probably as a result of the mitotic arrest induced by RABP1, thus indicating that RABP1 specifically affects the concentration of mitotic cyclins. Consistent with this conclusion, Cdc2 activity was also augmented in a concentration-dependent manner by overexpression of RABP1 (Fig. 3B). We determined the duration of the mitotic arrest induced by RABP1 overexpression in HeLa cells synchronized by the double thymidine block protocol (25). Whereas the amounts of cyclins A and B in control cells were greatly reduced by 10 h after release from the block, the accumulation of cyclins A and B apparent in RABP1-overexpressing cells was maintained for more than 14 h after release (Fig. 3C). These results thus indicated that overexpression of RABP1 induces an early mitotic arrest that is accompanied by accumulation of cyclins A and B.
Given that RABP1 interacts with RASSF1A and that the effects of RABP1 overexpression, including early mitotic arrest and accumulation of mitotic cyclins, are similar to those apparent in cells overexpressing RASSF1A (25), we then tested whether RASSF1A contributes to the mitotic arrest induced by RABP1. Overexpression of RABP1 failed to induce either mitotic arrest or accumulation of cyclins A and B in HeLa cells depleted of RASSF1A by RNAi (Fig. 3D). Given that RASSF1A binds to Cdc20 and inhibits APC-Cdc20 activity, we then examined the effect of RABP1 overexpression on the level of association between RASSF1A and Cdc20. The extent of the interaction between RASSF1A and Cdc20 was increased in cells overexpressing RABP1 compared with that apparent in control cells (Fig. 3D), suggesting that the induction of mitotic arrest by RABP1 is likely due to an increased level of association between RASSF1A and Cdc20. Together these results indicate that both the mitotic arrest and accumulation of mitotic cyclins induced by ectopic expression of RABP1 are dependent on RASSF1A-mediated inhibition of APC-Cdc20 activity.
Mislocalization of RASSF1A and Its Dissociation from Cdc20 Induced by Depletion of RABP1-Given that each of the two MAP domains of RABP1 is sufficient for its association with RASSF1A (Fig. 1A) and localization to the spindle poles (Fig.  2D), we determined whether RABP1 contributes to RASSF1A localization to the spindle poles during mitosis. We first depleted HeLa cells of endogenous RABP1 by RNAi. One of two RABP1-specific siRNAs tested stably suppressed RABP1 expression in these cells (Fig. 4A). We then examined the localization of RASSF1A during mitosis in cells depleted of RABP1. RASSF1A failed to localize to the centrosome during mitosis in RABP1-depleted cells (Fig. 4B), suggesting that RABP1 is required for the recruitment of RASSF1A to spindle poles.
RASSF1A negatively regulates the activity of APC-Cdc20 through association with Cdc20 on the spindle and at the spindle poles during prometaphase (25,26,30,31). We therefore determined whether RABP1 is required for the interaction FIG. 4. Mislocalization of RASSF1A and its dissociation from Cdc20 in cells depleted of RABP1. A, depletion of RABP1 in HeLa cells by RNAi. Cells transfected with a vector for one of two RABP1 siRNAs (or with empty vector) were subjected to immunoblot analysis with antibodies to RABP1 and to ␤-tubulin. B, loss of RABP1 results in mislocalization of RASSF1A. HeLa cells expressing RABP1 siRNA-1 or those transfected with empty vector were stained with antibodies to RASSF1A (green), antibodies to Cdc20 (red), and DAPI (blue). Scale bars, 10 m. C, loss of RABP1 prevents the interaction of RASSF1A with Cdc20. Lysates of HeLa cells expressing RABP1 siRNA-1 or of those transfected with empty vector were subjected to immunoprecipitation with antibodies to Cdc20, and the resulting precipitates were subjected to immunoblot analysis with antibodies to RASSF1A or to Cdc20. The asterisk indicates Ig heavy chain. IB, immunoblot; IP, immunoprecipitation.
between RASSF1A and Cdc20. RASSF1A was neither colocalized with Cdc20 on the spindle or at the spindle poles nor co-immunoprecipitated with Cdc20 in RABP1-depleted cells (Fig. 4, B and C). We did not detect a direct interaction between RABP1 and Cdc20 in an in vitro binding assay (data not shown). Together these results thus suggest that the recruitment of RASSF1A to spindle poles by RABP1 is required for the interaction of RASSF1A with Cdc20 during mitosis.
Acceleration of Mitotic Progression and Mitotic Abnormalities Induced by RABP1 Depletion-Given that depletion of RASSF1A accelerates mitotic progression as a result of premature APC activation (25), we examined whether the failure of RASSF1A to localize to spindle poles in RABP1-depleted cells also promotes mitotic progression. RABP1-depleted cells (those expressing RABP1 siRNA-1) did not differ from control cells (those transfected with empty vector or with the vector for RABP1 siRNA-2, which does not suppress RABP1 expression) in the kinetics of cell cycle progression from S phase to prometaphase. However, RABP1-depleted cells had already progressed to G 1 phase within 12 h after release from thymidine block, whereas control cells did not reach G 1 until 14 h (Fig.  5A), indicating that loss of RABP1 indeed accelerated mitotic progression. Consistent with these observations, the destruction of cyclins A and B was initiated prematurely in RABP1depleted cells (at 12 h after release from thymidine block) compared with that apparent in control cells (Fig. 5B). Time lapse microscopy also showed that mitotic progression from metaphase to telophase was ϳ15-20 min faster in cells depleted of RABP1 than in control cells (Fig. 5C). Together with our demonstration that RASSF1A failed to localize to spindle poles or to bind to Cdc20 in RABP1-depleted cells (Fig. 4), these observations suggest that depletion of RABP1 accelerates mitotic progression as a result of premature APC activation.
Acceleration of mitosis by premature APC activation increases the frequency of mitotic abnormalities (24,25,32). We therefore examined whether depletion of RABP1 also induces mitotic abnormalities in HeLa cells. Immunostaining of cells with antibodies to ␥-tubulin to label the centrosome and staining of DNA with DAPI revealed the presence of various mitotic spindle abnormalities, including the formation of multipolar spindles and misalignment of chromosomes on the metaphase plate, in cells depleted of RABP1 (Fig. 6). Quantitative analysis showed that the frequency of abnormal mitosis was increased about 4-fold to ϳ20% in RABP1-depleted cells compared with the corresponding value for control cells. Together these results suggest that RABP1 contributes to regulation both of the timing of normal mitotic progression and of equal chromosomal segregation by recruiting RASSF1A to spindle poles, allowing RASSF1A to inhibit APC-Cdc20 activity during mitosis.

DISCUSSION
The findings that RASSF1 is frequently silenced in a variety of solid tumors and that restoration of RASSF1A expression in such tumor cells inhibits cell growth both in vitro and in vivo suggest that RASSF1A is a tumor suppressor protein. We have previously shown that RASSF1A localizes to spindle poles and spindle microtubules during mitosis and that RASSF1A overexpression results in mitotic arrest before the metaphase-anaphase transition (25). RASSF1A associates with Cdc20 at the spindle apparatus and inhibits the ability of APC-Cdc20 to promote both the degradation of cyclins A and B and mitotic progression.
Our previous data suggested that the inhibition of APC-Cdc20 activity by RASSF1A is mediated by a mechanism more complex than simple association of RASSF1A with Cdc20 (25,26). We have now isolated a MAP-related protein, RABP1 (previously known as C19ORF5), that interacts with RASSF1A in vivo. While this manuscript was in preparation, Dallol et al. (33) also showed that RASSF1A interacts with MAP1A and with C19ORF5, although the functional significance of the interaction between RASSF1A and C19ORF5 was not demon-strated. Our data now show that RABP1 is a centrosomal protein and is required for RASSF1A function in mitosis. We thus found that RABP1 associates with RASSF1A through its MAP domains and colocalizes with RASSF1A at the centrosome. Overexpression of RABP1 induced mitotic arrest at prometaphase in a manner dependent on RASSF1A function. Conversely depletion of RABP1 by RNAi prevented the localization of RASSF1A to the spindle apparatus as well as induced the dissociation of RASSF1A from Cdc20, consistent with the notion that RABP1 is required for the recruitment of RASSF1A to spindle poles. Finally we showed that depletion of RABP1 resulted in acceleration of mitotic progression, premature destruction of mitotic cyclins, and an increase in the frequency of mitotic abnormalities, suggesting that RABP1 depletion induces premature activation of APC-Cdc20. Given that all of these effects of RABP1 depletion are similar to those previously observed in cells depleted of RASSF1A (25), they are likely due to loss of RASSF1A-mediated regulation of APC-Cdc20. Although we observed that overexpression of RASSF1A both delayed mitotic progression and induced mitotic abnormalities in RABP1-depleted cells (data not shown), this finding is likely attributable to the fact that RASSF1A stabilizes microtubules and binds to MAP1A as well as to RABP1 (13,14,33). Indeed we observed such a stabilizing effect of RASSF1A or MAP1A, whereas neither overexpression nor depletion of RABP1 affected microtubule stability in nocodazole-treated cells (Ref. 33 and data not shown). The stabilization of microtubule dynamics by RASSF1A thus appears to be independent of RABP1mediated RASSF1A function. We therefore postulate that RABP1 is responsible for recruiting RASSF1A to spindle poles where RASSF1A functions to inhibit APC-Cdc20 activity rather than in regulation of microtubule stability.
Mitotic progression is tightly regulated by the timed destruction of mitotic regulators, which is dependent on APC activity. The activity of the APC is itself temporally controlled by negative regulators, including Emi1, RASSF1A, and Mad2, that associate differentially with Cdc20 during mitotic progression (19 -25, 30, 31). Emi1 inhibits the APC during G 1 and prophase (24,34). After most Emi1 molecules are degraded during prophase, RASSF1A is responsible for inhibiting APC activity and preventing degradation of cyclins A and B until the spindle checkpoint becomes fully operational and inhibits the APC in a Mad2-and BubR1-dependent manner.
In addition to its temporal regulation, the activity of APC-Cdc20 appears to be regulated in a spatially dependent manner as indicated by the observations that cyclin B disappears first from the spindle poles in human cells (35) and Drosophila embryos (36) and that activation of the APC is detected first at spindle poles (37). The APC is thus likely activated initially at spindle poles, and the activated APC then moves into the spindle and mediates degradation of cyclin B, which is restricted to the spindle and spindle poles. The current model of spindle checkpoint activation by unattached kinetochores of chromosomes or unbalanced tension between sister chromatids might not be sufficient to explain how activated APC both mediates cyclin A degradation and initiates cyclin B degradation from spindle poles rather than from kinetochores (26,30). Although Mad2-BubR1 contributes to prevention of premature onset of anaphase, cyclin A destruction occurs before activation of the spindle checkpoint. On the basis of the results of our previous (25) and present studies, we propose that RASSF1A is localized predominantly to spindle poles as a result of its association with RABP1 at the prophase-prometaphase transition and thereby inhibits APC-Cdc20 activity and prevents the destruction of mitotic cyclins at the spindle poles until the spindle checkpoint monitors unattached kinetochores (Fig. 7) (26,30,31).
Important questions regarding regulation of APC-Cdc20 by RASSF1A-RABP1 remain to be answered. We still do not know how depletion of either RABP1 or RASSF1A induces multiple spindle poles as well as chromosome misalignment in cells. The most likely explanation is that premature activation of APC-Cdc20 due to loss of either RABP1 or RASSF1A disrupts the proper timing of mitotic regulation. Alteration of APC regulators, including Emi1, Mad2, Bub1, and BubR1, induces similar mitotic abnormalities (24,32,38), the molecular basis of which also remains to be determined. Given that RASSF1A also blocks cell cycle progression at the G 1 -S transition (11,12), we cannot absolutely rule out the possibility that manipulation of RABP1 or RASSF1A results in an abnormal number of centrosomes by affecting centrosome duplication at this transition.
It also remains to be determined when and how RASSF1Amediated inhibition of APC-Cdc20 activity is relieved. In addition, given that RABP1 possesses two potential phosphorylation sites for Cdc2, it will be important to determine whether phosphorylation of RABP1 by Cdc2 affects mitotic progression. Furthermore, given that loss of RABP1, like that of RASSF1A, induced both acceleration of mitotic progression and chromosomal instability in HeLa cells, it is possible that RABP1 also functions as a tumor suppressor protein.