Control of Microtubule Stability by the RASSF1A Tumor Suppressor

Control of microtubule stability by the RASSF1A tumor suppressor

Limin Liu1, Stella Tommasi1, Dong-Hyun Lee1, Reinhard Dammann2 and Gerd P Pfeifer*,1

1Division of Biology, Beckman Research Institute of the City of Hope; Duarte, CA 91010, USA; 2Institut fu¨r Humangenetik und Medizinische Biologie, Martin-Luther-Universita¨t Halle-Wittenberg, D-06097 Halle/Saale, Germany

The RAS association domain family 1A (RASSF1A) gene by DNA methylation of its promoter CpG island at a is silenced by DNA methylation in over 50% of all solid high frequency, as observed originally in lung cancers tumors of different histological types. However, the (Dammann et al., 2000; Burbee et al., 2001). RASSF1A biochemical function of the RASSF1A protein is un- is the most frequently methylated gene so far described known. We show that RASSF1A colocalizes with micro- in human cancer (Pfeifer et al., 2002; Dammann et al., tubules in interphase and decorates spindles and 2003). RASSF1A methylation occurs in a broad centrosomes during mitosis. RASSF1A has a strong spectrum of carcinomas and other solid tumors (Dam- cytoprotective activity against the microtubule-destabiliz- mann et al., 2000, 2001a, b; Agathanggelou et al., 2001; ing drug nocodazole, and against cold-treatment in vivo. Burbee et al., 2001; Byun et al., 2001; Dreijerink et al., Conversely, loss of RASSF1 in RASSF1À/À mouse 2001; Lee et al., 2001; Lo et al., 2001; Maruyama et al., embryonic fibroblasts renders the cells more sensitive to 2001; Morrissey et al., 2001; Kuzmin et al., 2002; Liu nocodazole-induced depolymerization of microtubules. et al., 2002; Lusher et al., 2002; Schagdarsurengin et al., The domain required for both microtubule association 2002; Toyooka et al., 2002; van Engeland et al., 2002; and stabilization was mapped to a 169 amino-acid Wagner et al., 2002; Zhang et al., 2002; Chan et al., fragment that contains the RAS association domain. 2003; Dammann et al., 2003; Honorio et al., 2003; Overexpression of RASSF1A induces mitotic arrest at Schagdarsurengin et al., 2003; Spugnardi et al., 2003). metaphase with aberrant mitotic cells reminiscent of such Re-expression of RASSF1A reduces the growth of produced by the microtubule-stabilizing drug paclitaxel human cancer cells supporting a role for RASSF1 as a (taxol), including monopolar spindles, or complete lack of tumor suppressor gene (Dammann et al., 2000; Burbee a mitotic spindle. Altered microtubule stability in cells et al., 2001; Dreijerink et al., 2001; Kuzmin et al., 2002). lacking RASSF1A is likely to affect spindle assembly and The biochemical function of the RASSF1A protein is chromosome attachment, processes that need to be unknown. The homology of RASSF1A with the carefully controlled to protect cells from genomic mammalian RAS effector NORE1 suggested that the instability and transformation. In addition, knowledge of RASSF1A gene product may function in signal trans- the microtubule-targeting function of RASSF1 may aid in duction pathways involving RAS-like proteins. Recent the development of new anticancer drugs. data indicate that RASSF1A itself binds to RAS only Oncogene (2003) 22, 8125–8136. doi:10.1038/sj.onc.1206984 weakly, and that binding to RAS may require heterodimerization of RASSF1A and NORE1 (Ortiz-Vega Keywords: RASSF1A; mitosis; DNA methylation; et al., 2002). There is evidence for an association of both microtubules; taxol NORE1 and RASSF1A with the proapoptotic kinase MST1 and that this interaction is involved in apoptosis (Khokhlatchev et al., 2002). Other investigations have Introduction uncovered a role of RASSF1A in suppression of cyclin D accumulation and cell cycle progression (Shivakumar Recently, we have cloned and characterized the RAS et al., 2002). association domain family 1 (RASSF1) gene (Dammann Here, we present evidence that RASSF1A is a et al., 2000). The RASSF1 gene is located at 3p21.3 microtubule binding protein that can stabilize micro- within an area of common heterozygous and homo- tubules from depolymerization in vivo. RASSF1A zygous deletions, which occur frequently in a variety of localizes to the mitotic apparatus and controls mitotic human solid tumors (Kok et al., 1997; Sekido et al., progression. 1998; Lerman and Minna, 2000). None of the eight genes in the common homozygous deletion area at 3p21.3 is mutated at a significant frequency in cancer Results (Lerman and Minna, 2000). However, one of the isoforms of the RASSF1 gene, RASSF1A, is inactivated RASSF1A associates with microtubules in both interphase and mitosis *Correspondence: GP Pfeifer, Division of Biology, City of Hope, 1500 East Duarte Road, Duarte, CA 91010, USA; E-mail: [email protected] To analyse the function of RASSF1A, we investigated Received 28 May 2003; revised 1 July 2003; accepted 3 July 2003 the subcellular localization of this protein in COS-7 Microtubule stabilization by RASSF1A LLiuet al 8126 cells. Using EGFP-tagged RASSF1A, we observed that with microtubules was observed in the interphase cells the RASSF1A signal formed a meshwork-like pattern in (Figure 1a–a00). However, no significant colocalization the cytoplasm with brighter signals surrounding the of RASSF1A with F-actin was observed (data not nucleus (Figure 1A). When EGFP-RASSF1A was shown). To further demonstrate the association of introduced into other mammalian cell lines, including RASSF1A with microtubules, we treated EGFP-RASS- A549, LNCaP, HeLa, and NIH-3T3, the same pattern F1A-transfected COS-7 cells with nocodazole, a micro- was observed (data not shown). When we transfected tubule depolymerizer, and with taxol, a microtubule EGFP-RASSF1A into COS-7 cells and costained with stabilizer. Upon disruption of microtubules by nocoda- an anti-a-tubulin antibody, colocalization of RASSF1A zole (2 mm, 20 h), the fibrous structure of EGFP-

a a' a''

b b' b''

c c' c''

Figure 1 RASSF1A associates with microtubules in the interphase cells. COS-7 cells transfected with EGFP-RASSF1A (green) were left untreated (a, a0,a00) or treated with nocodazole (b, b0,b00;2mm, 20 h), or taxol (c, c0,c00;5mm, 15 h). Cells were stained with an anti-a- tubulin antibody (red). The arrows indicate the MTOC. Scale bars, 10 mm. (d) 293T cells were transfected with pEBG (GST) or GST- RASSF1A. Total cellular extracts were incubated with glutathione sepharose 4B, and bound proteins were analysed by Western blotting using antibodies against GST, b-tubulin, or g-tubulin

Oncogene Microtubule stabilization by RASSF1A LLiuet al 8127 RASSF1A disappeared. Instead, the RASSF1A signal F1A staining accumulated to spindle poles and only was diffused throughout the cytoplasm with about 15– slightly decorated the interpolar spindles. In telophase, 25% entering the nucleus (Figure 1b–b00). Conversely, RASSF1A intensely stained both the spindle poles and RASSF1A colocalized with bundled microtubules in the interzone region. During cytokinesis, the cytoplas- cytoplasmic protrusions when taxol was used (Figure mic microtubules reappear, and RASSF1A staining was 1c–c00). These results suggest that the localization of distributed along the astral microtubules and concen- RASSF1A depends on the integrity of the microtubule trated at the spindle poles. Strikingly, the midbody network. Next, we performed an in vivo GST pull-down region, where the microtubules overlap and the cleavage assay. b-tubulin and g-tubulin could only be detected in furrow forms, was also intensely stained by RASSF1A. the pull-down product from GST-RASSF1A-trans- These results suggest that RASSFIA may play a role fected 293T cells, but not in the pull-down product during mitosis and cytokinesis. from pEGB (GST)-transfected cells (Figure 1d), suggesting that RASSF1A can associate with b-tubulin and RASSF1A stabilizes microtubules in vivo g-tubulin physically. Since g-tubulin is a centrosome- speciﬁc protein, this is consistent with our results We treated COS-7 cells transfected with EGFP-RASS- showing that RASSF1A intensely stained the micro- F1A with nocodazole, a tubulin-depolymerizing drug. tubule-organizing center (MTOC) in interphase cells When untransfected or EGFP-transfected COS-7 cells (Figure 1a–a00). were incubated with 20 mm nocodazole for 60 min, their To further explore the association of RASSF1A with microtubular array was completely destructed (Figure microtubules, we studied cells throughout mitosis 3b–b00). In contrast, the microtubule structure was (Figure 2). In prophase and prometaphase, RASSF1A largely intact in over 75% of the COS-7 cells with localized to spindle poles and attached spindles. When forced expression of EGFP-RASSF1A (Figure 3a–a00). cells progressed into metaphase and anaphase, RASS- To further explore the microtubule stabilizing effect of

Figure 2 RASSF1A localizes to the mitotic apparatus during mitosis. COS-7 cells transfected with EGFP-RASSF1A (green) were ﬁxed, permeabilized and costained with an anti-a-tubulin antibody (red) and DAPI (blue). Cells at each mitotic stage are as indicated

Oncogene Microtubule stabilization by RASSF1A LLiuet al 8128

a a' a''

b b' b''

c c' c''

d d' d''

Figure 3 RASSF1A stabilizes microtubules against disruption by nocodazole or ice treatment. COS-7 cells transfected with EGFP- RASSF1A (a, a0,a00) or EGFP alone (b, b0,b00) were incubated with 20 mm nocodazole for 1 h. EGFP-RASSF1A-transfected cells (c, c0, c00) or EGFP-transfected cells (d, d0,d00) were held on ice for 30 min. After treatment, cells were ﬁxed, permeabilized and stained with an anti-a-tubulin antibody (red). Arrows indicate cells transfected with the indicated constructs and arrowheads indicate untransfected cells

RASSF1A, we held COS-7 cells on ice for 30 min, wild type cells (Figure 5). In RASSF1À/À MEFs, conditions that resulted in loss of microtubules in all the microtubules began to disrupt at 0.5 mm nocodazole untransfected or EGFP-transfected cells (Figure 3d–d00). and microtubular fibers could hardly be seen at 2 mm Surprisingly, the microtubules still remained intact in nocodazole (Figure 5). ice-treated COS-7 cells that were transfected with Compared with cytoplasmic microtubules, mitotic EGFP-RASSF1A (Figure 3c–c00). To support these spindles are more dynamic structures. Overexpression of data, we generated RASSF1À/À mouse embryonic RASSF1A cannot protect spindles from depolymeriza- fibroblasts (MEFs) (Figure 4). The sensitivity of these tion by cold treatment for 30 min. Interestingly, the cells to nocodazole was analysed. Consistent with the centrosome spots were still visible in these cells and protective effect seen in the transfection studies, RASSF1A staining coincided with these spots (Figure loss of RASSF1 rendered the MEFs more sensitive to 6a–a000, b–b000). Since RASSF1A can interact with g- nocodazole treatment. When different concentrations of tubulin (Figure 1d) and RASSF1A concentrates at the nocodazole were used for 30 min, we found that spindle pole during mitosis (Figure 2), the data strongly microtubules began to partially depolymerize at 1 mm indicate that RASSF1A can interact with the centro- but microtubule fibers were still detectable at 6 mm in some.

Oncogene Microtubule stabilization by RASSF1A LLiuet al 8129

Figure 4 RASSF1 knockout mice. (a) Strategy for targeting the mouse RASSF1 locus. (b) Southern blot analysis of targeted ES cell clones. After correct targeting and elimination of exons 1–6, the expected size of the ScaI and SpeI fragments is 7.5 and 7.3kb, respectively. These bands are present in several of the clones. ES cell clone #52 was used for the production of chimeric mice, followed by germline transmission and breeding to homozygosity. Lanes labeled ‘C’ are nonrecombinant ES cell clones. (c) RT–PCR analysis of RASSF1A in MEFs from wild-type, heterozygous, and homozygous knockout mice

When EGFP-RASSF1A-transfected COS-7 cells were by fluorescence microscopy. Only two mutants, RASS- incubated with 2 mm nocodazole for 20 h, over 80% of F1A 120-340, and RASSF1A 1-288, could bind to the cells lose the integrity of microtubules. When a 5 min microtubule-like structures in the cytoplasm (Figure 7b), time period was allowed for the cells to recover after and they colocalized with microtubules in the interphase removal of nocodazole, the microtubular array began to cells (data not shown). Unlike full-length RASSF1A and nucleate from the newly forming centrosome (Figure RASSF1A 120–340, a nuclear accumulation of RASS- 6c0). Strikingly, 90% of RASSF1A staining was solely F1A 1–288 is predominant in COS-7 cells. This suggests concentrated at this star-like structure (Figure 6c–c000), that the C-terminal 52 amino acids play a role in suggesting that RASSF1A may promote microtubule retaining full-length RASSF1A in the cytosol. It is less polymerization in vivo. likely that this region or this region combined with an adjacent region contain a nuclear export signal (NES), Microtubule-binding and -stabilizing domain mapping of because no typical NES was detected. When this region RASSF1A alone (RASSF1A 289–340) or this region plus an adjacent region (RASSF1A 194–340) was cloned into The RASSF1A cDNA encodes a protein of 340 amino an EGFP vector, the signal largely localized to the acids with a calculated MW of 38.8 kDa. The C- nucleus (Figure 7b). Like RASSF1A full-length protein, terminus of RASSF1A contains a putative Ras-associa- both mutants can localize to the spindle apparatus tion domain (R194 to S288), which shows high during mitosis (data not shown). Similarly, these two homology to the murine Ras-effector protein Nore1 mutants were also capable of protecting microtubules (Vavvas et al., 1998) and may mediate interaction with from disruption by nocodazole (Figure 7c, d) and by Ras oncoproteins. The N-terminus of RASSF1A has cold treatment (data not shown). The other mutants did high homology to a cystein-rich diacyl-glycerol/phorbol not significantly colocalize with microtubules and were ester-binding domain (Newton, 1995). RASSF1A also incapable of protecting microtubules from depolymer- has a putative ATM kinase phosphorylation consensus ization (data not shown). These results indicate that the motif (Kim et al., 1999) spanning amino acids W125 to minimal overlapping domain required for association K138. To determine the domain(s) that is (are) required with and stabilization of microtubules is a 169 amino- for RASSF1A to associate with and to stabilize acid fragment from D120 to S288, which includes the microtubules, we generated a series of deletion mutants intact RAS-association domain (RA) and putative of RASSF1A tagged with EGFP (Figure 7a). The ATM phosphorylation site of RASSF1A. This sequence subcellular distribution of these mutants was analysed shares no significant homology with any other known

Oncogene Microtubule stabilization by RASSF1A LLiuet al 8130 microtubule-binding protein, deﬁning a novel micro- domain in RASSF1C would suggest that this protein tubule interacting domain. Although we have done no may bind to microtubules as well. In addition, these detailed study with RASSF1C, the other major isoform sequences are highly conserved in the two closest expressed from the RASSF1 locus, the presence of this RASSF1 homologues, NORE1 and RASSF3(Tommasi et al., 2002), implying a conserved function of RASSF1- related proteins.

RASSF1A regulates mitotic progression To determine whether RASSF1A has any effect on mitosis, EGFP-tagged RASSF1A and its truncated mutant constructs were transfected into COS-7 cells. Cells were double-stained with anti-a-tubulin antibody and 4,6-diamidino-2-phenylindole (DAPI), and mitotic cells were scored. Strikingly, overexpression of RASS- F1A resulted in over 50% of mitoses with only monopolar spindles (Figure 8a–a000, c). In these cells, RASSF1A colocalized with the monopolar spindles. Typically these cells have highly condensed ball-shape chromatin (Figure 8a00). Similarly, ectopic expression of RASSF1A 120–340 also led to 46% mitoses with monopolar spindles. In contrast, only 5 and 12% monopolar mitoses were observed in untransfected and EGFP-transfected COS-7 cells, respectively (Figure 8c). In RASSF1A 1–288-transfected cells, 18% of all mitoses were monopolar, but 51% of all mitoses were without any spindle structure (Figure 8b–b000, d). In these spindle-less mitoses, the chromosomes were also highly condensed and the only tubulin-containing structures were some punctate aggregates of tubulin throughout the cell. RASSF1A 1–288 also distributed without any definitive spindle structure in these mitotic cells (Figure 8b–b000). In all, 10% of mitoses were found to be without spindle in RASSF1A 120–340-transfected cells. How- ever, less than 2% of spindle-less mitoses were found in cells untransfected or transfected with EGFP, full-length RASSF1A, RASSF1A 1–193, RASSF1A 1–220, RASS- F1A 194–340, or RASSF1A 152-340, respectively (Figure 8D). Since a large percentage of abnormal mitoses appeared in cells overexpressing RASSF1A or two of its truncated mutants that can bind to microtubules, we calculated the ratio of cells in anaphase to cells in metaphase. In untransfected COS-7 cells, this ratio was 0.16. A significant increase in this ratio may indicate an anaphase arrest, while a significant decrease in this ratio may suggest a metaphase arrest. As shown in Figure 8e, this ratio dramatically dropped to 0.016, 0.015, and 0.002 in RASSF1A-, RASSF1A 120-340-, and RASSF1A 1–288- transfected cells, respectively, suggesting a metaphase arrest in these cells.

Figure 5 RASSF1À/À MEFs are more sensitive to nocodazole-induced microtubule depolymerization than the wild-type MEFs. Wild-type embryonic fibroblasts (WT MEF) and RASSF1À/À embryonic fibroblasts (RASSF1À/À MEF) were incubated with nocodazole at the indicated concentrations for 30 min. Cells were fixed, permeabilized and stained with an anti-a-tubulin antibody. A total of 100–150 cells were scored for microtubule integrity, and images are representations of 470% of the cells at each concentration of nocodazole

Oncogene Microtubule stabilization by RASSF1A LLiuet al 8131

a a' a'' a'''

b b' b'' b'''

c c' c'' c'''

Figure 6 RASSF1A colocalizes with centrosomes. COS-7 cells transfected with EGFP-RASSF1A were treated with ice for 30 min (a, a0,a00,a000;b,b0,b00,b000) or incubated with 2 mm nocodazole for 20 h (c, c0,c00,c000). Cells were ﬁxed immediately after cold treatment or 5 min after removal of nocodazole and were double-stained with an anti-a-tubulin antibody (red) and DAPI (blue). Arrows indicate centrosomes

Discussion cellular localization was not signiﬁcantly changed compared to that in the paired untransformed cell line We have shown here that the RASSF1A protein (data not shown), suggesting that microtubule targeting localizes to the microtubule network. This result was of RASSF1A is independent of active k-RAS. somewhat unexpected since RASSF1 is endowed with a Recently, several other tumor suppressor gene pro- RAS association domain, a domain found in proteins ducts have been reported to associate with the micro- that can bind to activated (GTP bound) RAS family tubule network. The adenomatous polyposis coli (APC) members (Ponting and Benjamin, 1996). However, tumor suppressor protein localizes to and stabilizes previous studies have shown that RASSF1A is able to microtubules (Zumbrunn et al., 2001). The von Hippel– bind to activated RAS only indirectly through its closest Lindau tumor suppressor protein (pVHL) also is a homologue and heterodimerization partner NORE1 microtubule-associated protein that can protect micro- (Ortiz-Vega et al., 2002). The intracellular localization tubules from depolymerization (Hergovich et al., 2003). of NORE1 has not yet been reported. Activated k-RAS BRCA1, a tumor suppressor for breast cancer, has been proteins are normally localized at the plasma membrane shown to associate with centrosomes (Hsu and White, although an interaction of k-RAS with the microtubule 1998) and mitotic spindles (Lotti et al., 2002). It remains network has been described (Chen et al., 2000). An to be seen if microtubule association and stabilization by intact RAS association domain is required for RASS- several important tumor suppressor proteins can be F1A to bind to microtubules (Figure 7). It remains to be functionally linked to a common pathway that needs to determined if (indirect) k-RAS interaction and the be disabled in tumorigenesis. proapoptotic effect of RASSF1A (Khokhlatchev et al., We have shown here that expression of RASSF1A 2002) take place in the setting of microtubule associa- and several of its deletion mutants arrests cells in mitosis tion of RASSF1A. Interestingly, a putative Ras effector and promotes the formation of aberrant spindle in Schizosaccharomyces pombe, Scd1, was reported to structures. The microtubule stabilizing effect of RASS- localize to mitotic spindles (Li et al., 2000; Segal and F1A in interphase cells and the effects of its over- Clarke, 2001). Loss of Scd1 resulted in hypersensitivity expression on mitotic cells are surprisingly similar to to microtubule depolymerizing drugs, while overexpres- those produced by the commonly used cancer che- sion of Scd1 induced mitoses with abnormal spindles motherapy drug paclitaxel (taxol). Taxol produces and missegregated chromosomes. Our report is the ﬁrst mitotic arrest by stabilization of microtubules leading one to describe microtubule and spindle association for to a loss of microtubule dynamics (Jordan et al., 1993). a putative mammalian RAS effector. When RASSF1A The effects of taxol on the mitotic spindle are dose- was introduced into a K-ras transformed cell line, its dependent (Jordan et al., 1996). At the lowest doses, a

Oncogene Microtubule stabilization by RASSF1A LLiuet al 8132 a

c c' c''

d d' d''

Figure 7 Microtubule-binding and -stabilizing domain mapping of RASSF1A. (a) Schematic representation of RASSF1A and its truncated mutants. (B) COS-7 cells were transfected with the indicated EGFP-tagged RASSF1A deletion constructs and the subcellular localization of theses mutants was visualized using ﬂuorescence microscopy. (c–c00; d-d00) COS-7 cells transfected with the indicated EGFP-tagged RASSF1A deletion constructs were incubated with 20 mm nocodazole for 1 h. Cells were ﬁxed, permeabilized and double-stained with an anti-a-tubulin antibody (red) and DAPI (blue). Arrows indicate cells transfected with the indicated mutant constructs and arrowheads indicate untransfected cells

bipolar spindle is still formed but some of the chromo- (Jordan et al., 1992), a phenomenon observed with the somes fail to align along the metaphase plate (Jordan C-terminal RASSF1 deletion mutant (RASSF1A 1–288) et al., 1996). At higher concentrations of taxol, in our studies (Figure 8d). The mechanistic basis of the monopolar spindles, with congression of the chromo- taxol-like effect of RASSF1A is not known, but we somes into a ball-shaped structure, has been observed speculate that RASSF1A overexpression reduces micro- (Jordan et al., 1993; Jordan et al., 1996). We see exactly tubule dynamics, thus leading to a defect in centrosome the same effect with overexpression of full-length separation and bipolar spindle formation (monopolar or RASSF1A (Figure 8). At the highest concentrations of spindle-less mitoses). Knowledge of the microtubule- taxol or other microtubule-targeting drugs, the drugs targeting function of RASSF1 may be exploited for the promote destruction and loss of the mitotic spindle development of new anticancer drugs. In addition, it

Oncogene Microtubule stabilization by RASSF1A LLiuet al 8133

a a' a'' a'''

b b' b'' b'''

Figure 8 Overexpression of RASSF1A induces metaphase arrest and aberrant spindles. COS-7 cells were transfected with the indicated EGFP-tagged RASSF1A or its deletion constructs. Cells were ﬁxed, permeabilized and double-stained with an anti-a-tubulin antibody (red) and DAPI (blue). Resultant monopolar mitoses (a, a0,a00,a000 for full-length RASSF1A) or mitoses lacking spindle structures (b, b0, b00,b000 for RASSF1A 1-288) are shown. (c) The percentage of monopolar mitoses in each indicated transfection. (d) The percentage of mitoses without spindles in each indicated construct transfection. (E) The ratio (A/M) of cells in anaphase to cells in metaphase in each indicated transfection may be important to determine if patients with tumors 2001; Morrissey et al., 2001; Kuzmin et al., 2002; Liu harboring RASSF1 deﬁciencies may respond differently et al., 2002; Lusher et al., 2002; Schagdarsurengin et al., to treatment regimens involving spindle poisons. 2002; Toyooka et al., 2002; van Engeland et al., 2002; Most solid tumors are characterized by frequent Wagner et al., 2002; Zhang et al., 2002; Chan et al., epigenetic inactivation of the RASSF1A gene (Dam- 2003; Dammann et al., 2003; Honorio et al., 2003; mann et al., 2000, 2001a, b; Agathanggelou et al., 2001; Schagdarsurengin et al., 2003; Spugnardi et al., 2003). Burbee et al., 2001; Byun et al., 2001; Dreijerink et al., We have prepared RASSF1 knockout mice in order to 2001; Lee et al., 2001; Lo et al., 2001; Maruyama et al., investigate the role of this gene in tumorigenesis.

Oncogene Microtubule stabilization by RASSF1A LLiuet al 8134 Homozygous knockout mice are viable and fertile. They PCR from genomic 129/SvImJ DNA. Appropriate fragments are currently being observed for spontaneous and flanking exon 1a and exon 6 of mouse RASSF1 (3.9 and 5.3 kb induced tumor formation. Fibroblasts were prepared in length, respectively) were cloned into the pKSneolox from day 13.5 embryos. These RASSF1À/À fibroblasts targeting vector and were confirmed by DNA sequencing. are only 60–70% the size of wild-type MEFs (data not The vector was linearized and introduced into 129/SvImJ ES cells by electroporation. Drug-resistant ES cell clones were shown) and are much more sensitive than fibroblasts isolated and Southern analysis was used to identify recombi- from wild-type littermates towards microtubule destruc- nant clones. After production of chimeras and germline tion by nocodazole (Figure 5), suggesting cytoskeletal transmission, the RASSF1-targeted mice were bred to homo- defects. This confirms that the stabilizing effect seen in zygosity (129S1 Â B6J background). Absence of RASSF1 the cell line studies is not due to expression of the transcripts in the mouse tissues and embryonic fibroblasts protein at nonphysiological levels. from RASSF1À/À mice was confirmed by RT–PCR (Figure 4c). Loss of RASSF1A expression by promoter methyla- We used previously reported methods (Abbondanzo et al., tion is a very common phenomenon observed in almost 1993) to isolate and culture primary MEFs derived from 13.5 every type of cancer so far described (Dammann et al., day embryos. 2003). One notable exception is cervical carcinoma for which an inverse association between RASSF1A methy- Plasmid constructs lation and presence of high-risk papillomavirus To generate EGFP- or GST-tagged RASSF1A, the cDNA (HPV16/HPV18) sequences has been seen (Kuzmin encoding human RASSF1A was subcloned into EGFP-C2, et al., 2003). Human papillomavirus oncoproteins E6 EGFP-N2 (Clontech, Palo Alto, CA, USA), or pEGB (kindly and E7 promote mitotic infidelity by promoting centro- provided by Dr J Avruch; Massachusetts General Hospital) some duplication errors, aberrant mitoses, and aneu- vectors. Each experiment concerning EGFP-RASSF1A was carried out using both EGFP-C2- and EGFP-N2-tagged ploidy (Duensing and Mu¨ nger, 2002). Aberrations in RASSF1A. Similar results were obtained with C-terminal- chromosome numbers (aneuploidy, polyploidy) are a and N-terminal-tagged constructs. The truncated mutant common feature of most solid tumors for which RASSF1A sequences were generated by a single-step PCR RASSF1A inactivation has been reported, suggesting using EGFP-RASSF1A as template and cloned into EGFP-C2 that loss of RASSF1A and papillomavirus infection may or EGFP-N2. engage the same tumorigenic pathway. Thus, one hypothesis to be tested would be that lack of RASSF1A Immunofluorescence microscopy affects mitotic spindle function. In preliminary analysis At 24 h after transfection, COS-7 cells were trypsinized and of a limited number of metaphase spreads of RASSF1 seeded at 60% confluence into six-well plates containing knockout fibroblasts, we have not detected gross coverslips coated with poly-l-lysine (Sigma; St Louis, MO, chromosomal aberrations. However, one requirement USA). Cells were allowed to grow for an additional 24 h. The for such mitotic aberrations to result in viable daughter cells were washed with PBS, fixed with formaldehyde, and cells is that the mitotic spindle checkpoint, which permeabilized with 0.1% Triton X-100 for 5 min. The slides monitors unaligned chromosomes (Rieder et al., 1995), were treated with 1% BSA in PBS for 30 min. For tubulin also needs to be defective and/or that the apoptotic staining, the cells were incubated with a mouse monoclonal mechanism that eliminates such defective cells is anti-a-tubulin antibody (Molecular Probes; Eugene, OR, abrogated. This could be accomplished, for example, USA) for 1 h at room temperature. After washing with PBS, by mutation or epigenetic inactivation of checkpoint Alexa Fluor 568-conjugated anti-mouse secondary antibody (Molecular Probes) was used and DNA was counterstained genes such as BUB1, BUB3, MAD2, or others (Cahill with DAPI (Sigma). Images were obtained with an Olympus et al., 1998; Michel et al., 2001) or by overexpression of IX81 automated inverted microscope equipped with a Spot RT the serine–threonine kinase AURORA-A (Anand et al., Slider high-resolution cooled CCD color camera (Olympus; 2003). We are currently testing this hypothesis by Melville, NY, USA) or with a Zeiss LSM310 laser scanning analysing tumor samples for simultaneous aberrations confocal microscope with Zeiss MC100 35 mm camera. Color in RASSF1A and in known spindle checkpoint genes. images were processed using Photoshop 7.0 (Adobe Systems; San Jose, CA, USA) or ImagePro Plus (Media Cybernetics; Silver Spring, MD, USA). Materials and methods In vivo GST pull-down assay Cell culture and transfection At 60–72 h after transfection with GST-RASSF1A or pEBG COS-7, 293T, HeLa, A549, LNCaP, and NIH3T3 cells were (GST) vector alone, 293T cells were harvested and lysed in ice- m obtained from the ATCC and were cultured in high-glucose cold cytoskeleton lysis buffer (0.5% Nonidet P-40, 10 m m m m DMEM with 10% fetal bovine serum. Cells were transfected PIPES, pH 7.0, 50 m NaCl, 300 m sucrose, 3m MgCl2, m m m using Lipofectamine 2000 (Invitrogen, Carlsbad, CA, USA). 1m EGTA, 1 m DTT, 1 m Na3VO4) with protease inhibitor cocktail (Roche; Indianapolis, IN, USA). The lysates were incubated with glutathione sepharose 4B (Amersham; Knockout mice and embryonic fibroblasts Piscataway, NJ, USA) for 45 min at room temperature. After To construct RASSF1 knockout mice, we sequenced 18 kb of centrifugation and washing with PBS, the pellet was collected the mouse 129/SvImJ RASSF1 locus (Genbank Accession and the bound proteins were eluted with SDS loading buffer number AF333027). All RASSF1 exons are highly conserved by heating to 951C for 5 min. The eluted proteins were between mouse and human with 490% identity at the amino- fractionated on a 10% SDS–polyacrylamide gel and trans- acid level. We generated mouse PCR products by high-fidelity ferred to a PVDF membrane. The blots were probed with

Oncogene Microtubule stabilization by RASSF1A LLiuet al 8135 monoclonal antibodies against GST (Amersham), b-tubulin, Mitotic index or g-tubulin (Santa Cruz Biotechnologies; Santa Cruz, CA, COS-7 cells transfected with EGFP-RASSF1A or its truncated USA) followed by appropriate secondary antibodies conju- mutant constructs were double-stained with anti-a-tubulin gated with horseradish peroxidase. The signals were detected antibody (Molecular Probes) and DAPI (Sigma). Mitotic cells using ECL-Plus (Amersham). were scored and counted under a fluorescence microscope, based on both spindle structure and chromosome configura- Microtubule stability and repolymerization analysis tion. Mitotic cells with highly condensed chromosomes but without spindle or with a monopolar spindle were categorized To evaluate microtubule stability, coverslip cultures of COS-7 as metaphase according to the convention in the literature. For cells transfected with the indicated constructs were incubated each transfection, at least 300 mitotic cells were enumerated. with 20 mm nocodazole (Sigma) for 1 h. The coverslips were COS-7 cells, either untransfected or transfected with EGFP washed twice briefly with cold PBS and then were double- only, were analysed in parallel as controls. stained with anti-a-tubulin antibody (Molecular Probes) and DAPI (Sigma) as described above. For MEFs, the cells were seeded onto lysine-coated eight-chamber cover slides and were allowed to grow for 24 h. Nocodazole was used to treat the Abbreviations cells at concentrations of 0.5, 1, 2, 4, 6, and 8 mm for 30 min. DAPI, 4,6-diamidino-2-phenylindole; MEF, mouse embryonic The cover slides were washed twice with cold PBS and were fibroblast; MTOC, microtubule-organizing center; RASSF1A, stained with anti-a-tubulin as described above. At each RAS association domain family 1A. concentration of nocodazole, 100–150 cells were scored for microtubule integrity. The micrographs shown are representa- Acknowledgements tive of 475% of the cells in terms of microtubule integrity at We thank R Barber, M Lee, and P Salvaterra for assistance each concentration of nocodazole. To study repolymerization with microscopy, S Bates for culturing cells, W Tsark and F of microtubules, coverslip cultures of COS-7 cells transfected Silva for ES cell electroporation and blastocyst injection, L with EGFP-RASSF1A were incubated with 2 mm nocodazole Brown and C Spalla for FACS analysis, S-G Jin for plasmid for 20 h. Cells were incubated at 371C for 5 min after removal construction, and J Shively, CJ Chen, and CL Jiang for of nocodazole and then were processed for immunofluores- instructive discussions. This work was supported by NIH cence microscopy as described above. Grant CA88873(to GPP) and by BMBF Grant FKZ01ZZ0104 (to RD).

References

Abbondanzo SJ, Gadi I and Stewart CL. (1993). Methods Protopopov A, Li J, Kashuba V, Klein G and Zabarovsky Enzymol., 225, 803–823. ER. (2001). Proc. Natl. Acad. Sci. USA, 98, 7504–7509. Agathanggelou A, Honorio S, Macartney DP, Martinez A, Duensing S and Mu¨ nger K. (2002). Oncogene, 21, Dallol A, Rader J, Fullwood P, Chauhan A, Walker R, 6241–6248. Shaw JA, Hosoe S, Lerman MI, Minna JD, Maher ER and Hergovich A, Lisztwan J, Barry R, Ballschmieter P and Krek Latif F. (2001). Oncogene, 20, 1509–1518. W. (2003). Nat. Cell Biol., 5, 64–70. Anand S, Penrhyn-Lowe S and Venkitaraman AR. (2003). Honorio S, Agathanggelou A, Wernert N, Rothe M, Maher Cancer Cell, 3, 51–62. ER and Latif F. (2003). Oncogene, 22, 461–466. Burbee DG, Forgacs E, Zochbauer-Muller S, Shivakumar L, Hsu LC and White RL. (1998). Proc. Natl. Acad. Sci. USA, 95, Fong K, Gao B, Randle D, Kondo M, Virmani A, Bader S, 12983–12988. Sekido Y, Latif F, Milchgrub S, Toyooka S, Gazdar AF, Jordan MA, Thrower D and Wilson L. (1992). J. Cell Sci., 102 Lerman MI, Zabarovsky E, White M and Minna JD. (2001). (Part 3), 401–416. J. Natl. Cancer Inst., 93, 691–699. Jordan MA, Toso RJ, Thrower D and Wilson L. (1993). Proc. Byun DS, Lee MG, Chae KS, Ryu BG and Chi SG. (2001). Natl. Acad. Sci. USA, 90, 9552–9556. Cancer Res., 61, 7034–7038. Jordan MA, Wendell K, Gardiner S, Derry WB, Copp H and Cahill DP, Lengauer C, Yu J, Riggins GJ, Willson JK, Wilson L. (1996). Cancer Res., 56, 816–825. Markowitz SD, Kinzler KW and Vogelstein B. (1998). Khokhlatchev A, Rabizadeh S, Xavier R, Nedwidek M, Chen Nature, 392, 300–303. T, Zhang XF, Seed B and Avruch J. (2002). Curr. Biol., 12, Chan MW, Chan LW, Tang NL, Lo KW, Tong JH, Chan 253–265. AW, Cheung HY, Wong WS, Chan PS, Lai FM and To KF. Kim ST, Lim DS, Canman CE and Kastan MB. (1999). (2003). Int. J. Cancer, 104, 611–616. J. Biol. Chem., 274, 37538–37543. Chen Z, Otto JC, Bergo MO, Young SG and Casey PJ. (2000). Kok K, Naylor SL and Buys CH. (1997). Adv. Cancer Res., 71, J. Biol. Chem., 275, 41251–41257. 27–92. Dammann R, Li C, Yoon JH, Chin PL, Bates S and Pfeifer Kuzmin I, Gillespie JW, Protopopov A, Geil L, Dreijerink K, GP. (2000). Nat. Genet., 25, 315–319. Yang Y, Vocke CD, Duh FM, Zabarovsky E, Minna JD, Dammann R, Schagdarsurengin U, Strunnikova M, Rastetter Rhim JS, Emmert-Buck MR, Linehan WM and Lerman MI. M, Seidel C, Liu L, Tommasi S and Pfeifer GP. (2003). (2002). Cancer Res., 62, 3498–3502. Histol. Histopathol., 18, 665–677. Kuzmin I, Liu L, Dammann R, Geil L, Stanbridge EJ, Dammann R, Takahashi T and Pfeifer GP. (2001a). Oncogene, Wilczynski SP, Lerman MI and Pfeifer GP. (2003). Cancer 20, 3563–3567. Res., 63, 1888–1893. Dammann R, Yang G and Pfeifer GP. (2001b). Cancer Res., Lee MG, Kim HY, Byun DS, Lee SJ, Lee CH, Kim JI, Chang 61, 3105–3109. SG and Chi SG. (2001). Cancer Res., 61, 6688–6692. Dreijerink K, Braga E, Kuzmin I, Geil L, Duh FM, Angeloni Lerman MI and Minna JD. (2000). Cancer Res., 60, D, Zbar B, Lerman MI, Stanbridge EJ, Minna JD, 6116–6133.

Oncogene Microtubule stabilization by RASSF1A LLiuet al 8136 Li YC, Chen CR and Chang EC. (2000). Genetics, 156, Schagdarsurengin U, Gimm O, Hoang-Vu C, Dralle H, Pfeifer 995–1004. GP and Dammann R. (2002). Cancer Res., 62, 3698–3701. Liu L, Yoon JH, Dammann R and Pfeifer GP. (2002). Schagdarsurengin U, Wilkens L, Steinemann D, Flemming P, Oncogene, 21, 6835–6840. Kreipe HH, Pfeifer GP, Schlegelberger B and Dammann R. Lo KW, Kwong J, Hui AB, Chan SY, To KF, Chan AS, Chow (2003). Oncogene, 22, 1866–1871. LS, Teo PM, Johnson PJ and Huang DP. (2001). Cancer Segal M and Clarke DJ. (2001). BioEssays, 23, 307–310. Res., 61, 3877–3881. Sekido Y, Ahmadian M, Wistuba II, Latif F, Bader S, Wei Lotti LV, Ottini L, D’Amico C, Gradini R, Cama A, Belleudi MH, Duh FM, Gazda AF, Lerman MI and Minna JD. F, Frati L, Torrisi MR and Mariani-Costantini R. (2002). (1998). Oncogene, 16, 3151–3157. Genes Chromosomes Cancer, 35, 193–203. Shivakumar L, Minna J, Sakamaki T, Pestell R and White Lusher ME, Lindsey JC, Latif F, Pearson AD, Ellison DW MA. (2002). Mol. Cell. Biol., 22, 4309–4318. and Clifford SC. (2002). Cancer Res., 62, 5906–5911. Spugnardi M, Tommasi S, Dammann R, Pfeifer GP and Hoon Maruyama R, Toyooka S, Toyooka KO, Harada K, Virmani DS. (2003). Cancer Res., 63, 1639–1643. AK, Zochbauer-Muller S, Farinas AJ, Vakar-Lopez F, Tommasi S, Dammann R, Jin SG, Zhang XF, Avruch J and Minna JD, Sagalowsky A, Czerniak B and Gazdar AF. Pfeifer GP. (2002). Oncogene, 21, 2713–2720. (2001). Cancer Res., 61, 8659–8663. Toyooka S, Carbone M, Toyooka KO, Bocchetta M, Michel LS, Liberal V, Chatterjee A, Kirchwegger R, Pasche B, Shivapurkar N, Minna JD and Gazdar AF. (2002). Gerald W, Dobles M, Sorger PK, Murty VV and Benezra R. Oncogene, 21, 4340–4344. (2001). Nature, 409, 355–359. van Engeland M, Roemen GM, Brink M, Pachen MM, Morrissey C, Martinez A, Zatyka M, Agathanggelou A, Weijenberg MP, de Bruine AP, Arends JW, van den Brandt Honorio S, Astuti D, Morgan NV, Moch H, Richards FM, PA, de Goeij AF and Herman JG. (2002). Oncogene, 21, Kishida T, Yao M, Schraml P, Latif F and Maher ER. 3792–3795. (2001). Cancer Res., 61, 7277–7281. Vavvas D, Li X, Avruch J and Zhang XF. (1998). J. Biol. Newton AC. (1995). Curr. Biol., 5, 973–976. Chem., 273, 5439–5442. Ortiz-Vega S, Khokhlatchev A, Nedwidek M, Zhang XF, Wagner KJ, Cooper WN, Grundy RG, Caldwell G, Jones C, Dammann R, Pfeifer GP and Avruch J. (2002). Oncogene, Wadey RB, Morton D, Schoﬁeld PN, Reik W, Latif F and 21, 1381–1390. Maher ER. (2002). Oncogene, 21, 7277–7282. Pfeifer GP, Yoon JH, Liu L, Tommasi S, Wilczynski SP and Zhang YJ, Ahsan H, Chen Y, Lunn RM, Wang LY, Chen SY, Dammann R. (2002). Biol. Chem., 383, 907–914. Lee PH, Chen CJ and Santella RM. (2002). Mol. Carcinog., Ponting CP and Benjamin DR. (1996). Trends Biochem. Sci., 35, 85–92. 21, 422–425. Zumbrunn J, Kinoshita K, Hyman AA and Nathke IS. (2001). Rieder CL, Cole RW, Khodjakov A and Sluder G. (1995). Curr. Biol., 11, 44–49. J. Cell Biol., 130, 941–948.

Oncogene