Oncogene (2008) 27, 6175–6186 & 2008 Macmillan Publishers Limited All rights reserved 0950-9232/08 $32.00 www.nature.com/onc ORIGINAL ARTICLE RASSF1A interacts with and activates the mitotic Aurora-A

L Liu1, C Guo1, R Dammann2, S Tommasi1 and GP Pfeifer1

1Division of Biology, Beckman Research Institute, City of Hope Center, Duarte, CA, USA and 2Institute of Genetics, University of Giessen, Giessen, Germany

The RAS association domain family 1A (RASSF1A) tumorigenesis and carcinogen-induced tumorigenesis is located at 3p21.3 within a specific area of (Tommasi et al., 2005; van der Weyden et al., 2005), common heterozygous and homozygous deletions. RASS- supporting the notion that RASSF1A is a bona fide F1A frequently undergoes promoter methylation-asso- tumor suppressor. However, it is not fully understood ciated inactivation in human . Rassf1aÀ/À mice how RASSF1A is involved in tumor suppression. are prone to both spontaneous and carcinogen-induced The biochemical function of the RASSF1A is tumorigenesis, supporting the notion that RASSF1A is a largely unknown. The homology of RASSF1A with the tumor suppressor. However, it is not fully understood how mammalian Ras effector novel Ras effector (NORE)1 RASSF1A is involved in tumor suppression pathways. suggests that the RASSF1A gene product may function Here we show that overexpression of RASSF1A inhibits in signal transduction pathways involving Ras-like separation. RASSF1A interacts with Aurora-A, . However, recent data indicate that RASSF1A a mitotic kinase. Surprisingly, knockdown of RASS- itself binds to RAS only weakly and that binding to F1A by siRNA led to reduced activation of Aurora-A, RAS may require heterodimerization of RASSF1A and whereas overexpression of RASSF1A resulted in in- NORE1 (Ortiz-Vega et al., 2002). In addition, there is creased activation of Aurora-A, suggesting that RASS- evidence for an association of both NORE1 and F1A is involved in Aurora-A activation. Like other RASSF1A with the pro-apoptotic kinase MST1 and Aurora-A activators, RASSF1A was also a substrate of that this interaction is involved in apoptosis induced by Aurora-A in vitro. The failure of recombinant RASSF1A activated Ras (Khokhlatchev et al., 2002). to activate recombinant Aurora-A indicates that RASS- Recently, several groups, including ours, have re- F1A may not activate Aurora-A directly and suggests that ported that RASSF1A is a - and centro- RASSF1A may function as a scaffold to bring together some-associated protein, which regulates mitotic Aurora-A and its activator(s). Inhibition of centrosome progression (Liu et al., 2003, 2005c; Dallol et al., 2004; separation by RASSF1A overexpression is most likely a Rong et al., 2004; Song et al., 2004; Vos et al., 2004). We consequence of hyperstabilization of by this have shown that overexpressed RASSF1A co-localized protein. with microtubules in interphase cells and with spindles Oncogene (2008) 27, 6175–6186; doi:10.1038/onc.2008.220; and during (Liu et al., 2003). published online 21 July 2008 Forced expression of RASSF1A induced mitotic arrest with aberrant mitotic cells (Liu et al., 2003). Similar Keywords: RASSF1A; Aurora-A; centrosome; mitosis results been have been obtained by several other groups (Dallol et al., 2004; Rong et al., 2004; Song et al., 2004; Vos et al., 2004). Song et al. (2004) have reported that RASSF1A regulated mitosis by inhibiting the - Introduction promoting complex (APC) through binding to CDC20 and induces G2–M arrest at pro-. The The RAS association domain family 1A (RASSF1A) function of RASSF1A was independent of the protein gene is located at chromosome 3p21.3 within a specific early mitotic inhibitor 1, and therefore Song et al. (2004) area of common heterozygous and homozygous dele- proposed that RASSF1A acted in early tions (Dammann et al., 2000; Lerman and Minna, 2000). to prevent the degradation of mitotic cyclins and to The RASSF1A gene is one of the most frequently delay mitotic progression. However, our laboratory has methylation-silenced described in human cancer been unable to confirm an interaction between RASS- to date (Agathanggelou et al., 2005; Dammann et al., F1A and CDC20 (Liu et al., 2007). RASSF1A may 2005). Rassf1aÀ/À mice are prone to both spontaneous function as a tumor suppressor through promoting apoptosis and/or by controlling mitotic division (Agathanggelou et al., 2005; Dammann et al., 2005). Correspondence:Dr GP Pfeifer, Division of Biology, Beckman However, its exact mechanism of action remains to be Research Institute, City of Hope Cancer Center, City of Hope, 1500 determined. East Duarte Road, Duarte, CA 91010, USA. E-mail:[email protected] Aurora-A belongs to a family of highly conserved Received 14 April 2008; revised 4 June 2008; accepted 10 June 2008; /threonine , which includes Aurora-A, -B published online 21 July 2008 and -C in mammals. Aurora kinases combined with RASSF1A interacts with and activates Aurora-A LLiuet al 6176 cyclin-dependent kinase 1 (or p34cdc2), Polo-like kinases 2005; Weaver and Cleveland, 2006). Being a putative and NIMA-related kinases are four families of known oncogene, Aurora-A has attracted a great deal of mitotic kinases that regulate mitotic progression. All attention as a potential antitumor target (Keen and three members of the family share a Taylor, 2004; Mortlock et al., 2005; Naruganahalli conserved C-terminal catalytic domain and have been et al., 2006). found to have multiple mitotic functions (Andrews Here we demonstrate that endogenous RASSF1A et al., 2003; Carmena and Earnshaw, 2003). Among interacts with and activates Aurora-A. them, Aurora-A is the most intensively studied one. Originally Aurora-A was identified in a screen for melanogaster mutants that have abnormal spindle-pole phenotypes (Glover et al., 1995). Aurora-A Results is located on centrosomes from late S/early G to the 2 RASSF1A regulates centrosome separation time when the cells exit mitosis (Carmena and Earn- shaw, 2003). Aurora-A is important in centrosome RASSF1A is a microtubule- and centrosome-associated maturation and separation as well as in spindle assembly protein in interphase cells and is found at the spindle et al (Andrews et al., 2003; Carmena and Earnshaw, 2003; poles and midzone region in mitotic cells (Liu ., et al et al Marumoto et al., 2005). Moreover, activity of Aurora-A 2003, 2005a, c; Dallol ., 2004; Rong ., 2004; Vos et al et al is required for timely mitotic entry (Marumoto et al., ., 2004; van der Weyden ., 2005). Endogenous 2002, 2003; Hirota et al., 2003; Hachet et al., 2007; RASSF1A is primarily a centrosomal protein in et al Portier et al., 2007). Both upregulation and down- interphase cells (Guo ., 2007) and it colocalizes regulation of Aurora-A results in failure with mitotic spindles in mitosis (Figure 1a). Using (Meraldi et al., 2002; Anand et al., 2003; Marumoto EGFP-tagged deletion fragments of RASSF1A, we et al., 2003; Zhang et al., 2004), indicating that the determined that the centrosome and spindle pole timing of Aurora-A activity is important for proper localization of RASSF1A requires a minimal domain cytokinesis. fragment from amino acid D120 to G193 (Figure 1b). Expression and localization of Aurora-A is cell-cycle This domain is distinct from the RAS association regulated. Its protein level peaks in the middle of mitosis domain present in RASSF1A (R194–S288) and only and is diminished in late mitosis/early G by the Cdh1- partially overlaps with the minimum microtubule- 1 et al activated APC/C (APC/cyclosome) (Honda et al., 2000; binding domain (D120-S288) (Liu ., 2003). Castro et al., 2002; Taguchi et al., 2002). The activity of Previously, we found that overexpression of RASS- Aurora-A is also controlled by phosphorylation. Phos- F1A resulted in over 50% of mitoses with monopolar et al phorylation of Thr288 of human Aurora-A is essential spindles (Liu ., 2003). Generally, monopolar for its kinase activity (Walter et al., 2000; Littlepage spindles may result from defects in centrosome duplica- et al., 2002). Several activators of Aurora-A have been tion or from failure of centrosome separation. When identified, which include TPX2, Ajuba, HEF1, Bora and these cells were stained with an anti-g- , PAK1 (Kufer et al., 2002; Hirota et al., 2003; Crane we observed that there were two closely spaced g-tubulin et al., 2004; Satinover et al., 2004; Pugacheva and spots in 96% of the monopolar spindle containing cells, Golemis, 2005; Zhao et al., 2005; Hutterer et al., 2006). which indicates that the centrosomes have duplicated Activation of Aurora-A by TPX2 is required for but were not separated far enough to organize a bipolar bipolar-spindle assembly (Trieselmann et al., 2003; Tsai spindle (Figure 2). et al., 2003), whereas activation of Aurora-A by Ajuba is Microtubules are important in centrosome separation et al et al et al essential for mitotic entry (Hirota et al., 2003). It is (Waters ., 1993; Gaglio ., 1996; Uzbekov ., tempting to propose that Aurora-A activators may at 2002). Therefore, we speculated that inhibition of least partially control the temporal and spatial distribu- microtubule dynamics by RASSF1A may at least in tion of Aurora-A activity along the cell division cycle. part account for its effect on centrosome separation. Aurora-A is frequently overexpressed in various Indeed, the monopolar spindles were hyperstabilized human tumors (Bischoff et al., 1998; Zhou et al., and resistant to nocodazole-induced depolymerization 1998). Overexpression of Aurora-A was sufficient to (Figure 2). transform NIH3T3 and Rat1 cells (Bischoff et al., 1998; Zhou et al., 1998). However, the same regimen did not RASSF1A interacts with Aurora-A transform primary murine embryonic fibroblasts We asked if there are other mechanisms by which (MEFs), suggesting that upregulation of Aurora-A itself RASSF1A may regulate centrosome separation. It is is insufficient for oncogenesis (Anand et al., 2003). It is known that Eg5 and Aurora-A are required for still unclear whether overexpression of Aurora-A centrosome separation (Blangy et al., 1995; Glover induces mammary tumors in transgenic mouse models et al., 1995; Gaglio et al., 1996). We reasoned that (Zhang et al., 2004; Wang et al., 2006). Being a proteins, which can affect Eg5 or/and Aurora-A multifunctional protein, it is not known which function function, may affect centrosome separation. To test if of Aurora-A contributes to its oncogenic activity. RASSF1A interacts with Aurora-A and/or Eg5, we Current thought is that overexpression induces centro- transfected GST-RASSF1A or GST-only into COS-7 some defects, mitotic abnormalities and , cells. Aurora-A but not Eg5 was detected in the pull- which leads to cancer (Meraldi et al., 2004; Giet et al., down product from GST-RASSF1A-expressing cells

Oncogene RASSF1A interacts with and activates Aurora-A LLiuet al 6177

Figure 1 Centrosome association of RAS association domain family 1A (RASSF1A). (a) Endogenous RASSF1A localizes to centrosomes and their associated mitotic spindles in GM004427F human fibroblasts. GM004427F cells were co-stained with anti-RASSF1A antibody, anti-g-tubulin antibody and 4,6-diamidino-2-phenylindole (DAPI). Although RASSF1 predominantly colocalizes with centrosome in interphase cells, it resides in both centrosome and mitotic spindle in mitosis. (b) Centrosome association domain mapping of RASSF1. COS-7 cells were transfected with the indicated EGFP-tagged RASSF1 deletion constructs and the subcellular localization of theses mutants was visualized using florescence microscopy.

(Figure 3a), indicating an interaction between RASS- Aurora-A directly (Figure 3b). Surprisingly, MBP- F1A and Aurora-A. To rule out the possibility that tagged recombinant RASSF1C can only bind to His- this interaction might be an artifact of RASSF1A tagged Aurora-A weakly (Figure 3b). As RASSF1A and overexpression, we performed a co-immunoprecipitation RASSF1C only differ in their N termini, we reasoned assay using anti-RASSF1A . Anti-RASSF1A that either the RASSF1A N terminus is required to bind antibodies, but not the control IgG, co-immunoprecipi- to Aurora-A or that the RASSF1C N terminus prevents tated endogenous Aurora-A (Figure 3c). To determine RASSF1C from binding to Aurora-A. The C-terminal whether RASSF1A interacts with Aurora-A directly or catalytic domain of Aurora-A was both sufficient and indirectly, we carried out an in vitro binding assay. required for Aurora-A to bind to RASSF1A Maltose-binding protein (MBP)-tagged recombinant (Figure 4a). As this catalytic domain is well conserved RASSF1A but not MBP alone can bind to His-tagged among Aurora-A family members, we propose that

Oncogene RASSF1A interacts with and activates Aurora-A LLiuet al 6178

Figure 2 Monopolar spindles induced by overexpression of RAS association domain family 1A (RASSF1A) or its deletion mutants are resistant to nocodazole-induced depolymerization. Top row, monopolar spindles induced by overexpression of RASSF1A. Centrosomes are duplicated but not separated. Middle and bottom rows, COS-7 cells transfected with EGFP-RASSF1A (middle) or the EGFP-RASSF1A 120-288 deletion mutant (bottom) were incubated with 20 mM nocodazole for 1 h. Cells were double stained with anti-a-tubulin antibody (red) and 4,6-diamidino-2-phenylindole (DAPI) (blue).

RASSF1A may also interact with Aurora-B and RASSF1A could inhibit Aurora-A. First, RASSF1A Aurora-C. Indeed, we found that both Aurora-B and may repress Aurora-A expression or promote Aurora-A Aurora-C could bind to RASSF1A when they were degradation. To test this possibility we quantitated overexpressed in COS-7 cells (Figures 4b and c). Aurora-A expression in both Rassf1A þ / þ MEFs and To further explore the interaction between RASSF1A Rassf1AÀ/À MEFs using western blot. We did not find and Aurora-A, we co-stained GM004427F human any significant difference in terms of Aurora-A expres- fibroblasts with an anti-RASSF1A antibody and an sion at the total protein level although we did notice a anti-Aurora-A antibody. In interphase cells, RASSF1A slight reduction of the slower migrating (presumable locates predominantly at centrosomes, but Aurora-A phosphorylated) protein band in Rassf1AÀ/À MEFs after largely resides in the cytoplasm. When cells progress nocodazole treatment (Supplementary Figure 1). Sec- into , Aurora-A starts to accumulate on ond, RASSF1A may regulate localization of Aurora-A. centrosomes where RASSF1A stays (Figure 5). In Aurora-A was not mislocalized when RASSF1A was , both RASSF1A and Aurora-A were found absent in Rassf1aÀ/À MEFs (Supplementary Figure 1). on centrosomes (spindle poles) as well as the associated Third, RASSF1A may inhibit Aurora-A kinase activity spindles. In anaphase and telpophase, both RASSF1A directly. In an in vitro kinase assay, we did not detect and Aurora-A colocalized on the centrosome and any inhibitory effect of RASSF1A on Aurora-A midzone region (Figure 5). The same mitotic stage- (Supplementary Figure 2), though recombinant RASS- specific co-localization pattern was observed when F1A is a substrate of Aurora-A in vitro (Supplementary EGFP-tagged RASSF1A and endogenous Aurora-A Figure 2) consistent with a recent report (Rong et al., were co-localized (data not shown). 2007). Fourth, RASSF1A may inhibit Aurora-A activa- tion that subsequently leads to negative regulation of Aurora-A. To test this, we knocked down RASSF1A by RASSF1A regulates Aurora-A activation siRNA in HeLa cells. Consistent with what we found in As overexpression of RASSF1A and inactivation of RASSF1A-deficient MEFs, downregulation of RASS- Aurora-A leads to the same phenotype, known as F1A in HeLa cells did not change the total protein level monopolar spindle mitoses, we hypothesized that of Aurora-A (Figure 6, compare lanes 8, 9 or 10 with RASSF1A negatively regulates Aurora-A function. lanes 13 or 14). Given that phosphorylation of Aurora- There are at least four possible ways by which A on Thr288 leads to activation of Aurora-A, we

Oncogene RASSF1A interacts with and activates Aurora-A LLiuet al 6179

pulldown 1/10 input

BP-RASSF1A BP-RA1-119BP-Paramyosin M MBP-RASSF1CM M

GST GST-RASSF1AGST-RASSF1CGST GST-RASSF1AGST-RASSF1C Anti-MBP

anti-GST Anti-Aurora-A

IP anti-Eg5

input 5-05 anti-Aurora-A 7% M3-04 M IgG

Anti-Aurora-A

Anti-RASSF1A

Figure 3 Aurora-A interacts with RAS association domain family 1A (RASSF1A) both in vitro and in vivo.(a) 293T cells were transfected with GST-RASSF1A, GST-RASSF1C or glutathione S- (GST). Total cellular extracts were incubated with glutathione sepharose 4B gel, and bound proteins were analysed by western blot using antibodies as indicated. (b) RASSF1A and Aurora-A interact in vitro. Purified maltose-binding protein (MBP)-RASSF1A and His-Aurora-A were incubated in a binding buffer at 4 1C for 1 h. The mixture was adsorbed onto amylose resins. The bound proteins were analysed by western blot using antibodies as indicated. (c) Aurora-A interacts with RASSF1A in vivo. Lysates from mitotic HeLa cells were incubated with two anti-RASSSF1A antibodies (M3-04, M5-05) or purified rabbit IgG. The immunocomplexes were precipitated by protein A/G agarose. The precipitated proteins were analysed by western blot.

3Myc-Aurora-A 131-403 --+ + IP: Flag Lysate 3Myc-Aurora-A 1-130 ++- - 3HA-Aurora-B ++++++ 3Flag-pcDNA3.0 -+-+ 3Flag-pcDNA3.0 --+--+ 3Flag-RASSF1A +-+- 3Flag-RASSF1C -+--+- 3Flag-RASSF1A +--+-- IP: Flag WB: Flag WB: HA

3Myc-Aurora-A WB: Flag IP: Flag 131-403 WB: Myc IP: Flag Lysate 3HA-Aurora-C ++++++ 3Flag-RASSF1A Lysate 3Flag-pcDNA3.0 --+--+ 3Flag-RASSF1C -+--+- WB: Flag &Myc 3Myc-Aurora-A 131-403 3Flag-RASSF1A +--+--

3M yc-Aurora-A WB: HA 1-130

WB: Flag

Figure 4 Interaction of RAS association domain family 1A (RASSF1A) with Aurora kinases. (a) RASSF1A associates with the C-terminal catalytic domain of Aurora-A. COS-7 cells were co-transfected with Flag-RASSF1A and Aurora-A deletion mutants tagged with c-Myc. Total cellular extracts were incubated with anti-Flag agarose, and bound proteins were analysed by western blot using antibodies as indicated. (b) RASSF1A interacts with Aurora-B. COS-7 cells were co-transfected with HA-Aurora-B and RASSFA or RASSF1C tagged with Flag. Total cellular extracts were incubated with anti-Flag agarose, and bound proteins were analysed by western blot using antibodies as indicated. (c) RASSF1A interacts with Aurora-C. COS-7 cells were co-transfected with HA-Aurora-C and RASSFA or RASSF1C tagged with Flag. Total cellular extracts were incubated with anti-Flag agarose, and bound proteins were analysed by western blot using antibodies as indicated.

Oncogene RASSF1A interacts with and activates Aurora-A LLiuet al 6180

Figure 5 Endogenous RAS association domain family 1A (RASSF1A) colocalizes with Aurora-A in mitosis. GM004427F, SV-40 immortalized human fibroblasts, were fixed and co-stained with an anti-Aurora-A antibody and an anti-RASSF1A antibody. DNA was counterstained with 4,6-diamidino-2-phenylindole (DAPI). Cells at each mitotic stage are as indicated.

examined the phosphorylation status of Thr288 by an tion change in the randomly growing cells (Figure 6, antibody specific for phosphorylated Thr288 of human compare lanes 1, 2, or 3 with lanes 6 or 7). To rule out Aurora-A. Surprisingly, downregulation of RASSF1A the possibility that cell-cycle distribution may be resulted in a marked decrease of Aurora-A phospho- involved in this decreased phosphorylation of Aurora- rylation on Thr288 (Figure 6, compare lanes 8, 9 or 10 A, we probed the same membrane with an antibody with lanes 13 or 14) when the cells are synchronized to against a mitotic marker, H3 phosphorylated on mitosis by nocodazole. As Aurora-A expression is Ser10. We did not detect any significant difference of diminished in interphase cells and Aurora-A is solely Ser10 phosphorylation of histone H3 between RASS- phosphorylated and activated in mitosis, we did not F1A siRNA-treated cells and control LacZ siRNA- expect to detect any significant Aurora-A phosphoryla- treated cells (Figure 6, compare lanes 8, 9 or 10 with

Oncogene RASSF1A interacts with and activates Aurora-A LLiuet al 6181 Untreated Nocodazole

725

RA196 RA1404 RA1430 AuA150 AuA LacZ untransfectedRA196 RA1404 RA1430 AuA150 AuA725 LacZ untransfected

RASSF1A

Aurora-A

P-Aurora-A (Thr288 )

P-H3 (Ser10)

Aurora-B

Tubulin

12345678 9 1011121314 Figure 6 RAS association domain family 1A (RASSF1A) is an activator of Aurora-A. Downregulation of RASSF1A by siRNA in HeLa cells inhibits Aurora-A activation. HeLa cells were transfected with siRNAs targeting RASSF1A (RA196, RA1404 and RA1430) and Aurora-A (AuA150 and AuA725), respectively. Cells were either unsynchronized or synchronized at mitosis by nocodazole (200 nM, 20 h) treatment. Cell lysates were analysed by western blot with the indicated antibodies. Untransfected HeLa cells and HeLa cells transfected with siRNA targeting the LacZ gene were used as controls. lanes 13 or 14), indicating that cell-cycle distribution did not contribute to the reduced phosphorylation of Aurora-A in RASSF1A siRNA-treated cells. To further investigate the role of RASSF1A in Aurora-A activation, we overexpressed either RASSF1A or RASSF1C in HeLa cells. Overexpression of RASSF1A promoted Aurora-A activation, as indicated by an increase of phosphorylation of Aurora-A on Thr288 in nocoda- zole-treated cells (Figure 7). However, such an increase was not proportionally dependent on RASSF1A expres- sion level. For instance, transfection of 1 mg RASSF1A plasmid had a greater effect on Aurora-A phosphorylation than transfection of 2 or 4 mg RASSF1A. One possible explanation would be that at high expression levels, RASSF1A may act differently compared to low RASS- F1A levels. For example, overexpression of RASSF1A stabilizes microtubules only when RASSF1A is expressed at low to moderate levels but not when it is expressed at Figure 7 Effect of RAS association domain family 1A (RASS- high levels (L Liu and GP Pfeifer, unpublished observa- F1A) overexpression on Aurora-A activation. Overexpression of tion). Further investigation is warranted to reveal whether RASSF1A promotes phosphorylation of Aurora-A. HeLa cells RASSF1A’s effect on microtubules has any relationship were transfected with Flag-RASSF1A, Flag-RASSF1C or Flag- only plasmids. Cells were either unsynchronized or synchronized at with RASSF1A’s effect on Aurora-A activation. Strik- mitosis by nocodazole (200 nM, 20 h) treatment. Cell lysates were ingly, overexpression of RASSF1C did not have an effect analysed by western blot with the indicated antibodies. on Aurora-A activation (Figure 7), indicating that RASSF1A and RASSF1C may play opposite roles in Aurora-A activation. Indeed, overexpression of RASSF1C Although downregulation of RASSF1A in HeLa cells inhibited Aurora-A phosphorylation compared to control did not change the protein level of Aurora-A (Figure 6), (Figure 7). Therefore, a balance between RASSF1A and downregulation of Aurora-A by siRNA did decrease RASSF1C might be important for Aurora-A regulation. RASSF1A protein levels, especially in synchronized In cancer cells, such a balance is disrupted by loss of mitotic HeLa cells (Figure 6), suggesting that Aurora-A RASSF1A expression. might regulate RASSF1A expression or stability.

Oncogene RASSF1A interacts with and activates Aurora-A LLiuet al 6182 However, the mechanism of this regulation remains 2005; van der Weyden et al., 2005), it is unlikely that unknown. RASSF1A has an essential role in cell-cycle control. Instead RASSF1A may fine-tune cell-cycle progression or its role is redundant with that of other proteins, for example, its closest homologue NORE1A. Cytokinesis Discussion failure occurs more frequently in Rassf1aÀ/À MEFs than in wild-type MEFs (Guo et al., 2007). The mechanistic The centrosome was first described by Boveri (1888). basis for cytokinesis failure in Rassf1aÀ/À MEFs is not Over the past 100 years centrosomes have been known clear although it is at least in part a consequence of lack as microtubule-organizing centers that nucleate micro- of activation of the mitotic exit kinase LATS1 (Guo tubule arrays in animal cells. Not until recently, we et al., 2007). As Aurora-B, which is required for started to appreciate that centrosomes function beyond cytokinesis (Giet and Glover, 2001; Guse et al., 2005), being structural scaffolds for microtubules. Instead, interacts with RASSF1A (Figure 4b), it is possible that centrosomes are the sites that anchor and integrate loss of RASSF1A impedes Aurora-B function. Our various regulatory activities, which are required for cell- failure to detect an endogenous interaction between cycle control (Doxsey, 2001a, b; Rieder et al., 2001; RASSF1A and Aurora-B (data not shown) might Doxsey et al., 2005). For instance, the centrosome is suggest that the abundance of such an interaction is required for both G1–S cell-cycle progression (Hinch- low or that this interaction is transient in nature. cliffe et al., 2001) and cytokinesis (Piel et al., 2001), Further investigation is required to dissect the func- whereas centrosomal localization of Aurora-A is essen- tional significance of the Aurora-B-RASSF1A inter- tial for mitotic entry (Hachet et al., 2007; Portier et al., action, which might shed light on how RASSF1A is 2007). involved in cytokinesis. Although it has been well established that over- It came as a surprise that RASSF1A is an activator of expressed RASSF1A colocalizes with microtubules in Aurora-A. RASSF1A has been proven to be a tumor interphase (Liu et al., 2003, 2005a, b; Dallol et al., 2004; suppressor (Tommasi et al., 2005; van der Weyden et al., Rong et al., 2004; Vos et al., 2004; van der Weyden 2005), whereas Aurora-A fits the criteria to be classified et al., 2005), the endogenous RASSF1A protein is as an oncoprotein (Bischoff et al., 1998; Zhou et al., primarily a centrosome-associated protein in interphase 1998; Ewart-Toland et al., 2003, 2005). This calls into cells. One simple explanation for such a discrepancy is question whether activation of Aurora-A by RASSF1A that overexpression of RASSF1A saturates its binding contributes to RASSF1A’s tumor suppressive function. sites on the centrosome; therefore excessive cytoplasmic However, it has been reported that both downregulation RASSF1A is forced to bind to cytoplasmic microtu- and upregulation of Aurora-A leads to cytokinesis bules. Consistent with such a notion, we found that defects (Meraldi et al., 2002; Anand et al., 2003; when RASSF1A expression was low in an inducible Marumoto et al., 2003; Zhang et al., 2004), suggesting system, it colocalized with centrosomes but not micro- that not only the level of expression but also the proper tubules in interphase cells (data not shown). It has been timing of Aurora-A activation and inactivation is reported that Aurora-A can phosphorylate RASSF1A important for cytokinesis. Activation and localization (Rong et al., 2007) (see also Supplementary Figure 2). of Aurora-A is tightly regulated. The temporal and This phosphorylation might modulate RASSF1A’s spatial distribution of Aurora-A activity is at least ability to bind to microtubules, as a phosphorylation- partially controlled by Aurora-A activators. For in- mimicking mutant of RASSF1A could not interact with stance, activation of Aurora-A by TPX2 is required for microtubules, whereas a phosphorylation-defective mu- bipolar-spindle assembly (Trieselmann et al., 2003; Tsai tant could (Rong et al., 2007). However, endogenous et al., 2003), whereas activation of Aurora-A by Ajuba is RASSF1A localizes predominantly at centrosomes but essential for mitotic entry (Hirota et al., 2003). Activa- not microtubules in interphase cells (Guo et al., 2007; tion of Aurora-A by HEF-1 in a nonmitotic pathway is Figure 1). As Aurora-A is inactivate in G1 (Honda et al., mandatory for the disassembly of the primary 2000; Castro et al., 2002; Taguchi et al., 2002) and (Pugacheva et al., 2007). RASSF1A is a residential expression of RASSF1A is constant throughout the cell centrosomal protein, whereas Aurora-A is recruited to cycle (L Liu and GP Pfeifer, unpublished observation), centrosomes during G2/M phase. It is not clear when it is very likely that something other than phosphoryla- and how RASSF1A activates Aurora-A. Recombinant tion by Aurora-A prevents RASSF1A from binding to RASSF1A did not activate recombinant Aurora-A in microtubules in G1. This could be another kinase that vitro (Supplementary Figure 2), suggesting that RASS- phosphorylates RASSF1A in interphase and prevents it F1A itself is not sufficient to activate Aurora-A. It is from binding to microtubules but cannot phosphorylate possible that RASSF1A functions as a scaffold to bring an excess of overexpressed RASSF1A. By controlling together Aurora-A and its activator(s). On the other the dynamics of microtubules, RASSF1A might regulate hand, we cannot exclude the possibility that posttransla- centrosome separation, at least upon overexpression tional modifications of RASSF1A, which could not be (Figure 2). achieved in Escherichia coli, are required for RASSF1A The centrosomal localization of RASSF1A might to activate Aurora-A. implicate RASSF1A in cell-cycle regulation. As Rass- Loss of RASSF1A leads to a downregulation of f1a-deficient mice are viable and fertile (Tommasi et al., Aurora-A activity. To achieve suitable Aurora-A

Oncogene RASSF1A interacts with and activates Aurora-A LLiuet al 6183 activity for , cells may compensate the tions. The construct of GST-Lats2-79-118 was a kind gift from Aurora-A activation deficit in cells not expressing Dr H Nojima (Osaka University, Japan) (Toji et al., 2004). RASSF1A by gene amplification. This gene amplifica- GST-Lats2-79-118 protein was purified using glutathione tion may finally override the loss of RASSF1A effect sepharose 4B (Amersham, Piscataway, NJ, USA) according and result in Aurora-A overexpression and subsequent to the manufacturer’s instructions. tumorigenesis. Therefore, it will be of interest to determine whether Aurora-A amplification and over- Antibodies expression correlates with RASSF1A methylation silen- To generate polyclonal antibodies against RASSF1A, the cing in primary human tumors. Aurora-A expression polypeptides MBP-RASSF1A 1-119 and MBP-RASSF1A levels and gene amplification in the tumors developing in 120-340 were used to immunize rabbits. The antibodies were Rassf1aÀ/À mice will also need to be determined. immunopurified by passing through Affi-Gel 10 (Bio-Rad) conjugated with MBP-RASSF1A. Three antibodies were generated, named M3-04, M5-05 and M5-06, respectively. M3-04 is against MBP-RASSF1A 1-119 that is specific for Materials and methods RASSF1A (Guo et al., 2007), whereas M5-05 and M5-06 are against MBP-RASSF1A 120-340, which is shared by both the Cell culture and transfection RASSF1A and RASSF1C isoforms. Monoclonal antibody COS-7, HeLa and 293T cells were obtained from the ATCC against RASSF1A (clone 3F3) was from Abcam (Cambridge, and the GM004427F cell line, an SV-40 immortalized human MA, USA) (used in western blotting). skin fibroblast line was purchased from the Coriell Institute for Polyclonal antibodies against Aurora-A (BL469, BL656) Medical Research (Camden, NJ, USA). All cell lines were were purchased from Bethyl Laboratories (Montgomery, TX, cultured in high glucose Dulbecco’s modified Eagle’s medium USA). BL469 recognizes human Aurora-A, whereas BL656 with 10% fetal bovine serum. COS-7 and 293T cells were reacts with mouse Aurora-A. Mouse monoclonal anti-Aurora- transfected using FuGENE6 (Roche, Indianapolis, IN, USA) A antibody (clone 35C1) was obtained from Abcam and following the manufacturer’s instructions. We used previously phospho-specific polyclonal antibody against Aurora-A reported methods (Abbondanzo et al., 1993) to isolate and (Thr288) was from Cell Signaling Technology (Danvers, culture primary MEF derived from 13.5 day embryos from MA, USA). Rabbit polyclonal antibodies against Flag and Rassf1aÀ/À and Rassf1a þ / þ mice (Tommasi et al., 2005). c-Myc tags and mouse monoclonal antibodies against b- tubulin (clone TUB 2.1), g-tubulin (clone GTU-88) and b-actin Plasmid constructs (clone AC-15) were from Sigma. Anti-phospho-histone H3 To generate EGFP-, GST-, Flag- and MBP-tagged RASSF1A/ (Ser10) was from Upstate (Lake Placid, NY, USA). Anti-MBP RASSF1C constructs, the cDNA encoding human RASSF1A/ antibody was from New England Biolabs. Anti-HA polyclonal RASSF1C was subcloned into EGFP-C2 (Clontech, Palo antibody was from Bethyl Laboratories. Alto, CA, USA), pEGB (kindly provided by Dr J Avruch, Massachusetts General Hospital), Flag-pcDNA3.0 and Immunofluorescence microscopy pMAL-c2 Â (New England Biolabs (NEB), Ipswich, MA, USA) vectors. The truncated mutant RASSF1A sequences Cells were cultured in six-well plates containing coverslips were generated by a single-step PCR using EGFP-RASSF1A coated with poly-L-lysine (Sigma). After 24–48 h, the cells were 1 as template and cloned into EGFP-C2 (Clontech) and pMAL- fixed in À20 C methanol for 5 min. After incubation with 1% bovine serum albumin in phosphate-buffered saline (PBS) for c2 Â (NEB) vectors. Aurora-A constructs were kindly provided by Erich Nigg (Max Planck Institute for Biochem- 30 min, the cells were incubated with the indicated primary istry, Martinsried, Germany). The cDNAs of Aurora-B and antibodies and appropriate secondary antibodies (Invitrogen, Aurora-C were purchased from Open Biosystems (Huntsville, Carlsbad, CA, USA) sequentially. DNA was counterstained AL, USA). Each of them was cloned into HA-pcDNA3.0 to with 4,6-diamidino-2-phenylindole (DAPI; Sigma). Images generate HA-Aurora-A, HA-Aurora-B and HA-Aurora-C. were obtained with an Olympus IX81 automated inverted The truncated mutant Aurora-A sequences were generated by microscope equipped with a Spot RT Slider high-resolution a single-step PCR and cloned into the c-Myc-pcDNA3.0 cooled CCD color camera (Olympus; Melville, NY, USA). vector. Color images were processed using Photoshop 9.0 (Adobe Systems; San Jose, CA, USA) or ImagePro Plus (Media Cybernetics; Silver Spring, MD, USA). Protein purification The pMAL protein fusion and purification system (NEB) was used to purify MBP-tagged RASSF1A, RASSF1C and their Western blot and immunoprecipitation truncated mutants according to the manufacturer’s instruc- Cells were lysed in lysis buffer (10 mM Tris-HCl, pH 8.0, 1% tions. ER2508 (NEB), a protease-deficient strain of E. coli, was NP40, 150 mM NaCl and 1 mM EDTA), containing phospha- used to express the fusion proteins. The cells were lysed by tase inhibitors (20 mM b-glycerolphosphate, 50 mM sodium sonication in column buffer (20 mM Tris-HCl, pH 7.4, 200 mM fluoride and 1 mM Na3VO4) and protease inhibitor cocktail NaCl, 1 mM EDTA and 1 mM dithiothreitol (DTT)) containing (Roche). protease inhibitor cocktail (Roche) and the lysate was cleared For immunoprecipitation with anti-Flag antibody, the by centrifugation at 9000 g at 4 1C for 30 min. After incubating EZview Red Anti-Flag (M2) affinity gel (Sigma) was used the supernatant with amylose resin (NEB) for 2 h at 4 1C, the according to the manufacturer’s instruction. For immuno- mixture was loaded onto a chromatography column (Bio-Rad, precipitation with anti-RASSF1A antibodies (M3-04, M5-05), Hercules, CA, USA) and washed with 12 column volumes of HeLa cells were synchronized by thymidine (2 mM,17h) column buffer. Fusion protein was eluted with column buffer followed by nocodazole (200 nM, 20 h). Total cell lysates were containing 10 mM maltose. incubated with these antibodies for 4 h followed by incubation His-Aurora-A was purified using Ni-NTA agarose (Qiagen, with Protein A/G plus agarose (Santa Cruz Biotechnology, Valencia, CA, USA) according to the manufacturer’s instruc- Santa Cruz, CA, USA) for 1 h at 4 1C.

Oncogene RASSF1A interacts with and activates Aurora-A LLiuet al 6184 For western blots, proteins were fractionated by electro- lysed in lysis buffer (10 mM Tris-HCl, pH 8.0, 1% NP40, phoresis on a 10% or 6–18% gradient SDS–polyacrylamide 150 mM NaCl, 1 mM EDTA) containing phosphatase inhibi- (PAGE) gel and transferred to an immunoblot polyvinylidene tors (20 mM b-glycerolphosphate, 50 mM sodium fluoride and fluoride (PVDF) membrane (Bio-Rad). The blots were probed 1mM Na3VO4) and protease inhibitor cocktail (Roche). The with the indicated primary antibodies followed by appropriate lysates were incubated with EZview Red Anti-Flag (M2) secondary antibodies conjugated with horseradish peroxidase affinity gel (Sigma). After washing with kinase buffer, the (Jackson ImmunoResearch, West Grove, PA, USA). The pellets were incubated with 2 mg GST-Lats2-79-118 in kinase signals were detected using ECL Plus (Amersham). buffer supplemented with 20 mM ATP and 10 mCi [g-32P]ATP at 30 1C for 30 min. Reactions were stopped by the addition of In vivo GST pull-down assay 2 Â SDS–PAGE sample buffer. Proteins were analysed by At 60–72 h after transfection with GST-RASSF1A or pEBG 15% SDS–PAGE followed by autoradiography. At last, the (GST) vector alone, 293T cells were harvested and lysed in gel was stained with Coomassie blue (Bio-Rad). ice-cold cytoskeleton lysis buffer (0.5% Nonidet P-40, 10 mM piperazine-1,4-bis(2-ethanesulfonic acid) (PIPES), pH 7.0, 50 mM siRNA NaCl, 300 mM sucrose, 3 mM MgCl2,1mM ethylene glycol For siRNA the following sequences were used:RASSF1A, tetraacetic acid, 1 mM DTT, 1 mM Na3VO4)withprotease RA196, 50-GACCUCUGUGGCGACUUCAtt-30 (nucleotides inhibitor cocktail (Roche, Indianapolis, IN, USA). The lysates 196–214 relative to the start codon) (Shivakumar et al., 2002), were incubated with glutathione sepharose 4B gel (Amersham) RA1404, 50-GGUUUUGGAUCUUGAAUGUtt-30 (nucleo- for 45 min at room temperature and the gel was sedimented at tides 1404–1422 relative to the start codon) and RA1430, 500 g for 5 min. After five times of washing with PBS, the 50-GGAUAUCCUUAUCAGAGCUtt-30 (nucleotides 1430– glutathione sepharose pellet was collected and the bound proteins 1448 relative to the start codon). Aurora-A, AUA725, were eluted with 1 Â SDS loading buffer with heating at 95 1Cfor 50-AUGCCCUGUCUUACUGUCA-30 (Kufer et al., 2002) 5 min. The eluted proteins were fractionated by electrophoresis on and AUA155, 50-AUUCUUCCCAGCGCGUUCC-30 (Hirota a 10% SDS–PAGE gel and transferred to an immunoblot PVDF et al., 2003). An siRNA sequence targeting the LacZ gene membrane (Bio-Rad). The blots were probed with the indicated (LacZ,50-CGUACGCGGAAUACUUCGAtt-3) was used as antibodies. a control. RA196 specifically targets RASSF1A, whereas RA1404 and RA1430 target both RASSF1A and RASSF1C. In vitro protein binding assay All oligonucleotides were purchased from Ambion and Purified MBP-RASSF1A, MBP-RASSF1C or their truncated delivered into HeLa cells using Oligofectamine (Invitrogen). mutants were incubated with His-Aurora-A in binding buffer HeLa cells were harvested 72 h after transfection. (20 mM Tris-HCl, pH 7.4, 150 mM NaCl, 1 mM EDTA, pH 8.0, 0.025% NP-40, 1 mM DTT, 10% glycerol) at 4 1C for 1 h. The mixture was adsorbed onto amylose resins (NEB). The resins were washed five times with binding buffer. The bound Abbreviations proteins were analysed by western blot using antibodies against MBP and Aurora-A. APC, anaphase-promoting complex; DAPI, 4,6-diamidino-2- phenylindole; MBP, maltose-binding protein; MEF, mouse embryonic fibroblast; MTOC, microtubule-organizing center; In vitro kinase assay NORE, novel Ras effector; RASSF1A, RAS association Recombinant His-Aurora-A (0.5 mg) was incubated with either domain family 1A. MBP-RASSF1A (4 mg) or MBP2 (4 mg) in kinase buffer (25 mM Tris-HCl, pH 7.4, 10 mM MgCl2,5mM MnCl2,1mM NaF, 5 mM b-glycerophosphate, 0.1 mM Na3VO4,1mM DTT) Acknowledgements supplemented with 20 mM ATP, 10 mCi [g-32P]ATP and 2 mg GST-Lats2-79-118 as substrate at 30 1C. COS-7 cells co- We thank E Nigg and H Nojima for the Aurora-A and LATS2 transfected with Flag-Aurora-A and HA-RASSF1A were plasmid constructs.

References

Abbondanzo SJ, Gadi I, Stewart CL. (1993). Derivation of embryonic Carmena M, Earnshaw WC. (2003). The cellular geography of aurora stem cell lines. Methods Enzymol 225:803–823. kinases. Nat Rev Mol Cell Biol 4:842–854. Agathanggelou A, Cooper WN, Latif F. (2005). Role of the Ras- Castro A, Arlot-Bonnemains Y, Vigneron S, Labbe JC, Prigent C, association domain family 1 tumor suppressor gene in human Lorca T. (2002). APC/Fizzy-related targets Aurora-A kinase for cancers. Cancer Res 65:3497–3508. proteolysis. EMBO Rep 3:457–462. Anand S, Penrhyn-Lowe S, Venkitaraman AR. (2003). AURORA-A Crane R, Gadea B, Littlepage L, Wu H, Ruderman JV. (2004). Aurora amplification overrides the mitotic spindle assembly checkpoint, A, and mitosis. Biol Cell 96:215–229. inducing resistance to Taxol. Cancer Cell 3:51–62. Dallol A, Agathanggelou A, Fenton SL, Ahmed-Choudhury J, Hesson L, Andrews PD, Knatko E, Moore WJ, Swedlow JR. (2003). Mitotic Vos MD et al. (2004). RASSF1A interacts with microtubule- mechanics:the auroras come into view. Curr Opin Cell Biol 15:672–683. associated proteins and modulates microtubule dynamics. Cancer Res Bischoff JR, Anderson L, Zhu Y, Mossie K, Ng L, Souza B et al. 64:4112–4116. (1998). A homologue of Drosophila aurora kinase is oncogenic and Dammann R, Li C, Yoon JH, Chin PL, Bates S, Pfeifer GP. (2000). amplified in human colorectal cancers. EMBO J 17:3052–3065. Epigenetic inactivation of a RAS association domain family Blangy A, Lane HA, d’Herin P, Harper M, Kress M, Nigg EA. (1995). protein from the lung tumour suppressor 3p21.3. Nat Genet Phosphorylation by p34cdc2 regulates spindle association of human 25:315–319. Eg5, a -related motor essential for bipolar spindle formation Dammann R, Schagdarsurengin U, Seidel C, Strunnikova M, in vivo. Cell 83:1159–1169. Rastetter M, Baier K et al. (2005). The tumor suppressor RASSF1A Boveri T. (1888). Zellenstudien II. Die Befruchtung und Teilung des in human carcinogenesis:an update. Histol Histopathol 20: Eies von Ascaris megalocephala. Jena Zeit Naturw 22:685–882. 645–663.

Oncogene RASSF1A interacts with and activates Aurora-A LLiuet al 6185 Doxsey S. (2001a). Re-evaluating centrosome function. Nat Rev Mol Liu L, Tommasi S, Lee DH, Dammann R, Pfeifer GP. (2003). Control Cell Biol 2:688–698. of microtubule stability by the RASSF1A tumor suppressor. Doxsey S, Zimmerman W, Mikule K. (2005). Centrosome control of Oncogene 22:8125–8136. the . Trends Cell Biol 15:303–311. Liu L, Vo A, Liu G, McKeehan WL. (2005a). Distinct structural Doxsey SJ. (2001b). Centrosomes as command centres for cellular domains within C19ORF5 support association with stabilized control. Nat Cell Biol 3:E105–E108. microtubules and mitochondrial aggregation and Ewart-Toland A, Briassouli P, de Koning JP, Mao JH, Yuan J, destruction. Cancer Res 65:4191–4201. Chan F et al. (2003). Identification of Stk6/STK15 as a candidate Liu L, Vo A, Liu G, McKeehan WL. (2005b). Putative tumor low-penetrance tumor-susceptibility gene in mouse and human. suppressor RASSF1 interactive protein and inducer Nat Genet 34:403–412. C19ORF5 is a DNA binding protein. Biochem Biophys Res Commun Ewart-Toland A, Dai Q, Gao YT, Nagase H, Dunlop MG, 332:670–676. Farrington SM et al. (2005). Aurora-A/STK15 T+91A is a general Liu L, Vo A, McKeehan WL. (2005c). Specificity of the methylation- low penetrance cancer susceptibility gene:a meta-analysis of suppressed A isoform of candidate tumor suppressor RASSF1 for multiple cancer types. Carcinogenesis 26:1368–1373. microtubule hyperstabilization is determined by cell death inducer Gaglio T, Saredi A, Bingham JB, Hasbani MJ, Gill SR, Schroer TA C19ORF5. Cancer Res 65:1830–1838. et al. (1996). Opposing motor activities are required for the Marumoto T, Hirota T, Morisaki T, Kunitoku N, Zhang D, Ichikawa organization of the mammalian mitotic spindle pole. J Cell Biol Y et al. (2002). Roles of aurora-A kinase in mitotic entry and G2 135:399–414. checkpoint in mammalian cells. Genes Cells 7:1173–1182. Giet R, Glover DM. (2001). Drosophila is required for Marumoto T, Honda S, Hara T, Nitta M, Hirota T, Kohmura E et al. histone H3 phosphorylation and recruitment during (2003). Aurora-A kinase maintains the fidelity of early and late chromosome condensation and to organize the central spindle mitotic events in HeLa cells. J Biol Chem 278:51786–51795. during cytokinesis. J Cell Biol 152:669–682. Marumoto T, Zhang D, Saya H. (2005). Aurora-A—a guardian of Giet R, Petretti C, Prigent C. (2005). Aurora kinases, aneuploidy and poles. Nat Rev Cancer 5:42–50. cancer, a coincidence or a real link? Trends Cell Biol 15:241–250. Meraldi P, Honda R, Nigg EA. (2002). Aurora-A overexpression Glover DM, Leibowitz MH, McLean DA, Parry H. (1995). Mutations reveals tetraploidization as a major route to centrosome amplifica- in aurora prevent centrosome separation leading to the formation of tion in p53À/À cells. EMBO J 21:483–492. monopolar spindles. Cell 81:95–105. Meraldi P, Honda R, Nigg EA. (2004). Aurora kinases link Guo C, Tommasi S, Liu L, Yee JK, Dammann R, Pfeifer GP. (2007). chromosome segregation and cell division to cancer susceptibility. RASSF1A is part of a complex similar to the Drosophila Curr Opin Genet Dev 14:29–36. Hippo/Salvador/Lats tumor-suppressor Network. Curr Biol 17: Mortlock A, Keen NJ, Jung FH, Heron NM, Foote KM, Wilkinson R 700–705. et al. (2005). Progress in the development of selective inhibitors of Guse A, Mishima M, Glotzer M. (2005). Phosphorylation of ZEN-4/ Aurora kinases. Curr Top Med Chem 5:199–213. MKLP1 by aurora B regulates completion of cytokinesis. Curr Biol Naruganahalli KS, Lakshmanan M, Dastidar SG, Ray A. (2006). 15:778–786. Therapeutic potential of Aurora kinase inhibitors in cancer. Hachet V, Canard C, Gonczy P. (2007). Centrosomes promote timely Curr Opin Investig Drugs 7:1044–1051. mitotic entry in C. elegans embryos. Dev Cell 12:531–541. Ortiz-Vega S, Khokhlatchev A, Nedwidek M, Zhang XF, Dammann Hinchcliffe EH, Miller FJ, Cham M, Khodjakov A, Sluder G. (2001). R, Pfeifer GP et al. (2002). The putative tumor suppressor Requirement of a centrosomal activity for cell cycle progression RASSF1A homodimerizes and heterodimerizes with the Ras-GTP through G1 into . Science 291:1547–1550. binding protein Nore1. Oncogene 21:1381–1390. Hirota T, Kunitoku N, Sasayama T, Marumoto T, Zhang D, Nitta M Piel M, Nordberg J, Euteneuer U, Bornens M. (2001). Centro- et al. (2003). Aurora-A and an interacting activator, the LIM some-dependent exit of cytokinesis in animal cells. Science 291: protein Ajuba, are required for mitotic commitment in human cells. 1550–1553. Cell 114:585–598. Portier N, Audhya A, Maddox PS, Green RA, Dammermann A, Honda K, Mihara H, Kato Y, Yamaguchi A, Tanaka H, Yasuda H Desai A et al. (2007). A microtubule-independent role for centrosomes et al. (2000). Degradation of human Aurora2 protein kinase by and aurora a in nuclear envelope breakdown. Dev Cell 12:515–529. the anaphase-promoting complex-ubiquitin-proteasome pathway. Pugacheva EN, Golemis EA. (2005). The focal adhesion scaffolding Oncogene 19:2812–2819. protein HEF1 regulates activation of the Aurora-A and Nek2 Hutterer A, Berdnik D, Wirtz-Peitz F, Zigman M, Schleiffer A, kinases at the centrosome. Nat Cell Biol 7:937–946. Knoblich JA. (2006). Mitotic activation of the kinase Aurora-A Pugacheva EN, Jablonski SA, Hartman TR, Henske EP, Golemis EA. requires its binding partner Bora. Dev Cell 11:147–157. (2007). HEF1-dependent Aurora A activation induces disassembly Keen N, Taylor S. (2004). Aurora-kinase inhibitors as anticancer of the primary cilium. Cell 129:1351–1363. agents. Nat Rev Cancer 4:927–936. Rieder CL, Faruki S, Khodjakov A. (2001). The centrosome in Khokhlatchev A, Rabizadeh S, Xavier R, Nedwidek M, Chen T, vertebrates:more than a microtubule-organizing center. Trends Cell Zhang XF et al. (2002). Identification of a novel Ras-regulated Biol 11:413–419. proapoptotic pathway. Curr Biol 12:253–265. Rong R, Jiang LY, Sheikh MS, Huang Y. (2007). Mitotic kinase Kufer TA, Sillje HH, Korner R, Gruss OJ, Meraldi P, Nigg EA. Aurora-A phosphorylates RASSF1A and modulates RASSF1A- (2002). Human TPX2 is required for targeting Aurora-A kinase to mediated microtubule interaction and M-phase cell cycle regulation. the spindle. J Cell Biol 158:617–623. Oncogene 26:7700–7708. Lerman MI, Minna JD. (2000). The 630-kb homozygous Rong R, Jin W, Zhang J, Sheikh MS, Huang Y. (2004). Tumor deletion region on human chromosome 3p21.3:identification and suppressor RASSF1A is a microtubule-binding protein that evaluation of the resident candidate tumor suppressor genes. The stabilizes microtubules and induces G2/M arrest. Oncogene 23: International Lung Cancer Chromosome 3p21.3 Tumor Suppressor 8216–8230. Gene Consortium. Cancer Res 60:6116–6133. Satinover DL, Leach CA, Stukenberg PT, Brautigan DL. (2004). Littlepage LE, Wu H, Andresson T, Deanehan JK, Amundadottir LT, Activation of Aurora-A kinase by protein phosphatase inhibitor-2, a Ruderman JV. (2002). Identification of phosphorylated residues that bifunctional signaling protein. Proc Natl Acad Sci USA 101: affect the activity of the mitotic kinase Aurora-A. Proc Natl Acad 8625–8630. Sci USA 99:15440–15445. Shivakumar L, Minna J, Sakamaki T, Pestell R, White MA. (2002). Liu L, Baier K, Dammann R, Pfeifer GP. (2007). The tumor The RASSF1A tumor suppressor blocks cell cycle progres- suppressor RASSF1A does not interact with Cdc20, an activator sion and inhibits cyclin D1 accumulation. Mol Cell Biol 22: of the anaphase-promoting complex. Cell Cycle 6:1663–1665. 4309–4318.

Oncogene RASSF1A interacts with and activates Aurora-A LLiuet al 6186 Song MS, Song SJ, Ayad NG, Chang JS, Lee JH, Hong HK et al. Vos MD, Martinez A, Elam C, Dallol A, Taylor BJ, Latif F et al. (2004). The tumour suppressor RASSF1A regulates mitosis by (2004). A role for the RASSF1A tumor suppressor in the regulation inhibiting the APC-Cdc20 complex. Nat Cell Biol 6:129–137. of tubulin polymerization and genomic stability. Cancer Res 64: Taguchi S, Honda K, Sugiura K, Yamaguchi A, Furukawa K, 4244–4250. Urano T. (2002). Degradation of human Aurora-A protein kinase is Walter AO, Seghezzi W, Korver W, Sheung J, Lees E. (2000). The mediated by hCdh1. FEBS Lett 519:59–65. mitotic serine/threonine kinase Aurora2/AIK is regulated by Toji S, Yabuta N, Hosomi T, Nishihara S, Kobayashi T, Suzuki S phosphorylation and degradation. Oncogene 19:4906–4916. et al. (2004). The centrosomal protein Lats2 is a phosphorylation Wang X, Zhou YX, Qiao W, Tominaga Y, Ouchi M, Ouchi T et al. target of Aurora-A kinase. Genes Cells 9:383–397. (2006). Overexpression of aurora kinase A in mouse mammary Tommasi S, Dammann R, Zhang Z, Wang Y, Liu L, Tsark WM et al. epithelium induces genetic instability preceding mammary tumor (2005). Tumor susceptibility of Rassf1a knockout mice. Cancer Res formation. Oncogene 25:7148–7158. 65:92–98. Waters JC, Cole RW, Rieder CL. (1993). The force-producing Trieselmann N, Armstrong S, Rauw J, Wilde A. (2003). mechanism for centrosome separation during spindle formation in Ran modulates spindle assembly by regulating a subset of TPX2 vertebrates is intrinsic to each . J Cell Biol 122:361–372. and Kid activities including Aurora A activation. J Cell Sci 116: Weaver BA, Cleveland DW. (2006). Does aneuploidy cause cancer? 4791–4798. Curr Opin Cell Biol 18:658–667. Tsai MY, Wiese C, Cao K, Martin O, Donovan P, Ruderman J et al. Zhang D, Hirota T, Marumoto T, Shimizu M, Kunitoku N, Sasayama (2003). A Ran signalling pathway mediated by the mitotic kinase T et al. (2004). Cre-loxP-controlled periodic Aurora-A overexpres- Aurora A in spindle assembly. Nat Cell Biol 5:242–248. sion induces mitotic abnormalities and hyperplasia in mammary Uzbekov R, Kireyev I, Prigent C. (2002). Centrosome separation: glands of mouse models. Oncogene 23:8720–8730. respective role of microtubules and actin filaments. Biol Cell 94: Zhao ZS, Lim JP, Ng YW, Lim L, Manser E. (2005). The GIT- 275–288. associated kinase PAK targets to the centrosome and regulates van der Weyden L, Tachibana KK, Gonzalez MA, Adams DJ, Ng BL, Aurora-A. Mol Cell 20:237–249. Petty R et al. (2005). The RASSF1A isoform of RASSF1 promotes Zhou H, Kuang J, Zhong L, Kuo WL, Gray JW, Sahin A et al. (1998). microtubule stability and suppresses tumorigenesis. Mol Cell Biol Tumour amplified kinase STK15/BTAK induces centrosome amplifi- 25:8356–8367. cation, aneuploidy and transformation. Nat Genet 20:189–193.

Supplementary Information accompanies the paper on the Oncogene website (http://www.nature.com/onc)

Oncogene