sign in sign up
<<
Home , Microtubule

Oncogene (2007) 26, 4357–4371 & 2007 Nature Publishing Group All rights reserved 0950-9232/07 $30.00 www.nature.com/onc ORIGINAL ARTICLE A novel role of phospho-b- in regrowth at

P Huang, T Senga and M Hamaguchi

Department of Oncology, Division of Cancer Biology, Nagoya University Graduate School of Medicine, Showa ku, Nagoya, Japan

b-Catenin is a biologically important molecule playing expression (Huelsken and Behrens, 2002). Recent critical roles in both adhesion and transcriptional studies, however, have shown that some components regulation in the Wnt pathway. Here, we show that of Wnt pathway regulate microtubule (MT) function phospho-b-catenin (phosphorylatedat Ser33/37/Thr41), (Gundersen and Cook, 1999; Harwood and Braga, which is reported to be degraded immediately after its 2003). GSK3b itself is present along spindle MTs , accumulatedin the centrosome. Whereas (Wakefield et al., 2003) and can regulate the phospho- phospho-mimicking mutant, S33/37/T41E-b-catenin, rylation of some MT-binding such as tau, couldlocalize to the centrosome, S33A- b-catenin that adenomatous polyposis coli (APC), MAP1B to lacks the phosphorylation site lost its localization to the affect the stability of MTs (Lucas et al., 1998). Another centrosome. Phospho-b-catenin localizedmainly to key component of Wnt pathway, APC, binds to the mother centrosome during the and was plus ends of MTs and stabilizes the growing ends recruitedto daughter centrosome in M-phase. Depletion (Zumbrunn et al., 2001). Similarly, Dishevelled-1, an of b-catenin with small interfering RNA or inhibition of its upstream molecule of b-catenin, is involved in the phosphorylation by LiCl treatment causeddisruption of regulation of MT stability (Krylova et al., 2000; Ciani radial microtubule (MT) array and retardation of the MT et al., 2003). b-Trcp1 may also regulate the timely order regrowth during the recovery from treatment. of meiotic and mitotic events (Guardavaccaro et al., In addition, these treatments increased the frequency of 2003). mono-astral MT reorganization. Furthermore, overex- Many b-catenin-binding proteins are also involved in pression of the nonphosphorylatable b-catenin, but not the the regulation of the function of MTs. binds to phospho-mimicking b-catenin, markedly disrupted radial b-catenin and may tether MTs at adherens junctions. MT array andrepressedthe MT regrowth. In contrast, Overexpression of b-catenin disrupts the cellular loca- phospho-mimicking b-catenin localizedto both of the lization of dynein and dramatically perturbs both the duplicated with aberrant larger and denser organization of MTs at the centrosome and the radial MTs array formation. In addition, some of the cells tethering of MTs at adherens junctions (Ligon et al., overexpressing phospho-mimicking b-catenin hadmultiple 2001). b-Catenin, APC, KAP3 and KIF3A-KIF3B are centrosomes. Taken together, this study demonstrates a in the same complex in vivo, and both APC and novel role of phospho-b-catenin in MT organization at the b-catenin are transported along MTs assisted by KAP3- centrosomes. KIF3A-KIF3B (Jimbo et al., 2002). Another b-catenin- Oncogene (2007) 26, 4357–4371; doi:10.1038/sj.onc.1210217; binding , EB1, is required for the formation and published online 29 January 2007 maintenance of the radial MT array anchored at the centrosome (Askham et al., 2002). A recent paper has Keywords: phospho-b-catenin; microtubule regrowth; provided direct evidence that b-catenin is associated centrosome with the establishment of bipolar mitotic spindle (Kaplan et al., 2004). In addition, b-catenin itself is reported to be associated with the regulation of the . The level of b-catenin increases in S phase, reaching maximum accumulation at late G2/M. Over- Introduction expression of a nonphosphorylatable mutant b-catenin in epidermal keratinocyte induces G2 arrest (Olmeda Traditional Wnt/b-catenin pathway et al., 2003). All these studies above indicate a possible plays key role in embryonic development and role of b-catenin on the regulation of the MT function cancer progression through b-catenin-regulated during the cell cycle. In this report, we show that phospho-b-catenin is localized to mother centrosome during the interphase Correspondence: Professor M Hamaguchi, Division of Cancer and distributed to both mother and daughter centro- Biology, Nagoya University Graduate School of Medicine, 65 somes in M-phase. Furthermore, we found phospho-b- Tsurumaicho, Showa ku, Nagoya, Aichi 466-8550, Japan. E-mail: [email protected] catenin plays a role on the maintenance of radial MT Received 3 October 2005; revised 9 October 2006; accepted 10 November array during the interphase by anchoring MT to the 2006; published online 29 January 2007 centrosome. Phospho-b-catenin and microtubule regrowth P Huang et al 4358 Results proteasome-dependent pathway. However, we found that phospho-b-catenin was localized at microtubule Phospho-b-catenin localizes mainly to mother centrosome organizing center (MTOC) (Figure 1a). Interestingly, during the interphase phospho-b-catenin was localized to only one of the two We first investigated the subcellular localization of centrosomes in cells (Figure 1b). Deconvolution analysis b-catenin and its phosphorylated form in rat fibroblast showed that phospho-b-catenin enveloped the centro- cell line, 3Y1. In normal cells, b-Catenin appears to be some (Figure 1b), which is similar to the localization distributed mainly in the membrane and diffusely of MT anchor proteins such as and CEP100 throughout the (Sadot et al., 2002). b-catenin (Ou et al., 2002). As phospho-b-catenin is localized to only is phosphorylated by GSK-3b at Ser33, 37 and Thr41, one of the two centrosomes (Figure 1b), we checked and ubiquitinated by b-Trcp1 and degraded in a whether it is mother centrosome or not by staining with

Figure 1 Localization of phospho-b-catenin in fibroblast cells. (a and b) Rat 3Y1 cells were immunostained with anti-phospho-b- catenin. MTs and centrosomes were visualized with FITC-conjugated anti-a- and anti-g- , respectively. Phospho-b- catenin (red) localized to centrosomes (arrowhead). Zaxis reconstructions (b, bottom panels) were generated along the bars indicated in the upper panels. Bars, 20 mm. (c) Mother centrosome (arrowhead) was visualized with mother centrosome marker e-tubulin in human foreskin fibroblasts (HFF). Bar, 20 mm. (d) Rat 3Y1 cells were immunostained with anti-phospho-b-catenin and anti-EB1. Insets show higher magnification of the centrosome. Arrowheads point mother centrosome. Bar, 20 mm.

Oncogene Phospho-b-catenin and microtubule regrowth P Huang et al 4359 a mother centrosome marker, e-tubulin. We found the detect mutant b-catenin (S33A) whose Ser33 was co-localization of phospho-b-catenin with e-tubulin replaced with alanine (Figure 2a). To confirm phos- (Figure 1c). EB1 localizes to the mother centrosome and phorylation-dependent localization of b-catenin to the C-terminus of EB1 interacts with b-catenin (Askham centrosome, we constructed a phospho-mimicking et al., 2002). We also observed co-localization of EB1 mutant b-catenin whose Ser33, 37 and Thr41 were and phospho-b-catenin (Figure 1d). However, we replaced to (S33/37/T41E) and transfected could not detect direct interaction of phospho-b-catenin it to COS7 cells. Substitution of serine or threonine to with e-tubulin and EB1 by immunoprecipitation (Supple- glutamic acid is known to mimic phosphorylation. mentary Figure S1a). Although S33/37/T41Eb-catenin localized to the centro- The phospho-b-catenin antibody has been reported to some, we could not see any localization of S33Ab- detect specifically phosphorylated b-catenin (Kielhorn catenin to the centrosome (Figure 2b arrowhead). These et al., 2003), and we confirmed that this antibody did not results with mutant b- are consistent with our

Figure 2 S33/37/T41E b-catenin localizes to the centrosome. (a) Lysates of COS7 cells expressing b-catenin WT or mutant were subjected to immunoblotting with rabbit anti-phospho-b-catenin antibody. Then the membrane was reprobed with anti-b-catenin or anti-GFP antibody. Line 1, GFP; Line 2, b-catenin-GFP; Line 3, b-catenin-S33A-GFP; Line 4, b-catenin-S33/37/T41E-GFP. (b) COS7 cells transfected with b-catenin-S33/37/T41E-GFP or b-catenin-S33A-GFP were stained with anti-g-tubulin and the localization of b-catenin mutant was visualized with GFP. Bar, 20 mm. (c) COS7 cells transfected with b-catenin-S33/37/T41E-GFP were stained with anti-g-tubulin. Bar, 20 mm. Insets show higher magnification of the centrosomes.

Oncogene Phospho-b-catenin and microtubule regrowth P Huang et al 4360 observation that phospho-b-catenin localizes to the onset of M-phase, phospho-b-catenin was distributed to centrosome. In contrast to endogenous wild-type both mother and daughter centrosomes (Figure 3a). phospho-b-catenin localized to the mother centrosome, Consistent with this observation, there was a dramatic S33/37/T41E mutant unexpectedly localized to both of increase of phospho-b-catenin at G2/M-phase, as the duplicated centrosomes visualized with g-tubulin. compared with that at G1-phase (Figure 3b). We also Moreover, we observed that some of the cells over- found that phospho-b-catenin was present at both expressing S33/37/T41Eb-catenin had multiple centro- centrosomes throughout the whole M-phase (Figures 3c somes, every one of which had association of S33/37/ and Supplementary Figure S2a). This distinctive distri- T41Eb-catenin (Figure 2c). bution pattern led us to speculate that it may take a role in daughter centrosome maturation and spindle pole Phospho-b-catenin is recruited to daughter centrosome in formation. As cells progressed to late , phospho- M-phase b-catenin accumulated at mid-body, indicating a possible We next examined the localization of phospho-b-catenin localization of phospho-b-catenin to the plus ends of in various cell-cycle stages. We found that, upon the the MTs.

a γ-Tubulin Phospho-β-Catenin Merge

b 0h 3h 6h 9h 12h Phospho-β-Catenin

β-Catenin

Cyclin B1

α-Tubulin

c α-Tubulin Phospho-β-Catenin DAPI Merge

telophase

Figure 3 Phospho-b-catenin is recruited to daughter centrosome in M-phase. (a) 3Y1 cells at interphase and mitotic phase in one field were stained with anti-g-tubulin (green) and anti-phospho-b-catenin (red), respectively. The white arrows indicate the cells in interphase, the red arrows indicate the cells in mitotic phase. Bar, 20 mm. (b) Phospho-b-catenin level increased during cell cycle. 3Y1 cells were synchronized at G1 phase using double thymidine block. Cells were harvested at 0, 3, 6, 9 and 12 h after the release from double thymidine block. Cell lysates were subjected to immunoblotting with against phospho-b-catenin, b-catenin and B1. (c) 3Y1 cells at different mitotic phase were investigated for the localization of a-tubulin (green), phospho-b-catenin (red) and DNA (blue). Bar, 10 mm.

Oncogene Phospho-b-catenin and microtubule regrowth P Huang et al 4361 Phospho-b-catenin pool is upregulated in was impaired by the depletion of b-catenin (Figures 5d cell cycle-dependent manner and Supplementary Figure S1d). Deconvolution As b-catenin is hardly detectable owing to the immediate analysis clearly showed that centrosomal localization degradation in the cytoplasm after its phosphorylation of EB1 significantly reduced in cells depleted of at S33 (Sadot et al., 2002), we carefully investigated the b-catenin. Although 76% of cells (n ¼ 113) transfected subcellular localization and the protein level of with luciferase siRNA had clear centrosomal EB1, only b-catenin and phospho-b-catenin during the cell cycle. 25% of cells (n ¼ 103) treated with b-catenin siRNA had The proportion of each cell phase at 0, 3, 6, 9 and 12 h centrosomal EB1. Centrosomal localization of ninein after release from double thymidine block was indicated was also reduced in the cells treated with b-catenin in Supplementary Figure S3. b-Catenin relocalized to siRNA (Figure 5e). By treatment with b-catenin siRNA, the nucleus from S phase. This relocalization continued the relative level of cells with reduced centrosomal from S phase to G2/M-phase so that nuclear b-catenin ninein increased to 69% (n ¼ 81), whereas those of clearly increased in G2/M-phase cells, as compared with control cells were 19% (n ¼ 105). those in G1-phase cells (Figure 4a). Phospho-b-catenin started to accumulate in the nucleus at S phase and reached plateau at G2/M-phase. Consistently, the Depletion of b-catenin retards the MT regrowth at the protein level of phospho-b-catenin increased 3 h after centrosome the release from thymidine block and decreased 12 h To obtain more clues, we next examined the dynamics of after the release when cyclin B1 has also decreased phospho-b-catenin localization during MT reorganiza- (Figure 3b). As the phosphorylation of b-catenin is tion by treatment with nocodazole. In control cells, part regulated by GSK-3b, we determined the localization of of phospho-b-catenin was localized at the perinuclear phospho-b-catenin and GSK-3b during the cell cycle. As region, which almost overlapped with the localization of shown in Figure 4c, GSK-3b started to accumulate in radial MT array. This perinuclear phospho-b-catenin the nucleus at S phase and localized exclusively in the was disappeared when MTs were depolymerized by nucleus at G2/M-phase as reported previously (Tang nocodazole. After washing out of nocodazole, phospho- et al., 2003). The accumulation of GSK-3b in the b-catenin was recruited to the peripheral area of the nucleus preceded that of phospho-b-catenin as well as centrosome along with the regrowth of MT network b-catenin. LiCl that inhibits GSK-3b activity impeded (Figure 6a, arrowhead). This localization mode of the cell cycle-dependent increase of phospho-b-catenin phospho-b-catenin is similar to that of proteins asso- (Figure 4b). Taken together, cell cycle-dependent ciated with spindle pole formation such as dynein and phosphorylation of b-catenin is likely to be regulated (Quintyne and Schroer, 2002). by GSK-3b. We next examined the effect of b-catenin depletion on MT reorganization. Although, 2 min after nocodazole washout, nucleation of MTs was observed in cells Depletion of b-catenin suppresses radial MT array depleted of b-catenin, it was smaller than that of control anchoring at centrosome cells. Fifteen minutes after nocodazole washout, control As fibroblast radial MTs anchor at mother centrosome cells showed fine MT formation. In contrast, MT where phospho-b-catenin is localized during the regrowth was clearly retarded in cells depleted of interphase, we next examined the effect of depletion of b-catenin (Figure 6b, arrowhead). MT regrowth was b-catenin on the formation of radial MT array. quantified when the intensity of MT aster reached As shown in Figure 5a, the total protein level of maximum after 30 min of nocodazole washout. The b-catenin was reduced greatly by treatment with fluorescent intensity of visualized MT (a-tubulin staining) small interfering RNA (siRNA) targeting b-catenin. in cells treated with b-catenin siRNA was decreased to We found that in cells whose b-catenin was not depleted, 56% of normal cells (Figure 6c and d). MT asters were the majority of MTs focused on the centrosome then organized into radial MT arrays at 60 min after and was organized into well-defined radial arrays. nocodzole washout (Figure 6e). In normal cells at 60 In contrast, in cells whose b-catenin was substantially and 90 min after nocodzole washout, we observed a fine depleted, MTs did not form radial arrays nor converge radial MT array formation focusing at centrosome on the centrosome but were randomly organized in where phospho-b-catenin was localized. In contrast, the cytoplasm (Figure 5b). Sixty-nine percent of cells MTs did not form clear radial arrays but were randomly treated with b-catenin siRNA showed defects in MT formed in the cytoplasm in b-catenin-depleted cells anchoring at the centrosome, but only 8.7% of cells (Figure 6e). treated with control siRNA had the defects of MT MTs are generally associated only with mother anchoring (Figure 5c). These results suggest that centrosomes and young daughter centrosomes are not b-catenin is required for MTs anchoring at the centro- yet competent to organize MTs (Piel et al., 2000). We some. We next examined whether the depletion of visualized the centrosome by staining g-tubulin and b-catenin had the effect on the localization of other investigated the MTs reorganization in cells that has two centrosomal proteins. As shown in Supplementary centrosomes (the distance between two centrosomes Figure S1b and c, both g- and e-tubulin localized separated over 5 mm). Sixty-three percent of control normally to the centrosome in b-catenin-depleted cells treated with siRNA had MT reorganization from cells. In contrast, localization of EB1 to the centrosome both centrosomes, whereas 37% of cells had MT

Oncogene Phospho-b-catenin and microtubule regrowth P Huang et al 4362

Figure 4 Phospho-b-catenin pool is upregulated in cell cycle-dependent manner. (a) After release from double thymidine block for 0, 3, 6 and 9 h, cells were stained with anti-b-catenin (top panels), anti-phospho-b-catenin (middle panels) and DAPI (bottom panels). Bar, 20 mm. (b) 3Y1 cells synchronized by double thymidine block were treated with either NaCl (20 mM) or LiCl (20 mM) after release from the double thymidine block, then were lysed and subjected to immunoblotting. (c) 3Y1 cells synchronized by double thymidine block were stained with anti-GSK-3b and anti-phospho-b-catenin after release from the double thymidine block. Bar, 20 mm.

reorganization from only one of the centrosomes. When somes (Figure 6f and g). These results suggest that cells were treated with b-catenin siRNA, 65% of cells maturation of centrosomes was inhibited by depletion of had MT reorganization from only one of the centro- b-catenin and thus mono-astral MTs increased.

Oncogene Phospho-b-catenin and microtubule regrowth P Huang et al 4363 a c 100 48h 72h

24h lls(%) 80 - + - + - + β-Catenin siRNA + - + - + - Luciferase siRNA 60 β-Catenin 40 α-Tubulin 20 Diffused astral MT ce Diffused 0 Luciferase β-Catenin siRNA α β b -Tubulin -Catenin Merge

F D

d EB1 γ-Tubulin Merge

β-Catenin siRNA

EB1 γ-Tubulin Merge

Luciferase siRNA

e γ-Tubulin Ninein Merge

Luciferase siRNA

γ-Tubulin Ninein Merge

β-Catenin siRNA

Figure 5 Depletion of b-catenin suppresses radial MT array anchoring at centrosome. (a) b-Catenin siRNA suppresses the expression of b-catenin effectively. Lysates of 3Y1 cells transfected with 50 nM b-catenin siRNA (combination of two kinds of siRNA targeting different sequences) or luciferase siRNA were analysed with anti-b-catenin and anti-a-tubulin. (b) 3Y1 cells treated with 50 nM b-catenin siRNA for 24 h were stained with anti-b-catenin (red in merge) and anti-a-tubulin (green in merge). The cell with MT minus ends focused at the centrosome is marked as ‘F’ and the cell with diffused MT is marked as ‘D’ for diffuse. Bar, 20 mm. (c) Quantification of cells with diffused radial MTs. Two hundred cells from three independent experiments were scored for the presence of incomplete radial MTs. (d) 3Y1 cells treated with 50 nM b-catenin siRNA for 24 h were stained with EB1 (green) and anti-g-tubulin (red). Z axis reconstructions of the cross-section and the vertical section were generated along the bars indicated in the centrosome. Bar, 10 mm. (e) 3Y1 cells treated with 50 nM b-catenin siRNA for 24 h were stained with ninein (red) and anti-g-tubulin (green). The regions inside the frames were shown in higher magnification (inset). Inset right panel showed the z axis reconstruction along the bar indicated in the inset left panel. Bar, 10 mm.

Oncogene Phospho-b-catenin and microtubule regrowth P Huang et al 4364 a α-Tubulin Phospho-β-Catenin Merge

Control

Nocodazole treatment

regrowth 30min

b α-Tubulin β-Catenin Merge α-Tubulin Luciferase siRNA RNA Regrowth 2 min Regrowth -Catenin si β Luciferase siRNA RNA Regrowth 15 min Regrowth -Catenin si β

Figure 6 Knockdown of b-catenin inhibits the reorganization of the astral MTs. (a) MTs of 3Y1 cells were depolymerized by treating with 33 mM nocodazole on ice for 45 min. Cells were washed with PBS for three times followed by serum-free medium for another three times, and then incubated in serum-free medium at 371C for 30 min. Cells at each stage were examined for the localization of phospho-b-catenin (red in merge) and MTs (green in merge). Bar, 20 mm. (b) Cells transfected with siRNA (50 nM) targeting b-catenin were treated with nocodazole and released for 2 or 15 min. Cells were immunostained with a-tubulin and b-catenin. To facilitate the visualization of MTs, green color MTs were changed to black color (right columns). Bar, 20 mm. (c) Cells transfected with siRNA (50 nM) targeting b-catenin were treated with nocodazole and released for 30 min. Cells were immunostained for a-tubulin. MTs fluorescent intensity was measured within a 5-mm-diameter disk at the center of the nucleated MTs, relative intensity map was shown in inset (yellow color means high intensity and blue color means low intensity, intensity range is available 0–4095). Bars, 20 mm. (d) Quantification of the rate of MT regrowth by immunofluorescence intensity. The average relative intensity of the cells stained with a-tubulin was indicated in graph (mean s.d., n ¼ 100, three independent experiments). (e) Cells transfected with siRNA (50 nM) targeting b-catenin were treated with nocodazole and released for 60 or 90 min. Cells were immunostained with a-tubulin and phospho-b-catenin. Bar, 20 mm. (f) Cells transfected with siRNA (50 nM) targeting b-catenin were treated with nocodazole and released for 30 min. Cells were immunostained for a- and g-tubulin. Inset shows the cell containing two centrosomes. Bar, 10 mm. (g) Quantification of cells with astral MT reorganization from only one centrosome. Thirty cells (the distance between two centrosomes is over 5 mm) were scored for the presence of mono-astral MT cells. The experiment was performed three times independently.

Oncogene Phospho-b-catenin and microtubule regrowth P Huang et al 4365 c d

β-Catenin siRNA 1200

800 fluorescence

400 Intensities of

Luciferase 0 Luciferase β-Catenin siRNA siRNA siRNA

e Regrowth 60 min Regrowth 90 min

α-Tubulin

Phospho-β-Catenin

Merge

Luciferase siRNAβ-Catenin siRNA Luciferase siRNA β-Catenin siRNA g 100 f ) 80

60 l MT cells(% 40

20 Mono astra

0 β Luciferase β-Catenin Luciferase siRNA -Catenin siRNA siRNA siRNA

Figure 6 Continued.

LiCl treatment suppresses radial MT array anchoring at found that cells whose centrosomal phospho-b-catenin centrosome and retards MT regrowth was reduced by LiCl treatment had no organized radial LiCl inhibits phosphorylation of b-catenin by GSK-3b. MT array anchoring at the centrosome, whereas cells As shown in Supplementary Figure S4a, phosphoryla- with phospho-b-catenin at the centrosome had orga- tion of b-catenin was suppressed by treatment with nized radial Mt arrays (Figure 7a). 20 mM LiCl. LiCl induced the accumulation of b-catenin We also examined the effect of LiCl treatment and disrupted the formation of radial MT array on the reorganization of MTs after nocodazole (Supplementary Figure S4b). We checked whether the washout. Twenty-four hours after LiCl treatment, MT centrosomal localization of phospho-b-catenin was regrowth was remarkably repressed (Supplementary associated with the formation of radial MT array. We Figure S5). MT fibers regrew from the centrosome

Oncogene Phospho-b-catenin and microtubule regrowth P Huang et al 4366

Figure 7 LiCl treatment inhibiting phosphorylation of b-catenin suppresses radial MT array anchoring at centrosome and retard MT regrowth. (a) After incubating for 24 h with LiCl (20 mM), 3Y1 cells were immunostained with anti-a-tubulin and phospho-b-catenin. To facilitate the visualization of MTs, green color MTs were changed to black color (bottom right). Bar, 20 mm. (b) After incubating for 12 h with LiCl (20 mM) or NaCl (20 mM) as a control, 3Y1 cells were treated with nocodazole and released for 30 min. Cells were immunostained for a- and g-tubulin. Insets showed the cell containing two centrosomes. Bar, 10 mm. (c) Quantification of cells with astral MT reorganization from only one centrosome. Thirty cells (the distance between two centrosomes is over 5 mm) were scored for the presence of mono-astral MT cells. The experiment was performed three times independently.

where phospho-b-catenin remained (Supplementary Figure S5, arrow), and many polymerized MT fibers were Figure S5, arrowhead), but not from the centrosomes distributed freely in the cytoplasm detaching from the where phospho-b-catenin disappeared (Supplementary centrosome (Supplementary Figure S5). Interestingly,

Oncogene Phospho-b-catenin and microtubule regrowth P Huang et al 4367 short-time LiCl treatment caused an acceleration of the Nonphosphorylatable mutant b-catenin (S33A) represses reorganization of MT array (data not shown), which may the regrowth of MTs reflect a stabilization of the MT owing to the decreased We next examined the MT formation under the phosphorylation of tau and APC (Lucas et al., 1998). We expression of nonphosphorylatable S33Ab-catenin or also observed that LiCl treatment increased the number phospho-mimicking S33/37/T41Eb-catenin. Green of mono-aster MT, a phenotype similarly observed in fluorescent protein (GFP)-tagged S33Ab-catenin or b-catenin-depleted cells (Figure 7b and c), possibly by S33/37/T41Eb-catenin was transfected to COS7 cells inhibiting the maturation of the centrosome. and stained for a-tubulin to evaluate the formation of

a GFP α-Tubulin Merge S33A S33/37/T41E

b GFP α-Tubulin Merge S33A S33/37/T41E S33/37/T41E

Figure 8 Nonphosphorylatable b-catenin S33A represses the regrowth of MTs. (a) COS7 cells transfected with b-catenin-S33A-GFP or b-catenin-S33/37/T41E-GFP were stained with anti-a-tubulin. The localization of b-catenin-S33A or b-catenin-S33/37/T41E-GFP were visualized by GFP. The regions inside the frames were shown in higher magnification (inset). Inset right panel showed the z axis reconstruction along the bar indicated in the inset left panel. Bars, 20 mm. (b) Cells transfected with b-catenin-S33A-GFP or b-catenin- S33/37/T41E-GFP were treated with nocodazole and then released for 60 min. Cells were immunostained for a-tubulin. Bar, 20 mm. The regions inside the frames were shown in higher magnification (inset). Inset right panel showed the z axis reconstruction along the bar indicated in the inset left panel. Bars, 20 mm.

Oncogene Phospho-b-catenin and microtubule regrowth P Huang et al 4368 radial MT array. The formation of radial MT array was expressing phospho-mimicking mutant had aberrant disrupted in cells expressing GFP-S33Ab-catenin, larger and denser radial MTs array formation. In whereas the radial MT array was well formed in cells , phospho-b-catenin localizes to both centro- expressing GFP-S33/37/T41Eb-catenin (Figure 8a). somes/spindle poles. b-Catenin siRNA-induced mono- Whereas 90% of cells (n ¼ 100) expressing GFP- astral spindle, and both b-catenin siRNA and LiCl S33Ab-catenin could not form radial MT array, only increased frequency of mono-astral MT reorganization 27% of cells (n ¼ 100) expressing GFP-S33/37/T41Eb- after nocodazole washout, suggesting that phospho-b- catenin could not form radial MT array as measured by catenin has a potentially important function on centro- the quantitative assay. We also observed larger and some maturation. In fact, when GSK-3b is inhibited, denser formation of radial MT array in some cells additional long MTs were formed from one pole expressing GFP-S33/37/T41Eb-catenin (Figure 8a, bot- and many of the were mono-oriented tom panel and Supplementary Figure S6a), suggesting (Wakefield et al., 2003). Our findings perhaps provide that this phospho-mimicking mutant b-catenin may more reasonable explanation for this asymmetrical strengthen the formation of radial MT array. We could effect of GSK-3b inhibition on centrosomes. clearly observe that S33/37/T41E mutant displayed two GSK-3b-dependent phosphorylation of b-catenin is dots distribution at the center of radial MT array upregulated in G2/M-phase. On the other hand, the (Figure 8a, inset). Strikingly, in some cases, we found inactive phosphorylated form of GSK-3b is also that S33/37/T41E mutant formed a ring structure but increased in G2/M-phase. We examined the localization not the dot one at the centrosome (Supplementary of phospho-GSK-3b and compared with that of GSK- Figure S6b), overlapping the localization of MTOC, 3b, which is mostly unphosphorylated. Whereas most whose structure was also deformed from the dot type to of the phospho-GSK-3b localized in the cytoplasm a ring one, strongly supporting that phospho-b-catenin throughout the cell cycle (Supplementary Figure S2b), was involved in the MTOC function. We next examined unphosphorylated GSK-3b was translocated to the the effect of both mutants on MT reorganization. COS7 nucleus from S phase to G2/M-phase. It is likely that cells were transfected with S33Ab-catenin or S33/37/ active form of GSK-3b in nucleus increases from S phase T41Eb-catenin and treated with nocodazole. As shown to G2/M-phase. Indeed, GSK-3b appears to phosphor- in Figure 8b, at 60 min after nocodazole washout, cells ylate cyclin D in the nucleus in S phase which in turn expressing S33/37/T41E showed clear formation of translocates phosphorylated cyclin D to the cytoplasm radial MT array, whereas radial MT array formation (Diehl et al., 1998). These results support the idea that was clearly impaired in cells that express S33Ab-catenin. GSK-3b in the nucleus remains active. It is likely that We also observed that overexpression of S33A, but not b-catenin is phosphorylated by GSK-3b in the nucleus, S33/37/T41E, resulted in the delay or complete abroga- where it is protected from the degradation and tion of MTs regrowth at 30 min after nocodazole accumulated until mitosis and then released to the washout (Supplementary Figure S6c). cytoplasm when nuclear membrane is broken. Although phospho-b-catenin has a short half-life, degradation of phospho-b-catenin appears to be determined by the subcellular localization (Sadot et al., 2002). It should be Discussion noted that the subcellular distribution of the b-catenin destruction complex is mostly localized in the cytoplasm In addition to its critical role in cell adhesion and (Reinacher-Schick and Gumbiner, 2001). Nuclear accu- transcriptional regulation, Kaplan et al. (2004) reported mulation of phospho-b-catenin in G2/M may, therefore, that b-catenin localized to spindle poles and played a protect it from degradation. In fact, when cells enter role in the establishment of a bipolar mitotic spindle. In into the next G1-phase, phospho-b-catenin released to this study, we showed that phospho-b-catenin was the cytoplasm is rapidly degraded (Figure 3b). It is present at mother centrosome during the interphase consistent with a previous report that soluble b-catenin and localized to both of the two centrosomes upon entry levels showed dynamic changes: a five to six-fold into M-phase. This distribution pattern is similar to the increase at G2/M-phase followed by a quick decrease localization of the MT anchors such as ninein and of b-catenin after reentering into G1-phase (Olmeda CEP100 (Ou et al., 2002), which indicates a potential et al., 2003). function of phospho-b-catenin on MT anchoring and It remains to be clarified whether b-catenin is centrosome maturation. Depletion of b-catenin by phosphorylated at the centrosome or translocated to siRNA or inhibition of the phosphorylation of b-catenin the centrosome after phosphorylation. It has been by LiCl causes loss of radial MTs array focusing at the reported that phospho-GSK-3b that is inactive is centrosome, indicating a role of phospho-b-catenin on abundant at the centrosome and spindle poles. We MT array anchoring at the centrosome. Furthermore, found that GSK-3b is not present at the centrosome overexpression of nonphosphorylatable S33Ab-catenin during the interphase when phospho-b-catenin is present causes disruption of radial MTs array focusing at the at the centrosome. Phospho-b-catenin may be, there- centrosome, whereas overexpression of phospho-mi- fore, translated to the centrosome after phosphorylation micking S33/37/T41Eb-catenin does not disrupt radial by unknown mechanism. MTs array focusing at the centrosome. Although we As b-catenin plays critical roles in both cell adhesion could not clarify the mechanism, some of the cells and transcriptional regulation in the Wnt pathway, its

Oncogene Phospho-b-catenin and microtubule regrowth P Huang et al 4369 function has been studied in this context. b-Catenin controlled by the cell cycle. Our results are consistent localizes to the plasma membrane where it associates with the previous report (Olmeda et al., 2003). with dynein and controls MT anchoring at the Interestingly, we found that most of the cells over- adherence junction (Ligon et al., 2001). Because of expressing wild-type b-catenin or nonphosphorylatable these observations, a possibility arose that disruption of S33Ab-catenin formed abnormal nuclei (data not MT anchoring at the adherence junction subsequently shown). On the other hand, cells overexpressing caused disorganization of MT. In a single isolated cell, phospho-mimicking S33/37/T41Eb-catenin also fre- however, b-catenin is not involved in cell–cell adhesion; quently displayed aberrant MT network and abnormal yet, MT arrays were clearly organized in these cells nucleus (data not shown). These results suggest that (Supplementary Figure S4c). After transfection of aberrant accumulation of phospho-b-catenin may also siRNA against b-catenin, some cells lost b-catenin at cause aberrant MT formation, which, in turn, yields the the plasma membrane but MT array were still clearly formation of abnormal nucleus. Indeed, cell transforma- organized (Supplementary Figure S4d). Therefore, the tion with v-src causes aberrant accumulation of disruption of MT array may be a direct effect of phospho-b-catenin and abnormal nucleus formation phospho-b-catenin depletion rather than the secondary with multiple centrosomes (data not shown). Aberrant ex- effect caused by the loss of cell–cell adhesion. Moreover, pression of centrosomal proteins is frequently observed it should be noted here that cells treated with LiCl had in cancers and overexpression of centrosomal kinase large amount of b-catenin localized at adherence such as Nek2 causes accumulation of multinucleated junction but failed to form the radial MT array cells with supernumerary centrosomes (Hayward et al., (Supplementary Figure S4b), suggesting again that b- 2004). It is widely accepted that centrosomal abnorm- catenin localized at adherence junction alone is not alities increases the genomic instability which in turn, sufficient for the formation of radial MT array. contributes to tumorigenicity (Pellman, 2001; Nigg, In the canonical Wnt pathway, b-catenin plays a 2002; Sluder and Nordberg, 2004). In fact, high levels central role in regulating . GSK-3b and of nuclear phospho-b-catenin are associated with APC are involved in Wnt pathway by regulating significantly worse survival in malignant melanoma b-catenin degradation. However, in the noncanonical (Kielhorn et al., 2003). Transcriptional upregulation of Wnt pathway, GSK-3b and APC are localized at the cell proliferation such as cyclin D1 and c-Myc are centrosome and have a direct role on the regulation of attributed to the oncogenic function of b-catenin. In MT (Harwood and Braga, 2003). Inhibi- addition to the transcriptional control, our findings tion of GSK-3b or truncations of APC cause abnormal indicate that b-catenin may also contribute to tumor- mitosis, which is considered as an effect on the igenesis by causing centrosomal abnormalities and formation of mitotic spindle but not an effect on gene genetic instability. expression mediated by b-catenin. Inhibition of GSK-3b or truncations of APC promotes the elevation of b-catenin and its mediated gene expression. Since Materials andmethods phospho-b-catenin localizes at the centrosome, the MT disorganization caused directly by the depletion of Cell culture and synchronization b-catenin. It has been reported that a dominant-negative A rat fibroblast cell line, 3Y1, was cultured in Dulbecco’s- mutant of T-cell factor (TCF) that blocked b-catenin- modified Eagle’s medium (DMEM) supplemented with 5% mediated transcriptional activation could not induce the newborn bovine serum. For double thymidine block and formation of mono-astral spindles (Kaplan et al., 2004). release experiment, cells were incubated in fresh DMEM Furthermore, blocking all transcriptional activity by containing 2 mM thymidine (Sigma, Saint Louis, MO, USA) for 12–16 h, released for 12 h in DMEM containing 10% fetal treatment of cells with actinomycin D did not abrogate bovine serum, and then incubated again in 2 mM thymidine for the MTs stabilization function of dishevelled (DVL) 12–16 h. After release from double thymidine block, cells were (Ciani et al., 2003). To rule out the involvement of subjected to immunofluorescence and immunoblot. b-catenin transcriptional activity in the MT reorganiza- tion, we examined the cells treated with actinomycin D. Reagents and antibodies We found that actinomycin D neither disrupt MT Nocodazole was from Cell Signaling Technology (Beverly, MA, focused at the centrosome nor inhibited the MT USA). Anti-a- and anti-g-tubulin antibodies were purchased regrowth after nocodazole washout (data no shown). from Calbiochem (Darmstadt, Germany). Other antibodies In addition, both LiCl treatment and overexpression of included mouse monoclonal antibodies against b-catenin, S33A mutant, which increased the accumulation of GSK-3b, EB1 (Transduction Laboratories, San Jose, CA, b-catenin thereby upregulated b-catenin-mediated gene USA), rabbit polyclonal antibodies against phospho-b- expression, resulted in disorganized MT and suppres- catenin, phospho-GSK-3b (Cell Signaling), ninein (Biolegend, sion of MT regrowth. We, therefore, suggest that the San Diego, CA, USA). Fluorescein isothiocyanate (FITC)- transcriptional process controlled by b-catenin is dis- conjugated a-tubulin, FITC-conjugated g-tubulin, e-tubulin (Sigma) or cyclin B1 (Santa Cruz Biotechnology, Santa Cruz, pensable for its function in MT dynamics. CA, USA). Rat b-catenin cDNA was isolated using reverse The localization and phosphorylation of b-catenin transcription-polymerase chain reaction (RT–PCR), and then were regulated in a cell cycle-dependent manner ligated into pcDNA3 vector. S33A and S33/37/T41E mutation (Figure 4a), suggesting that the localization and phos- in b-catenin cDNA was made by site-directed mutagenesis phorylation of b-catenin should be spatiotemporally using PCR, and then ligated into pEGFP-N1 vector.

Oncogene Phospho-b-catenin and microtubule regrowth P Huang et al 4370 SiRNA treatment laser-scanning (Ar 488 nm, HeNe 543 nm) confocal microscope For siRNA treatment, siRNA oligonucleotides with two (BX50; Olympus). FLUOVIEW (version 4.3; Olympus) was thymidine residues (dTdT) at the 30-end were produced by used for fluorescence intensity measurements of a-tubulin DHARMACON (Lafayette, Colorado, USA). The sequences staining at the nucleated MTs. A 1280 Â 1024-pixel, 16-bit targeting b-catenin were 50-CUAUCAGGAUGACGCGGAA-30 image was collected with acquisition parameters (Objective (corresponding to nucleotides 427–436) and 50-AGAGUAG Lens: PLAPO 60 Â 0; PMT Voltage: 620; offset: 6; Gain: 2; CUGCAGGGGUCC-30 (corresponding to nucleotides 1338– Frame Filter: 4 frame Kalman filter). Imaging conditions were 1347). A luciferase-targeting siRNA was used as a control. identical for all cells. Best focus was based on highest pixel SiRNA was transfected into cells using gene Porter (Gene intensity at the center of the nucleated MTs. The relative Therapy Systems, San Diego, CA, USA) or Magnetofection intensity map within a 5-mm-diameter disk was generated by Kit (OZ Biosciences, Marseille cedex, France). using FLUOVIEW (yellow color means high intensity and blue color means low intensity, intensity range is available Immunoblotting and immunoprecipitation 0–4095). The average relative pixel intensity within a 5-mm- Cells were washed with ice-cold phosphate-buffered saline diameter disk in each cell was calculated by using FLUOVEW. (PBS) and scraped with sample buffer as reported previously Later, the average relative intensity of 100 cells in three (Ruhul Amin et al., 2003). Equal amount of proteins were independent experiments was calculated and used for quanti- electrophoresed on 10% SDS–PAGE (SDS–polyacrylamide fication of MT regrowth. gel electrophoresis) and transferred to polyvinylidene difluor- ide membrane. The membrane was blocked in 5% nonfat Microtubule regrowth skimmed milk, incubated with the primary antibodies followed were depolymerized by treating with 33 mM by the secondary antibodies. Proteins were detected using nocodazole on ice for 45 min. For regrowth, cells were washed enhanced chemiluminescence system (Amersham Biosciences, in PBS for three times followed by serum-free medium for Bockinghamshire, England). For immunoprecipitation experi- another three times, and then incubated in serum-free medium ment, cells were lysed in a lysis buffer containing 50 mM Tris- at 371C for 2, 15, 30 or 60 min. HCl (pH 7.4), 150 mM NaCl, 1 mM ethylenediaminetetraacetic acid, 1 mM phenylmethylsulfonyl fluoride and 1% Triton X-100. Lysates were centrifugalized at 15 000 r.p.m. for 20 min at 41C, then the supernatants were incubated with antibodies Abbreviations coupled to protein A-agarose for 2–4 h at 41C. The immuno- precipitates were washed six times with lysis buffer and then phospho-b-catenin, phosphorylated b-catenin at serine33/37/ subjected to immunoblotting. threonine41; GSK-3b, Glycogen synthase kinase 3b; MT, microtubule. Immunofluorescence and microscopy Cells grown on glass coverslips (Matsunami Glass IND, Acknowledgements Osaka, Japan) were fixed in 4% paraformaldehyde for 15 min, permeabilized with 0.5% Triton X-100 in PBS for 5 min as We thank members and staffs of the Hamaguchi laboratory previously reported (Liu et al., 2001). They were then exposed for their technical assistance and helpful discussion. This work consecutively to primary antibodies and rhodamine-conju- was supported by a Grant-in-Aid for COE research from the gated or FITC-conjugated secondary antibodies. Images were Ministry of Education, Science, Culture and Technology, obtained with microscope (BX60; Olympus, Tokyo, Japan) or Japan.

References

Askham JM, Vaughan KT, Goodson HV, Morrison EE. elevated levels of protein expression in human breast cancer. (2002). Evidence that an interaction between EB1 and Cancer Res 64: 7370–7376. p150Glued is required for the formation and maintenance of Huelsken J, Behrens J. (2002). The . J a radial microtubule array anchored at the centrosome. Mol Cell Sci 115: 3977–3978. Biol Cell 13: 3627–3645. Jimbo T, Kawasaki Y, koyama R, Sato R, Takada S, Ciani L, Krylova O, Smalley MJ, Dale TC, Salinas PC. (2003). Haraguchi K et al. (2002). Identification of a link between A divergent canonical WNT-signaling pathway regulates the tumour suppressor APC and the superfamily. microtubule dynamics: dishevelled signals locally to stabilize Nat Cell Biol 4: 323–327. microtubules. J Cell Biol 164: 243–253. Kaplan DD, Meigs TE, Kelly P, Casey PJ. (2004). Identifica- Diehl JA, Cheng M, Roussel MF, Sherr CJ. (1998). Glycogen tion of a role for b-catenin in the establishment of a bipolar synthase kinase-3beta regulates cyclin D1 proteolysis and mitotic spindle. J Biol Chem 279: 10829–10832. subcellular localization. Genes Dev 12: 3499–3511. Kielhorn E, Provost E, Olsen D, D’Aquila TG, Smith BL, Guardavaccaro D, Kudo Y, Boulaire J, Barchi M, Busino L, Camp RL et al. (2003). Tissue microarray-based analysis Donzelli M et al. (2003). Control of meiotic and mitotic shows phospho-beta-catenin expression in malignant mela- progression by the F box protein beta-Trcp1 in vivo. Dev Cell noma is associated with poor outcome. Int J Cancer 103: 4: 799–812. 652–656. Gundersen GG, Cook TA. (1999). Microtubules and signal Krylova O, Messenger MJ, Salinas PC. (2000). Dishevelled-1 transduction. Curr Opin Cell Biol 11: 81–94. regulates microtubule stability: a new function mediated by Harwood A, Braga VMM. (2003). Cdc42 and GSK-3: signals glycogen synthase kinase-3b. J Cell Biol 151: 83–94. at the crossroads. Nat Cell Biol 5: 275–277. Ligon LA, Karki S, Tokito M, Holzbaur ELF. (2001). Dynein Hayward DG, Clarke RB, Faragher AJ, Pillai MR, Hagan IM, binds to b-catenin and may tether microtubules at adherens Fry AM. (2004). The centrosomal kinase Nek2 displays junctions. Nat Cell Biol 3: 913–917.

Oncogene Phospho-b-catenin and microtubule regrowth P Huang et al 4371 Liu Y, Hiraiwa Y, Liu E, Kurata H, Thant AA, Matsuda S Reinacher-Schick A, Gumbiner BM. (2001). Apical membrane et al. (2001). Suppression of cell spreading v-Crk localization of the adenomatous polyposis coli tumor requires Ras-MEK-MAP kinase signaling. Oncogene 20: suppressor protein and subcellular distribution of the beta- 5908–5912. catenin destruction complex in polarized epithelial cells. Lucas FR, Goold RG, Gordon-Weeks PR, Salinas PC. J Cell Biol 152: 491–502. (1998). Inhibition of GSK-3b leading to the loss of Ruhul Amin AR, Oo ML, Senga T, Suzuki N, Feng GS, phosphorylated MAP-1B is an early event in axonal Hamaguchi M. (2003). SH2 domain containing protein remodelling induced by WNT-7a or lithium. J Cell Sci phosphatase 2 regulates concanavalin A-dependent 111: 1351–1361. and activation of matrix metalloproteinase 2 via Nigg EA. (2002). Centrosome aberrations: cause or conse- the extracellular signal-regulated kinase and p38 pathways. quence of cancer progression? Nature Reviews Cancer 2: Cancer Res 63: 6334–6339. 815–825. Sadot E, Conacci-Sorrell M, Zhurinsky J, Shnizer D, Lando Z, Olmeda D, Castel S, Vilaro S, Cano A. (2003). b-Catenin Zharhary D et al. (2002). Regulation of S33/S37 phosphoryl- regulation during the cell cycle: implications in G2/M and ated b-catenin in normal and transformed cells. J Cell Sci . Mol Biol Cell 14: 2844–2860. 115: 2771–2780. Ou YY, Mack GJ, Zhang M, Rattner JB. (2002). CEP110 and Sluder G, Nordberg JJ. (2004). The good, the bad and the ninein are located in a specific domain of the centrosome ugly: the practical consequences of centrosome amplifica- associated with centrosome maturation. J Cell Sci 115: tion. Curr Opin Cell Biol 16: 49–54. 1825–1835. Tang QQ, Otto TC, Lane MD. (2003). Mitotic clonal Pellman D. (2001). A CINtillating new job for the APC tumor expansion: a synchronous process required for adipogenesis. suppressor. Science 291: 2555–2556. Proc Natl Acad Sci U S A 100: 44–49. Piel M, Meyer P, Khodjakov A, Rieder CL, Bornens M. Wakefield JG, Stephens DJ, Tavare JM. (2003). A role for (2000). The respective contributions of the mother and glycogen synthase kinase-3 in mitotic spindle dynamics and daughter to centrosome activity and behavior in alignment. J Cell Sci 116: 637–646. cells. J Cell Biol 149: 317–329. Zumbrunn J, Kinoshita K, Hyman AA, Nathke IS. (2001). Quintyne NJ, Schroer TA. (2002). Distinct cell cycle-depen- Binding of the adenomatous polyposis coli protein to dent roles for dynactin and dynein at centrosomes. J Cell microtubules increases microtubule stability and is regulated Biol 159: 245–254. by GSK3 beta phosphorylation. Curr Biol 11: 44–49.

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

Oncogene