3738 Short Report AMER1 regulates the distribution of the tumor suppressor APC between microtubules and the plasma membrane

Annette Grohmann, Kristina Tanneberger, Astrid Alzner, Jean Schneikert and Jürgen Behrens* Nikolaus-Fiebiger-Center for Molecular Medicine, University Erlangen-Nuremberg, Glückstr. 6, 91054 Erlangen, Germany *Author for correspondence (e-mail: [email protected])

Accepted 6 August 2007 Journal of Cell Science 120, 3738-3747 Published by The Company of Biologists 2007 doi:10.1242/jcs.011320

Summary APC is a multifunctional tumor suppressor that redirects APC from microtubule ends to the plasma negatively controls Wnt signaling, but also regulates cell membrane of epithelial cells. Conversely, siRNA-mediated adhesion and migration by interacting with the plasma knockdown of AMER1 reduces the overall levels of APC, membrane and the microtubule cytoskeleton. Although the promotes its association with microtubule ends in cellular molecular basis for the microtubule association of APC is protrusions and disturbs intercellular junctions. These well understood, molecular mechanisms that underlie its data indicate that AMER1 controls the subcellular plasma membrane localization have remained elusive. We distribution of APC between membrane- and microtubule- show here that APC is recruited to the plasma membrane associated pools, and might thereby regulate APC- by binding to APC membrane recruitment 1 (AMER1), a dependent cellular morphogenesis, cell migration and cell- novel membrane-associated protein that interacts with cell adhesion. the ARM repeat domain of APC. The N-terminus of AMER1 contains two distinct phosphatidylinositol(4,5)- Supplementary material available online at bisphosphate [PtdIns(4,5)P2]-binding domains, which http://jcs.biologists.org/cgi/content/full/120/21/3738/DC1 mediate its localization to the plasma membrane. Overexpression of AMER1 increases APC levels and Key words: APC, Cytoskeleton, Tumor suppressor

Introduction interacts with the microtubule-associated protein EB1

Journal of Cell Science The tumor suppressor APC is mutated at an early stage (MAPRE1) (Askham et al., 2000; Smith et al., 1994). APC in the development of most sporadic and inherited colorectal appears to regulate microtubule dynamics as well as the tumors (Polakis, 2000). Wild-type APC negatively regulates interplay of microtubules with the actin network through the Wnt signaling pathway by promoting the proteasomal interactions with cytoskeletal regulators such as the Rho degradation of ␤-catenin, thereby preventing TCF/␤-catenin effector mammalian Diaphanous-related (mDia), the Rac- dependent target gene transcription (Lustig and Behrens, specific guanine nucleotide exchange factor Asef (Arhgef4), 2003). Evidence from a large number of studies indicates that and the Rac effector IQGAP1, the latter two associating with aberrant activation of the Wnt pathway resulting from the N-terminal armadillo (ARM) repeat domain of APC truncating mutations of APC is a central event in colorectal (Kawasaki et al., 2000; Watanabe et al., 2004; Wen et al., tumor formation (Polakis, 2000; Schneikert and Behrens, 2004). 2007). Besides its role in the Wnt pathway, APC has been APC is connected to the plasma membrane at different sites implicated in a variety of cellular functions based on its within the cell (reviewed in Bienz and Hamada, 2004; Hanson association with regulators and components of the and Miller, 2005). It can be detected in cortical clusters at the cytoskeleton. Most prominently, APC interacts with and basal plasma membrane, in association with microtubules and stabilizes microtubules, thereby regulating diverse biological at the tips of cellular protrusions characterized by microtubule processes such as axon outgrowth, cell migration and mitosis ends in migrating cells (Barth et al., 2002; Reilein and Nelson, (Hanson and Miller, 2005; Zumbrunn et al., 2001). Several of 2005). At cellular protrusions, APC can be anchored to the these functions appear to be altered by tumor-associated APC membrane via interaction of its C-terminal PDZ-binding motif mutations, indicating that the tumor suppressor activity of APC with the peripheral membrane protein DLG1 (Etienne- extends beyond the regulation of Wnt signaling (Fodde, 2003). Manneville et al., 2005; Mimori-Kiyosue et al., 2007). The N- Functional evidence in different cellular contexts suggests terminal ARM repeats and the ␤-catenin-binding domains of that the microtubule plus-end association of APC is required APC also play a role in the localization of APC at the tip for the establishment of cell polarity during directed cell structures (Sharma et al., 2006). Outside of these microtubule- migration (Etienne-Manneville et al., 2005; Kawasaki et al., associated clusters, APC has a more global distribution at the 2003; Nathke et al., 1996; Watanabe et al., 2004; Wen et al., plasma membrane. The ARM repeat domain was shown to be 2004). A basic domain in the C-terminal third of APC can involved in the lateral membrane targeting of APC both in directly associate with microtubules, and an adjacent domain Drosophila and mammalian epithelial cells (Hamada and AMER1 regulates the distribution of APC 3739

Bienz, 2002; Langford et al., 2006; McCartney et al., 1999; molecules capable of linking APC to the plasma membrane McCartney et al., 2006). This localization can be promoted by have not yet been identified, precluding the in-depth analysis destruction of microtubules, suggesting that APC can exchange of the APC function in this compartment. Here, we report the between the microtubule-associated and the lateral plasma identification and characterization of the AMER , membrane pools (Rosin-Arbesfeld et al., 2001). Missense which link APC to the plasma membrane thereby preventing mutations or deletions of individual amino acids in the ARM association of this protein with microtubule ends. repeats of Drosophila epithelial APC (E-APC)/APC2 abolish the membrane localization of E-APC and result in defects of Results cadherin/catenin-based cell-cell adhesion and in the tethering AMER1 binds to the ARM repeat domain of APC of mitotic spindles to cortical actin (Hamada and Bienz, 2002; We identified the novel protein AMER1 as an APC McCartney et al., 2001). The mechanisms by which APC interaction partner in a yeast two-hybrid screen using the associates with the plasma membrane outside microtubule- ARM repeat domain of human APC (APC-ARM, amino acids associated clusters are largely unknown. In particular, 308-789, Fig. 1C) as a bait. We isolated three independent Journal of Cell Science

Fig. 1. Structure of AMER1, AMER1(short) and AMER2, and their interaction with the ARM repeats of APC. (A) Scheme of human AMER1, AMER1(short) and AMER2. The APC-interacting sequences are indicated by gray shading. In AMER1(short), the C-terminal amino acids that are different from AMER1 are indicated in black. (B) Amino acid sequence of human AMER1 and AMER1(short). APC-interacting sequences are shaded, glutamic acid-rich and proline-rich sequences are underlined, and the REA repeats are in italics. For AMER1(short), only the amino acids from position 786 onwards, which are different to AMER1, are shown. This sequence is encoded by a separate 3Ј exon (see text for details). (C, upper panel) Interaction of murine Amer1 sequences #1-3 with the human APC ARM repeat region (APC-ARM) as well as with an asparagine-to-lysine substitution mutant (APC- ARMN507K) in quantitative yeast two-hybrid assays. The seven ARM repeats are indicated by shaded boxes. Values represent ␤- galactosidase units of representative experiments. (C, lower panel) Interaction of human AMER2 APC-binding sites #1 and 2 with APC- ARM in quantitative yeast two-hybrid assays. Empty DNA-binding- domain vector was used as a control. Values represent ␤-galactosidase units of representative experiments. Binding sites #1 and 2 produced similar high ␤-galactosidase values upon interaction with APC-ARM when tested in the DNA-binding-domain vector (not shown). 3740 Journal of Cell Science 120 (21)

sequences (#1-3) of Amer1, which are all located in the the yeast two-hybrid data (cf. Fig. 1C), the APC-ARMN507K central region of the protein but do not overlap and share no mutant showed diminished interaction with AMER1 and amino acid sequence similarity (Fig. 1A,B). The sequences AMER1(short) in co-immunoprecipitation experiments as interacted with similar efficiency with the ARM repeats of compared with wild-type APC-ARM (Fig. 2D). APC-ARM APC-ARM (Fig. 1C, upper panel). Interestingly, an co-immunoprecipitated with AMER1 fragments containing at asparagine-to-lysine substitution in the second ARM repeat least one of the APC-interacting domains [i.e. with AMER1, of APC-ARM abolished or significantly reduced its AMER1(2-839), AMER1(2-601) and AMER1(319-716)] but interaction with the AMER1 sequences (Fig. 1C, upper not with fragments lacking such domains [i.e. with AMER1(2- panel). The corresponding mutation of Drosophila E-APC 321), AMER1(531-716) and AMER1(833-1135)] (Fig. 2E). has been shown to disrupt its membrane localization (Hamada Together, our results show that AMER1 specifically interacts and Bienz, 2002). with APC via three independent binding domains, and that the The coding sequences of both human and mouse AMER1 two proteins form endogenous complexes in mammalian cells. reside on single exons located on syntenic regions of the X APC was also co-immunoprecipitated with EGFP-tagged of both species (Xq11.1 and XC3, respectively). AMER2 in 293T cells (supplementary material Fig. S2C). AMER1 consists of 1135 amino acids. It lacks known functional protein domains, but contains conspicuous glutamic AMER1 controls APC levels acid-rich and proline-rich stretches as well as a series of Cells transiently expressing AMER1 showed increased levels arginine-glutamic acid-alanine (REA) repeats (Fig. 1B). A of co-expressed APC (Fig. 2A), as well as APC-ARM (Fig. human cDNA clone (FLJ39827) annotated in the NCBI 2E), as compared with control transfectants, whereas the Reference Sequence (RefSeq) database is identical to the amount of a co-expressed unrelated protein, FKBP8, remained AMER1 coding sequence up to 2356, after which 59 unchanged (Fig. 2E, lower panel and lower scheme). AMER1 coding base pairs, present on a downstream exon, are added by fragments lacking APC interaction domains, such as splicing (supplementary material Fig. S1A). This isoform lacks AMER1(2-321), AMER1(531-716) and AMER1(833-1135), a major part of the third APC-interacting fragment and was did not increase APC-ARM levels, indicating that AMER1 named AMER1(short) (Fig. 1A,B). requires direct interaction for stabilization of APC (Fig. 2E). AMER1 is ubiquitously expressed in human tissues similar Importantly, the amounts of endogenous APC were also to APC (supplementary material Fig. S1B). Flag-tagged and significantly increased in MDCK cell clones stably expressing endogenous AMER1 protein ran at the same position of about EGFP-tagged AMER1 as compared to EGFP-expressing 190 kDa in anti-AMER1 western blots (supplementary controls, or untransfected MDCK cells (Fig. 2F and data not material Fig. S1C). Moreover, expression of the endogenous shown). Conversely, knockdown of AMER1 in 293T cells by 190 kDa protein was reduced after treatment with siRNA different siRNA oligonucleotides led to a reduction of oligonucleotides against AMER1, thus proving the authenticity endogenous APC protein levels (Fig. 2G), whereas mRNA of the 190 kDa band as AMER1. AMER1 was detected in levels of APC were not changed (data not shown). lysates from several human cancer cell lines by western

Journal of Cell Science blotting (supplementary material Fig. S1C). AMER1 localizes to the plasma membrane and recruits The NCBI RefSeq database contains another protein APC sequence with high similarity to AMER1 (FAM123A), which In MCF-7 cells, EGFP-tagged AMER1 was localized at the we named AMER2 (Fig. 1A). The coding sequence of AMER2 plasma membrane, whereas EGFP was diffusely distributed is located on 13. According to the database, two (Fig. 3A, upper panels). Flag-tagged AMER1 also localized to isoforms of AMER2 are generated by alternative splicing the membrane as determined by immunofluorescence staining, (transcript variants 1 and 2; supplementary material Fig. S2A). which required permeabilization of fixed cells, indicating that We cloned transcript variant 1, which encodes a protein sharing AMER1 associates with the cytoplasmic side of the plasma 26% amino acid identity with AMER1. Interestingly, this membrane (data not shown). Similarly, EGFP- or Flag-tagged isoform contains stretches with 100% amino acid identity to AMER2 localized to the plasma membrane of MCF-7 cells AMER1 within two regions analogous to the APC-binding (Fig. 3A, and data not shown). AMER proteins were enriched sites #1 and #2 (supplementary material Fig. S2B). These in but not restricted to lateral membranes. AMER1 fragments regions of AMER2 interacted with APC-ARM in yeast (Fig. comprising amino acids 2-601 and 2-209 associated with the 1C, lower panel). By BLAST searches, homologs of AMER1 plasma membrane similar to full-length AMER1 (Fig. 3A, and AMER2 were found in mouse, chicken, Xenopus and upper panel), whereas deletion fragments of AMER1 lacking zebrafish, but not in Drosophila. the N-terminus, such as AMER1(207-839), did not localize to the plasma membrane but were distributed throughout the cell. AMER1 forms complexes with APC in mammalian cells Thus, the N-terminal domain of AMER1, containing amino In SW480 colon carcinoma cells, endogenous (truncated) APC acids 2-209, is necessary and sufficient for the membrane (APCmut) as well as exogenous wild-type APC (APC) could localization of this protein. be specifically co-immunoprecipitated with Flag-tagged Importantly, both AMER1 and AMER1(short), as well as AMER1 (Fig. 2A). Reciprocally, endogenous as well as Flag- AMER2, recruited co-expressed APC from filamentous tagged AMER1 were co-immunoprecipitated with EGFP- structures, which probably represented microtubules tagged APC-ARM in 293T cells (Fig. 2B). Importantly, (Munemitsu et al., 1994; Smith et al., 1994), to the plasma AMER1 was also co-immunoprecipitated with APC from non- membrane (Fig. 3A, lower panels and data not shown). transfected 293T or SW480 cells, demonstrating the presence AMER1-mediated membrane recruitment of APC required the of endogenous AMER1-APC complexes (Fig. 2C). In line with APC-binding sites and the membrane docking N-terminus of AMER1 regulates the distribution of APC 3741

AMER1, because the mutants AMER1(2-209) and associated with the tips of microtubules, in contrast to AMER1(207-839), which lack these domains, respectively, did filamentous staining in control cells. The C-terminal half of not recruit APC to the plasma membrane. Interestingly, in AMER1 was not required for APC recruitment, as AMER1(2-209)-expressing cells, APC was frequently demonstrated by AMER1(2-601) (Fig. 3A). Both AMER1 and

Fig. 2. Interaction between AMER1 and APC. (A) Co-immunoprecipitation of wild-type APC (APC) and endogenous mutant APC (APCmut) with Flag-tagged AMER1 after transient transfections of SW480 cells, as indicated. Western blottings were performed using anti-APC Ab1 (for APCmut) and Ab2 (for APC), and anti- Flag antibodies. The double band for Flag-AMER1 is observed in some but not all experiments (cf. B) and might result from incomplete denaturation of the protein in the gel sample buffer. (B) Co-immunoprecipitation of endogenous AMER1 and Flag-tagged AMER1 with EGFP-tagged APC- ARM after transient transfections of 293T cells, as indicated. Western blottings were performed using anti- AMER1 or anti-GFP antibodies. (C) Co-immunoprecipitation of endogenous AMER1 with APC from lysates of nontransfected 293T and SW480 cells. Immunoprecipitations were performed with anti-GFP antibody as a control or with anti-APC antibody Ali followed by western blotting using the anti-AMER1 antibody or Ali. (D) Co- immunoprecipitation of EGFP-tagged APC-ARM or APC-ARMN507K with Journal of Cell Science Flag-tagged AMER1 or AMER1(short) after transient transfections of 293T cells as indicated. Western blottings were performed using anti-GFP or anti-Flag antibodies. (E) Co- immunoprecipitation of Flag-tagged APC-ARM with EGFP-tagged AMER1 and AMER1 deletion fragments after transient transfections of 293T cells as indicated. Western blottings were performed using anti- Flag or anti-GFP antibodies. The bottom blot shows levels of the Flag- FKBP8 control protein in lysates of 293T cells after co-expression with the indicated AMER1 constructs. The scheme below shows the structure of the deletion mutants of AMER1 and quantification from separate experiments of the fold change of APC-ARM or FKBP8 protein levels (as a control) after co-transfection with the indicated AMER1 constructs, relative to EGFP transfection. (F) Western blotting for APC (antibody Ali) in stable clones of MDCK cells expressing EGFP (EGFP#1, EGFP#2) or EGFP-tagged AMER1 (EGFP-AMER1#1, EGFP-AMER1#2). (G) Western blotting for APC (antibody Ali), AMER1, Pan- cadherin and FKBP8 from lysates of 293T cells transiently transfected with siRNA oligonucleotides against GFP (as a control), or two different siRNA oligonucleotides against AMER1 (siAMER1a, c). The numbers below the lanes indicate relative protein levels normalized to Pan- cadherin, with siGFP controls set to 100%. 3742 Journal of Cell Science 120 (21)

AMER1(short) also recruited the ARM repeat domain of APC 285), but not GST-AMER1(260-545), bound to alone (APC-ARM) to the plasma membrane. The N507K phosphatidylinositol mono-, bis- and tris-phosphate species, substitution of APC-ARM barely reduced the recruitment by and to 3-sulfogalactosylceramide (sulfatide) in vitro (Fig. 3B). AMER1 but impaired the recruitment by AMER1(short) Interestingly, GST-AMER1 fragments comprising amino acids (supplementary material Fig. S3). Notice that AMER1(short) 2-142 or 143-209 also bound to membrane lipids, indicating lacks a large part of the third APC-binding domain of AMER1. that AMER1 contains two lipid-binding sites (Fig. 3B). Because AMER1 lacks any obvious membrane-anchoring Phosphatidylinositol(4,5)-bisphosphate [PtdIns(4,5)P2] is the domains, we next investigated whether it would bind directly predominant phosphoinositide at the plasma membrane (Di to membrane lipids. Recombinant GST-tagged AMER1(2- Paolo and De Camilli, 2006) and therefore is a good candidate

Fig. 3. AMER1 localizes to the membrane via binding to PtdIns(4,5)P2. (A) Double staining of EGFP, EGFP-AMER1, EGFP- AMER1 deletion constructs as in Fig. 2E or EGFP-AMER2 (upper panels, GFP fluorescence), and APC (lower panels, antibody Ali immunofluorescence) in Journal of Cell Science MCF-7 cells transiently transfected as indicated above the panels. Arrowheads denote the membrane, and arrows the filamentous localizations. (B) Membrane lipid-binding assays of AMER1 deletion mutants. Membrane lipid strips were incubated with the indicated recombinant GST-AMER1 fusion proteins, revealing two phosphoinositide binding sites. DAG, diacylglycerol; PA, phosphatidic acid; PS/PE/PC/PG, phosphatidyl-serine/- ethanolamine/-choline/-glycerol. (C) Localization of AMER1 (a-e) and APC (aЈ-cЈ) or AMER1 deletion mutants (f-k) in transiently transfected MCF7 cells. Double staining of EGFP-AMER1 (a-c, GFP fluorescence) and APC (aЈ-cЈ, anti-M-APC immunofluorescence) with (b,bЈ) and without (a,aЈ) prior ionomycin treatment (Iono), or with ionomycin treatment followed by EGTA treatment (c,cЈ). Staining of EGFP-AMER1 in cells treated with wortmannin (Wm) prior to ionomycin/EGTA treatment (d) or with neomycin (Neo) prior to ionomycin (e). Staining of EGFP-AMER1(2-142) (f-h) or EGFP-AMER1(143-209) (i-k) in cells treated with ionomycin/EGTA as indicated. AMER1 regulates the distribution of APC 3743

for a physiological membrane link for AMER1. We analyzed thereby prevents resynthesis of PtdIns(4,5)P2 (Nakanishi et al., whether cleavage of PtdIns(4,5)P2 by phospholipase C (PLC) 1995) (Fig. 3Cd). Furthermore, pretreatment with neomycin, affects AMER1 localization in cells. Membrane localization of which affects the turnover of PtdIns(4,5)P2 (Gabev et al., ectopically expressed full-length AMER1 and APC, and of the 1989), impaired the ionomycin-induced dissociation of deletion fragments AMER1(2-142) and AMER1(143-209), in AMER1 from the plasma membrane (Fig. 3Ce). These results MCF-7 cells was completely abolished upon treatment with suggest that AMER1 localizes to the plasma membrane by 2+ ionomycin, which induces Ca influx and thereby activates direct interaction with PtdIns(4,5)P2 via two distinct binding PLC (Varnai and Balla, 1998) (Fig. 3Cb,bЈ,g,j). AMER1 and domains at the N-terminus. APC localized again to the plasma membrane when Ca2+ was chelated with EGTA after ionomycin treatment (Fig. AMER1 controls the distribution of APC between 3Cc,cЈ,h,k). Relocalization at the membrane was prevented by microtubules and the plasma membrane treatment of cells with wortmannin, which, at high In MDCK clones stably expressing AMER1, endogenous APC concentrations, inhibits phosphatidylinositol 4-kinases and delocalized from focal cellular protrusions that had been previously characterized as microtubule ends (Nathke et al., 1996; Rosin-Arbesfeld et al., 2001) and became recruited to the plasma membrane (Fig. 4Aa,b). As has been shown previously, disruption of microtubules by nocodazole treatment resulted in membrane association of endogenous APC in MDCK cells [Fig. 4B (cf. Rosin-Arbesfeld et al., 2001)]. Treatment with ionomycin, which blocks AMER1 membrane association (cf. Fig. 3C), prevented the nocodazole-induced membrane localization of APC (Fig. 4B), which is in line with APC being linked to the membrane by endogenous AMER1 in these cells. Next, we analyzed whether loss of endogenous AMER1 in MCF-7 cells by RNAi would promote microtubule association of APC. Transient transfection of the siAMER1c oligonucleotide in MCF-7 cells resulted in an overall downregulation of AMER1 mRNA by about 30%, which corresponds to the transfection efficiency obtained in these cells (supplementary material Fig. S5A). After transient knockdown of AMER1 with siAMER1c, 13.2% (n=657) of cells at the edges of colonies showed staining Journal of Cell Science Fig. 4. AMER1 controls the distribution of APC between microtubules and the plasma membrane. (A) Localization of APC (a,b) in MDCK cells stably expressing EGFP (a,aЈ) or EGFP-tagged AMER1 (b,bЈ). (a,b) Immunofluorescence stainings using the anti-APC antibody Ali; (aЈ,bЈ) corresponding EGFP fluorescence. Notice the membrane association of EGFP-AMER1, which is not observed for EGFP. Arrowheads point to APC at cytoplasmic clusters at cellular protrusions in the EGFP transfectants (a) and to colocalization of AMER1 and APC at the plasma membrane in the EGFP-AMER1 transfectants (b,bЈ). Insets in upper panels represent higher magnifications. (B) Immunofluorescence staining of APC (anti-M-APC) in MDCK cells treated with solvent (DMSO), nocodazole (Noco), or nocodazole followed by ionomycin (Noco/Iono). Arrowheads indicate lateral plasma membranes. (C) Immunofluorescence staining of APC (anti-M-APC) in MCF-7 cells treated with siRNA against either GFP as a control (siGFP), AMER1 (siAMER1c), or AMER1 and APC (siAMER1c+siAPC). Arrowheads indicate tips of cellular protrusions. (D) Staining of transiently transfected APC (a-c), microtubules (aЈ,bЈ, ‘MT’) or AMER1 (cЈ) in MCF-7 cells transiently transfected with APC together with siGFP (a,aЈ), siAMER1c (b,bЈ), or siAMER1c and the siAMER1c-insensitive EGFP-rAMER1 expression construct (c,cЈ). (a,aЈ;b,bЈ;c,cЈ) Double stainings. (a-c) Anti-M-APC immunofluorescence; (aЈ,bЈ) anti-␣-tubulin immunofluorescence; (cЈ) GFP fluorescence. Broken lines indicate the edge of colonies. 3744 Journal of Cell Science 120 (21)

of APC in tips of cellular protrusions, often corresponding to microtubule ends (Nathke et al., 1996). This staining was specific because it was abolished by concomitant treatment with siAPC (Fig. 4C, supplementary material Fig. S5B). In cells treated with control siRNA (siGFP), only 3.9% of cells (n=751) showed such tip staining. Alterations in the membrane association of APC could not be reliably determined, because the membrane staining with the anti-M-APC serum appeared to be nonspecific because it was not affected by siRNA against APC, in contrast to APC staining at the protrusions. Transiently transfected APC localized to filamentous structures partially overlapping with microtubules in most of the siGFP-treated MCF-7 control cells (Fig. 4Da,aЈ and Fig. 3A). In 25.5% of transfected cells at the edge of colonies, APC was found at the tips of cellular protrusions (n=106). This localization was markedly increased to 56.8% of the cells in siAMER1c-treated cultures (Fig. 4D,b; n=111). Moreover, in these structures APC overlapped with the ends of microtubules, which were frequently reoriented perpendicular to the membrane indicating a rearrangement of the microtubular network (Fig. 4DbЈ). To verify that tip localization of APC was caused by the specific depletion of AMER1, we additionally expressed a mutated AMER1 cDNA resistant to knockdown by siAMER1c (rAMER1, supplementary material Fig. S5C,D). Tip localization of APC was completely abolished in rAMER1- expressing cells (n=30), and APC was exclusively observed at the cell membrane co-localizing with rAMER1 (Fig. 4Dc,cЈ). These data demonstrate that silencing of endogenous AMER1 Fig. 5. AMER1 controls intercellular junctions together with APC. (A) Immunofluorescence staining of E-cadherin in MCF-7 cells promotes association of APC with the tips of cellular treated with a control siRNA against GFP (siGFP), siRNA against protrusions, in which it overlaps with microtubule ends. AMER1 (siAMER1c), or siAMER1c together with an siRNA against APC (siAMER1c+siAPC). Arrowheads indicate disrupted cell AMER1 controls intercellular junctions together with junctions. (B) Relative number of gaps between MCF-7 cells APC transfected with siRNA oligonucleotides as indicated below the bars. MCF-7 cells are epithelial and show the classical cobblestone Gaps were counted in E-cadherin- and ␤-catenin-stained samples in ϫ Journal of Cell Science pattern when stained for E-cadherin or ␤-catenin (Fig. 5A). 20 optical fields using the 40 objective. Results show the When AMER1 was transiently knocked down by transfection mean±s.d. of at least two independent experiments. with the siAMER1c oligonucleotide, the cell junctions of MCF-7 cells were frequently disrupted, resulting in ‘gaps’ between cells (Fig. 5A,B). This was most apparent at sites at of APC associated with microtubule ends. Finally, a point which multiple cells met (e.g. at tricellular junctions). mutation, which was shown to reduce membrane association Interestingly, membrane association of E-cadherin, ␤-catenin of Drosophila E-APC, in the ARM repeat domain also and other junctional proteins was still preserved (Fig. 5A and diminished the AMER1-APC interaction. data not shown). The additional knockdown of APC led to a AMER1 was localized at the plasma membrane, and the N- more severe disruption of cellular junctions, with cells terminus of AMER1 is necessary and sufficient for this frequently showing complete detachment from each other (Fig. localization. We could identify two distinct sites (amino acids 5A,B), whereas knockdown of APC alone had only a minor 2-142 and 143-209) that mediate membrane association and effect (Fig. 5B). These data indicate a role for AMER1 in directly bind PtdIns(4,5)P2. These sites have high isoelectric maintaining the integrity of intercellular junctions, probably by points (pI=9.3 and 10.55, respectively; ProtParam) compared mediating the membrane localization of APC. with full-length AMER1 (pI=4.77), contain highly conserved basic and aromatic residues and are particularly enriched in Discussion lysine residues (11.3 and 14.9%, respectively, compared with In this paper, we provide a novel molecular mechanism by 4.1% in full-length AMER1; supplementary material Fig. S2B). which APC can be recruited to the plasma membrane by Basic and aromatic amino acids have been shown to mediate interacting with the AMER1 and AMER2 proteins. When PtdIns(4,5)P2 binding in other proteins (Kagan and Medzhitov, overexpressed, AMER proteins recruited both exogenous and 2006). The majority of these lysine residues are conserved in endogenous APC to the plasma membrane and prevented its AMER2 (supplementary material Fig. S4), which can also interaction with microtubules. Depletion of PtdIns(4,5)P2 by interact with the plasma membrane. We could further show that ionomycin treatment, which abolishes AMER1 localization at activation of phospholipase C by ionomycin-induced Ca2+ plasma membranes, also reduced the association of APC with influx leads to the delocalization of AMER1 and APC from the the membrane. Endogenous AMER1-APC complexes were plasma membrane, probably because of PtdIns(4,5)P2 cleavage. identified, and knockdown of AMER1 increased the fraction These data strongly support a role for PtdIns(4,5)P2 as the AMER1 regulates the distribution of APC 3745

physiological membrane anchor for AMER1. AMER1 was not predominantly seen at vertices in which more than two cells restricted to a specific membrane compartment in the cell lines contact each other. Possibly, adherens junctions at these used in this study, which is in line with its association with locations are more vulnerable than at the lateral domains of two membrane lipids. However, local PtdIns(4,5)P2 concentrations neighboring cells. AMER1 might compete with cytoskeletal are regulated by signaling events and are significantly enriched regulators, such as Asef (Kawasaki et al., 2003), for binding to at the apical plasma membrane in polarized epithelial cells APC, and loss of AMER1 would allow signaling by these APC (Martin-Belmonte et al., 2007; McLaughlin et al., 2002). Such complexes, leading to the observed effects. Alternatively, changes might influence AMER1 localization and thus control localization of APC at the plasma membrane might be directly the subcellular distribution of APC. required for cell junction formation (e.g. by promoting ␤- AMER1 directly interacted with the ARM repeat domain of catenin recruitment to cadherins), as was suggested previously APC via three separate binding sites, which were identified in (Hamada and Bienz, 2002). Because both E-cadherin and ␤- our initial yeast two-hybrid screen. The sites interact with catenin membrane localization were not strongly altered in similar efficiency with APC yet do not share any obvious siAMER1-treated cells, such a function is rather unlikely in our sequence similarity. Other known interaction partners of the experimental setting. The fact that simultaneous knockdown of ARM repeat domain of APC, such as Asef, KAP3 (KIFAP3) APC aggravated the junctional defects of the AMER1 and IQGAP1, also show no sequence similarity to each other knockdown suggests a direct and positive role of both proteins or to AMER1, indicating that the ARM repeats can contact in junction formation and/or maintenance. The identification various partners using different binding modes. This is similar of AMER1 as a plasma membrane link of APC will allow us to ␤-catenin, in which the ARM repeats can interact with to analyze more specifically the function of this important various interaction partners using distinct amino acids (von tumor suppressor in polarized cell migration and cell-cell Kries et al., 2000). It is possible that the presence of multiple adhesion. APC-binding domains in AMER1 increases the overall affinity After this paper had been submitted, a report described of the interaction. Moreover, the differences in sequence WTX as a negative regulator of the Wnt signaling pathway, between these domains might allow for individual modulation which forms complexes with APC, Axin, ␤-catenin and ␤- of APC binding by biochemical modifications or competing TRCP (Major et al., 2007). WTX is identical to AMER1, interactions. Interestingly, the third APC-interaction domain of indicating that the ␤-catenin destruction complex as a whole AMER1 retained significant capacity to interact with the might become recruited to the plasma membrane, which might N507K mutant of the ARM repeat domain in the yeast two- affect Wnt signaling. WTX/AMER1 is mutated in Wilms hybrid assays. In AMER1(short), this domain is largely tumors, leading to truncations or deletions of the complete missing and membrane recruitment of the N507K mutant by gene (Rivera et al., 2007). It was suggested that loss of WTX AMER1(short) was reduced, whereas it was largely preserved leads to aberrant activation of Wnt signaling and thereby in AMER1, at least under conditions of overexpression. This promotes tumorigenesis (Major et al., 2007). From our data, indicates that, depending on the expression of the AMER1 these mutations might additionally contribute to tumorigenesis isoform, APC might be differentially recruited to the plasma by abolishing functions of APC at the plasma membrane and

Journal of Cell Science membrane. activating those at microtubules. Two main cellular functions have been assigned to peripheral APC. First, several reports have indicated a role for Materials and Methods APC in the polarization of migrating cells via its association DNA constructs and transfections with the cytoskeleton (Etienne-Manneville et al., 2005; For EGFP-tagging, cDNAs were inserted into pEGFP-C3 (Clontech); for GST- Kawasaki et al., 2003; Kroboth et al., 2007; Nathke et al., 1996; tagging they were inserted into pGEX-4T3 (Amersham Pharmacia Biotech); and for Watanabe et al., 2004; Wen et al., 2004). siAMER1 treatment Flag-tagging into pcDNA-Flag, which was derived from pcDNA3.1 (Invitrogen) by inserting the Flag peptide coding sequence into the multiple cloning site. APC increased the fraction of cells harboring APC at microtubules cDNA fragments were obtained by PCR amplification using pCMV-APC (Smith et in cellular protrusions, which was reverted by expression of an al., 1994) as a template, which was also used for expression of human full-length siRNA-resistant cDNA of AMER1. Plasma membrane APC. The N507K mutant of APC-ARM was created by PCR mutagenesis. Full- length human AMER1 was cloned by replacing the 3Ј coding sequence of clone recruitment of APC by AMER1 might therefore block cell FLJ39827 (NITE Biological Resource Center, Chiba, Japan) at nt 2073 with the polarization and counteract cell migration. Second, it was corresponding coding sequence of AMER1 obtained by reverse transcriptase (RT)- suggested that plasma membrane association of APC controls PCR from mRNA of 293T cells (cf. supplementary material Fig. S1A). Deletion cell-cell adhesion and adherens junction formation (Faux et al., mutants of AMER1 were generated by restriction digests or PCR amplification. rAMER1 cDNA was generated by PCR mutagenesis exchanging three nucleotides 2004; Hamada and Bienz, 2002). Mutations within the ARM of the targeting sequence of siAMER1c without changing the amino acid sequence. repeat domain leading to reduced membrane association of Full-length AMER2 transcript variant 1 was obtained by RT-PCR from mRNA of Drosophila E-APC results in the disruption of cell junctions 293T cells. APC-binding sites #1 and #2 of AMER2 were obtained by RT-PCR from mRNA of Caki cells. Transient transfections of plasmids were performed using (Hamada and Bienz, 2002). However, this might result from a ESCORT IV (Sigma), and of siRNA using TransIT-TKO (Mirus, Madison, WI, dominant action of the mutants, because null mutations did not USA). affect cell junctions (McCartney et al., 2006). Furthermore, ARM repeat mutations also diminish the interaction with Asef, siRNA oligonucleotides The sequences of the siRNA oligonucleotides are: siAPC, 5Ј-GACGUUGCG - which could have an impact on the regulation of the AGAAGUUGGAdTdT-3Ј; siGFP, 5Ј-GCUACCUGU UC CAUGGCCAdTdT-3Ј cytoskeleton and thereby affect cellular junctions (Watanabe et (Eurogentec, Seraing, Belgium). siAMER1a, siAMER1b and siAMER1c were al., 2004). We observed disturbance of the integrity of cell purchased from Dharmacon (catalog numbers D-016075-01, D-016075-02 and D- junctions in MCF-7 cells treated with the siAMER1c 016075-03). oligonucleotide, which could result from diminished Cell culture and drug treatments membrane localization of APC. Interestingly, the effect was Cells were cultured in 10% CO2 at 37°C in DMEM supplemented with 10% FCS 3746 Journal of Cell Science 120 (21)

and 1% penicillin/streptomycin. For stable expression of EGFP-AMER1 and EGFP, Behrens, J., von Kries, J. P., Kuhl, M., Bruhn, L., Wedlich, D., Grosschedl, R. and MDCK cells were transfected with pEGFP-AMER1 or pEGFP-C3, respectively, and Birchmeier, W. (1996). Functional interaction of beta-catenin with the transcription subsequently selected in medium containing 1 mg/ml geneticin (G418, Invitrogen). factor LEF-1. Nature 382, 638-642. For drug treatments, cells were incubated for 5 minutes at 37°C with 10 ␮M Behrens, J., Jerchow, B. A., Wurtele, M., Grimm, J., Asbrand, C., Wirtz, R., Kuhl, ionomycin (Calbiochem), followed by 10 minutes at 37°C with 2 mM EGTA in M., Wedlich, D. and Birchmeier, W. (1998). Functional interaction of an axin culture medium where indicated, or cells were incubated for 30 minutes at 37°C homolog, conductin, with beta-catenin, APC, and GSK3beta. Science 280, 596-599. with 10 ␮M wortmannin, or 10 mM neomycin (Calbiochem) in culture medium Bienz, M. and Hamada, F. (2004). Adenomatous polyposis coli proteins and cell prior to ionomycin treatment. For microtubule disruption, cells were incubated with adhesion. Curr. Opin. Cell Biol. 16, 528-535. Di Paolo, G. and De Camilli, P. (2006). Phosphoinositides in cell regulation and 10 ␮g/ml nocodazole (Sigma) for 15 minutes on ice followed by 60 minutes at 37°C, membrane dynamics. Nature 443, 651-657. or with solvent containing medium as controls. Etienne-Manneville, S., Manneville, J. B., Nicholls, S., Ferenczi, M. A. and Hall, A. (2005). Cdc42 and Par6-PKCzeta regulate the spatially localized association of Dlg1 Antibodies and APC to control cell polarization. J. Cell Biol. 170, 895-901. The AMER1-specific monoclonal mouse antibody was raised against amino acids Faux, M. C., Ross, J. L., Meeker, C., Johns, T., Ji, H., Simpson, R. J., Layton, M. J. 2-285 of recombinant human AMER1 generated as a GST fusion in bacteria. The and Burgess, A. W. (2004). Restoration of full-length adenomatous polyposis coli anti-APC antibody Ali and the rabbit anti-M-APC serum were a gift from Inke S. (APC) protein in a colon cancer cell line enhances cell adhesion. J. Cell Sci. 117, 427- Näthke (University of Dundee, Dundee, UK). The rabbit anti-FKBP8 antibody was 439. a gift of Frank Edlich and Gunter Fischer (Max-Planck Research Unit for Fodde, R. (2003). The multiple functions of tumour suppressors: it’s all in APC. Nat. Enzymology of Protein Folding, Halle, Germany). Commercial antibodies were Cell Biol. 5, 190-192. obtained from Calbiochem (APC Ab-1 and Ab-2), Sigma (␤-Actin: clone AC-15; Frangioni, J. V. and Neel, B. G. (1993). Solubilization and purification of enzymatically rabbit anti-Flag polyclonal antibody; mouse anti-GST monoclonal antibody; rabbit active glutathione S-transferase (pGEX) fusion proteins. Anal. Biochem. 210, 179-187. anti-Pan-cadherin serum), Roche (GFP, mixture of clones 7.1 and 13.1), Serotec (␣- Gabev, E., Kasianowicz, J., Abbott, T. and McLaughlin, S. (1989). Binding of tubulin, clone YL1/2) and TaKaRa (mouse anti-E-cadherin monoclonal antibody, neomycin to phosphatidylinositol 4,5-bisphosphate (PIP2). Biochim. Biophys. Acta 979, 105-112. clone HECD-1). Secondary antibodies (Jackson ImmunoResearch, Cambridgeshire, Hamada, F. and Bienz, M. (2002). A Drosophila APC tumour suppressor homologue UK) were Cy2, Cy3 or Cy5 conjugates for immunofluorescence and HRP functions in cellular adhesion. Nat. Cell Biol. 4, 208-213. conjugates for western blotting. Hanson, C. A. and Miller, J. R. (2005). Non-traditional roles for the Adenomatous Polyposis Coli (APC) tumor suppressor protein. Gene 361, 1-12. Yeast two-hybrid screen Kagan, J. C. and Medzhitov, R. (2006). Phosphoinositide-mediated adaptor recruitment Yeast two-hybrid and ␤-galactosidase assays were performed in the L40 yeast strain controls Toll-like receptor signaling. Cell 125, 943-955. using pBTM116 as a bait vector and a mouse embryonic day 10.5 cDNA library in Kawasaki, Y., Senda, T., Ishidate, T., Koyama, R., Morishita, T., Iwayama, Y., pVP16 as described previously (Behrens et al., 1998). Higuchi, O. and Akiyama, T. (2000). Asef, a link between the tumor suppressor APC and G-protein signaling. Science 289, 1194-1197. Preparation of protein lysates, immunoprecipitation and Kawasaki, Y., Sato, R. and Akiyama, T. (2003). Mutated APC and Asef are involved in the migration of colorectal tumour cells. Nat. Cell Biol. 5, 211-215. western blotting Kroboth, K., Newton, I. P., Kita, K., Dikovskaya, D., Zumbrunn, J., Waterman- Cells were washed three times with PBS and lysed in Triton X-100 buffer (20 mM Storer, C. M. and Nathke, I. S. (2007). Lack of adenomatous polyposis coli protein Tris-HCl, pH 7.4, 150 mM NaCl, 5 mM EDTA, 1% Triton X-100, 1 mM DTT and correlates with a decrease in cell migration and overall changes in microtubule stability. 1 mM PMSF) at 4°C for 10 minutes. Lysates were cleared at 16,000 g for 10 minutes Mol. Biol. Cell 18, 910-918. at 4°C. For co-immunoprecipitation, lysates were incubated for 4 hours at 4°C with Langford, K. J., Askham, J. M., Lee, T., Adams, M. and Morrison, E. E. (2006). the appropriate antibody and protein A/G-Sepharose beads (Santa Cruz Examination of actin and microtubule dependent APC localisations in living Biotechnology), or with anti-FLAG M2 affinity gel beads (Sigma). mammalian cells. BMC Cell Biol. 7, 3. Immunoprecipitates were collected, washed four times in Triton X-100 buffer, Lustig, B. and Behrens, J. (2003). The Wnt signaling pathway and its role in tumor eluted with SDS sample buffer and subjected to western blotting (Lustig et al., development. J. Cancer Res. Clin. Oncol. 129, 199-221. 2002). Proteins were visualized using Enhanced Chemiluminescence reagent Lustig, B., Jerchow, B., Sachs, M., Weiler, S., Pietsch, T., Karsten, U., van de (Perkin Elmer) and a LuminoImager (LAS-3000, Fuji), and quantified using the Wetering, M., Clevers, H., Schlag, P. M., Birchmeier, W. et al. (2002). Negative AIDA image analyzer software v. 3.52 (Raytest, Straubenhardt, Germany). feedback loop of Wnt signaling through upregulation of conductin/axin2 in colorectal

Journal of Cell Science and liver tumors. Mol. Cell. Biol. 22, 1184-1193. Major, M. B., Camp, N. D., Berndt, J. D., Yi, X., Goldenberg, S. J., Hubbert, C., Lipid-binding assays Biechele, T. L., Gingras, A. C., Zheng, N., Maccoss, M. J. et al. (2007). Wilms tumor Expression of GST-AMER1 fragments in pGEX-4T3 was induced in Escherichia suppressor WTX negatively regulates WNT/beta-catenin signaling. Science 316, 1043- coli BL21 with 0.2 mM IPTG for 3 hours at 37°C. GST-AMER1 fusion proteins 1046. were freshly purified before the experiments using glutathione Sepharose 4B beads Martin-Belmonte, F., Gassama, A., Datta, A., Yu, W., Rescher, U., Gerke, V. and (Amersham) as described previously (Frangioni and Neel, 1993). Beads were Mostov, K. (2007). PTEN-mediated apical segregation of phosphoinositides controls washed three times in ice-cold PBS, and GST-AMER1 fusion proteins were eluted epithelial morphogenesis through Cdc42. Cell 128, 383-397. with glutathione elution buffer (20 mM Tris-HCl, pH 7.5, 8 mM glutathione and 5 McCartney, B. M., Dierick, H. A., Kirkpatrick, C., Moline, M. M., Baas, A., Peifer, mM DTT) and quantified by SDS-PAGE/Coomassie staining. Membrane Lipid M. and Bejsovec, A. (1999). Drosophila APC2 is a cytoskeletally-associated protein Strips (Echelon) were incubated with the GST fusion proteins at a concentration of that regulates wingless signaling in the embryonic epidermis. J. Cell Biol. 146, 1303- 1 ␮g/ml at 4°C overnight and detected by a mouse anti-GST antibody, according 1318. to the manufacturer’s instructions. McCartney, B. M., McEwen, D. G., Grevengoed, E., Maddox, P., Bejsovec, A. and Peifer, M. (2001). Drosophila APC2 and Armadillo participate in tethering mitotic Immunofluorescence microscopy spindles to cortical actin. Nat. Cell Biol. 3, 933-938. McCartney, B. M., Price, M. H., Webb, R. L., Hayden, M. A., Holot, L. M., Zhou, Immunofluorescence staining was performed as described previously using 0.5% M., Bejsovec, A. and Peifer, M. (2006). Testing hypotheses for the functions of APC Triton X-100 for cell permeabilization (Behrens et al., 1996). Photographs were family proteins using null and truncation alleles in Drosophila. Development 133, 2407- taken with a CCD camera (Visitron, Munich, Germany) on a Zeiss Axioplan 2 2418. ϫ microscope (Zeiss, Oberkochen, Germany, 63 objective) and MetaMorph software McLaughlin, S., Wang, J., Gambhir, A. and Murray, D. (2002). PIP(2) and proteins: (Molecular Devices). Images were processed using Adobe Photoshop CS software. interactions, organization, and information flow. Annu. Rev. Biophys. Biomol. Struct. 31, 151-175. We thank Thomas Winkler for help in antibody generation, Inke Mimori-Kiyosue, Y., Matsui, C., Sasaki, H. and Tsukita, S. (2007). Adenomatous polyposis coli (APC) protein regulates epithelial cell migration and morphogenesis via Näthke for providing anti-APC antibodies, Eva Kohler for the FKBP8 PDZ domain-based interactions with plasma membranes. Cells 12, 219-233. expression construct and Angela Döbler for secretarial assistance. Munemitsu, S., Souza, B., Muller, O., Albert, I., Rubinfeld, B. and Polakis, P. (1994). This work was supported by a grant of the Marohn Stiftung of the The APC gene product associates with microtubules in vivo and promotes their University of Erlangen-Nuremberg. assembly in vitro. Cancer Res. 54, 3676-3681. Nakanishi, S., Catt, K. J. and Balla, T. (1995). A wortmannin-sensitive phosphatidylinositol 4-kinase that regulates hormone-sensitive pools of References inositolphospholipids. Proc. Natl. Acad. Sci. USA 92, 5317-5321. Askham, J. M., Moncur, P., Markham, A. F. and Morrison, E. E. (2000). Regulation Nathke, I. S., Adams, C. L., Polakis, P., Sellin, J. H. and Nelson, W. J. (1996). The and function of the interaction between the APC tumour suppressor protein and EB1. adenomatous polyposis coli tumor suppressor protein localizes to plasma membrane Oncogene 19, 1950-1958. sites involved in active cell migration. J. Cell Biol. 134, 165-179. Barth, A. I., Siemers, K. A. and Nelson, W. J. (2002). Dissecting interactions between Polakis, P. (2000). Wnt signaling and cancer. Genes Dev. 14, 1837-1851. EB1, microtubules and APC in cortical clusters at the plasma membrane. J. Cell Sci. Reilein, A. and Nelson, W. J. (2005). APC is a component of an organizing template for 115, 1583-1590. cortical microtubule networks. Nat. Cell Biol. 7, 463-473. AMER1 regulates the distribution of APC 3747

Rivera, M. N., Kim, W. J., Wells, J., Driscoll, D. R., Brannigan, B. W., Han, M., Kim, homology domains: calcium- and agonist-induced dynamic changes and relationship J. C., Feinberg, A. P., Gerald, W. L., Vargas, S. O. et al. (2007). An X chromosome to myo-[3H]inositol-labeled phosphoinositide pools. J. Cell Biol. 143, 501-510. gene, WTX, is commonly inactivated in Wilms tumor. Science 315, 642-645. von Kries, J. P., Winbeck, G., Asbrand, C., Schwarz-Romond, T., Sochnikova, N., Rosin-Arbesfeld, R., Ihrke, G. and Bienz, M. (2001). Actin-dependent membrane Dell’Oro, A., Behrens, J. and Birchmeier, W. (2000). Hot spots in beta-catenin for association of the APC tumour suppressor in polarized mammalian epithelial cells. interactions with LEF-1, conductin and APC. Nat. Struct. Biol. 7, 800-807. EMBO J. 20, 5929-5939. Watanabe, T., Wang, S., Noritake, J., Sato, K., Fukata, M., Takefuji, M., Nakagawa, Schneikert, J. and Behrens, J. (2007). The canonical Wnt signalling pathway and its M., Izumi, N., Akiyama, T. and Kaibuchi, K. (2004). Interaction with IQGAP1 links APC partner in colon cancer development. Gut 56, 417-425. APC to Rac1, Cdc42, and actin filaments during cell polarization and migration. Dev. Sharma, M., Leung, L., Brocardo, M., Henderson, J., Flegg, C. and Henderson, B. Cell 7, 871-883. R. (2006). Membrane localization of adenomatous polyposis coli protein at cellular Wen, Y., Eng, C. H., Schmoranzer, J., Cabrera-Poch, N., Morris, E. J., Chen, M., protrusions: targeting sequences and regulation by beta-catenin. J. Biol. Chem. 281, Wallar, B. J., Alberts, A. S. and Gundersen, G. G. (2004). EB1 and APC bind to 17140-17149. mDia to stabilize microtubules downstream of Rho and promote cell migration. Nat. Smith, K. J., Levy, D. B., Maupin, P., Pollard, T. D., Vogelstein, B. and Kinzler, K. Cell Biol. 6, 820-830. W. (1994). Wild-type but not mutant APC associates with the microtubule Zumbrunn, J., Kinoshita, K., Hyman, A. A. and Nathke, I. S. (2001). Binding of the cytoskeleton. Cancer Res. 54, 3672-3675. adenomatous polyposis coli protein to microtubules increases microtubule stability and Varnai, P. and Balla, T. (1998). Visualization of phosphoinositides that bind pleckstrin is regulated by GSK3 beta phosphorylation. Curr. Biol. 11, 44-49. Journal of Cell Science