Loss of AF6/Afadin, a Marker of Poor Outcome in Breast Cancer, Induces Cell Migration, Invasiveness and Tumor Growth

Oncogene (2011) 30, 3862–3874 & 2011 Macmillan Publishers Limited All rights reserved 0950-9232/11 www.nature.com/onc ORIGINAL ARTICLE Loss of AF6/afadin, a marker of poor outcome in breast cancer, induces cell migration, invasiveness and tumor growth

G Fournier1,4,5, O Cabaud1,4,5, E Josselin1,4,5, A Chaix2,4,5, J Ade´laı¨de4,5, D Isnardon4,5, A Restouin3,4,5, R Castellano3,4,5, P Dubreuil2,4,5, M Chaffanet1,4,5, D Birnbaum1 and M Lopez1,4,5

1Inserm, UMR 891, Laboratoire d’oncologie molećulaire, CRCM, Marseille, France; 2Inserm, UMR 891, Laboratoire d’he´matopoıë`se fonctionnelle et molećulaire, CRCM, Marseille, France; 3Inserm, UMR 891, Plateforme TrGET essais pre´-cliniques, CRCM, Marseille, France; 4Institut Paoli-Calmettes, Marseille, France and 5Univ. Me´diterraneé, Marseille, France

Afadin/AF6, an F-actin-binding protein, is ubiquitously leading to improvement of classification and prognosis expressed in epithelia and has a key role during develop- (Sorlie et al., 2001). Despite this, adverse evolution can ment, through its regulatory role in cell–cell junction still be observed for patients with apparent good organization. Afadin loss of expression in 15% of breast prognosis at diagnosis and vice versa. New molecular carcinoma is associated with adverse prognosis and markers that help define prognosis are needed. We increased risk of metastatic relapse. To determine the previously showed that afadin expression is lost in 15% role of afadin in breast cancer, we studied the functional of breast cancer irrespective of the molecular subtype consequences of afadin protein extinction using in vitro (Letessier et al., 2007). We further found that afadin is and in vivo models. Three different breast cancer cell lines an independent adverse prognosis factor for patients representative of the major molecular subtypes were without axillary lymph node invasion. Absence of afadin stably repressed for afadin expression (knockdown of expression in tumors correlates with reduced 5-year afadin (afadin KD)) using RNA interference. Collective metastasis-free survival (MFS). and individual migrations as well as Matrigel invasion The MLLT4 gene encoding afadin is located on were markedly increased in afadin KD cells. Heregulin-b1 chromosome 6 band q27 (Saha et al., 1995). No paralog (HRG-b1)-induced migration and invasion were increased gene exists in mammals. Loss of afadin expression in by twofold in afadin KD cells. Conversely, ectopic breast tumors correlates with break at the FRA6E expression of afadin in the afadin-negative T47D cell line common fragile site located at 6q26 (Letessier et al., inhibited spontaneous and HRG-b1-induced migrations. 2007). Afadin is a multidomain F-actin-binding protein RAS/MAPK and SRC kinase pathways were activated in expressed in almost all epithelial tissues, and in a variety afadin KD cells. Activation levels positively correlated of other cell types including neurons, fibroblasts, with migration and invasion strength. Use of MEK1/2 endothelial and epithelial cells (Mandai et al., 1997; (U0126) and SRC kinases (SU6656) inhibitors reduced Takai et al., 2008). From the N- to C-terminus, the long afadin-dependent migration and invasion. Afadin extinc- isoform of afadin comprises two RAS/RAP1 binding, a tion in the SK-BR-3 cell line markedly accelerated tumor Forkhead, a Dilute, a PDZ and an actin-binding growth development in mouse mammary gland and lung domains (Hofmann and Bucher, 1995; Ponting, 1995; metastasis formation. These results may explain why the Mandai et al., 1997; Linnemann et al., 1999; Yamamoto loss of afadin expression in tumors correlates with high et al., 1999). The short isoform lacks the actin-binding tumor size and poor metastasis-free survival in patients. domain (Saito et al., 1998). Afadin role is complex, and Oncogene (2011) 30, 3862–3874; doi:10.1038/onc.2011.106; still not precisely known. Afadin interacts with various published online 11 April 2011 proteins (receptor tyrosine kinases, cell adhesion molecules, cytoskeletal components, regulatory and signaling Keywords: afadin; breast cancer; migration; invasion; molecules). Studies of afadin have been carried out in tumorigenicity; SRC and MAPK different species, at different developmental stages and in different cellular models in vitro. In mouse, afadin knockout results in embryonic lethality with defects starting at gastrulation. This includes disorganization of Introduction the ectoderm, impaired migration of the mesoderm and loss of somites (Ikeda et al., 1999; Zhadanov et al., Breast cancer is a heterogeneous disease. Different 1999). In Drosophila, afadin homolog Canoe (Cno) molecular subtypes of breast tumors have been defined, (Miyamoto et al., 1995) is essential during morphogenesis to maintain adherens junctions by connecting adherens junctions to actin (Sawyer et al., 2009). Correspondence: Dr M Lopez, Laboratoire d’oncologie molećulaire, However, in contrast to mammals, Cno is dispen- UMR891 Inserm, 27 Bd Leı¨Roure, 13009 Marseille, France. E-mail: [email protected] sable for cadherin–catenin-based adherens junctions Received 3 September 2010; revised and accepted 15 February 2011; assembly and establishment of cell polarity in the fly published online 11 April 2011 (Sawyer et al., 2009). Cno contributes to regulate signal Afadin loss in breast cancer and tumorigenicity G Fournier et al 3863 transduction during morphogenesis and genetically interacts with RAS, JNK, NOTCH and WNT pathways (Miyamoto et al., 1995; Takahashi et al., 1998; Matsuo et al., 1999; Boettner et al., 2003; Carmena et al., 2006). Cno regulates cell shape changes during dorsal closure, asymmetric divisions and cell fate choice (Takahashi et al., 1998; Carmena et al., 2006; Speicher et al., 2008). The precise mechanisms that support afadin and Cno function are still obscure, but the data have assigned a role for afadin in cell–cell adhesion and/or in the regulation of signal transduction. Afadin has also been described as a regulator of integrin-mediated cell adhesion (Su et al., 2003). More recently, afadin has been described as a regulator of cell migration. However, afadin contribution as positive or negative regulator of migration is still largely controversial (Lorger and Figure 1 (a) Afadin knockdown in breast tumor cell lines was Moelling, 2006; Miyata et al., 2009; Severson et al., obtained by stable RNAi strategy. SK-BR-3, MCF7 and MDA- 2009). The role of afadin in oncogenesis has been docu- MB-231 cell lines were tested by western blot using two different monoclonal antibodies. Top: the anti-AF6 (BD) directed against mented in some acute lymphoid and myeloid leukemias. the PDZ domain of afadin. Bottom: the anti-AF6 (HBT) directed The MLLT4 gene is a fusion partner of the mixed against the last C-terminus amino-acid residues present in the long lineage leukemia gene MLL (Prasad et al., 1993). Self- isoform. Both antibodies were used at a concentration of 1 mg/ml, association of afadin activates the oncogenic potential gave similar pattern and revealed the 220-kDa long isoform of afadin. (À): miRNA LacZ control vector; ( þ ): miRNA afadin of MLL–afadin fusion protein (Liedtke et al., 2010). vector. The P85 regulatory subunit of phosphoinositide-3 kinase To determine the functional consequences of afadin was used as loading charge control. (b) Afadin knockdown using loss of expression in breast cancer physiopathology, we three different miRNA afadin vectors (1, 2 and 3) gave variable studied in vitro and in vivo models. We inhibited afadin levels of inhibition compared with the miRNA LacZ control vector expression in breast tumor cells expressing afadin using (4). Afadin extinction level in cell population was controlled to be homogenous based on GFP signal, and immunofluorescence RNA interference (RNAi). We demonstrated that stable studies using anti-afadin antibody (data not shown). knockdown of afadin (afadin KD) expression led to a marked increase in collective and individual cell migrations, as well as invasion. In afadin KD cells, increased the F-actin-binding domain and a 190-kDa isoform migration and invasion were concomitant and depen- missing the F-actin-binding domain (Mandai et al., dent of the activation of the RAS/MAPK and SRC 1997). The latter is usually less frequently expressed than kinase pathways. In vivo, afadin extinction led to the former. Because functional differences have been accelerated tumor growth development in mouse mam- reported between these two isoforms (Buchert et al., mary gland and lung metastasis formation. Our results 2007) we first characterized the isoforms of our cellular assigned afadin as a new breast cancer biomarker models. We used two different antibodies. One was involved in tumor progression. directed against the C-terminus F-actin-binding domain and thus specific for the 220-kDa isoform. The other, directed against the PDZ domain, recognized both Results isoforms. Both antibodies revealed a major band at 220 kDa, suggesting that the 220-kDa isoform is Afadin knockdown using miRNA sequences prominently expressed in the three cell lines To analyze the functional consequences of afadin loss of (Figure 1a). The knockdown expression of afadin in expression, we used an RNAi strategy. This was done the three cell lines was equally detected by both using an expression vector that enables the expression of antibodies. This ascertained the specificity of the engineered microRNA (miRNA) sequences and Emer- engineered miRNA vectors and the two antibodies ald GFP (EmGFP) as reporter (Letessier et al., 2007). (Figure 1a). Afadin knockdown efficiency was quanti- Three different afadin miRNA sequences were designed fied for the three different miRNAs in the three cell and cloned in this vector. These miRNA sequences were lines. Residual expression of afadin after knockdown located at positions 2515 (AF6-1), 3481 (AF6-2) and ranged from 12 to 38% for MCF7, from 0.5 to 37% for 4088 (AF6-3) in the afadin cDNA sequences. Valida- SK-BR-3 and from 33 to 67% for MDA-MB-231 tions of the respective vectors were done in Cos cells (Figure 1b). Immunofluorescence analysis showed that after transient expression (data not shown). Because expression was homogeneously inhibited within every afadin loss of expression in breast tumors is not cell population (data not shown). correlated with any molecular subtype, we selected three cell lines representative of different breast cancer Afadin negatively controls ‘collective’ cell migration subtypes: MCF7 (luminal), SK-BR-3 (ERBB2) and Afadin loss of expression in breast cancer correlates with MDA-MB-231 (basal) cell lines. high pathological size and reduced 5-year MFS. This Two isoforms of afadin have been described in suggests that afadin loss may lead to the deregulation rodents and humans: a 220-kDa isoform containing of different biological processes such as proliferation,

Oncogene Afadin loss in breast cancer and tumorigenicity G Fournier et al 3864

Figure 2 Increase of collective migration velocity in afadin KD cells. (a) Collective cell migration was measured using wound-healing assay in serum-starved conditions. Speed of recovery increases in the three cell types and inversely correlates with afadin extinction level. (b) Collective cell migration was done in the absence (À) or in the presence ( þ )of1nM HRG-b1. Each histogram bar corresponds to the mean of speed recovery determined by 15 measurements for each condition. This is representative of at least three experiments done in parallel with western blot. The Mann–Whitney test was used in these experiments (**Po0.005, ***Po0.0005).

adhesion and migration. To test these hypotheses, we Heregulin-b1 (HRG-b1) is a potent inducer of cell developed different assays using the three cell lines migration through ERBB2-dependent activity (Spencer expressing variable levels of afadin. We thus expected to et al., 2000). To analyze the strength of cell migration in observe a gradual relationship between phenotype and afadin KD cells, we compared the collective migration afadin expression level. of SK-BR-3 afadin KD cells with that of SK-BR-3 To measure the migration properties of cells, we ﬁrst control cells in the presence of HRG-b1. Using the used the haptotaxis wound-closure assay that provides wound-closure assay, we found that the velocity of insights into ‘collective’ cell motility. Speed of wound wound closure of SK-BR-3 afadin KD cells was similar recovery was measured by time-lapse videomicroscopy to that of non-KD cells treated with 1 nM HRG-b1 at 16 h post-wound for MCF7 and SK-BR-3 and 7 h (Figure 2b). This concentration gives maximum effect in post-wound for MDA-MB-231. These time periods limit this test. Interestingly, activation of afadin KD cells with artefacts of wound closure due to an expanding 1nM HRG-b1 induced an additive effect by increasing population rather than motility. We found similar velocity of wound closure by twofold compared with results for the three cell lines tested, that is, an increase HRG-b1 on control cells. of the velocity of wound closure ranging from 20 to In summary, afadin KD led to (i) increased collective 180% compared with control cells (Figure 2). Notice- cell migration in three different cellular models, (ii) ably, speed of wound recovery was inversely correlated migration in the low-migrating SK-BR-3 cells and (iii) a with afadin extinction level, especially in MCF7 and SK- twofold increase of HRG-b1-mediated collective cell BR-3 cells. Indeed, MCF7 AF6-2 and AF6-3, which had migration. These results suggest that afadin exerts a the lowest levels of afadin, showed the highest velocity negative regulation on spontaneous and inducible of wound closure (Figure 2a). The same was true collective cell migration. for SK-BR-3 AF6-1 and AF6-3. The MDA-MB-231 cell line is intrinsically highly motile. In this cell line, Afadin negatively controls ‘individual’ cell migration afadin KD led to a signiﬁcant increase of the velocity The wound-closure assay is related to ‘collective’ of wound closure. The whole of these results and motility. Compared with control SK-BR-3 cells we time-lapse videomicroscopy videos are presented in observed that the morphology of SK-BR-3 afadin KD Supplementary Figure S1. Together, these results estab- cells was markedly different during wound closure. lished a relationship between afadin expression levels SK-BR-3 afadin KD cells adopted morphology of and velocity of wound closure. migrating cells, with the formation of lamellipodia

Oncogene Afadin loss in breast cancer and tumorigenicity G Fournier et al 3865

Figure 3 Afadin KD SK-BR-3 cells exhibit shapes of migrating cells. (a) Comparison between control and afadin KD SK-BR-3 cells at the leading edge during wound healing. Top: inverted microscope view ( Â 1000). Bottom: immunofluorescence staining for a-tubulin ( Â 400). (b) Comparison between control and afadin KD SK-BR-3 cells in culture. Top: inverted microscope view ( Â 630). Bottom: immunofluorescence staining for a-tubulin ( Â 400). (c) Quantification of cells with extended lamellipodia in the different conditions (visualization of F-actin with TRITC-labeled phalloidin). structures identified by microtubule cytoskeleton stain- tests, correlation between migration and afadin expres- ing by a-tubulin antibody (Figure 3a). This difference sion levels of SK-BR-3 RNAi AF6-1 and AF6-3 was less was also seen in standard cultures. Afadin KD cells had marked than for wound-closure experiments. This may an elongated shape compared with control cells be attributed to a variation of afadin extinction at the (Figure 3b). The number of cells with an elongated time of the experiments. Together, these results show shape was increased from B20% in control cells to 70% that afadin negatively controls individual cell migration. in afadin KD cells (Figure 3c). Results were similar with the three different afadin KD cells. Addition of HRG-b1 Afadin expression in afadin-negative breast tumor cells to culture induced the formation of lamellipodia in To further demonstrate that afadin is a negative 100% of afadin KD cells compared with 70% of control regulator of migration, we looked for a cellular model cells (Figure 3c). with an afadin gene alteration such as the one we found We further analyzed SK-BR-3 cell migration on in breast tumors. We screened for afadin-negative breast collagen type I precoated (8 mm) transwell filters and tumor cell lines associated with an MLLT4 gene compared afadin KD with control cells. Before incuba- deletion. Among 16 cell lines, we found 3 cell lines with tion in upper chambers, cells were controlled to be no or low level of afadin by western blot, namely individualized. We found that migration of SK-BR-3 SUM159, T47D and ZR75.1 (Figure 5a). Low expres- control cells was low. Migration was dramatically sion was found at transcriptional level for these cells increased in afadin KD cells, ranging from 2.5- to (data not shown). Genome analysis using array com- 6-fold respective to afadin extinction (Figure 4a). These parative genomic hybridization showed that one allele of results suggest that afadin extinction leads to increase the MLLT4 gene was altered in T47D. This was the individual cell migration. Analysis of velocity on cells alteration we described in tumors and therefore we seeded at low density by cell tracking showed that afadin chose T47D as a model to carry out afadin-related KD cells had an increase velocity compared with control functional studies (Figure 5b). cells (Figure 4b). No difference in directionality was Full-length L-afadin cDNA was transduced in this observed (data not shown). It is of note that, for both cell line (T47D/afadin) (Figure 6a) and spontaneous

Oncogene Afadin loss in breast cancer and tumorigenicity G Fournier et al 3866 increased by 5.6-fold cell invasiveness as seen in Figure 7a (top) and 7b. As expected, addition of 1 nM HRG-b1 strongly increased invasiveness of control cells. More importantly, this effect was strengthened by 2.7-fold in afadin KD cells (Figure 7a bottom and 7b). These results demonstrated that loss of afadin expression in SK-BR-3 cells leads to increase invasiveness of the cells. As described for migration, afadin KD strengthened HRG-b1-mediated invasion.

Afadin negatively regulates SRC kinase and RAS/MAPK activation Activation of ERBB2 signaling pathways through HRG-b1 activation leads to the activation of the PI3K, p38/MAPK, RAS/MAPK and SRC kinase pathways (Marone et al., 2004). The RAS/MAPK and SRC kinase pathways are negatively regulated by afadin (Radziwill et al., 2003, 2007; Carmena et al., 2006). These pathways have been implicated in cytoskeleton remodeling, cell motility and invasion (Keely et al., 1997; Klemke et al., 1997; Vadlamudi et al., 1999). We showed that afadin KD results in increased velocity and invasiveness of the cells in the absence or presence of 1nM HRG-b1 (Figures 2b and 7b). These conclusions are also strengthened by the experiments performed on the T47D cell line (Figure 6b). The enhanced effect observed with 1 nM HRG-b1 is surprising, as this dose is expected to give a full biological response in this assay (Marone et al., 2004). This may indicate that afadin could control the extent of these biological processes by exerting a permanent negative regulation. We analyzed Figure 4 (a) Increased migration of afadin KD cells in Boyden chamber. Individualized SK-BR-3 cells were incubated as described RAS/MAPK and SRC kinase signaling pathways in in Materials and methods. Migration of SK-BR-3 afadin KD cells SK-BR-3 cells and compared SRC and ERK1/2 expressing different levels of afadin (AF6-1, AF6-2 and AF6-3) was activation in control and afadin KD cells, in the absence measured and compared with control cells. Each histogram bar or presence of 1 nM HRG-b1. We found a significant corresponds to the mean of luminescence determined by six and reproducible increase in SRC and ERK1/2 activa- measurements for each condition. This is representative of at least three experiments done in parallel. (b) Cell velocity of SK-BR-3 tion in non-stimulated afadin KD cells (Figure 8a, top- cells. Cells were plated at low density. Velocity was measured using left panel). As already described by others, stimulation the MetaMorph software. The Mann–Whitney test was used in of cells with HRG-b1 leads to an increase of SRC and these experiments (**Po0.005, ***Po0.0005). ERK1/2 activation in control cells. Interestingly, activation levels increased in afadin KD cells (Figure 8a, top-right panel). The strength of SRC and ERK1/2 activation and HRG-b1-induced migration was measured using the correlated with cell invasiveness, highlighting their probable wound-closure assay (Figure 6b). A 35% reduction of implication in this process (Figure 8a, bottom). spontaneous collective migration was observed in T47D/ To test the implication of these pathways in the afadin compared with T47D control cells (Figure 6b, afadin-regulated migration processes, we used MEK1/2 histogram bars 1 and 2). T47D/afadin showed a 26% (U0126) and SRC kinases (SU6656) inhibitors on afadin reduction of HRG-b1-induced migration (Figure 6b, KD vs control cells. Used at 5 mM on SK-BR-3 cells, histogram bars 3 and 4). Remarkably, comparison of these inhibitors markedly blocked phosphorylation of T47D and SK-BR-3 cell lines showed that T47D their targets with a very good specificity (data not collectively migrated at a velocity similar to SK-BR-3/ shown). As shown in Figure 8b, invasiveness of afadin afadin KD and conversely that T47D/afadin migrated KD cells was inhibited by 46 and 52% with U0126 and similarly to SK-BR-3 (Figure 6c). These results further SU6656 treatments, respectively. U0126 treatment led to demonstrated the role of afadin as an important a 36% inhibition of the migration. SU6656 had no effect regulator of cell migration. on migration (Figure 8c). These results suggested that afadin knockdown leads Afadin negatively regulates invasion to the activation of ERK1/2 and SRC kinase and to To further characterize the role of afadin, we compared their ‘hyperactivation’ upon stimulation with HRG-b1. the Matrigel invasiveness of SK-BR-3 afadin KD cells Activation of these pathways is implicated, at least in with that of SK-BR-3 control cells in the presence of a part, in the increase of invasiveness and migration. Our chemotactic gradient of serum. Afadin KD significantly results assigned afadin as a probable negative regulator

Oncogene Afadin loss in breast cancer and tumorigenicity G Fournier et al 3867

Figure 5 Afadin expression and a-CGH analysis in breast-derived cell lines. (a) Expression was tested by western blot using two different monoclonal antibodies (see Figure 1). (b) Chromosome 6 profiles of SUM159, T47D and ZR75-1 breast cell lines were established with CGH Analytics software and analyzed as previously described (Adelaide et al., 2007) (left part). Color profiles (golden brown, gray and red) correspond to the different cell lines (SUM159, T47D and ZR75-1, respectively). They show 6q losses in T47D and ZR75-1 comparatively to SUM159; the latter does not present any copy number aberration in that region. The genomic profiles focused on the 6q 164.5–171.3 Mb genomic interval centered on the MLLT4 locus (right part) show a breakage within the telomeric region to the 50part of MLTT4 locus (ZR75-1) and a downstream heterozygous loss of the MLLT4 gene (T47D). * None specific band. of cell migration and invasiveness at least in tumor cells marked luminescent signal in lungs of mice xenografted and a putative negative modulator of these pathways. with SK-BR-3 afadin KD not with control mice (Figure 9c). Analysis on isolated organs confirmed the Afadin loss of expression increases tumor development presence of metastases in lung and not at other sites in xenografted mice (Figure 9d). Similar results were obtained with MCF7 In breast cancer patients, afadin loss of expression cells (data not shown). correlates with tumor size and reduced 5-year MFS. To These results indicated that downregulation of afadin further document the oncogenic role of afadin loss of expression contributes to increase tumorigenicity and expression, we selected the SK-BR-3 cell line that is metastasis formation. poorly tumorigenic in mice. We used two assays. First, we measured colony formation in agar, which is a common method to monitor anchorage-independent Discussion growth. Colony formation was slightly but significantly increased in SK-BR-3 afadin KD cells. This was mainly Afadin is an independent prognosis marker for breast associated with a colony size increase (Figure 9a). cancer. Loss of afadin is associated with poor outcome Second, SK-BR-3 afadin KD cells were xenografted in the population of patients with favorable prognosis orthotopically and compared with control cells. As (lymph node-negative patients). Afadin-negative tumors shown in Figure 9b, afadin knockdown markedly have a larger size and patients have a reduced 5-year accelerated tumor onset in the mouse mammary gland. MFS. The detection of metastases was assessed by biolumines- In this study, we have developed and analyzed cence. Lung metastases were, respectively, detected different models to determine the functional conse- between the 12th and the 16th weeks and between the quences of loss of afadin in breast cancer. 18th and the 21th weeks in 100% of mice xenografted Knockdown of afadin expression in breast tumor with SK-BR-3 afadin KD cells and SK-BR-3 control cell lines led to an increase of cell migration, cell inva- cells. Analyses of viable mice at the 13th week showed a siveness, colony formation and tumor growth in mice.

Oncogene Afadin loss in breast cancer and tumorigenicity G Fournier et al 3868

Figure 7 Afadin KD increased invasiveness of SK-BR-3 cells. Figure 6 Afadin expression reduces motility in T47D cells. (a) Analysis of cell invasiveness in two different Matrigel coated (a) Afadin expression in T47D (T47D/afadin) transduced with inserts (#1 and #2). Top: in the absence of HRG-b1; Bottom: in the full-length afadin cDNA by western blot. ‘*’ This band may correspond presence of 1 nM HRG-b1. (b) Quantification of invasive cells to protein degradation. (b) Spontaneous and HRG-b1-induced per field (magnification Â 200). Values correspond to fold increase migration is reduced in T4D/afadin cells compared with T47D compared with SK-BR-3 RNAi control cells. The human fibro- transduced with empty vector. (c) Comparative migration of sarcoma HT-1080 control cell line is highly invasive and is used as T47D and SK-BR-3 cells expressing or not afadin. Wound-closure a positive control, as recommended by the manufacturer. This is experiments were performed as described in Figure 2. The Mann– representative of three experiments done in duplicate. The Mann– Whitney test was used in these experiments (***Po0.0005). Whitney test was used in these experiments (**Po0.005, ***Po0.0005). These results are in accordance with the histoclinical status of patients with afadin-negative tumors. They afadin KD cells. We found no qualitative and quantitative also emphasize a key role of afadin in controlling difference of E-cadherin and b-catenin localization in the migration and invasion processes at least in tumor cells. presence or absence of afadin expression (Supplementary Collective and individual cell migration increase was Figure S2). This indicates that, at least in MCF7, the noted in the three cell lines using three different RNAi cadherin–catenin complex is not affected by afadin KD sequences, and was inversely correlated with afadin and that the increase of collective cell migration is mainly levels. Conversely, complementation of the afadin- the consequence of increased individual cell migration. deficient T47D breast tumor cell line reduced cell Afadin interacts with a-catenin and indirectly with other migration. It has been shown that the F-actin-binding F-actin-binding proteins that link afadin to E-cadherin/ region of afadin is important to negatively regulate the catenin complex (Takai et al., 2008). We cannot exclude collective cell migration of the MCF10A breast-derived that loss of afadin expression may influence cell migration cell line, through the maintenance of E-cadherin- through deregulation of E-cadherin/catenin signalization. dependent intercellular adhesion (Lorger and Moelling, Together, our results are in line with those of Lorger and 2006). In our study, we used breast tumor cell lines that Moelling (2006). express (MCF7) or not (MDA-MB-231 and SK-BR-3) We found a significant activation of the SRC kinase E-cadherin. We analyzed the junctional localization and ERK1/2 in afadin KD cells and an ‘hyperactivation’ of E-cadherin and b-catenin in MCF7 cells during upon stimulation with HRG-b1. These results are fully wound healing and compared between control and in accordance with the repressive role of afadin or its

Oncogene Afadin loss in breast cancer and tumorigenicity G Fournier et al 3869

Figure 8 (a) Top: activation of SRC kinase and ERK1/2 in control SK-BR-3 cells, in afadin KD SK-BR-3 cells and in response to HRG-b1. Cell extracts were analyzed by western blot and membranes were probed with afadin (BD Transduction Laboratories), Phospho-SRC and Phospho-ERK1/2 antibodies. SRC and ERK1/2 antibodies were probed to control loading charge. ‘*’ Two different time exposures with the anti-Phospho-ERK1/2 antibody. Bottom: levels of SRC and ERK1/2 activation and strength of invasion. (b) The MEK inhibitor U0126 and the SRC inhibitor SU6656 (5 mM) partially inhibit invasion process. (c) The MEK inhibitor U0126, not the SRC inhibitor SU6656 (5 mM) partially inhibits collective cell migration. The Mann–Whitney test was used in this experiments (**Po0.005, ***Po0.0005). ortholog Cno on the RAS/MAPK pathway and the migration and invasiveness by regulating RAS/MAPK negative role on SRC kinase activation in mammalian and SRC signaling pathways. Afadin is thought to have cells (Radziwill et al., 2003, 2007; Carmena et al., 2006). a key role in the stabilization and maintenance of cell– We observed a correlation between the strength of SRC cell junction integrity in polarized epithelial cells by and ERK phosphorylation and strength of biological exerting a dual role as a cytoskeletal molecular adaptor responses. Inhibition of RAS/MAPK and SRC path- and as a negative regulator of RAS/MAPK and SRC ways reduced afadin KD-mediated effects. Together, kinase (Radziwill et al., 2003, 2007). This dual role could these results assigned afadin as a key negative regulator also be involved in the control of cell migration. It has of tumor cell migration and invasiveness through been hypothesized that RAS-GTP binding to the RA regulation of these two pathways. domain of afadin negatively regulates the RAS/MAPK Because afadin negatively regulates the strength of pathway. This binding is regulated by the BCR kinase migration and invasion, we studied the status of afadin (Radziwill et al., 2003). SRC activity is restricted by during HRG-b1 stimulation, that is, expression level binding with the PDZ domain of afadin (Radziwill and localization. We found no modulation of afadin et al., 2007). In Drosophila, it has been shown that EGF/ expression level upon stimulation with 1 nM HRG-b1, RAS signaling may regulate non-muscle myosin II showing that afadin is not downregulated during through Canoe (Gaengel and Mlodzik, 2003). Thus, activation (Figure 8A, top). During activation, we interactions with transmembrane, cytoskeletal, regulat- found a marked localization of afadin at the cell leading ing and signaling proteins have a pivotal role in the edge (Supplementary Figure S3). Thus, afadin localized function of this scaffold molecule. It remains to under- at leading edge may control the strength of cell stand how different partners can bind to one speciﬁc

Oncogene Afadin loss in breast cancer and tumorigenicity G Fournier et al 3870

Figure 9 Afadin downregulation increases tumorigenicity of SK-BR-3 cells in NOD/SCID mice. (a) Soft agar assay: (left) SK-BR-3 AF6-3 cells form signiﬁcantly larger colonies than control SK-BR-3. (Right) Each histogram bar corresponds to the mean of luminescence determined by three measurements for each condition. This is representative of at least three experiments. The t-test was used in this experiment (*Po0.05). (b) Tumor development in the mammary fat pads was accelerated in SK-BR-3 AF6-3 (0.3 Â 106 cells) compared with SK-BR-3 LacZ control cells. (c) Ventral view of mice 13 weeks after mammary fat pad engraftment. SK-BR-3 cells used express luciferase. Luminescence counting was done in the same conditions. Signal in SK-BR-3 RNAi AF6-3 is localized to the lungs. No signal was detected with SK-BR-3 RNAi control. These images are representative of the 10 mice used in the study. (d) Detection of SK-BR-3 AF6-3 metastasis at day 14 was restricted to lung. No luminescence is detected in other organs.

domain. It is possible that these interactions are ﬁnely mediates self-association of the MLL–AF6 fusion regulated in a spatio-temporal manner during dynamic protein via a region different from RAS interaction processes such as cell–cell junction formation and (Liedtke et al., 2010). Thus, multimerization of afadin migration processes. Recently, it has been demonstrated molecules could be an attractive means by which afadin that the RA1 domain of afadin or its ortholog Cno may exert its role of scaffold molecule.

Oncogene Afadin loss in breast cancer and tumorigenicity G Fournier et al 3871 We extended our observations to the colon carcino- TCAGTCAGTGGCCAAAACGCAAATCAGCCTCCTAGT ma-derived cell line SK-CO-15 and found that knock- CCTC, respectively. down of afadin, with three different RNAi, increased the AF6_4088S: TGCTGAATCCATGGACATTCCATCGAGT velocity of these cells (data not shown). This result tends TTTGGCCACTGACTGACTCGATGGAGTCCATGGATT to generalize the impact of loss of afadin expression in and AF6_4088AS: CCTGAATCCATGGACTCCATCGAGT tumors. However, for an unclear reason, this result CAGTCAGTGGCCAAAACTCGATGGAATGTCCATGGA TTC, respectively. These sequences are located, respectively, contrasts with a recently published work (Severson at residues 2515, 3481 or 4088 in afadin cDNA and are present et al., 2009). This could be attributed to the origin of the in the different afadin isoforms. cell line used. Like afadin, the expression of several actin-binding proteins (a-actinin, gelsolin, tensin-3 and Vector for afadin expression profilin-1) weakens the phenotype of transformed cells Full-length cDNA of human L-afadin was amplified by PCR (Tanaka et al., 1995; Nikolopoulos et al., 2000; Zou from the pCDNA3-FLAG-L-AF6 vector and cloned in the et al., 2007; Martuszewska et al., 2009). However, effects pRRL-iGFP-GW lentiviral vector using following primers: can be opposite in non-epithelial cells. For example, GWAF6.5, TCGCTCGGATCACCACCATGGACTAC and whereas profilin-1 depletion increases motility and GWAF6.3FL, CTCACTTTGTGTTCAGTTCATTC. The re- invasiveness of breast cancer cells, it inhibits motility sulting pRRL-AF6FL vector was used to produce virion stock of normal vascular endothelial cells (Bae et al. 2009, Zou (see below). et al., 2007, Ding et al., 2006). Similarly, whereas we found that afadin depletion increases motility of breast Cells and culture conditions and colon cancer cells, a recent study showed that it MCF7, MDA-MB-231, SK-BR-3 and T47D breast carcinoma inhibits directional movement induced by platelet- cells were grown in RPMI 1640 medium supplemented with derived growth factor in mouse NIH3T3 fibroblasts 10% fetal calf serum (FCS) (heat inactivated 30 min at 56 1C), (Nakata et al., 2007). Profilin-1 interacts physically with 1mM pyruvate, 2 mML-glutamine, 0.1 mM non-essential amino afadin and it is possible that the two molecules interact acid, 10 U/ml penicillin, 10 mg/ml streptomycin (Invitrogen); for MCF7 cells, 10 mg/ml of insulin was added (Invitrogen). also functionally to regulate actin modeling at the Cell lines were maintained in a humidified 5% CO2 atmo- leading edge (Boettner et al., 2000). Our results are close sphere at 37 1C. to those described for profilin-1 not only in vitro but also in vivo. Indeed, we have shown for the first time that loss Cell transfection and cell sorting of afadin expression increases Matrigel invasiveness and MCF7, MDA-MB-231 and SK-BR-3 cells were transfected by accelerates orthotopic tumorigenicity and metastasis nucleofection with 2 mg of the pCDmiRAF6-2515, 3481, 4088 formation. or lacZ control plasmid as recommended by the manufacturer Afadin loss of expression contributes to foster the (Lonza, Walkersville, MD, USA) and selected in the presence major steps of tumor progression, that is, loss of cell of 5 mg/ml blasticidin (Invitrogen). polarity, proliferation and migration through deregula- High GFP expressing cells were selected using a FACSAria tion of master molecular events such as SRC and II cell sorter (Becton Dickinson, CA, USA). MAPK activation, F-actin functions and probably other non-elucidated mechanisms. These data assign afadin as Production of lentiviral virions a tumor progression suppressor gene and explain why a The virus particles were produced by transient transfection loss of afadin expression in breast tumors impacts on into 293T cells seeded the day before in 10 cm plates with 12 ml both migration and tumor growth, translating into high culture media, with 8.1 mg of the pRRL-AF6FL vector, 8.1 mg tumor size and poor MFS in patients. of the packaging construct and 2.7 mg of the construct that express the vesicular stomatitis virus glycoprotein by using the calcium-phosphate method. The following day, cells were fed with 6 ml of Dulbecco’s modified eagle medium and further Materials and methods cultured for 24 h. The supernatant was then collected, cleared by centrifugation and passed through a 0.45-mm filter. The Vectors for RNAi supernatant was then concentrated by ultracentrifugation over RNA silencing was done using the BLOCK-iT Pol II RNAi a 20% sucrose cushion. Viral pellet was resuspended in expression vector kit as recommended by the manufacturer Dulbecco’s phosphate-buffered saline supplemented with 1% (Invitrogen, Carlsbad, CA, USA). Artificial afadin miRNA glycerol in 1/100 of the initial volume of the viral supernatant, 1 was cloned in the pcDNA 6.2-GW/EmGFP-miR leading to aliquoted and stored at –80 C. cocistronic expression of EmGFP with afadin miRNA. Different sequences of afadin miRNA were designed using Antibodies an algorithm developed by Invitrogen. Sense and antisense Afadin (clone 35), E-cadherin (clone 36) and b-catenin (clone DNA sequences were 14) antibodies were from Becton Dickinson. Afadin antibodies AF6_2515S: TGCTGATCAGGTGCATACTTATCCATG were also purchased from Abcam (ab11337, Cambridge, UK) TTTTGGCCACTGACTGACATGGATAAATGCACCTGAT and HyCult Biotechnology (clone 3, Uden, The Netherlands). and AF6_2515AS: CCTGATCAGGTGCATTTATCCATGT Mouse monoclonal antibody anti human a-tubulin was used CAGTCAGTGGCCAAAACATGGATAAGTATGCACCTG as recommended by the manufacturer (clone B-5-1-2, Sigma, ATC, respectively. St Louis, MO, USA). Antibodies against P85 (Millipore, AF6_3481S: TGCTGAGGACTAGGAGGCTGATTTGCG Boston, MA, USA), ERK1/2 (154, Santa Cruz Biotechnology, TTTTGGCCACTGACTGACGCAAATCACTCCTAGTCCT Santa-Cruz, CA, USA), phosphor-ERK1/2 (V803A, Promega and AF6_3481AS: CCTGAGGACTAGGAGTGATTTGCG Corporation, Madison, WI, USA), SRC (2110, Cell Signaling

Oncogene Afadin loss in breast cancer and tumorigenicity G Fournier et al 3872 Technology, Danvers, MA, USA), phospho-SRC (2101, Cell microscope (Zeiss Axiovert 200, Zeiss Carl Zeiss S.A.S. Signaling Technology) were used as recommended by the Microscopy Division, Le Pecq, France). Images of wounded manufacturers. HRP- and fluorophore-conjugated secondary monolayers were acquired for 7 or 16 h (5 or 15 min intervals) antibodies were from ThermoFisher Scientific (Rockford, IL, using a digital camera (charge-coupled device camera Cool- USA) and Jackson Immuno-Research (West Grove, PA, snap HQ; Photometrics Headquarters (Tucson, AZ, USA)) USA), respectively. driven by Metamorph software (Molecular Devices, Down- ingtown, PA, USA). Three independent experiments were done for each experimental condition. Measurements of cell Western blot analysis velocity were calculated using the MetaMorph software. Cells were lyzed in ice-cold lysis buffer containing 50 mM Hepes, pH 7.5, 150 mM NaCl, 1.5 mM MgCl2,1mM EGTA, 1% Triton X-100, 0.2 mM sodium ortho-vanadate and 10% Cell migration assay in Boyden chamber glycerol. ‘Complete’ Protease Inhibitor Cocktail was added The migration of SK-BR-3 cells was assayed using transwell to cold lysis buffer as recommended by the manufacturer 96-well permeable supports (8 mm—Corning HTS) precoated (Roche Diagnostics, Meylan, France). with rat tail collagen I (25 mg/ml—Roche) as described Cell lysates (30 or 60 mg) were separated using NuPAGE previously (ALMA). In brief, 0.25 ml of 10% FCS–RPMI 4–12% Novex Bis-Tris gels according to the manufacturer’s medium was added in the well of the culture plate (the bottom recommendations (Invitrogen) and transferred onto Hybond- compartment). Serum-starved SK-BR-3 cells (5 Â 104) were C membrane (GE Healthcare Bio-Sciences, Piscataway, NJ, suspended in 50 ml of 10% FCS–RPMI medium and seeded to USA). Membranes were blocked in phosphate buffered saline the upper chamber. After incubation at 37 1C for 24 h in a cell (PBS) supplemented with 5% milk or 5% bovine serum culture incubator, the cells that migrated from the top to the albumin for 1 h and then incubated with indicated antibodies. bottom compartment were detached by transferring the After washing, membranes were incubated with the secondary transwell insert in an opaque 96-feeder well plate containing antibody conjugated with horseradish peroxidase (Pierce). The prewarmed trypsin-EDTA (37 1C). After 30 min of incubation signal was detected using the Amersham ECL system at 37 1C, 10% FCS was added to inactivate trypsin and (Amersham Pharmacia Biotech, Uppsala, Sweden). Celltiter-Glo reagent was added to determine the cell number by luminescence assay. Luminescence was recorded after 10 min of incubation using a microplate luminometer (Fluo a-CGH hybridization and analysis Star Optima, BMG Labtech GmbH, Offenburg, Germany). DNA digestion, labeling and hybridization on Human Three independent experiments were done for each experi- Genome CGH 244A Oligo Microarray (Agilent Technologies, mental condition. Santa Clara, CA, USA) was performed as previously described (Adelaide et al., 2007). Briefly, 1 mg of DNA for each sample and reference (a mix of DNA from peripheral blood leukocytes Cell tracking assay of 13 healthy men) was, first digested by AluI and RsaI, then Cells plated at low density (5 Â 104) were seeded in 6-well plates labeled with BioPrime Array CGH Genomic Labeling System precoated with rat tail collagen I (25 mg/ml—Roche). Indivi- (Invitrogen) and hybridized with Oligo aCGH Hybridization dual cell tracking was acquired for 24 h at 5 min intervals using Kit (Agilent Technologies). After washing, the slides were a digital camera driven by MetaMorph software. Three scanned on an Agilent Microarray Scanner G2505B, captured independent experiments were done for each experimental images were analyzed with Feature Extraction Software v condition. Measurements of cell velocity were calculated using 10.5.1.1 and created text files were processed with CGH the MetaMorph software. Analytics v 3.4 (Agilent Technologies, Santa Clara, CA, USA) to generate genomic alteration profiles. Invasion assay The invasion assay was performed using the 24 cell culture Immunofluorescence inserts with 8 mm pore size membrane coated with Matrigel Cells were plated on coverslips in 10% FCS–RPMI for 24 h (BD BioCoat Growth Factor reduced Matrigel Invasion and incubated with or without 1 nM heregulin (R&D Systems, Chamber (Becton Dickinson, MA, USA)). Chemoattractant Minneapolis, MN, USA) for 30 min. Then, cells were fixed in (10% FCS or 1 nM HRG-b1) was added to the wells and methanol at À20 1C for 6 min and rinsed in PBS. After 105 cells were loaded in the insert in RPMI 0.1% BSA. Plates blocking in PBS, 10% FCS, cells were incubated overnight were incubated for 72 h at 37 1C. After migration, non- with primary antibodies diluted in PBS–10% FCS. After migratory cells on the upper membrane surface were removed washing in PBS-0.1% Tween-20, primary antibodies were with cotton tipped swabs (four times per insert) and the detected using donkey anti-rabbit or anti-mouse Ig secondary migratory cells attached to the bottom surface of the antibodies conjugated to Cy3 or Cy5. DNA was stained with membrane were stained for 20 min with 0.2% crystal violet 250 ng/ml 4’,6-diamidino-2-phenylindole for 2 min. Cells were in ethanol (Sigma). mounted in Prolong Gold anti-fade reagent (Invitrogen) and After five washes with 200 ml distilled water, invasive cells examined on an LSM-510 Carl Zeiss confocal microscope. were counted with an inverted microscope. Each determina- tion represents the average of three individual experiments done in duplicate. Wound healing assay MCF7, MDA-MB-231, SK-BR-3 and T47D cells (1 Â 105) were seeded into culture insert (Ibidi, Martinsried, Germany) Soft agar assay in 6-well plates and grown to 100% confluency for 24 h and The 96-well plate was covered first with bottom agar (DMEM, then serum-starved for 24 h. Culture inserts were carefully 10% FCS with 0.6% agar). Cells were suspended at a density removed to create a cell-free gap of 400 mm. After three washes of 3 Â 104 cells/ml in top agar (DMEM, 10% FCS with 0.3% with PBS, the wounded monolayers were then incubated in agar) and seeded to wells of the 96-well plate. After incubation 10% FCS–RPMI medium in the presence of 1 nM heregulin at 37 1C for 10 days, colonies were quantified using CellTiter- (R&D Systems). Cells were observed using a fluorescence Glo (Promega). This is a luminescent method that determines

Oncogene Afadin loss in breast cancer and tumorigenicity G Fournier et al 3873 the number of viable cells in culture based on quantiﬁcation of (Photon Imager, Biospace Lab, France). Tumor volume the ATP present. Luminescence was recorded after 10 min of was determined weekly by measuring the tumors in two incubation using a microplate luminometer (FluoStar Optima, dimensions using a calliper. Volume (V) was determined by BMG Labtech). Three independent experiments were done for the following equation, where L is length and W is width: each experimental condition each times. V ¼ (L Â W2) Â 0.5.

Animal models and xenograft assay Before xenograft experiments, luciferase gene was transduced Statistical analyses ± in SK-BR-3 (AF6-4088 or lacZ) cells. Approximately 70% Data are presented as mean s.e.m. and were analyzed by confluent cells (106 cells) were incubated with a 1:3 precipitated Mann–Whitney tests using Prism software as indicated. Po0.05 mixture of lentiviral supernatants Lenti-LUC-vesicular stoma- was considered statistically significant. Tumor diameter and titis virus G (Vector Core, University of Michigan, MI, USA) corresponding standard errors were determined for each experi- and culture medium, for overnight incubation. Fresh medium ment. Statistical analyses were based on paired t-tests. was replenished at this time. The following day, the cells were harvested by trypsin/EDTA and subcultured at a ratio of 1:3. After 1-week incubation, the luciferase expression was detected by adding 2 ml D-luciferin (Promega). Conflict of interest Tumorigenicity of SK-BR-3 (AF6-4088 or lacZ) cells was assessed in independent sets of five immunodeficient non-obese The authors declare no conflict of interest. diabetic/severe combined immunodeficiency/gamma cnull (NOD/SCID/gamma cnull or NOG) mice (Jackson Labora- tory, Bar Harbor, ME, USA). Six-week-old NOG mice were anesthetized with ketamine (100 mg/kg) and xylazine (10 mg/kg) Acknowledgements during the implantation procedure. For tumor formation, 0.3 Â 106 SK-BR-3 cells in FBS were mixed with Matrigel We thank F Birg for helpful discussions. We are grateful of (Becton Dickinson; 1:1) and injected orthotopically into the M Buchert for providing us with the pCDNA3-FLAG-L-AF6 fourth abdominal mammary gland. To ensure that equiv- vector. This work was supported by INSERM, Institut Paoli- alent amount of cells were xenografted in mice, biolumines- Calmettes, and grants from Ligue Nationale Contre le Cancer cence quantification was performed on anesthetized animals (LNCC) (Label 2009–2011) and Institut National du Cancer 5 days after xenograft. Bioluminescence images were acquired (AO translationnelle). In 2009–2010, GF has been supported using a cooled intensified charge-coupled device camera by a fellowship from LNCC.

References

Adelaide J, Finetti P, Bekhouche I, Repellini L, Geneix J, Sircoulomb organization of cell-cell junctions of polarized epithelia during F et al. (2007). Integrated profiling of basal and luminal breast embryogenesis. J Cell Biol 146: 1117–1132. cancers. Cancer Res 67: 11565–11575. Keely PJ, Westwick JK, Whitehead IP, Der CJ, Parise LV. (1997). Bae YH, Ding Z, Zou L, Wells A, Gertler F, Roy P. (2009). Loss of Cdc42 and Rac1 induce integrin-mediated cell motility and profilin-1 expression enhances breast cancer cell motility by Ena/ invasiveness through PI(3)K. Nature 390: 632–636. VASP proteins. J Cell Physiol 219: 354–364. Klemke RL, Cai S, Giannini AL, Gallagher PJ, de Lanerolle P, Boettner B, Govek EE, Cross J, Van Aelst L. (2000). The junctional Cheresh DA. (1997). Regulation of cell motility by mitogen- multidomain protein AF-6 is a binding partner of the Rap1A activated protein kinase. J Cell Biol 137: 481–492. GTPase and associates with the actin cytoskeletal regulator profilin. Letessier A, Garrido-Urbani S, Ginestier C, Fournier G, Esterni B, Proc Natl Acad Sci USA 97: 9064–9069. Monville F et al. (2007). Correlated break at PARK2/FRA6E and Boettner B, Harjes P, Ishimaru S, Heke M, Fan HQ, Qin Y et al. loss of AF-6/Afadin protein expression are associated with poor (2003). The AF-6 homolog canoe acts as a Rap1 effector during outcome in breast cancer. Oncogene 26: 298–307. dorsal closure of the Drosophila embryo. Genetics 165: 159–169. Liedtke M, Ayton PM, Somervaille TC, Smith KS, Cleary ML. (2010). Buchert M, Poon C, King JA, Baechi T, D’Abaco G, Hollande F et al. Self-association mediated by the Ras association 1 domain of AF6 (2007). AF6/s-afadin is a dual residency protein and localizes to a activates the oncogenic potential of MLL-AF6. Blood 116: 63–70. novel subnuclear compartment. J Cell Physiol 210: 212–223. Linnemann T, Geyer M, Jaitner BK, Block C, Kalbitzer HR, Carmena A, Speicher S, Baylies M. (2006). The PDZ protein Canoe/ Wittinghofer A et al. (1999). Thermodynamic and kinetic AF-6 links Ras-MAPK, Notch and Wingless/Wnt signaling path- characterization of the interaction between the Ras binding ways by directly interacting with Ras, Notch and Dishevelled. PLoS domain of AF6 and members of the Ras subfamily. J Biol Chem One 1: e66. 274: 13556–13562. Ding Z, Lambrechts A, Parepally M, Roy P. (2006). Silencing profilin- Lorger M, Moelling K. (2006). Regulation of epithelial wound closure 1 inhibits endothelial cell proliferation, migration and cord and intercellular adhesion by interaction of AF6 with actin morphogenesis. J Cell Sci 119: 4127–4137. cytoskeleton. J Cell Sci 119: 3385–3398. Gaengel K, Mlodzik M.. (2003). Egfr signaling regulates ommatidial Mandai K, Nakanishi H, Satoh A, Obaishi H, Wada M, Nishioka H rotation and cell motility in the Drosophila eye via MAPK/Pnt et al. (1997). Afadin: A novel actin filament-binding protein with signaling and the Ras effector Canoe/AF6. Development 130: one PDZ domain localized at cadherin-based cell-to-cell adherens 5413–5423. junction. J Cell Biol 139: 517–528. Hofmann K, Bucher P. (1995). The FHA domain: a putative nuclear Marone R, Hess D, Dankort D, Muller WJ, Hynes NE, Badache A. signalling domain found in protein kinases and transcription (2004). Memo mediates ErbB2-driven cell motility. Nat Cell Biol 6: factors. Trends Biochem Sci 20: 347–349. 515–522. Ikeda W, Nakanishi H, Miyoshi J, Mandai K, Ishizaki H, Tanaka M Martuszewska D, Ljungberg B, Johansson M, Landberg G, Oslakovic et al. (1999). Afadin: A key molecule essential for structural C, Dahlback B et al. (2009). Tensin3 is a negative regulator of cell

Oncogene Afadin loss in breast cancer and tumorigenicity G Fournier et al 3874 migration and all four Tensin family members are downregulated in cytoskeleton to adherens junctions during apical constriction. J Cell human kidney cancer. PLoS One 4: e4350. Biol 186: 57–73. Matsuo T, Takahashi K, Suzuki E, Yamamoto D. (1999). The Canoe Severson EA, Lee WY, Capaldo CT, Nusrat A, Parkos CA. (2009). protein is necessary in adherens junctions for development of Junctional adhesion molecule A interacts with Afadin and PDZ- ommatidial architecture in the Drosophila compound eye. Cell GEF2 to activate Rap1A, regulate beta1 integrin levels, and Tissue Res 298: 397–404. enhance cell migration. Mol Biol Cell 20: 1916–1925. Miyamoto H, Nihonmatsu I, Kondo S, Ueda R, Togashi S, Hirata K Sorlie T, Perou CM, Tibshirani R, Aas T, Geisler S, Johnsen H et al. et al. (1995). canoe encodes a novel protein containing a GLGF/ (2001). Gene expression patterns of breast carcinomas distinguish DHR motif and functions with Notch and scabrous in common tumor subclasses with clinical implications. Proc Natl Acad Sci USA developmental pathways in Drosophila. Genes Dev 9: 612–625. 98: 10869–10874. Miyata M, Ogita H, Komura H, Nakata S, Okamoto R, Ozaki M et al. Speicher S, Fischer A, Knoblich J, Carmena A. (2008). The PDZ (2009). Localization of nectin-free afadin at the leading edge and its protein Canoe regulates the asymmetric division of Drosophila involvement in directional cell movement induced by platelet- neuroblasts and muscle progenitors. Curr Biol 18: 831–837. derived growth factor. J Cell Sci 122: 4319–4329. Spencer KS, Graus-Porta D, Leng J, Hynes NE, Klemke RL. Nakata S, Fujita N, Kitagawa Y, Okamoto R, Ogita H, Takai Y. (2000). ErbB2 is necessary for induction of carcinoma cell (2007). Regulation of platelet-derived growth factor receptor invasion by ErbB family receptor tyrosine kinases. J Cell Biol 148: activation by afadin through SHP-2: implications for cellular 385–397. morphology. J Biol Chem 282: 37815–37825. Su L, Hattori M, Moriyama M, Murata N, Harazaki M, Kaibuchi K Nikolopoulos SN, Spengler BA, Kisselbach K, Evans AE, Biedler JL, et al. (2003). AF-6 controls integrin-mediated cell adhesion by Ross RA. (2000). The human non-muscle alpha-actinin protein regulating Rap1 activation through the specific recruitment of encoded by the ACTN4 gene suppresses tumorigenicity of human Rap1GTP and SPA-1. J Biol Chem 278: 15232–15238. neuroblastoma cells. Oncogene 19: 380–386. Takahashi K, Matsuo T, Katsube T, Ueda R, Yamamoto D. (1998). Ponting CP. (1995). AF-6/cno: neither a kinesin nor a myosin, but a bit Direct binding between two PDZ domain proteins Canoe and ZO-1 of both. Trends Biochem Sci 20: 265–266. and their roles in regulation of the jun N-terminal kinase pathway in Prasad R, Gu Y, Alder H, Nakamura T, Canaani O, Saito H et al. Drosophila morphogenesis. Mech Dev 78: 97–111. (1993). Cloning of the ALL-1 fusion partner, the AF-6 gene, Takai Y, Ikeda W, Ogita H, Rikitake Y. (2008). The immunoglobulin- involved in acute myeloid leukemias with the t(6;11) chromosome like cell adhesion molecule nectin and its associated protein afadin. translocation. Cancer Res 53: 5624–5628. Annu Rev Cell Dev Biol 24: 309–342. Radziwill G, Erdmann RA, Margelisch U, Moelling K. (2003). The Tanaka M, Mullauer L, Ogiso Y, Fujita H, Moriya S, Furuuchi K Bcr kinase downregulates Ras signaling by phosphorylating AF-6 et al. (1995). Gelsolin: a candidate for suppressor of human bladder and binding to its PDZ domain. Mol Cell Biol 23: 4663–4672. cancer. Cancer Res 55: 3228–3232. Radziwill G, Weiss A, Heinrich J, Baumgartner M, Boisguerin P, Vadlamudi R, Adam L, Talukder A, Mendelsohn J, Kumar R. (1999). Owada K et al. (2007). Regulation of c-Src by binding to the PDZ Serine phosphorylation of paxillin by heregulin-beta1: role of p38 domain of AF-6. EMBO J 26: 2633–2644. mitogen activated protein kinase. Oncogene 18: 7253–7264. Saha V, Lillington DM, Shelling AN, Chaplin T, Yaspo ML, Ganesan Yamamoto T, Harada N, Kawano Y, Taya S, Kaibuchi K. (1999). TS et al. (1995). AF6 gene on chromosome band 6q27 maps distal to in vivo interaction of AF-6 with activated Ras and ZO-1. Biochem the minimal region of deletion in epithelial ovarian cancer. Genes Biophys Res Commun 259: 103–107. Chromosomes Cancer 14: 220–222. Zhadanov AB, Provance Jr DW, Speer CA, Coffin JD, Goss D, Blixt Saito S, Matsushima M, Shirahama S, Minaguchi T, Kanamori Y, JA et al. (1999). Absence of the tight junctional protein AF-6 Minami M et al. (1998). Complete genomic structure DNA disrupts epithelial cell-cell junctions and cell polarity during mouse polymorphisms, and alternative splicing of the human AF-6 gene. development. Curr Biol 9: 880–888. DNA Res 5: 115–120. Zou L, Jaramillo M, Whaley D, Wells A, Panchapakesa V, Das T et al. Sawyer JK, Harris NJ, Slep KC, Gaul U, Peifer M. (2009). The (2007). Profilin-1 is a negative regulator of mammary carcinoma Drosophila afadin homologue Canoe regulates linkage of the actin aggressiveness. Br J Cancer 97: 1361–1371.

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

Oncogene