The lipid-transfer protein Nir2 enhances epithelial-mesenchymal transition and facilitates breast cancer metastasis

Increasing lines of evidence suggest that epithelial-mesenchymal transition (EMT), a crucial developmental process implicated in organ development, tissue repair and organ fibrosis, plays a major role in cancer progression, invasion and metastasis (Xiao and He, 2010). EMT is defined by the loss of epithelial characteristics and the acquisition of a motile, invasive and migratory mesenchymal phenotype (Iwatsuki et al., 2010). This conversion is accompanied by loss of cell–cell adhesion and cell polarity, decrease in the expression of epithelial markers such as E-cadherin, increase in the expression of mesenchymal markers such as vimentin, fibronectin, N-cadherin, alpha-smooth muscle actin (α-SMA), as well as increase in the activity of matrix metalloproteinases (MMPs) (Thiery and Sleeman, 2006). In carcinoma cells, EMT is generally associated with increased aggressiveness, invasion and metastasis, in particular, in the early stages of metastasis (invasion and intravasation), when tumor cells detach from the primary tumor to invade the surrounding stroma and enter the circulation (Savagner, 2001; Thiery, 2002; Gupta and Massagué, 2006).

EMT is a transcriptionally regulated process that is mediated, at least in part, by a number of specific transcription factors including Snai1 and Snai2 (hereafter referred to as Snail1 and Snail2, respectively), ZEB1 and ZEB2, Twist1 and Twist2. Expression of these transcription factors in untransformed epithelial cells efficiently induces EMT (Lee et al., 2006; Savagner, 2010). The activity and expression levels of these transcription factors are controlled by multiple signaling cascades, including the TGFβ signaling pathway, the Wnt, Notch and Hedgehog pathways, as well as several growth factor receptor (GFR) signaling cascades, including the receptors of epidermal growth factor (EGF), insulin-like growth factor 1 (IGF1), fibroblast growth factor (FGF) and hepatocyte growth factor (HGF; also known as MET) (Xiao and He, 2010). The TGFβ pathway, which has been most extensively studied in the context of EMT, cooperates with other signaling cascades including the Wnt, Hedgehog, Notch, Ras–MAPK and the PI3K/AKT pathway to induce EMT (Huber et al., 2005; A. Bellacosa, 2010). The PI3K/AKT pathway is known to play a critical role in human cancer initiation and progression, and is also associated with the induction of EMT (Irie et al., 2005). Expression of activated AKT in epithelial cells causes loss of cell–cell adhesion, loss of apical–basolateral cell polarization, induction of cell motility, and changes in the expression or the distribution of various epithelial and/or mesenchymal markers (Grille et al., 2003).

Recently, we found that the lipid-transfer protein Nir2 (also known as PITPNM1) positively regulates the PI3K/AKT and the MAPK signaling pathways in response to EGF treatment in HeLa and MCF7 cells (Kim et al., 2013). We further showed that Nir2, which contains an N-terminal phosphatidylinositol (PtdIns)-transfer domain (Lev et al., 1999) and mainly localizes at the Golgi complex (Litvak et al., 2005), translocates to the plasma membrane upon stimulation with EGF. At the plasma membrane, Nir2 appears to regulate the PtdIns(4,5)P₂ level and, consequently, the production of PtdIns(3,4,5)P₃, thereby activating the ERK1 and ERK2 (hereafter referred to as ERK1/2), and AKT pathways (Kim et al., 2013). These findings introduced Nir2 as an upstream regulator of the PI3K/AKT and the MAPK signaling pathways, which are activated in response to GFR stimulation.

Here, we show that Nir2 is also an important regulator of cell migration and invasion, and a new modulator of EMT in mammary epithelial and breast carcinoma cells. We further show that depletion of Nir2 expression inhibits breast cancer metastasis in animal models, and that Nir2 expression correlates with high tumor grade and poor disease outcome in breast cancer patients. These new findings demonstrate the important physiological roles of lipid-transfer proteins, their implication in human diseases, and highlight Nir2 as a potential target for therapeutic intervention.

We have previously shown that Nir2 translocates from the Golgi complex to the plasma membrane of HeLa and MCF7 cells in response to EGF treatment (Kim et al., 2013). More recently we observed that Nir2 is expressed in many human breast carcinomas, including MDA-MB-231, T47D and SKBR3 (supplementary material Fig. S1A). Furthermore, several growth factors such as EGF, neuregulin 1 (NRG1) and insulin-like growth factor 1 (IGF1, hereafter referred to as IGF-I) trigger its translocation from the Golgi complex to the plasma membrane (Fig. 1A; supplementary material Fig. S1B,C). These growth factors also enhance the migration of various breast cancer cell lines including that of the highly invasive MDA-MB-231 line (Price et al., 1999; Bartucci et al., 2001; Tsai et al., 2003). Hence, we examined whether Nir2 affects the migration of MDA-MB-231 cells in response to migratory stimuli. The cells were infected with lentiviruses expressing three different shRNAs of Nir2. Two of them (shRNA number 1 and shRNA number 2) efficiently reduced the expression of Nir2 (Fig. 1B) and were used in all the subsequent experiments. The effect of Nir2 depletion on cell migration was assessed in wound-healing (Fig. 1B,C) and transwell migration assays (Fig. 1D). As shown, Nir2 depletion substantially inhibited wound closure in response to EGF, NRG1 and IGF-I as well as to 10% serum. The migration of Nir2-depleted MDA-MB-231 cells through a transwell chamber was also markedly (∼63%) attenuated (Fig. 1D).

Previous studies suggested that the PI3K/AKT pathway mediates the migratory responses of IGF-I and EGF in MDA-MB-231 cells (Price et al., 1999; Bartucci et al., 2001). The MAPK pathway is also involved in EGF-induced MDA-MB-231 cell migration (Harrison et al., 2013). We, therefore, examined the influence of Nir2 depletion on these signaling cascades. As shown, depletion of Nir2 markedly reduced the phosphorylation of AKT in response to EGF (Fig. 1E; supplementary material Fig. S1D), consistent with our previous results in HeLa and MCF7 cells (Kim et al., 2013). However, we could not detect any effect on ERK1/2 phosphorylation, possibly owing to the activating mutations within the B-Raf proto-oncogene (BRAF) and the Kirsten rat sarcoma viral oncogene homolog (KRAS) in this specific cell line (Hollestelle et al., 2007). These results suggest that Nir2 regulates the migration of MDA-MB-231 cells mainly through the PI3K/AKT pathway.

Although reduction of Nir2 expression in MDA-MB-231 cells markedly attenuated cell migration, Nir2 overexpression had no obvious effects on the migratory response of these highly motile cells. Nonetheless, overexpression of Nir2 in either the human breast carcinoma cell line T47D or the non-transformed human mammary cell line MCF10A enhanced (approximately twofold) NRG1- or EGF-induced cell migration, respectively (supplementary material Fig. S2A,B). Nir2 overexpression also enhanced TGFβ-induced migration (Fig. 2C) in MCF10A cells, whereas silencing of Nir2 markedly attenuated collective as well as individual cell migration in response to either TGFβ or EGF, as assessed in wound-healing and transwell migration assays (Fig. 2A-C), respectively. These results suggest that Nir2 plays an important role in the migratory responses of MCF10A cells.

Previous studies suggested that the migratory response to EGF and TGFβ in MCF10A cells are mediated, mainly, through the ERK pathway (Kim et al., 2004; Tarcic et al., 2012). We, therefore, examined the effects of Nir2 on ERK1/2 phosphorylation. As shown, overexpression of Nir2 in MCF10A cells enhanced the phosphorylation of ERK1/2 (Fig. 2D), whereas Nir2 depletion reduced ERK1/2 phosphorylation in response to either EGF or TGFβ (Fig. 2D; supplementary material Fig. S2B), consistent with its stimulatory or inhibitory effects on cell migration, respectively (Fig. 2A–C). Furthermore, inhibition of ERK1/2 phosphorylation by the MEK1/2 inhibitor U0126 (supplementary material Fig. S2C), markedly attenuated the migration of MCF10A cells in response to EGF (by 70%) and TGFβ (by 50%) and, apparently, abolished the stimulatory effect of Nir2 overexpression on MCF10A migration (Fig. 2E), suggesting that Nir2 affects the migration of MCF10A cells primarily through the ERK pathway.

The marked effect of Nir2 on TGFβ-mediated responses in MCF10A cells (Fig. 2C-E) led us to examine its involvement in EMT, because TGFβ is a potent inducer of EMT (Tumbarello and Turner, 2007). We first examined the effects of TGFβ and EGF, two known EMT inducers of mammary cells (Hardy et al., 2010; Imamura et al., 2012), on the protein and mRNA expression levels of Nir2. TGFβ induces EMT in MCF10A cells (Tumbarello and Turner, 2007) and enhances the mesenchymal properties of MDA-MB-231 cells (Romagnoli et al., 2012), whereas EGF induces EMT in the human breast cancer cell line MDA-MB-468 (Lo et al., 2007). As shown in Fig. 3A, both TGFβ and EGF increased the levels of Nir2 protein in MCF10A, MDA-MB-231 and MDA-MB-468 cells. These elevated levels of Nir2 protein were accompanied by an increase in fibronectin or a decrease in E-cadherin levels, characteristic markers of mesenchymal and epithelial cells, respectively. TGFβ and EGF also enhanced the mRNA levels of Nir2 (Fig. 3B), suggesting that Nir2 is transcriptionally regulated during EMT.

Next, we asked whether Nir2 affects the mesenchymal or epithelial properties of these mammary cell lines. Lentiviruses encoding Nir2-Myc or shRNA targeting Nir2 were used to infect MCF10A, MDA-MB-231 and MDA-MB-468 cells. The effect of Nir2 overexpression or its downregulation on the level of mesenchymal (fibronectin, vimentin, CD44) and epithelial (E-cadherin, ZO-1) protein markers was assessed by western blotting. As shown, overexpression of Nir2 increased the level of mesenchymal markers or decreased the level of epithelial markers, whereas downregulation of Nir2 had opposite effects (Fig. 3C,E; supplementary material Fig. S3A,B). In MCF10A cells, Nir2 overexpression markedly decreased the levels of the tight-junction protein ZO-1 and, consequently, disrupted the integrity of cell–cell contacts as shown by immunostaining of the tight-junction protein ZO-1 and the adherens-junction protein E-cadherin (Fig. 3D). Furthermore, Nir2 expression affected EGF- and TGFβ-induced EMT markers in MDA-MB-468 (Fig. 3F) and in MCF10A cells (Fig. 3G; supplementary material Fig. S3C), respectively. Collectively these results suggest that Nir2 positively regulates EMT.

Downregulation of E-cadherin expression is a hallmark of EMT, and is regulated by specific transcription factors including Snail1 and Snail2, Twist1 and Twist2, ZEB1 and ZEB2 and E47 (TCF3) (Batlle et al., 2000; Bolós et al., 2003). These factors are involved in most physiological EMT processes, and their overexpression in epithelial cell lines often induces EMT (Savagner, 2010). We, therefore, examined the influence of Nir2 on the expression levels of these transcription factors using RT-PCR. Depletion of Nir2 expression in MDA-MB-231 cells by two different shRNAs substantially reduced the expression of Snail1 and ZEB1 and concomitantly increased the expression level of E-cadherin (Fig. 3H), whereas overexpression of Nir2 in MCF10A cells enhanced the expression of Snail1, Twist1, Twist2 and ZEB2 (Fig. 3I). Importantly, expression of the wild-type Nir2 (Nir2-Myc) in Nir2-depleted MDA-MB-231 cells restored the expression levels of Snail1 and ZEB1, thus demonstrating the specificity of the shRNAs effects (Fig. 3H). Furthermore, expression of a myristoylated (constitutively active) form of AKT (Myr-AKT) also restored the expression of vimentin and CD44 in Nir2-depleted MDA-MB-231 and MCF10A cells (Fig. 4A,B). These results suggest that the PI3K/AKT pathway plays a central role in Nir2-enhanced EMT, consistent with the stimulatory effect of Nir2 on this pathway (Figs 1E, 2D).

The marked effects of Nir2 on the expression levels of EMT markers (Fig. 3), suggest that Nir2 influences EMT-related cellular processes including cell shape, migration and invasion. Indeed, silencing of Nir2 in MDA-MB-231 cells, which display mesenchymal cell morphology, induced morphological changes characteristic of epithelial cells (Fig. 5A); the cells lost their spindle-like morphology and grew in clusters with established cell–cell contacts. However, overexpression of Nir2 in MCF10A cells, which display epithelial cell morphology, reduced cell–cell contacts and triggered cell scattering (Fig. 5A) – characteristic features of mesenchymal cells.

An additional characteristic feature of EMT is increased invasiveness. We, therefore, examined the influence of Nir2 on the invasion of MCF10A and MDA-MB-231 cells. As shown in Fig. 5B, silencing of Nir2 in either MCF10A or MDA-MB-231 cells reduced cell invasion through Matrigel by ∼55% or ∼63%, respectively. Since matrix metalloproteinases (MMPs) play a major role in cell invasion, we examined the activity of two prominent enzymes; MMP2 and MMP9, in control and Nir2-depleted MCF10A or MDA-MB-231 cells by using the gelatin zymography assay. As shown in Fig. 5C, the activity of MMP2 and MMP9 was significantly reduced in Nir2-depleted cells as compared to control cells. The MMP9 protein level in the conditioned media was also reduced (Fig. 5C), suggesting that Nir2 influences the level of these enzymes rather than their activity. To determine whether Nir2 affects the transcription of these enzymes, we examined the mRNA levels of MMP2 and MMP9 as well as of MMP14 in control and Nir2-depleted MCF10A and MDA-MB-231 cells. As shown, depletion of Nir2 reduced the expression levels of these MMPs (Fig. 5D). Interestingly, previous studies suggest that MMP2, MMP9 and MMP14 are involved in breast cancer progression (Somiari et al., 2006; Têtu et al., 2006; Köhrmann et al., 2009), and that AKT activity promotes MMP9 transcription in MDA-MB-231 cells (Cho et al., 2008). Hence, the reduced activation of AKT in Nir2-depleted MDA-MB-231 and MCF10A cells (Figs 1E, 2D; supplementary material Figs S1D, S2B), may cause a decrease in MMP9 transcription. Overall, our results suggest that Nir2 positively regulates EMT and, consequently markedly affects cell morphology and invasion in vitro.

The profound effects of Nir2 on cell migration and invasion in vitro, led us to examine its influence on breast cancer metastasis in vivo. MDA-MB-231 cells, which are commonly used in experimental metastasis models, form tumors in the lungs and other organs after tail vein injection. Control and Nir2-depleted MDA-MB-231 cells were injected intravenously into the lateral tail vein of 5-week-old SCID mice. The mice were sacrificed 10 weeks later, the lungs were removed, and the number of metastatic foci on the surface of the lungs was assessed. As shown in Fig. 6A, control MDA-MB-231 cells produced large and numerous metastatic foci (∼200 lung metastases/mouse). Depletion of Nir2, however, markedly reduced the number of lung metastases (∼20 lung metastases/mouse) (Fig. 6B), and only few micrometastases were detected following H&E staining of lung sections (Fig. 6C). This profound reduction in the number of both metastatic and micrometastatic lesions suggests that loss of Nir2 mainly affects the formation of metastatic foci. It could be, however, that Nir2 also slightly affects metastatic outgrowth as it inhibits the proliferation of the cells by ∼28% (supplementary material Fig. S3D).

The striking effect of Nir2 depletion on lung metastasis in animal models led us to hypothesize that the expression levels of Nir2 correlate with poor patient prognosis. To explore this possibility, we used the expression data of 1693 breast cancer patients that had been described by Curtis et al., (Curtis et al., 2012) and found that breast cancer patients who showed high Nir2 expression had a poorer survival rate than patients with low expression (Fig. 7A). Furthermore, immunohistochemistry analysis of paraffin-embedded sections of human breast tissue samples revealed that all samples from healthy individuals had no or a very weak staining for Nir2, whereas the majority (43/56) of cancer samples had moderate to strong Nir2 staining (Fig. 7B,C). Remarkably, positive Nir2 staining was observed in 91% of high-grade tumors, whereas only 64% of low- to intermediate-grade tumors stained positively for Nir2 (P = 0.029, χ²-statistics, Fig. 7D), indicating a correlation between high-grade, more aggressive tumors with poorer prognosis and Nir2 protein expression. Moreover, analysis of Nir2 expression and EMT markers in clinical breast cancer samples (n = 4467) suggests that Nir2 expression adversely affects the prognosis of patients with an elevated EMT marker score (supplementary material Table S1; Fig. S3E,F). Samples that showed high expression of both Nir2 and EMT markers, as compared to those with only high expression of EMT markers, correlated with a statistically significant decreased survival of the patient. Collectively, our results suggest that high expression of Nir2 enhances EMT in vitro, facilitates cancer metastasis in animal models and correlates with poor prognosis of human patients.

Metastasis, the spreading of cancer cells from a primary tumor to seed secondary tumors in distant sites, is the main cause of mortality in breast cancer patients (Oppenheimer, 2006). Recent studies suggest that EMT plays a central role in motility and dissemination of cancer cells (Thiery et al., 2009). This dynamic and reversible process is necessary for efficient metastatic colonization (Brabletz, 2012) and is regulated by several signaling cascades that activate specific transcription factors (Huber et al., 2005).

In this study, we show that the PtdIns-transfer protein Nir2 potentiates EMT in mammary cells (Fig. 3) and is required for lung metastasis of breast cancer cells (Fig. 6). Remarkably, the expression level of Nir2 is upregulated in response to EMT inducers (Fig. 3A), such as EGF and TGFβ, and is markedly enhanced in human breast cancer tissue (Fig. 7). These results introduce Nir2 as a new positive regulator of EMT and breast cancer metastasis. Indeed, depletion of Nir2 by shRNA substantially inhibits cell migration and invasion, whereas its overexpression had opposite effects (Figs 1, 2 and Fig. 5B). Consistent with these results, previous studies have demonstrated that Nir2 expression is upregulated in invading endothelial cells grown in 3D collagen matrices as well as in mobile hematopoietic stem cells (Gan et al., 2008; Su et al., 2008). However, the underlying mechanisms of Nir2 function in these cells remain unknown.

Our studies suggest that Nir2 mediates its effects in mammary epithelial cells predominantly through the PI3K/AKT and the ERK signaling pathways (Fig. 1E and Fig. 2D; supplementary material Figs S1D, S2B). These pathways contribute to all aspects of breast cancer progression, including cell growth, survival, migration, invasion and metastasis (Whyte et al., 2009; Pal and Mandal, 2012). Furthermore, the PI3K/AKT pathway is abnormally activated in about 30% of breast cancer patients (Cancer Genome Atlas Network, 2012), and is required for EMT. Specific inhibitors of the PI3K/AKT pathway, dominant-negative AKT or a Myr-AKT mutant markedly affect EMT in mammary cells (Bakin et al., 2000; Grille et al., 2003). We show here that Myr-AKT can rescue the effects of Nir2 depletion on the expression of vimentin and CD44 (Fig. 4), two characteristic EMT markers, thus demonstrting the central role of AKT activation in Nir2-mediated EMT effects. Interestingly, previous studies have shown that CD44s also enhances EMT through activation of AKT (Brown et al., 2011), suggesting a positive feedback during EMT.

The influence of Nir2 on PI3K/AKT and ERK signaling pathways is possibly related to its PtdIns-transfer activity and its effect on PtdIns(4,5)P₂ and PtdIns(3,4,5)P₃ production at the plasma membrane – as described in a recent publication of ours (Kim et al., 2013). Our findings were further supported by an independent study demonstrating that Nir2, through its PtdIns-transfer activity, promotes the replenishment of PtdIns(4,5)P₂ at the plasma membrane after receptor-induced PtdIns(4,5)P₂ hydrolysis (Chang et al., 2013). Both studies suggest that Nir2 influences PtdIns(4,5)P₂ levels through its plasma-membrane-associated pool, whereas its Golgi-associated pool possibly ensures efficient transport (Kim et al., 2013). In this context, it is important to mention that depletion of Nir2 by using shRNA in either MCF10A or MDA-MB-231 cells had no effect on Golgi-to-plasma membrane transport (supplementary material Fig. S4A).

Although Nir2 may influence cell migration and invasion through PtdIns-mediated pathways, we cannot exclude other mechanisms such as interaction with other cellular proteins. Consistent with this hypothesis, we have previously shown that Nir2 interacts with proteins that regulate cell migration and invasion, such as the small GTPase RhoA (Tian et al., 2002) and the tyrosine kinase PYK2 (Lev et al., 1999). Although we could not detect an association with either RhoA or PYK2 in these cells, we found Nir2 in focal adhesions after re-plating MDA-MB-231 cells on fibronection-coated coverslips (not shown). Thus, it could be that Nir2, through its multiple structural domains (Lev, 2004), can regulate different migratory events, such as actin organization, focal adhesion turnover and cell spreading. Indeed, depletion of Nir2 in MDA-MB-231 cells markedly affected the organization of the actin cytoskeleton and the formation of the leading edge (supplementary material Fig. S4B). Furthermore, recent studies suggest that VAPB, an integral ER-protein that interacts with Nir2 through a specific sequence motif known as FFAT (two phenylalanine residues in an acidic tract) (Amarilio et al., 2005), regulates the proliferation of breast tumor cells and activation of AKT (Rao et al., 2012). VAPB expression levels are also elevated in primary and metastatic breast cancer specimens, and correlate negatively with patient survival (Rao et al., 2012). Similarily, we show here that high expression of Nir2 correlates with poor prognosis and bad disease outcome (Fig. 7). Furthermore, depletion of Nir2 expression in MDA-MB-231 markedly reduced the number and size of lung metastases in a mice xenograft tail-vain-injection model. Although the tail-vain-injection model is commonly used to study lung metastasis of breast cancer (Rashid et al., 2013), it bypassess the early steps of the metaststic cascade, including local invasion and intravasation, and addresses late metastic events such as extravasion and local outhgrowth (Khanna and Hunter, 2005). We assume that Nir2 is involved in both early and late metaststic steps, as it markedly affects EMT-associated events (Figs 3, 5) that regulate early metastatic steps and also influences late events, as shown in Fig. 6.

Taken together, our results demonstrate for the first time that Nir2 is a regulator of EMT in breast cancer cells, and that Nir2 expression affects cell migration and invasion in vitro, lung metastasis in animal models and disease outcome in human patients. Further studies on Nir2 and its role in cell migration, invasion and metastasis could shed light on the mechanisms by which lipid-transfer proteins regulate key cellular processes and contribute to cancer metastasis.

MCF10A cells were grown in DMEM∶F12 (1∶1) medium supplemented with EGF (10 ng/ml), insulin (10 µg/ml), cholera toxin (1 µg/ml), hydrocortisone (1 µg/ml) and heat-inactivated horse serum (5%). MDA-MB-231, T47D, SKBR3 and MDA-MB-468 cells were grown in RPMI (Gibco BRL; Grand Island, NY) medium supplemented with 10% fetal calf serum (FCS). Human embryonic kidney (HEK-293T) cells were grown in DMEM (Gibco BRL; Grand Island, NY) supplemented with 10% heat-inactivated fetal bovine serum (FBS) (Gibco BRL, Grand Island, NY), 1 mM sodium pyruvate and a penicillin-streptomycin mixture (100 units/ml; 0.1 mg/ml; Beit Haemek, Israel).

TGFβ and IGF-I were purchased from ProSpec (Israel), NRG1 was purchased from PeproTech (Israel). EGF, Hoechst 33342, U0126 and other chemicals were purchased from Sigma-Aldrich. Monoclonal antibodies against Myc, ERK1/2 and phosphorylated ERK1/2 (pERK1/2) were purchased from Santa Cruz Biotechnology (Santa Cruz, CA). Antibodies against AKT and phosphorylated AKT (pAKT T308 and pAKT S473) were purchased from Cell Signaling Technologies (Beverly, MA). Antibodies against fibronectin, CD44 and vimentin were purchased from Developmental Studies Hybridoma Bank (Iowa City, IA) Monoclonal antibodies against general vesicular transport factor p115 (USO1) as well as rabbit polyclonal antibodies against Nir2 have been described previously (Peretti et al., 2008). Goat polyclonal antibody against Nir2 was purchased from Abcam (Cambridge, UK). Monoclonal antibodies against α-tubulin and Rac1 were purchased from Sigma and BD (San Jose, California), respectively. Alexa-Fluor-488 donkey anti-mouse as well as anti-rabbit immunoglobulin Gs (IgGs) were purchased from Invitrogen (Carlsbad, CA). Cyanine (Cy)3-conjugated goat anti-rabbit and goat anti-mouse IgGs, as well as Cy5-conjugated goat anti-mouse IgG were purchased from Jackson ImmunoResearch Laboratories (West Grove, PA).

Lentiviruses encoding shRNAs targeting Nir2 or Myc-tagged wild-type Nir2 (Nir2-Myc) have been described previously (Kim et al., 2013). Lentiviral vector encoding myristoylated (constitutively active) AKT (Myr-Akt) was established by subcloning the Myr-AKT cDNA from the pCIS2-AKT-Myr vector (kindly provided by Michael J. Quon, University of Maryland) into the PHAGE lentiviral vector. Lentivirus production and infection were conducted essentially as previously described (Kim et al., 2013). Infected MCF10A or T47D cells were grown in selection medium containing 100 µg/ml hygromycin, whereas MDA-MB-231 cells were grown in the presence of 400 µg/ml hygromycin for 72 hr.

The MTT [3-(4, 5-dimethylthiazolyl-2)-2,5-diphenyltetrazolium bromide] assay was used to measure the proliferation of MDA-MB-231 and MCF10A, essentially as previously described (Mosmann, 1983).

Cells were grown on coverslips, washed with PBS and fixed, as described in the figure legends. To analyse localization of Nir2, cells grown on coverslips were fixed in 1% PFA in a hypotonic buffer (10 mM 2-(N-morpholino) ethanesulfonic acid pH 6.2, 10 mM NaCl, 1.5 mM MgCl₂ and 2.5% glycerol) for 20 min at room temperature and immunostained as described previously (Peretti et al., 2008). The cells were then incubated for 15 min in PBS containing 0.1 M glycine, incubated in blocking buffer containing 0.1% Triton X-100, 10% goat serum and 2% BSA in TBS for 30 min, followed by 1-hour incubation with the primary antibody, and 1-hour incubation with the secondary antibody. In specific experiments, coverslips were coated with 20 µg/ml of fibronectin for 16 hours at 4°C. The specimens were analyzed by using a confocal laser scanning microscope (Zeiss 510; Carl Zeiss, Jena, Germany).

RNA was purified using the EZ-RNA kit (Beit Haemek, Israel). cDNA was generated using the SuperScriptII first-strand synthesis kit (Invitrogen; Carlsbad, CA). Reverse transcriptase (RT)-PCR was performed using Red-Taq mix (Sigma-Aldrich, Israel). Real-time PCR analysis was performed using SYBR Green I as a fluorescent dye, according to the manufacturer's guidelines (Invitrogen). All experiments were carried out in triplicates and normalized to actin RNA levels. Real-time PCR primers were designed using the Primerexpress software of Applied Biosystems (Invitrogen).

To detect MMP2 and MMP9 activity, conditioned medium was separated electrophortically on 10% polyacrylamide/0.1% gelatin-embedded gels. The gels were then washed in 2.5% Triton X-100, and incubated at 37°C for 24 hours in 50 mM Tris-HCl (pH 7.5), containing 0.2 M NaCl, 5 mM CaCl₂, 0.02% Brij 35, and stained using Coomassie Brilliant Blue (Pierce).

Wound-healing assays were performed according to the manufacturer's protocol (iBidi, Germany). Briefly, cells were trypsinized and diluted in normal medium at 10⁶ cells/ml and 70 µl were plated into each well, resulting in a confluent layer within 16 hr. Thereafter, the insert was removed and cells were allowed to migrate for different durations in the presence or absence of growth factor. Graphical presentation of the percentage of the wound healing was measured with the following formula: (initial width − time point width of wound) ÷ initial width of wound. The experiments were done in the presence of 10 µM mitomycin C to inhibit cell proliferation.

Cells [MDA-MB-231 (45×10³ cells/insert), MCF10A (50×10³ cells/insert) and T47D (80×10³ cells/insert)] were plated in the upper compartment of a Transwell tray (BD Bioscience, San Jose, California). For T47D cells, the lower compartment was coated with 25 ug/ml collagen for 2 h prior to cell plating. Cells were allowed to migrate through the intervening nitrocellulose membrane for 18 hr in the presence of growth factor in the lower chamber. Thereafter, cells were fixed in saline-containing paraformaldehyde (PFA; 3%), permeabilized in Triton X-100 (0.05%) and stained with Methyl Violet (0.02%). Non-migrating cells, growing on the upper side of the filter, were removed and cells that had migrated photographed. For cell invasion assays, 60×10³ cells/insert (MDA-MB-231) or 80×10³ cells/insert (MCF10A) were plated in the upper compartment of BioCoat Matrigel Chambers (BD Bioscience, San Jose, California) and processed as above.

All animal procedures were carried out in accordance with the Guidelines for the Care and Use of Research Animals at the Weizmann Institute. Female CB-17 severe combined immunodeficient (SCID) mice were maintained and treated under specific pathogen-free conditions. Cells (7.5×10⁵/mouse) were injected into the tail vein of 5-week-old female mice (15 mice/group). Ten weeks post injection, mice were sacrificed, the lungs were removed and fixed in Bouin solution for 24 hours. The lungs were imaged and nodules were counted. Representative slices were stained with hematoxylin and eosin (H&E). Statistical analyses used t-test.

Kaplan-Meyer analysis of the data from breast cancer samples (Curtis et al., 2012) was done using a software written by Assif Yitzhaky (The Weizmann Institute of Science).

Tissue samples of invasive breast cancer cases were obtained with institutional review board approval and informed consent from patients undergoing surgical resection at the Department of Gynecology and Obstetrics at the Goethe-University in Frankfurt am Main (Germany). Samples were characterized according standard pathology, including estrogen receptor, progesterone receptor and HER2 status. Formalin-fixed paraffin-embedded sections (3 µm) were mounted on Superfrost Plus slides. Immunohistochemistry was performed as follows; 3-µm serial sections from formalin-fixed, paraffin-embedded breast carcinoma were deparaffinized, rehydrated and pre-treated for antigen retrieval by microwave treatment for 10 min in 10 µM of citrate buffer pH 6.0. Samples were then incubated with goat anti-Nir2 antibody (Abcam) overnight at RT. Biotinylated rabbit anti-goat antiserum (DakoCytomation) served as secondary antibody. Strept/AB was used according to the manufacturer's protocol (DakoCytomation). After rinsing with phosphate buffer, sections were counterstained with DAB (Merck), dehydrated and mounted on coverslips.

Samples from 56 breast cancer patients and from seven healthy individuals were stained. The intensity of specific staining was characterized as not present (0), weak (1+), moderate (2+) and strong (3+). For each sample, a summary value, referred to as H-Score, was calculated (Kraus et al., 2012). This consists of the sum of the percentages of positively stained cells multiplied by a weighted intensity of staining. H-Score = Σ P_i, (i+1), where P_i, is the percentage of stained cells in each intensity category, and i is the intensity for i = 0, 1, 2, 3. Tumor grade was evaluated according clinical and pathological data (supplementary material Table S2). Statistics were performed using χ²-analysis.

The densitometric analysis of western blot bands was determined using NIH ImageJ software. Results are presented as means ± s.d. Differences between experimental conditions were determined by two-tailed Student's t test. *P<0.05, **P<0.01 and ***P<0.001 denote statistical relevance.

S.L. is the incumbent of the Joyce and Ben B. Eisenberg Chair of Molecular Biology and Cancer Research. We thank Assif Yitzhaky and Dvir Netanely for bioinformatics assistance and Nir Ben-Chetrit for excellent suggestions.