Oncogene (2017), 1–11 © 2017 Macmillan Publishers Limited, part of Springer Nature. All rights reserved 0950-9232/17 www.nature.com/onc

ORIGINAL ARTICLE

Slug/Pcad pathway controls epithelial cell dynamics in mammary gland and breast carcinoma Y Idoux-Gillet1, M Nassour1, E Lakis1, F Bonini, C Theillet, S Du Manoir and P Savagner Mammary gland morphogenesis results from the coordination of proliferation, cohort migration, apoptosis and stem/progenitor cell dynamics. We showed earlier that the transcription repressor Slug is involved in these functions during mammary tubulogenesis. Slug is expressed by a subpopulation of basal epithelial cells, co-expressed with P-cadherin (Pcad). Slug-knockout mammary glands showed excessive branching, similarly to Pcad-knockout. Here, we found that Slug unexpectedly binds and activates Pcad promoter through E-boxes, inducing Pcad expression. We determined that Pcad can mediate several functions of Slug: Pcad promoted clonal mammosphere growth, basal epithelial differentiation, cell–cell dissociation and cell migration, rescuing Slug depletion. Pcad also promoted cell migration in isolated cells, in association with Src activation, focal adhesion reorganization and cell polarization. Pcad, similarly to Slug, was required for in vitro 3D tubulogenesis. Therefore, Pcad appears to be responsible for epithelial–mesenchymal transition-linked plasticity in mammary epithelial cells. In addition, we found that genes from the Slug/Pcad pathway components were co-expressed and specifically correlated in human breast carcinomas subtypes, carrying pathophysiological significance. Oncogene advance online publication, 9 October 2017; doi:10.1038/onc.2017.355

INTRODUCTION Complex signaling pathways orchestrate breast morphogenesis. Recently, we and others described how Slug/Snail2, a member of the snail family of epithelial–mesenchymal transition (EMT) genes, was required for proper mammary tubulogenesis.1,2 We defined distinct functions for Slug in cell proliferation, differentiation and survival. In brief, Slug appeared to control stem/progenitor cell dynamics. As an EMT transcription factor (EMT-TF) gene, Slug is known to induce cell–cell dissociation, motility and morphogenesis in distinct cell types.3–7 Slug is also linked to invasiveness and to basal cell phenotype in human breast carcinoma.8–11 However, few direct transcriptional targets have been well characterized. The best defined target, E-cadherin (Ecad), is downregulated in Slug-overexpressing transformed cell lines.9,12 In mammary gland, Ecad is mostly expressed by luminal epithelial cells, whereas P-cadherin (Pcad) is exclusively expressed by basal/myoepithelial cells, similarly to Slug. Correct expression pattern is required for tubulogenesis.13 These cadherins can partially compensate each other in mediating cell–cell adhesion. Pcad promotes collective cell migration14,15 and is correlated with poor prognosis in breast cancer.16,17 Here, we scrutinized possible links between Slug and Pcad in mammary gland and breast cancer. RESULTS Slug controls Pcad gene expression in mouse mammary gland We looked at Pcad expression pattern in Slug-KO mammary gland. Immunostaining showed a marked downregulation of Pcad in the basal layer of Slug-KO mammary tubules (Figure 1a). Myoepithelial cells appeared normal, based on smooth muscle actin and cytokeratin 5 (CK5) expression. This was confirmed by rtPCR from

Slug-KO mammary glands (Figure 1b) showing a striking Pcad downregulation. We transfected mammary epithelial CommaDβ cells with SiRNAs targeting Slug. Slug repression resulted in a significant decrease of Pcad expression detected by rtPCR analysis (Figure 1b). Western blot analysis showed a transient and significant decrease in Slug but also in Pcad protein levels (Figures 1c and d), culminating around 48 h. Slug and Pcad protein levels were strongly correlated along this timeline (R = 0.69, P o0.05). We obtained similar results in basal keratinocyte HaCaT cells transfected with siRNA-targeting Slug or Pcad (Figure 1e). Finally, we overexpressed Slug, inducing a significant increase of Pcad expression (Figure 1f). These results indicate that Slug can activate Pcad expression, directly or indirectly. In addition, we investigated Ecad and Ncad/Cdh2, another classic cadherin typically upregulated during EMT processes (Figure 1g). We found that Ecad expression was significantly downregulated following Slug overexpression, but also Pcad overexpression. Pcad overexpression triggered a significant Ncad downregulation as estimated by a Student's test, considering the variance were similar here 4 and in other experiments. As Slug is known to target specific E2-boxes (CAGGTG)9,12 we screened Pcad mouse promoter and found three such E-boxes in the proximal promoter: − 110, − 166 and − 1480 from ATG site (Genbank seq D12688). Considering that proximal paired E2-boxes are essential in Ecad promoter regulation by Slug,12,18 we focused on the two proximal Pcad promoter E2-boxes. We carried out chromatin immunoprecipitation assays, adding a Ecad promoter sequence as positive control.9,12 As antibodies against Slug failed to immunoprecipitate endogenous proteins, we used a fusion protein Slug-GFP validated earlier1 and anti-GFP antibodies. SlugGFP overexpression resulted in a twofold protein overexpression

Institut de Recherche en Cancérologie de Montpellier (IRCM), Inserm U1194, Université de Montpellier, Institut Régional du Cancer de Montpellier (ICM), Montpellier, France. Correspondence: Dr P Savagner, Institut de Recherche en Cancérologie de Montpellier, Université de Montpellier, Institut Régional du Cancer de Montpellier (ICM), Montpellier Cedex 5, 34298 France. Email: [email protected] 1 These authors contributed equally to this work. Received 15 February 2016; revised 16 August 2017; accepted 23 August 2017

Slug/Pcad in mammary epithelial cells Y Idoux-Gillet et al

2

as compared to untransfected cells1 (Supplementary Figure 1). Chromatin immunoprecipitation analysis corroborated the bond between Slug-GFP and Ecad promoter but also to Pcad promoter region including E-boxes (−298 to − 100 from ATG). No significant amplification was obtained with a neighboring promoter region without E-boxes (−634 to − 275 from ATG) (Figure 2a). We also carried out luciferase reporter gene experiments. We used one negative control (pxp1), one short Pcad promoter (−200, +47: pxp-200), and one larger Pcad promoter region (−1060, +70: pxp-1060, Figure 2b). Both constructs include proximal E-boxes. We co-transfected CommaDβ cells with these constructions and control vector (Ctr) or Slug expression vector (Slug). Slug overexpression resulted in a significant increase of luciferase activity (Figure 2b). Longer reporter construct provided stronger activity, suggesting cooperating sites. To determine E-boxes role, we produced three inactivating mutant constructs, where residues 1 (C) and 6 (G) from the CAGGTG E-boxes were switched for an A residue, as described previously.19 Surprisingly, inactivating one E-box increased luciferase activity. But inactivating both E-boxes strongly repressed the luciferase activity (Figure 2c). Finally, we identified in the upstream sequence two sites: CACATGT (923 bp from ATG) and CTACCT (758 bp from ATG) described as alternative binding sites for Snail and Twist in drosophila20 and potentially involved in the pxp-1060 activity.

Figure 1. Slug controls Pcad in mouse mammary epithelial cells. (a) Immunofluorescence analysis of the mouse mammary epithelium from wild-type (WT) and knockout (KO) Slug mice revealed loss of Pcad in basal epithelial cells, expressing SMA and CK5 (arrowheads). Scale bar = 10 μm. (b) RtPCR analysis of Pcad gene expression in WT and Slug-KO (KO) mammary gland and in CommaDβ cells treated with siRNA control (Ctr) and two distinct anti-Slug siRNA (siSlug1 and siSlug2). Columns indicate statistical means and error bars point to s.e.m. Star (*): Two-sided Student's test, Po0.01. (c) Western blot analysis of Slug and Pcad protein levels in CommaDβ cells treated with siRNA controls (Ctr1 and Ctr2) and two Slug siRNA (siSlug1 and siSlug2). (d) Slug and Pcad protein quantification at 24, 48 and 72 h after siSlug transfection in CommaDβ cells. (e) Slug and Pcad protein quantification 48 h after siSlug or siPcad transfection in HaCaT keratinocytes. (f) RtPCR analysis of Pcad gene expression in CommaDβ cells transfected with control vector (Ctr) or Slug expression vector (Slug). (g) Impact of Slug and Pcad on Ncad and Ecad expression level. CommaDβ cells were transfected with Slug (black columns) or with Pcad (white columns). RtPCR analysis was reported related to Control plasmid transfection. Star (*): two-sided Student's test, Po0.05.

Oncogene (2017) 1 – 11

Pcad mediates Slug effects on mammosphere growth, differentiation and migration As Slug upregulates Pcad expression, we hypothesized that Pcad could mediate Slug functional impact. We and others previously demonstrated that Slug controls stem cell growth.1,21 Therefore, we evaluated Pcad influence on mammosphere growth, a recognized test to evaluate stemness. We treated CommaDβ cells with siRNA against Pcad in clonal mammosphere assay as described previously.1 After 3 days, we found no effect in early survival phase (Figures 3a and b). However, we observed a significant decrease in the number of clonal mammospheres growing after 3 weeks from siPcad cells (Figure 3a), suggesting a role in stem/progenitor cell proliferation. We also investigated a putative role for Pcad in cell differentiation. We overexpressed or downregulated Pcad and Slug in CommaDβ cells and followed luminal and basal cell fate commitment monitoring specific cytokeratins: CK5 for basal cells and cytokeratin 8 (CK8) for luminal cells (Figures 3c and d). Slug and Pcad induced a significant increase in the percentage of CK5+ cells, accompanied by a significant decrease of CK8+ cells, expressed by 25% of control cells. SiRNA had an opposite effect increasing the CK8/CK5 ratio, suggesting that Pcad impacts early differentiation phases, as underscored by the similar overbranching phenotype observed in Slug-KO and Pcad-KO mammary glands.1,22 We also found that Pcad overexpression was able to partially restore CK8/CK5 ratio when overexpressed in siSlug treated cells, suggesting a rescue mechanism. Slug is known to favor cell migration in mammary epithelial cells and keratinocytes.3,8,9,11,12,23 Therefore, we looked for Pcad role in mediating Slug-induced cell motility. We performed wound-healing experiments monitoring wound coverage by migrating cell sheets (Figure 4). As found earlier,24 Slug overexpression increased motility, mostly in a cohort migration mode (Figure 4). Pcad overexpression induced cell individualization and motility, but migration index did not take into account individual cells, resulting in a non-significant index rise. SiRNA targeting Slug and Pcad both inhibited migration but Pcad overexpression was able to rescue siSlug-induced inhibition (Figure 4a). Reciprocally, siPcad repressed Slug-induced migration increase. To better analyze cell response, we specifically counted individualized migrating cells in the wound area. We found that Pcad overexpression enhanced this value, and that siRNA-targeting © 2017 Macmillan Publishers Limited, part of Springer Nature.

Slug/Pcad in mammary epithelial cells Y Idoux-Gillet et al

3

Figure 2. Slug binds and activate Pcad promoter. (a) ChIP analysis reveals Slug binding to the Ecad and Pcad promoter in CommaDβ cells. Cells were transfected with a Slug-GFP construct, allowing immunoprecipitation with an anti-GFP antibody. QPCR was performed with primers specific to the Ecad and Pcad promoter, including (black columns) or not including (white column) E-box region for Pcad promoter. (b) Reporter gene analysis. Three luciferase reporter gene constructions are depicted, one without Pcad promoter upstream (pxp1), one with a short Pcad promoter (−200, +47) containing two E-boxes (pxp-200), and one with a large Pcad promoter (−1060, +70) (pxp-1060). Quantitative analysis of luciferase expression in CommaDβ cells co-transfected with the reporter constructs, and with control vector (Ctr: white column) or Slug expression vector (Slug: black column). (c) Mutated E-boxes were introduced in the pxp-1060 Pcad promoter construct including Ebox1 mutation (E1), Ebox2 mutation (E2) and combined mutations E1/E2. Columns indicate statistical means and error bars point to s.e.m. Star (*): Two-sided Student's test, P o0.05.

Slug and Pcad decreased it (Figure 4c). Here also, Pcad could rescue Slug deficit (Figure 4c). Finally, we recorded isolated cells using time-lapse video for 20 h. Pcad overexpression induced individual motility when siRNA treatment reduced this motility. Quantification indicated that Pcad-overexpressing cells migrated about three times further than siPcad-treated cells (Figures 4b and d and movies M1 and M2). Globally, these results support the hypothesis that Pcad is mediating Slug-induced migration. P120cas and Src participate to Slug/Pcad-induced cell migration Similarly to other cadherins, Pcad clusters a sub-membrane complex including catenins. Among them, P120cas/CTNND1 is © 2017 Macmillan Publishers Limited, part of Springer Nature.

Figure 3. Pcad reproduces most of Slug impact on mammosphere growth and cell differentiation and rescues Slug deprivation. (a) Schematic representation of mammosphere generation. After 3 days suspension culture, some isolated cells gave rise to microspheres. Three weeks later, some microspheres evolved to mammospheres. Bar = 25 μm. (b) Sphere-forming efficiency for CommaDβ cells, transfected with siRNA control (siCt) and siRNA against Pcad (SiPcad). (c) Percentage of CommaDβ expressing a basal (CK5+) or luminal (CK8+) phenotype, after treatment with control vector (Ctr), Slug expression vector (Slug) or Pcad expression vector (Pcad). (d) Ratio CK8/CK5 in CommaDβ cells after treatment, reported to untreated (Ctr) cells. Columns indicate statistical means and error bars point to s.e.m. Star (*): Two-sided Student's test, P = 0.05.

described to mediate Pcad activity and to be important for mammary tubulogenesis.25–27 We immunolocalized P120cas in CommaDβ cells after modulation of Pcad expression. Overexpression of Pcad was associated with cell–cell dissociation and disappearance of membrane-bound P120cas and a cytoplasmic translocation (Figure 5a). Pcad overexpression induced a drop in overall P120cas protein amount as compared with Slug overexpression (Figure 5b). Both siRNA increased the level of P120cas, in link with a cohesive phenotype implying cell junctions. However, Pcad overexpression rescued siSlug treatment in decreasing P120cas protein level, in link with a dissociated phenotype (Figure 4). As P120cas is a major Src substrate,28 we investigated Src implication. In accordance with previous work,27,29–31 we found by western analysis that Pcad gene repression induced a significant drop in Src protein expression (Figure 5c and d, P o 0.05). We also checked for other members of Src family: Lyn and Fyn were not significantly regulated (Figure 5c). Finally, we verified Src requirement during migration. Slug/Pcad dependent migration was inhibited by 5 μM Src inhibitor PP2, without affecting cell Oncogene (2017) 1 – 11

Slug/Pcad in mammary epithelial cells Y Idoux-Gillet et al

4 viability (Figures 5e and f). Src inhibitor also strongly inhibited HaCaT keratinocyte Slug-mediated motility (Figure 5g and Supplementary Figure 2).

Slug/Pcad pathway modulates focal adhesions in cohesive and isolated cells Src activation is known to increase CDC42 and Rac1 activity.30,31 This activation is reflected by cell-matrix adhesion reorganization and cell migratory status. We monitored two focal adhesions components: Paxillin and Vinculin, during slug/Pcad-mediated wound healing.32–34 After 48 h Slug/Pcad overexpression or suppression, we located Paxillin and Vinculin by immunofluorescence, paired with phalloidin labeling to monitor focal adhesions. Those were located within large lamellipodia at the front of migrating cells, emphasizing a recognizable migrating polarized phenotype. Overexpression of Slug and Pcad did not alter this pattern (Figure 6a). But inhibition of Slug or Pcad significantly reduced the number of margin cells expressing focal adhesions and increased the number of cells exhibiting little or no prominent focal adhesions. Both Paxillin and Vinculin studies provided similar results. We then focused on fully isolated cells following the same transfection scheme. As previously, we sorted and counted polarized (migratory) and unpolarized (non-migratory) cells based on their phenotype and focal adhesion pattern (Supplementary Figure 3). Overexpression of Slug and Pcad did not alter this pattern significantly compared to the control vectors (Figure 6b). Conversely, siRNA-targeting Slug and Pcad decreased significantly the number of cells expressing a motile phenotype. Remarkably, siEcad treatment did not regulate the expression of the motile phenotype. In contrast, Ecad overexpression decreased the percentage of polarized, motile cells (Figure 6b). Pcad controls early tubulogenesis in vitro in 3D assay We developed an in vitro 3D assay designed to recapitulate tubulogenesis steps. We used microsupport beads to obtain an organized and polarized mammary epithelium fully covering suspended beads. Cell-loaded beads were embedded in a mixed collagen/matrigel 3D matrix and we could follow emerging cells, aligned to initiate a tubulogenesis-like process in a 3D environment (Figure 7a). More than 40% microsupport beads grew tubule-like structures. Among those, around 13% (from original loaded beads) harbored more than three tubule-like structures. Slug/Pcad involvement was examined in this process as previously. Slug overexpression drove more tubule-like processes. Conversely, suppression of both Slug or Pcad strongly and significantly decreased the number of emerging tubules. However, siSlug-induced inhibition was rescued by Pcad overexpression (Figure 7b). Figure 4. Pcad mediates Slug impact on cell migration. (a) Representative experiment from a wound-healing assay in CommaDβ cells, transfected with control siRNA (SiCt) and control vector (Ctr), Slug or Pcad expression (Slug or Pcad), siRNA against Slug or Pcad (SiSlug or SiPcad), and combined expression vector and siRNA. Wound areas were photographed 48 h post transfection, and solid line represents the original wound borders. Bar = 50 μm. Insert: Migration front at higher magnification. Arrowheads show cells with mesenchymal (Ctr, SiSlug+Pcad) or epithelial (siSlug) phenotypes. Bar = 10 μm. (b) Quantification of the wound healing assay for CommaDβ cells. (c) Quantification of cell individualization after transfections. (d) Pcad induces cell motility in individual cells. CommaDβ cells were seeded in low number, transfected with control vector (Ctr), Pcad expression vector (Pcad) and SiRNA (SiPcad) before monitoring by video recording (Online Video m1 and m2). Arrowhead: cell monitored. (e) Quantification of migration extent by transfected individualized CommaDβ cells. Columns indicate statistical means and error bars point to s.e.m. Star (*): Two-sided Student test comparing column/Ctr or bracketed column indicates a significant difference, P = 0.05.

Oncogene (2017) 1 – 11

© 2017 Macmillan Publishers Limited, part of Springer Nature.

Slug/Pcad in mammary epithelial cells Y Idoux-Gillet et al

5

Figure 6. Motile phenotype and individualization are induced by Pcad during wound-healing cohesive migration. (a) At the leading edge during in vitro ‘wound-healing’ experiments, cohesive cells expressing a polarized front extension expressing paxillin-rich focal adhesions (leading cells) were typed and counted. Value was reported to cells showing no such extension (non-leading cell) in Slug and Pcad gain or loss experiments. Bar = 5 μm. (b) Motility phenotype in isolated cells is also controlled by Slug/Pcad pathway. Motile phenotype with an elongated polarized cell expressing paxillin in the front extension was quantified versus unpolarized spreading cell, categorized as a non-motile phenotype as described in Material and methods section. Columns indicate statistical means and error bars point to s.e.m. Star (*) signals a significant difference (Two-sided Student' test, Po 0.05).

Figure 5. Slug/Pcad pathway promotes cell migration via P120cas internalization and Src activation. (a) P120cas was immunolocalized in CommaDβ cells treated with siRNA-targeting Pcad (siPcad) or with Pcad plasmid. Bar = 10 μm. (b) P120cas protein expression was detected by western analysis and quantified in cells treated as indicated. Statistically significant difference evaluated by a two-sided Student's test (P ⩽ 0.05) was indicated by a star. (c, d) Src kinase was detected and quantified by western analysis in cells treated with siRNA-targeting Pcad (SiPcad) or with Pcad, showing significant difference using two-sided Student's test. Star (*): Po 0.05. In a distinct set of experiments, Fyn and Lyn were also detected by western blotting. Based on apparent molecular weight, specific band for Lyn is indicated by the arrow. No significant modulation was detected. (e, f) Inhibition of Src kinase impedes Slug/ Pcad-mediated migration. Migration extent was estimated as previously described in untreated (Ctr), mock (DMSO) or PP2-treated CommaDβ (d) or HaCaT (e) at 48 h (d, e) and 72 h (d). cells. (f) Bar = 50 μm. (g) Src inhibitor treatment inhibits CommaDβ migration. Representative experiment from a wound-healing assay in CommaDβ cells, untreated (Ct), treated with DMSO or with Src inhibitor PP2 for 48 h. Wound areas were photographed 24 h later. Columns indicate statistical means and error bars point to s.e.m.

© 2017 Macmillan Publishers Limited, part of Springer Nature.

Slug/Pcad pathway components are co-expressed and correlate with differentiation markers in breast carcinoma samples To explore the relevance of Slug/Pcad pathway in human cancer samples, we analyzed Slug, Snail, Ecad, Pcad, Src, Lyn expression pattern in the METABRIC breast cancer dataset (2509 samples). We observed a similar profile for Slug, Pcad, CK5, Lyn and CEBP, more abundant in Basal and HER2 carcinoma subtypes (Figure 8). We observed a highly significant correlation pattern between Pcad and Slug in Her2 and Lum A subgroups. Pcad expression was also correlated with members of the Src family in Basal, Lum B and HER2 subtypes and with CKs and CEBPB in most subgroups (Figure 8 and Supplementary Table 1). A negative correlation was observed with ESR1 in all cases. Slug/Pcad pathway activation is linked to clinical evolution in distinct breast carcinoma subtypes We examined the clinical impact of snail family/Pcad pathway. High Pcad expression has been linked to poorer outcome earlier.14,35–40 Based on the recurrence-free survival (RFS) rate, we confirmed these observations and observed a subtype specificity (Figure 9a): high Pcad expression was marginally associated with a poorer outcome in Lum A (P = 0.035; HR = 1.62) and strongly in Lum B (P = 0.0067; HR = 1.67). Similarly, Slug higher Oncogene (2017) 1 – 11

Slug/Pcad in mammary epithelial cells Y Idoux-Gillet et al

6

Figure 7. Pcad and Slug are necessary for in vitro 3D tubulogenesis and Pcad can rescue Slug deprivation. (a) Cells were grown in suspension at the surface of microcarrier cytodex beads and transfected with siRNA as indicated. Cell-loaded beads were immerged in a 3D collagen/matrigel mix. Bar = 20 μm. (b) Growing tubule-like structures were quantified after 48 h. Columns (mean+ s.e.m) represent the percentage of beads harboring no extending processes (left panel) or at least three tubule-like processes (right panel). Cells were treated with expression vectors or siRNA before 3D growth. Columns indicate statistical means and error bars point to s.e.m. Star (*) signals a significant difference (Two-sided Student's test, P o0.05).

expression level was linked to a poorer survival index in Lum B (P = 0.014; HR = 1.47). We analyzed variable independence for RFS by Cox regression analysis and by plotting multigroup Kaplan–Meier graphs. Slug remained an independent parameter in predicting RFS even when taking into account the most significant clinical parameters, that is, lymph node involvement (P = 1.8e-10) and tumor size (P = 2.5e-03). In combined Kaplan– Meier graph, the Pcad high/Slug high group expressed a worst survival rate in Lum A and B groups, characterized by a stronger significance (Lum A: P = 0.015; HR = 2.7 (1.18-6.25); Lum B, P = 0.003; HR = 2.17 (1.28-3.71) compared with the Pcad low/Slug low group than for each variable alone (Figure 9a). Neither Pcad nor Slug was significantly associated with RFS (Figure 9a) in the Basal subtype, but combining both parameters provided a clear trend (P = 0.13). DISCUSSION Control of cell–cell adhesion and cell motility pathways is a key to morphogenesis and metastasis. In this report, we describe a new role for Pcad, a classic cadherin, in mediating several functions of EMT master gene Slug (Figure 9b), including motility and differentiation. Most EMT studies link motility induction to Ecad downregulation by EMT-TF. However, Ecad suppression alone does not induce motility as shown by our results and others41,42 (Figure 6). In drosophila for example, tracheal tubulogenesis involves cell plasticity, controlled by a cadherin (DE-cadherin), upregulated by a snail gene (escargot).43 Here, we found in mammary epithelial cells that Pcad is upregulated by Slug and mediates Slug-induced cell motility. The impact of Pcad and Ecad Oncogene (2017) 1 – 11

co-expression pattern has been discussed in mammary epithelial cells.44–46 In accordance, we suggest that they can compete in mammary basal epithelial cells expressing both cadherins. This balance could control a partial (cohort migration) to total EMT, reflecting a metastable phenotype47 depending on local microenvironment, as described during carcinoma progression. Distinct junctional complex configurations could explain this plasticity, resulting in distinctive adhesion forces, as described for Ecad.42 Conversely, Pcad appears to mediate predominantly cell–cell adhesion in cells expressing little Ecad, evoking compensation mechanisms.46 In some cases, Pcad downregulation was reported to enhance stroma-driven motility.48 We also observed a downregulation of Ncad gene expression following Pcad overexpression. Ncad is linked to cell motility and morphogenesis in vivo.49,50 It is generally considered as an EMT marker.51 However, no E2-box was found in the promoter region (1260 bp, NCBI: NC_000018.10). Our description of Slug/Pcad pathway provides a comprehensive mechanism for EMT-driven cell motility. In addition, we found that several of Slug roles that we defined earlier1 in supporting mammosphere growth, myoepithelial differentiation, motility and tubulogenesis were sustained by Pcad, typically rescuing an induced Slug deficit. We went one step further by describing the induction of individual cell motility by Pcad. This novel result was unsuspected for a cell junction protein44,45 but probably reflects permissive in vitro microenvironment and substrate. The downstream pathway linking Pcad to P120cas, Src, Rac and cell motility has been well documented.27,29,30,38 We confirmed in this report that P120cas and Src pathways are involved in mediating Slug-mediated cell migration: P120cas was destabilized from adherens junctions and downregulated following Pcad overexpression and cell individualization. Interestingly, Slug had a different impact, apparently linked to the active maintenance of a rather cohesive phenotype expressed during a cohort migration, in accordance to previous work from us and others.1,8,21,24 Indeed, we described originally that Slug overexpression induced only a partial cell–cell dissociation, with suppression of desmosomes, but persistence of Ecad expression.3 Cohesion and high P120cas expression level was also found in siSlug and siPcad-treated cells that remained stationary. These observations suggest that Pcad induced cell–cell dissociation is the trigger to P120cas downregulation. Further down, we found that Src inhibition suppressed a Slug/Pcad-mediated migration in two cell types. Src family is involved in multiple other situations, but here, it is specifically triggered by the Slug/Pcad activation, mediating cytoskeleton reorganization and motility. We identified two proximal E2-boxes, binding sites for Snail family in the promoter of Pcad. We established by point mutations that these sites mediated Slug transcriptional activation, without excluding distinct indirect mechanisms. Inactivating only one E-box increased the promoter activity, suggesting that each box alone could recruit nuclear factors with up- or downregulating activity and that higher-level molecular complex was required for global activation. Co-factors recruitment has been described for members of the Snail family before, sometimes resulting in transcriptional activation, similarly to our finding.20,52,53 We also located non-classic Snail binding putative sites recently described in drosophila in Pcad promoter. More work will be necessary to assess their putative role in Pcad regulation. In addition to Slug, specific binding sites for β-catenin, CEBP, P63, HoxA9 and Foxc154–61 are integrated into Pcad promoter region. We found CEBP to be overexpressed in Basal breast carcinoma and to strongly correlate with Pcad in all cancer subtypes in accordance with previous work,60,61 supporting an involvement in Pcad regulation. CEBP is also described to downregulate ESR1,61 which we found to be strongly and negatively correlated to Pcad and Slug expression (Figure 8 and Supplementary Table 1). © 2017 Macmillan Publishers Limited, part of Springer Nature.

Slug/Pcad in mammary epithelial cells Y Idoux-Gillet et al

7

Figure 8. Distinctive expression of Slug/Pcad pathway genes in breast carcinoma subtypes. (a) Expression pattern box-plots compare Slug/ Pcad pathway gene expression levels among breast carcinoma subtypes. Bsl: Basal, H2: HER2, LA:Luminal A, LB:Luminal B, Nl:Normal-like. (b) Correlation between Pcad and other pathway components expression level is analyzed for distinct carcinoma subtypes. Pearson’s correlation indexes between Pcad and other components expression were calculated for each gene. Strong positive correlation (P o0.005) is displayed in dark green. Significant correlation (P o0.05) is displayed with a pale green. Negative correlation (P o0.01) is indicated with an pink (Po0.05) or red (P o0.005) link. Other genes are reported in Supplementary Table 1.

Pcad overexpression has been associated to differential clinical outcome in breast tumors.14,26,35–40 More specifically, we found that it correlates with negative outcome in Lum A and Lum B breast cancers, similarly to Slug. It also correlates with CK5 and CEBPB in luminal subtypes, suggesting that the expression of these two genes, commonly associated with Basal tumors in more aggressive luminal tumors. Both Pcad and Slug behave as independent variables for RFS, but co-expression added a prognostic value. Importantly, this was also found when including other clinical parameters such as lymph node invasion and tumor size. In conclusion, we characterized a new Slug/Pcad pathway, depicting Slug for the first time as a transcriptional activator. This pathway appears to control several EMT associated-properties, such as motility, stemness properties, differentiation, and to have a role during mammary gland morphogenesis and breast cancer aggressiveness. Clinical relevance in breast cancer is suggested by the striking and contrasted correlation patterns we observed, linked to tumor subtype molecular characteristics and clinical outcome. MATERIALS AND METHODS Cell culture CommaDβ62,63 were graciously provided by Dr Medina (Baylor college of Medicine, Houston, TX, USA). They grew in DME F12 (Gibco, Carlsbad, CA, USA) with 2% fetal calf serum (Gibco), 10 μg/ml bovine insulin (Sigma, St © 2017 Macmillan Publishers Limited, part of Springer Nature.

Louis, MO, USA) and 5 ng/ml murine epidermal growth fator (Sigma). For microsupport culture, cells were trypsinized and resuspended, then seeded at 106 cells/well in 24-well culture dish onto a monolayer of Cytodex Microsupports (type 3, Sigma). Cells were maintained in agitation for 48 h. Immortalized human keratinocyte cell line HaCaT64 was a kind gift from Professor Fusenig (University of Heidelberg, Germany) and were maintained in DMEM+10% fetal calf serum, glutamine, and antibiotics. All cells were grown at 37 °C in a humidified 5% CO2 incubator. All cells were regularly tested for mycoplasma contamination.

Transfections Plasmids. CommaDβ cells were transiently transfected using INTERFERin (Polyplus, Illkirch, France). After 48 h, cells were harvested for RNA or protein extraction, or fixed for immunofluorescence. Control vector pEGFPN2 was used to assess transfection efficiency. MPCR3.Slug vector3 was used for mouse Slug overexpression. Plasmids pCDNA3.H.Slug and Mβcat Pcad were kindly provided by T Ip (University Massachusetts Medical School, MA, USA) and Professor M Takeichi (Riken Center for Developmental Biology, Japan). For siRNA transfections, INTERFERin (Polyplus) was added to 10 nM duplex siRNA per well (Eurogentec, Angers, France and Qiagen, Valencia, CA, USA). Two siRNA oligoribonucleotides were used for Slug and for Pcad (Supplementary Table 2). Four were combined for Ecad. Control uncoding siRNA oligoribonucleotides were randomly synthesized and sometimes mixed. After 48 h incubation, cells were harvested for RNA and protein extraction, or fixed for immunofluorescence. Slug, Slug-GFP and Pcad Oncogene (2017) 1 – 11

Slug/Pcad in mammary epithelial cells Y Idoux-Gillet et al

8

Figure 9. (a) Slug/Pcad pathway components have distinct clinical impact among carcinoma subtypes. Kaplan–Meyer survival curves for Pcad and Slug examining breast cancer molecular subtypes Lum A, Lum B and Basal. Combined curves (right panel) analyze both genes. (b) General hypothesis: the Slug/ Pcad /Src pathway controls cell motility. transfection efficiency was demonstrated by RtPCR or western analysis (Supplementary Figure 1). For microsupport culture, cells were transfected before embedding with Lipofectamine 2000 Reagent (Invitrogen, Carlsbad, CA, USA) directly onto Cytodex according to manufacturer’s instructions.

Mammary epithelial differentiation monitoring Cytokeratins were used to differentiate basal/myoepithelial cells (CK5) and luminal cells (CK8). Cells were evaluated by immunofluorescence 48 h after transfection. About 500 cells were scrutinized for each condition, using ImageJ (http://developer.imagej.net). More than 95% cells expressed only one cytokeratin type. Experiments were repeated three times.

Immunofluorescence Samples fixed for 2 h (paraformaldehyde 4%) at RT (room temperature), were included in paraffin. A total of 5 μm sections were deparaffinized then processed for antigen retrieval (citrate buffer, pH 6.0) for 30 min at 96 °C. Cells grown for 48 h on glass coverslips were fixated (methanol for CK immunolabeling, 4% paraformaldehyde+0.05% triton otherwise). After phosphate-buffered saline washing, cells/tissue sections were incubated for 1 h (RT) with primary antibodies (Supplementary Table 2) and 10% goat serum, then secondary antibodies for 45 min. Images were acquired with a CoolSNAP HQ camera (Roper, Tucson, AZ, USA) using Volocity software (Perkin Elmer, Waltham, MA, USA). Oncogene (2017) 1 – 11

RNA extractions and reverse transcription RNA was extracted from cell lines using RNeasy Mini Kit (Qiagen). Quality was checked by spectrometer analysis and gel migration. One μg total RNA was reverse transcribed using hexanucleotides and Superscript II Reverse Transcriptase (Invitrogen).

Real time-quantitative PCR RtPCR was carried out using an ABIPRISM 7300 (Applied Biosystems, Foster City, CA, USA). Amplifications used SyberGreen master mix (Applied Biosystems) in 25 μl standard PCR conditions (40 cycles with an annealing/ © 2017 Macmillan Publishers Limited, part of Springer Nature.

Slug/Pcad in mammary epithelial cells Y Idoux-Gillet et al

9 elongation step at 60 °C). Full experiment was repeated three times. Primary data were displayed as the difference in the number of cycles between the studied gene and 18 S, during the linear amplification phase.

Chromatin immunoprecipitation analysis Chromatin immunoprecipitation assays were performed as described.65 CommaDβ cells transfected with Slug-GFP construct were fixed for 10 min (1% formaldehyde, phosphate-buffered saline). Cells were centrifuged for 5 min at 2000 g/4 °C, resuspended for 30 min in lysis buffer (50 mM 4-(2Hydroxyethyl)piperazine-1-ethanesulfonic acid (pH 7.5), 140 mM NaCl, 1% Triton X-100, protease inhibitor), then sonicated and centrifugated for 10 min. Supernatants were cleared by incubation with 2.5 mg protein G agarose (ROCHE, Indianapolis, IN, USA), sonicated salmon sperm DNA (2 μg), 1 mg/ml BSA for 2 h at 4 °C before analyzing. Immunoprecipitation was carried out overnight at 4 °C using purified IgGs (negative control), GFP (4 μg, Abcam, Cambridge, UK), or acetylated histone H4 (positive control) antibodies. After centrifugation, washing, and elution, cross-linking was reversed by heat treatment (65 °C overnight). DNA was then purified (Qiagen kit), and PCR performed with appropriate primers. Full experiment was repeated twice.

Luciferase assay CommaDβ cells were co-transfected with reporter gene constructs and control (Ctr) or Slug expression vector (Slug). Luciferase analyses were performed with Renilla Luciferase (Promega, Madison, WI, USA). A total of 100 μl per well of lysis buffer were added on cells and shaken for 15 min (RT). Then 50 μl of cell lysate were mixed with 50 μl of Luciferase reagent. Measurements were performed with a Luminometer Centro LB960. Experiments were repeated four times, each in duplicate.

Mammosphere culture After trypsinization, CommaDβ cells were seeded as single cells in ultralow attachment plates (Corning, Corning, NY, USA) at a density of 2 cells/well as described.1 All wells were checked after three days then 3 weeks of suspension culture to report small aggregates (one to four cohesive cells) called microspheres, and larger aggregates (diameter450 μm) after 3 weeks. Monitoring was done blindly by examiners unaware of the treatment. Experiments were performed three times for each condition. Values are reported to the number of seeded cells.

Western blot analysis Cells were grown to confluence in six-well plates, transfected with control siRNA (SiCt) and vector (Ct), Slug or Pcad expression vector (Slug or Pcad), siRNA against Slug or Pcad (SiSlug or SiPcad), and combined expression vector and siRNA (Slug+SiPcad or Pcad+SiSlug). Cell dishes were used alternatively for Western blot analysis or for wound-healing experiments. For Western blot, cells were lysed in a Tris 10 mM buffer (SDS 1%, 5 mM ethylenediaminetetraacetic acid, 10 mM sodium pyrophosphate 10 mM βglycerophosphate, protease inhibitors Complete-Roche). Protein concentration was determined with BCA Protein assay (Pierce, Thermofisher Scientific, Carlsbad, CA, USA). 40 μg were resolved by 12% SDS-PAGE gel, then transferred to a nitrocellulose membrane (Amersham, Little Chalfont, UK). Membranes incubated with 5% BSA for 1 h (RT), then primary antibodies at 4 °C overnight. Immune complexes were detected by chemiluminescence with horseradish peroxidase-conjugated secondary antibodies. Each western blot analysis was repeated at least three times.

In vitro wound-healing assay Cells were prepared as for western blot analysis. Wound areas were created using a P1000 plastic pipette. Reepithelialization area was quantified as the difference between uncovered wound area at t = 0 and t = 48 h as estimated using ImageJ. Measurements were performed blindly, from three independent experiments, examiners being unaware of the treatment. A total of 5 μM PP2 were used for Src inhibition studies.

Motile phenotype monitoring During wound-healing experiments, motile phenotype was defined by the presence of elongated focal adhesions expressing vinculin and paxillin, located at the front of border cells on the leading edge. In isolated cells, it was defined by a polarized bipolar phenotype involving focal adhesions at © 2017 Macmillan Publishers Limited, part of Springer Nature.

extremities. Non-motile phenotype was defined by the lack of visible polarized focal adhesions. Cells showing unequivocally one of these two phenotypes were counted along the wound-healing borders for each condition, and sorted into migratory and non-migratory cells. We integrated ~ 20–40% of margin cells into the two categories. Experiments were performed three times blindly. Examiners were unaware of the treatment.

Cell individualization monitoring Wound-healing experiments were analyzed after 48 h migration. Three to seven wound area microscopy fields from two distinct experiments were screened to determine number of isolated cells by phase microscopy (Leica N Plan PH1 × 10/0.25 objective), ranging from 30 to 100 per field. Values were reported to control cell population.

Videomicroscopy Videomicroscopy was performed using a Leica DMIRB inverted microscope with a Leica N Plan PH1 × 10/0.25 objective. Cells were maintained at 37 °C. Frames were taken every 10 min, starting 6 h after transfection and lasting for 20 h. Images were analyzed using Volocity software (Perkin Elmer). Quantification of cell migration was performed using ImageJ software (NIH). Experiments were performed twice, monitoring 24 cells for each experiment.

3D tubulogenesis model Cells were trypsinized, resuspended and seeded at 1 million cells/well in 24-well culture dish onto a monolayer of Cytodex Microsupports (type 3, SIGMA) for 48 h. Medium was changed every day and Cytodex beads were maintained in suspension. Cells were transfected with Lipofectamine 2000 Reagent (Invitrogen) onto Cytodex. The composite collagen (Corning)–BME (Cultrex, AMSBIO) matrix was prepared by diluting collagen to 2 mg/ml in phosphate-buffered saline × 1 (Gibco). After pH balancing, Cultrex was added (10% V/V) and the mixture was vortexed and briefly centrifuged. Cells on Cytodex transfected with SiSlug, SiPcad and Si-Ctrl were added to the mixture and loaded in a Lab-Tek Slide system (eight wells, Sigma). After complete polymerization (15 min, 37 °C), medium was added on the gel and culture proceeded for 48 h. Experiments were performed in duplicates at least three times. Counting was performed blindly by at least two examiners unaware of the treatment, based on the number of tubule-like structures (minimal 100 μm) budding from the Cytodex. Results were normalized to the total number of Cytodex monitored, reported to the values obtained with Si-Ctrl.

Patients and tumors: survival analysis We downloaded the 2509 samples Breast cancer dataset from METABRIC (Molecular Taxonomy of Breast Cancer International Consortium).66,67 Survival analysis (Kaplan–Meier) and Cox regression for CDH3 and Slug were processed using R program (http://www.R-project.org/) and the Györffy script68 to calculate optimal thresholds.

ABBREVIATIONS ChIP, Chromatin immunoprecipitation; CK5, Cytokeratin 5; CK8, Cytokeratin 8; EMT, Epithelial–mesenchymal transition; Ecad, E-cadherin; ESR1, Estrogen receptor 1; H2, HER2; KO, Knocked-out; Lum A, Luminal A; Lum B, Luminal B; Ncad, N-cadherin; Nl, Normal-like; Pcad, P-cadherin; RFS, Recurrence-free survival; SMA, Smooth muscle actin.

CONFLICT OF INTEREST The authors declare no conflict of interest.

ACKNOWLEDGEMENTS Financial support was provided by the Fondation de France (no. E 2009 006685), the Ligue Nationale contre le Cancer, the Ligue Régionale contre le Cancer (LanguedocRoussillon and Ardèche) and Aide à la Recherche en Partenariat avec Entreprises (ARPE-Languedoc-Roussillon). In addition, the authors gratefully acknowledge support from the Ligue Régionale (Ardèche) and Nationale contre le Cancer for Y Idoux-Gillet and Association pour la Recherche sur le Cancer for M Nassour. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript. We are grateful to L Larue for plasmid vector Ecad,

Oncogene (2017) 1 – 11

Slug/Pcad in mammary epithelial cells Y Idoux-Gillet et al

10 Professor M Takeichi for Pcad vector, Marisa M Faraldo and Marie-Ange Deugnier for Luciferase gene reporter and constructive discussions.

REFERENCES 1 Nassour M, Idoux-Gillet Y, Selmi A, Côme C, Faraldo M-LM, Deugnier M-A et al. Slug controls stem/progenitor cell growth dynamics during mammary gland morphogenesis. PLoS One 2012; 7: e53498–12. 2 Guo W, Keckesova Z, Donaher JL, Shibue T, Tischler V, Reinhardt F et al. Slug and Sox9 cooperatively determine the mammary stem cell state. Cell 2012; 148: 1015–1028. 3 Savagner P, Yamada KM, Thiery JP. The zinc-finger protein slug causes desmosome dissociation, an initial and necessary step for growth factor-induced epithelial-mesenchymal transition. J Cell Biol 1997; 137: 1403–1419. 4 del Barrio MG, Nieto MA. Overexpression of Snail family members highlights their ability to promote chick neural crest formation. Development 2002; 129: 1583–1593. 5 Arnoux V, Nassour M, L'Helgoualc'h A, Hipskind RA, Savagner P. Erk5 controls Slug expression and keratinocyte activation during wound healing. Mol Biol Cell 2008; 19: 4738–4749. 6 Onodera T, Sakai T, Hsu JC-F, Matsumoto K, Chiorini JA, Yamada KM. Btbd7 regulates epithelial cell dynamics and branching morphogenesis. Science 2010; 329: 562–565. 7 Shields MA, Krantz SB, Bentrem DJ, Dangi-Garimella S, Munshi HG. Interplay between β1-integrin and Rho signaling regulates differential scattering and motility of pancreatic cancer cells by snail and Slug proteins. J Biol Chem 2012; 287: 6218–6229. 8 Come C, Magnino F, Bibeau F, De Santa Barbara P, Becker KF, Theillet C et al. Snail and slug play distinct roles during breast carcinoma progression. Clin Cancer Res 2006; 12: 5395–5402. 9 Hajra KM, Chen DYS, Fearon ER. The SLUG zinc-finger protein represses E-cadherin in breast cancer. Cancer Res 2002; 62: 1613–1618. 10 Proia TA, Keller PJ, Gupta PB, Klebba I, Jones AD, Sedic M et al. Genetic predisposition directs breast cancer phenotype by dictating progenitor cell fate. Cell Stem Cell 2011; 8: 149–163. 11 Storci G, Sansone P, Trere D, Tavolari S, Taffurelli M, Ceccarelli C et al. The basallike breast carcinoma phenotype is regulated by SLUG gene expression. J Pathol 2008; 214: 25–37. 12 Bolós V, Peinado H, Pérez-Moreno MA, Fraga MF, Esteller M, Cano A. The transcription factor Slug represses E-cadherin expression and induces epithelial to mesenchymal transitions: a comparison with Snail and E47 repressors. J Cell Sci 2003; 116: 499–511. 13 Daniel CW, Strickland P, Friedmann Y. Expression and functional role of E- and P-cadherins in mouse mammary ductal morphogenesis and growth. Dev Biol 1995; 169: 511–519. 14 Albergaria A, Ribeiro AS, Vieira A-F, Sousa B, Nobre AR, Seruca R et al. P-cadherin role in normal breast development and cancer. Int J Dev Biol 2011; 55: 811–822. 15 Ribeiro AS, Albergaria A, Sousa B, Correia AL, Bracke M, Seruca R et al. Extracellular cleavage and shedding of P-cadherin: a mechanism underlying the invasive behaviour of breast cancer cells. Oncogene 2010; 29: 392–402. 16 Paredes J, Figueiredo J, Albergaria A, Oliveira P, Carvalho J, Ribeiro AS et al. Epithelial E- and P-cadherins: role and clinical significance in cancer. Biochim Biophys Acta 2012; 1826: 297–311. 17 Vieira A-F, Ricardo S, Ablett MP, Dionísio MR, Mendes N, Albergaria A et al. P-cadherin is coexpressed with CD44 and CD49f and mediates stem cell properties in basal-like breast cancer. Stem Cells 2012; 30: 854–864. 18 Behrens J, Löwrick O, Klein HL, Birchmeier W. The E-cadherin promoter: functional analysis of a GC-rich region and an epithelial cell-specific palindromic regulatory element. Proc Natl Acad Sci USA 88: 11495–11499. 19 Batlle E, Sancho E, Francí C, Domínguez D, Monfar M, Baulida J et al. The transcription factor snail is a repressor of E-cadherin gene expression in epithelial tumour cells. Nat Cell Biol 2000; 2: 84–89. 20 Rembold M, Ciglar L, Yáñez-Cuna JO, Zinzen RP, Girardot C, Jain A et al. A conserved role for Snail as a potentiator of active transcription. Genes Dev 2014; 28: 167–181. 21 Ye X, Tam WL, Shibue T, Kaygusuz Y, Reinhardt F, Ng Eaton E et al. Distinct EMT programs control normal mammary stem cells and tumour-initiating cells. Nature 2015; 525: 256–260. 22 Radice GL, Ferreira-Cornwell MC, Robinson SD, Rayburn H, Chodosh LA, Takeichi M et al. Precocious mammary gland development in P-cadherindeficient mice. J Cell Biol 1997; 139: 1025–1032. 23 Chen H, Zhu G, Li Y, Padia RN, Dong Z, Pan ZK et al. Extracellular signal-regulated kinase signaling pathway regulates breast cancer cell migration by maintaining slug expression. Cancer Res 2009; 69: 9228–9235.

Oncogene (2017) 1 – 11

24 Savagner P, Kusewitt DF, Carver EA, Magnino F, Choi C, Gridley T et al. Developmental transcription factor slug is required for effective re-epithelialization by adult keratinocytes. J Cell Physiol 2005; 202: 858–866. 25 Kurley SJ, Bierie B, Carnahan RH, Lobdell NA, Davis MA, Hofmann I et al. p120catenin is essential for terminal end bud function and mammary morphogenesis. Development 2012; 139: 1754–1764. 26 Paredes J, Correia AL, Ribeiro AS, Milanezi F, Cameselle-Teijeiro J, Schmitt FC. Breast carcinomas that co-express E- and P-cadherin are associated with p120catenin cytoplasmic localisation and poor patient survival. J Clin Pathol 2008; 61: 856–862. 27 Taniuchi K, Nakagawa H, Hosokawa M, Nakamura T, Eguchi H, Ohigashi H et al. Overexpressed P-cadherin/CDH3 promotes motility of pancreatic cancer cells by interacting with p120ctn and activating rho-family GTPases. Cancer Res 2005; 65: 3092–3099. 28 Reynolds AB, Daniel J, McCrea PD, Wheelock MJ, Wu J, Zhang Z. Identification of a new catenin: the tyrosine kinase substrate p120cas associates with E-cadherin complexes. Mol Cell Biol 1994; 14: 8333–8342. 29 Vieira A-F, Ribeiro AS, Dionísio MR, Sousa B, Nobre AR, Albergaria A et al. P-cadherin signals through the laminin receptor α6β4 integrin to induce stem cell and invasive properties in basal-like breast cancer cells. Oncotarget 2014; 5: 679–692. 30 Dohn MR, Brown MV, Reynolds AB. An essential role for p120-catenin in Src- and Rac1-mediated anchorage-independent cell growth. J Cell Biol 2009; 184: 437–450. 31 Noren NK, Liu BP, Burridge K, Kreft B. p120 catenin regulates the actin cytoskeleton via Rho family GTPases. J Cell Biol 2000; 150: 567–580. 32 Ezzell RM, Goldmann WH, Wang N, Parashurama N, Parasharama N, Ingber DE. Vinculin promotes cell spreading by mechanically coupling integrins to the cytoskeleton. Exp Cell Res 1997; 231: 14–26. 33 Miyamoto S, Teramoto H, Coso OA, Gutkind JS, Burbelo PD, Akiyama SK et al. Integrin function: molecular hierarchies of cytoskeletal and signaling molecules. The Journal of Cell Biology 1995; 131: 791–805. 34 Turner CE, Glenney JR, Burridge K. Paxillin: a new vinculin-binding protein present in focal adhesions. The Journal of Cell Biology 1990; 111: 1059–1068. 35 Sarrió D, Palacios J, Hergueta-Redondo M, Gómez-López G, Cano A, MorenoBueno G. Functional characterization of E- and P-cadherin in invasive breast cancer cells. BMC Cancer 2009; 9: 74. 36 Gamallo C, Moreno-Bueno G, Sarrió D, Calero F, Hardisson D, Palacios J. The prognostic significance of P-cadherin in infiltrating ductal breast carcinoma. Mod Pathol 2001; 14: 650–654. 37 Bernardes N, Ribeiro AS, Abreu S, Mota B, Matos RG, Arraiano CM et al. The bacterial protein azurin impairs invasion and FAK/Src signaling in P-cadherinoverexpressing breast cancer cell models. PLoS One 2013; 8: e69023. 38 Plutoni C, Bazellieres E, Le Borgne-Rochet M, Comunale F, Brugues A, Séveno M et al. P-cadherin promotes collective cell migration via a Cdc42-mediated increase in mechanical forces. J Cell Biol 2016; 212: 199–217. 39 Turashvili G, McKinney SE, Goktepe O, Leung SC, Huntsman DG, Gelmon KA et al. P-cadherin expression as a prognostic biomarker in a 3992 case tissue microarray series of breast cancer. Mod Pathol 2010; 24: 64–81. 40 Stefansson IM, Salvesen HB, Akslen LA. Prognostic impact of alterations in P-cadherin expression and related cell adhesion markers in endometrial cancer. J Clin Oncol 2004; 22: 1242–1252. 41 Kümper S, Ridley AJ. p120ctn and P-cadherin but not E-cadherin regulate cell motility and invasion of DU145 prostate cancer cells. PLoS One 2010; 5: e11801. 42 Rakshit S, Zhang Y, Manibog K, Shafraz O, Sivasankar S. Ideal, catch, and slip bonds in cadherin adhesion. Proc Natl Acad Sci USA 2012; 109: 18815–18820. 43 Tanaka-Matakatsu M, Uemura T, Oda H, Takeichi M, Hayashi S. Cadherin-mediated cell adhesion and cell motility in Drosophila trachea regulated by the transcription factor Escargot. Development 1996; 122: 3697–3705. 44 Hirai Y, Nose A, Kobayashi S, Takeichi M. Expression and role of E- and P-cadherin adhesion molecules in embryonic histogenesis. II. Skin morphogenesis. Development 1989; 105: 271–277. 45 Nose A, Takeichi M. A novel cadherin cell adhesion molecule: its expression patterns associated with implantation and organogenesis of mouse embryos. J Cell Biol 1986; 103: 2649–2658. 46 Ribeiro AS, Sousa B, Carreto L, Mendes N, Nobre AR, Ricardo S et al. P-cadherin functional role is dependent on E-cadherin cellular context: a proof of concept using the breast cancer model. J Pathol 2013; 229: 705–718. 47 Savagner P. Epithelial–mesenchymal transitions. In: Cellular Adhesion in Development and Disease. Elsevier: Amsterdam, Netherlands, 2015, pp 273–300. 48 Nguyen-Ngoc K-V, Cheung KJ, Brenot A, Shamir ER, Gray RS, Hines WC et al. ECM microenvironment regulates collective migration and local dissemination in normal and malignant mammary epithelium. Proc Natl Acad Sci USA 2012; 109: E2595–E2604.

© 2017 Macmillan Publishers Limited, part of Springer Nature.

Slug/Pcad in mammary epithelial cells Y Idoux-Gillet et al

11 49 Hatta K, Takeichi M. Expression of N-cadherin adhesion molecules associated with early morphogenetic events in chick development. Nature 1986; 320: 447–449. 50 Nieman MT, Prudoff RS, Johnson KR, Wheelock MJ. N-cadherin promotes motility in human breast cancer cells regardless of their E-cadherin expression. J Cell Biol 1999; 147: 631–644. 51 Lamouille S, Xu J, Derynck R. Molecular mechanisms of epithelial-mesenchymal transition. Nat Rev Mol Cell Biol 2014; 15: 178–196. 52 Uygur B, Wu WS. SLUG promotes prostate cancer cell migration and invasion via CXCR4/CXCL12 axis. Mol Cancer 2011; 10: 139. 53 Tang Y, Feinberg T, Keller ET, Li XY, Weiss SJ. Snail/Slug binding interactions with YAP/TAZ control skeletal stem cell self-renewal and differentiation. Nat Cell Biol 2016; 18: 917–929. 54 Faraldo MM, Teulière J, Deugnier M-A, Birchmeier W, Huelsken J, Thiery JP et al. beta-Catenin regulates P-cadherin expression in mammary basal epithelial cells. FEBS Lett 2007; 581: 831–836. 55 Albergaria A, Resende C, Nobre AR, Ribeiro AS, Sousa B, Machado JC et al. CCAAT/ enhancer binding protein β (C/EBPβ) isoforms as transcriptional regulators of the Pro-invasive CDH3/P-cadherin gene in human breast cancer cells. PLoS One 2013; 8: e55749–9. 56 Shimomura Y, Wajid M, Shapiro L, Christiano AM. P-cadherin is a p63 target gene with a crucial role in the developing human limb bud and hair follicle. Development 2008; 135: 743–753. 57 Ko SY, Naora H. HOXA9 promotes homotypic and heterotypic cell interactions that facilitate ovarian cancer dissemination via its induction of P-cadherin. Mol Cancer 2014; 13: 1–13. 58 Ray PS, Wang J, Qu Y, Sim M-S, Shamonki J, Bagaria SP et al. FOXC1 is a potential prognostic biomarker with functional significance in basal-like breast cancer. Cancer Res 2010; 70: 3870–3876.

59 Thuault S, Hayashi S, Lagirand-Cantaloube J, Plutoni C, Comunale F, Delattre O et al. P-cadherin is a direct PAX3-FOXO1A target involved in alveolar rhabdomyosarcoma aggressiveness. Oncogene 2013; 32: 1876–1887. 60 Albergaria A, Ribeiro AS, Pinho S, Milanezi F, Carneiro V, Sousa B et al. ICI 182,780 induces P-cadherin overexpression in breast cancer cells through chromatin remodelling at the promoter level: a role for C/EBPbeta in CDH3 gene activation. Hum Mol Genet 2010; 19: 2554–2566. 61 Grimm SL, Rosen JM. The role of C/EBPbeta in mammary gland development and breast cancer. J Mammary Gland Biol Neoplasia 2003; 8: 191–204. 62 Medina D, Oborn CJ, Kittrell FS, Ullrich RL. Properties of mouse mammary epithelial cell lines characterized by in vivo transplantation and in vitro immunocytochemical methods. J Natl Cancer Inst 1986; 76: 1143–1156. 63 Deugnier M-A, Faraldo MM, Teulière J, Thiery JP, Medina D, Glukhova MA. Isolation of mouse mammary epithelial progenitor cells with basal characteristics from the Comma-Dbeta cell line. Dev Biol 2006; 293: 414–425. 64 Boukamp P, Petrussevska RT, Breitkreutz D, Hornung J, Markham A, Fusenig NE. Normal keratinization in a spontaneously immortalized aneuploid human keratinocyte cell line. J Cell Biol 1988; 106: 761–771. 65 Annicotte J-S, Fayard E, Swift GH, Selander L, Edlund H, Tanaka T et al. Pancreaticduodenal homeobox 1 regulates expression of liver receptor homolog 1 during pancreas development. Mol Cell Biol 2003; 23: 6713–6724. 66 Curtis C, Shah SP, Chin SF, Turashvili G, Rueda OM, Dunning MJ et al. The genomic and transcriptomic architecture of 2000 breast tumours reveals novel subgroups. Nature 2012; 486: 346–352. 67 Pereira B, Chin SF, Rueda OM, Vollan HK, Provenzano E, Bardwell HA et al. The somatic mutation profiles of 2,433 breast cancers refines their genomic and transcriptomic landscapes. Nat Commun 2016; 7: 11479. 68 Györffy B, Lanczky A, Eklund AC, Denkert C, Budczies J, Li Q et al. An online survival analysis tool to rapidly assess the effect of 22,277 genes on breast cancer prognosis using microarray data of 1,809 patients. Breast Cancer Res Treat 2010; 123: 725–731.

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

© 2017 Macmillan Publishers Limited, part of Springer Nature.

Oncogene (2017) 1 – 11

Pcad pathway controls epithelial cell dynamics in mammary gland and breast carcinoma.

Mammary gland morphogenesis results from the coordination of proliferation, cohort migration, apoptosis and stem/progenitor cell dynamics. We showed e...
3MB Sizes 1 Downloads 10 Views