(A) Establishment of MCF10A stable cell lines overexpressing ErbB2 alone, 14-3-3ζ alone or both ErbB2 and 14-3-3ζ. Multiple stable clones were established for each sublines, and most experiments were repeated with different clones to rule out clonal effects. Immunoblot analysis of one representative of each subline for the indicated proteins is shown.
(B) Co-overexpression of ErbB2 and 14-3-3ζ led to anchorage-independent growth of MCF10A cells in soft agar assay.
(C) Distinct acinar structures of MCF10A sublines in 3D culture. Top: Phase-contrast images of acinar structures (19 days); bar: 200 μm. Inset: Amplified view of individual 10A.ErbB2.ζ cell invading into surrounding matrigel; bar: 100 μm.
(D) Loss of basement membrane integrity in 10A.ErbB2.ζ acini. MCF10A sublines were cultured in 3D matrigel for 28 days and then stained for laminin V (red) and DAPI (blue). Representative images were shown. Bar: 100 μm.

(A) Transwell migration assay of the indicated MCF10A sublines. Cells that migrated to the bottom of the chamber were counted in five fields under 20X magnification. Experiments were done three times with triplicates, error bar represents SEM.
(B) Wound healing assay of the indicated MCF10A sublines. Wound closures were photographed at 0 and 6 hours after wounding. Bar: 200 μm.
(C) The four MCF10A sublines exhibited different morphologies in 2D culture. Bar: 100 μm.
(D) IFS of EMT markers in MCF10A sublines. E-cadherin (top, red), vimentin (bottom, red), DAPI (blue). Bar: 100 μm.
(E) Immunoblot analysis of the indicated EMT markers in MCF10A sublines. Expression of epithelial cell markers (E-cadherin, β-catenin, α-catenin, p120 catenin) and mesenchymal cell marker (N-cadherin) were examined by immunoblot analysis in both 2D and 3D culture cell lysates.

(A) E-cadherin mRNA was dramatically reduced in 10A.ErbB2.ζ and 10A.14-3-3ζ cells.
(B) Expression of EMT regulators in the four MCF10A sublines. Top: Immunoblot analysis of EMT regulators, including snail, twist, and slug. Bottom: RT-PCR analysis of EMT regulators, including E12, E47, and delta EF1.
(C) RT-PCR and immunoblot analysis of ZFHX1B levels in the four MCF10A sublines. MDA-MB-435 cells served as the positive control for ZFHX1B protein expression.
(D) 14-3-3ζ overexpression led to transcriptional repression of E-cadherin promoter activity. Top: Schematic representation of the luciferase reporter driven by E-cadherin proximal promoter containing two ZFHX1B binding E-box motif: pGL3.Ecad. Bottom: Relative luciferase activity of pGL3.Ecad in the four MCF10A sublines, P<0.05. Error bars indicate SEM.
(E) Downregulation of ZFHX1B in 14-3-3ζ–overexpressing MCF10A sublines by siRNA partially relieved suppression of E-cadherin promoter-driven luciferase activity. Top: Control siRNA and ZFHX1B siRNA were transfected into 10A.ErbB2.ζ cells and 10A.14-3-3ζ cells. After 48 hours, downregulation of ZFHX1B was examined by RT-PCR with GAPDH as internal control. Bottom: Cells were pre-transfected with ZFHX1B siRNA or control siRNA. After 48 hours, pGL3.Ecad was co-transfected with pRL.TK plasmid as transfection efficiency control. Relative luciferase activity was determined 36 hours later. Error bars indicate SEM.
(F) A reverse correlation between ZFHX1B and E-cadherin expression in breast cancer cell lines. RNA was extracted from four E-cadherin–negative breast cancer cell lines (BT549, Hs578T, MDA-MB-231, MDA-MB-435), five E-cadherin–positive breast cancer cell lines (MCF-7, BT474, T47D, MDA-MB-361, MDA-MB-453), and a non-transformed breast epithelial cell line (MCF10A), then reverse transcribed into cDNA, followed by PCR with ZFHX1B specific primers. GAPDH served as loading control.

(A) Immunoblot and RT-PCR analysis of TβRI and TβRII expression in MCF10A sublines.
(B) Immunoblot analysis of nuclear p-smads (p-smad2/p-smad3), T-smad2/3, with lamin C as nuclear fraction loading control.
(C) Chromatin immunoprecipitation assay with the indicated antibodies showed smad3 binding with ZFHX1B promoter in 10A.ErbB2.ζ and 10A.14-3-3ζ cells (arrows), not in 10A.Vec and 10A.ErbB2 cells (empty arrows).
(D) Knockdown of 14-3-3ζ by siRNA reduced TβRI and ZFHX1B expression in 10A.ErbB2.ζ and 10A.14-3-3 ζ cells. TβRI level and ZFHX1B level were analyzed by immunoblot and RT-PCR, respectively, 48 hours after siRNA transfection.
(E) 14-3-3ζ inhibited TβRI ubiquitination. 10A.ErbB2.ζ and 10A.ErbB2 cells were co-transfected with vectors expressing myc-TβRI and HA-ubiquitin (left), 10A.ErbB2.ζ and Hela cells were co-transfected with myc-TβRI, HA-ubiquitin, and control/14-3-3ζ siRNA (middle), Hela cells were co-transfected with myc-TβRI, HA-ubiquitin, and pcDNA3.Vec/pcDNA3.14-3-3ζ (right).After 48 hours, cell lysates were collected and subjected to immunoprecipitation and immunoblot with myc and HA antibodies.
(F) 10A.ErbB2 and 10A.ErbB2.ζ cells were treated with DMSO or 20 μg/ml MG132 for 6 hours; TβRI levels were analyzed by immunoblot.
(G) 14-3-3ζ associated with TβRI. 10A.ErbB2.ζ and 10A.14-3-3ζ cell lysates were immunoprecipitated by anti-HA antibody, followed by immunoblot analysis of TβRI.
(H) Schematic representation of different TβRI mutants.
(I) 14-3-3ζ binds to TβRI at its kinase domain between amino acid 210 and 370. 10A.ErbB2.ζ cells were infected with lentivirus expressing different TβRI mutants as in (H). Then the cell lysates were subjected to pull-down assay with GST or GST-14-3-3ζ, followed by immunoblot with TβRI antibody.

(A) LY2109761 efficiently inhibited smad2/3 phosphorylation in 10A.ErbB2.ζ cells. Cells were treated with TGFβ1 (5 ng/ml) or TGFβ1 (5 ng/ml) plus TβR inhibitor LY2109761 (10 μM) for 60 minutes, then analyzed by immunoblot with indicated antibodies (left). Cells were treated with increasing dose of LY2109761, then analyzed by immunblot for p-Akt, T-Akt, p-P42, T-P42 and β-actin (right).
(B) Treatment of 10A.ErbB2.ζ cells with TGFβ receptor inhibitor partially reversed the EMT morphology in 2D culture and inhibited the invasiveness of acini in 3D culture. Left: Cells were treated with LY2109761 for 48 hours, and images were photographed under a phase-contrast microscope. Middle: 10A.ErbB2.ζ cells were cultured on 3D matrigel for 13 days to form invasive acinar structures, and then treated with DMSO or LY2109761 for 6 days. Right: 10A.ErbB2.ζ cells were treated with DMSO or LY2109761 and subjected to immunofluorescence staining of E-cadherin. Bar: 100 μm. Cells lysates were also collected after treatment, and subjected to immunoblot with E-cadherin, α-catenin and β-actin antibodies.
(C) Prolonged LY2109761 treatment (15days) reduced mesenchymal protein expression (fibronectin, vimentin).
(D) E-cadherin expression was stably re-introduced into 10A.ErbB2.ζ cells.
(E) E-cadherin restoration led to increased epithelial protein expression (α-catenin, β-catenin, p120-catenin) and decreased mesenchymal protein expression (N-cadherin, vimentin).
(F) E-cadherin restoration inhibited 10A.ErbB2.ζ cell invasion. Phase-contrast microscopy of acinar structures of 10A.ErbB2.ζ.Vec and 10A.ErbB2.ζ.Ecad cells in 3D culture. Bar: 200 μm.

(A) 140 DCIS (top) and 107 IBC (bottom) samples were subjected to IHC staining for TβRI and 14-3-3ζ, of which 138 DCIS and 100 IBC cases were satisfactory for analysis. Chi-square test indicated a correlation between TβRI and 14-3-3ζ expression in both cohorts, P<0.05.
(B) Representative IHC staining for ErbB2, 14-3-3ζ, TβRI, E-cadherin, and N-cadherin in a DCIS case with microinvasion and a low-grade DCIS. Bar: 50 μm.
(C) 14-3-3ζ overexpression in TM15 cells increased lung metastasis. TM15 cells were stably transfected with HA-14-3-3ζ or empty vector. TM15.Vec and TM15.14-3-3ζ cells were injected into mammary fat pad of nude mice (four mice each group). When the primary tumors reached the size of 150 mm³, mice were sacrificed and their lungs underwent histological analysis to measure metastasis, P<0.05. Error bar indicates SEM. Experiment was repeated twice and similar results were observed.
(D) Co-overexpression of ErbB2 and 14-3-3ζ in IBC correlated with higher rates of death and disease recurrence. Tumor samples from 107 breast cancer patients were IHC-stained for ErbB2 and 14-3-3ζ, Kaplan-Meier plots of overall survival (left) and disease-free survival (right) according to ErbB2 and 14-3-3ζ expression are shown. P<0.05 (log-rank test).

(A) A model of how co-overexpression of ErbB2 and 14-3-3ζ promotes invasion. ErbB2 overexpression increased cell proliferation and motility; 14-3-3ζ overexpression decreased cell-cell adhesion via induction of EMT. Collectively, co-overexpression of ErbB2 and 14-3-3ζ promoted cell invasion. Co-overexpression of ErbB2 and 14-3-3ζ may promote progression from DCIS to IBC via this mode of cooperation. ↑: increase; ↓: decrease; −: no significant effect.
(B) A model of 14-3-3ζ–mediated E-cadherin repression. 14-3-3ζ interacts with and stabilizes TβRI, leading to smad2/3 phosphorylation and translocation to the nucleus, where smads bind to ZFHX1B promoter to increase its transcription. ZFHX1B then represses E-cadherin transcription by binding to its promoter.