Foxa1

Chromatin immunoprecipitation (ChIP) was performed following the method previously described [54 (link)]. Briefly, crosslinked chromatin was collected from MCF7 cells asynchronously grown in full media at 80% confluence. ZNF217 antibody (previously described [4 (link)]), H3K4me3, H3K9me3, or H3K27me3 antibodies were incubated with sonicated chromatin overnight. Libraries were constructed and analyzed using an Illumina Hiseq2000. ChIP-seq experiments were performed in duplicate using MCF7 cells grown on different days using independently performed ChIP assays. Replicate input chromatin samples were also included as a control. Additional ChIP assays were performed for ERα (Santa Cruz Biotechnology; cat #scbt8002) and FOXA1 (Abcam #23738). ChIP-qPCR was performed using iQ SYBR Green Supermix (BioRad). Primers used to validate TF binding are listed in Additional file 15: Supplementary Methods.

Total protein was extracted using a RIPA kit (Beyotime, Shanghai, China) containing a 1% dilution of the protease inhibitor PMSF (Beyotime). Protein concentrations were determined by the enhanced BCA Protein Assay kit (Beyotime). Equal amounts of protein in each lane were separated by 8% SDS-PAGE and transferred to a PVDF membrane (Millipore, Billerica, MA, USA). After blocking the membrane in blocking buffer (5% milk powder in 20 mM Tris–HCl pH 7.5, 500 mM NaCl, 0.1% (v/v) Tween 20), the membrane was incubated with primary antibodies against FOXA1 (1:1000; Abcam), AR (1:2000; Cell Signaling Technology, Danvers, MA, USA), Notch1 (1:2000; Epitomics, Burlingame, CA, USA), Hes1 (1:2000; Epitomics), and β-actin (1:2000, Cell Signaling Technology) at 4°C overnight. Peroxidase-linked secondary anti-rabbit or anti-mouse antibodies were used to detect the bound primary antibodies.

Cells were lysed with NP40 Lysis buffer (Beyotime, China) and protein concentration was tested by BCA kits (Thermo, USA). Protein lysates were separated by SDS-PAGE, then transferred to PVDF membranes (Millipore, USA) and blocked with 5% fat-free milk. The membranes were incubated overnight at 4 °C with the following primary antibodies: DAB2IP (Proteintech, USA), PI3K-p85 (CST, USA), p-AKT (CST, USA), AKT (CST, USA), p-JNK (CST, USA), JNK (CST, USA), Nanog (Proteintech, USA), Bak (CST, USA), Bax (CST, USA), Bcl-2 (Proteintech, USA), β-actin (Servicebio, China), FOXA1 (Abcam, UK), and GAPDH (Servicebio, China). Anti-rabbit IgG-HRP (Servicebio, China) secondary antibody was then applied. Chemiluminescent signals were detected by Enhanced Chemiluminescence (Thermo, USA). Results were confirmed by at least three independent experiments.

Tissue microarrays (1.0 mm cores, 4 μm sections) were analyzed with antibodies against CCNB1 (Y106, Epitomics), EGFR (3C6, Ventana), FOXA1 (2F83, Abcam), FOXM1 (C-20, Santa Cruz), GATA3 (D13C9, Cell Signaling), phospho-HISTH3 (Ser10) (#9701, Cell signaling), KRT5 (EP1601Y, Thermo Scientific), LAMA5 (4C7, Dako), PLK1 (208G4, Cell signaling), PPARG (C26H12 Cell Signaling), RXRA (F-1, Santa Cruz), STAT3 (124H6, Cell signaling), phospho-STAT3 (Tyr705) (D3A7, Cell signaling). Cores were evaluated as blinded digitalized image files. For quantitatively staining markers (EGFR, FOXA1, GATA3, KRT5, PPARG, RXRA, STAT3) a tumor cell score (TCS) was defined as described in Sjödahl et al. [4 (link)]. For discrete cellular labeling (CCNB1, FOXM1, PLK1, p-STAT3), fractions of positive tumor cells was recorded. The mean tumor cell score of core pairs from the same sample was calculated. The number of cores evaluated for each marker ranged from 480 to 524. Mitotic figures were identified by the phospho-HISTH3 (Ser10) antibody and basal lamina by anti-Laminin α-5 staining. When possible, lines were drawn through the plane of the mitotic figure (in the direction of the cell division), and tangentially along the nearest basal lamina (Photoshop CS5 version 12.1), and the distance, in cell layers, to the basal membrane recorded. A total of 416 mitoses from 112 different tumors were analyzed.