Human tissue samples were obtained from the archive of The Pathology Department of Hospital del Mar with the approval of the Bank Tumor Committee, according to the Spanish Ethical regulations. This study was approved by the local Ethics Board (Clinical Research Ethical Committee of the Parc de Salut Mar, CEIC-IMAS2009/3515/) and followed the guidelines of the Declaration of Helsinki and patient identity of the pathological specimens remained anonymous in the context of this study. Informed written consent was obtained from all subjects.
All animal work has been performed at IRB, Parc Científic, Barcelona and followed the protocols approved by the Animal Care and Use Committee from Parc Científic de Barcelona. Decree 214/1997 regulates the use of animals for research. Along these lines the Autonomous Government of Catalonia approved the Animal Welfare Committee of the Parc Científic de Barcelona, responsible for evaluating activities undertaken in the centre.
Cell lines and cultures
Seven human breast cancer cell lines (MDA-MB-231, MDA-MB-453, MDA-MB-468, SK-BR-3, BT-474, MCF-7 and T47D), a human cervical cancer cell line (HeLa) and a mammary epithelial one as control (MCF-10A) were purchased from the American Type Culture Collection and cultured at 37°C with 5% CO2 in DMEM (MDA-MB-231, MDA-MB-453, MDA-MB-468, BT-474, SK-BR-3, MCF-7, 2 supplemented with 10% fetal bovine serum. For BT-474, insulin (0.01 mg/ml) was added. Line MCF-10A was cultured in MEMB Basal Medium (Clonetics, cat. No. CC-3151) supplemented with MEGM Singlequots (Clonetics, cat. No. CC-4136): 2 ml BPE, 500 ng/ml hydrocortisone, 10 ng/ml hEGF and 5 mg/ml insulin. Cell lines were transfected using PPEI (PolySciences).
Antibodies and Reagents
Anti p65 (sc-109 and sc-372), p50 (sc-7178), IκBα (sc-371), PAN-14-3-3 (sc-629), 14-3-3β (sc-17288) and 14-3-3σ (sc-7681 and sc-7683) were from Santa Cruz. α-α-tubulin was from Sigma, and α-ki67 from Novocastra. Secondary HRP-conjugated antibodies were from DAKO and fluorescein-conjugated secondary antibodies from Molecular Probes. BAY11-7082 (Calbiochem) was used at 100 µM and BAY65-5811 is a gift from Bayer. TNFα from Preprotech was used at 40 ng/ml.
PD assays were performed as previously described 
, Briefly, GST fusion proteins were purified and incubated with cell lysates for 45 min in a rotary shaker at 4°C. Precipitates were washed and analyzed by western blot.
Subcellular fractionation and western blotting
For cytoplasm-nuclear separations, cells were lysed in 10 mM HEPES, 1.5 mM MgCl2, 10 mM KCl, 0.05%NP40 at pH 7.9, 10 min on ice and centrifuged at 3.000 rpm. Supernatants were recovered as cytoplasmic fraction and the pellets lysed in 5 mM HEPES, 1.5 mM MgCl2, 0.2 mM EDTA, 0.5 mM DTT and 26% glycerol and sonicated 5 min three times to recover the soluble nuclear fractions. Protein lysates were run in SDS/PAGE and transferred to immobilon-P transfer membranes (Millipore). Membranes were blocked with 4% dehydrated milk and blotted with the indicated antibodies O/N at 4°C. Secondary HRP-linked secondary antibodies were developed using the ECL system from Amersham.
Precipitations were performed in 500 µl of PBS containing 0.5% Triton X-100, 1 mM EDTA, 100 µM sodium orthovanadate, 0.2 mM PMSF, 20 mM β-glycerophosphate and complete protease inhibitor cocktail (Roche). Lysates were incubated for 2 hours at 4°C with primary antibodies coupled to Protein A-Sepharose beads, washed and analyzed by western blot.
Cells were fixed in 4% paraformaldehyde and permeabilized with 0.3% Triton X-100 and 4% non-fat dry milk in PBS. Primary antibodies were incubated overnight and secondary antibodies for 90 min. Slides were mounted in Vectashield with DAPI (Vector) and visualized in an Olympus BX61. Images were analyzed using the Cel
B Software from Olympus.
Paraffin-embedded 4 µm sections were kept one hour at 60°C and re-hydrated. Antigen retrieval was done in citrate buffer (pH 6.0) at 80°C, and primary antibodies were incubated overnight at 4°C. Samples were washed in ChemMate buffer solution and developed with Envision signal detection system (DAKO). Two investigators, blinded to clinical data, scored each sample, for nuclear p65 and 14-3-3σ staining. Concordance between investigators was higher than 95%.
Electrophoretic mobility shift assay (EMSA)
Nuclear extracts were prepared as described 
. Briefly, 3 µg of the indicated nuclear extracts were incubated with labeled double-stranded κB-binding consensus oligonucleotide (5-AGTTGAGGGGACTTTCCCAGGC
-3) 15 min at room temperature in a final volume of 20 µL. Samples were run in 0.5% tris–borate–EDTA buffer on non-denaturating 4% polyacrylamide gels, which were vacuum-dried and subjected to autoradiography. Densitometric analysis of the bands was performed using the Adobe Photoshop CS4 software.
Total RNA was isolated using RNeasy kit (Qiagen) and cDNA was obtained with RT-First Strand cDNA Synthesis kit (Amersham). qRT-PCR was performed in LightCycler 480 system sing SYBR Green I master kit (Roche).
Control or 14-3-3σ-reconstituted MCF7 cells were incubated with TNFa for 0, 20 and 90 min in two independent experiments. RNA was obtained and RNA integrity assessed using Agilent 2100 Bioanalyzer (Agilent Technologies, Palo Alto, CA). All samples had an RNA integrity number (RIN)≥8.5. Amplification, labeling and hybridizations were performed according to protocols from Ambion and Affymetrix. Briefly, 250 ng of total RNA were amplified using the Ambion® WT Expression Kit (Ambion/Applied Biosystems, Foster City, CA, USA), labeled using the WT Terminal Labeling Kit (Affymetrix Inc., Santa Clara, CA, USA), and then hybridized to Human Gene 1.0 ST Array (Affymetrix) in a GeneChip® Hybridization Oven 640. Washing and scanning were performed using the Hybridization Wash and Stain Kit and the GeneChip® System of Affymetrix (GeneChip® Fluidics Station 450 and GeneChip® Scanner 3000 7G).
Microarray data processing
Microarray data was processed with R statistical framework 
. Aroma.affymetrix 
, corpcor 
and Limma 
packages were used to preprocess, batch adjust and determine the top 220 most differentially expressed genes across conditions with an adjusted p-value (False Discovery Rate, FDR) <0.05 
. The analysis was designed to identify genes that were induced by TNFα in control cells but not in 14-3-3σ-expressing MCF7 cells (group 1); genes induced in both control and 14-3-3σ-expressing MCF7 cells (group 2) including genes that were differentially induced in control and 14-3-3σ-expressing MCF7 cells (group 2.1). Groups 1 and 2.1 were included as 1_2.1 in subsequent analysis. Values represented for each probe set correspond to standard deviations after mean-centering the values in each row. Hierarchical clustering was performed on samples and also on the 220 most differentially expressed genes using Pearson correlation as distance measure and the average method, and the obtained clusters were used to generate the groups of genes for further analysis.
Functional enrichment analysis of 14-3-3σ gene signatures
We performed enrichment analysis (EA) of the different groups (see microarray data processing) with Gene Ontology terms. This analysis reveals the enrichment of a particular functional term in a group of genes. The EA was done with binomial test using Gitools 
, a statistical framework for the analysis and visualization of genomics data. The annotation of genes to corresponding Gene Ontology Biological Process, Cell Location and Molecular Function terms were downloaded from Ensembl version 60 
using Gitools importer. The p-values resulting from the binomial test were adjusted for multiple testing using the Benjamini and Hochberg's method of FDR and we set the cut-off for significance at a 0.05 of the corrected p-value. Results are represented in a heat-map of p-values where colors towards red indicate significance and grey indicates no significant enrichment.
14.3.3σ dependent signature up-regulation in cancer gene modules
To test whether genes in the 14-3-3σ-dependent signature (group 1_2.1) are enriched in cancer samples, we used sets of genes significantly upregulated in different cancer genomic experiments as obtained from IntOGen 
. Briefly, IntOGen contains a collection of independent cancer genomic datasets from different tumor types, which are processed to identify genes altered in a significant number of samples. Enrichment analysis was performed with Gitools as explained above.
Relapse-free survival analysis
To test whether expression of genes in the 14-3-3σ dependent signature correlates with tumor relapse we used Sample Level Enrichment Analysis (SLEA) as described in Gundem and Lopez-Bigas 2012 
with a dataset containing 204 primary tumors from breast cancer patients with known site of relapse (GEO id GSE12276, 
. Briefly, SLEA assesses the transcriptional status of a set of genes in each sample using Z-score test (see 
for details), the result is a Z-score per sample indicating the transcriptional status of the gene set in each sample. High Z-scores indicate that the set of genes tend to have high expression, low Z-scores indicate that the set of genes tend to have low expression and Z-scores around 0 indicate that they do not show a clear tendency to be misregulated in any direction. We used the Z-score to stratify the 204 samples according to the expression levels of genes in 14-3-3σ dependent signature and generated two groups: samples with significant up-regulation of 14-3-3σ dependent signature (n
38) and the rest of samples (n
166). Analysis of the relapse-free survival of patients in the two groups was performed using R Bioconductor survival package 
Metastasis Assay in nude mice
A cell suspension of 0.5×106
MDA-MB-231 breast cancer cells was injected in the left cardiac ventricle of anesthetized 6–7-week-old nude mice. Tumor development was monitored by weekly bioluminescence imaging using the IVIS-200 imaging system from Xenogen 
. Bone metastatic lesions were localized by ex vivo bioluminescence imaging and confirmed by histological analysis after necropsy. Bones were decalcified, fixed and processed for histological analysis (Hematoxylin & Eosin staining, H&E) and IHC. All animal work followed the protocols approved by the Animal Care and Use Committee of the Barcelona Science Parc.
Computational promoter analysis
Presence of NF-κB binding sites in the promoters of interest (from −3000 bp to +100 bp to the transcription start site (TSS)) was predicted using the Genomatix software.