Human colon cancer cells, SW480 and HT29 (American Type Culture Collection, Manassas, VA), were cultured in RPMI (Cellgro, Mediatech, Inc, Manassas, VA) containing 10% FBS and Pen/Strep at 37°C and 5% CO2. Both SW480 and HT29 cells are part of the NCI-60 panel of cancer cell lines and represent two colon cancer lines with a wealth of biochemical, molecular, proteomic and genomics data, providing an opportunity for meta-analysis [61–63]. Cells were transfected with either rat par-4 cDNA in pCB6+ or with empty vector using Fugene 6 reagent (Roche Diagnostics, Indianapolis, IN, USA). Transfectants were selected with G418 (Gibco, Carlsbad, CA) and colonies expanded and assayed for Par-4 expression. HT29 cells were transfected with 1 μg/ml DROSHA siRNA or the corresponding scrambled siRNA (Thermo Scientific Dharmacon, Lafayette, CO) using Lipofectamine 2000 (Invitrogen, Carlsbad, CA).
Subcellular fractionation, immunoprecipitation and Western blotting
Antibodies used were Par-4, NFκB p 50, p 65, BCL2, MCL1, BCL2L11 rabbit polyclonal, SGK1, VDAC1 goat polyclonal, CDN1B (p27) mouse monoclonal (Santa Cruz, Santa Cruz, CA, USA), AKT1 mouse monoclonal (Cell Signaling, Danvers, MA, USA), PKA goat polyclonal, PARP rabbit polyclonal (Upstate Cell Signaling Solutions, Charlottesville, VA, USA), and β-actin mouse monoclonal (Sigma, Saint Louis, MO, USA). Subcellular fractionations were performed using the NE-PER Nuclear and Cytoplasmic extraction reagent kit (Pierce Biotechnology, Rockford, IL, USA) according to the manufacturer's instructions. Western blotting and immunoprecipitation assays were performed as previously documented 
Synthesis of ISC-4
ISC-4 was synthesized following a method recently developed by Sharma et al. Briefly, a solution of triphosgene (1.48 g, 5.0 mmol) in CH2Cl2 (15 mL) was added dropwise, for a period of 1 h, to a refluxing mixture of phenylbutyl formamide (1.77 g, 10.0 mmol), triethylamine (4.35 g, 6.0 mL, 43.0 mmol) and 4Å molecular sieves in CH2Cl2 (50 mL). The mixture was refluxed for an additional 2.5 h. Selenium powder (1.58 g, 20 mmol) was added and the resulting mixture was refluxed for additional 7 h. The mixture was purified by silica gel column chromatography (EtOAc/hexanes 5:95) to yield 1.7 g (71%) of ISC-4 as viscous oil.
Assays of apoptosis
For caspase-3 activity assay, cells were cultured in 6-well plates with half of the wells treated with 100 μM 5-FU. After 24 hrs cells were harvested and assayed for apoptosis using the Caspase-3 Assay kit from BD Biosciences Pharmingen (San Diego, CA). For annexinV/7AAD assay, cells were cultured in a 12 well plate and treated as above. Cells were trypsinized, washed, and stained with Annexin V, conjugated with PE, and 7AAD (BD Biosciences, San Jose, CA). Staining was detected by flow cytometry. MTS assay was performed in 96 well plates according to manufacturer's protocol (Promega Corp, Madison, WI). After addition of MTS solution, plates were incubated at 37° for 4 h and read at 490 nm using a plate reader.
Treatment with AKT inhibitors
HT29 cells were treated with 3 to 50 μM ISC-4 for 48 hours. In vitro cytotoxic efficacy was measured using 3-(4,5-dimethylthiazol-2-yl)-5-(3-carboxymethoxyphenyl)-2-(4-sulfophenyl)-2H-tetrazolium (MTS) viability assay (Promega, Madison, WI, USA).
NFκB reporter activity assay
NFκB activity was assessed using the Cignal™ NFκB Reporter Assay Kit (SuperArray, Bioscience Corporation, Frederick, MD, USA). Cells were transfected with 6.6 μg/ml of the reporter plasmids. Cells were incubated 48 hours, and assayed using the Dual-Luciferase Reporter Assay System (Promega Corporation, Madison, WI, USA). Cells were lysed, proteins quantitated, and 26 μg protein added to each well of a 96-well opaque white plate. The plate was read in a Synergy plate reader, with KC4 software (Bio-Tek Instruments, Winooski, VT, USA). Firefly luciferase substrate was injected to assess luciferase activity under the control of the NFκB promoter. Renilla luciferase substrate was added to assess transfection efficiency. Each well was read 11 times after a 2 second delay over a period of 10 seconds.
Gene expression profiling analysis
Gene (mRNA) expression profiling experiments were performed on a 39,936 human cDNA microarray employing a common reference design as previously described [66, 67]. LOWESS data normalization, experimental noise determination, and statistical analysis with false discovery rate (FDR) to correct for multiple testing were performed as described previously [66–68]. Biological themes associated with the differentially expressed genes were identified using gene ontology (GO) categories in the Expression Analysis Systematic Explorer (EASE) application , which is executable using TIGR Multi Experiment Viewer (TMEV; available at http://www.tigr.org/softlab). A Fisher's Exact score p < 0.05 was considered significant.
Real-time RT-PCR validation of mRNA expression
Quantitative RT-PCR assay was performed on the ABI 7900 HT Sequence Detection System using Assay on Demand primers and probes and TaqMan Universal PCR Master Mix (Applied Biosystems, Foster City, CA). PCR conditions were 2 minutes at 50°C, 10 minutes at 95°C and 40 cycles of 15 seconds at 95°C and 1 minute at 60°C. ABI SDS 2.2.2 software and the 2-ΔΔCt analysis method  were used to quantitate relative amounts of product using beta-actin as an endogenous control.
Promoter analysis for NFκB binding sites
For each differentially regulated gene identified by microarray analysis, 5,000 bases of the proximal promoter region (-5,000 to -1) were extracted from ENSEMBL database http://www.ensembl.org. Position weight matrix (PWM) models representing the binding sites for NF-κB family members p50 and p65 were taken from version 7.0 of the TRANSFAC database , using matrix IDs 'V$NFKAPPAB50_01' and 'V$NFKAPPAB65_01', respectively. Matches to each PWM were identified in promoter regions using a slightly modified version of tffind, with default matrix similarity thresholds.
Chromatin immunoprecipitation analysis
Chromatin immunoprecipitation (ChIP) reagents were purchased from Upstate Cell Signaling (Billerica, MA, USA). HT29 colon cancer cells (1 × 106) were transfected with empty vector or Par-4 expressing vector and treated with 1% formaldehyde for 15 min at 37°C to crosslink protein-chromatin complexes. The fixed cells were washed twice with cold PBS, and chromatin DNA was harvested and sonicated for ChIP assays. The ChIP assays were performed according to the manufacturer's protocol. Anti-NFκB p50 and p65 antibodies used for ChIP assays were from Abcam (Cambridge, MA, USA) and Santa Cruz (Santa Cruz, CA, USA), respectively. In ChIP experiments, quantitative real-time PCR (qPCR) analysis with SYBR Green PCR Master Mix (Applied Biosystems, Foster City, CA, USA) were calculated by measuring the ratios of ChIP-to-Input, and the non-antibody-treated chromatin immunoprecipitated samples were used as a negative control. NFκB occupancy at IL8 and ACTB promoters served as additional positive and negative controls, respectively, for the ChIP-qPCR experiments. All primers used for quantitative PCR are listed in Supplemental data (Additional file 9).
Immunocytochemistry with confocal fluorescence detection
Empty vector-transfected HT29 cells, Par-4-overexpressing HT29 cells, and native HT29 cells transfected with scrambled or pshAkt1 were grown in 6-well plates on glass coverslips. After culturing for 24 h (empty vector- and Par-4-transfected cells) or 48 h (scrambled- and pshAkt1-transfected cells), cells were fixed with 3.7% formaldehyde in phosphate-buffered saline (1×PBS) for 15 min at room temperature and washed once with 1×PBST (1×PBS with 0.1% Tween-20). Cells were permeabilized with 0.1% Triton X-100 in 1×PBS for 5 min followed by two washes with 1×PBST. The cells were then treated with blocking buffer (5% FBS in 1×PBS) for 1 h, and incubated with the rabbit polyclonal anti-NF-κB p65, Par-4 (Santa Cruz Biotechnology, Santa Cruz, CA, USA) or p105/p50 (Abcam, Cambridge, MA, USA) antibody, or the mouse monoclonal anti-Par-4 or IκB-antibody (Santa Cruz Biotechnology, Santa Cruz, CA, USA) at 1:400 dilution for overnight at 4°C, followed by two washes with 1×PBST. For the secondary antibodies, Alexa Fluor 488 donkey anti-rabbit IgG and Alexa Fluor 594 goat anti-mouse IgG (Molecular Probe, Carlsbad, CA, USA) were used at 1:1000 dilutions and incubated for 1 h at room temperature in the dark, and then washed three times with 1×PBST. Coverslips were mounted onto glass slides with Prolong Gold antifade reagent with DAPI (Molecular Probe, Carlsbad, CA, USA) to detect the cell nuclei. Confocal microscopy was performed with a Zeiss Axioplan fluorescence microscope coupled with a Zeiss LSM710 Laser Scanning System (Zeiss, Berlin, Germany). Colocalization analysis was performed using Volocity 5.0 software (PerkinElmer, Waltham, MA, USA). Positive PDM (product of the difference from the mean) channels were generated to visualize the highly-correlated colocalization of two fluorescence-labeled proteins (Figure 5A-v, B-v and 5B-x). Images were processed with Volocity software (PerkinElmer, Waltham, MA, USA).
Recurrent co-regulated gene network analysis
Recurrent co-regulated (ReCo) gene networks were constructed from 334 colon cancer gene expression microarray studies downloaded from the Expression Project for Oncology repository https://expo.intgen.org/geo/listPublicGeoTransactions.do
and Gene Expression Omnibus (GEO) [73
]. In a separate analysis, we also inspected the ReCo links in prostate-disease networks given that Par-4 is known to have an important role in prostate cancer. Using only data corresponding to a specific tissue and disease state (i.e. colon cancer or prostate cancer), a co-regulated network was constructed by linking gene pairs that were significantly coexpressed across the corresponding samples. Network analysis was restricted to the set of genes initially identified as Par-4-targets containing NFκB binding sites in colon cancer cell lines. A Pearson correlation r
was calculated for all pairs of genes (i, j):
where N is the number of samples, E
is the expression level of gene i in sample s, μ
is the mean expression level of gene i across all samples, and σ
is the standard deviation of the expression level of gene i.
Pearson correlations were then transformed to Z
-scores using the Fisher Z
Intuitively, the Fisher Z transformation assigns higher significance to genes with strong correlations across a greater number of samples. Zij's can be interpreted as Z-scores which provides an estimate of the significance of two gene's correlation across the set of conditions. While the Z-score would be exact if the expression levels across the conditions were independent and normally distributed, which is clearly not true in our case, it still provides a good measure of relative correlation for a single experiment useful for ranking gene pairs against one another.
We calculated a tissue-disease-specific ReCo network by combining the Z scores of gene pairs across individual studies of the same tissue-disease state. We first converted the Z
-scores for each gene pair (i, j
) in each study S
values using the following rank ratio transformation:
is the total number of gene pairs in the dataset. Then, the P
values were combined across studies using the Mudhalker-George's t
-statistic to derive a recurrent co-regulated (RECO)
is the total number of studies. The resulting RECO score gives a single value quantifying the strength of coexpression of each gene pair (i, j) across all available studies of the same tissue-disease state.
microRNA microarray hybridization and data analysis
Total RNA from HT29 cells was harvested in QIAzol Lysis reagent (Qiagen, Valencia, CA, USA), and microRNA isolated using miRNeasy Mini Kit (Qiagen, Valencia, CA, USA) according to the manufacturers' instructions. RNA quality was assessed and quantified using the RNA 6000 Nano Kit (Agilent Technologies, Inc., Santa Clara, CA). microRNeasy-isolated RNA (250 ng) was used as input in the labeling reaction, and the entire reaction was hybridized onto an Agilent human genome microRNA microarray V1 (Agilent Technologies, Inc., Santa Clara, CA) containing 20-40 probes for each of 470 microRNAs for 20 hours at 55°C. Hybridization signal intensities were extracted using the Agilent Feature Extraction software. Raw mean signal, total probe intensities and total gene intensities were uploaded into GeneSpring GX 10.0 software (Agilent Technologies, Santa Clara, CA). Poor spots, as reported in the raw data file, were flagged as A (absent). The background-subtracted signal intensities were log2 transformed and quantile normalized. GeneSpring GX 10.0 and Partek Genomic Suite 6.2 (St. Louis, MO) were used for principal component analysis (PCA), statistical analysis, and hierarchical clustering. Differential expression was assessed using a two sample t-test (double-sided). Corrected P-values were adjusted for multiple testing using Benjamini and Hochberg's FDR at 10% .
Real-time RT-PCR validation of microRNA expression
qRT-PCR was performed using the NCode™ EXPRESS SYBR® GreenER™ microRNA qRT-PCR Kit (Invitrogen, Carlsbad, CA). Poly(A) tailing and RT reactions consisted of 4 μl 5× reaction mix, 2 μl 10× SuperScript enzyme mix, 0.5 mL and 200 ng total RNA in a final volume of 20 μl. Following poly(A) tailing and RT steps, 0.17 μl of the RT product was transferred into a PCR reaction mixture consisting of 10 μl Express SYBR green qPCR SuperMix, 0.4 μl microRNA-specific forward primer (10 μM), 0.4 μl universal qPCR primer (10 μM) in a final volume of 20 μl. PCR cycling began with template denaturation and hot start Taq activation at 95°C for 2 min, then 40 cycles of 95°C for 15 sec, and 60°C for 1 min performed in a 7300 Real-Time PCR System (Applied Biosystems, Foster City, CA). MiR-103 was used as the internal standard reference in the qRT-PCR reaction . Normalized expression was calculated using the comparative Ct method and fold change was derived from the equation 2-ΔΔCt for each microRNA.
Identification of microRNA target genes and pathway analyses
Target Scan database (version 4.2, http://www.targetscan.org) was integrated to GeneSpring GX and used for the identification of target genes. Target Scan allows identification of target mRNAs for any specific microRNA, based on the context score percentile . Differentially expressed microRNAs with 10% FDR were imported into the TargetScan algorithm and the target mRNAs with high context scores (context percentile of 80) were retained for further analysis. Predicted target mRNAs were imported to GeneSpring GX 10.0 and Ingenuity Pathway Analysis programs to identify significant biological pathways, associated network functions, and associated molecular and cellular functions.
Inhibition of miR-34a expression
Cells grown to 60-70% confluence in 6-well plates were transfected with 150 pmole anti-miR-34a inhibitor or negative-control oligonucleotide (Ambion, Austin, TX) using Lipofectamine 2000 according to manufacturer's protocol. The medium was replaced with fresh medium after 24 hours, and cells were allowed to grow for another 48 hours prior to functional analysis.