Marchionni Lab Surfing the Genome

Analysis of the genomic response of human prostate cancer cells to histone deacetylase inhibitors (Supplementary data)

Original manuscript

Analysis of the genomic response of human prostate cancer cells to histone deacetylase inhibitors.
Analysis of the genomic response of human prostate cancer cells to histone deacetylase inhibitors.
Madeleine S Q Kortenhorst, Michel D Wissing, Ronald Rodríguez, Sushant K Kachhap, Judith J M Jans, …, Paul O Aiyetan, Elsken van der Wall, Michael A Carducci, Paul J Van Diest, Luigi Marchionni
Epigenetics  ·  19 Jul 2013  ·  pubmed:23880963
prostate-cancer epigenetics combination-therapy gene-expression

Supplementary Methods

In the following “Supplementary Methods” section we extensively describe all methods and data sources used in our study. In particular we provide details about the source of gene expression datasets, the platforms and annotation packages employed, the pre-processing procedures adopted for each dataset, the statistical details of the analysis performed, and the gene set databases used for the Analysis of Functional Annotation (AFA).

The “Supplementary Methods” section can be downloaded here:

Download “Supplementary material”

Supplementary Figures

Gene membership for the differentially expressed Functional Gene Set (FGS) associated with HDACI-treatment.

Heatmaps visualizing gene membership to FGS enriched in DU-145 and PC3 cells upon HDACI treatment. Rows in the heatmaps correspond to described in Figure 1 of the main paper, while columns represent the most differentially expressed genes (FDR < 0.1%) annotated to such FGS. In the heatmaps gene membership to specific FGS is highlighted in white, while black is used when genes do not belong to the FGS. Hierarchical clustering of rows and columns was obtained using the binary distance and the Ward clustering method.

Social network analysis of enriched FGS upon HDACI-treatment.

Weighted undirected networks based on differentially regulated genes in common among the enriched FGS from Figure 1 of the main paper (see also Supplementary Figures S11, S12 and S13 above). In these networks vertexes represent specific FGS, while edges (and their weights) are based on the number of differentially expressed genes in common among the FGS. Social network analysis was used to identify distinct FGS ”communities” (i.e. subgraphs of FGS sharing common subsets of genes), and are shown in the figures with distinct colors.

Heatmaps showing samples gene expression correlation upon HDACI treatment versus control.

The color code used for the columns highlights the different inhibitors that were used: VPA in yellow, SAHA in red, Buphenyl in blue, TSA in green, CG1521 in purple, and miR transfection in cyan. The color code used for the rows highlights the different cell lines that were analyzed: DU-145 in red, PC3 in blue, and LNCaP in green. Correlations were computed between t-statistics and log2 fold-change (Supplementary Figures 1 and 2 respectively), and they express the degree of differential gene expression between treatment and control.

Correspondence At the Top (CAT) curves for differential gene expression in HDACi-treatment.

Correspondence At the Top (CAT) curves for differential gene expression in HDACi-treatment. Genes were ranked in decreasing (or increasing) order by t-statistics (or log2 fold-change) expressing differential gene expression between treated and untreated cells. Each CAT curve represents the proportion of differentially expressed genes that are in common between comparisons from distinct studies. In particular results for DU-145 or PC3 cells from GSE34452 are compared to results from other studies (GSE8645, GSE31620, and Connectivity Map). CAT curves in the white area above the gray shading indicate agreement, while CAT curves below it indicate disagreement between experiments. The grey shading represents the 99% probability intervals of agreement by chance.

Heatmaps showing enriched FGS after HDACI-treatment across studies.

Heatmap visualizing FGS enriched in DU-145 and PC3 cells upon HDACI treatment (GSE34452), along with the level of enrichment for similar experiments in other studies (GSE8645, GSE31620, and Connectivity Map). These heatmaps correspond to Figures 1A, 1B, and 1C in the main paper. Each row represents a distinct FGS, while each column represents a distinct comparison from our linear model analysis. Color scales representing the enrichment correspond to absolute log10 adjusted p-values (i.e. the number on the color scale increases with decreasing adjusted p-values). Enriched FGS were selected from different collections in order to encompass distinct and complementary biological concepts, as shown by the color bar on the left of each heatmap. Cell signaling FGS are highlighted in red and yellow (Pathway Commons Reactome and NCI pathways, respectively), signaling pathway target gene sets in green (Human Protein Reference Database, HPRD), protein-protein-interaction networks in cyan (PPI, as compiled in the NCBI Entrez Gene database), FGS for shared transcriptional factor binding sites (TFBS) in blue, and microRNA (MIR) targets gene sets in pink (both from the Broad Institute Molecular Signature Database collections).

Flow cytometry graphs after treatment of PC3 or DU-145 cells.

Treatments included combinations of Colcemid with HDACIs, or HDACIs alone, across several time points. PC3 cells were treated for variable periods with either 9 μM SAHA alone, or a combination of 0.1 μg/ml Colcemid with 4 μM SAHA or 4 mM VPA. DU-145 cells were treated for variable periods with either 9 μM SAHA alone or a combination of 0.04 μg/ml Colcemid with 4 μM SAHA or 4 mM VPA. Every 2-3 hours cells were harvested and stained with propidium iodide and flow cytometry analyses were performed. Combining HDAC-inhibitors with Colcemid resulted in mitotic accumulation of PC3 cells.

Differential gene expression analysis

Nore that columns in the HTML files can be sorted by clicking the column header.

Gene annotation table headers

Analysis os Functional Annotation (AFA), Top 200 FGS, FDR < 5%

Columns in the HTML files can be sorted by clicking the column header.

Functional Gene Set (FGS) tables headers

Heatmaps for the Analysis of Functional Annotation

Figure Legends: In all heatmaps the color scale represents the negative base 10 logarithm of the q-values obtained after adjusting the Wilcoxon rank-sum test P-values using with the Benjamini-Hochberg method. Functional Gene Sets are represented on the rows and were clustered using the Euclidian distance and the average clustering method, while the comparisons for the different HDACIs treatments were represented by column and were not reordered.