Fig. 2

Transcriptional signatures of inflammation and angiogenesis mark the molecular pathology of TSC. a Correspondence analysis of RNA-Seq results from control (blue), TSC2 heterozygous (orange), and homozygous mutant (red) cell lines after six weeks of differentiation. Each dot represents a biological replicate. b Venn diagram showing overlap of genes deregulated in two mouse models of astrocyte activation (blue) [36] with those in TSC2-deficient cells (red). Significance of overlap between gene lists (p values of 4.9E-25 and 5.5E-10) was determined by Fisher’s exact test. c Pathway map for glycolysis enzymes indicating log2 fold changes of transcript levels in TSC2-deficient cells. d Whisker box plot for target gene expression changes of listed upstream regulators/signaling pathways that were found significantly (false discovery rate (FDR) < 1E-4) and consistently over-expressed in TSC2-deficient cells. Vertical lines (median); error bars (1.5 times the interquartile range); (−): negatively regulated targets. e, f Gene Ontology gene set enrichment in in vivo (x-axis) and in vitro (y-axis) samples. Gene set analysis based on Fisher’s exact test was applied to the comparison of cortical tubers versus normal cortex autopsies from controls (Boer et al. [37]) or SEGAs versus periventricular control tissue from non-TSC patients against TSC2-deficient cells versus control cells. Genes that are significantly differentially expressed (cutoff: FDR < 0.05) are tested for over-representation of Gene Ontology biological process terms. The p values are log10 transformed and given a sign according to the direction of gene expression changes in the absence of TSC2. Only significantly enriched terms (p < 0.05) are reported and no extra filtering was applied. All changes that we observed were consistent in both conditions. Selected points are text labeled; numerical values and labels of all points can be found in the accompanying source data