Fig 2: ZEB2 promotes the malignant phenotype of glioma by downregulating miR-637 in vitro(A) Quantitative real-time PCR assays demonstrated increased miR-637 levels under siZEB2 transfection. (B) The cell viability of U251 or U87 cells transfected with a ZEB2 expression plasmid and/or miR-637 lentiviral expression particles was evaluated by the MTT assay. (C–H) The proliferative rate, migration, and invasion of U251 or U87 cells transfected with a ZEB2 expression plasmid and/or miR-637 mimics were evaluated by the EdU incorporation assay (C and D), Transwell assay (E and F), and Boyden assay (G and H), respectively. Scale bar indicates 100 μm. Data are presented as mean ± SD (∗p < 0.05, ∗∗p < 0.01, and ∗∗∗p < 0.001).

Fig 3: Human enhancer design by in silico evolution.a, Distribution of GC content in GC-adjusted random sequences (green) and human genomic regions (red). b, Average number of motif hits at each mutational step compared to genomic enhancers. c, Delta number of motifs in each mutational step. The TF-Modisco patterns and the most similar position weight matrices from the cisTarget motif database are shown at the top of each plot. ZEB2 upside-down pattern is contributing negatively to the model’s prediction and is destroyed by the model on each step. d, Bar plot showing the mean luciferase signal (log2 fold-change over Renilla) in a MES melanoma line (MM047) of the synthetic MEL enhancers (generated by in silico sequence evolution), showing no activity compared to positive control genomic MES enhancers. The bar shows the mean (n = 2 biological replicates). e, MM001 (left) and MM099 (right) ATAC-seq profiles of all integrated lentiviral EFS reporters. f, MM001 ATAC-seq profile of 3 integrated EFS reporters: initial (top), evolved (middle) and post-evolution with repressive sites (bottom). g, DeepMEL2 prediction score (left), luciferase activity levels in MM001 (middle) and correlation between prediction score and activity (right) for EFS-1 (top) and EFS-8 (bottom) sequences after incremental mutation steps. In e, f, red dashed lines indicate boundaries of the enhancer. In g, the error bars show the standard error of the mean (n = 3 biological replicates).

Fig 4: Human enhancer design.a,b, Prediction score distribution (MEL class, n = 4,000 sequences (a) and ten selected sequences (b)) after each mutation. c, Nucleotide contribution scores of a synthetic sequence pre (top) and post (bottom) 15 mutations, with binding site names, mutation positions (dashed lines) and orders (between top and bottom plots). d, Mean luciferase signal (log2 fold-change (FC) over Renilla) of synthetic sequences from in silico sequence evolution and genomic enhancers. e, MM001 ATAC-seq profile of three integrated EFS reporters: initial, evolved and evolved with repressor sites. Red lines, enhancer boundaries. f, DeepMEL2 prediction scores (left), luciferase activity (middle) and their correlation (right) for EFS-4 sequences. g, MM001 ATAC-seq, SOX10 and ZEB2 ChIP–seq tracks for IRF4 gene; enhancer location in red. h, ZEB2 ChIP–seq signal (x axis), SOX10 ChIP–seq signal (y axis) and ATAC-seq signal (colour) for top ZEB2 regions in MM001. i, In vitro and in silico saturation mutagenesis values of the IRF4 enhancer. j, Enformer predictions for EFS-4 sequences replacing IRF4 enhancer: initial score and score changes postmutations. Ref. pred., reference predictions; Alt. - ref. pred., delta alternative versus reference predictions. k, Enformer predictions per mutation step and after repressor addition for MEL EFS sequences. l, Prediction scores for top 50 DNase tracks for EFS-4 sequences. Four first tracks are foreskin melanocyte male newborn and SK-MEL-5 tracks. m, ChromBPNet ATAC MM001 (MEL) and MM047 (MES) prediction scores for EFS sequences, across mutations and postrepressor addition. n, Prediction score distribution for MEL class (n = 2,000 sequences) after motif implantation. o, Relative TF locations distribution (n = 2,000). p, Luciferase signal (log2 FC over Renilla) comparison of motif-implanted sequences and genomic enhancers. In a,n, box plots show the median (centre line), interquartile range (box limits) and 5th and 95th percentile range (whiskers). Error bars in d,f,p denote mean standard error (n = 3 biological replicates). S, SOX10; M, MITF; T, TFAP2.

Fig 5: Intermediate steps of in silico evolution and generation of repressor sites in human generated enhancers.Nucleotide contribution scores of EFS-4 at different mutational steps; 0 (random sequence), 3, 4, 7, 8, 12, 15, 15+Repressors. ZEB2 motif annotation is indicated with strong (s) or weak (w) motif instances.