Fig. 2

EphB6 positive cells are characterized by high transcriptional levels of genes related to cell proliferation and significantly selected in minimal residual disease (MRD) of human T-ALL. (A) Hierarchical clustering of EphB6 positive and EphB6 negative subpopulations using 20 differentially expressed genes from M71 (PDX#1), H3255 (PDX#2) and F1313 (PDX#3), three independent PDX clones. A dual-color code represents markers over-(red) and under-represented (blue). (B) Flow cytometric analysis depicting the enrichment of CCNB1 and CDCA3 in EphB6-overexpressing cells (EphB6/GFP) as compared to control (Empty vector/GFP). (C) Heatmap showing the Spearman’s rank correlation between the EphB6 fractions and 15 differentially expressed genes. The higher the statistical correlation the bigger the circles. A dual-color code reports positive (blue) and negative (red) correlation and color intensity represents the increase of correlation. *P < 0.05, by Student’s t-test. (D) tSNE plots based on the single cell RNA-Sequencing (scRNA-Seq) data from 4 primary T-ALL samples, CSS12401, CSS13693, CSS15501 and CSS20705 at the diagnosis (d0) and up to 30 days after the start of therapy (MRD) without passaging into immunocompromised mice. In the map, EphB6 negative and positive leukemia cells are colored in grey and red, respectively. (E) Graphical representation of cell percentage in the EphB6 positive cell population at the diagnosis (d0) and up to 30 days after the start of therapy (MRD) according to each sample group. (F) mRNA expression level of CDCA3, CCNB1 and KIF20A genes in EphB6 positive vs. EphB6 negative cells as determined by scRNA-seq data. ***, P < 0.001 (Student’s t-test). (G) Higher KIF20A and CCNB1 genes are significantly associated with lower patients’ survival according to the COG TARGET dataset (P = 0.032 and P = 0.029 from log-rank test, n = 264). Kaplan-Meier survival plots shows the same trend for CDCA3 gene although without statistical significance (P = 0.075, n = 264)