Fig. 8
HOXB genes confer growth advantage in human T-ALL initiation and maintenance. a HOXB mRNA expression level in 17 CB leukemias and 22 PDX samples. Rlog values were calculated from RNA-seq data using DESeq2 and corrected for batch effects using ComBat. Each datapoint represents an individual sample with mean ± SD indicated. b Volcano plot of HOX shRNA depletion/enrichment. Primary CB leukemia cells were transduced with a 59-plex lentiviral shRNA pool targeting HOXA/B genes plus controls and cultured on OP9-DL1 feeders (two independent replicates; Spearman correlation 0.58). Genomic DNA was prepared from FACS-sorted cells on days 2 (t0) and 9/11 (t1) post transduction. Proviral shRNAs were PCR amplified and enumerated by NGS. Each datapoint represents a different shRNA species. Genes targeted by at least two different shRNA species with FDR < 0.05 are indicated in color. c, d Flow cytometric tracking of shRNA-transduced cell fraction over time in culture. c Primary NLTB CB leukemia cells and d human T-ALL cell lines. Mean ± SD fraction of the initial transduction value are plotted for experiments performed at least in triplicate. Decreasing NGFR+ or GFP+ fraction over time indicates shRNA-associated growth disadvantage. e, f Limiting dilution growth assays. CB cells were transduced with N + LTB viruses on days 0 and 4 and cultured until day 19, then transduced with shHOXB3 or control (shScr) lentiviruses and cultured until day 25. Triply transduced cells (N/GFP+, LTB/Cherry+, shRNA/NGFR+) were FACS sorted into individual wells of a 96-well plate containing OP9-DL1 feeders and cultured for ~3 weeks. The entire contents of each well were then harvested with trypsin and assayed by flow cytometry. e Total yields of viable, triply transduced cells per well (n = 12–16 for 50 and 100 cells per well; n = 24–32 for 25, 10, 3, and 1 cell(s) per well). Dotted line indicates threshold yield of 500 cells used for calculating well-initiating cell (WIC) frequencies. Box and whiskers are as defined by Tukey. ****p < 0.0001; ***p < 0.001; ns not significant (Kruskal−Wallis test with Dunn’s correction for multiple comparisons). f WIC frequencies according to single-hit/Poisson model. Dotted lines indicate 95% CI. ****p < 0.0001 (chi-square test)
