Extended Data Fig. 1: Specification of chemoresistant Mex3a+ cells in MTOs.

a, Quantification of organoid growth of APT MTOs (Apc^mut, p53^mut, Tgfbr2^mut) in absence of EGF compared to complete stem cell medium (SC med.). Dashed lines indicate the day EGF was re-added. n = 4 technical replicates (independent tissue culture wells), linear model on log-transformed values with no adjustment for multiple comparisons. Scale bars: 50 µm. b, Viability (ATP measurements) of APT MTOs grown in complete SC medium versus medium lacking EGF upon treatment with increasing concentrations of SN38. Trend line was calculated using LOESS model, n = 3 technical replicates (independent tissue culture wells). c, Heat map summarizing the effect on organoid growth under suboptimal niche conditions (absence of EGF or presence of TGFβ1) vs growth in SC complete medium (with EGF and no TGF-β1) in MTOs with the indicated genotypes. d, Heat map summarizing the increase in SN38 IC50 values promoted by absence of EGF or presence of TGFβ1 in MTOs with different genotypes. Color scale shows normalized IC50s compared to IC50s in SC medium. e, Viability (ATP quantification) of APS, AKP and AKP + S MTOs grown in complete SC medium versus no EGF + TGF-β1 medium upon treatment with increasing concentrations of 5FU, Oxaliplatin and SN38. Trend line was calculated using LOESS model, n = 3 technical replicates (independent tissue culture wells). f. Scheme and viability of APS organoids in increasing concentrations of SN38 in complete or medium with no EGF in latent-recovered organoids. LOESS model, n = 3 replicates. g, Relevant GSEAs obtained after RNA sequencing analysis of APT MTOs grown in SC medium versus medium with no EGF respectively. Profiling was performed at day 7. Left panels indicate signatures related to functional pathways (Function); right panels indicate GSEAs related to cell identity of different cell populations (Identity). Gene sets are ordered based on their -log10-pvalue, dashed line indicates p-value = 0.05, calculated with rotation gene set testing for linear models. Signatures are detailed in Supplementary Table 6. LRC, label-retaining cell. h, Mex3a mRNA expression levels in APT MTOs grown in the absence of EGF vs control SC medium (n = 4 independent experiments). Boxes represent the first and third quartiles. The line within the box indicates adjusted values of Mex3a mRNA corrected by experimental batch. Whiskers indicate maximum and minimum values. Linear model with no adjustment for multiple comparisons. Data in panels a, b, e and f are presented as mean values +/− SEM. Experiments in a, b, e and f were conducted 3 times with equivalent results.