(A) After injection of 70,000 sorted GFP⁺ cells plus 1×10⁶ supporter cells, no significant difference in survival was evident in primary recipient mice of homozygous floxed (HO) or wildtype (WT) BM cells. β-catenin (Ctnnb1) was excised after administration of Tamoxifen (TAM) and Corn Oil served as an ‘empty control’. (B) Blood counts of moribund animals display significant differences between recipients of homozygous (Ctnnb1^fl/fl Cre⁺) (p=.0149*) and wildtype (Ctnnb1^+/+ Cre⁺) controls. (C) BM analysis reveals reduction in GFP⁺ (~3 fold reduction in Ctnnb1^-/- recipients) and GFP⁺Sca-1⁺ cells (~3fold reduction in Ctnnb1^-/- recipient mice). The left panel demonstrates GFP % and the right panel absolute number of GFP⁺ cells. (D) Organ infiltration (predominantly lung infiltration) was the cause of death in the majority of recipient mice without differential infiltration patterns between the groups investigated (displayed are representative HE-stains of each genetic group). Loss of β-catenin did not influence proliferation (Ki67) of the majority of transformed cells in spleen and BM. Error bars indicate standard deviation (SD) of 3 cohorts investigated. Excision control PCR on sorted GFP⁺ bone marrow cells from moribund primary recipient mice is indicated in Supplemental Figure S1.

(A) Treatment schedule of primary recipient mice after transplantation with 40,000 GFP⁺ cells. (B-D) Endpoint analysis of white blood count (WBC), BM and spleen weight have been performed at disease onset in the first relapsed animal after discontinuation of treatment. At this point all remaining animals were sacrificed and analyzed. Disease burden (BCR-ABL transformed GFP⁺ cells) was analyzed as a correlate for relapse (‘cytogenetic or hematologic relapse after treatment discontinuation’) (B) WBCs are significantly reduced in recipients of Ctnnb1^-/- BM, compared to Ctnnb1^+/+ controls. Apc^min mice do not display any differences in WBC or spleen weight compared to Ctnnb1^+/+ animals. (C, D) BM analysis reveals significant reduction of GFP⁺ cells and GFP⁺LSK in the Ctnnb1^-/- recipient cohort. Apc^min mice show similar amounts of GFP⁺ cells as WT controls with a minor reduction in the GFP⁺Lin^- compartment. (Error bars indicate standard deviation of WBCs n>4 per cohort displayed).

(A) Treatment schedule of secondary recipient mice after transplantation with 150,000 GFP⁺Sca-1⁺ cells plus supporter cells. (B) Representative analysis plot of GFP⁺Lin^- and GFP⁺LSK compartments of secondary recipient mice. (C) Genetic deletion of β-catenin (Ctnnb1) leads to a significant decrease of GFP⁺Lin^- cells (p=.0129*) and GFP⁺LSK (p=.005**) in secondary recipient mice when combined with imatinib (IM) treatment. Deletion of β-catenin in combination with IM led to a reduction of GFP⁺Lin^- cells from 72988 in Ctnnb1^+/+ recipient animals to 9231 in Ctnnb1^-/- recipient animals (calculated as percentage of total harvested BM cells). These numbers correlated with the reduction of GFP⁺LSK-cells from 2560 to 194, respectively. Error bars indicate the standard deviation of n=6 mice investigated in the cohort displayed. (D) Transplantation of 1.5×10⁶ whole BM cells leads to a significant survival advantage in tertiary recipients of Ctnnb1^-/- BM when compared to Ctnnb1^+/+ controls. Transplantation of higher cell numbers (6×10⁶ whole BM cells) is displayed in Supplemental Figure S2. Analysis of disease burden at different time points by flow cytometry (E) or nested RT-PCR (F) showed a significant decrease in disease load in Ctnnb1^-/- tertiary recipient mice. These results could be recapitulated using Mx1-Cre (Supplemental Figure S2 panels C-E).

(A) Western blotting on whole cell lysates of K562 cells. Incubation of K562 cells with either IM (1μM) or indomethacin (40μM) led to reduction of β-catenin protein levels, while combined treatment revealed profound decrease of β-catenin. Increase of β-catenin levels were detectable upon GSK3β inhibition using BIO (0.5μM) and partially reversible by co-treatment with IM and indomethacin. (B) Using low cell numbers, these results could be validated using intracellular flow cytometry in K562 cells (left panel) and in GFP⁺Sca1⁺ leukemic cells from primary mice (right panel). (C). When plated in methylcellulose, indomethacin treated GFP⁺Sca1⁺ cells sorted from primary recipient mice (n=3 mice /group, error bars indicate SD) showed significantly impaired colony formation compared to DMSO, BIO or PGE₂ treated cells (p=.0109). Evaluation of β-catenin modulation in CML-LIC derived from recipient mice after drug treatment is shown in Supplemental Figure S3.

(A) Treatment schedule of secondary recipient mice (n ≥ 8 mice/cohort) injected with 150,000 GFP⁺Sca-1⁺ cells out of primary recipient mice. (B) Representative analysis plot of GFP⁺Lin^- and GFP⁺LSK compartments of secondary recipient mice after treatment. (C) Co-treatment of secondary recipient mice with IM and indomethacin leads to reduction of GFP⁺Lin^- cells (p=.0635) and to significant depletion of GFP⁺LSK cells (p=.0159) in the BM. GFP⁺Lin^- cells were reduced from 42,560 (IM only group) to 9,867 (IM+INDO treated cohort). This translated into a significant reduction of GFP⁺LSK cells from 823 to 146, respectively. (Error bars indicate SD of GFP positive cells; n>4 per cohort investigated). Evaluation of secondary recipients after drug combination therapy is displayed in Supplemental Figure S4.

(A-C) Pharmacologic inhibition of PGE₂-signaling in combination with IM treatment leads to delayed onset of disease in tertiary recipient mice. Injection of 1.5×10⁶ whole BM cells leads to delayed onset of disease in tertiary recipients of IM plus indomethacin treated BM cells in comparison to IM only treated controls (A). Endpoint analysis of moribund animals revealed no difference in disease phenotype in the different treatment arms with regard to white blood count, spleen weight (B) or peripheral blood/organ involvement (C). Error bars indicated standard deviation of all mice (n>13) investigated in both cohorts investigated. Gene Expression Profiling of CML patient derived CD34+ cell samples (D, E) reveals a prognostically relevant PGE₂-signature. Expression levels of prostaglandin synthetase 2 correlate with molecular response in the dataset investigated (p=0.0004, McWeeney et al. 2010). Moreover, sample clustering based on the “REACTOME_PROSTANOID_HORMONES” MSigDB gene set correlates also with response/non-response in this dataset (p=0.0067) (E). Figure S5 shows survival of tertiary recipient mice after combination of imatinib and BIO in secondary donor animals.