Fig 2: EREG rescues the growth defects in Sod1 deficient organoids.A Western blot analysis of EREG and SOD1 expression using the separated stromal compartment from Sod1f/f and Sod1f/f; Vil-creERT2 mice after 24 h post 5 consecutive tamoxifen injections. n = 4 mice per genotype. B Western blot analysis of SOD1 expression in Sod1f/f and Sod1f/f; Vil-creERT2 organoids after 5 days of tamoxifen with or without 0.5 μg ml-1 of recombinant EGF/AREG/EREG induction. n = 3 mice per genotype. C Representative images of organoids after 5 days of tamoxifen with or without EGF/AREG/EREG induction. n = 6 wells per group; n = 3 mice per genotype. Red arrowhead indicates dead organoids. D Percentage of organoids with 0, 1, 2, 3, or ≥4 crypts formed after 5 days of tamoxifen with or without EGF/AREG/EREG induction. Data represent mean ± SEM; n = 6 wells per group; n = 3 mice per genotyping. Statistical significances were tested by two-way ANOVA. *P < 0.05, ***P < 0.001. Number of organoids counted: n = 249 (Sod1f/f), n = 287 (Sod1f/f; Vil-creERT2), n = 307 (Sod1f/f; Vil-creERT2 + EGF), n = 309 (Sod1f/f; Vil-creERT2 + AREG), n = 272 (Sod1f/f; Vil-creERT2 + EREG) organoids per group; one of three experiments. E Percentage of dead organoids assayed after 5 days of tamoxifen with or without EGF/AREG/EREG induction. Data represent mean ± SEM; n = 6 wells per group; n = 3 mice per genotyping. Statistical significance was tested by one-way ANOVA. ***P < 0.001. Number of organoids counted: n = 393 (Sod1f/f), n = 475 (Sod1f/f; Vil-creERT2), n = 537 (Sod1f/f; Vil-creERT2 + EGF), n = 532 (Sod1f/f; Vil-creERT2 + AREG), n = 413 (Sod1f/f; Vil-creERT2 + EREG) organoids per group; one of three experiments. All results are representative of at least three independent experiments.

Fig 3: Myoepithelial layer has fewer cells and is thinner in thickness in AREG−/− mice. a Representative images of 12-week-old AREG+/+ (left) and AREG−/− (right) mammary ducts immunostained with K8 (red) and K14 (green) in merged channel. Arrow indicates discontinuous myoepithelial layer. Scale bar shows 50 μm. At both 6 weeks (b, c) and 12 weeks (d, e), AREG−/− glands have thinner myoepithelial layer (b, d) and smaller percentage of K14+ cells (c, e). At least three animals used in each analysis. f Proportion of K14+ cells in ducts and mature duct termini of 12-week-old AREG+/+ mammary ducts compared (N = 7). g PyMT expression in AT-3 cells suppressed by myoepithelial cells when cocultured. AT-3 cells either cultured alone or with primary myoepithelial cells overnight. RNA extracted and PyMT expression assessed by RT-qPCR. h AT-3 cells cultured without addition of growth factors (control), with 100 ng/ml AREG, 10 ng/ml EGF, or 10 ng/ml bFGF, or with both EGF and bFGF. Statistical analysis performed using t test. *p < 0.05, **p < 0.01, ***p < 0.001. AREG amphiregulin, bFGF basic fibroblast growth factor, EGF epidermal growth factor, PYMT polyoma middle-T antigen

Fig 4: AREG−/− tumors are less necrotic and tumor cells in papillary regions are less proliferative. a Representative images of H&E (left column) and Ki67 (right column) staining in AREG+/+ PyMT and AREG−/− PyMT 1-cm tumors. Scale bar shows 500 μm for H&E stains and Ki67 stains. b Percentage of necrotic areas calculated as average of five fields per AREG+/+ (N = 32) and AREG−/− (N = 22) 1-cm tumors. c, d Ki67+ proliferating cells in solid vs papillary areas in 1-cm AREG+/+ PyMT and AREG−/− PyMT tumors compared using HistoQuant. Evaluation performed on at least five separate areas from at least three different tumors per genotype. Statistical analysis performed using t test. *p < 0.05, ***p < 0.001, ****p < 0.0001. AREG amphiregulin, H&E hematoxylin and eosin, PYMT polyoma middle-T antigen

Fig 5: Working model to explain increased tumor initiation and morphological changes in AREG−/− tumors. (a) In mammary duct in absence of PyMT (top), myoepithelial cells (green) form continuous layer around luminal epithelial cells. In mature duct termini, myoepithelial layer is discontinuous. Luminal epithelial cells secrete AREG that binds to EGFR on stromal cells (yellow). Stimulated stromal cells produce FGFR ligands that bind to FGFR luminal epithelial cells and myoepithelial cells [31, 56]. In absence of AREG (bottom), EGFR/FGFR paracrine loop is interrupted and impairs proper mammary ductal development. (b) In AREG+/+ PyMT animals (top), PyMT initiates transformation of luminal epithelial cells in mature duct termini where there are fewer myoepithelial cells. Myoepithelial cells as well as secreted growth factors such as AREG and bFGF suppress PyMT expression in mammary duct. In AREG−/− PyMT mice (bottom), PyMT expression is more widespread. Due to global reduction in myoepithelial cells and reduced AREG and FGF expression, oncogenic transformation takes place more broadly in ductal tree. AREG amphiregulin, FGF basic fibroblast growth factor, PYMT polyoma middle-T antigen