Fig 2: SOX11 regulates growth of ER- breast cancer cells.(A) qRT-PCR results for several potential SOX11 targets in CAL-148 cells transduced with shRNA to SOX11 or shRNA NS cells with and without DOX induction in cells grown in 2D. (B) Western blot of SOX11, MEX3A and TUBB3 in total cell lysates of CAL-148 cells transduced with shRNA SOX11 in presence or absence of 1 µM DOX after 48 hr. GAPDH was used as loading control. Densitometry results normalised against no DOX are shown in brackets. (C) Cell growth assay results for CAL-148 shRNA SOX11 cells induced with 1 µM DOX at 3 and 7 days. Experiments were performed three times. Error bars represent SEM. *p=0.0106 (day 7). (D) Quantification of clonogenicity in 2D and 3D from single CAL-148 shRNA SOX11 cells plated in presence or absence of DOX after 21 days. The number in brackets represents the mean in each group of the three experimental replicates. DOX: doxycycline, NS: non-silencing.

Fig 3: Inducible expression of SOX11 leads to changes in cell state profiles of DCIS.com cells.(A) Western blot of SOX11 in cytoplasmic and nuclear fractions of DCIS.com cells containing the pInducer21 empty vector in presence (iEV) or absence (niEV) of 1 µM Doxycycline (DOX) or the pInducer21SOX11 with (iSOX11) or without DOX (niSOX11). GAPDH and LAMIN B1 were used as loading control of cytoplasmic and nuclear fractions, respectively. Densitometry results normalised against niSOX11 are shown in brackets. (B) SOX11 expression detected in iSOX11 cells stained by IF after 48 hr of DOX induction. Scale Bar: 200 µm. (C) ER- DCIS case sample showing SOX11 staining in DCIS and adjacent normal breast tissue. Scale Bar: 200 µm. (D) Results from flow cytometry analysis of Aldefluor assays of niEV and niSOX11 cells (day 0) and iEV and iSOX11 after 2 days treatment with 1 µM DOX. Results show the % of ALDH+ cells normalised against niEV. Error bars represent SD. *p = 0.0223. n = 5. (E) Results from flow cytometry analysis of CD24 and CD44 of niEV and niSOX11 cells (day 0) and iEV and iSOX11 after treatment with 1 µM DOX for 2 days. Results show the average % of cells CD44+/CD24+ in each condition. Error bars represent SD. ***p = 0.0005 (iSOX11 vs niSOX11) and p = 0.0009 (iSOX11 vs iEV) n = 3. (F) Western blot of CD24 in cytoplasmic and nuclear fractions of niEV, niSOX11, iEV and iSOX11 cells. GAPDH and LAMIN B1 were used as loading control of cytoplasmic and nuclear fractions, respectively. In brackets densitometry results normalised against niSOX11. (G) Confocal images of IF staining of E-CADHERIN, N-CADHERIN, VIMENTIN in niSOX11 and iSOX11 cells. Cells were grown in slides covered with Collagen I. Scale Bar: 200 µm. (H) Western blot of N-CADHERIN and VIMENTIN in cytoplasmic and nuclear fractions of niEV, niSOX11, iEV and iSOX11 cells. GAPDH and LAMIN B1 were used as loading control of cytoplasmic and nuclear fractions, respectively. Densitometry results normalised against niSOX11 are shown in brackets. (I) Confocal IF images of iEV and iSOX11 spheroids treated with 1 µM DOX, stained with luminal marker, K8, and basal markers K14 or SMA, and DAPI. Scale bar: 200 µm. DOX: doxycycline, IF: Immunofluorescence.

Fig 4: SOX11 expression promotes expression of developmental pathways frequently activated in cancer.(A) Volcano plot representing the RNAs with a log2 fold-change > +/- 0.585 in the RNA-sequencing results of iSOX11 cells grown in 2D compared with the controls [(iSOX11-niSOX11)- (iEV-niEV)] to account for effects of DOX treatment on DCIS.com cells. (B) Gene ontology results from A. (C) List of genes overexpressed log2 fold-change >+/-1.585 times in all three RNA-sequencing (cells grown in: 2D, 3D for 2 days, 3D for 5 days) results comparing iSOX11 versus iEV. (D) qRT-PCR results for several potential SOX11 targets in EV and SOX11 cells with and without DOX induction in cells grown in 2D. Experiment was repeated three times. (E) Western blot of MEX3A and TUBB3 in cytoplasmic and nuclear fractions of EV or SOX11 cells in presence or absence of 1 µM DOX. GAPDH and LAMIN B1 were used as loading control of cytoplasmic and nuclear fractions respectively. In brackets, densitometry results normalised against niEV and niSOX11. (F) IF staining of DCIS iEV and DCIS iSOX11 cells with TUBB3 (green) and MEX3A (red). Scale: 200 µm. (G) Western blot of MEX3A and TUBB3 in SOX11+ breast cancer cell lines and SOX11- DCIS.com and MCF10A from the MCF10A mammary cell progression series. (H) Pie charts representing the percentage of breast cancer samples with a log2 fold-change greater than two in the levels of MEX3A or TUBB3 RNA when SOX11 increased between 0.5- and 2-fold, 2- and 4-fold, or greater than 4-fold in the TCGA dataset. DOX: doxycycline, qRT-PCR: Quantitative real time PCR, IF: Immunofluorescence.

Fig 5: MEX3A, a novel SOX11 downstream effector, regulates cell growth and (E/M) state of ER- breast cancer cells.(A) Invasion assay results after overlaying BT-20 and HCC1187 spheroids with Collagen I, at day two after siRNA transfection specific for MEX3A or control (NS). Graph shows the area invaded in pixel2 normalised against NS siRNA. *p = 0.0181, **p = 0.0014 for BT-20 and *p = 0.0220 for HCC1187 cells. (B) Representative images of BT-20 spheroids 48 hr after adding Collagen I and at time 0 hr (shown in small inset, 2 days after transfection) made from cells transfected with either control siRNA or MEX3A siRNAs. Scale bar: 200 µm. (C) Cell growth assays 5 days after CAL-148 and HCC1187 cells were transfected with siRNA specific for MEX3A or NS controls (siRNAs require 48 hr for efficient knockdown; this was taken into account to select day five as final point). Relative results to NS siRNA are shown. Experiments performed three times. *p = 0.0342, **p = 0.0052 ***p = 0.0005 (siRNA2), ****p < 0.0003 (siRNA3) for CAL-148 cells and *p=0.0337 (siRNApool) and *p = 0.0477 for HCC1187 cells. (D) Sphere size measured 3, and 5 days after HCC1187 cells were transfected with siRNA specific for MEX3A. Relative results to NS siRNA are shown. Experiments performed three times *p = 0.0240. (E) Cell cycle analysis performed by flow cytometry at day 4 after siRNA transfection specific for MEX3A in CAL-148 cells. Graph shows % of cell in each phase of the cell cycle. *p = 0.0147 (phase S siRNApool vs siRNA NS) and *p = 0.0094 (phase S siRNA3 vs siRNA NS) and *p = 0.0419 (phase subG1 NS vs siRNA3). Experiments performed four times. (F) Western blot of RRM2 and SURVIVIN in total cell lysates of CAL-148 at day four after siRNA transfection specific for MEX3A. ß-ACTIN was used as loading control. Densitometry results normalised against NS siRNA are shown in brackets. (G) Examples of morphological changes observed in CAL-148 cells after 4 days of MEX3A knockdown compared to NS control. Scale bar: 200 µm. (H) Western Blot of EPCAM and E-CADHERIN in CAL-148 cells transfected with siRNA specific for MEX3A or NS control. ß-ACTIN was used as loading control. Densitometry results normalised against NS siRNA are in brackets. NS: non-silencing.