Figure S1. Inhibition of TORC1/2 Promotes a CSC-Like Phenotype in TNBC Cells

SUPPLEMENTARY FIGURE LEGENDS

Figure S1. Inhibition of TORC1/2 promotes a CSC-like phenotype in TNBC cells.

(A) SUM159 and BT549 cells were treated with 1 µM of the PI3K inhibitor BKM120 for 72 h. FACS analysis for the CSC markers ALDH and CD44hi/PROCR+ was performed in SUM159 and BT549 cells, respectively (*P<0.02). (B)SUM159 cells were treated with 25 nM of Cell Tracker Green dye for 45 min and split into 3 plates, each treated with vehicle, 250 nM BEZ235 or 100 nM MLN128. FACS analysis for FITC was performed at Day1 and Day10. Each bar represents the percentage of FITC-positive cells at Day 10 (*P<0.01). Error bars represent mean ± SEM.

Figure S2. TORC1/2 inhibition increases expression of Notch1, FGF1 and Notch1 activity.

(A) SUM159 cells were treated with vehicle or 250 nM BEZ235 for 48 h. mRNA was collected and used in a Stem Cell-specific PCR Array. Fold regulation of gene expression in MLN128-treated cells vs untreated control cells was determined using SA Biosciences PCR Array Software.(B) SUM159 and (C) BT549 cells were treated with 250 nM BEZ235 or 100 nM MLN128 for 3 h or 48 h. Q-RTPCR was performed for Notch1 and FGF1 (n=2). (D) SUM159, BT549, MDA231 and (E) CAL51, CAL120 and MDA468 cells were treated with 250 nM BEZ235 or 100 nM MLN128 for 72 h as indicated qPCR was performed for FGF1 (*P<0.03; **P=0.004). (F) BT549 cells transfected with the RBP-Jk luciferase reporter and the CMV-renilla construct were treated with 250nM BEZ235 and 100nM MLN128 for 17 h. Cells were treated with 5mM EDTA for 30 min (before lysis) as a positive control for Notch1 activation. Dual luciferase reporter assay was performed as described in Materials and Methods (*P<0.01). (G) SUM159 cells transfected with the 4XCSL-luciferase reporter and the CMV-renilla construct were treated with 250nM BEZ235 and 100nM MLN128 for 48 h (*P<0.025).BT549 and MDA468 cells were transfected with the RBP-Jk luciferase reporter and CMV-renilla constructs and treated with 250nM BEZ235 and 100nM MLN128 for 48 h. Dual luciferase reporter assay was performed as described in Materials and Methods (*P<0.03). Error bars represent mean ± SEM.

Figure S3. TORC1/2 inhibition and paclitaxel sustains Notch1 activity and a CSC-like phenotype.

(A)Nanostring analysis of Notch1 (P=0.0036) and JAG1 (P=0.0021) in mRNA extracted from paired pre and post-chemotherapy (Tx) breast cancer biopsies (n=17). (B) SUM159 cells were treated with vehicle or 10 nM paclitaxel (TAX) for 72 h. Lysates were collected and analyzed for NICD, C-MYC, Actin and phospho-S6 expression by immunoblot analysis. (C) SUM159 cells were treated with 250 nM BEZ235, 100 nM MLN128, 10 nM paclitaxel or combinations of each BEZ235 and MLN128 with paclitaxel. After 72 h, Alamar Blue assay was performed to determine cell viability (*P<0.001). (D) SUM159 cells were treated with 250 nM BEZ235, 100 nM MLN128, 10 nM paclitaxel or combinations described in (C). After 72 h, cells were allowed to recover for 24 h; viable cells were seeded as mammospheres and mammosphere number was determined 7 days later using Gelcount. (E) Similar to D, the ALDEFLUOR assay was performed (*P< 0.05). (F) SUM159 cells were treated with MLN128, paclitaxel (TAX), GSI-IX or combinations of MLN with paclitaxel and GSI-IX. Lysates were prepared and resolved by SDS-PAGE followed by immunoblot analysis for NICD, phospho-S6 and Actin. Error bars represent mean ± SEM.

Figure S4. Inhibiting Notch1 decreases the CSC-like populationinduced by inhibition of TORC1/2.

(A) SUM159 MLNR cells were transfected with Control and Notch1 siRNA. After 72 h, FACS analysis for ALDH activity was performed. (B) SUM159 cells were transfected with 2 different Jagged1 (JAG1) siRNAs and treated in the presence or absence of 100 nM MLN128. After 72 h, immunoblot analysis of cell lysated was performed or single cells were seeded as mammospheres. Mammospshere number was determined after 5 days using GelCount and GelCount software (*P=0.01). (C) BT549 cells were treated with 100 nM MLN128 ± 5 µM GSI-IX for 72 h. After 72 h, FACS analysis was performed for the indicated CSC markers as described in Materials and Methods (*P<0.01). (D) Mice with MDA468 xenografts were divided into 4 treatment groups, vehicle (n=4), MLN128 (1mg/kg x3/week, p.o.; n=8), GSI-IX (10 mg/kg, 3 days on, 4 days off, i.p; n=8) MLN128 + GSI-IX (n=8). Tumors were measured with calipers twice weekly. Individual and mean ± SEM tumor volumes in mm3 on day 21 of treatment are represented in the figure. (E) In vivo limiting dilution assay to determine tumor-initiating capacity of MDA468 cells from xenografts that had been treated with vehicle, MLN128, GSI-IX and MLN128 + GSI-IX. P-values apply to pairwise statistical analysis of treatment groups. Error bars represent mean ± SEM.

Figure S5. TORC1/2 inhibition induces mitochondrial metabolism.

(A) SUM159 cells were treated with vehicle or 100 nM MLN128 for 48 h. Cells were collected and prepared for Transmission Electron microscopy imaging. Inset represents a section of a MLN128-treated cell displaying the location of mitochondria (blue triangles) and lipid droplets (red triangles). Magnification: 11,000x. (B) BT549 cells treated with 100 nM MLN128 for 48 h were analyzed for changes in mitochondrial metabolism genes using the Mitochondrial Metabolism PCR Array. Red bars indicate mitochondrial polycistronic transcripts. (C) SUM159 and BT549 cells treated with MLN128 for 3 h were analyzed for mitochondrial metabolism gene alterations using the Mitochondrial Metabolism PCR Array. (D) SUM159 cells were transfected with control or TFAM siRNA and treated with or without 250 nM BEZ235 or 100 nM MLN128. After 72 h, Alamar Blue assay was performed to determine percent cell viability (*P=0.004; **P=0.009). (E) BT549 cells were transfected with control vector or human NICD followed by transfection with control or TFAM siRNA. Cells were then treated with 100 nM MLN128 for 72 h. FACS analysis for the CD44hi population was performed (*P< 0.001). GFP expression to confirm transfection of hNICD-GFP construct is displayed to the right. Error bars represent mean ± SEM.

Figure S6. Induction of NICD and CSCs upon TORC1/2 inhibition is FGFR-dependent.

(A) MDA468 cells were treated with 100 nM MLN128 ± 2 µM lucitanib) for 72 h.

Immunoblot analysis for NICD, phospho-S6 and Actin was performed.

(B) SUM159 cells were transfected with control or FRS2 siRNA±100nM MLN128 for 72 h. Immunoblot analysis for NICD, TFAM, phospho-S6 and Actin was performed.