Combination of ERK and autophagy inhibition as a treatment approach for pancreatic cancer

Abstract

Pancreatic ductal adenocarcinoma (PDAC) is characterized by KRAS- and autophagy-dependent tumorigenic growth, but the role of KRAS in supporting autophagy has not been established. We show that, to our surprise, suppression of KRAS increased autophagic flux, as did pharmacological inhibition of its effector ERK MAPK. Furthermore, we demonstrate that either KRAS suppression or ERK inhibition decreased both glycolytic and mitochondrial functions. We speculated that ERK inhibition might thus enhance PDAC dependence on autophagy, in part by impairing other KRAS- or ERK-driven metabolic processes. Accordingly, we found that the autophagy inhibitor chloroquine and genetic or pharmacologic inhibition of specific autophagy regulators synergistically enhanced the ability of ERK inhibitors to mediate antitumor activity in KRAS-driven PDAC. We conclude that combinations of pharmacologic inhibitors that concurrently block both ERK MAPK and autophagic processes that are upregulated in response to ERK inhibition may be effective treatments for PDAC.

Conflict of interest statement

Competing interests

C.J.D. is on the Scientific Advisory Board of Mirati Therapeutics. A.C.K. has financial interests in Vescor Therapeutics, LLC. A.C.K. is an inventor on patents pertaining to KRAS-regulated metabolic pathways, redox control pathways in pancreatic cancer, targeting GOT1 as a therapeutic approach, and the autophagic control of iron metabolism. A.C.K. is on the Scientific Advisory Board of Cornerstone/Rafael Pharmaceuticals.

Figures

Extended Data Fig. 1. Genetic suppression of mutant KRAS increases autophagic flux in PDAC cell lines.

a, Signaling diagram displaying the multiple ways the KRAS ERK MAPK and autophagy pathways were perturbed in this study, as well as the specific components in autophagic signaling that were inhibited and monitored. Red text denotes phosphorylation sites that were assayed, and genetic and pharmacologic inhibitors that were used. b, A panel of human PDAC lines were transduced with a lentiviral vector to stably express EGFP-LC3B and transiently transfected with siRNA targeting KRAS (KRAS) or a mismatch (MM) control oligo (72 h) and treated with chloroquine (12 h) to assess flux. Quantification of EGFP+ punctae area normalized to cell area (autophagic index). Mean autophagic index is plotted, with each individual data point representing one field containing at least ten analyzed cells. Data are representative of two biological replicates for Pa02C and PANC-1 cells, and one biological replicate for Panc10.05, HPAC, Pa14C and Pa04C cells. P values shown are from unpaired, two-sided t-test; error bars represent s.e.m. c, Representative images of EGFP-LC3B expressing cells described and quantified in b. Scale bar, 20 μm..

Extended Data Fig. 2. Pharmacological inhibition of ERK1/ERK2 increases autophagic flux in PDAC cell lines.

a, PDAC lines were transduced with a lentiviral vector to stably express EGFP-LC3B and treated with DMSO or SCH772984 (ERKi, 1 μM) for 24 h, and chloroquine was added (12 h) to assess flux. Mean autophagic index is plotted, with each individual data point representing one field containing at least ten analyzed cells. Data are representative of two independent experiments for HPAC and PANC-1 cells, and one independent experiment for Pa02C, Pa04C, Pa16C and HPAF-II cells. P values are from unpaired t-test; error bars denote s.e.m. b, Representative images of EGFP-LC3B-expressing cells described and quantified in a. Scale bar, 20 μm. c, PDAC cells were treated with SCH772984 (concentrations in µM shown) for 24 h, RNA was extracted and BECN1 and PRKAA1 gene expression was measured. mRNA levels were normalized to ACTB mRNA. Relative expression quantified via the double delta Ct method is plotted. Data are from one independent experiment; error bars represent s.d. of three technical replicates. d, PDAC cell lines were treated with SCH772984 (1 µM, ERKi) for a time course of 1, 4, 12 and 24 h. Shown are immunoblots to determine levels of phosphorylated Beclin-1 (pBeclin-1), total Beclin-1, phosphorylated ERK (pERK), total ERK and vinculin, and are representative of two independent experiments..

Extended Data Fig. 3. ERK inhibition influences the transcription and activation of upstream mediators of autophagic signaling.

a, Mean comparisons of RPPA data for markers of ERK activity. All 12 cell lines were averaged based on treatment condition and treated as biological replicates. P values are from two-sided, unpaired t-test (RSK-3) or Wilcoxon test (all other proteins) comparing the mean of cells treated with SCH772984 (ERKi, 1 μM, 24 h) to untreated vehicle control (DMSO); error bars denote s.e.m. b, Mean comparisons of RPPA data for apoptotic markers. Means were calculated as described in a; P values are from Wilcoxon test; error bars denote s.e.m. c, Signaling pathway correlation analysis of RPPA data after treatment with vehicle (DMSO) or SCH772984 (1 μM, 24 h). Spearman’s correlation coefficients among 14 signaling proteins were used to create the heatmap. All 12 cell lines were averaged based on treatment condition and treated as biological replicates. Unsupervised hierarchical clustering analysis was performed on correlation coefficients. Red color represents positive correlation, gray represents no correlation and blue represents negative correlation (***P < 0.0005, **P < 0.005, *P < 0.05 from two-sided, unpaired t-test). d, Normalized mean-peak intensities for IMP, AMP and hypoxanthine identified from triplicate LC–MS/MS experiments. The mean metabolite concentrations from SCH772984-treated (ERKi, 1 μM) samples are normalized to DMSO controls; P values are from two-sided, unpaired t-test; error bars denote s.e.m. of mean peak intensity across three biological replicates. e, PDAC cells were treated with SCH772984 (concentrations in µM shown) for 24 h and RNA was extracted to measure GABARAPL1, WIPI1, SQSTM1 and MYC (control for ERK inhibition) gene expression. mRNA levels were normalized to ACTB mRNA. Relative expression quantified via the double delta Ct method is plotted. Data are from one independent experiment; error bars represent s.d. of three technical replicates.

Extended Data Fig. 4. Genetic silencing of KRAS and ERK inhibition reduces glycolytic flux in PDAC.

a, Normalized mean-peak intensities for indicated glycolytic metabolites identified from triplicate LC–MS/MS experiments. The mean metabolite concentrations from SCH772984-treated (ERKi, 1 μM) samples are normalized to DMSO controls and plotted for the seven PDAC lines assayed; P values are from two-sided, unpaired t-test; error bars denote s.e.m. Mean of three independent experiments is plotted. b, PDAC cells stably expressing EGFP-mCherry-LC3B were treated with SCH772984 (ERKi, 1 µM, 16 h) or cultured in medium containing dialyzed FBS and no glucose (16 h). Mean autophagic index is plotted, with each individual data point representing one field containing at least ten analyzed cells. Data for all cell lines are representative of two biological replicates; P values are from two-sided, unpaired t-test; error bars denote s.e.m.

Extended Data Fig. 5. Genetic silencing of KRAS and ERK inhibition alters mitochondrial dynamics in PDAC.

a, PDAC cells were transduced with the mitophagy probe COX8-mCherry-EGFP and treated with DMSO or SCH772984 (ERKi, 1 µM, 24 h). To assess mitophagy, the proportion of mCherryonly vesicles was quantified in relation to the total area of mitochondria (that is, EGFP+ and mCherry+). This proportion is represented as percentage mitophagy, with each individual data point representing one field containing at least ten analyzed cells; P values are from two-sided, unpaired t-test; error bars denote s.e.m. b, Representative images of HPAC and Pa14C cell lines expressing COX8-EGFP-mCherry quantified in l. Scale bar, 5 μm. c, Immunoblot analysis of PDAC cells treated with DMSO or SCH772984 (ERKi, 1 μM, 24 h) to determine the expression levels of PINK1, VDAC, phosphorylated ERK (pERK), total ERK and vinculin, representative of three independent experiments. d, PDAC cells were transiently transfected with siRNA constructs targeting KRAS (KRAS 1 and KRAS 2) or a mismatch (MM) control construct (72 h). Shown are representative images of mitochondrial morphologies after KRAS knockdown. Green, Anti-TOMM20; blue, DAPI. Scale bar, 20 μm. e, Quantification of mitochondrial morphologies observed in cells shown in d. Some 50 cells per condition per repetition were blindly scored and data are the mean of four independent experiments; error bars denote s.e.m.

Extended Data Fig. 6. Genetic silencing of KRAS decreases oxygen consumption in PDAC.

a, iKRAS 192 cells, derived from the doxycycline (Dox)inducible iKRAS mouse model, were cultured without doxycycline (24 h) to turn off Kras G12D expression. TMRE (200 nM) was added to the medium for 20 min and staining was analyzed by FACS. Median TMRE signal normalized to doxycycline ON is plotted and is the mean of six independent experiments; P values are from two-sided, unpaired t-test; error bars denote s.e.m. b, OCR response after doxycycline withdrawal (24 h) in iKRAS 192 cells. Mean oxygen consumption is plotted and is the mean of six independent experiments; P values are from two-sided, unpaired t-test comparing –Dox to +Dox, which was normalized to 1 for each measurement; error bars denote s.e.m. c, PDAC cell lines were transiently transfected with two siRNA constructs targeting KRAS (KRAS 1 or KRAS 2) or a mismatched control (MM). OCR response of PDAC cells after knockdown (60 h). Data are the mean of three independent experiments for each line, P values are from two-sided, unpaired t-test, comparing KRAS KD to MM, which was normalized to 100 for each measurement; error bars denote s.e.m.

Extended Data Fig. 7. ERK and MEK inhibition synergizes with the inhibition of autophagy to reduce proliferation in human PDAC cells.

a, A panel of PDAC cell lines (indicated lines shown as representative) were treated with a range of BVD-523 concentrations (BVD, 0.0195–10 µM) and a constant concentration of chloroquine (CQ, 6.25 μM) for 72 h and proliferation was monitored by the addition of MTT (3 h) at end of experiment. Normalized absorbance at 590 nm, comparing treatment conditions to BVD-only DMSO control, which was normalized to 100, is plotted and error bars denote s.d. of three technical replicates. Curves are representative of four independent experiments. b, PDAC cells (indicated) were treated with a range of binimetinib concentrations (0.005–2.5 µM) and indicated constant concentrations of chloroquine, and proliferation was quantified via live cell counting after 5 d of treatment. Shown are dose response curves and heatmaps representing BLISS independence scores. Plots are representative of three independent experiments. c, CI values for a panel of indicated PDAC cell lines treated with BVD-523 were calculated using Compusyn. d, A panel of PDAC cell lines were treated with a range of SCH772984 concentrations (0.0195–10 µM) and a constant concentration of chloroquine (6.25 μM) for 72 h, and proliferation was monitored by live cell counting. CI values for a panel of indicated PDAC cell lines were calculated using Compusyn.

Extended Data Fig. 8. ERK inhibition synergizes with the inhibition of autophagy to reduce proliferation in subject-derived PDAC organoid models.

a, Representative images of organoids from experiment described in Fig. 5e,f. Scale bar, 100 μm. Images are representative of five independent experiments. b, Quantification of organoid viability in representative wells shown in a, representative of five independent experiments. c, hT2 subject–derived organoids were grown for 10 d in the presence of indicated concentrations of chloroquine (CQ) and SCH772984 (ERKi). Growth curves shown are representative of five independent experiments. d, BLISS independence scores from experiment described in c. e, Representative images of organoids from d; scale bar, 100 μm. Images are representative of five independent experiments. f, Quantification of organoid viability in representative wells shown in e, representative of five independent experiments.

Extended Data Fig. 9. ERK inhibition synergizes with the inhibition of autophagy to reduce tumor growth in PDAC PDX mouse models.

a, Kaplan-Meier survival curves for tumor-bearing NSG mice described in Fig. 5g. b, Relative mouse weights throughout the course of the experiment described in Fig. 5g. Plotted is the mean percentage change in weight, normalized to weight at day 0 of treatment. c, Immunocompromised (NSG) mice with implanted KRAS-mutant PDX tumors AZ97 were treated with SCH772984 (ERKi) and hydroxychloroquine (HCQ) alone or in combination for the indicated days. Mean tumor volume normalized to tumor volume at day 0 of treatment is plotted over time; error bars denote s.e.m. The control and single-agent data are the mean of four independent tumors and the combination is the mean of five independent tumors. d, Kaplan-Meier survival curves for tumor-bearing NSG mice described in c. e, Images of representative tumors from AZ1013 PDAC PDX treatment groups. Images of control and ERKi-treated tumors are reproduced from Vaseva et al.

Extended Data Fig. 10. Inhibition of upstream regulators of autophagy reduces PDAC proliferation and synergizes with ERK inhibition.

a, Additional PDAC cell lines (Pa14C and PANC-1) were treated with a two-fold dilution range (beginning at 50 µM) of the indicated autophagy inhibitors. Proliferation was assessed using live cell labeling with calcein AM. b, Additional PDAC cell lines were treated with a range of SCH772984 concentrations (ERKi, 0.0195–10 µM) and a constant concentration of each autophagy inhibitor: SBI-0206965 (SBI, 2 μM), MRT68921 (MRT, 500 nM) and Spautin-1 (SP-1, 1.25 μM). Proliferation was assessed by live cell counting. Normalized mean cell number, comparing treatment conditions to SCH (ERKi)-only DMSO control, which was normalized to 100, is plotted and error bars denote s.d. of three technical replicates. Curves are representative of three independent experiments. c, CI values for Spautin-1 + SCH772984 combination were calculated from representative data displayed in Fig. 6f for multiple cell lines. d, MIA PaCa-2 cells were treated with ARS-1620 (RASi, 3 µM), SBI-0206965 (SBI, 2 µM) or MRT68921 (MRT, 1 µM) alone or in combination (5 d). Apoptosis was monitored using FACS analysis of Annexin-V/propidium iodide-labeled cells. Percentage apoptotic cells (normalized to vehicle (DMSO) control) is plotted. Data are representative of two biological replicates.

Fig. 1. KRAS suppression increases autophagic flux in KRAS-mutant PDAC cell lines.

a, PDAC cell lines were stably infected with a lentiviral vector encoding mCherry-EGFP-LC3B and then transiently transfected with siRNA targeting KRAS (KRAS) or a mismatch control oligo (MM) (72 h). To quantify autophagic flux, the area ratios of mCherry+ punctae to EGFP+ punctae (autophagic index) were determined. Mean autophagic index is plotted, with each individual data point representing one field containing at least ten analyzed cells. Data for HPAC, PANC-1, Pa01C and Pa14C represent two independent experiments; data for HPAF-II, Pa04C and Pa16C cells represent one independent experiment. b, Representative images of cells described and quantified in a, which visualize autophagosomes (EGFP+ and mCherry+) and autophagolysosomes (mCherry+ only). Scale bar, 20 μm. c, MIA PaCa-2 cells expressing mCherry-EGFP-LC3B were transfected with siRNA targeting KRAS as described in a or treated with ARS-1620 (10 μM, 24 h), and the mean autophagic index was determined. Data are representative of three independent experiments. d, PDAC cell lines were transiently transfected with siRNA as described in a and treated with bafilomycin A1 (Baf, 200 nM, 2 h) to assess flux. Immunoblot analyses of cell lysates were done to determine the levels of LC3B, KRAS and vinculin (control for total protein) and are representative of three independent experiments. e, Three cell lines derived from the iKRAS doxycycline-inducible PDAC mouse model were stably infected with a lentiviral vector encoding mCherry-EGFP-LC3B. Doxycycline was removed (–Dox) from the medium for 24 h, and mean autophagic flux was assessed as described in a. Data are representative of two independent experiments. f, iKRAS cells were deprived of doxycycline for 24 h to silence Kras G12D expression and treated with Baf (200 nM, 2 h) to assess autophagic flux. Immunoblots were performed as in d and are representative of three independent experiments. g, PDAC cell lines were stably infected with lentiviral vectors encoding shRNA targeting KRAS or a nonspecific control (96 h). Immunoblots were done to determine the levels of phosphorylated AMPKα (pAMPKα), total AMPKα, phosphorylated Beclin-1 (pBeclin-1), total Beclin-1, KRAS and vinculin and are representative of three independent experiments. h, BxPC-3 cells were processed and imaged as described in a. Mean autophagic index of five independent experiments is plotted. i, Representative images of BxPC-3 cells as described in h (scale bar, 20 μm), representative of five independent experiments. j, BxPC-3 cells were transiently transfected with siRNA as in a and immunoblotted to determine the levels of pAMPKα, KRAS and β-actin (control for total cellular protein); immunoblots are representative of three independent experiments. All P values shown are from unpaired, two-sided t-test. Error bars represent s.e.m.

Fig. 2. Pharmacological inhibition of ERK1/ERK2 increases autophagic flux in KRAS-mutant PDAC cell lines.

a, PDAC cell lines were transduced to stably express mCherry-EGFP-LC3B and then treated with the ERK1/ERK2 inhibitor, SCH772984 (ERKi, 1 μM) or DMSO for 24 h. Autophagic index indicates the ratio of the areas of mCherry+ punctae to EGFP+ punctae. Mean autophagic index is plotted, with each individual data point representing one field containing at least ten analyzed cells. Data for HPAC, PANC-1 and Pa14C cells are representative of three independent experiments; two independent experiments were performed for Pa01C and Pa16C cells; and one was performed for CFPAC-1, SW1990 and Pa04C cells. b, Representative images of cells described and quantified in a; scale bar, 20 μm. c, PDAC cell lines were treated with DMSO or SCH772984 (ERKi, 1 μM, 24 h); Baf was added (200 nM, 2 h) to assess flux. Immunoblot analyses of cell lysates were done to determine the levels of LC3B and vinculin, and are representative of three independent experiments. d, Three cell lines derived from the iKRAS doxycycline-inducible PDAC mouse model were stably infected with a retroviral vector encoding mCherry-EGFP-LC3B. Cells were treated with SCH772984 (ERKi, 1 μM, 24 h) and mean autophagic flux was assessed as described in a. Data are representative of two independent experiments. e, Representative images of cells described and quantified in d; scale bar, 20 μm. f, PDAC cell lines were treated with DMSO or SCH772984 (ERKi, 250 nM, 500 nM or 1 μM) for 24 h. Immunoblot analyses of cell lysates were done to determine the levels of pAMPKα, total AMPKα, pBeclin-1, total Beclin-1 and β-actin, and are representative of three independent experiments. g, BxPC-3 cells were prepared as described in a. Mean autophagic index of five independent experiments is plotted. h, Representative images of cells quantified in g; scale bar, 20 μm. i, BxPC-3 cells treated as described in g. Immunoblot analyses of cell lysates were done to determine the levels of pAMPKα, total AMPKα, pBeclin-1, total Beclin-1 and β-actin, and are representative of three independent experiments. All P values shown are from unpaired, two-sided t-test. Error bars represent s.e.m.

Fig. 3. Pharmacological inhibition of ERK1/ERK2 increases autophagic flux via multiple mechanisms.

a, PDAC cell lines were treated with vehicle (DMSO) or SCH772984 (ERKi, 1 μM) for 1 or 24 h. Three independent experiments were prepared, and RPPA analysis was performed. RPPA expression data for 1 h of SCH772984 treatment were standardized to no-treatment samples, and 24 h SCH772984-treatment samples were standardized to 24 h DMSO-treated samples. Data were log2 transformed, and medians of three independent experiments are represented in the heatmap. Only proteins with statistically significant (P < 0.05, FDR corrected, Wilcoxon test) changes at 1 or 24 h are shown. Samples and proteins were arranged via hierarchical clustering. b, Mean comparisons of autophagy-and mTOR pathway-related proteins from RPPA data in which all 12 cell lines (shown in a) are averaged based on treatment condition and treated as biological replicates. P values are from unpaired, two-sided, t-test (phosphorylated mTOR (s2448)) or Wilcoxon test (all other proteins) comparing the mean of cells treated with SCH772984 (ERKi, 1 μM, 24 h) to vehicle control (DMSO) are shown; error bars denote s.e.m. c, Normalized mean-peak intensities for IMP, AMP and hypoxanthine identified from triplicate LC–MS/MS experiments. The average metabolite concentrations from SCH772984-treated (ERKi, 1 μM) samples are normalized to DMSO controls; P values are from two-sided, unpaired t-test; error bars denote s.e.m. Mean of three independent experiments is plotted. d, A panel of seven PDAC cell lines (HPAC, PANC-1, HPAF-II, SW1990, Pa01C, Pa04C and Pa14C) were treated with SCH772984 (1 µM) for 1, 4, 12 or 24 h. Cells were collected, and RNA-seq analysis was performed. GSVA was completed to estimate the variation of different metabolic pathway activities over time in a nonsupervised manner. Each of the seven cell lines was treated as a biological replicate. Shown are the GSVA scores for a gene set composed of autophagy-related genes. Significance was determined based on a repeated measures ANOVA. Box plot components include minimum, 25th percentile, median, 75th percentile and maximum. e, Samples were prepared and analyzed as described in d. Shown are the GSVA scores for a gene set composed of lysosomal membrane proteins and hydrolases. Significance was determined as in d. f, Samples were prepared and analyzed as described in d. Shown are the GSVA scores for a gene set composed of glycolysis-related genes. Significance was determined as in d. g, Samples were prepared and analyzed as described in d. Shown are the GSVA scores for a gene set composed on mitochondrial biogenesis-related genes. Significance was determined as in d.

Fig. 4. ERK inhibition and KRAS silencing impairs mitochondrial function in PDAC cells.

a, Cells were treated with vehicle (DMSO) or SCH772984 (ERKi, 1 μM) for 24 h. Shown is quantification of mean fluorescence of MitoTracker green staining, normalized to DMSO for each cell line across three independent experiments. P values are from two-sided, unpaired t-test; error bars denote s.e.m. b, Mitochondrial morphologies of PDAC cells treated with SCH772984 (ERKi, 1 µM) for 24 h. Green, anti-TOMM20; blue, DAPI. Scale bar, 20 μm. c, Quantitation of mitochondrial morphologies observed in cells shown in b. Some 50 cells per condition per repetition were blindly scored and data are averages of four independent experiments; error bars denote s.e.m. d, Pa14C cells were treated with SCH772984 (ERKi, 1 µM) for 1 h. Immunoblot of phosphorylated DRP1 (S616) (pDRP1), total DRP1 and β-actin is representative of three independent experiments. e, Mitochondrial morphologies of PDAC cells treated with SCH772984 (ERKi, 1 µM) for 24 h. Green, Anti-TOMM20; blue, DAPI; scale bar, 20 μm. f, Quantitation of mitochondrial morphologies observed in cells shown in e. Some 50 cells per condition per repetition were blindly scored; data are means of four independent experiments; error bars denote s.e.m. g, PDAC cells were treated with SCH772984 (ERKi, 1 μM) for 24 h. Immunoblot of pDRP1, total DRP1, MYC, phosphorylated RSK (pRSK) and β-actin is representative of three independent experiments. h, Cell lines derived from the iKRAS PDAC mouse model were deprived of doxycycline (Dox) to turn off Kras G12D for 24 h. Representative transmission electron microscopy (TEM) micrographs displaying mitochondrial morphologies in the presence and absence of doxycycline; images are representative of six independent experiments; scale bar, 500 nm. i, Cells were treated as described in a. Quantification of relative mitochondrial potential via labeling with the JC-1 mitochondrial dye. Data are the mean of three independent experiments; P values are from two-sided, unpaired t-test; error bars denote s.d. of median ratios of JC-1 aggregate fluorescence to JC-1 monomer fluorescence. j, Representative histograms from Pa01C and Pa14C cells quantified in i; as a control, the uncoupler CCCP was added to acquired DMSO samples to establish minimum potential. k, OCR response of PDAC cells treated with DMSO or SCH772984 (ERKi, 1 μM) for 1.5 h; shown is mean OCR of three independent experiments; error bars denote s.e.m. of mean OCR across replicates. l, OCR response of PDAC cells treated with DMSO or SCH772984 (ERKi, 1 μM) for 24 h. Data for HPAC and Pa01C lines are the mean of nine independent experiments; data for Pa14C are the mean of eight independent experiments, and data for PANC-1 are the mean of three independent experiments. P values are from two-sided, unpaired t-test, comparing treatment conditions to DMSO, which was normalized to 100 for each measurement; error bars denote s.e.m.

Fig. 5. Dual inhibition of ERK signaling and autophagy synergistically impairs PDAC proliferation.

a, PDAC cell lines (indicated lines shown as representative) were treated with a range of SCH772984 concentrations (SCH, 0.0195–10 µM) and a constant concentration of chloroquine (CQ, 6.25 μM) for 72 h, and proliferation was monitored by the addition of MTT for 3 h at end of experiment. Normalized absorbance at 590 nm, comparing treatment conditions to DMSO control, which was normalized to 100, is plotted, and error bars denote s.d. of three technical replicates. Curves are representative of four independent experiments. b, A panel of PDAC cell lines was treated with ERK inhibitors SCH772984 (SCH), BVD-523 (BVD) or chloroquine (CQ), alone or in combination. Cells were stained with crystal violet to visualize colonies of proliferating cells 14 d after plating. Average colony number, normalized to the DMSO control, of three technical replicates is shown; error bars represent s.d. Data are representative of two independent experiments. c, Combination indices (CI values) were calculated using Compusyn. Average CI values for a panel of indicated PDAC cell lines treated as described in a are shown; values are mean of four independent experiments; error bars denote s.e.m. of CI value across replicates. d, A panel of indicated human PDAC cell lines treated with SCH772984 (ERKi) or CQ alone or in combination for 5 d. Apoptosis was monitored using FACS analysis of Annexin-V and propidium iodide–labeled cells. Percentage apoptotic cells (normalized to vehicle (DMSO) control) is plotted. Data are the mean of four independent experiments for each cell line and condition; error bars denote s.e.m. of mean percentage apoptotic cells. BLISS scores were calculated to determine whether combination was synergistic. e, hM1A subject–derived organoids were grown for 10 d in the presence of indicated concentrations of CQ and SCH772984 (ERKi, 0.039 to 2.5 µM). Growth curve is representative of five independent experiments. f, Heatmap representing BLISS independence scores corresponding to representative growth curve shown in e. g, Immunocompromised (NSG) mice with implanted KRAS-mutant PDX tumor (AZ1013) were treated with SCH772984 (ERKi) alone or together with hydroxychloroquine (HCQ) for 22 d. Mean tumor volume is plotted over time; error bars denote s.e.m. The control and single-agent data are the mean of four independent tumors, and the combination is the mean of five independent tumors. h, Quantitation of AZ1013 tumor weights; P value is from two-sided, unpaired t-test comparing ERKi-treated to ERKi plus hydroxychloroquine (HCQ)-treated tumors; error bars denote s.e.m.; sample sizes are as in g.

Fig. 6. Synergy with ERK inhibition is observed when autophagy is inhibited at multiple levels.

a, A panel of PDAC cell lines was stably transduced with two distinct shRNA constructs (ATG 5–1 or ATG 5–2) or a nonspecific control construct (NS). Cell lines were treated with a range of SCH772984 concentrations (ERKi, 0.0195–10 µM, 72 h), and proliferation was assessed by live cell counting. Normalized mean cell number, comparing treatment conditions to SCH (ERKi)-only control, which was normalized to 100, is plotted, and error bars denote s.d. of three technical replicates. Curves are representative of three independent experiments. b, A panel of PDAC cell lines were stably transduced with three distinct shRNA constructs (ATG 7–1, ATG7–2 or ATG 7–3) or NS and were treated and analyzed as in a. c, Immunoblot analyses of cell lysates were done to determine the levels of ATG5 and vinculin, and are representative of three independent experiments. d, Immunoblot analyses of cell lysates were done to determine the levels of ATG7 and β-actin, and are representative of three independent experiments. e, A panel of PDAC cell lines (HPAC data shown as representative) were treated with a two-fold dilution range (beginning at 50 µM) of indicated autophagy inhibitors for 72 h. Proliferation was assessed using live cell counting. f, A panel of PDAC cell lines were treated with a range of SCH772984 concentrations (ERKi, 0.0195–10 µM) and a constant concentration of each autophagy inhibitor: SBI-0206965 (SBI, 2 μM), MRT68921 (MRT, 500 nM) and Spautin-1 (SP-1, 1.25 μM). HPAC data shown are representative of three independent experiments. Proliferation was assessed using live cell counting. Data were analyzed and plotted as in a. g, CI values calculated from representative data displayed in f for multiple cell lines. h, MIA PaCa-2 cells were treated with a range of ARS-1620 concentrations (0.039–20 µM) and constant concentrations of SBI-0206965 (SBI) or MRT68921 (MRT) for 5 d, and proliferation was quantified by live cell counting. Data were analyzed and plotted as in a, and the plot is representative of three independent experiments. i, Representative CI values for each combination described in h are shown.