Co-targeting of epidermal growth factor receptor (EGFR) and PI3K overcomes PI3K/Akt oncogenic dependence in pancreatic ductal adenocarcinoma

Matthew H. Wong,1,2*Aiqun Xue1*, Sohel M. Julovi1, Nick Pavlakis2, Jaswinder S. Samra3, Thomas J. Hugh3, Anthony J. Gill4, Lyndsay Peters5, Robert C. Baxter6, Ross C. Smith1,3

1Cancer Surgery, Kolling Institute of Medical Research, University of Sydney;2Department of Medical Oncology, 3Department of Upper Gastrointestinal Surgery, 4Department of Pathology, Royal North Shore Hospital, Sydney; 5Flow Cytometry Unit, Haematology Division;6Hormones and Cancer Division, Kolling Institute of Medical Research, University of Sydney. *These authors contributed equally.

Corresponding author: Ross C Smith, Cancer Surgery, Level 8, Kolling Institute,

Royal North Shore Hospital, St Leonards, NSW, Australia, 2065.

Phone: 61-2-94373511

Email:

Keywords: Pancreatic ductal adenocarcinoma, erlotinib, NVP-BYL-719, PI3K/Akt, oncogenic dependence

Total number of figures and tables:5 (main), 10 (supplementary)

Supplemental Data

Appendix A: Supplemental Methods

RTK Phosphoarray

Protein concentrations were measured for all samples after lysate preparation and equalization (Thermo Scientific Fisher, Australia). The manufacturer’s instructions for the PathScan RTK Signaling Antibody Array Kit (Cell Signaling, Arundel, QLD) was strictly followed. Slides were read on high-resolution setting on the LAS-4000 (Fujifilm). The chemiluminescence intensity of 39 phospho-proteins was first quantified with Multigauge v3.0 software (Fujifilm), adjusted to both negative and positive controls, and graphed as per manufacturer’s recommendations.

Quantitative Real-time Reverse-Transcriptase Polymerase Chain Reaction (qRT-PCR)

Methods for qRT-PCR have been described previously.1 After RNA extraction (Minieasy, Qiagen, Chadstone, VIC), quantification and quality assessment (Nanodrop spectrometer, Thermo Scientific Fisher, Wilmington, BE) and reverse transcription (ReadyScript, Sigma, Castle Hill, NSW), cDNA was robotically loaded (EP Motion 5070, Eppendorf, North Ryde, NSW) onto 96-well PCR plate and run on the 7900HT PCR system (Applied Biosystems, Mulgrave, VIC). A two-step qRT-PCR using SYBR green primers (Sigma) and SYBR mastermixes (Life Technologies, Mulgrave, VIC) was used. PI3Kα, PI3Kγ, Akt-1, Akt-2, Akt-3, MAPK1 and MAPK3 genes were tested previously in PCR array and standard curve experiments. The final pairs of primers selected for qRT-PCR validation in this study were 5’-GAAACACAAGGAAAGGGAAC-3’ (AKT2 sense), 5’-AGGTCTTGATGTATTCACCTC-3’ (AKT2 anti-sense), 5’-TAACATCTGGAGACTGTGAG-3’ (MAPK1 sense), 5’-TGAAGTTAGTGGTTTCATGC-3’ (MAPK1 anti-sense), 5’-GAGTAACAGACTAGCTAGAGAC-3’ (PIK3CA sense) and 5’-AGAAAATCTTTCTCCTGCTC-3’ (PIK3CA anti-sense). Relative quantification using the delta-delta Ct method against 18S control gene was used to measure fold changes against BxPC-3 cell line.

Supplementary Table Legends

Table S1: IC50 values for inhibition of proliferation of (A) 5 PDAC cell lines and (B) 2 derived erlotinib-resistant (ER) cell lines, in response to inhibitors of EGFR (erlotinib, gefitinib), IGF1R (NVP-AEW541), MEK (PD-98059), PI3K (LY-294002), PI3K (NVP-BYL719) and PI3K/mTOR (NVP-BEZ235). Average  SEM of 6 experiments. Highly insensitive cell lines are shaded in grey in (A). Paired t-test statistics was calculated for resistance in (B).

Table S2: Cell cycle dual blockade experiments. (A) Cell cycle: Quantitative analysis of proportion of cells in G1, S and G2/M phases, calculated as mean +/-SEM of 3 experiments (at 24 hours). Average CV is provided: a CV <5% is acceptable for cell cycle analysis. (B) Apoptosis assay, average total apoptosis + necrosis by flow cytometry, expressed as a percentage of total cells;mean of 5 experiments with SEM.NT: untreated; E10: erlotinib 10 µM; A1: AEW541 1 µM; B5: BYL719 5 µM.

Table S3: In-vivo dual blockade in the subrenal xenograft model. (A) Primary endpoint: tumor volume comparisons among 4 treatment groups: no treatment (NT), gemcitabine 100mg/kg twice weekly, erlotinib 50mg/kg/d, or erlotinib 50mg/kg/d plus BYL-719 25mg/kg/d (ERL+ BYL) (n=90), with t- statistics, P-value and 95% confidence interval around parameter estimate. (B) Secondary endpoint: kidney weight (n=90). All statistics were calculated by two-way ANOVA adjusted by experiments (SAS 9.2). *P<0.05

Supplementary FigureLegends

Figure S1: EGFR and PI3K signaling in 5 PDAC cell lines. (A) Densitometric quantitation of Western blotting of pAkt/ total Akt(left panel) and pERK/ total ERK (right panel), with no treatment (NT), or stimulation by EGF, IGF-1, or both growth factors: average  SEM of 3 experiments. (B) Quantitation of Western blot of pEGFR, pIGF1R, pERK and pAkt in response to increasing dose of erlotinib or NVP-AEW541, average of 3 experiments. Drug doses were selected based on relative potency from proliferation assay and investigator brochure. Note that all signals were adjusted to EGF-stimulated control, so for PANC-1 which had a much more highly EGF-stimulated pAkt, the relative fold-change with erlotinib inhibition appeared smaller than for BxPC-3. * P<0.05, ** P<0.01

Figure S2:Quantitation of immunofluorescence by confocal microscopy, average of 4 experiments: (A) E-cadherin and vimentin immunofluorescence. (B) Total ERK and Akt on immunofluorescence. (C, D)pAkt and pERK immunofluorescence. *P<0.05; **P<0.001; ***P<0.0001

Figure S3:Phospho-RTK array on ER versus parental cell lines(A) Phosphorylated protein chemiluminescence intensity in untreated (NT) parental and resistant cells after adjusting to positive and negative controls, with the largest differences in signal intensity shaded in grey (B) Quantitation of Western blotting results of pERK, pAkt and pS6 adjusted to total proteins inER and parental cell lines, either untreated (NT) or after stimulation by EGF, IGF-1, or both growth factors.* P<0.05

Figure S4: Western blotting dual blockade isobolograms: Isobolograms constructed for pERK, pAkt and pS6 inhibition by ERL* AEW (A) and ERL* BYL (B) for BxPC-3 and PANC-1. An IC50additivity line was extrapolated based on the predicted single drug IC50 in three separate experiments. The combined IC50values of dual inhibitors were plotted on the same graph. If the observed points lied below the additivity line, then the two drugs acted synergistically in inhibiting the respective signal.

Figure S5: Effect of single and dual blockade (DB) on cell proliferation (A, B) 3D contour maps for inhibition by ERL+AEW (A) and ERL+BYL (B). Synergy index (SI) was calculated by non-linear regression (PROC NLIN) using the Bliss synergy/antagonism formula. SI>1 indicates synergy and SI<1 indicates antagonism between two drugs. If the 95% confidence interval does not cross 1, the effect is significant. (C) Proliferation of control and ER cell lines was measured in real-time over 70 h of drug treatment. Treatments are control (black solid), erlotinib 10 µM (E10- black dotted), NVP-AEW541 1 µM (A1- black dashed), NVP-BYL719 5 µM (B5- grey solid), E10A1 (E10+ A1- grey dotted), and E10B5 (E10+ B5- grey dashed). Mean values ± SEM for 4 experiments.

Figure S6: Effect of single and dual blockade (DB) on apoptosis in resistant and parental cell lines. (A) Percentage of cells undergoing apoptosis and necrosis measured by flow cytometry: paired t-test statistics calculated for DB versus no treatment. (NT)(B) Representative confocal microscopy of cleaved caspase-3 (red), after 24 hours of treatment. Blue = DAPI (nuclei). (C) Cleaved caspase-3 immunofluorescence, mean intensity averaged from 4 experiments, with t-test statistics *P<0.05; **P<0.001; ***P<0.0001. (D) Relative migration rate, expressed as increasing percentage cell density across a monolayer wound over 44 hours, mean values from 3 experiments ± SEM. Treatments are control (black solid), erlotinib 10 µM (E10- black dotted), NVP-AEW541 1 µM (A1- black dashed), NVP-BYL719 5 µM (B5- grey solid), E10A1 (E10+ A1- grey dotted), and E10B5 (E10+ B5- grey dashed). Proliferation was inhibited by 10 µg/ml mitomycin.

Figure S7: Translational study endpoints (IHC) and subcutaneous in-vivo model. (A) Representative photos of primary tumorpEGFR and pAktimmunohistochemistry in each of the 6 subrenal model experiments. (B)Mean and sem oftumor volume of dual blockade (ERL+ BYL) versus single blockade (ERL, BYL) versus control (NT) over 28 days in the subcutaneous model (n=5 in each treatment group).

1