BIM and mTOR expression levels predict outcome to erlotinib in EGFR-mutant non-small-cell lung cancer

BIM is a proapoptotic protein that initiates apoptosis triggered by EGFR tyrosine kinase inhibitors (TKI). mTOR negatively regulates apoptosis and may influence response to EGFR TKI. We examined mRNA expression of BIM and MTOR in 57 patients with EGFR-mutant NSCLC from the EURTAC trial. Risk of mortality and disease progression was lower in patients with high BIM compared with low/intermediate BIM mRNA levels. Analysis of MTOR further divided patients with high BIM expression into two groups, with those having both high BIM and MTOR experiencing shorter overall and progression-free survival to erlotinib. Validation of our results was performed in an independent cohort of 19 patients with EGFR-mutant NSCLC treated with EGFR TKIs. In EGFR-mutant lung adenocarcinoma cell lines with high BIM expression, concomitant high mTOR expression increased IC50 of gefitinib for cell proliferation. We next sought to analyse the signalling pattern in cell lines with strong activation of mTOR and its substrate P-S6. We showed that mTOR and phosphodiesterase 4D (PDE4D) strongly correlate in resistant EGFR-mutant cancer cell lines. These data suggest that the combination of EGFR TKI with mTOR or PDE4 inhibitors could be adequate therapy for EGFR-mutant NSCLC patients with high pretreatment levels of BIM and mTOR.

. The relationship between the EGFR pathway, apoptosis and the DGKα -PDE4-cAMP-mTOR pathway was designed using the Ingenuity Pathway Analysis (IPA) software (https://www.ingenuity. com/). EGFR stimulates intracellular signalling cascades, such as the RAS/RAF/ERK (MAPK) pathwaywhich induces BIM proteosomal degradation-and the PI3K/AKT/mTOR pathway. mTOR nucleates a rapamycin and nutrient -sensitive multiprotein complex called mTORC1, and a second growth-factorsensitive but nutrient-insensitive mTOR-containing complex called mTORC2. Besides mTOR, mTORC1 contains Raptor, mLST8 (also known as Gβ L), and PRAS40 (proline-rich AKT substrate 40 kDa). mTORC2, like mTORC1, also includes the mLST8 protein, but instead of Raptor, mTORC2 contains the Rictor and mammalian stress-activated protein kinase [SAPK]-interacting (mSIN1) proteins. mTORC2 also contains Protor (protein observed with RICTOR). Ribosomal S6 kinase 70kDa (p70S6K) and eIF4E-binding protein 1 (4EBP1)-both regulators of mRNA translation-are the only extensively described mTORC1 substrates. Phosphorylation of the translational repressor 4EBP1 results in its dissociation from the eukaryotic initiation factor 4E (eIF4E), thereby allowing eIF4E to assemble with other translation initiation factors and initiate cap-dependent translation. mTORC2 directly phosphorylates and activates AKT. BIM activates BAK and BAX, causing activation and mitochondrial outer membrane permeabilization (MOMP). Anti-apoptotic BCL-2 proteins prevent MOMP by binding BIM and other BH3-only proteins as well as activated BAX or BAK. Following MOMP, release of various proteins from the mitochondrial intermembrane space promotes caspase activation and apoptosis. DGKα is a lipid kinase that phosphorylates the lipid diacylglycerol (DAG), transforming it into phosphatidic acid. Phosphatidic acid activates mTOR signalling via a unique pathway involving cAMP. The cAMP-degrading PDE4 enzymes also activate mTOR signalling. mTORC1 promotes survival through translational control of Mcl-1.
Scientific RepoRts | 5:17499 | DOI: 10.1038/srep17499 signalling, is a mediator of tumour cell death in response to targeted therapies 7 (Fig. 1). Faber et al., were the first to demonstrate that patients with EGFR-mutant NSCLC and low BIM expression derive less clinical benefit from EGFR inhibitors 5 . We identified high levels of BIM mRNA expression as a predictive marker of response, progression-free survival (PFS) and overall survival (OS) in erlotinib-treated EGFR-mutant NSCLC patients 8 .
RAF or MEK inhibitors inhibit ERK phosphorylation (P-ERK) and induce BIM levels in BRAF-mutant melanoma cell lines. In resistant melanoma cell lines, vemurafenib (BRAF inhibitor) or selumetinib (MEK inhibitor) either fail to suppress P-ERK or resistance emerges through the activity of mammalian target of rapamycin (mTOR), despite P-ERK suppression and BIM induction 9 . This suggests that BIM regulation is MAPK-dependent, but mTOR-independent, and BIM up-regulation is not always sufficient to promote apoptosis 9 . Combining vemurafenib with an mTOR or PI3K inhibitor improved cell killing in BRAF-mutant melanomas with ERK-independent resistance to MAPK inhibition 9 . mTOR, a multifunctional 293-kDa serine/threonine protein kinase encoded by the gene MTOR, is a downstream effector of PI3K/AKT and promotes cell growth, division, angiogenesis and metabolic reprogramming 9 . The mTOR kinase serves as the catalytic subunit of two multiprotein complexes with distinct functions: mTOR complex 1 (mTORC1), a rapamycin and nutrient-sensitive complex, defined by the regulatory associated protein of mTOR (Raptor), and mTOR complex 2 (mTORC2), a growth-factor-sensitive but nutrient-insensitive complex, defined by the rapamycin-insensitive companion of mTOR (Rictor) (Fig. 1) 10 .
The activity of mTORC1 is regulated by the integration of many signals. For instance, increases in circulating branched-chain amino acids as a result of a high-fat diet, induce mTOR signalling independent of PI3K signalling 11 . In glioblastoma and melanoma cells, diacylglycerol kinase α (DGKα ), a lipid kinase converting diacylglycerol to phosphatidic acid, regulates both mTOR activity and MTOR mRNA levels via modulation of cyclic adenosine monophosphate (cAMP) ( Fig. 1) 12,13 . The inhibitory effect of cAMP on mTOR can be also neutralized by phosphodiesterase 4 (PDE4), an enzyme in which two of four isoforms (PDE4A and PDE4D) are increased under hypoxia in lung adenocarcinoma cell lines ( Fig. 1) 13,14 . Once activated, mTORC1 phosporylates ribosomal S6 kinase 70 kDa (p70S6K) and eIF4E-binding protein 1 (4EBP1) to promote cap-dependent translation and cell growth (Fig. 1).
To further understand the clinical implications of mTOR in EGFR-mutant patients, we assessed baseline mRNA levels of MTOR by quantitative real-time polymerase chain reaction (qRT-PCR) in 57 EGFR-mutant erlotinib or chemotherapy treated NSCLC patients from the EURTAC trial from whom tumour tissue was available 8 . Herein we present updated results of the correlation of BIM mRNA alone and in combination with MTOR with OS, PFS and response in these 57 EGFR-mutant NSCLC patients (training cohort). An independent group of 19 EGFR-mutant patients treated with EGFR TKIs was included in the study as a validation cohort for which BIM expression and the phosphorylation state of ribosomal protein S6 (P-S6) were additionally determined by immunohistochemistry. Finally, BIM and mTOR expression were determined in our panel of EGFR-mutant lung adenocarcinoma cell lines and correlated with the half maximal inhibitory concentration (IC 50 ) of gefitinib. We investigated the effect of gefitinib treatment on BIM expression and mTOR expression and activity. DGKa, PDE4A and PDE4D expression were examined in our cell lines and the results correlated with mTOR expression.

Results
The EURTAC study enrolled 173 patients with EGFR mutations who were randomized to receive erlotinib or standard intravenous chemotherapy with cisplatin or carboplatin plus docetaxel or gemcitabine 1 . Pretreatment tumour specimens were available from 57 of these patients for assessment of MTOR mRNA expression. Table 1 shows patient characteristics of the 57 patients included in the present subanalysis. The EURTAC was approved by the Institutional Review Board of each participating centre and written informed consent was obtained from all patients. Among the 48 patients whose MTOR mRNA was successfully examined, MTOR expression was low ( Although not statistically significant, a trend for a positive correlation was found between BIM mRNA and protein expression (Wilcoxon test two-side P value = 0.1161) as well as MTOR mRNA and P-S6 expression (Wilcoxon test two-side P value = 0.4048) ( Supplementary Fig. 1a (Fig. 2a). No significant differences in PFS were observed according to MTOR mRNA levels. Among the seven erlotinib treated patients with high BIM and evaluable MTOR expression levels, median PFS was NR (95% CI, 9.7-NR) for those with low/intermediate MTOR vs 9.7 months (95% CI, NR) for those with high MTOR (P = 0.0894). MTOR did not affect PFS in patients with low/intermediate BIM (Fig. 2b). In the univariate analysis, erlotinib (hazard ratio [HR] = 0.48; 95% CI, 0.25-0.93; P = 0.0265) and high BIM expression (HR = 0.40; 95% CI, 0.20-0.80; P = 0.0095) were associated with longer PFS (Supplementary Table 2   BIM and mTOR expression and in vitro sensitivity to gefitinib. We examined the in vitro sensitivity of five EGFR-mutant lung adenocarcinoma cell lines to gefitinib (Table 2). Gefitinib-sensitive PC-9 cells harbour a small in-frame deletion in exon 19 that leads to elimination of an LREA motif in the protein (Del E746-A750). Gefitinib-sensitive H3255 and 11-18 cells harbour a point mutation in exon 21 that substitutes an arginine for leucine at position 858 in the protein (L858R). Gefitinib-insensitive H1975 and H1650 cells, although harbouring the same kinase domain mutations (L858R and Del E746-A750), have additional changes such as T790M (H1975) or phosphatase and tensin homologue (PTEN) loss (H1650).
By matching cell line sensitivity to BIM and mTOR expression, we observed that inhibition concentration of 50% cell viability (IC 50 ) induced by gefitinib was increased as mTOR expression increased, in the three sensitive and high BIM expressing EGFR-mutant lung adenocarcinoma cell lines, H3255, PC-9 and 11-18. In fact, H3255 cells with high BIM and low mTOR expression (both protein and mRNA) are hypersensitive to gefitinib, yielding IC 50 values at 10-fold lower concentrations compared to PC-9 and at 100-fold lower concentrations compared to 11-18 (Fig. 3).
To test the ability of gefitinib to induce BIM and inhibit mTOR expression in EGFR-mutant cells, we treated cells with gefitinib and performed western blotting and qRT-PCR. Treatment of PC-9 and H3255 cells with gefitinib increased BIM protein expression even at a concentration of 5nM. However changes in mTOR expression levels were not observed in these cells (Fig. 4a). Treatment of PC-9 cells with gefitinib increased BIM mRNA expression in a dose-and time-dependent manner but MTOR expression was not affected (Fig. 4b). In contrast, gefitinib changed neither BIM nor mTOR expression in the less gefitinib-sensitive 11-18 cells as well as the gefitinib-resistant H1975 and H1650 cells (Fig. 4c). Furthermore in PC-9 and H3255 cells, gefitinib treatment inhibited the phosphorylation of mTOR and p70S6K, while phosphorylation levels of mTOR and p70S6K could not be inhibited below basal levels in 11-18, H1975 and H1650 cells (Fig. 5).
In an exploratory analysis, the protein and mRNA expression levels of DGKa, PDE4A and PDE4D were examined in the five EGFR-mutant lung adenocarcinoma cell lines in an effort to explore whether DGKa regulates MTOR transcription through modulation of cAMP levels. We also wished to elucidate the role of PDE4 as the predominant cAMP-degrading enzyme. Immunoblotting confirmed that the protein levels of PDE4D and mTOR are similarly increased in 11-18, H1975 and H1650 cells ( Supplementary  Fig. S3a). By qRT-PCR, MTOR mRNA expression showed significant positive correlation with PDE4D mRNA expression, with a Pearson correlation coefficient of r = 0.92; P = 0.0244 ( Supplementary Fig.  S3b).

Discussion
Although expression and degradation of BIM are regulated mainly by the MAPK pathway, a variety of other mechanisms can also regulate BIM function, including transcriptional and posttranscriptional regulation to posttranslational modification and epigenetic silencing 3 . For instance, an inverse relationship has been reported between miR-494 and BIM expression 15 . AKT may also phosphorylate and suppress the BIM transcription factor FOXO3 3,16 . Our findings highlight that pre-treatment assessment of BIM levels is able to identify EGFR-mutant patients who will benefit more from EGFR TKI treatment. An additional aim of our study was identification of MAPK-independent mechanisms that may not affect BIM induction but may still affect efficacy of EGFR TKI monotherapy. Among the 29 patients of the training cohort treated with erlotinib, PFS was 18.5 months for those with high BIM compared with 3.6 months for those with low/intermediate BIM mRNA expression (P = 0.0145). Median PFS was not reached for patients with high BIM and low/intermediate MTOR compared to 9.7 months for those with both high BIM and MTOR, though differences were not statistically significant (P = 0.0894).
In the validation cohort of 19 patients receiving treatment with erlotinib, gefitinib or afatinib, PFS was 15.0 months for those with high BIM compared with 9.2 months for those with low/intermediate BIM mRNA expression (P = 0.02). Among the 7 patients with high BIM and evaluable MTOR expression levels, PFS was 18.5 months for the four patients with low/intermediate MTOR compared to 13.0 months for the three with high MTOR, though differences were not statistically significant (P = 0.0939).
Interestingly, when we matched gefitinib sensitivity to BIM and mTOR mRNA and protein expression in EGFR-mutant lung adenocarcinoma cell lines, we observed that the IC 50 values of gefitinib increase as the mTOR levels increase in the three sensitive and high BIM expressing cell lines (PC-9, H3255 and 11-18 in Fig. 4). In cells with high mTOR expression, gefitinib did not induce BIM expression and did not suppress mTOR activity (11-18, H1975 and H1650 in Figs 4c and 5b). mTOR serves as a key signalling hub that integrates signals from several important upstream pathways, making it a bona fide target for molecular therapy 17 . RAF and MEK inhibitor combination has been found to be less effective in BRAF-mutant melanoma tumours with MAPK-independent resistance in which ERK is adequately suppressed but alternatively mTOR is activated as estimated by the phosphorylation of p70S6 kinase 1 (S6K1) 9 . Additionally mTOR activity can predict sensitivity of PIK3CA-mutant breast tumours to PI3K p110α inhibitors 18 . MTOR mutations have also been described as biomarkers for predicting tumour responses to mTOR inhibitors 19,20 .
Only rarely does single-agent therapy for cancer result in durable disease control. Patients with low BIM expression could derive only a meagre benefit from treatment with EGFR TKIs alone but could benefit from synthetic lethality combinations, including small molecules that mimic the BH3 motif. A previous study has demonstrated that gefitinib combined with the BH3 mimetic ABT-737 (an analog of navitoclax) substantially increases apoptosis compared with each agent alone in EGFR-mutant H1650 cells with low BIM expression 21 . Selective Bcl-xL family inhibitors like venetoclax have improved safety and efficacy profiles, compared to their less selective predecessor, navitoclax 22 . Patients with high BIM expression could benefit from EGFR TKIs but analysis of MTOR could further improve outcomes by selecting patients with high MTOR for combination therapy with EGFR TKIs and mTOR inhibitors. Interestingly, the addition of an mTOR inhibitor to BH3 mimetics reduces the expression of the antiapoptotic protein Mcl-1 and allows high BIM levels to "prime" tumour cells for apoptosis 23 .
A better understanding of the DGKα -PDE4-cAMP-mTOR pathway can indicate novel approaches to mTOR inhibition using DGKα or PDE4 inhibitors (Fig. 1) 3,12-14 . In the present study, in an exploratory in vitro analysis MTOR mRNA expression showed significant positive correlation with the PDE4D mRNA expression. By immunoblotting, mTOR expression was mainly related with PDE4D expression. Currently the effects of PDE4D in cancer are not fully understood and few studies have examined the role of PDE4D and its inhibitors in cancer therapy. A study revealed that hypoxia via hypoxia-inducible factor 1α regulates PDE4D in lung cancer cell lines, including H1975, and treatment with the first-generation PDE4D inhibitor rolipram decreased cell proliferation 14 . Roflumilast is an oral PDE4 inhibitor used for patients with chronic obstructive pulmonary disease 24 .
The limitations of our study are its retrospective nature and small sample size which limit statistical power. However, the data presented herein provide important biological insights and may be used to refine the predictive role of BIM for outcomes to EGFR TKIs. Pretreatment levels of BIM and mTOR can lead to adding mTOR or PDE4 inhibitors to EGFR TKIs 3 . Also, it is tempting to speculate that PDE4 could be a theranostic marker that warrants further research.

Methods
The Methods were carried out in accordance with the guidelines defined in the EURTAC study, which was approved by the Institutional Review Board of each participating centre. Written informed consent was obtained from all patients.

Gene expression analyses. All analyses were carried out centrally at the ISO 15189-certified Pangaea
Biotech oncology laboratory located in the Quirón Dexeus University Hospital (Barcelona, Spain). Gene expression analysis of MTOR was performed on RNA isolated from the tumour tissue specimens and cell lines. Gene expression analysis of DGKA, PDE4A and PDE4D was performed on RNA isolated from the cell lines. RNA extraction, retrotranscription analysis, and RT-PCR were performed as previously described and gene expression was examined by quantitative PCR using β-actin as housekeeping gene 25 .  20 min. The detection was performed with DAB detection kit (Ventana Medical Systems) according to manufacturer instruction. Slides were counterstained with hematoxylin and mounted. BIM staining was considered positive when either strong (3+ ) or moderate (+ 2) cytoplasmic staining was observed. P-S6 staining was considered positive when only strong (3+ ) cytoplasmic staining was observed. In addition, protein expression was quantified using the histoscore (HS) method. Briefly, each tumour specimen was scored on a semiquantitative scale ranging from 0 to 300, with the final score resulting from the percentage of tumour cells staining positively (range 0-100) multiplied by staining intensity graded as negative, weak, moderate or strong (range 0-3). The median HS value was used as a cutoff level to discriminate high vs low expression of each biomarker. All cell lines were maintained in RPMI medium supplemented with 10% FBS, 50 μ g/mL penicillin-streptomycin and 2 mM L-Glutamine. All cells were grown in a humidified atmosphere with 5% CO 2 at 37 °C. EGFR exons 19 and 21 of all cell lines were sequenced to confirm their status. Cell viability was assessed by the Thiazolyl Blue Tetrazolium Bromide (MTT) (Sigma, St Louis, MO) assay. Cells from each cell line were seeded at 2000 to 6000 per well in 96-well plates. The concentration of gefitinib required for IC 50 after a 72 h treatment was assessed. After treatment, cells were incubated with medium containing MTT (0.75 mg/mL in medium) for 1-2 h at 37 °C. Culture medium with MTT was removed and formazan crystals reabsorbed in 100 μ L DMSO (Sigma, St. Louis, MO). Cell viability was determined by measuring absorbance at 590 nm using a microplate reader (BioWhittaker, Walkersville, MD).

Statistical analysis.
The primary endpoint of the study was to examine the potential effects of BIM and MTOR mRNA expression levels on survival. On December 9 th 2013, 135 PFS events had occurred and the results reported here are based on data analyses from that cutoff date. For the OS analysis, patients were not censored at crossover, whereas all patients were censored at crossover for the analysis of PFS. PFS and OS were estimated by means of the Kaplan-Meier method and compared with a nonparametric log-rank test. Based on our previous experience [26][27][28] , in addition to analysing gene expression as a continuous variable, expression levels were divided into three groups according to their tertiles (inter-quartile ranges [Q1-Q3] were used to describe the data) to explore the risk trend of the gene variable and easily identify groups of gene expression with different risk. A multivariate Cox proportional hazard model was applied with treatment and potential risk factors as covariates, obtaining HRs and their 95% CI. Response rates were compared with the χ 2 test or Fisher exact test, as required. Each analysis was performed with the use of a two-sided 5% significance level and a 95% CI. Association between BIM expression levels and response was evaluated using logistic regression analysis. Association between biomarkers was assessed using a Pearson correlation analysis. The correlation between immunohistochemical and RNA expression analysis has been investigated with the non parametric Mann-Whitney Wilcoxon Two-Sample test; significance was defined at the p < 0.05 level. The statistical analyses were performed using SAS version 9.3 and SPSS version 18.0. The EURTAC study is registered with ClinicalTrials.gov, number NCT00446225.