Abstract
Background/Aim: The acquisition of resistance to epidermal growth factor receptor (EGFR) tyrosine kinase inhibitors (TKIs) remains a major challenge in lung cancer medicine. We sought to identify biomarkers for the early detection of resistance to TKIs. Materials and Methods: Capillary electrophoresis time-of-flight mass spectrometry analysis was performed to identify the metabolic signatures associated with erlotinib resistance in erlotinib-resistant PC-9ER NSCLC cells established from the EGFR-mutant NSCLC cell line PC-9. Results: PC-9ER cells showed metabolic signatures indicative of enhanced glutamine metabolism. Copy number gains in v-myc avian myelocytomatosis viral oncogene homolog (MYC), glutathione-S-transferase theta 2 (GSTT2), gamma-glutamyltransferase 1 (GGT1), and GGT5 were also detected, suggesting that amplification of these genes confers glutamine addiction in PC-9ER cells. Conclusion: Enhanced glutamine metabolism may be a surrogate marker that can be used to predict the likelihood of patients to respond to EGFR-TKIs.
- EGFR-TKI resistance
- metabolomics
- glutamine metabolism
- copy-number alteration
- MYC
- non-small cell lung cancer
The epidermal growth factor receptor (EGFR) tyrosine kinase inhibitors (TKIs) gefitinib and erlotinib are effective against non-small cell lung cancer (NSCLC) harbouring EGFR activating-mutations (1). However, the majority of responders eventually develop resistance within one year of treatment (2, 3). Although the occurrence of genetic alterations, activation of bypass signaling, and phenotypic transformation have been implicated (3), the precise mechanisms responsible for the acquisition of resistance to TKIs remain poorly-understood. Identification of biomarkers that allow early detection of the acquisition of resistance to EGFR-TKIs will significantly reduce morbidity and mortality in patients with NSCLC.
Omics technologies have been widely used to identify novel therapeutic targets and biomarkers in solid tumors (4, 5). Among these technologies, metabolomics capture the endogenous metabolic profile as precise snapshots of the dynamic changes in the overall cellular state (6, 7). A number of variables, including environmental factors, genetic alterations, transcriptional regulation, levels of enzymes, significantly influence metabolic profiles (6, 7). Thus, metabolomics has the potential to deliver more in-depth information regarding the phenotypic and physiological changes than other omics technologies (6, 7). Additionally, metabolic de-regulation, including Warburg-like glycolytic state and increased fatty acid synthesis, as well as glutamine metabolism, is regarded as one of the hallmarks of cancer, and is known to be related to drug resistance (8). Metabolomics has been utilized to predict drug efficacy and toxicity (9).
Chronic myelogenous leukaemia cells that acquire resistance to the BCR (breakpoint cluster region)–ABL (Abelson murine leukemia virus oncogene) tyrosine kinase inhibitor imatinib exhibit elevated glucose uptake and enhanced activity of the pentose phosphate pathway. This phenotype could be used as a marker for early detection of imatinib resistance in BCR–ABL-positive cells (10).
Serial biopsy is a widely adapted method for real-time monitoring of drug efficacy after the initial treatment of tumors. However, this method is highly invasive for monitoring cancer such as lung cancer. Furthermore, such methods require ample yield of tumor cells for precise diagnosis. In contrast, metabolomics allows analysis of metabolites obtained not only directly from tissues, but also from surrogate tissues such as blood, urine, and body fluids collected using minimally invasive procedures (6, 7). Therefore, specific metabolite signatures identified by metabolomics provides unique opportunities for serial and real-time monitoring of cancer with precision.
We previously established erlotinib-resistant PC-9ER cells from PC-9 non-small cell lung cancer (NSCLC) cells carrying EGFR-activating mutation (11). Its drug resistance was due to the constitutive activation of v-crk avian sarcoma virus CT10 oncogene homolog-like (CRKL)/ phosphatidylinositol 3-kinase (PI3K)/v-akt murine thymoma viral oncogene homolog (AKT) induced by CRKL amplification (12). We also found significant copy number gains in genomic regions, including five genes associated with glutamine and glutathione metabolic pathways–v-myc avian myelocytomatosis viral oncogene homolog (MYC) (8q24.12-q24.21), glutathione-S-transferase theta 2 (GSTT2), glutathione-S-transferase theta-1 (GSTT1), gamma-glutamyltransferase-5 (GGT5), and gamma-glutamyltransferase-1 (GGT1) (22q11.21-q12.1 (Figure 1A) in PC-9ER cells (12). These results suggest that there are changes in the levels of metabolites in PC-9ER cells.
In the current study, using capillary electrophoresis time-of-flight mass spectrometry (CE-TOF-MS) (13), we attempted to identify the metabolic profiles relevant to EGFR-TKI resistance of erlotinib-resistant PC-9ER cells, and examined the association between the metabolic profile and copy-number alterations by integrating the data from metabolomics and genomics.
Materials and Methods
Cell culture and reagents. The human lung adenocarcinoma cell line PC-9 harbouring an EGFR-activating mutation (E746-A750del at exon 19) was provided by Dr. Fumiaki Koizumi (National Cancer Center Hospital, Tokyo, Japan). Erlotinib-resistant PC-9ER cells (PC-9ER1 and PC-9ER4) were derived from PC-9 cells as reported elsewhere (11). All cell lines used in this study were cultured in RPMI-1640 (Invitrogen, Carlsbad, CA, USA) supplemented with 10% heat-inactivated foetal bovine serum (FBS; Invitrogen) and maintained at 37°C in a humidified atmosphere of 5% CO2.
Nucleic acid sample preparation. Genomic DNA from PC-9 and PC-9ER cells was extracted using the QIAamp DNA mini kit (Qiagen, Hilden, Germany) according to the manufacturer's instructions. DNA concentrations were determined with the help of a double-strand DNA quantification kit (Quant-iT™ PicoGreen dsDNA Assay kit; Invitrogen). The concentration of DNA in samples was adjusted to 1 ng/μl.
Copy number analysis. The relative copy numbers of MYC, GSTT2, GSTT1, GGT5, and GGT1 were estimated by employing quantitative real-time polymerase chain reaction (qPCR) on a StepOnePlus™ Real-time PCR system (Applied Biosystems, Foster City, CA, USA). The reaction mixture contained 2 ng genomic DNA, PCR primers for each gene, and SYBR® Premix Ex Taq™ II (Tli RNaseH Plus) (Takara Bio, Otsu, Shiga, Japan). To generate the standard curve for the quantification of target gene copies, serial dilutions of the PCR amplicons of each gene were used (copy numbers from 10 to 107). The copy number of each gene was normalized against Long interspersed nuclear element 1 (LINE1) and human genomic DNA (Clontech, Palo Alto, CA, USA). The following primer sequences were used: LINE1 (forward: 5’-AAAGCCGCTCAACTACATGG-3’; reverse: 5’-TGCTTTG AATGCGTCCCAGAG-3’), MYC (forward: 5’-CGAGAAGCC GCTCCACATA-3’; reverse: 5’-TGCTTTGAATG CGTCCCAGAG-3’), GSTT2 (forward: 5’-AGCTCGGCCAT CCTGATT-3’; reverse: 5’-TCTCCAGATGGCTCTCCTCAC-3’), GSTT1 (forward: 5’-GACGCGCAAATATAAGGTCCC-3’; reverse: 5’-CCACATT CCCAGCCTCACC-3’), GGT5 (forward: 5’-AGCGACTCTGTG CCATGTCC-3’; reverse: 5’-GCCCACCTGTTG TCACATTGT-3’) and GGT1 (forward: 5’-GGGAGATCCGA GGCTATGA-3’; reverse: 5’-GATGACGGTCCGCTTGTT-3’).
Sample collection for the extraction of metabolites. Cells (1.8×106 cells/well) were seeded into 100-mm dishes (Corning costar, Cambridge, MA, USA) and incubated overnight. After replacing the spent culture medium with fresh medium, cells were incubated for an additional 8 h and harvested. Subsequent sample preparation for CE-TOF-MS analysis was performed according to the methods of Urakami et al. (14).
CE-TOF-MS and data analysis. CE-TOF-MS and data analysis were performed following the methods of Urakami et al. (14). Metabolite profiles were evaluated using Agilent CE Capillary electrophoresis System (Agilent Technologies, Palo Alto, CA, USA). Cationic and anionic metabolites were analysed with the help of an HMT Metabolomics Solution Package (Human Metabolome Technologies Inc., Tsuruoka, Yamagata, Japan), and quantified using Master-Hands software (Keio University, Tsuruoka, Yamagata, Japan).
Statistical analyses. Welch's t-test was performed to detect metabolites whose concentrations differed significantly between PC-9 and PC-9ER cells. A p-value that was lower than 0.01 indicated statistically significant differences. Multivariate statistical investigation with orthogonal partial least-squares-discriminant analysis (OPLS-DA) was performed with the help of SIMCA-P+ software (v12.0.1.0; Umetrics, Umeå, Sweden). The quality of the OPLS-DA model was evaluated by the explained parameter (R2) and the predictive parameter (Q2). Values of 0.5 indicated acceptable OPLS-DA model (15).
Results
Amplification of MYC, GSTT2, GGT5, and GGT1 relevant to glutamine and glutathione metabolism in PC-9ER cells. Since results of array comparative genomic hybridization detected a significant copy number gain in the genomic regions, including MYC (8q24.12-q24.21), GSTT2, GSTT1, GGT5, and GGT1 (22q11.21-q12.1) in PC-9ER cells (Figure 1A), we performed qPCR analysis to confirm these gene copy number gains (Figure 1B) (12). Significant gains in the copy numbers of MYC, GSTT2, GGT5, and GGT1, but not that of GSTT1, were detected in PC-9ER cells (Figure 1B). These results suggest that there is enhanced metabolism of glutamine and glutathione in PC-9ER cells.
A: Array comparative genomic hybridization karyograms of the commonly detected significant amplifications 22q11.21-q12.1 in PC-9ER1 and PC-9ER4 cells. Regions of copy number gain and loss are indicated by yellow and blue squares, respectively. Black arrows indicate the locations of glutathione-S-transferase theta 2 (GSTT2), glutathione-S-transferase theta 1 (GSTT1), gamma-glutamyltransferase 5 (GGT5) and gamma-glutamyltransferase 1 (GGT1). B: Copy numbers of v-myc avian myelocytomatosis viral oncogene homolog (MYC), GSTT2, GSTT1, GGT5 and GGT1 in PC-9 and PC-9ER cells. The copy number of each gene was normalized against that of long interspersed nuclear element 1 (LINE1) and human genomic DNA. Copy numbers of MYC, GSTT2, GGT5 and GGT1, but not of GSTT1, in PC-9ER cell lines were significantly higher than those in PC-9 cells (p<0.01, Student's t-test).
Results of multivariate statistical investigation with orthogonal partial least-squares-discriminant analysis (OPLS-DA). A: OPLS-DA score plots. The explained parameter R2 and the predictive ability Q2 were 0.9999 and 0.9994, respectively, indicating a reliable OPLS-DA model. “1” and “2” in each cell line correspond to the “Set 1” and “Set 2”, respectively, shown in Table I. B: Loading S-plots were derived from metabolic profiles of each cell line. Twenty-four metabolites (listed in Table II) indicated by red and blue spheres with p(corr)<|0.9| were selected and were considered to have strongly contributed to the differences between PC-9 and PC-9ER cells.
Comparison of metabolic profiles of PC-9 and PC-9ER cells. Results from metabolic profiling of PC-9 and PC-9ER cells with CE-TOF-MS identified the metabolites responsible for the differences between PC-9 and PC-9ER cells (Figure 2). The metabolites detected by CE-TOF-MS analysis are presented in Table I. The OPLS-DA score plot shows two significant components corresponding to PC-9 and PC-9ER cells (Figure 2A). The OPLS-DA loading S-plot, a plot of the covariance versus the correlation in conjunction with the variable trend plots, allows for easier identification of significant metabolites that contribute to the differences between PC-9 and PC-9ER cells (Figure 2B). Twenty-four metabolites with p(corr)>|0.9| were selected and were considered to have strongly contributed to the differences between PC-9 and PC-9ER cells (Figure 2B, Table II). Among the 24 metabolites, 18 contributed significantly (Table II). Compared to that in PC-9 cells, the concentration of glucose 6-phosphate, the first metabolite of glycolysis, and those of subsequent metabolites, including fructose 6-phosphate and fructose 1,6-bisphosphate, were significantly lower in PC-9ER cells (Figure 3) (Table I). In contrast, the concentration of metabolites derived from glutamine, including glutamate, proline, γ-aminobutyric acid, ornithinine, and citrulline were significantly higher in PC-9ER cells than in PC-9 cells (Figure 3) (Table II). Additionally, the difference in the glutamate concentration between PC-9 and PC-9ER was more pronounced than those of other metabolites (Table II). These results clearly indicated that when compared with glycolysis, glutamine metabolism was enhanced in PC-9ER cells.
Pathway map of metabolites from glycogenesis, glycolysis, tricarboxylic acid (TCA) cycle and glutamine metabolism. Metabolites in PC-9ER cells showing significant increase or decrease compared to those in PC-9 cells are shown in red and blue, respectively. Metabolites not detected in the present analysis are shown in parentheses. Red arrowheads indicate genes with copy gains. Blue arrowheads indicate genes reported to be under MYC regulation.
Concentration of metabolites detected in each cell line.
Discussion
Tumor cells expressing high levels of the transcriptional regulator MYC depend on glutamine metabolism for proliferation (16, 17). This process is mediated by MYC-induced expression of glutamine transporters, solute carrier family 1 member 5 [SLC1A5 (ASCT2)], solute carrier family 38, member 5 [SLC38A5 (SN2)], and GLS encoding glutaminase that are involved in the conversion of glutamine to glutamate (18). This MYC-induced metabolic signature of “glutamine addiction” (19) was also found in erlotinib-resistant PC-9ER cells (Figure 3), suggesting that components of glutamine metabolism may be a useful therapeutic target and this metabolic profile may be used as a biomarker to predict the acquisition of resistance to EGFR-TKIs. In addition to MYC gene amplification, PC-9ER cells had gene copy number gains in GSTT2, GGT5, and GGT1–genes that are relevant in the metabolism of glutathione derived from glutamate (Figure 1B) (12). GSTT2 is responsible for catalysing the glutathione conjugation, a major step in detoxification processes (Figure 3) (20). GGT5 and GGT1 play key roles in the hydrolysis of glutathione conjugate, leading to the production of glutamate (Figure 3) (21). Therefore we conclude that gene amplifications of GSTT2, GGT5, and GGT1 also contribute to the high concentration of glutamate seen in PC-9ER cells and to the dependency of PC-9ER cells on glutamine metabolism for energy production (Figure 1B, Figure 3).
Based on our results, it appears that the metabolic pathways responsible for glutamine addiction could be useful targets for cancer therapeutics. Previous studies have shown that therapeutic targeting of metabolic pathways improves the response to cancer therapeutics (8). Inhibition of glycolysis by 2-deoxy-D-glucose has been reported to enhance the sensitivity of NSCLC cells harbouring EGFR T790M mutation to irreversible EGFR-TKIs (22). However, anti-glutaminolysis agents such as phenylacetate, L-γ-glutamyl-p-nitroanilide, 6-diazo-5-oxo-L-norleucine, and azaserine have been developed (19, 23, 24), the majority of these compounds had toxic side-effects (19, 23). Development of improved inhibitors of pathways responsible for glutamine metabolism is much awaited. The mammalian target of rapamycin (mTOR) is recognized as a therapeutic target in order to attenuate the effects of enhanced glutamine metabolism (25). We previously reported that the dual PI3K/mTOR inhibitor NVP-BEZ235 effectively inhibited the growth of PC-9ER cells by suppressing the activation of the PI3K/AKT signalling pathway induced by CRKL amplification in PC-9ER cells (12). One possibility is that the inhibitory effect of NVP-BEZ235 on glutaminolysis-induced mTOR activation also affected the growth of PC-9ER cells. Therefore, the evaluation of the effects of mTOR inhibitors on tumours with ‘glutamine addiction’ may pave the way for the development of promising cancer therapeutics.
Twenty-four metabolites that contributed to the difference between PC-9 and PC-9ER cells.
Metabolites that can be detected in samples collected using minimally invasive procedures are of particular interest in diagnostics (6, 7). This is mainly because the existing methods for obtaining adequate amounts of tissue for the diagnosis of lung cancer are primarily invasive.
Our study demonstrates that the integration of metabolomics and genomics data is highly effective in elucidating the molecular mechanisms responsible for complex processes such as the acquisition of drug resistance. Cavill et al. demonstrated that the integration of analysis of the metabolome and transcriptome allows the detection of metabolic pathways associated with platinum sensitivity (26). Multi-omics analyses, a combination of metabolomics and other omics technologies, are a highly effective approach not only for biomarker discovery, but also for uncovering unidentified processes in cancer biology.
A limitation in our study is that the metabolic profile of PC-9ER cells with CRKL-amplification-induced EGFR-TKI resistance was analyzed. To examine the clinical relevance of our findings, metabolic profiles of NSCLC cells collected from patients with EGFR-TKI resistance, as well as those from other instances of EGFR-TKI resistance such as EGFR T790M mutation, need to be analyzed.
To our knowledge, this is the first report to show the specific metabolic profiles of erlotinib-resistant NSCLC cells established from EGFR-mutant NSCLC cells. Our results indicate ‘glutamine addiction’ caused by genetic alterations in erlotinib-resistant NSCLC cells. Profiling of glutamine metabolism may be used as a surrogate marker to identify patients who are unlikely to respond to EGFR-TKIs.
Acknowledgements
This work was supported by JSPS KAKENHI Grant Number 25871225 (M.S.). The Authors thank Yoko Masuda, Junko Suzuki, and Akane Naruoka for technical assistance.
Footnotes
-
Conflicts of Interest
The Authors declare that no conflicts of interest exist.
- Received March 6, 2014.
- Revision received April 10, 2014.
- Accepted April 11, 2014.
- Copyright© 2014 International Institute of Anticancer Research (Dr. John G. Delinassios), All rights reserved
References
In this issue
Jump to section
Related Articles
Cited By...
- No citing articles found.