Metabolic and transcriptional profiling reveals pyruvate dehydrogenase kinase 4 as a mediator of epithelial-mesenchymal transition and drug resistance in tumor cells
© Sun et al.; licensee BioMed Central Ltd. 2014
Received: 31 March 2014
Accepted: 11 September 2014
Published: 3 November 2014
Accumulating preclinical and clinical evidence implicates epithelial-mesenchymal transition (EMT) in acquired resistance to anticancer drugs; however, mechanisms by which the mesenchymal state determines drug resistance remain unknown.
To explore a potential role for altered cellular metabolism in EMT and associated drug resistance, we analyzed the metabolome and transcriptome of three lung cancer cell lines that were rendered drug resistant following experimental induction of EMT. This analysis revealed evidence of metabolic rewiring during EMT that diverts glucose to the TCA cycle. Such rewiring was at least partially mediated by the reduced expression of pyruvate dehydrogenase kinase 4 (PDK4), which serves as a gatekeeper of the TCA cycle by inactivating pyruvate dehydrogenase (PDH). Overexpression of PDK4 partially blocked TGFβ-induced EMT; conversely, PDK4 inhibition via RNAi-mediated knockdown was sufficient to drive EMT and promoted erlotinib resistance in EGFR mutant lung cancer cells. We identified a novel interaction between PDK4 and apoptosis-inducing factor (AIF), an inner mitochondrial protein that appears to play a role in mediating this resistance. In addition, analysis of human tumor samples revealed PDK4-low as a predictor of poor prognosis in lung cancer and that PDK4 expression is dramatically downregulated in most tumor types.
Together, these findings implicate PDK4 as a critical metabolic regulator of EMT and associated drug resistance.
Acquired drug resistance has emerged as a major challenge to effective cancer therapy. Tumor-targeted therapies, such as anti-EGFR treatments for lung cancer and anti-HER2 treatments for breast cancer, are highly effective in disease management in biomarker-defined patient populations. However, the vast majority of these treated patients’ cancers ultimately become drug resistant due to the presence of a subpopulation(s) of malignant cells that survive the therapy and repopulate the relapsed tumor, thereby limiting the long-term effectiveness of these targeted agents .
Epithelial-mesenchymal transition (EMT) is a biological process during which epithelial cells lose the expression of epithelial markers and gain the expression of mesenchymal markers, imparting cells with distinct morphological, migratory and other properties. EMT has frequently been observed in drug-resistant cancer cells in both preclinical models and clinical samples , including resistance to anti-EGFR therapy in lung cancer [3, 4], resistance to androgen-deprivation therapy in prostate cancer , and resistance to chemotherapy in breast cancer . In addition to its role in drug resistance, EMT also converts epithelial cancer cells into a more metastatic and “stem-like” state .
Altered metabolism has been recognized as a hallmark of cancer . Tumor cells need to consume more glucose and glutamine to satisfy the energy demand and biosynthesis requirement for rapid proliferation. Although considered as auxiliary to tumorigenesis for many years, recent studies have revealed oncogenic mutations in the core metabolism network that could potentially drive tumor initiation and progression . In addition, accumulating evidence has revealed that altered metabolism is essential for activated oncogenes and inactivated tumor suppressors to drive malignant transformation . Considering the profound changes in cellular metabolism in cancer, we hypothesized that tumors with acquired drug resistance might resort to alternative mechanisms to fuel their growth and survival under drug treatment. In this study, we report a diversion of glucose metabolism towards the TCA cycle during EMT in cancer cells. We also provide evidence that pyruvate dehydrogenase kinase 4 (PDK4) is a critical regulator of EMT and associated drug resistance.
Antibodies and other reagents
Antibodies used are listed as follows: E-cadherin (Cell Signaling, 24E10), N-cadherin (Cell Signaling, no. 4061); Vimentin (Cell Signaling, D21H3), GAPDH (Cell Signaling, 14C10), Zeb1 (Cell Signaling, D80D3), Snail (Cell Signaling, C15D3), VDAC (Cell Signaling, D73D12), apoptosis-inducing factor (AIF) (Cell Signaling, D39D2), and FLAG (Sigma, M2). PDK4 polyclonal antibody was generated by immunizing rabbits against a peptide corresponding to amino acids 264–277 of human PDK4 (VEHQENQPSLTPIE) (Yenzyme).
Unless otherwise specified, all cell lines were cultured in RPMI1640 (Gibco) containing 2 g/l sodium bicarbonate, 10% FBS, 2 mM P/S, and 4 mM glutamine (Gibco). To generate the TGFβ-induced mesenchymal cells, cells were cultured in the presence of 2 ng/ml TGFβ (Cell signaling) for 2 to 5 weeks. A549 cells were cultured with TGFβ for 2 to 3 weeks. HCC827 and NCI-H358 cells were cultured with TGFβ for 3 to 5 weeks prior to experiments. Fresh media containing TGFβ was replenished every 3–4 days. The day before the experiment, cells were re-seeded in the appropriate plates without TGFβ. Cell lines are maintained by a core facility at Genentech that routinely uses STR fingerprinting to verify cell line identity.
Unless otherwise specified, all data plotting and statistical analysis was performed using Prism Graph Pad 5.0, and the error bars represent SEM. Student's t test was used to assess the statistical significance of the differences between groups (two-tail *p value <0.05; two-tail **p value <0.01.
Survival analyses were performed with the Kaplan-Meier method and Cox proportional-hazard model. Results across the three data sets (GSE42127, GSE8894, and GSE3141) were combined in a meta-analysis, using the R package meta. The overall combined estimate of the hazard ratio was obtained from their values and standard errors in the individual data sets.
PDK4 expression data in normal lung, lung adenocarcinoma and squamous cell carcinoma of the lung was generated from TCGA RNA-seq data, which was obtained from the Cancer Genomics Hub at UC Santa Cruz and preprocessed and aligned with HTSeqGenie . PDK4 expression data in multiple cancer indications was from the Gene Logic database of microarray data using GeneChip human genome U133 Plus 2.0 array (Affymetrix). Expression summary values for all probe sets were calculated using the RMA algorithm as implemented in the affymetrix package from Bioconductor.
Global metabolomic profiling
The parental and TGFβ-induced mesenchymal cells were rinsed with PBS, scraped in PBS, and spun down. The cell pellets were snap-frozen and submitted to Metabolon Inc for global metabolomic analysis . Briefly, a combination of GC-MS and LC-MS methods were used, and each metabolite amount was normalized to total protein amount of the individual cell pellets. Each sample consisted of cells collected from two 15-cm plates at approximately 60% confluence, and each condition included five replicates.
Glycolysis/OXPHOS ratio measurement
Real-time Glycolysis/OXPHOS rate was measured using the Seahorse metabolic analyzer, following manufacturer's protocols. Briefly, cells were plated in six replicates in 96-well Seahorse assay plates. The seeding cell numbers were adjusted based on cell growth rate, with the goal to reach similar cell density at the time of the real-time measurement. The next day, cells were washed twice and incubated in 100 μl of modified RPMI1640 growth media for 2 h. The modified RPMI1640 growth media did not contain sodium bicarbonate, and contained dialyzed FBS (Gibco) instead of standard FBS. Proton production rate (PPR) and oxygen consumption rate (OCR) were recorded.
Mass isotopologue distribution analysis using C-13 stable isotopes
Cells were plated in a 15-cm plate overnight, and then switched to tracing media. The tracing media was based on standard RPMI1640 growth media containing 10% dialyzed FBS, with either glutamine substituted by 13C-U5-glutamine or glucose substituted by 13C-U6-glucose (Cambridge Isotope). After being cultured in the tracing media for 24 h, cells were harvested and processed for mass spectrometry. A detailed description of the mass spectrometry analysis is provided in ‘Extended Methods.’
Microarray gene expression analysis
Gene expression profiling comparing TGFβ-treated mesenchymal cells and corresponding parental cells was performed using GeneChip human genome U133 Plus 2.0 array (Affymetrix), following standard protocols. Data were normalized using the R package RMA from Bioconductor and analyzed with the R limma package. The expression microarray data has been deposited in the Gene Expression Omnibus (GEO) database under accession number GSE49644.
Description of additional methods is provided in Additional file 1.
Experimentally-induced EMT in lung cancer cell lines is associated with metabolic reprogramming
Consistent with previous studies implicating EMT in drug resistance, the derived mesenchymal cells became significantly less sensitive to targeted therapies. The mutant EGFR-driven HCC827 parental cells were very sensitive to erlotinib, whereas the TGFβ-induced mesenchymal derivatives had lost their EGFR ‘addiction’ and became largely resistant to erlotinib (Figure 1C). Likewise, the mesenchymal derivatives of A549 and NCI-H358 cells became significantly less sensitive to the MEK inhibitor GDC-0973 (Figure 1C) and KRAS inhibition via RNAi (see Figure three F in reference ), respectively. Taken together, these findings demonstrate an EMT upon chronic TGFβ treatment of these ‘oncogene-addicted’ lung cancer cells and provide a model system for studying the role of EMT in mediating drug resistance in vitro.
In the course of culturing these cells, we observed slower acidification of the media in the mesenchymal derivatives relative to the corresponding untreated parental cell lines, suggesting a possible metabolic rewiring event during EMT. A switch from oxidative phosphorylation (OXPHOS) to aerobic glycolysis, the so-called Warburg effect, is a key feature of many cancer cells . Whereas glycolysis is a relatively rapid mechanism by which cells can produce ATP, OXPHOS is much more energy efficient as it produces more ATP per glucose molecule. We therefore measured real-time glycolysis and OXPHOS levels in these cells and observed a consistent decrease in the glycolysis/OXPHOS ratio upon EMT in all three cell lines (Figure 1D). Moreover, this glycolysis-to-OXPHOS switch was not restricted to TGFβ-induced EMT: EGFR-mutant HCC4006 cells that had been selected for resistance to erlotinib and which gained mesenchymal features , similarly, exhibited a decreased glycolysis/OXPHOS ratio relative to their parental, erlotinib-sensitive counterpart (Additional file 2: Figure S1A). These collective data reveal a metabolic remodeling process commensurate with EMT.
Metabolomic profiling reveals amino acid accumulation associated with EMT
The observed glycolysis-to-OXPHOS switch during EMT prompted us to examine the global metabolomic profiles of these cells before and after EMT using mass spectrometry (Additional file 3: Table S1). Using an absolute fold-change >2 and a p value <0.05 as the cutoff, we observed only ten metabolites (out of approximately 400 known metabolites measured) that showed consistently significant changes during EMT across all three cell lines (Additional file 2: Figure S1B). Among them, five metabolites—glutamine, tryptophan, reduced glutathione, 4-guanidinobutanoate, and cysteinylglycine—belong to the amino acid or peptide category. We further examined the levels of the 20 standard amino acids before and after EMT and found an overall increase in amino acid levels in the mesenchymal cells in all three models (Additional file 2: Figure S1C).
We were particularly intrigued by the observed changes in glutamine (Figure 1E) and glutamate (Figure 1F) due to the direct interconversion between these two metabolites. The accumulation of intracellular glutamate was not due to decreased glutamate efflux. In fact, we observed a significant (p < 0.01) increase in glutamate secretion across all three mesenchymal cell populations relative to their corresponding parental controls (Figure 1G). Notably, although the TGFβ-treated cells grew about two fold slower than their parental counterparts (Additional file 2: Figure S1D), it is unlikely that the observed glutamate accumulation is simply a consequence of reduced proliferation rate since metabolomic profiling of the NCI-60 cell line panel did not reveal any association between glutamate accumulation and cell proliferation rate . Together, these findings demonstrate glutamate accumulation associated with the conversion of epithelial cancer cells to a mesenchymal state.
Glucose diversion to the TCA cycle contributes to increased intracellular glutamate in mesenchymal cells
PDK4 is downregulated in mesenchymal cells and regulates glucose contribution to the TCA cycle
To further explore the molecular mechanisms underlying the observed metabolic rewiring upon EMT, we compared microarray gene expression profiles of A549, HCC827, and NCI-H358 cells before and after EMT. Using an absolute fold-change >5 and a FDR-adjusted p value <0.00001 as the cutoff, we observed approximately 30 metabolism-associated genes that consistently demonstrated differential expression in all three mesenchymal and parental cell line comparisons (Additional file 4: Figure S2A). Among them, PDK4, but not the other related PDKs (PDK1-3), was consistently and dramatically downregulated in all three mesenchymal cell line models (Figure 2B).
The pyruvate dehydrogenase (PDH) complex is a mega-protein complex that functions at the interface of glycolysis and the TCA cycle, critically controlling entry into the TCA cycle by catalyzing the conversion of pyruvate to acetyl-CoA, thereby supplying acetyl-CoA to the TCA cycle . The regulatory mechanisms for PDH activity include its inactivation by four kinases, PDK1-4, and its activation by two phosphatases, PDP1-2 . Decreased expression of PDK4 would therefore be expected to activate PDH and divert glucose to the TCA cycle, consistent with the observed diversion of glucose to TCA cycle/glutamate (Figure 2A).
To validate the microarray findings, we confirmed the reduction in PDK4 expression in TGFβ-treated mesenchymal cells at both the mRNA (Figure 2C) and protein (Figure 2D) levels. Of note, we were only able to reliably detect endogenous PDK4 protein using isolated mitochondria preparations. In addition to lung cancer cell lines, we examined the pancreatic cancer cell line PANC-1 as well as the mammary cell line MCF10A and its oncogenically transformed derivative MCF10AT. All of these cell lines have previously been reported to undergo TGFβ-induced EMT [22, 23] and similarly demonstrated decreased PDK4 expression following TGFβ treatment (Additional files 4: Figure S2B, C). We also analyzed the time-course of TGFβ treatment and observed decreased PDK4 mRNA levels as early as 30′ after TGFβ treatment (Additional files 4: Figure S2D, E), suggesting that PDK4 is regulated by TGFβ signaling at the mRNA level and that PDK4 downregulation is likely an early event in the induction of EMT by TGFβ.We next examined whether the decreased expression of PDK4 may be relevant in other models in which EMT was not directly induced by TGFβ. A549 cells that had been rendered resistant to a combination of MEK and PI3K inhibitors (GDC-0973 and GDC-0941, respectively; Shiuh-Ming Luoh and Marcia Belvin, unpublished data) displayed hallmark features of EMT, including increased expression of Vimentin and N-cadherin as well as decreased expression of E-cadherin (Figure 2E). In this model, we similarly observed dramatically reduced PDK4 expression in the resistant, mesenchymal derivatives (Figure 2E).
Finally, we re-introduced PDK4 into the mesenchymal derivative of the HCC827 cells and found that PDK4 overexpression completely reversed glucose diversion to the TCA cycle/glutamate (Figure 2F), indicating that PDK4 loss is indeed responsible for the diversion of glucose to the TCA cycle. Consistent with PDK4’s functional role in such metabolic rewiring, we also anticipated and observed increased PDH activity in the mesenchymal cells as shown by the increased contribution of pyruvate to the TCA cycle/glutamate (Figure 2G). This diversion of pyruvate to the TCA cycle/glutamate was similarly rescued following PDK4 reintroduction into the mesenchymal derivatives (Figure 2G). In contrast, we did not observe any changes in glucose contribution to pyruvate or lactate in the mesenchymal derivatives of HCC827 cells (Additional file 4: Figure S2F). Taken together, these findings strongly implicate the PDH regulatory component PDK4 in EMT.
PDK4 expression partially prevents TGFβ-induced EMT
Inhibiting PDK4 drives an EMT associated with erlotinib resistance
Next, we asked whether blocking PDK4 was sufficient to drive EMT. We specifically knocked down PDK4 using a siRNA smart pool. As shown in Figure 3B, we achieved efficient PDK4 knockdown in A549 and HCC827 cells at both the mRNA and protein level. Notably, in A549 cells, PDK4 knockdown decreased the expression of E-cadherin and modestly increased the expression of Vimentin and Zeb1. In HCC827 cells, PDK4 knockdown markedly increased the expression of Vimentin and Zeb1, although it had no effect on E-cadherin (Figure 3B). To rule out potential off-target effects of the siRNA pool, we performed a deconvolution analysis and found that at least three single siRNAs targeting PDK4 efficiently decreased PDK4 mRNA levels (Additional file 5: Figure S3A) and E-cadherin in A549 cells (Additional file 5: Figure S3B). These same single siRNAs also increased Vimentin and Zeb1 in HCC827 cells (Additional file 5: Figure S3C, H). To determine whether downregulation of other PDKs could also affect EMT, we individually knocked down PDK1-PDK3 using their respective siRNA smart pool. Under conditions of comparable knockdown efficiency (Additional file 5: Figure S3D), while knockdown of these other PDKs led to a modest effect on the expression of EMT markers, PDK4 knockdown clearly caused the most pronounced effect on the expression of classical EMT markers in A549 (Additional file 5: Figure S3E) and HCC827 (Additional file 5: Figure S3F) cells.
Accumulating evidence has linked EMT to drug resistance [2–6]. Consistent with these findings, HCC827 cells that have been experimentally induced to undergo EMT demonstrate erlotinib resistance (Figure 1C), and conversely, HCC827 cells that have been selected for erlotinib resistance through chronic drug exposure gain mesenchymal features . To address a potential role for PDK4 in EMT-associated erlotinib resistance, we performed PDK4 RNAi studies and examined colony formation capacity in the presence of erlotinib in mutant EGFR-driven HCC827 cells. In the absence of erlotinib, PDK4 knockdown had no inhibitory effect on cell growth (Figure 3C). In contrast, PDK4 knockdown significantly promoted colony formation in the presence of erlotinib (Figure 3C and Additional file 5: Figure S3G), suggesting that its downregulation enhanced erlotinib resistance. This effect was also most pronounced with PDK4 (Additional file 5: Figure S3I) compared to other PDKs. Notably, PDK4 knockdown also promoted erlotinib resistance in EGFR-mutant HCC4006 cells (Figure 3D and Additional file 5: Figure S3J), although it inhibited their growth in the absence of erlotinib (Figure 3D). In addition to drug resistance, increased migration and/or invasion are hallmarks of EMT. We examined the migratory and invasive capacity of these cells using scratch wound and boyden chamber assays, respectively, and observed increased migration (Additional file 6: Figure S4A) and invasion (Additional file 6: Figure S4B) in A549 cells following PDK4 knockdown. Collectively, these data strongly suggest that PDK4 inhibition can promote erlotinib resistance and EMT.
Considering PDK4’s role as a metabolic inhibitor of PDH that controls entry into the TCA cycle, we examined the effect of siPDK4 on metabolism using the MIDA and Seahorse assays. We observed an increased contribution to glutamate from glucose (Additional file 7: Figure S5A) as well as a modestly decreased glycolysis/OXPHOS ratio (Additional file 7: Figure S5B) in siPDK4 cells, indicating that PDK4 knockdown promotes a metabolic reprogramming event that favors the TCA cycle. Considering that OXPHOS is more efficient in glucose utilization, we hypothesized that siPDK4 would enhance cell survival under low-glucose conditions. Indeed, in A549 cells in which PDK4 had been knocked down, we observed a robust increase in cell survival under low-glucose conditions (Additional file 7: Figure S5C), but only modestly enhanced survival under low-glutamine conditions (Additional file 7: Figure S5D).
AIF interacts with PDK4 and promotes EMT, erlotinib resistance, and ROS production
Consistent with PDK4’s role in mitochondria, AIF is a mitochondrial protein that can translocate to the nucleus and promote caspase-independent apoptosis . Biochemical analysis of AIF has revealed two major functional domains: 1) a DNA binding domain that promotes chromatin condensation and DNA fragmentation and 2) an oxidoreductase domain that is involved in cellular redox metabolism and mitochondrial bioenergetics . Of note, AIF inhibition has also been shown to promote drug resistance to the multi-kinase inhibitor sorafenib and the IκB kinase inhibitor BMS-345541 in melanoma [27, 28].
To establish a potential role for AIF in EMT, we knocked down AIF using a siRNA smart pool and observed EMT in both A549 and HCC827 cells. Similar to PDK4, knockdown of AIF resulted in increased Zeb1 and Vimentin (A549 and HCC827 cells) and decreased E-cadherin (A549 cells) (Figure 4C) and promoted erlotinib resistance (Figure 4D), all of which are consistent with both AIF and PDK4 having a role in EMT. To rule out a potential off-target effect of siAIF on EMT, we deconvoluted the siRNA smart pool in A549 cells and observed that each of the four individual siRNA oligos decreased E-cadherin expression (Additional file 9: Figure S6). Considering AIF’s role in redox regulation, we also measured reactive oxygen species (ROS) levels and detected increased ROS following knockdown of either PDK4 or AIF (Figure 4E). These collective data support a functional requirement for the interaction between AIF and PDK4 in EMT and associated drug resistance.
PDK4-low expression predicts poor prognosis in lung cancer and is frequently down-regulated in human cancer
We further examined PDK4 expression levels in NSCLC biopsies from the Cancer Genome Atlas (TCGA) using RNA-seq data. We observed dramatically decreased PDK4 in lung cancer biopsies compared to the corresponding normal tissue (Figure 5B). Of the two different NSCLC subtypes, adenocarcinoma (ADC) and squamous cell carcinoma (SCC), we found that PDK4 expression was particularly low in SCC, a subtype lacking good treatment options.
To explore PDK4 expression levels in a broad range of cancer types, we surveyed Gene Logic microarray data covering multiple types of human tumor biopsies and normal tissues. Most notably, we observed dramatically decreased PDK4 expression in the majority of cancer types examined, including breast, colorectal, lung, lymphoid, ovary, and skin cancers (Figure 5C). Finally, we analyzed the global gene expression changes of 19 cancer types compared to corresponding normal tissue in the Gene Logic database. We ranked the approximately 19,000 genes according to the average fold-change and found that PDK4 was one of the genes that showed the most dramatic expression loss in cancer (ranked no. 28 overall, data not shown).
Cancer cells preferably use aerobic glycolysis to generate energy, which has been recognized as a hallmark of cancer. Our findings reveal a metabolic rewiring event that drives cancer cells back to an OXPHOS state during the process of EMT or during the acquisition of drug resistance. This observation is consistent with three recent studies: Firstly, Haq et al. reported that BRAF inhibitor-resistant cells are more addicted to OXPHOS ; secondly, Roesch et al. demonstrated that the multi-drug resistant JARID1Bhigh subpopulation of melanoma cells expressed more OXPHOS enzymes ; and most recently, Viale et al. showed that pancreatic tumor cells surviving oncogene ablation depend on mitochondria . In this study, we analyzed metabolic activity, metabolite profiles, and mass isotopologue distribution to reveal that cancer cells that have undergone EMT divert more glucose to the TCA cycle compared to their parental epithelial cells, which presumably enables the mesenchymal cells to use the metabolites of the TCA cycle as the backbone to produce more amino acids. We further speculate that the increased supply of macromolecules provides the building blocks for de novo protein synthesis and extracellular matrix remodeling, which is essential for EMT . Additionally, since OXPHOS is a more efficient process for energy production, shifting to OXPHOS might enhance the ability of cancer cells to survive under conditions of stress, such as drug treatment.
Our study reveals PDK4 as a novel metabolic regulator of EMT and drug resistance. A previous study comparing basal and luminal subtypes of breast cancer demonstrated loss of the metabolic enzyme FBP1 in the more mesenchymal, basal subtype; however, inhibition of FBP1 alone was not sufficient to regulate EMT . Unlike FBP1, we show that inhibition of PDK4 alone is sufficient to induce EMT, and ectopic expression of PDK4 could partially prevent TGFβ-induced EMT, although PDK4 is not differentially expressed between the basal and luminal subtypes of breast cancer (data not shown). As a phenotypic readout of EMT, drug resistance is also enhanced by PDK4 knockdown. In the absence of drug, PDK4 inhibition does not promote, and even inhibits, cell growth in some cell lines; in the presence of erlotinib, PDK4 inhibition dramatically promotes colony formation. This functional profile is reminiscent of a recent study examining MED12, a component of the transcriptional MEDIATOR complex that regulates TGFβ receptor II . Similar to PDK4 inhibition, MED12 inhibition impeded cell growth in the absence of drug, but promoted colony formation in the presence of drug. Whereas MED12 regulates TGFβ signaling, PDK4 appears to be regulated by TGFβ signaling.
Four isoforms of PDK (PDK1-PDK4) with similar structure negatively regulate the activity of PDH. PDK4 may have some unique functions and may also be subject to distinct regulatory mechanisms from those affect PDK1-3. PDK4 expression has been shown to be regulated by various metabolic stimuli (such as starvation, exercise, and diabetes), the transcription factors FOXO and E2F1 , and epigenetic mechanisms such as promoter methylation and histone acetylation [37, 38]. Although PDK1 and PDK2 have been previously proposed as cancer drug targets [39, 40], our findings suggest that PDK4 may function as a metabolic tumor suppressor. We observed that depletion of PDK4 by RNAi promoted colony formation in EGFR-mutant cell lines upon EGFR inhibition. Conversely, Grassian et al. reported that PDK4 overexpression suppressed the proliferation of normal mammary epithelial MCF10A cells . In that study, increased PDK4 levels and decreased flux through PDH was induced upon extracellular matrix (ECM) detachment, leading to the metabolic impairment of these cells. The fact that PDK4 emerged as a key factor during EMT in our study further establishes the role of this enzyme in regulating ECM/tumor crosstalk and bioenergetics. Significantly, we observed widespread loss of PDK4 expression in tumor cells compared to normal tissue. This loss of PDK4 expression in cancer is more substantial and prevalent than that of most known tumor suppressors, and collectively, these findings suggest that PDK4 may function as a metabolic tumor suppressor, a possibility that would need to be further explored in an in vivo tumorigenesis study.
The mechanism(s) by which PDK4 regulates EMT and drug resistance is not completely clear. We speculate that PDK4 depletion plays a role in the survival of drug-tolerant persisters, but may not be essential for the proliferative expansion of these cells. Furthermore, since PDK4 knockdown induces Zeb1 expression, PDK4 likely functions upstream of the EMT-associated transcription factors. Our data also suggest a novel interaction between AIF and PDK4, both physically and functionally. We show the knockdown of either AIF or PDK4 promoted EMT, erlotinib resistance, and ROS production. Indeed, AIF inhibition has been previously shown to promote resistance to serum deprivation-induced cell death in embryonic stem cells  as well as to a multi-kinase inhibitor and an IκB kinase inhibitor in melanoma [27, 28]. A hypothesis that remains to be tested is that PDK4 or AIF inhibition promotes EMT and drug resistance through ROS generation, especially given the established role of ROS in physiological conditions [43, 44].
Our collective observations implicate a glycolysis-to-OXPHOS shift in drug-resistant cancer cells that have undergone EMT. We provide evidence that downregulation of PDK4 is responsible for such metabolic rewiring, is sufficient to drive EMT, and promotes erlotinib resistance in EGFR mutant lung cancer cells. In addition, we have identified a novel PDK4-interacting protein, AIF, which plays a role in EMT and drug resistance. Finally, analysis of human lung adenoma tumor samples reveals PDK4-low as a predictor of poor prognosis, consistent with PDK4’s role in EMT. Although establishing the precise mechanism by which PDK4 regulates EMT will require further investigation, the findings described here implicate a specific metabolic reprogramming event in EMT associated with resistance to cancer treatments, thereby revealing novel potential therapeutic opportunities.
We thank Shiuh-ming Luoh and Marcia Belvin for providing the A549 MEK/PI3K inhibition-resistant cell line. We thank Xiaofen Ye and Robert Yauch for providing the HCC4006 erlotinib-resistant cell line. We also thank Zora Modrusan for microarray services, David Peterson and Marie Evangelista for assistance with Seahorse studies, James Kiefer for structural biology support, Peng Yue for data analysis services, and the Genentech FACS laboratory for service support.
- Lackner MR, Wilson TR, Settleman J: Mechanisms of acquired resistance to targeted cancer therapies. Future Oncol. 2012, 8: 999-1014. 10.2217/fon.12.86.View ArticlePubMedGoogle Scholar
- Singh A, Settleman J: EMT, cancer stem cells and drug resistance: an emerging axis of evil in the war on cancer. Oncogene. 2010, 29: 4741-4751. 10.1038/onc.2010.215.View ArticlePubMedPubMed CentralGoogle Scholar
- Sequist LV, Waltman BA, Dias-Santagata D, Digumarthy S, Turke AB, Fidias P, Bergethon K, Shaw AT, Gettinger S, Cosper AK, Akhavanfard S, Heist RS, Temel J, Christensen JG, Wain JC, Lynch TJ, Vernovsky K, Mark EJ, Lanuti M, Iafrate AJ, Mino-Kenudson M, Engelman JA: Genotypic and histological evolution of lung cancers acquiring resistance to EGFR inhibitors. Sci Transl Med. 2011, 3: 75ra26-View ArticlePubMedPubMed CentralGoogle Scholar
- Yauch RL, Januario T, Eberhard DA, Cavet G, Zhu W, Fu L, Pham TQ, Soriano R, Stinson J, Seshagiri S, Modrusan Z, Lin CY, O'Neill V, Amler LC: Epithelial versus mesenchymal phenotype determines in vitro sensitivity and predicts clinical activity of erlotinib in lung cancer patients. Clin Cancer Res. 2005, 11: 8686-8698. 10.1158/1078-0432.CCR-05-1492.View ArticlePubMedGoogle Scholar
- Sun Y, Wang BE, Leong KG, Yue P, Li L, Jhunjhunwala S, Chen D, Seo K, Modrusan Z, Gao WQ, Settleman J, Johnson L: Androgen deprivation causes epithelial-mesenchymal transition in the prostate: implications for androgen-deprivation therapy. Cancer Res. 2012, 72: 527-536. 10.1158/0008-5472.CAN-11-3004.View ArticlePubMedGoogle Scholar
- Creighton CJ, Li X, Landis M, Dixon JM, Neumeister VM, Sjolund A, Rimm DL, Wong H, Rodriguez A, Herschkowitz JI, Fan C, Zhang X, He X, Pavlick A, Gutierrez MC, Renshaw L, Larionov AA, Faratian D, Hilsenbeck SG, Perou CM, Lewis MT, Rosen JM, Chang JC: Residual breast cancers after conventional therapy display mesenchymal as well as tumor-initiating features. Proc Natl Acad Sci U S A. 2009, 106: 13820-13825. 10.1073/pnas.0905718106.View ArticlePubMedPubMed CentralGoogle Scholar
- Thiery JP, Acloque H, Huang RY, Nieto MA: Epithelial-mesenchymal transitions in development and disease. Cell. 2009, 139: 871-890. 10.1016/j.cell.2009.11.007.View ArticlePubMedGoogle Scholar
- Hanahan D, Weinberg RA: Hallmarks of cancer: the next generation. Cell. 2011, 144: 646-674. 10.1016/j.cell.2011.02.013.View ArticlePubMedGoogle Scholar
- Schulze A, Harris AL: How cancer metabolism is tuned for proliferation and vulnerable to disruption. Nature. 2012, 491: 364-373. 10.1038/nature11706.View ArticlePubMedGoogle Scholar
- Cairns RA, Harris IS, Mak TW: Regulation of cancer cell metabolism. Nat Rev Cancer. 2011, 11: 85-95. 10.1038/nrc2981.View ArticlePubMedGoogle Scholar
- Pau GB, Reeder C, Lawrence J, Degenhardt M, Wu J, Huntley T, Brauer MM: HTSeqGenie: a software package to analyse high-throughput sequencing experiments. Bioconductor package. 2012, 1-12.Google Scholar
- Reitman ZJ, Jin G, Karoly ED, Spasojevic I, Yang J, Kinzler KW, He Y, Bigner DD, Vogelstein B, Yan H: Profiling the effects of isocitrate dehydrogenase 1 and 2 mutations on the cellular metabolome. Proc Natl Acad Sci U S A. 2011, 108: 3270-3275. 10.1073/pnas.1019393108.View ArticlePubMedPubMed CentralGoogle Scholar
- Huang S, Holzel M, Knijnenburg T, Schlicker A, Roepman P, McDermott U, Garnett M, Grernrum W, Sun C, Prahallad A, Groenendijk FH, Mittempergher L, Nijkamp W, Neefjes J, Salazar R, Ten Dijke P, Uramoto H, Tanaka F, Beijersbergen RL, Wessels LF, Bernards R: MED12 controls the response to multiple cancer drugs through regulation of TGF-beta receptor signaling. Cell. 2012, 151: 937-950. 10.1016/j.cell.2012.10.035.View ArticlePubMedPubMed CentralGoogle Scholar
- Shipitsin M, Campbell LL, Argani P, Weremowicz S, Bloushtain-Qimron N, Yao J, Nikolskaya T, Serebryiskaya T, Beroukhim R, Hu M, Halushka MK, Sukumar S, Parker LM, Anderson KS, Harris LN, Garber JE, Richardson AL, Schnitt SJ, Nikolsky Y, Gelman RS, Polyak K: Molecular definition of breast tumor heterogeneity. Cancer Cell. 2007, 11: 259-273. 10.1016/j.ccr.2007.01.013.View ArticlePubMedGoogle Scholar
- Singh A, Greninger P, Rhodes D, Koopman L, Violette S, Bardeesy N, Settleman J: A gene expression signature associated with "K-Ras addiction" reveals regulators of EMT and tumor cell survival. Cancer Cell. 2009, 15: 489-500. 10.1016/j.ccr.2009.03.022.View ArticlePubMedPubMed CentralGoogle Scholar
- Koppenol WH, Bounds PL, Dang CV: Otto Warburg's contributions to current concepts of cancer metabolism. Nat Rev Cancer. 2011, 11: 325-337. 10.1038/nrc3038.View ArticlePubMedGoogle Scholar
- Suda K, Tomizawa K, Fujii M, Murakami H, Osada H, Maehara Y, Yatabe Y, Sekido Y, Mitsudomi T: Epithelial to mesenchymal transition in an epidermal growth factor receptor-mutant lung cancer cell line with acquired resistance to erlotinib. J Thorac Oncol. 2011, 6: 1152-1161. 10.1097/JTO.0b013e318216ee52.View ArticlePubMedGoogle Scholar
- Jain M, Nilsson R, Sharma S, Madhusudhan N, Kitami T, Souza AL, Kafri R, Kirschner MW, Clish CB, Mootha VK: Metabolite profiling identifies a key role for glycine in rapid cancer cell proliferation. Science. 2012, 336: 1040-1044. 10.1126/science.1218595.View ArticlePubMedPubMed CentralGoogle Scholar
- Hiller K, Metallo CM, Kelleher JK, Stephanopoulos G: Nontargeted elucidation of metabolic pathways using stable-isotope tracers and mass spectrometry. Anal Chem. 2010, 82: 6621-6628. 10.1021/ac1011574.View ArticlePubMedGoogle Scholar
- Roche TE, Hiromasa Y: Pyruvate dehydrogenase kinase regulatory mechanisms and inhibition in treating diabetes, heart ischemia, and cancer. Cell Mol Life Sci. 2007, 64: 830-849. 10.1007/s00018-007-6380-z.View ArticlePubMedGoogle Scholar
- Patel MS, Korotchkina LG: Regulation of the pyruvate dehydrogenase complex. Biochem Soc Trans. 2006, 34: 217-222. 10.1042/BST20060217.View ArticlePubMedGoogle Scholar
- Brown KA, Aakre ME, Gorska AE, Price JO, Eltom SE, Pietenpol JA, Moses HL: Induction by transforming growth factor-beta1 of epithelial to mesenchymal transition is a rare event in vitro. Breast Cancer Res. 2004, 6: R215-R231. 10.1186/bcr778.View ArticlePubMedPubMed CentralGoogle Scholar
- Ellenrieder V, Hendler SF, Boeck W, Seufferlein T, Menke A, Ruhland C, Adler G, Gress TM: Transforming growth factor beta1 treatment leads to an epithelial-mesenchymal transdifferentiation of pancreatic cancer cells requiring extracellular signal-regulated kinase 2 activation. Cancer Res. 2001, 61: 4222-4228.PubMedGoogle Scholar
- Wynn RM, Kato M, Chuang JL, Tso SC, Li J, Chuang DT: Pyruvate dehydrogenase kinase-4 structures reveal a metastable open conformation fostering robust core-free basal activity. J Biol Chem. 2008, 283: 25305-25315. 10.1074/jbc.M802249200.View ArticlePubMedPubMed CentralGoogle Scholar
- Chung JH, Rho JK, Xu X, Lee JS, Yoon HI, Lee CT, Choi YJ, Kim HR, Kim CH, Lee JC: Clinical and molecular evidences of epithelial to mesenchymal transition in acquired resistance to EGFR-TKIs. Lung Cancer. 2011, 73: 176-182. 10.1016/j.lungcan.2010.11.011.View ArticlePubMedGoogle Scholar
- Modjtahedi N, Giordanetto F, Madeo F, Kroemer G: Apoptosis-inducing factor: vital and lethal. Trends Cell Biol. 2006, 16: 264-272. 10.1016/j.tcb.2006.03.008.View ArticlePubMedGoogle Scholar
- Panka DJ, Wang W, Atkins MB, Mier JW: The Raf inhibitor BAY 43–9006 (Sorafenib) induces caspase-independent apoptosis in melanoma cells. Cancer Res. 2006, 66: 1611-1619. 10.1158/0008-5472.CAN-05-0808.View ArticlePubMedGoogle Scholar
- Yang J, Amiri KI, Burke JR, Schmid JA, Richmond A: BMS-345541 targets inhibitor of kappaB kinase and induces apoptosis in melanoma: involvement of nuclear factor kappaB and mitochondria pathways. Clin Cancer Res. 2006, 12: 950-960. 10.1158/1078-0432.CCR-05-1220.View ArticlePubMedPubMed CentralGoogle Scholar
- Bild AH, Yao G, Chang JT, Wang Q, Potti A, Chasse D, Joshi MB, Harpole D, Lancaster JM, Berchuck A, Olson JA, Marks JR, Dressman HK, West M, Nevins JR: Oncogenic pathway signatures in human cancers as a guide to targeted therapies. Nature. 2006, 439: 353-357. 10.1038/nature04296.View ArticlePubMedGoogle Scholar
- Lee ES, Son DS, Kim SH, Lee J, Jo J, Han J, Kim H, Lee HJ, Choi HY, Jung Y, Park M, Lim YS, Kim K, Shim Y, Kim BC, Lee K, Huh N, Ko C, Park K, Lee JW, Choi YS, Kim J: Prediction of recurrence-free survival in postoperative non-small cell lung cancer patients by using an integrated model of clinical information and gene expression. Clin Cancer Res. 2008, 14: 7397-7404. 10.1158/1078-0432.CCR-07-4937.View ArticlePubMedGoogle Scholar
- Tang H, Xiao G, Behrens C, Schiller J, Allen J, Chow CW, Suraokar M, Corvalan A, Mao J, White MA, Wistuba II, Minna JD, Xie Y: A 12-gene set predicts survival benefits from adjuvant chemotherapy in non-small cell lung cancer patients. Clin Cancer Res. 2013, 19: 1577-1586. 10.1158/1078-0432.CCR-12-2321.View ArticlePubMedPubMed CentralGoogle Scholar
- Haq R, Shoag J, Andreu-Perez P, Yokoyama S, Edelman H, Rowe GC, Frederick DT, Hurley AD, Nellore A, Kung AL, Wargo JA, Song JS, Fisher DE, Arany Z, Widlund HR: Oncogenic BRAF regulates oxidative metabolism via PGC1alpha and MITF. Cancer Cell. 2013, 23: 302-315. 10.1016/j.ccr.2013.02.003.View ArticlePubMedPubMed CentralGoogle Scholar
- Roesch A, Vultur A, Bogeski I, Wang H, Zimmermann KM, Speicher D, Korbel C, Laschke MW, Gimotty PA, Philipp SE, Krause E, Patzold S, Villanueva J, Krepler C, Fukunaga-Kalabis M, Hoth M, Bastian BC, Vogt T, Herlyn M: Overcoming intrinsic multidrug resistance in melanoma by blocking the mitochondrial respiratory chain of slow-cycling JARID1B(high) cells. Cancer Cell. 2013, 23: 811-825. 10.1016/j.ccr.2013.05.003.View ArticlePubMedGoogle Scholar
- Viale A, Pettazzoni P, Lyssiotis CA, Ying H, Sanchez N, Marchesini M, Carugo A, Green T, Seth S, Giuliani V, Kost-Alimova M, Muller F, Colla S, Nezi L, Genovese G, Deem AK, Kapoor A, Yao W, Brunetto E, Kang Y, Yuan M, Asara JM, Wang YA, Heffernan TP, Kimmelman AC, Wang H, Fleming JB, Cantley LC, DePinho RA, Draetta GF: Oncogene ablation-resistant pancreatic cancer cells depend on mitochondrial function. Nature. 2014, advance online publicationGoogle Scholar
- Xiao Q, Ge G: Lysyl oxidase, extracellular matrix remodeling and cancer metastasis. Cancer microenvironment. 2012, 5: 261-273. 10.1007/s12307-012-0105-z.View ArticlePubMedPubMed CentralGoogle Scholar
- Dong C, Yuan T, Wu Y, Wang Y, Fan TW, Miriyala S, Lin Y, Yao J, Shi J, Kang T, Lorkiewicz P, St Clair D, Hung MC, Evers BM, Zhou BP: Loss of FBP1 by Snail-mediated repression provides metabolic advantages in basal-like breast cancer. Cancer Cell. 2013, 23: 316-331. 10.1016/j.ccr.2013.01.022.View ArticlePubMedPubMed CentralGoogle Scholar
- Barres R, Yan J, Egan B, Treebak JT, Rasmussen M, Fritz T, Caidahl K, Krook A, O'Gorman DJ, Zierath JR: Acute exercise remodels promoter methylation in human skeletal muscle. Cell Metab. 2012, 15: 405-411. 10.1016/j.cmet.2012.01.001.View ArticlePubMedGoogle Scholar
- Kwon HS, Huang B, Ho Jeoung N, Wu P, Steussy CN, Harris RA: Retinoic acids and trichostatin A (TSA), a histone deacetylase inhibitor, induce human pyruvate dehydrogenase kinase 4 (PDK4) gene expression. Biochim Biophys Acta. 2006, 1759: 141-151. 10.1016/j.bbaexp.2006.04.005.View ArticlePubMedGoogle Scholar
- Fujiwara S, Kawano Y, Yuki H, Okuno Y, Nosaka K, Mitsuya H, Hata H: PDK1 inhibition is a novel therapeutic target in multiple myeloma. Br J Cancer. 2013, 108: 170-178. 10.1038/bjc.2012.527.View ArticlePubMedPubMed CentralGoogle Scholar
- Bonnet S, Archer SL, Allalunis-Turner J, Haromy A, Beaulieu C, Thompson R, Lee CT, Lopaschuk GD, Puttagunta L, Harry G, Hashimoto K, Porter CJ, Andrade MA, Thebaud B, Michelakis ED: A mitochondria-K + channel axis is suppressed in cancer and its normalization promotes apoptosis and inhibits cancer growth. Cancer Cell. 2007, 11: 37-51. 10.1016/j.ccr.2006.10.020.View ArticlePubMedGoogle Scholar
- Grassian AR, Metallo CM, Coloff JL, Stephanopoulos G, Brugge JS: Erk regulation of pyruvate dehydrogenase flux through PDK4 modulates cell proliferation. Genes Dev. 2011, 25: 1716-1733. 10.1101/gad.16771811.View ArticlePubMedPubMed CentralGoogle Scholar
- Joza N, Susin SA, Daugas E, Stanford WL, Cho SK, Li CY, Sasaki T, Elia AJ, Cheng HY, Ravagnan L, Ferri KF, Zamzami N, Wakeham A, Hakem R, Yoshida H, Kong YY, Mak TW, Zuniga-Pflucker JC, Kroemer G, Penninger JM: Essential role of the mitochondrial apoptosis-inducing factor in programmed cell death. Nature. 2001, 410: 549-554. 10.1038/35069004.View ArticlePubMedGoogle Scholar
- Schieber MS, Chandel NS: ROS links glucose metabolism to breast cancer stem cell and EMT phenotype. Cancer Cell. 2013, 23: 265-267. 10.1016/j.ccr.2013.02.021.View ArticlePubMedGoogle Scholar
- Sena LA, Chandel NS: Physiological roles of mitochondrial reactive oxygen species. Mol Cell. 2012, 48: 158-167. 10.1016/j.molcel.2012.09.025.View ArticlePubMedPubMed CentralGoogle Scholar
This article is published under license to BioMed Central Ltd. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/4.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly credited. The Creative Commons Public Domain Dedication waiver (http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated.