Succinate dehydrogenase inhibition leads to epithelial-mesenchymal transition and reprogrammed carbon metabolism
- Paul-Joseph P Aspuria1Email author,
- Sophia Y Lunt†2, 3,
- Leif Väremo†4,
- Laurent Vergnes5,
- Maricel Gozo1, 6,
- Jessica A Beach1, 6,
- Brenda Salumbides1,
- Karen Reue5,
- W Ruprecht Wiedemeyer1, 7,
- Jens Nielsen4,
- Beth Y Karlan1, 7 and
- Sandra Orsulic1, 7
© Aspuria et al.; licensee BioMed Central Ltd. 2014
Received: 20 May 2014
Accepted: 4 September 2014
Published: 15 December 2014
Succinate dehydrogenase (SDH) is a mitochondrial metabolic enzyme complex involved in both the electron transport chain and the citric acid cycle. SDH mutations resulting in enzymatic dysfunction have been found to be a predisposing factor in various hereditary cancers. Therefore, SDH has been implicated as a tumor suppressor.
We identified that dysregulation of SDH components also occurs in serous ovarian cancer, particularly the SDH subunit SDHB. Targeted knockdown of Sdhb in mouse ovarian cancer cells resulted in enhanced proliferation and an epithelial-to-mesenchymal transition (EMT). Bioinformatics analysis revealed that decreased SDHB expression leads to a transcriptional upregulation of genes involved in metabolic networks affecting histone methylation. We confirmed that Sdhb knockdown leads to a hypermethylated epigenome that is sufficient to promote EMT. Metabolically, the loss of Sdhb resulted in reprogrammed carbon source utilization and mitochondrial dysfunction. This altered metabolic state of Sdhb knockdown cells rendered them hypersensitive to energy stress.
These data illustrate how SDH dysfunction alters the epigenetic and metabolic landscape in ovarian cancer. By analyzing the involvement of this enzyme in transcriptional and metabolic networks, we find a metabolic Achilles’ heel that can be exploited therapeutically. Analyses of this type provide an understanding how specific perturbations in cancer metabolism may lead to novel anticancer strategies.
KeywordsSuccinate dehydrogenase SDH Ovarian cancer EMT Carbon metabolism Epigenetics
Ovarian cancer is the most lethal gynecological malignancy in the United States, resulting in approximately 22,000 new diagnoses and 14,000 deaths in 2013 . The prognosis of ovarian cancer is poor and has seen limited improvement in the past few decades with a 5-year survival rate of roughly 44% . High-grade serous ovarian cancer is the most common and most lethal subtype of ovarian cancer, contributing to >70% of deaths from ovarian cancer . Integrative large-scale analyses of patient tumors with serous histology have revealed that the disease is highly heterogeneous but can be clustered into several molecular subtypes with different clinical outcomes [4, 5]. One of the poor prognosis subtypes contains an epithelial-mesenchymal transition (EMT) signature . Therefore, understanding and characterizing EMT in ovarian cancer is of great importance.
EMT is a biological process that converts epithelial cells to motile, mesenchymal cells . It is largely defined by the loss of plasma membrane E-cadherin and conversion from a cobblestone-like morphology to a more spindle-shaped and less adherent state. It was initially characterized as a fundamental event in morphogenesis during embryonic tissue and organ development . However, emerging evidence suggests that cancer cells utilize the EMT program to enhance invasiveness, increase migratory ability, and acquire stem-like characteristics such as chemoresistance. EMT reprogramming involves several layers of regulation and has been the subject of intense research, especially in the context of cancer . One area that has not been extensively studied is the role of metabolic enzymes in the regulation and maintenance of EMT.
Alterations in metabolism have been implicated in cancer, with the main focus on the Warburg effect, a phenomenon in which cancer cells upregulate glycolysis and lactate production while decreasing glucose contribution to the citric acid (TCA) cycle in the mitochondria, even in the presence of sufficient oxygen [8–10]. However, mitochondrial metabolism is also important for cancer proliferation. For example, mutations in the mitochondrial enzyme succinate dehydrogenase (SDH) have been reported in a wide variety of cancers [11, 12]. Also known as mitochondrial respiratory Complex II, SDH is the only enzyme to participate in both the TCA cycle and electron transport chain (ETC.). In the TCA cycle, SDH catalyzes the oxidation reaction where succinate is converted to fumarate, which is coupled to the reduction of ubiquinone to ubiquinol in the ETC. . SDH is a holoenzyme consisting of four essential subunits: SDHA (a flavoprotein), SDHB (iron-sulfur protein), and two membrane anchor units SDHC and SDHD. SDH assembly requires two factors, SDHAF1 and SDHAF2 , while its function is influenced by the deacetylase activity of SIRT3 . SDHx loss-of-function mutations have been found to predispose individuals to hereditary pheochromocytoma (PCC), paraganglioma (PGL), gastrointestinal stromal tumor (GIST), and renal cell carcinoma . Interestingly, it has been shown that SDHx mutations are associated with hallmarks of EMT in PCC and PGL as a result of alterations to the epigenome [15, 16]. Furthermore, the SDH regulators SDHAF1 and SIRT3 have been identified as tumor suppressor genes in lung and breast cancer, respectively [14, 17]. This has led to the definition of SDH as a mitochondrial tumor suppressor and a key player in cancer cell differentiation.
Ovarian cancer is characterized by a remarkable degree of genomic disarray but extremely low frequency of recurrent genomic alterations . The dearth of common therapeutic targets in ovarian cancer samples highlights the need for individualized therapy or different approaches to identifying patient groups with common characteristics that can be therapeutically exploited. Interestingly, succinate was identified as a potential metabolic urinary biomarker for ovarian cancer , indicating that altered SDH activity may be important in ovarian cancer progression.
Here, we investigate whether alterations in SDH function also impact ovarian cancer biology. We find that dysregulation of SDH members occurs frequently in serous ovarian cancer. Utilizing a mouse ovarian cancer cell line as a model system, we find that knockdown of one of the SDH subunits, Sdhb, increases aggressiveness and results in EMT. Sdhb knockdown cells were found to have a hypermethylated epigenome whereby increased H3K27 methylation was sufficient to induce EMT. Using a metabolomic approach to investigate the metabolic effects of decreased SDH activity, we found that Sdhb knockdown leads to altered glucose and glutamine utilization through central carbon metabolism as well as mitochondrial dysfunction. These data demonstrate the metabolic and epigenetic consequences of decreased activity of a mitochondrial metabolic enzyme in ovarian cancer.
Plasmids and reagents
pLKO.1-based short hairpin constructs specific for mouse Sdhb and human SDHB, as well as scrambled control sequences, were obtained from Sigma-Aldrich (St. Louis, MO, USA). shRNA sequences are listed in Additional file 1: Table S1. GSK-J1 and GSK-J4 were obtained from Tocris Bioscience (Bristol, Avon, UK). Metformin, p-iodonitrotetrazolium, oligomycin, carbonyl cyanide 4-(trifluoromethoxy)phenylhydrazone chloride (FCCP), and rotenone-myxothiazol were purchased from Sigma-Aldrich.
C1 mouse ovarian cancer cells were generated as previously described [19–21]. HEY cells were obtained from American Type Culture Collection (ATCC). Both cell lines were grown in Dulbecco’s modified Eagle’s medium (DMEM) supplemented with 10% FBS unless otherwise indicated.
Lentiviral-mediated gene knockdown
pLKO.1-sh1-Sdhb, pLKO.1-sh2-Sdhb, pLKO.1-sh1-SDHB, pLKO.1-sh2-SDHB, and pLKO.1 scrambled shRNA plasmids were co-transfected with the delta 8.9 packaging plasmid and VSVG plasmid into Lenti-X™ 293 T cells using Lipofectamine 2000 (Invitrogen, Carlsbad, CA, USA). After overnight incubation, the cell culture medium was changed to DMEM supplemented with 30% FBS. The medium was harvested 48 h later and filtered through a 0.45-μM filter. The viral supernatant was used for infection of C1 and HEY cells with polybrene (4 μg/ml). Cells were infected overnight, and polyclonal populations of stable transductants were selected with puromycin (5 μg/ml) for 72 h.
Western blot analysis
Cell lysates were prepared using RIPA lysis buffer (Sigma-Aldrich) containing a protease inhibitor cocktail (Roche, Basel, Switzerland). Protein concentrations were quantified by the BCA protein assay (Thermo Scientific, Rockford, IL, USA). Fifty micrograms of lysate was loaded onto a gradient gel (BioRad, Hercules, CA, USA) and subjected to gel electrophoresis. Semi-dry protein transfer to a PVDF nitrocellulose membrane (BioRad) was performed using the Transblot Turbo transfer system (BioRad). The membrane was then blocked in Odyssey blocking buffer (LiCOR, Lincoln, NE, USA) for 30 min and incubated with the appropriate antibody overnight at 4°C. Antibodies included E-cadherin (BD Biosciences, San Jose, CA, USA), β-actin (Abcam, Cambridge, MA, USA), Complex II Western Blot Antibody Cocktail that detects SDHB (30 kDa) and Complex Va (60 kDa) (Abcam, Burlungame, CA, USA), Methyl-Histone H3 Antibody Sampler Kit (Cell Signaling, Danvers, MA, USA), and SDHB (Santa Cruz Biotechnology, Dallas, TX, USA). Membranes were then washed with TBS-T, incubated with the appropriate secondary antibody for 1 h at room temperature and then washed with TBS-T. The membrane signal was subsequently analyzed by the LiCOR Odyssey system (LiCOR Biosciences, Frankfurt, Germany).
RNA was extracted using the RNeasy Miniprep Kit (Qiagen, Valencia, CA, USA) and was reverse transcribed to cDNA using the Quantitect Reverse Transcription Kit (Qiagen). For qRT-PCR, 50 ng of cDNA was mixed with the appropriate primers and the iQ SYBR-Green Supermix (BioRad) in a 96-well plate format. The qRT-PCR reaction was performed using the CFX96 Real-Time System (BioRad). Primer sequences are listed in Additional file 1: Table S1.
Cells grown on coverslips were fixed with 3% paraformaldehyde for 20 min, washed with PBS, then permeabilized with 0.1% Triton X-100/PBS for 10 min. This was followed by incubation for 1 h at 37°C with E-cadherin antibody (1:1,000 dilution; BD Biosciences) or SDHB antibody (1:100 dilution; Santa Cruz Biotechnology). After washing with PBS, cells were incubated with appropriate fluorescent secondary antibody (1:1,000 dilution; Invitrogen) for 30 min at 37°C. The coverslips were mounted on slides with Vectashield mounting medium with DAPI (Vector Labs, Burlingame, CA, USA).
Succinate dehydrogenase assay
The SDH assay was performed using the Succinate Dehydrogenase Activity Colorimetric Assay Kit (Biovision, Milpitas, California, USA). Briefly, 1 × 107 cells were dounce homogenized in 1 ml SDH assay buffer. Five microliters of cell lysate were added to a mixture containing SDH assay buffer, SDH substrate mix, and SDH probe. Absorbance readings at 600 nm were taken every 3 min for a total of 30 min. SDH activity was determined by the formula B/(ΔT × V) × Dilution factor, where B = amount of reduced DCIP from standard curve (nmol), ΔT = reaction time (min), V = sample volume added into the reaction well (μl), and D = dilution factor.
Cell proliferation and drug treatment
Cells were seeded onto 6-well or 96-well plates at 5 × 104 or 1 × 103 cells/well, respectively. For the glucose and glutamine withdrawal experiments, the medium was changed to DMEM + 10% FBS with indicated concentrations of glucose or glutamine 24 h post-seeding. To assay cell proliferation, triplicate cultures were trypsinized at the indicated time points and counted using the trypan blue exclusion method. To assay metformin sensitivity, various concentrations of metformin were added to cultures in a 6-well plate 24 h post-seeding and cells were stained with crystal violet at the indicated time points.
Soft agar growth
To assay anchorage-independent growth, 1 × 104 cells were mixed with 0.5% agarose and overlayed on a 0.9% agarose base layer in a 6-well plate. Plates were incubated at 37°C for 3 weeks in a 5% CO2 incubator. Colonies were stained with p-iodonitrotetrazolium chloride overnight and counted using a stereomicroscope. Colony size was measured using QCapture Pro™ (Q Imaging, Surrey, BC, Canada).
Isotope labeling studies
For carbon tracing through central carbon metabolism, cells were cultured for approximately 24 h in 10-cm plates in glucose- and glutamine-free DMEM (Sigma D5030, Sigma-Aldrich) containing 10% dialyzed FBS, 1% streptomycin/penicillin, naturally labeled 4 mM glutamine or 25 mM glucose, and the appropriate tracer, [U-13C6]glucose or [U-13C5]glutamine (Cambridge Isotopes Laboratories, Inc., Cambridge, MA, USA). After incubation, the medium was aspirated and each plate was rinsed with 10 ml ice-cold saline. The saline was aspirated, and cells were quenched with 2.89 ml of −20°C HPLC-grade methanol. After adding 1.74 ml of ice water, the cells were scraped with a cell lifter and collected in 15-ml conical tubes. Chloroform (2.89 ml at −20°C) was added to each tube and vortexed for 10 min at 4°C. Extracts were centrifuged at 4,000 × g for 15 min at 4°C. The upper aqueous phase was collected in a separate tube and evaporated under nitrogen for polar metabolite analysis. The metabolites were analyzed using liquid chromatography tandem mass spectrometry (LC-MS/MS) using a variation of the method described previously . A Paradigm MS4 HPLC (Michrom Bioresources, Auburn, CA, USA) and a Synergi Hydro column (4 μm particle size, 80 Å, 150 mm × 2 mm, from Phenomenex, Torrance, CA, USA) were used for the separation of metabolites by polarity. Prior to column separation, the samples were loaded onto a trapping column (C18, 4 mm × 2 mm, from Phenomenex) and washed for 30 s with HPLC grade water containing 10 mM tributylamine and 15 mM acetic acid for rapid desalting. HPLC separation was coupled with negative-mode electrospray ionization (ESI) to a TSQ Vantage triple stage quadrupole mass spectrometer (Thermo Scientific, Waltham, MA, USA) operating in multiple reactions monitoring (MRM) mode. The LC parameters were as follows: autosampler temperature, 10°C; injection volume, 10 μl; column temperature, room temperature; and flow rate, 200 μl/minute. The LC solvents were Solvent A: 10 mM tributylamine and 15 mM acetic acid in 97:3 water:methanol (pH 4.95), and Solvent B: methanol. Elution from the column was performed over 50 minutes with the following gradient: t = 0, 0% B; t = 5, 0% B; t = 10, 20% B; t = 20, 20% B; t = 35, 65% B; t = 38, 95% B; t = 42, 95% B, t = 43, 0% B; t = 50, 0% B. ESI spray voltage was 3,000 V. Nitrogen was used as sheath gas at 30 psi and as the auxiliary gas at 10 psi, and argon as the collision gas at 1.5 mTorr, with the capillary temperature at 325°C. Scan time for each MRM transition was 0.1 second with a scan width of 1 m/z. The LC runs were divided into time segments, with the MRM scans within each time segment containing compounds eluting during that time interval. For compounds eluting near boundaries between time segments, the MRM scan corresponding to the compound was conducted in both time segments. Instrument control, chromatographic control, and data acquisition were performed by the Xcalibar software (Thermo Scientific). Data analysis was performed using MAVEN [25, 26]. The protein content of extracted cells was determined from the dried protein fraction from each extraction, which was incubated overnight in 2 ml of 0.2 M KOH and quantified by Bradford assay. Isotope labeling data was corrected for the natural abundance of different isotopes using IsoCor .
Glucose consumption and lactate excretion rates
and X is the cell number and t is the time in hours.
Cellular oxygen consumption and extracellular acidification rates
Cellular metabolic rates were measured using an XF24 Analyzer (Seahorse Bioscience, North Billerica, MA, USA) as previously described . Mixing, waiting, and measurement times were 4, 2, and 2 min and 2, 3, and 3 min for C1 and HEY cells, respectively. The measurements were normalized against protein concentration. The concentrations for compounds injected during the analysis were 100 μM 2,4 dinitrophenol, 0.5 μM oligomycin, 1 μM FCCP, 1 μM rotenone-myxothiazol and 0.75 μM oligomycin, 0.5 μM FCCP, 0.75 μM rotenone-myxothiazol for C1 cells and HEY cells, respectively. Data was obtained using the XF24 Analyzer software.
Extracted RNA was enriched using the Ribominus™ Eukaryote Kit (Invitrogen). The Ion PI™ Template OT2 200 Kit (Invitrogen) was used to prepare the RNA library for sequencing. Sequencing was performed using the Ion Proton™ Sequencer (Life Technologies, Grand Island, NY, USA). Fragments per kilobase of exon per million reads (FPKM) were calculated using TopHAT and Cufflinks .
Gene set analysis
Genes with FPKM smaller than 5 were discarded, leaving a total of 6,503 genes for further analysis. The log2 fold change (i.e., log2 of the FPKM ratio) was used as the gene-level statistic. The gene set analysis was carried out using the R/Bioconductor package piano. Three types of gene set collections were used, describing the association between (i) genes and metabolites, (ii) genes and metabolic pathways, and (iii) Gene Ontology (GO) terms.
The human genome-scale metabolic network HMR2 was used to define the metabolite gene sets . HMR2 contains gene associations for its enzymatic reactions, so for each metabolite, all genes that are associated with reactions in which the metabolite takes part are grouped into a gene set. The metabolic pathway gene sets were defined by all genes associated to reactions belonging to an individual pathway. Each set of human genes was translated into a set of mouse genes by mapping the human Ensembl gene identifiers to mouse Ensembl gene identifiers using the R/Bioconductor package biomaRt. This package was also used to acquire the GO term gene. Gene set analysis was carried out separately for the metabolites, pathways, and GO terms. In all cases, the gene set statistics were calculated using piano with the median method and the gene set p values were calculated using gene permutation (10,000 permutations). The piano package calculates scores for several directionality classes. We primarily relied on the distinct-directional class for the identification of gene sets that are coordinately regulated in a distinct direction (up or down). Supplemental materials and methods are found in Additional file 2.
SDH pathway members are downregulated in serous ovarian cancer
Knockdown of Sdhb results in EMT in mouse ovarian cancer cells
Knockdown of Sdhb leads to histone hypermethylation
To further validate the transcriptional influence on histone methylation, we performed an additional gene set analysis using GO terms as gene sets. In Figure 3C, GO terms significantly enriched with upregulated genes (p < 0.05) and relevant to histone methylation are shown, clearly indicating a pattern of increased histone methylation regulated on the transcriptional level (complete results are presented in Additional file 6: Table S2). In summary, Sdhb knockdown appears to be orchestrating a transcriptional program in which histone methylation is promoted or maintained, possibly through upregulation of the SAM cycle.
Sdhb knockdown leads to reprogrammed glucose and glutamine utilization
To further probe metabolic alterations caused by Sdhb knockdown, we used isotope-labeled nutrients combined with mass spectrometry-based techniques to trace glucose and glutamine through central carbon metabolism. Analysis of intracellular metabolites after culturing cells in the presence of uniformly labeled 13C-glucose ([U-13C6]glucose) revealed that Sdhb knockdown increased glucose contribution to sedoheptulose-7-phosphate and adenosine triphosphate (ATP); in contrast, glucose contribution to TCA cycle intermediate succinate was decreased (Figure 5B). Consistently, succinate and glutamate labeling from uniformly labeled 13C-glutamine ([U-13C5]glutamine) was increased (Figure 5C). This increased glutamine utilization by Sdhb knockdown cells is highlighted by their increased uptake of glutamine relative to the control cells (Figure 5D). These data suggest that Sdhb knockdown increases glucose contribution to the pentose phosphate pathway and nucleotide biosynthesis to support rapid proliferation. Increased nucleotide synthesis from glucose is achieved by decreasing glucose contribution to the TCA cycle, which is compensated by increasing glutamine contribution to the TCA cycle. Intracellular pool levels also support a model in which nucleotide biosynthesis is upregulated, as many glycolytic and pentose phosphate pathway intermediates are increased in response to Sdhb knockdown (Additional file 7: Figure S4).
Given the increased use of glutamine in Sdhb knockdown cells, we hypothesized that these cells have an augmented dependency on glutamine for survival relative to the control cells. To test this, we assayed the effect of varying glutamine concentration on cell proliferation. Indeed, in lower glutamine concentrations, C1-sh1-Sdhb cells grew slower in comparison to the control C1-scr cells (Figure 5E). Therefore, knockdown of Sdhb leads to a greater reliance on glutamine for cell survival.
Sdhb knockdown results in a mitochondrial reserve capacity defect that can be exploited by metformin
The reserve respiratory capacity must be sufficient to provide enough ATP during times of energy stress. If ATP demands are not met, affected cells risk being driven into senescence or cell death . To determine whether the decreased reserve capacity of Sdhb knockdown cells rendered them sensitive to energy stress, C1-scr and C1-sh1-Sdhb cells were cultured in normal media then switched to media with varying amounts of glucose. Despite the previous observation that Sdhb knockdown cells grew at a faster rate than control cells under normal growth conditions (Figure 2B), C1-sh1-Sdhb cells grew at a slower rate than C1-scr cells upon glucose withdrawal (Figure 6B). In fact, the majority of Sdhb knockdown cells appeared to undergo apoptosis (Figure 6C). Consistent with the increased dependence of C1-sh1-Sdhb cells on glucose relative to C1-scr cells, C1-sh1-Sdhb cells were more glycolytic (Figure 6D), took up glucose and excreted lactate at increased levels relative to C1-scr cells (Figure 6E). We suspect that the effect of glucose starvation on the Sdhb knockdown cells is twofold, a failure to adapt to the decrease in glycolytic ATP and its reliance on the pentose phosphate pathway for nucleotide biosynthesis.
Since C1-sh1-Sdhb cells are hypersensitive to glucose starvation, we were interested if this bioenergetics deficiency could be therapeutically exploited. Metformin, a commonly used antidiabetic drug, inhibits Complex I in the ETC . This reduces the cellular energy charge, a measure of the energetic state of the cell’s adenine nucleotide pool defined as ([ATP] + 0.5[ADP]) / ([ATP] + [ADP] + [AMP]) . To determine if metformin-induced energy stress had a differential effect on C1-scr and C1-sh1-Sdhb cells, cells were exposed to various concentrations of metformin. C1-sh1-Sdhb cells were hypersensitive to metformin treatment relative to C1-scr cells (Figure 6F). Specifically, metformin inhibited growth of the control and Sdhb knockdown cells with IC50 values of 29.73 and 14.73 mM, respectively (Figure 6F). Taken together, these results implicate a crucial role for Sdhb in the response to bioenergetic insult.
Various mechanisms have been proposed for SDH-associated malignancy. Due to SDH’s role in the ETC, it has been suggested that reactive oxygen species (ROS) can be induced by SDH-inactivating mutations, which contributes to genomic instability, DNA damage, and other tumorigenic phenomena . However, Sdhb knockdown did not lead to altered ROS levels, and the antioxidant N-acetylcysteine (NAC) had no effect on various Sdhb knockdown phenotypes in our system (Additional file 8: Figure S5A). Interestingly, many studies have focused on the idea that succinate itself can act as an oncometabolite. Due to its structural similarity to α-ketoglutarate (α-KG), succinate has been found to competitively inhibit α-KG-dependent dioxygenases. This phenomenon was first described linking succinate to HIF1α stabilization; succinate was found to directly inhibit prolyl hydroxylases which serve to degrade HIF1α [48, 49]. However, we did not find altered HIF signaling between the control and Sdhb knockdown cells (Additional file 8: Figure S5B). This may be due in part to the modestly elevated levels of succinate in our Sdhb knockdown cells, approximately six-fold vs. control (Figure 5A), in comparison to previous studies (approximately 100-fold in tumors and approximately 300-fold upon complete knockout of Sdhb in mouse cells) [16, 50]. It is also possible that HIF signaling by SDHx mutations is tissue-dependent, thus explaining the specific range of cancers induced by SDH inactivation. Recently, the repertoire of α-KG-dependent dioxygenases targeted by succinate has expanded to include epigenome-modifying enzymes such as the TET family of 5-methylcytosine hydroxylases and histone demethylases . Inhibition of these enzymes by elevated levels of succinate leads to genome-wide histone and DNA methylation [51, 52]. In support of this idea, SDHx- related PGL tumors were found to display a hypermethylator phenotype, where epigenetic silencing was particularly severe in SDHB-mutated tumors . This association between SDHx mutations and hypermethylation was also found in GIST . Our data further support this model in which SDH inhibition leads to a hypermethylated epigenome. It has been suggested that a stronger hypermethylator phenotype, which associates with stronger downregulation of target genes, may contribute to the metastatic phenotype . Our results are in line with this idea as aberrant hypermethylation of H3K27 in mouse ovarian cancer cells was sufficient for EMT, thereby implicating the link between epigenetic remodeling and cell state . However, overexpression of the H3K27 demethylases, UTX and JMJD3, in Sdhb knockdown cells failed to reverse EMT, thereby indicating a higher complexity to the maintenance of the mesenchymal state (Additional file 9: Figure S6). This is likely due to transcriptional upregulation of metabolic and epigenetic networks to support this hypermethylated state. Therefore, SDH dysfunction is a contributor of the tumorigenic process via mitochondrion-to-nuclear signaling.
In addition to SDH activity being linked to epigenetic reprogramming, we also found that Sdhb knockdown leads to reprogrammed glucose and glutamine utilization. Specifically, increased glucose flux into the pentose phosphate pathway and increased glutamine usage in the TCA cycle. Interestingly, these phenomena mimic disruption of another TCA cycle enzyme, fumarate hydratase (FH), in mouse kidney cells . The mechanism by which SDH dysfunction leads to these effects is unknown. One possibility is that the global epigenetic changes induced by Sdhb knockdown may lead to an altered transcriptome as evidenced by pathway enrichment analysis of the RNA-seq data (Additional file 7: Figure S4). The analysis indicated that the pentose phosphate pathway and nucleotide metabolism gene sets were both upregulated. The gene set analysis also showed that genes associated with oxidative phosphorylation were significantly downregulated (Additional file 10: Figure S7). In similar fashion, the MYC proto-oncogene has been shown to directly control the transcription of various genes involved in several metabolic pathways including glycolysis, the pentose phosphate pathway, and glutaminolysis [55, 56]. Also, the glycolytic enzyme fructose-1,6-biphosphatase (FBP1) has been shown to influence metabolic reprogramming to impact EMT progression . How SDH dysfunction affects the regulation of these enzymes will be an interesting subject of future research. Interestingly, an association between glutamine dependence and cancer invasiveness has been recently found in ovarian cancer, thereby providing further support for carbon source reprogramming to support different ovarian cancer states .
Although SDH dysfunction leads to enhanced cancer aggressiveness, it also results in metabolic vulnerabilities due to its involvement in both the TCA cycle and ETC. We showed that altered carbon source utilization in Sdhb knockdown cells renders them hypersensitive to both glucose and glutamine starvation. Also, by analyzing mitochondrial function, we found that SDH dysfunction leads to a decreased mitochondrial reserve capacity that can be exploited by energy stress caused by either glucose withdrawal or metformin treatment. Interestingly, it has been shown that combinatorial treatment with metformin and the glycolytic inhibitor 2-deoxyglucose are effective in a broad spectrum of pre-clinical cancer models . Therefore, it is worth exploring the possibility that tumors with SDHx mutations can be therapeutically targeted by exploiting their metabolic state.
the citric acid cycle
electron transport chain
succinate dehydrogenase assembly factor
gastrointestinal stromal tumor
carbonyl cyanide 4-(trifluoromethoxy)phenylhydrazone
the Cancer Genome Atlas
copy number alterations
human genome-scale metabolic network
oxygen consumption rate
extracellular acidification rate
reactive oxygen species
This work was supported by the Foundation for Women’s Cancer Lynette Medlin-Brumfield/Mary-Jane Welker Ovarian Cancer Early Detection Research Grant (P-J.A.). Work in the lab of S.O. is supported, in part, by grants from Alex’s Lemonade Stand Foundation, Margaret E. Early Medical Research Trust, Sandy Rollman Foundation, Department of Defense Ovarian Cancer Clinical Translational Leverage Award (W81XWH-13-OCRP-TLA) and the American Cancer Society (RSG-10-252-01-TBG). Work in the lab of K.R. is supported, in part, by the National Institutes of Health grants P01 HL28481 and NCRR S10RR026744. W.R.W. is supported, in part, by the Ovarian Cancer Research Fund Liz Tilberis Scholarship. S.Y.L. is supported by the Department of Defense CDMRP Visionary Postdoctoral Award under award number W81XWH-12-1-0466. Views and opinions of, and endorsements by, the authors do not reflect those of the US Army or the Department of Defense. The authors thank the MSU Mass Spectrometry Core and Gavin Reid and Eran Andrechek for access to their laboratories. We would also like to thank Kristian Helin for the UTX and JMJD3 constructs.
- Siegel R, Naishadham D, Jemal A: Cancer statistics, 2013. CA Cancer J Clin. 2013, 63: 11-30. 10.3322/caac.21166.View ArticlePubMedGoogle Scholar
- SEER Cancer Statistics Review, 1975–2010.http://seer.cancer.gov/csr/1975_2010/,
- Seidman JD, Horkayne-Szakaly I, Haiba M, Boice CR, Kurman RJ, Ronnett BM: The histologic type and stage distribution of ovarian carcinomas of surface epithelial origin. Int J Gynecol Pathol. 2004, 23: 41-44. 10.1097/01.pgp.0000101080.35393.16.View ArticlePubMedGoogle Scholar
- Cancer Genome Atlas Research N: Integrated genomic analyses of ovarian carcinoma. Nature. 2011, 474: 609-615. 10.1038/nature10166.View ArticleGoogle Scholar
- Zhang W, Liu Y, Sun N, Wang D, Boyd-Kirkup J, Dou X, Han JD: Integrating genomic, epigenomic, and transcriptomic features reveals modular signatures underlying poor prognosis in ovarian cancer. Cell Rep. 2013, 4: 542-553. 10.1016/j.celrep.2013.07.010.View ArticlePubMedGoogle Scholar
- De Craene B, Berx G: Regulatory networks defining EMT during cancer initiation and progression. Nat Rev Cancer. 2013, 13: 97-110. 10.1038/nrc3447.View ArticlePubMedGoogle Scholar
- Nieto MA: The ins and outs of the epithelial to mesenchymal transition in health and disease. Annu Rev Cell Dev Biol. 2011, 27: 347-376. 10.1146/annurev-cellbio-092910-154036.View ArticlePubMedGoogle Scholar
- Warburg O: Origin of cancer cells. Oncologia. 1956, 9: 75-83. 10.1159/000223920.View ArticlePubMedGoogle Scholar
- Lunt SY, Vander Heiden MG: Aerobic glycolysis: meeting the metabolic requirements of cell proliferation. Annu Rev Cell Dev Biol. 2011, 27: 441-464. 10.1146/annurev-cellbio-092910-154237.View ArticlePubMedGoogle Scholar
- Gatto F, Nookaew I, Nielsen J: Chromosome 3p loss of heterozygosity is associated with a unique metabolic network in clear cell renal carcinoma. Proc Natl Acad Sci U S A. 2014, 111: E866-E875. 10.1073/pnas.1319196111.View ArticlePubMedPubMed CentralGoogle Scholar
- Bardella C, Pollard PJ, Tomlinson I: SDH mutations in cancer. Biochim Biophys Acta. 2011, 1807: 1432-1443. 10.1016/j.bbabio.2011.07.003.View ArticlePubMedGoogle Scholar
- King A, Selak MA, Gottlieb E: Succinate dehydrogenase and fumarate hydratase: linking mitochondrial dysfunction and cancer. Oncogene. 2006, 25: 4675-4682. 10.1038/sj.onc.1209594.View ArticlePubMedGoogle Scholar
- Hoekstra AS, Bayley JP: The role of complex II in disease. Biochim Biophys Acta. 2013, 1827: 543-551. 10.1016/j.bbabio.2012.11.005.View ArticlePubMedGoogle Scholar
- Finley LW, Haas W, Desquiret-Dumas V, Wallace DC, Procaccio V, Gygi SP, Haigis MC: Succinate dehydrogenase is a direct target of sirtuin 3 deacetylase activity. PLoS One. 2011, 6: e23295-10.1371/journal.pone.0023295.View ArticlePubMedPubMed CentralGoogle Scholar
- Loriot C, Burnichon N, Gadessaud N, Vescovo L, Amar L, Libe R, Bertherat J, Plouin PF, Jeunemaitre X, Gimenez-Roqueplo AP, Favier J: Epithelial to mesenchymal transition is activated in metastatic pheochromocytomas and paragangliomas caused by SDHB gene mutations. J Clin Endocrinol Metab. 2012, 97: E954-E962. 10.1210/jc.2011-3437.View ArticlePubMedGoogle Scholar
- Letouze E, Martinelli C, Loriot C, Burnichon N, Abermil N, Ottolenghi C, Janin M, Menara M, Nguyen AT, Benit P, Buffet A, Marcaillou C, Bertherat J, Amar L, Rustin P, De Reyniès A, Gimenez-Roqueplo AP, Favier J: SDH mutations establish a hypermethylator phenotype in paraganglioma. Cancer Cell. 2013, 23: 739-752. 10.1016/j.ccr.2013.04.018.View ArticlePubMedGoogle Scholar
- Liu J, Gao L, Zhang H, Wang D, Wang M, Zhu J, Pang C, Wang C: Succinate dehydrogenase 5(SDH5) regulate (GSK)-3beta-beta-catenin-mediated lung cancer metastasis. J Biol Chem. 2013, 288: 29965-29973. 10.1074/jbc.M113.450106.View ArticlePubMedPubMed CentralGoogle Scholar
- Zhang T, Wu X, Ke C, Yin M, Li Z, Fan L, Zhang W, Zhang H, Zhao F, Zhou X, Lou G, Li K: Identification of potential biomarkers for ovarian cancer by urinary metabolomic profiling. J Proteome Res. 2013, 12: 505-512. 10.1021/pr3009572.View ArticlePubMedGoogle Scholar
- Miao J, Wang Z, Provencher H, Muir B, Dahiya S, Carney E, Leong CO, Sgroi DC, Orsulic S: HOXB13 promotes ovarian cancer progression. Proc Natl Acad Sci U S A. 2007, 104: 17093-17098. 10.1073/pnas.0707938104.View ArticlePubMedPubMed CentralGoogle Scholar
- Orsulic S: An RCAS-TVA-based approach to designer mouse models. Mamm Genome. 2002, 13: 543-547. 10.1007/s00335-002-4003-4.View ArticlePubMedGoogle Scholar
- Xing D, Orsulic S: A genetically defined mouse ovarian carcinoma model for the molecular characterization of pathway-targeted therapy and tumor resistance. Proc Natl Acad Sci U S A. 2005, 102: 6936-6941. 10.1073/pnas.0502256102.View ArticlePubMedPubMed CentralGoogle Scholar
- Cerami E, Gao J, Dogrusoz U, Gross BE, Sumer SO, Aksoy BA, Jacobsen A, Byrne CJ, Heuer ML, Larsson E, Antipin Y, Reva B, Goldberg AP, Sander C, Schultz N: The cBio cancer genomics portal: an open platform for exploring multidimensional cancer genomics data. Cancer Discov. 2012, 2: 401-404. 10.1158/2159-8290.CD-12-0095.View ArticlePubMedGoogle Scholar
- Beroukhim R, Mermel CH, Porter D, Wei G, Raychaudhuri S, Donovan J, Barretina J, Boehm JS, Dobson J, Urashima M, Mc Henry KT, Pinchback RM, Ligon AH, Cho YJ, Haery L, Greulich H, Reich M, Winckler W, Lawrence MS, Weir BA, Tanaka KE, Chiang DY, Bass AJ, Loo A, Hoffman C, Prensner J, Liefeld T, Gao Q, Yecies D, Signoretti S: The landscape of somatic copy-number alteration across human cancers. Nature. 2010, 463: 899-905. 10.1038/nature08822.View ArticlePubMedPubMed CentralGoogle Scholar
- Munger J, Bennett BD, Parikh A, Feng XJ, McArdle J, Rabitz HA, Shenk T, Rabinowitz JD: Systems-level metabolic flux profiling identifies fatty acid synthesis as a target for antiviral therapy. Nat Biotechnol. 2008, 26: 1179-1186. 10.1038/nbt.1500.View ArticlePubMedPubMed CentralGoogle Scholar
- Clasquin MF, Melamud E, Rabinowitz JD: LC-MS data processing with MAVEN: a metabolomic analysis and visualization engine. Curr Protoc Bioinformatics. 2012, Chapter 14: Unit14 11-PubMedGoogle Scholar
- Melamud E, Vastag L, Rabinowitz JD: Metabolomic analysis and visualization engine for LC-MS data. Anal Chem. 2010, 82: 9818-9826. 10.1021/ac1021166.View ArticlePubMedGoogle Scholar
- Millard P, Letisse F, Sokol S, Portais JC: IsoCor: correcting MS data in isotope labeling experiments. Bioinformatics. 2012, 28: 1294-1296. 10.1093/bioinformatics/bts127.View ArticlePubMedGoogle Scholar
- Plaisier CL, Bennett BJ, He A, Guan B, Lusis AJ, Reue K, Vergnes L: Zbtb16 has a role in brown adipocyte bioenergetics. Nutr Diabetes. 2012, 2: e46-View ArticlePubMedPubMed CentralGoogle Scholar
- Trapnell C, Roberts A, Goff L, Pertea G, Kim D, Kelley DR, Pimentel H, Salzberg SL, Rinn JL, Pachter L: Differential gene and transcript expression analysis of RNA-seq experiments with TopHat and Cufflinks. Nat Protoc. 2012, 7: 562-578. 10.1038/nprot.2012.016.View ArticlePubMedPubMed CentralGoogle Scholar
- Varemo L, Nielsen J, Nookaew I: Enriching the gene set analysis of genome-wide data by incorporating directionality of gene expression and combining statistical hypotheses and methods. Nucleic Acids Res. 2013, 41: 4378-4391. 10.1093/nar/gkt111.View ArticlePubMedPubMed CentralGoogle Scholar
- Mardinoglu A, Agren R, Kampf C, Asplund A, Uhlen M, Nielsen J: Genome-scale metabolic modeling of hepatocytes leads to identification of serine deficiency in non-alcoholic fatty liver disease. Nat Comm. 2013, 5: 3083-Google Scholar
- Durinck S, Moreau Y, Kasprzyk A, Davis S, De Moor B, Brazma A, Huber W: BioMart and Bioconductor: a powerful link between biological databases and microarray data analysis. Bioinformatics. 2005, 21: 3439-3440. 10.1093/bioinformatics/bti525.View ArticlePubMedGoogle Scholar
- Solimini NL, Xu Q, Mermel CH, Liang AC, Schlabach MR, Luo J, Burrows AE, Anselmo AN, Bredemeyer AL, Li MZ, Beroukhim R, Meyerson M, Elledge SJ: Recurrent hemizygous deletions in cancers may optimize proliferative potential. Science. 2012, 337: 104-109. 10.1126/science.1219580.View ArticlePubMedPubMed CentralGoogle Scholar
- Davoli T, Xu AW, Mengwasser KE, Sack LM, Yoon JC, Park PJ, Elledge SJ: Cumulative haploinsufficiency and triplosensitivity drive aneuploidy patterns and shape the cancer genome. Cell. 2013, 155: 948-962. 10.1016/j.cell.2013.10.011.View ArticlePubMedPubMed CentralGoogle Scholar
- Tam WL, Weinberg RA: The epigenetics of epithelial-mesenchymal plasticity in cancer. Nat Med. 2013, 19: 1438-1449. 10.1038/nm.3336.View ArticlePubMedPubMed CentralGoogle Scholar
- Shiota M, Izumi H, Onitsuka T, Miyamoto N, Kashiwagi E, Kidani A, Yokomizo A, Naito S, Kohno K: Twist promotes tumor cell growth through YB-1 expression. Cancer Res. 2008, 68: 98-105. 10.1158/0008-5472.CAN-07-2981.View ArticlePubMedGoogle Scholar
- Ansieau S, Bastid J, Doreau A, Morel AP, Bouchet BP, Thomas C, Fauvet F, Puisieux I, Doglioni C, Piccinin S, Maestro R, Voeltzel T, Selmi A, Valsesia-Wittmann S, Caron de Fromentel C, Puisieux A: Induction of EMT by twist proteins as a collateral effect of tumor-promoting inactivation of premature senescence. Cancer Cell. 2008, 14: 79-89. 10.1016/j.ccr.2008.06.005.View ArticlePubMedGoogle Scholar
- Qian J, Luo Y, Gu X, Zhan W, Wang X: Twist1 promotes gastric cancer cell proliferation through up-regulation of FoxM1. PLoS One. 2013, 8: e77625-10.1371/journal.pone.0077625.View ArticlePubMedPubMed CentralGoogle Scholar
- Mardinoglu A, Agren R, Kampf C, Asplund A, Uhlen M, Nielsen J: Genome-scale metabolic modelling of hepatocytes reveals serine deficiency in patients with non-alcoholic fatty liver disease. Nat Commun. 2014, 5: 3083-View ArticlePubMedGoogle Scholar
- Patil KR, Nielsen J: Uncovering transcriptional regulation of metabolism by using metabolic network topology. Proc Natl Acad Sci U S A. 2005, 102: 2685-2689. 10.1073/pnas.0406811102.View ArticlePubMedPubMed CentralGoogle Scholar
- Martinez-Garcia E, Licht JD: Deregulation of H3K27 methylation in cancer. Nat Genet. 2010, 42: 100-101. 10.1038/ng0210-100.View ArticlePubMedGoogle Scholar
- Kruidenier L, Chung CW, Cheng Z, Liddle J, Che K, Joberty G, Bantscheff M, Bountra C, Bridges A, Diallo H, Eberhard D, Hutchinson S, Jones E, Katso R, Leveridge M, Mander PK, Mosley J, Ramirez-Molina C, Rowland P, Schofield CJ, Sheppard RJ, Smith JE, Swales C, Tanner R, Thomas P, Tumber A, Drewes G, Oppermann U, Patel DJ, Lee K: A selective jumonji H3K27 demethylase inhibitor modulates the proinflammatory macrophage response. Nature. 2012, 488: 404-408. 10.1038/nature11262.View ArticlePubMedPubMed CentralGoogle Scholar
- Heytler PG: Uncouplers of oxidative phosphorylation. Methods Enzymol. 1979, 55: 462-442.View ArticlePubMedGoogle Scholar
- Desler C, Hansen TL, Frederiksen JB, Marcker ML, Singh KK, Juel Rasmussen L: Is there a link between mitochondrial reserve respiratory capacity and aging?. J Aging Res. 2012, 2012: 192503-View ArticlePubMedPubMed CentralGoogle Scholar
- El-Mir MY, Nogueira V, Fontaine E, Averet N, Rigoulet M, Leverve X: Dimethylbiguanide inhibits cell respiration via an indirect effect targeted on the respiratory chain complex I. J Biol Chem. 2000, 275: 223-228. 10.1074/jbc.275.1.223.View ArticlePubMedGoogle Scholar
- Miller RA, Birnbaum MJ: An energetic tale of AMPK-independent effects of metformin. J Clin Invest. 2010, 120: 2267-2270. 10.1172/JCI43661.View ArticlePubMedPubMed CentralGoogle Scholar
- Kaelin WG: SDH5 mutations and familial paraganglioma: somewhere Warburg is smiling. Cancer Cell. 2009, 16: 180-182. 10.1016/j.ccr.2009.08.013.View ArticlePubMedGoogle Scholar
- Pollard PJ, Briere JJ, Alam NA, Barwell J, Barclay E, Wortham NC, Hunt T, Mitchell M, Olpin S, Moat SJ, Hargreaves IP, Heales SJ, Chung YL, Griffiths JR, Dalgleish A, McGrath JA, Gleeson MJ, Hodgson SV, Poulsom R, Rustin P, Tomlinson IP: Accumulation of Krebs cycle intermediates and over-expression of HIF1alpha in tumours which result from germline FH and SDH mutations. Hum Mol Genet. 2005, 14: 2231-2239. 10.1093/hmg/ddi227.View ArticlePubMedGoogle Scholar
- Selak MA, Armour SM, MacKenzie ED, Boulahbel H, Watson DG, Mansfield KD, Pan Y, Simon MC, Thompson CB, Gottlieb E: Succinate links TCA cycle dysfunction to oncogenesis by inhibiting HIF-alpha prolyl hydroxylase. Cancer Cell. 2005, 7: 77-85. 10.1016/j.ccr.2004.11.022.View ArticlePubMedGoogle Scholar
- Imperiale A, Moussallieh FM, Sebag F, Brunaud L, Barlier A, Elbayed K, Bachellier P, Goichot B, Pacak K, Namer IJ, Taieb D: A new specific succinate-glutamate metabolomic hallmark in SDHx-related paragangliomas. PLoS One. 2013, 8: e80539-10.1371/journal.pone.0080539.View ArticlePubMedPubMed CentralGoogle Scholar
- Xiao M, Yang H, Xu W, Ma S, Lin H, Zhu H, Liu L, Liu Y, Yang C, Xu Y, Zhao S, Ye D, Xiong Y, Guan KL: Inhibition of alpha-KG-dependent histone and DNA demethylases by fumarate and succinate that are accumulated in mutations of FH and SDH tumor suppressors. Genes Dev. 2012, 26: 1326-1338. 10.1101/gad.191056.112.View ArticlePubMedPubMed CentralGoogle Scholar
- Cervera AM, Bayley JP, Devilee P, McCreath KJ: Inhibition of succinate dehydrogenase dysregulates histone modification in mammalian cells. Mol Cancer. 2009, 8: 89-10.1186/1476-4598-8-89.View ArticlePubMedPubMed CentralGoogle Scholar
- Killian JK, Kim SY, Miettinen M, Smith C, Merino M, Tsokos M, Quezado M, Smith WI, Jahromi MS, Xekouki P, Szarek E, Walker RL, Lasota J, Raffeld M, Klotzle B, Wang Z, Jones L, Zhu Y, Wang Y, Waterfall JJ, O'Sullivan MJ, Bibikova M, Pacak K, Stratakis C, Janeway KA, Schiffman JD, Fan JB, Helman L, Meltzer PS: Succinate dehydrogenase mutation underlies global epigenomic divergence in gastrointestinal stromal tumor. Cancer Discov. 2013, 3: 648-657. 10.1158/2159-8290.CD-13-0092.View ArticlePubMedPubMed CentralGoogle Scholar
- Frezza C, Zheng L, Folger O, Rajagopalan KN, MacKenzie ED, Jerby L, Micaroni M, Chaneton B, Adam J, Hedley A, Kalna G, Tomlinson IP, Pollard PJ, Watson DG, Deberardinis RJ, Shlomi T, Ruppin E, Gottlieb E: Haem oxygenase is synthetically lethal with the tumour suppressor fumarate hydratase. Nature. 2011, 477: 225-228. 10.1038/nature10363.View ArticlePubMedGoogle Scholar
- Dang CV: MYC, metabolism, cell growth, and tumorigenesis. Cold Spring Harb Perspect Med. 2013, 3: a014217-View ArticlePubMedPubMed CentralGoogle Scholar
- Liu W, Le A, Hancock C, Lane AN, Dang CV, Fan TW, Phang JM: Reprogramming of proline and glutamine metabolism contributes to the proliferative and metabolic responses regulated by oncogenic transcription factor c-MYC. Proc Natl Acad Sci U S A. 2012, 109: 8983-8988. 10.1073/pnas.1203244109.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
- Yang L, Moss T, Mangala LS, Marini J, Zhao H, Wahlig S, Armaiz-Pena G, Jiang D, Achreja A, Win J, Roopaimoole R, Rodriguez-Aguayo C, Mercado-Uribe I, Lopez-Berestein G, Liu J, Tsukamoto T, Sood AK, Ram PT, Nagrath D: Metabolic shifts toward glutamine regulate tumor growth, invasion and bioenergetics in ovarian cancer. Mol Syst Biol. 2014, 10: 728-10.1002/msb.20134892.View ArticlePubMedPubMed CentralGoogle Scholar
- Cheong JH, Park ES, Liang J, Dennison JB, Tsavachidou D, Nguyen-Charles C, Wa Cheng K, Hall H, Zhang D, Lu Y, Ravoori M, Kundra V, Ajani J, Lee JS, Ki Hong W, Mills GB: Dual inhibition of tumor energy pathway by 2-deoxyglucose and metformin is effective against a broad spectrum of preclinical cancer models. Mol Cancer Ther. 2011, 10: 2350-2362. 10.1158/1535-7163.MCT-11-0497.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.