Background
Following the discovery by Otto Warburg high glucose uptake and lactate production—even in the presence of oxygen—have been documented in nearly all tumours [
1,
2]. The glycolytic phenotype was interpreted as a response to damaged cellular respiration to satisfy the ATP demand of the tumours [
3,
4]. However, recently numerous studies have showed that ATP production in tumours could be related to both glycolysis and mitochondrion despite of the defected oxidative phosphorylation [
4‐
6]. These intriguing findings received support by demonstrating multiple auxiliary processes for ATP synthesis [
6‐
8].
However, the relative contribution of glycolysis and TCA cycle in ATP production is rather variable in different tumours during progression [
4,
5] because the metabolic profile of the tumours depends on both the changes in malignant cells and on the microenvironment [
9]. The survival of tumour cells in unfavourable conditions may be attributed to the reprogramming of the metabolism, as a crucial hallmark of the malignant cells [
8]. Thus changes of tumour metabolism require individual and regular monitoring. To define the dominant bioenergetic pathway as bioenergetic signature based on the ratio of metabolic markers of glycolysis and mitochondrial respiration has been reported previously [
10]. In the present study the bioenergetic signature was estimated by comparing the utilization of glucose and acetate in multiple assays. As acetate is mainly metabolized in TCA cycle contrary to glucose acetate labels only TCA cycle. [
11,
12]. Therefore, the comparison of acetate and glucose utilization can be useful to decide the dominant bioenergetic process. Since metabolic reprogramming is directed through regulatory factors it has been suggested that the relevant regulatory mechanisms should also be taken into account [
13]. Hence the used complex multiple assay system includes measurements and analysis on substrate utilization and certain related enzymes as the following: (a) measurements on CO
2 production derived from [1-
14C]-glucose and [1-
14C]-acetate; (b) estimation of adenylate energy charge in the presence and absence of glucose and acetate [
14]; (c) analyse the expression of metabolic enzymes (G6PDH, GAPDH, β-F1-ATPase); (d) use of stable isotope-mass spectrometric technique to compare glycolysis and TCA cycle activity after labelling tumour cells with [U-
13C]-glucose or [2-
13C]-acetate; (e) test the TCA cycle function by measuring the number of
13C-carbons in citrate, derived from [2-
13C]-acetate;.
To gain an insight into the regulation of the aberrant bioenergetic phenotype the expression of mTOR complex 1 and 2 (mammalian target of rapamycin) and of their activity related proteins [
15,
16], heme-oxigenase-1 (HO-1) [
17] and the proinflammatory interleukins (IL-1β), IL-6, IL-8)) [
18,
19] were examined in tumour cell cultures. In contrast to the traditional
14C-substrate labelling method, applying the combination of complex assays and measurements help to map the correlations and allow to get more information about the highly different metabolic and regulatory characteristics of cell lines—as it was observed in the two selected cell lines in which the highest difference was detected at their substrate utilization.
Methods
All materials were purchased from Sigma-Aldrich, except where indicated in the text. HT-1080 (human fibrosarcoma) and ZR-75.1 (human mammary adenocarcinoma) cell lines were selected for detailed characterisation from other human cells (MDA-MB231, BT747—breast and HepG2—hepatocellular carcinoma; Oscort—osteosarcoma, U937—histiocytic lymphoma, isolated fibroblasts) based on their different energy substrate oxidizing capacities.
To obtain subconfluent cultures 5 × 105 HT-1080 and 106 ZR-75.1 cells were seeded in 10 ml RPMI1640 medium with 10 % fetal bovine serum (FBS), 100 IU penicillin and 50 µg/ml streptomycin. In metabolic experiments the medium was replaced by energy substrate-free D5030 medium. All experiments were performed with previously established in vitro cell lines which are not requiring special ethical approvals, The Institutional Tissue- and Cell Culture Laboratory has official permission.
Measurement of energy-substrate oxidation
Cultures were incubated at 37 °C placed in an air-flow chamber and perfused with CO2 free air. Cells were labelled with 0.2 µCi/ml [1-14C]-glucose (specific activity 55 mCi/mmol) or [1-14C]-acetate (specific activity 57 mCi/mmol (Institute of Isotope-Zrt. Budapest) for 1 h. The CO2 released by the cells was trapped on solid alkaline adsorbent, attached to the air-flow chamber and its radioactivity measured with Geiger–Müller counter.
Expression of mRNA
Total RNAs were isolated with Trizol (Invitrogen). Reverse transcription and real time PCR were performed following the instructions of the manufacturer (M-MLV Reverse Transcriptase kit—Invitrogen; ABI power SYBR® Green PCR Master mix and ABI Prism 7000 Sequence Detection System—Applied Biosystems). PCR primers are summarized in Supplementary Table (ST1). Results were obtained as threshold cycle (CT) values. Expression levels were calculated by using the 2−ΔCT method and β-actin for normalization as reported previously.
Expression analysis of mTOR complex related proteins, certain metabolic enzymes and inflammatory cytokines at protein level
Proteins extracted from cells were quantitated using Quant-iT protein assay (Invitrogen), separated by SDS-PAGE. Proteins transferred to PVDF membrane were incubated with the following reagents: anti-phospho-mTOR (Ser2448), anti-phospho-S6 (Ser 235/236), anti-Rictor (Cell Signaling Technologies) and anti-β-actin (Sigma-Aldrich), anti-GLUT1 (Abcam), anti-β-F1-ATPase (anti-ATPB Abcam), anti-GAPDH (Serotec), anti-pan-Akt (Cell Signaling Technologies), anti-phospho-(Ser473)-Akt 1 (Abcam) finally with biotinylated secondary antibodies, avidin-HRP complex (Vectastain Elite ABC Kit, Vector) and enhanced chemiluminescence technique (Pierce ECL Western Blotting Substrate).
The concentrations of extra- and intracellular cytokines were determined by cytokine bead assay (CBA immunoassay) were measured by flow cytometry. The assay was carried out using the CBA Human Soluble Protein Master Buffer Kit (BD Biosciences, San Jose, CA, USA; 558264) and CBA Human IL-8 (558277), IL-6 (558276) and IL-1β (558279) Flex Sets (BD Biosciences) according to the manufacturer’s instructions. The samples were acquired on FACSCalibur (BD Biosciences) and analysed by FCAP Array software (BD Biosciences).
Stable isotope-mass spectroscopic detection of metabolites
Cells incubated in D5030 medium for one hour were labelled with 10 mM [U-
13C]-glucose or [2-
13C]-acetate (Cambridge Isotope Laboratories, Andover, MA, USA). The key metabolites of the glycolytic and TCA cycle pathways were extracted and analysed by applying liquid chromatography–mass spectrometry (LC–MS) based on the protocol described previously [
20,
21].
Determination of adenylate energy charge
To evaluate the distinct contributions of glucose and acetate on the level of adenylate energy charge (AEC) ATP, ADP and AMP were determined in starving cell cultures. To this end the RPMI1640 medium was replaced by the energy substrate deprived medium (D5030) and the cultures were divided into three groups as follows: (a) supplemented with 10 mM glucose, (b) supplemented with 10 mM acetate, (c) kept in starving conditions. Adenylate nucleotides were extracted from the cells with methanol-acetic acid (9:1) were separated by isocratic elution using Aquasil C18 column principally as reported previously [
22] and analysed by applying LKB PE NELSON 3000 interface (Perkin-Elmer Corporation).
Statistical analysis
The experiments were performed in triplicate with 3–5 parallel samples and the data evaluation was performed using Student’s t test (two-tailed); p < 0.05 was considered statistically significant.
Discussion
The accumulated evidences on the glycolytic phenotype and repressed mitochondrial function as contributors of tumour progression have initiated attempts to introduce simple but relevant assays to characterise the bioenergetic profile of human tumour samples [
23‐
25]. The majority of studies in this area involves measurements of labelled intermediates in the bioenergetic pathways using radioactive or stabile isotope substrates [
21,
25,
26] and different detecting instruments, which represent a great challenge.
Glucose and glutamine are considered the basic and indispensable nutrients subsequently they are the most widely used energetic substrates in metabolic analysis of tumour cells. It has been assumed that using two substrates would provide more information about the relative capacity of glycolysis and TCA cycle than either one alone. As the catabolic product of fatty acid oxidation acetate is channelled toward TCA cycle it appeared to be an ideal complementary energetic substrate beside glucose [
27,
28]. Thus acetate can fuel exclusively mitochondrial respiration therefore it is a promising way to detect the bioenergetic mechanism. Some recently published new studies from clinical samples using NMR illustrated that especially acetate could be energy substrate of glioblastomas and brain metastases [
29]. In our study we compared acetate and glucose labelling in the substrate oxidation and in the intracellular metabolite concentration analysis by LC–MS. Our results show that using two substrates such as glucose and acetate has more advantage to get detailed information about the relative capacity of glycolysis and TCA cycle. Moreover, our novel evaluation from LC–MS analysis after
13C-acetate labelling is capable for highlighting the impairment of TCA cycle in tumour cells. To underline the importance of this metabolic profile analysis we demonstrated some points of the potential regulatory background in the two studied cells, which correlate well to the alterations in bioenergetic phenotype of the cells.
The radioactive technology has still an important role in metabolic investigations including released 14CO2 measuring, detecting the oxidation of energy substrates offer a simple and cost-effective assay to estimate the dominant bioenergetic mechanism. Applying LC–MS technique with [U-13C]-glucose and [2-13C]-acetate to investigate tumour specific bioenergetic profile, we demonstrated that the highest level of the incorporated 13C atoms derived from [13C]-glucose into lactate both in HT-1080 and in ZR-75.1 tumour cells correlates to the other results related to high glucose oxidation and AEC measurement. The LC–MS technique allowed to raise the question whether an identical metabolic phenotype may also be recognized by measuring the unlabelled metabolites which could be interesting in future metabolic analysis of tumour samples. This may have special interest. We found that the unlabelled lactate also reflects to the differences in the glycolytic activity of the two investigated tumours in vitro. However, it should be considered that for example lactate production in tumour cells in vivo is not exclusively derived from actual glycolytic flux (e.g. uptake from microenvironment). The low number of 13C atoms in malate and citrate after labelling with [U-13C]-glucose or [2-13C]-acetate respectively may suggest that TCA cycle is less effective (as we detected in HT-1080 in contrast to ZR-75.1 cells). This provides example for detectable relative dominance of the bioenergetic pathways in tumour cells. The remarkably high unlabelled malate concentration may suggest the function of the anaplerotic route to replenish TCA cycle.
Elevated lactate and reduced concentration of TCA cycle metabolites were described in certain human tumour samples and in patients’ body fluids [
30‐
32] by metabolomics mainly with GC or LC–MS, NMR, but other analytical methods have been developing further. However, there is still no consensual advice in using the analytical techniques to measure metabolic profile of tumour samples (certainly these assays are expensive, time-consuming and need large sample size). In our assays we did not measured the complete metabolite-profile, we only measured and evaluated selected metabolites (G6P, R5P, lactate, citrate, malate, succinate). In that way we could rationalize our sources, we could save time, instrumental capacity and we optimised the sensitivity and apply this methods for smaller sample size. The latter was performed after not only glucose but acetate labelling, as well—the metabolism of malignant cells have not been investigated previously in that way in comparison of acetate and glucose consumption. Upon studying the distribution of
13C-carbon, derived from [
13C-U]-glucose, between glycolytic and TCA cycle intermediates high lactate and low malate labelling indicated glycolytic phenotype and impaired TCA cycle in the HT-1080 tumour cells, but not in the ZR-75.1 cells. Moreover, the highly limited labelling of citrate by [2-
13C]-acetate—representing a novel functional test—confirmed the above detected defect in TCA cycle of HT-1080 tumour cells. Consequently glucose and acetate differently elevate the reduced level of adenylate energy charge depending on the dominant bioenergetic mechanism.
An other suggestion comparing the expression of glyceraldehyde-3-phosphate dehydrogenase (GAPDH) and β-F1-ATPase an inverse relationship could be concluded between glycolysis and mitochondrial respiration in various human tumour tissues. In fact it was proposed that β-F1-ATPase could be used for monitoring progression of breast cancer [
10,
31]. We could confirm this and we can also suggest GAPDH/β-F1-ATPase expression studies in tumour tissue samples by immunohistochemistry based on our Western blot results and its harmony with the metabolic characterisation data.
Evaluating our results from different measurements in HT-1080 and ZR-75.1 cells we could summarise that assays–measuring: (a)
14CO
2 from glucose/
14CO
2 from acetate ratio; (b) AEC at the presence of glucose/AEC at the presence of acetate ratio; (c) lactate/malate from
13C-glucose labelling; d. the number of incorporated
13C into citrate from
13C-acetate and e. the GAPDH/β-F1-ATPase expression (summarized in Table
4)—could indicate the metabolic differences at distinct level. From these assays, especially from analysis of the selected metabolites (G6P, R5P, lactate, citrate, malate, succinate) by LC–MS after
13C-glucose and acetate labelling we could conclude the condition of mitochondrial function and TCA cycle capacity (impaired TCA or potentially functioning one in HT-1080 or ZR-75.1 respectively) beside the admitted high glycolytic activity.
Table 4
Bioenergetic signature based on the utilization of glucose and acetate in various assays
14CO2 from glucose/from acetate | 0.4 | 30.1 |
AEC at the presence of glucose/at the presence of acetate | 0.88 | 1.39 |
Labelled13C-lactate/13C-malate from13C-glucosea
| 0.87 | 13.74 |
Number of incorporated13C atoms into citrate from13C-acetate | 1–6 | 1–2 |
GAPDH/ß-F1-ATPaseb
| 1.27 | 9.81 |
Our experimental results allowed to find some correlations of certain regulatory factors and the effected bioenergetic pathways or invasive growth in HT-1080 contrary to ZR-75.1 (harbouring less metastatic potential) tumour cells [
33]. This expectation was met when high expression of proinflammatory interleukins and HO-1 were detected in the TCA cycle impaired and metastatic HT-1080 cells compared with the non-metastatic ZR-75.1 cells. It was reported that IL-1β and IL-6 stimulate glucose metabolism and promote tumour progression [
16]. The high expression of HO-1 in HT-1080 cells deserves special attention as recent studies documented the stimulation of invasive growth and metastatic behaviour associated with high expression of HO-1 in lung cancers (NSCLCs) [
17,
34].
At present we may speculate on the possibility that activated rescue mechanism related to HO-1 in TCA cycle impaired HT-1080 tumour cells may enable mitochondrial ATP synthesis through a potential alternative HO-1 dependent way as previously reported in fumarate hydratase 1 deleted kidney cells especially in kidney tumour cells [
6].
To study mTOR signalling—which has well documented role in the regulation of metabolic and survival functions—we compared the expression of related phospho-proteins and Rictor in HT-1080 and ZR-75.1 cells [
35]. It is well known that malignant transformation is often accompanied by deregulated mTOR activity (mainly higher mTOR activity status) which helps to synthesise and activate the elements/proteins of the cellular apparatus of altered phenotype [
36]. The high mTOR activity is providing the enzymatic background for glycolysis, pentose-phosphate pathway and glutaminolysis, moreover the HIF1α production [
35,
36]. Present study contributed fascinating data to the distinct function of two mTOR complexes in terms of bioenergetic regulation. The high glycolytic activity with impaired TCA cycle in HT-1080 cells was accompanied by dominant mTORC1 activity, which is in harmony with the reported mTORC1 activity related promotion of glycolysis and the detected GLUT-1 and HIF1α expression. These correlate to the previously described HIF1α expression in HT-1080 and ZR-75.1 at normal condition which could be induced further in hypoxia [
37,
38]. On the other hand mTORC2 is rather closely related to oxidative phosphorylation as it is localized in the endoplasmic reticulum sub-compartment which is bound to mitochondria [
39,
40]. In this respect it is noteworthy, that expression of mTORC2 complex related Rictor protein was dominant and the related high amount of p-Akt was also detected in ZR-75.1 tumour cells bearing intact TCA cycle. However, the glycolytic capacity and the related GLUT-1 expression, glucose consumption are also not negligible in ZR-75.1 cells.
Conclusion
In summary, the present study demonstrated the feasibility of novel and cost-effective assays to define the bioenergetic signature based on the comparison of the cellular utilisation of acetate and glucose and to study further specific elements in the regulation of the energetic homeostasis of the cells. The feasibility of this complex energetic profiling assay system (applying minimum two labelling substrates—
13C glucose and acetate in the measurements of selected metabolites and their ratio and evaluating the number of incorporated
13C atoms into certain metabolites—for example into citrate) could define the bioenergetic signature of tumour cells. Moreover, the applied experimental setting provided further evidence for the relationship between the altered phenotype and the impaired/functioning TCA cycle in HT-1080/ZR-75.1 cells. Furthermore the presented complex test system correlated well with the found regulatory differences at protein level—such as mTOR complexes, G6PDH, GAPDH, GLUT-1, β-F1-ATP-ase, HO-1, IL-1ß, IL-6—, and with substrate oxidation/AEC results (Table
4). Based on our observations the presented assays may offer a possibility to characterise metabolic subtypes of human tumours and provide guidelines to find biomarkers for prediction and development of new metabolism related targets in personalized therapy.