Intracellular ATP Levels are a Pivotal Determinant of Chemoresistance in Colon Cancer Cells

Abstract

Altered metabolism in cancer cells is suspected to contribute to chemoresistance but the precise mechanisms are unclear. Here we show that intracellular ATP levels are a core determinant in the development of acquired cross-drug resistance of human colon cancer cells that harbor different genetic backgrounds. Drug-resistant cells were characterized by defective mitochondrial ATP production, elevated aerobic glycolysis, higher absolute levels of intracellular ATP and enhanced HIF-1α-mediated signaling. Interestingly, direct delivery of ATP into cross-chemoresistant cells destabilized HIF-1α and inhibited glycolysis. Thus, drug-resistant cells exhibit a greater “ATP debt” defined as the extra amount of ATP needed to maintain homeostasis of survival pathways under genotoxic stress. Direct delivery of ATP was sufficient to render drug-sensitive cells drug resistant. Conversely, depleting ATP by cell treatment with an inhibitor of glycolysis, 3-bromopyruvate, was sufficient to sensitize cells cross-resistant to multiple chemotherapeutic drugs. In revealing intracellular ATP levels are a core determinant of chemoresistance in colon cancer cells, our findings may offer a foundation for new improvements to colon cancer treatment.

Introduction

Acquired chemoresistance is one of the major challenges in patients with advanced stage malignancies. Studies have revealed genetic and epigenetic alterations that cancer cells acquire to adapt to chemotherapeutic stress for survival purposes. Such alterations include increases in drug efflux, enhanced drug inactivation, enhanced DNA damage repair, mutations of survival-related genes, deregulated growth factor signaling pathways, increases in anti-apoptotic gene expressions, and activation of intracellular survival signaling. Different types of cancer cells may depend more or less on some of these survival mechanisms (1). Advances are being made in identifying gene signatures that may predict drug-specific resistance prior to chemotherapy (2, 3). However, cancers that are resistant to one type of chemotherapeutic are often resistant to other chemotherapeutics, which is often referred to as cross-resistance, ultimately leads to treatment failure in nearly all patients with metastatic disease (4). The phenomenon of cross-resistance indicates that there might be a more centralized mechanism that cancer cells utilize to resist chemotherapeutic stresses.

Cell death can be executed by different mechanisms, including apoptosis, autophagy, necrosis, or combinations of these processes. Although different cell death mechanisms are unique in their molecular signaling cascades, one molecule is involved in the processes that mediate all types of cell death; adenosine-5′-triphosphate (ATP). During late-stage apoptosis, ATP levels sharply drop, mostly because of the loss of mitochondrial function and consumption by ATP-dependent proteases. In autophagy, a rescue process of self-degradation to compensate for energy paucity is also featured with ATP insufficiency prior to cell death (5, 6). During necrosis, depletion of ATP precedes mitochondrial permeability changes (7). The fact that ATP deprivation occurs in all types of cell death suggests that energy metabolism may play a critical role in the survival of cancer cells under stress.

Altered energy metabolism, such as enhanced aerobic glycolysis, is not only one of the fundamental phenotypes of malignant tumors but also plays important roles during tumor progression, metastasis and relapse (8-10). The role of metabolic alterations of cancer cells in the development of acquired chemoresistance remains to be elucidated.

In this study, we developed cross-chemoresistant human colon cancer cell lines of different genetic backgrounds, HT29 (mutated BRAF and P53, microsatellite stable) and HCT116 (mutated KRAS and PIK3CA, microsatellite unstable) (11), by long-term culturing of cells in the presence of chemotherapeutic agent, oxaliplatin. We compared the glucose metabolism of drug-resistant cells with that of their parental cells, and investigated the role of absolute level of intracellular ATP in the development of drug resistance of these cells. We found that drug-resistant cells demonstrated defective mitochondrial ATP production, enhanced aerobic glycolysis, higher intracellular ATP levels, and enhanced hypoxia-inducible factor-1 alpha (HIF-1α) mediated signaling that was reverted by direct delivery of ATP into cells. Importantly, direct delivery of ATP into cells caused drug-sensitive parental cells to become drug resistant.

Materials and Methods

Cell Lines and In vitro Chemoresistant Model

The human colon cancer cell lines HT29 and HCT116 were obtained from the American Type Culture Collection (ATCC). The oxaliplatin-resistant cell lines HT29-OxR and HCT116-OxR were developed in our laboratory as previously described (12). Briefly, cells stably resistant to oxaliplatin were developed by exposing parental HT29 and HCT116 cells to an initial oxaliplatin dose of 0.1 μM and culturing surviving cells to a confluence of 80% for three passages (∼6 wks). The cells that survived initial oxaliplatin treatment were then exposed to 0.5 μM oxaliplatin for three passages (∼8 wks) and then 1.0 μM for three passages (∼8 wks). Finally, the oxaliplatin concentration was increased to the clinically relevant plasma concentration of 2 μM for 3 wk (∼10 wks). The surviving resistant cells were named HT29-OxR and HCT116-OxR. All cells were cultured in minimal essential medium (MEM) containing 5mM glucose and supplemented with 10% fetal bovine serum (FBS), vitamins, nonessential amino acids, penicillin-streptomycin, sodium pyruvate, and L-glutamine (Life Technologies). Oxaliplatin-resistant cells were continuously cultured in 2 μM oxaliplatin unless otherwise indicated. Cell viability was measured by a Vi-cell XR cell viability analyzer (Beckman Coulter Inc.). In vitro experiments were carried out at 70% cell confluence and confirmed in at least three independent experiments. All cell lines are authenticated by short tandem repeats (STR) sequencing and matched with 100% accuracy to the ATCC database.

MTT Assay for IC₅₀ Determination

Cell growth inhibition was determined by MTT assay in 96-well plates. First, 1500-3000 cells/well/100 μl were seeded in 96-well plates. On the same day, 100 μl working stock of drug solution of oxaliplatin or 5-fluorouracil (5-FU) with 2× concentration of the final concentration was added to the cell suspension. After 72h drug incubation, 50 μl MTT reagent (3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide, 3 mg/ml) was added to each well and incubated for 4h. After the supernatant was removed, the formazan precipitates in the cells were dissolved in 200 μl DMSO. Absorbance was determined using a MultiSkan plate reader (LabSystems) at 570 nm. Fractional survival was plotted against logarithm of drug dose, and IC₅₀ values were calculated by Prisms software (GraphPad Software). Oxaliplatin and 5-FU were purchased from the MD Anderson Cancer Center pharmacy. Both stock drugs were reconstituted in distilled water and maintained at room temperature.

Measurement of Cellular ATP, ADP and AMP

Relative cellular ATP content was measured using the ATP-based CellTiter-Glo Luminescent Cell Viability kit (Promega) with modifications from the manufacturer's protocol. Briefly, cells were plated in 24-well plates at 20,000 cells/well to allow for attachment overnight. At the desired harvest time, an equal volume of the single-one-step reagent provided by the kit was added to each well and rocked for 15 min at room temperature. Cellular ATP content was measured using a luminescent plate reader. An additional plate with the same setup was used for cell counting by hemocytometer to normalize the cell number for calculating ATP level. The absolute amounts of cellular ATP, ADP, and AMP content were measured using the HPLC-MS method. Briefly, exponentially growing cells were trypsinized and washed with 2.5% glycerol (V/V) once. Cell pellets were frozen immediately in 100% ethanol with dry ice. Cell pellets were resuspended in 1 ml distilled water for ultrasonic fragmentation to release cellular ATP, ADP and AMP. After centrifugation at 15,000 rpm for 5 min, the supernatant portions were collected to inject into the HPLC-MS machine for measurement of ATP, ADP, and AMP using the standard protocol. The total cellular ATP, ADP and AMP amount was normalized by cell number.

Spectrophotometric Assay for PFK Activity

PFK activity was determined as described previously (13) in an assay mixture containing 50 mm Tris-HCl (pH 8.0), 50mM KCl, 5 mm MgCl₂, 10 mm (NH₄)₂SO₄, 5mM DTT, 10mM fructose 6-phosphate, 0.28mM NADH, 3 units/ml aldolase, 30 units/ml triosephosphate isomerase, 10 units/ml α-glycerophosphate dehydrogenase, and cell lysate containing 50 μg/ml total protein in a final volume of 100 μl. The cell lysate was incubated with coupling enzymes at 37°C for 10min. The reaction was started by the addition of NADH and fructose-6-phosphate, and NADH oxidation was followed by measuring the decrease in absorbance at 340 nm measured by a microplate reader at 37°C. Blank controls do not contain the cell lysate were used to control nonspecific NADH oxidation. The reaction curves represent the average of triplicate results.

In-Plate Measurement of Mitochondrial Oxygen Consumption

Oxygen consumption rates (OCR) and extracellular acidification rates (ECAR) were measured by a Seahorse XF 24 analyzer (Seahorse Bioscience). Briefly, cells were seeded in 24-well cell culture microplates according to their growth rate and allowed to attach overnight. Approximately 30 min prior to the assay, culture medium was changed to unruffled Seahorse assay medium, and OCR/ECAR measurement was taken by the instrument every 10 min up to 2h. An average of quadruplicate readings for one cell line was recorded and normalized by the number of cells, which were counted afterward using a Vi-cell XR cell viability analyzer (Beckman Coulter).

Transmission Electron Microscopy

Parental and drug-resistant cells were plated in 6-well plates according to their growth rate and cultured to 70% confluency, at which time the cells were fixed with a solution containing 3% glutaraldehyde plus 2% paraformaldehyde in 0.1 M cocodlate buffer, pH 7.3, for 1h. After fixation, the samples were submitted to the high-resolution electron microscopy facility at MD Anderson Cancer Center for processing using a JEM-1010 transmission electron microscope (JEOL, USA, Inc.). Digital images were obtained using the AMT Imaging System (Advanced Microscopy Techniques). For each cell type, at least 20 individual cells were imaged and analyzed.

Mitochondrial Isolation and Substrate ATP Production

Mitochondrial ATP production was quantified using the ATP Bioluminescent Assay obtained from Sigma (FL-AA) with minor modifications to the manufacturer's protocol. ATP assay dilution buffer and ATP assay mix were prepared as directed using 40 μl of the ATP assay mix and 2 ml of the dilution buffer, aliquoted in advance, and stored at −80°C. The measurements were performed using a white 96-well plate with a 100-μl total assay volume in each well. To each well, 41 μl of mitochondria (0.2-1.0 μg) and 4 μl of an electron transport substrate (either 500 mM glutamate plus 500 mM malate, or 500 mM succinate) was added. To the mitochondrial substrate mixture, 5 μl of 100 μM ADP was added to each reaction to initiate ATP production at time zero. The reaction was incubated between 30 and 45s, followed by the addition of 50 μl of assay reaction mixture for bioluminescence measurements. Luminescence was recorded, in triplicate, using a Bio-Tek Instruments FLx-800 Fluorescent Microplate Reader equipped with KC4 BioTek software approximately 2 min after the addition of ADP (initial measurement), and measurements were continued for 10 min using the kinetic measurement mode at 1-min intervals at a sensitivity level of 100. A 0-0.25 nM ATP standard curve was generated using a standard solution of 10 μM ATP. ATP stock concentration was determined at 259 nm using E_mM = 15.4 mM⁻¹cm⁻¹(14). The value of ATP produced by mitochondria within 1 min was normalized by the protein content of the mitochondria, which is expressed as “nmol ATP/min/mg protein”.

Measurement of Cellular Glycolytic Activity

Glycolytic activity in the parental and resistant colon cancer cells was determined by measuring glucose consumption and lactate production. Cells were seeded in 6-well cell culture microplates at 0.5-1×10⁶ cells/well/3 ml according to their growth rate and allowed to attach overnight. Fresh medium was replaced the next morning approximately 3h prior to the assay. Cell culture medium was sampled at 200 μl at three time points at 3h intervals. The glucose and lactic acid concentrations of the cell culture medium were measured using a Dual-Channel Biochemistry Analyzer-2700D (YSI Life Sciences). For each time point, triplicate samples were measured to calculate the average values. Cell numbers were counted with a Beckman Coulter cell viability analyzer to normalize the glucose and lactic acid concentrations.

Western Blot and Antibodies

The antibody against PARP was purchased from Cell Signaling. HIF-1α antibody was purchased from BD Transduction Laboratories. Glut-1 antibody was purchased from Abcam. β-actin, p21, and hexokinase II antibodies were purchased from Santa Cruz Biotechnology. VEGFA antibody was purchased from R&D Systems. Whole-cell lysate was collected from cells cultured at 70% confluency before Western blot analysis was applied. VEGF-A secretion into culture medium was examined using conditioned medium collected and analyzed by Western blot as previously described (15).

Quantitative Real Time-PCR for VEGF-A

VEGF-A gene expression level in the parental and resistant cells was analyzed by quantitative real time-PCR using Taqman primers (Applied Biosystems) specific for the VEGF-A gene. Actin was used as an internal control. Total RNA was extracted from 60 to 70% confluent tumor cells growing in culture using TRIzol reagent (Life Technologies). Reverse transcription PCR was performed using the First-Step RT-PCR Kit (Invitrogen). Real time-PCR reaction was performed using PCR Master Mix (Roche) on ABI-7500 platform (Applied Biosystems).

Liposome-encapsulated ATP Delivery

Encapsulation of ATP into liposomes was carried out according to a published protocol (16). In brief, chloroform solution of phosphatidylcholine (0.26 mmol), cholesterol (0.12 mmol), and 1,2 dioleoyl-3-trimethyl-ammonium-propane (10.8 μmol) was evaporated, and the film formed was hydrated with/without 5 ml of 600 mM ATP in Krebs-Henseleit (KH) buffer. The dispersion was frozen at −80°C for 30 min followed by thawing at 45°C for 5 min; this cycle was repeated five times. Non-encapsulated ATP was separated by dialysis against the KH buffer at 48°C overnight. For each treatment of cells with liposomes, 10 μl of either control or ATP liposomes were added into one well of a 6-well cell culture plate; the well contained 500 μl cell culture medium. Cells were harvested for analyses at indicated time points after liposome treatment.

Results

Characterization of Chemoresistant Phenotype of Cross-resistant Cell Lines

Cells stably resistant to oxaliplatin were developed by exposing parental HT29 and HCT116 cells to increasing doses of oxaliplatin over successive passages as described in the Experimental Procedures section (12). The surviving resistant cell lines were named HT29-OxR and HCT116-OxR. The proliferation rates of both HT29-OxR and HCT116-OxR cells were slower than those of their parental cells, HT29 and HCT116 (Fig. 1A and Table S1), consistent with an upregulation of cell cycle regulator p21 (Fig. S1). To confirm the chemoresistant phenotype after chronic oxaliplatin exposure, we compared the effect of oxaliplatin on HT29-OxR and HCT116-OxR cells to the effect on their parental cells by a 72-hr MTT assay (Fig. 1B). As expected, HT29-OxR and HCT116-OxR cells were more resistant to the cytotoxic effects of oxaliplatin compared to parental, chemo-naïve cells.

Figure 1. Characterization of cross-chemoresistant cancer cell lines.

A, The drug resistant cells were continuously maintained in media with 2 μM oxaliplatin. Before plating for each experiment, media without drug was used for two passages. HT29-OxR and HCT116-OxR cells grew more slowly than their parental cells. See also Table S1. B, Survival inhibition curves from MTT assays showed a chemoresistant phenotype of HT29-OxR and HCT116-OxR cells to oxaliplatin (top panels) and 5-FU (bottom panels). C, HT29-OxR and HCT116-OxR showed decreased PARP cleavage (and thus decreased apoptosis) under 24h treatment with oxaliplatin (top panels) and 5-FU (bottom panels). Actin served as a loading control. All experiments were repeated at least three times. Representative data is shown.

To test the cross-chemoresistance of HT29-OxR and HCT116-OxR cells, we treated the resistant cells with 5-fluorouracil (5-FU), a drug with a different mechanism of action from that of oxaliplatin, and observed cross-resistance of both cell lines to 5-FU. The chemoresistant phenotype was also reflected by >130-fold and 30-fold increases in the IC₅₀ to oxaliplatin in the HT29-OxR and HCT116-OxR cells, respectively, and 3-fold and 5-fold increases in the IC₅₀ to 5-FU (Table 1). Furthermore, exposure of both parental and resistant cells to increasing concentrations of oxaliplatin (0.2, 2, and 20 μM) and 5-FU (0.2, 2, and 20 μg/ml) for 24h induced a concentration-dependent apoptotic event in the parental cells, but not in the resistant cells, as reflected by detection of PARP, caspase 3 cleavage by Western blot (Fig. 1C) and measurement of annexin V-binding population by flow cytometry (Fig. S2A and Fig. S2B) Thus, the stably selected HT29-OxR and HCT116-OxR cells display a phenotype of cross-chemoresistance to both oxaliplatin and 5-FU.

Table 1. IC₅₀ values for oxaliplatin and 5-FU in parental (Par) and oxaliplatin-resistant colon cancer cells.

IC50	HT29			HCT116
	Par	OxR	O/P	Par	OxR	O/P
Oxal:	0.15 μM	>20 μM	>130	0.3 μM	9 μM	30
5-FU:	1 μg/ml	3 μg/ml	3	0.4 μg/ml	2 μg/ml	5

Par, parental cells. O/P, ratio of OxR to parental cell values.

Metabolic Alterations and Mitochondrial Defects of Chemoresistant Cells

We hypothesized that under chronic exposure to genotoxic stressors, such as oxaliplatin and 5-FU, the surviving resistant cells will need to alter their energy metabolism to adapt to the continuous stress. Thus, we examined the cellular ATP levels of HT29-OxR, HCT116-OxR, and their respective parental cell lines using an ATP-based luminescent assay. Both resistant cell lines showed a 2-fold increase in total cellular ATP levels compared to their parental cells (Fig. 2A), indicating that a higher amount of ATP was produced in the resistant cells.

Figure 2. Measurement of intracellular ATP and mitochondrial activity.

A, HT29-OxR and HCT116-OxR cells displayed more than a 2-fold increase in intracellular ATP levels compared with their parental cells. B, HT29-OxR and HCT116-OxR cells consumed oxygen at a significantly higher rate than their parental cells as indicated by oxygen consumption rate (OXR, pmole/min). C, HT29-OxR and HCT116-OxR cells displayed morphologic changes as examined by TEM. Bar: 2 μm. At least 20 individual cells were analyzed for each cell type. Representative images are shown. D, Purified mitochondria from HT29-OxR and HCT116-OxR showed decreased substrate ATP production ability from both complex I (glutamate/malate as substrates) and complex II (succinate as a substrate) compared with that of their parental cells. All experiments were repeated at least three times. Representative data is shown.

Cells produce ATP through two related mechanisms: glycolysis and mitochondrial oxidative phosphorylation. To determine which was involved in the chemoresistant cells, we measured the oxygen consumption rate using a Seahorse Bioenergizer (Seahorse Bioscience), which is often an indicator of mitochondrial oxidative phosphorylation activity (17). Interestingly, the HT29-OxR and HCT116-OxR cells exhibited a significantly elevated cellular oxygen consumption rate, suggesting the OxR cells may produce more ATP through upregulation of their mitochondrial oxidative phosphorylation (Fig. 2B).

To further determine changes in mitochondria, we examined the ultra-structure of mitochondria by transmission electronic microscopy (TEM). As shown in Fig. 2C, the TEM images showed two different patterns of alterations in the mitochondria of HT29-OxR and HCT116-OxR cells. The morphology of mitochondria of HT29-OxR is similar to the mitochondria of its parental cells; however, the mitochondria of HCT116-OxR cells are more elongated than that of its parental cells. A more condensed mitochondrial form was observed as a common feature in both resistant cell lines. The number of mitochondria in HT29-OxR cells did not change but increased in in HCT116-OxR cells as compared to their parental cells, respectively (Fig. S3).

The data demonstrating enhanced oxygen consumption and the increased density of mitochondria in the resistant cells suggest that the mitochondria of these cells are more active, possibly for the purpose of generating more ATP. To test this, we measured the ATP-producing capacity of mitochondria isolated from the parental and resistant cells using complex-I substrates glutamate and malate and complex-II substrate succinate. Surprisingly, the mitochondria of resistant cells actually were significantly less capable of producing ATP than were their parental cells (Fig. 2D). In addition, the reserve mitochondrial oxidative phosphorylation capacity was significantly lower in the OxR cells after treatment with the uncoupler carbonilcyanide p-triflouromethoxyphenylhydrazone (FCCP), as measured by the Seahorse Bioenergizer, suggesting defective mitochondrial function (Fig. S4A). Taken together, these data suggest that the resistant cells utilized an alternative energy-producing mechanism to maintain a higher ATP levels.

Elevated Glycolysis of Chemoresistant Cells

To test whether the cross-chemoresistant cells also have altered aerobic glycolysis, we measured several key glycolytic parameters, including glucose consumption, lactate production, and expression levels of glycolytic enzymes in the resistant cells. As shown in Fig. 3A, both HT29-OxR and HCT116-OxR cells consumed more glucose than their parental cells. Accordingly, the resistant cells released more lactate into the media (Fig. 3B). Consistently, a comprehensive glycolytic parameter, extracellular acidification rate (ECAR), was significantly higher in the OxR cells compared with their parental cells, suggesting a higher glycolytic phenotype of the OxR cells (S4B) In addition, the protein expression of glucose transporter (Glut-1) and hexokinase (HK2) were all significantly upregulated in HT29-OxR and HCT116-OxR cells (Fig. 3C).

Figure 3. Characterization of glycolytic activity and HIF-1α activity.

A, HT29-OxR and HCT116-OxR cells consumed glucose at a higher rate than their parental cells. B, HT29-OxR and HCT116-OxR cells produced lactate at a higher rate than their parental cells. C, Key glycolytic enzymes GLUT1, HK2, LDHA, and HIF-1α were upregulated in HT29-OxR and HCT116-OxR cells compared with their parental (Par) cells, as shown by Western blot. D, HT29-OxR and HCT116-OxR produced more VEGFA into cell culture medium than their parental cells (top panel), which correlated with VEGFA mRNA level by qPCR (bottom panel). All experiments were repeated at least three times. Representative data is shown.

A master regulator of glycolysis is HIF-1α (18), which is regulated by oxygen-dependent and -independent mechanisms (19). We next examined HIF-1α expression and found that HIF-1α was upregulated in both HT29-OxR and HCT116-OxR cells under normoxic conditions. To further prove a functional role of HIF-1α in the resistant cells, we examined the expression of vascular endothelial growth factor A (VEGF-A), a known downstream gene of HIF-1α. We found several isoforms of VEGFA were increased in the HT29-OxR and HCT116-OxR conditioned media including the VEGFA₁₆₅ (Fig. 3D and S5). Consistently, the mRNA levels of VEGFA were significantly upregulated in both resistant cell lines (Fig. 3D), suggesting a functional HIF1α pathway was elevated in the OxR cells.

Intracellular ATP Regulates HIF-1α and Induces Drug Resistant Phenotype

The higher intracellular ATP level and defective mitochondrial ATP production in the chemoresistant cells as compared to the parental cells indicated that the resistant cells might need more ATP to survive under genotoxic stress. It is likely that the drug resistant cells may turn to glycolysis for more rapid ATP generation. It is known that the ADP/AMP ratio determines the glycolytic rate through stereotypic regulation of the key glycolytic enzyme phosphofructose kinase (PFK) and drives glycolysis (20). Therefore we measured the levels of intracellular ADP and AMP using high-performance liquid chromatography coupled mass spectrum (HPLC-MS). As shown in Fig. 4A, the ratio of ADP/AMP was much lower in HT29-OxR and HCT116-OxR cells than in their parental controls. We further measured the enzyme activity of PFK by a well-established in vitro biochemical assay (13) over a time course of 5-20 minutes using extracted cell lysates. In consistent with the decreased ADP/AMP ratio, the PFK enzyme activity was elevated in HT29-OxR cells and to a much higher level in HCT116-OxR cells as compared with their parental cells (Fig. 4B).

Figure 4. ATP regulates HIF-1α expression and induces a drug-resistant phenotype.

A, Ratio between absolute cellular ADP and AMP amount decreased in HT29-OxR and HCT116-OxR cells, as shown by HPLC-MS measurement. B, PFK enzyme activity is elevated in HT29-OxR and HCT116-OxR cells compared with parental cells. The enzyme activity is reflected by the decrease of absorbance at 340nm (the Y-axis, unit) over the indicated time course (the X-axis, minute). (Baseline, blank control). C, ATP liposomal delivery to parental HT29 and HCT116 cells induced resistance to oxaliplatin under 72h treatment, as shown by MTT assay. D, ATP liposomal delivery to HT29-OxR and HCT116-OxR cells decreased HIF-1α expression and reverted glycolytic enzyme HK2 expression. All experiments were repeated at least three times. Representative data is shown.

The activity of aerobic glycolysis, level of metabolic regulator HIF-1α, and cellular ATP level were higher in the resistant cells, which suggest that ATP has a protective role for cancer cells under genotoxic stress. To test this possibility, we artificially delivered ATP packaged in liposomal vehicles to the parental HT29 and HCT116 cells by transient transfection and examined the drug treatment effect on these cells. As shown in Fig. 4C, exogenous ATP supplementation partially blocked the cytotoxic effect of oxaliplatin in both cell lines, indicating the role of intracellular ATP level in mediating the drug-resistant phenotype.

We further hypothesized that the resistant cells are in need of more ATP; thus, a faster ATP production mechanism, namely aerobic glycolysis, is elevated, which in turn demands a higher level of HIF-1α to enhance/sustain aerobic glycolysis (Fig. 3C). To test whether exogenous ATP delivery could affect HIF-1α level and glycolysis, we treated the resistant cells with ATP liposomes and examined the expression levels of HIF-1α and glycolytic enzymes. Consistent with our hypothesis, ATP treatment decreased HIF-1α rapidly in both HT29-OxR and HCT116-OxR cells, and reversed the glycolytic phenotype, as evidenced by a concordant decrease in hexokinase II expression in both resistant cell lines (Fig. 4D). Interestingly, the level of Glut-1 was not significantly altered at 2h ATP treatment, which is probably due to the protein turnover kinetics of Glut-1 is different from that of hexokinase II.

If the intracellular ATP level plays a central role in sustaining the survival of cross-chemoresistant cells, depletion of ATP should sensitize these cells to chemotherapeutic agents. As shown in Fig. 5A, 3-bromopyruvate (3-BrPA), a widely used ATP-depleting agent (21), depleted cellular ATP in HT29-OxR and HCT116-OxR cells in a dose-dependent manner. We pre-treated the resistant cells with a moderate does of 3-BrPA (30 μM) for 24h, then tested their sensitivity to oxaliplatin and 5-FU. As shown in Fig. 5B-5C, ATP depletion partially reversed the drug-resistant phenotype and re-sensitized both resistant cell lines to oxaliplatin and 5-FU treatment. The IC₅₀ values of oxaliplatin and 5-FU in the OxR cells before and after ATP depletion by 3-BrPA treatment were calculated. There was a 2-3 fold decrease in the IC₅₀ in both OxR cell lines after ATP depletion (Supplemental Table 2). Taken together, our data suggest that intracellular ATP plays a central role in regulation of drug-resistant phenotype of colon cancer cells.

Figure 5. ATP depletion sensitizes drug-resistant cells to chemoreagent.

A, 3-Bromopyruvate (3-BrPA) depleted cellular ATP level in HT29-OxR and HCT116-OxR cells after 24h of treatment, as measured by ATP luminescent activity. B, Pre-treatment of 3-BrPA (30μM) for 24h sensitized HT29-OxR and HCT116-OxR cells to oxaliplatin, as shown in 72h MTT assays. C, Pre-treatment of 3-BrPA (30μM) for 24h sensitized HT29-OxR and HCT116-OxR cells to 5-FU, as shown in 72h MTT assays. All experiments were repeated at least three times. Representative data is shown.

Discussion

ATP, the energy currency of the cell, is produced from carbon fuels at two levels of metabolism, glycolysis in the cytosol and oxidative phosphorylation in the mitochondria. Glycolysis, besides producing pyruvate for oxidative phosphorylation under non-stressful conditions, generates two moles of ATP from one mole of glucose. In normal cells undergoing stress, for example, due to a sudden drop of intracellular ATP and hypoxia, cells will accelerate glycolysis to produce ATP in order to meet the immediate energy need; accelerated glycolysis leads to increased production of lactate from pyruvate (22, 23). Although the efficiency of ATP production via glycolysis is lower than that of oxidative phosphorylation, the ATP generation rate of glycolysis is nearly 100 times faster than that of oxidative phosphorylation. In tumors, elevated aerobic glycolysis differentiates malignant tumors from benign tumors and non-tumor cells, a phenomenon that was identified by Otto Warburg (24) and has been repeatedly confirmed. One possible driver of the aerobic glycolysis of malignant tumor cells could be a higher ATP demand for the purposes of survival and growth as compared to demand in non-malignant or normal cells (25). Our data support the hypothesis that metabolic shift driven by higher ATP demand is also applicable to the progression of acquired chemoresistance of cancer cells.

To cope with constant chemotherapeutic stress, chemoresistant cancer cells are known to do at least one of the following: increase drug efflux, enhance drug inactivation, enhance DNA damage repair, mutate survival-related genes, deregulate growth factor signaling pathways, increase expression of anti-apoptotic genes, and/or activate intracellular survival signaling (1, 4). All of these activities consume ATP. Our data showing elevated aerobic glycolysis in the chemoresistant cells and sensitization of the chemoresistant cells by inhibition of glycolysis further argue that enhanced aerobic glycolysis of cancer cells occurs to provide the extra amount of ATP needed for chemoresistant cells to survive under stress. It is worth noting that the chemoresistant cells consume more oxygen while their mitochondrial ATP production is defective (Fig. 3D). Oxygen consumption has often been used as an indicator of mitochondrial ATP productivity. Our data support the argument that oxygen consumption is not a reliable index for mitochondrial ATP productivity (26). The role of enhanced oxygen consumption by chemoresistant cells warrants further investigation.

Extracellular ATP can be taken up by cells via adenosine transporters (27); however, because of signaling effects triggered by adenosine receptors (28), treating cells with naked ATP results in a complex of signaling responses besides an increase in the intracellular ATP level. Liposome-encapsulated ATP has been used to deliver ATP into cells in vivo(29-32), which allows us to avoid interfering with the signaling mediated by adenosine receptors. The comprehensive effects of delivering ATP directly into cells, reverting the aerobic glycolytic activity and HIF-1α level of cross-chemoresistant cells, and converting chemosensitive cells to resistant cells, further indicate that the chemoresistant cells are in need of more ATP.

In tumor tissues, upregulation of HIF-1α cannot be fully attributed to hypoxia. Increased expression of HIF-1α is often found in cancer cells of tissues in which oxygen supply is sufficient (33-35), and HIF-1α is upregulated only in the malignant tumor cells and not the benign tumor cells with the same tissue origin (36-39). For example, glucose deprivation induces HIF-1α expression in the presence of an ample amount of oxygen (40). Our finding of the oxygen-independent degradation of HIF-1α induced by the increase of intracellular ATP is novel. While the molecular mechanism is currently under investigation, it bears important implications for cancer biology. The data suggest the possibility that insufficiency of intracellular ATP might be a driving force of oxygen-independent HIF-1α signaling in cancer cells, which in turn elevates aerobic glycolysis to produce ATP more efficiently to meet the increased demand of cancer cells. This possibility is supported by a recent finding that malignant tumor cells consume a higher amount of ATP via the endoplasmic reticulum enzyme UDPase ENTPD5, that, in turn enhances aerobic glycolysis (41).

Based on the contradiction between a higher amount of ATP, HIF-1α, and glycolysis in chemoresistant cells and downregulation of HIF-1α and glycolysis by exogenous ATP delivery, and the chemoresistant effect of ATP on chemosensitive cells, we propose the concept of “ATP debt”. ATP debt is the extra amount of ATP needed to maintain survival homeostatic pathways in cancer cells, which is equal to ATP needed for survival minus ATP produced. The decreased ADP/AMP ratio and increased PFK activity in the drug resistant cells as compared with the parental cells indicates a higher ATP consumption rate in the drug resistant cells. Thus the drug resistant cells demand a faster ATP generating mechanism to maintain survival, which is met by aerobic glycolysis/fermentation. Under these conditions, HIF-1α, as a key glycolysis regulator, is up-regulated independent of oxygen levels. The steady-state higher level of HIF-1α in the drug resistant cells and its down-regulation by increased ATP levels suggests that more ATP is needed to cope with chemotherapeutic stress, i.e. the drug resistant cells have higher “ATP debt”. The level of HIF-1α under normoxic conditions might be an indicator of ATP debt. Understanding the molecular mechanisms that cancer cells utilize to reduce their ATP debt may provide the foundation for the development of novel therapeutic strategies.