Abstract
Extracellular acidity is a hallmark of cancers and is independent of hypoxia. Because acidity potentiates malignant phenotypes, therapeutic strategies that enhance the targeting of oncogenic mechanisms in an acidic microenvironment should be effective. We report here that drugs which abrogate mitochondrial respiration show enhanced cytotoxicity against melanoma cells in a normoxic but acidic extracellular pH, independent from P53 mutations, BRAF (V600E) mutations, and/or resistance against BRAF inhibitors. Conversely, the cytotoxicity against melanoma cells of mitochondrial inhibitors is impaired by a neutral or alkaline extracellular pH, and in vivo systemic alkalinization with NaHCO3 enhanced subcutaneous tumor growth and lung metastasis of B16F10 cells in mice treated with the mitochondrial inhibitor phenformin. Intracellular calcium (Ca2+) was significantly increased in melanoma cells treated with mitochondrial inhibitors at an acidic extracellular pH and an intracellular Ca2+ chelator, BAPTA/AM, inhibited cytoplasmic Ca2+ as well as melanoma cell death. Surprisingly, ROS scavengers synergized with increased apoptosis in cells treated with mitochondrial inhibitors, suggesting that ROS contributes to cell survival in this context. Notably, the cytotoxic enhancement of mitochondrial inhibitors by acidity was distinct from PGC1alpha-driven mitochondrial addiction, from therapy-induced senescence, and from slow, JARID1B-high–associated cell cycling, all of which have been shown to promote vulnerability to mitochondrial inhibition. These data indicate that extracellular pH profoundly modulates the cytotoxicity of mitochondrial inhibitors against cancer cells. Mol Cancer Ther; 16(5); 936–47. ©2017 AACR.
Introduction
Melanoma, the most aggressive form of skin cancer, typically responds poorly to conventional cancer therapies, including chemotherapy and radiotherapy, after it progresses and subsequently metastasizes (1). Although new therapies such as BRAF inhibitors (e.g., vemurafenib) and immunotherapies prolong the survival of many patients, relapses occur in most cases (2, 3). Accordingly, identifying new therapies and ascertaining ways in which to combine and potentiate existing therapies are high research priorities in this disease.
Recently, a subpopulation was identified of proliferative but slow-cycling melanoma cells that express the histone3 K4 demethylase JARID1B (4, 5). These cells showed enhanced mitochondrial oxidative phosphorylation (OXPHOS) and were relatively resistant to various drugs including vemurafenib, thus playing a critical role in melanoma cell survival. Furthermore, it was reported that the mitochondrial master regulator gene PGC1alpha is regulated in melanoma cells through microphthalmia-associated transcription factor (MITF), the central melanocytic lineage regulator, resulting in mitochondrial addiction, tolerance to vemurafenib (6), and oxidative stress (7). As PGC1alpha-driven mitochondria-addicted melanoma cells (6) as well as JARID1B-high slow-cycling cells (5, 8) are sensitive to drugs that inhibit OXPHOS (OXPHOSi), OXPHOSi have therapeutic potential against melanoma (9–11).
Interest in OXPHOSi is also driven by links between mitochondrial activity and therapy-induced senescence (TIS; ref. 12). Therapies such as radiotherapy and chemotherapy can drive either apoptosis or senescence in cancer cells, with senescent cells appearing in tumor remnants that remain after therapy. These cells are important as they may possess a slow-cycling, drug-resistant phenotype that serves as a reservoir for recurrent disease (12). Importantly, senescent cancer cells also exhibit activation of OXPHOS and sensitivity to OXPHOSi (12, 13).
The extracellular pH (pHe) is acidic in most cancers due to hypoxia, insufficient perfusion, and aerobic glycolysis (the so-called “Warburg effect”), although hypoxia and extracellular acidity often lack spatial correlation (14). The pHe is reportedly 6.2 to 6.9 in tumor tissues, whereas the pHe is 7.3 to 7.4 in normal tissues (15, 16). Extracellular acidity, independently from hypoxia, facilitates malignant behaviors of cancer cells including migration, invasion, epithelial–mesenchymal transition, metastasis, lymphangiogenesis, autophagy, mutagenesis, chemoresistance, immune evasion, and dedifferentiation (17–29). Indeed, high intratumoral acidity is linked to poor patient prognosis (30), and alkalizer therapy that manipulates the acidic pHe in a tumor toward neutral was reported to inhibit tumor growth and metastasis in several types of cancers (20, 28, 31, 32). These observations highlight the importance of considering pHe in cancer treatment. In this study, we assessed the effects of an acidic pHe on melanoma cells and on modulating the antimelanoma effects of OXPHOSi.
Materials and Methods
Cell lines and cell culture
B16F10 (ATCC Number CRL-6475) mouse melanoma cells were purchased from the American Type Culture Collection (ATCC) in April 2009. Human melanoma cell lines MEWO (ATCC Number HTB-65, in January 2012), SK-MEL-28 (ATCC Number HTB-72, in October 2014), and A375 (ATCC Number CRL-1619, in November 2014) were purchased from the ATCC. Normal human dermal fibroblasts were derived from the scalp skin of a 14-year-old female patient in Jun 2003. All cell lines were stored in liquid nitrogen and authenticated for viability a month before use. Cells were grown in DMEM (MediaTech) supplemented with 4.5 g/L glucose, 2 mmol/Ll-glutamine, 5% penicillin–streptomycin, 2.5 μg/mL amphotericin B, and 10% FBS at pH 7.4. All cell lines were cultured in a humidified 5% CO2 atmosphere at 37°C. The pH of media was adjusted with hydrochloric acid, sodium hydroxide, or bicarbonate to 6.2, 6.7, and 8.0 for pHe-manipulation in vitro. The medium was changed every day to normal/acidic medium with/without oligomycin (1404-19-9, Wako Pure Chemical Industries, Ltd), phenformin (P7045, Sigma-Aldrich), rotenone (R8875, Sigma-Aldrich), 2-thenoyltrifluoroacetone (TTFA; T27006, Sigma-Aldrich), cisplatin (15663-27-1, Wako Pure Chemical Industries, Ltd.), vemurafenib (PLX4032, RG7204) (S1267, Selleckchem.com), Z-VAD-FMK (sc-3067, Santa Cruz Biotechnology), Z-DEVD-FMK (FMK004, R&D Systems), Trolox (sc-200810, Santa Cruz Biotechnology), sodium pyruvate (sc-208397, Santa Cruz Biotechnology), aurintricarboxylic acid (ATA; sc-3525, Santa Cruz Biotechnology), E64 (sc-201276, Santa Cruz Biotechnology), antipain (sc-291907, Santa Cruz Biotechnology), and/or BAPTA/AM (1,2-Bis(2-aminophenoxy)ethane-N,N,N′,N′-tetraacetic acid tetrakis(acetoxymethyl ester) (sc-202488, Santa Cruz Biotechnology). To derive vemurafenib-resistant sublines, SK-MEL-28 and A375 cells were seeded at low density, treated with vemurafenib at 10 μmol/L for 6 weeks and clonal colonies isolated and maintained. All in vitro experiments were performed with cells at 70% to 80% confluency in normoxic conditions.
Cell viability assay
The viability of drug-treated cells at each pH was calculated using a Cell Counting Kit-8 (CCK8; Dojindo Laboratories) by the following formula: relative viability = value of drug-treated cells/value of cells with no treatment.
Western blotting
Immunoblotting was performed as previously described (33). Antibodies used were: mouse monoclonal anti-ATP synthase beta IgG (1:10,000; 612518, BD Biosciences), rabbit polyclonal anti–PGC-1 IgG (1:200; sc-13067, Santa Cruz Biotechnology), rabbit polyclonal anti–cleaved caspase-3 p11 IgG (1:200; sc-22171-R, Santa Cruz Biotechnology), rabbit polyclonal anti–cleaved caspase-6 (Asp162) IgG (1:1,000; #9761, Cell Signaling Technology), and anti–β-actin (1:20,000; A5441, Sigma-Aldrich).
Detection of mitochondrial membrane potential
For detection of mitochondrial membrane potential, MitoRed (PromoKine, PK-CA707-70055) was used following the manufacturer's protocol.
Oxygen consumption
The oxygen consumption rate was measured using a Seahorse Bioscience XF96 Extracellular Flux Analyzer. Briefly, cells (2 × 104/well) were cultured in XF96 tissue culture plates at pH 7.4 or pH 6.7 for 72 hours, and then media were changed into unbuffered XF Base Medium 1 hour before the measurement. Maximal oxygen consumption rate was measured under a mitochondrial uncoupler trifluoromethoxy carbonyl cyanide phenylhydrazine (FCCP; 0.25 μmol/L for SK-MEL-28 and 0.1 μmol/L for MEWO, respectively). The number of cells was normalized using CCK8.
Immunocytochemistry
Cells were fixed in 4% paraformaldehyde for 10 minutes at room temperature, and then permeabilized with 0.5% Triton X-100 for 5 minutes. The expression of P53 and P21 was detected by incubating cells with rabbit polyclonal anti-p53 IgG (1:100; sc-6243, Santa Cruz Biotechnology) or rabbit polyclonal anti-p21 IgG (1:100; sc-756, Santa Cruz Biotechnology) for 1 hour at room temperature followed by incubation with Alexa fluor 488–conjugated goat anti-rabbit IgG. DAPI was added to label cell nuclei. The expression of senescence-associated β-galactosidase was detected using the Cellular Senescence Kit (GX0003, OZ Biosciences) following the manufacturer's protocol. Cells were observed using an All-in-one Fluorescence Microscope BZ-9000 (Keyence).
Immunohistochemistry
For TUNEL staining, a TACS2 TdT in situ Apoptosis Detection kit (Trevigen, 20877) and a ChemMate ENVISION/HRP kit (Dako) were used. In brief, 3-μm-thick tumor sections on the slides were covered with Proteinase K Solution for 15 minutes and were then immersed in Quenching Solution for 5 minutes. After washing with PBS, the slides were immersed in 1×TdT Labeling Buffer for 5 minutes and were then covered with Labeling Reaction Mix for 60 minutes at 37°C, followed by covering with Stop Buffer for 5 minutes. The slides were then washed 2 times in deionized H2O for 5 minutes each, followed by treatment with alkaline phosphatase–conjugated streptavidin (K1018; Dako) for 15 minutes. Reaction products were visualized using the Fuchsin+ Substrate-Chromogen System (K0625, Dako). One percent Methyl Green was used as a counterstain.
Ki67 staining was performed as previously described (34). Briefly, sections were autoclaved and stained with rabbit polyclonal anti-Ki67 IgG (1:500; Leica Microsystems), followed by incubation and visualization with a ChemMate ENVISION/HRP kit (Dako).
FACS analysis
Cellular ROS was labeled with CellROX Deep Red Reagent (Ex/Em = 644/665 nm; C10422, Life Technologies) and measured using a BD FACS Canto II (BD Biosciences). Briefly, cells were treated with the reagent (5 μmol/L) with/without the drug(s) at pH 7.4 or 6.7 and were then incubated at 37°C for 30 minutes and collected. Menadione (100 μmol/L, M5625, Sigma-Aldrich) was used for a positive control. After distinguishing dead cells, median fluorescence intensities (MFI) of cells were determined (Ex/Em = 650/660 nm). MFIs of cells in each condition were normalized to cells at pH 7.4 with no treatment.
For apoptosis assays, cells were cultured with/without the drug(s) at pH 7.4 or 6.7 for the indicated periods and then apoptotic cells were detected using FITC Annexin V (640905, BioLegend) according to the manufacturer's protocol using a BD FACS Canto II.
For intracellular calcium measurement, cells were cultured with/without the drug(s) at pH 7.4 or 6.7 for the indicated periods and then loaded with Fluo3/AM (sc-202612, Santa Cruz Biotechnology; 5 μmol/L) in PBS at 37°C for 1 hour. After 2 washes, cells were cultured in PBS at 37°C for 30 minutes (de-esterification) and then analyzed on a BD FACS Canto II with excitation wavelength of 488 mm. For cell sorting, cells stained by Fluo3/AM were analyzed on a BD FACS Aria and sorted by Fluo3/AM fluorescence.
For cell-cycle analysis, 1 × 105 cells were fixed with ice-cold 70% EtOH for 4 hours and then suspended in DNA staining solution, including 2.5 μg/mL propidium iodide and 0.5 mg/mL RNase A in PBS. They were then analyzed using a BD FACS Canto II.
For analysis of JARID1B expression after incubation with/without the drug(s) at pH 7.4 or 6.7 for the indicated periods, 1 × 106 cells were fixed in 70% ethanol for 10 minutes on ice and then washed with 2% BSA in PBS. Cells were then incubated with rabbit polyclonal anti-JARID1B antibody (1:1,000; 22260002, Novus Biologicals) for 30 minutes on ice. After washing, cells were incubated with a secondary fluorescein isothiocyanate–conjugated anti-rabbit antibody for 20 minutes on ice and analyzed using a BD FACS Canto II.
Measurement of lactate
Measurement of lactate was performed using a Lactate Colorimetric Assay Kit II (#K627, BioVision). Briefly, cells were seeded (5 × 103/well) in 96-well plates and were incubated at pH 7.4 or 6.7 for the indicated periods. The culture medium of each well was then recovered and mixed with Reaction Mix. After incubation for 30 minutes at room temperature, OD 450 nm was measured using a microplate reader. Values were normalized by the number of viable cells evaluated using the CCK8 Kit. The values of cells in acidic conditions were normalized to cells at pH 7.4.
Measurement of ATP
Cellular ATP was measured using “Cellno” ATP ASSAY reagent (CA50, Toyo Inc.). Briefly, cells were seeded (5 × 103/well) in 96-well plates and incubated at pH 7.4 or 6.7 with/without the drug(s) for the indicated periods. The reagent was added to each well and the plates were stirred for 1 minute. Luminescence was measured using a CentroXS3 LB960 plate reader (Berthold Technologies). Values were normalized as described in the section of Lactate measurement.
Animals
All animal studies were performed according to protocols approved by the Institutional Animal Care and Use Committee at Osaka University. C57BL/6J mice (7-week-old, female) were purchased from Nippon Charles River. Mice were maintained under light/dark (12/12 hours) cycles with free access to standard chow and water. For mice treated with oral administration of NaHCO3, water was replaced with 200 mmol/L NaHCO3.
Subcutaneous tumor growth
For studies of subcutaneous tumor growth, mice (n = 6 per group) were inoculated with 2.5 × 105 B16F10 cells in their flanks. Mice were randomly divided into four groups: (a) no treatment, (b) oral administration of 200 mmol/L NaHCO3ad libitum, (c) daily i.p. injection of 100 mg/kg phenformin, and (d) both NaHCO3 and phenformin. Treatments started from the day of inoculation and were continued for the duration of the experiment. Tumor diameters were measured at the indicated time points. Tumor volumes were estimated by V = (a2 × b)/2, where “a” is the short axis and “b” is the long axis of the tumor.
Metastasis assay
Mice were injected intravenously with 5 × 104 B16F10 cells in 0.1 mL PBS via the tail vein and were treated daily with phenformin (100 mg/kg, i.p.) and/or NaHCO3 (200 mmol/L, oral administration ad libitum) from the day of injection, which continued for the duration of the experiment (n = 5 per group). After 21 days, mice were sacrificed and their lungs resected and photographed. The numbers of metastases on the surfaces of the lungs were counted macroscopically.
Statistical analysis
Data are expressed as mean ± SE. To compare two mean values, an unpaired Welch t test was used. For comparisons of multiple mean values, a Tukey–Kramer test following a single-factor ANOVA was used. For subcutaneous tumor growth analysis, an unpaired Welch t test was used. For metastasis analysis, a nonparametric Mann–Whitney U test was used. Differences of *P < 0.05 were considered statistically significant.
Results
Extracellular pH modulates the cytotoxicity of OXPHOSi on melanoma cells in vitro
To elucidate whether cytotoxicity of mitochondrial inhibitors on melanoma cells is affected by pHe in vitro, we examined human melanoma cells at pH 7.4 or 6.7 treated with oligomycin (0.01 μg/mL), an inhibitor of mitochondrial ATP synthase, or phenformin (1 mmol/L), an antidiabetic drug from the biguanide class known to inhibit mitochondrial complex I. Increased numbers of apoptotic cells were detected when A375, SK-MEL-28, or MEWO cells were cultured with oligomycin or phenformin at an acidic pHe, whereas there were only modest numbers of apoptotic cells when cells were cultured without the drugs at an acidic pHe or with the drugs at pHe 7.4 (Fig. 1A to D; Supplementary Fig. S1A to S1C). A similar enhancement of apoptosis was observed in cells treated with the mitochondrial complex II inhibitor TTFA (0.1 mmol/L) at an acidic pHe (Supplementary Fig. S1D and S1E). An increase in apoptosis was also seen in A375 cells cultured with oligomycin at pH 6.2 rather than at pH 6.7, and even in A375 cells cultured at pH 6.2 without the drug (Supplementary Fig. S1F and S1G). The number of apoptotic A375 cells cultured at pH 8.0 was comparable with that at pH 7.4 (Supplementary Fig. S1F and S1G). These data indicate that extracellular acidity enhances the cytotoxic effect of OXPHOSi on melanoma cells in vitro. Interestingly, patient-derived normal human dermal fibroblasts were also sensitive to phenformin (1 mmol/L) at an acidic pHe (Supplementary Fig. S1J–S1L), indicating that this phenomenon occurs in normal as well as malignant cells.
pHe is a threshold that modulates the cytotoxicity of OXPHOSi on melanoma cells in vitro. A, Representative microscopic images of A375 and of MEWO cells treated with/without oligomycin (0.01 μg/mL) or phenformin (1 mmol/L) at pHs of 7.4 and 6.7 for 48 (A375) or 72 hours (MEWO; magnification, ×100). Bar, 100 μm. B–D, Apoptosis (late apoptotic + early apoptotic) detected in A375 and in MEWO cells after treatment with/without oligomycin (0.01 μg/mL) or phenformin (1 mmol/L) for 48 (A375) or 72 hours (MEWO; n = 3). E, Representative microscopic images of A375 cells treated with/without cisplatin (20 μmol/L) for 48 hours at pHs of 7.4 and 6.7 (magnification, ×100). Bar, 100 μm. F and G, Apoptosis (late apoptotic + early apoptotic) detected in A375 cells after treatment with/without cisplatin (20 μmol/L) at pHs of 7.4 and 6.7 for 72 hours (n = 3). Images are representative of three independent experiments. Bars, mean ± SE; **, P < 0.01; N.S., not significant.
Because cancer cells in an acidic microenvironment generally acquire multidrug resistance (15, 28), we tested the cytotoxicity of cisplatin (20 μmol/L) on melanoma cells at varying pHe. In contrast with the effects of OXPHOSi, A375 and SK-MEL-28 cells displayed less apoptosis following cisplatin exposure at pH 6.7 compared with pH 7.4, although comparable levels of apoptosis were detected across pHs after treatment of MEWO cells with cisplatin (Fig. 1E to G; Supplementary Fig. S1H and S1I). Because a low pHe augments the cytotoxicity of weak acids and as cisplatin is a weak acid (35), the cytotoxicity enhancement of OXPHOSi by acidity must not merely be because cells are damaged by acidity. Rather, OXPHOS inhibition by these drugs may specifically cooperate with an acidic pHe to augment cytotoxicity.
Systemic buffering attenuates the anticancer effect of phenformin on B16F10 cells in vivo
Given that an acidic pHe enhanced cytotoxicity of OXPHOSi (including phenformin) against melanoma cells in vitro (Fig. 1), we hypothesized that systemic alkali buffering with NaHCO3 would reduce antitumor effects of phenformin in vivo, which have been shown for cancers such as melanoma (8, 36). Because it was reported that extracellular acidity promotes immune evasion of cancer cells (26) and that host immune responses critically affect tumor development (37), we utilized B16F10 cells and a syngeneic allograft model using C57BL/6 mice to test this.
In preparatory in vitro work, phenformin decreased the viability of B16F10 cells at pHe 7.4 as well as pHe 6.7 (Supplementary Fig. S2A and S2B). However, far more apoptosis was induced in B16F10 cells at the acidic pHe compared with the normal pHe (Fig. 2A and B). Consistent with previous studies (8, 36), in vivo tumor growth was significantly inhibited in mice treated with phenformin (148 mm3 ± 28 SE) compared with untreated mice (1,504 mm3 ± 106 SE; Fig. 2C and D). As hypothesized, a significant increase in tumor growth was observed in phenformin + NaHCO3-treated mice compared with phenformin only–treated mice (303 mm3 ± 39 SE on day 16 after injection; Fig. 2C and D). Correspondingly, immunohistochemical analysis suggested that apoptotic cells were decreased while proliferating cells were increased in tumors from phenformin + NaHCO3-treated mice (Fig. 2E and F). Interestingly, tumor growth was also inhibited in mice treated with oral bicarbonate (452 mm3 ± 81 SE on day 16 after injection) compared with mice without treatment (Fig. 2C and D), regardless of phenformin administration.
Systemic buffering attenuates the anticancer effect of phenformin on B16F10 cells in vivo. A and B, Apoptosis (late apoptotic + early apoptotic) detected in B16F10 cells after treatment with/without phenformin (1 mmol/L) at pH 7.4 or 6.7 for 48 hours (n = 3). C, Representative images of tumors in each group of mice 15 days after the subcutaneous injection of B16F10 cells; dashed lines indicate areas of tumors. D, Tumor volumes after the subcutaneous injection of B16F10 cells in mice with the indicated treatments (n = 6). Bars, mean ± SE; *, P < 0.05; **, P < 0.01. E and F, TUNEL and Ki67 staining of tumors 16 days after treatment with phenformin with/without NaHCO3 (magnification, ×200). Bars, 100 μm. Arrows indicate Ki67-positive nucleuses. Images are representative of three independent experiments. G, Representative photographs of lungs resected from mice treated with/without phenformin and/or NaHCO3 for 21 days after the intravenous injection of B16F10 cells. H, Numbers of metastases on the surfaces of lungs (n = 5). Bars, mean ± SE; *, P < 0.05; **, P < 0.01; N.S., not significant.
We also evaluated experimental metastasis. Twenty-one days after injection, the number of lung at metastases was reduced in mice treated with daily phenformin, compared with mice with no treatment (Fig. 2G and H). Noticeably, metastasis was increased in mice with daily phenformin + NaHCO3 treatment compared with phenformin only–treated mice (Fig. 2G and H). Collectively, these data are consistent with the possibility that OXPHOS inhibition preferentially targets melanoma cells in the acidic in vivo tumor microenvironment that otherwise facilitates malignant progression.
OXPHOS inhibition causes the activation of caspase-dependent apoptosis at an acidic pHe
We next interrogated mechanisms of action of OXPHOSi targeting in an acidic microenvironment. Cleavage of caspase-3 as well as caspase-6 was increased in A375 cells after treatment with oligomycin at an acidic pHe (Fig. 3A and B). Although the substrate-specific caspase inhibitor Z-DEVD-FMK failed to suppress apoptosis in A375 cells after treatment with oligomycin at pH 6.7 (Supplementary Fig. S3A), the pan-caspase inhibitor Z-VAD-FMK suppressed apoptosis in these cells (Fig. 3C and D). In contrast, treatment with the apoptosis-inducing factor (AIF)–specific inhibitor ATA (38) and E64 and antipain both of which inhibit cathepsin-B (39) failed to suppress apoptosis in A375 cells treated with oligomycin at pH 6.7 (Supplementary Fig. S3B and S3C). These data indicate caspase-dependent apoptosis in the cytotoxicity of OXPHOSi elicited by acidity. Inhibition of both caspase-3 and caspase-6 by Z-VAD-FMK, but not caspase-3–specific inhibition by Z-DEVD-FMK, may be necessary to suppress apoptosis in A375 cells treated with oligomycin at an acidic pHe.
OXPHOS inhibition causes the activation of caspase-dependent apoptosis at an acidic pHe. A, Western blot analysis of cleaved caspase-3 p11 in A375 cells after treatment with/without oligomycin (0.01 μg/mL) and/or Z-DEVD-FMK (10 μmol/L) at pH 7.4 (N) or 6.7 (A) for 48 hours; 9 μg total protein extract was loaded in each lane. B, Western blot analysis of cleaved caspase-6 in A375 cells after treatment with/without oligomycin (0.01 μg/mL) and/or Z-VAD-FMK (100 μmol/L) at pH 7.4 (N) or 6.7 (A) for 48 hours; 9 μg total protein extract was loaded in each lane. C and D, Apoptosis (late apoptotic + early apoptotic) detected in A375 cells after treatment with oligomycin (0.01 μg/mL), phenformin (1 mmol/L), and/or Z-VAD-FMK (100 μmol/L) at pH 6.7 for 48 hours (n = 3). Images are representative of three independent experiments. Bars, mean ± SE; *, P < 0.05; N.S., not significant.
Intracellular Ca2+ level is increased in melanoma cells after treatment with OXPHOSi at an acidic pHe
Because oxidative stress increases the concentration of intracellular Ca2+ (40), leading to intracellular Ca2+ overload and apoptosis (41), we next tested if intracellular Ca2+ is altered in melanoma cells after treatment with OXPHOSi at an acidic pHe. When A375 cells were treated with the OXPHOSi at pH 6.7, large fractions of surviving cells showed high cytoplasmic Ca2+. This was not seen following OXPHOSi at pH 7.4 or at pH 6.7 without OXPHOSi (Fig. 4A and B). Similar results were obtained with SK-MEL-28 cells and MEWO cells (Supplementary Fig. S4A to S4D). The viable Ca2+-high cell population induced by OXPHOSi at pH 6.7 was impaired, as A375 cells sorted from the propidium iodide–excluding Ca2+-high gate showed markedly impaired in vitro proliferation (Fig. 4C to E). The importance of intracellular Ca2+ in this context was further revealed by treatment with BAPTA/AM, an intracellular calcium chelator, which decreased Ca2+-high cells as well as death in A375 cells treated with oligomycin at pH 6.7 (Fig. 4F to I). These results indicate that intracellular Ca2+ overload contributes substantially to the enhanced cytotoxicity of OXPHOSi in an acidic pH.
Intracellular Ca2+ level is increased in melanoma cells after treatment with OXPHOSi at an acidic pHe, leading to cell death. A, A375 cells were treated with indicated drugs at pH 7.4 or pH 6.7 for 16 hours and stained by Fluo3/AM. Oligomycin: 0.01 μg/mL, Phenformin: 1 mmol/L, TTFA: 0.1 mmol/L. Images are representative of three independent experiments. B, The ratio of A375 cells that show high (Ca-h) and low-intermediate (Ca-m) cytoplasmic Ca2+ levels after treatment with the OXPHOSi at pH 7.4 or pH 6.7. (n = 3). C and D, A375 cells were cultured with oligomycin (0.01 μg/mL) at pH 6.7 for 16 hours. Ca-m/Ca-h cells were sorted and cultured at pH 7.4 (2,000/well). E, Sorted cells were seeded in 24-well plates (2,000/well) and cultured at pH 7.4. The number of viable cells was counted and normalized to that of day 1 (n = 3). Images are representative of three independent experiments. F, A375 cells were treated with BAPTA/AM (1 μmol/L) 1 hour before treatment with oligomycin (0.01 μg/mL) for 24 hours at pH 6.7 and stained by Fluo3/AM. G, The ratio of dead cells in the cells treated with oligomycin and oligomycin + BAPTA/AM (n = 12). H and I, The ratio of Ca-h/Ca-m cells in the cells treated with oligomycin or oligomycin + BAPTA/AM (n = 12). Images are representative of 12 independent experiments. Bars, mean ± SE; *, P < 0.05; **, P < 0.01.
ROS scavenging synergizes with OXPHOSi against melanoma cells in an acidic pHe
Cytotoxicity of OXPHOSi against cancer cells may be due to deleterious effects of ROS production (12). Because extracellular acidosis enhances ROS in cancer cells (42, 43), we next studied whether increased ROS production in melanoma cells at acidic pHe modulates the cytotoxic effects of OXPHOSi. As expected, ROS production in A375 and MEWO cells was significantly increased after OXPHOS inhibition at pH 6.7 compared with pH 7.4, and Trolox and Sodium pyruvate, both ROS scavengers, suppressed ROS production at pH 6.7 (Fig. 5A; Supplementary Fig. S5A). However, when ROS production was suppressed after treatment with OXPHOSi at pH 6.7, more rather than less apoptosis was unexpectedly seen in both A375 (Fig. 5B to E) and MEWO cells (Supplementary Fig. S5B to S5E). As a moderate increase of ROS correlates with aggressive cancer cell phenotypes (44), these data raise the possibility that OXPHOSi in an acidic pH increases ROS to a threshold that provides a level of adaptive cytoprotection that may be exploited therapeutically by concurrent ROS scavengers (44).
ROS scavenging synergizes with the cytotoxicity enhancement of OXPHOSi elicited by acidity. A, Measurement of ROS production in A375 cells after treatment with/without oligomycin (0.01 μg/mL) and phenformin (1 mmol/L) with/without Trolox (100 μmol/L) or sodium pyruvate (10 mmol/L) for 30 minutes at pHs of 7.4 and 6.7 (n = 3). The MFI of cells in each condition was normalized to cells with no treatment at pH 7.4. B–E, Apoptosis (late apoptotic + early apoptotic) detected in A375 cells after treatment with/without oligomycin (0.01 μg/mL), phenformin (1 mmol/L), TTFA (0.1 mmol/L), with/without Trolox (100 μmol/L; B and C) or sodium pyruvate (10 mmol/L; D and E) at pH 6.7 for 48 hours (n = 3). Trolox and sodium pyruvate were pretreated 1 hour before OXPHOS inhibitors. Images are representative of three independent experiments. Bars, mean ± SE; *, P < 0.05; **, P < 0.01; N.S., not significant.
Enhanced cytotoxicity in melanoma cells of OXPHOSi in acid pHe is independent of PGC1alpha
PGC1alpha is upregulated in some melanoma cells, resulting in mitochondrial addiction, tolerance to oxidative stress (7), and vulnerability against OXPHOSi (6). As expected (7), we found that MEWO cells robustly expressed PGC1alpha (Fig. 6A). Interestingly, expression of PGC1alpha and ATP synthase were increased by acidification in MEWO cells, indicating the activation of OXPHOS (Fig. 6A). Consistent with this, mitochondrial membrane potential was activated (Fig. 6B), and oxygen consumption rate was increased (Fig. 6D) in acid pHe, suggesting that acidification potentiates mitochondrial activity in MEWO cells. However, although SK-MEL-28 and A375 cells were also sensitive to OXPHOSi at acidic pHe, they expressed very low levels of PGC1alpha (Fig. 6A) and had much lower mitochondrial membrane potentials and oxygen consumption rates after acidification (Fig. 6B and C). This argues against a consistent PGC1alpha dependency of the cytotoxicity elicited by OXPHOSi in acidic pHe (6, 7).
The cytotoxicity enhancement of OXPHOSi elicited by acidity is distinct from PGC1alpha-driven mitochondrial addiction and JARID1B-high slow-cycling cells. A, Western blot analysis of PGC1alpha and ATP synthase after acidification of melanoma cell lines; 9 μg total protein extract was loaded in each lane. B, MitoRed staining of melanoma cell lines cultured at pH 7.4 or 6.7 for 72 hours (magnification, ×200). Bar, 100 μm. C and D, Basal and maximal oxygen consumption rates in SK-MEL-28 or MEWO cells after acidification for 72 hours (n = 12). E, Viability of SK-MEL-28 and SK-MEL-28-R cells after treatment with vemurafenib (PLX4032, 10 μmol/L) at the indicated concentration for 72 hours. The value of vemurafenib-treated cells was normalized to cells with no treatment (n = 6). F and G, Apoptosis (late apoptotic + early apoptotic) in SK-MEL-28-R cells after treatment with/without oligomycin (0.01 μg/mL) at pH 7.4 or 6.7 for 72 hours (n = 3). H, FACS analysis of JARID1B-positive cells after acidification of SK-MEL-28 and SK-MEL-28-R cells for 72 hours. The ratio of JARID1B-positive cells is shown. I, FACS analysis for JARID1B-positive cells after treatment of SK-MEL-28-R and A375-R cells with/without oligomycin (0.01 μg/mL) for 48 hours at pH 6.7. The ratio of JARID1B-positive cells is shown. Images are representative of three independent experiments. Bars, mean ± SE; *, P < 0.05; **, P < 0.01.
Melanoma cells at an acidic pHe acquire senescent phenotypes that do not mediate the cytotoxicity of OXPHOSi
To examine further potential mechanisms of OXPHOSi cytotoxicity in acidic pHe and as cells undergoing senescence have increased sensitivity to OXPHOSi (12, 13), we tested whether melanoma cells that survive an acidic milieu are senescent (12). Some cell lines showed some but not all features of senescence in acidic pHe. Consistent with previous data (17), A375 cells were slow-cycling and had increased G1 phase cells (Supplementary Fig. S6A to S6C). SK-MEL-28 cells, but not A375 or MEWO cells, showed increased expression of senescence-associated β-galactosidase (Supplementary Fig. S6G; ref. 45). In metabolism studies, both lactate and ATP production were increased by acidification in A375, SK-MEL-28, and MEWO cells (Supplementary Fig. S6H and S6I), consistent with a hyperactive hybrid metabolism, characterized by increased glycolysis as well as mitochondrial respiration, that is linked to the senescent state (12, 13). However, A375 and SK-MEL-28 cells did not display increased mitochondrial membrane potential after acidification (Fig. 6B and C), which would be expected of senescent cells (12). As P53 and P21 upregulation characterize the senescent state, we also tested expression of these proteins in acidic pHe in A375, SK-MEL-28, and MEWO cells, with contrasting results. Although P53 and P21 were increased after acidification in A375 cells (which harbor wild-type p53; ref. 46; Supplementary Fig. S6D), neither protein was increased in MEWO cells (which carry p53 mutations; ref. 47; Supplementary Fig. S6F). In SK-MEL-28 cells, which are also p53 mutant (48), P53 but not P21 was increased after acidification (Supplementary Fig. S6E). Thus, P53 and P21 expression was not consistently linked to sensitivity to OXPHOSi. Altogether, these data indicate that although some melanoma cells at an acidic pHe show some features of senescence, this phenotype is not unequivocally and universally linked to the cytotoxicity conferred by OXPHOSi in an acidic pHe.
The JARID1B-high population is not increased in melanoma cells by an acidic pHe
Treatment with vemurafenib has been shown to enrich JARID1B-high slow-cycling melanoma cells, which are sensitive to OXPHOSi (5). Because melanoma cells exhibited enhanced sensitivity to OXPHOSi after acidification, we also tested whether JARID1B-high cells are increased by acidification. SK-MEL-28-R and A375-R cells, resistant to vemurafenib, were generated from SK-MEL-28 and A375 cells, which harbor a homozygous BRAF (V600E) mutation (Fig. 6E; Supplementary Fig. S6J). Apoptotic SK-MEL-28-R and in A375-R cells were significantly increased by treatment with oligomycin at pH 6.7, compared with oligomycin at pH 7.4 or untreated cells at pH 6.7 (Fig. 6F and G; Supplementary Fig. S6K and S6L). However, this was not associated with an increase in JARID1B-high cells at pH 6.7 (Fig. 6H), although JARID1B-high SK-MEL-28-R cells were decreased after treatment with oligomycin or rotenone at pH 6.7 (Fig. 6I). JARID1B-high A375-R cells were also decreased after treatment with oligomycin or rotenone at acidic pHe (Supplementary Fig. S6I and S6M). Thus, the augmented sensitivity of melanoma cells to OXPHOSi at an acidic pHe cannot be attributed to increased JARID1B-high cells in an acidic milieu. Despite this, the vulnerability of JARID1B-high cells to OXPHOSi is preserved after acidification.
Discussion
We demonstrate that the cytotoxicity of OXPHOSi against melanoma is strongly mediated by pHe. Because not only melanoma cells but also normal human dermal fibroblasts are relatively protected against phenformin by a neutral pH (Supplementary Fig. S1J to S1L), the neutral pHe in normal tissues (15, 16) should ameliorate potential side-effects in patients of OXPHOSi, while preserving cytotoxic effects that will be more specific for acidic tumor microenvironments. On the other hand, when tissue physiologic oxygenation is disturbed in patients with pathologic conditions like severe diabetes, coronary heart disease, or stroke (49), acidosis can occur in involved organs and tissues. It is thus possible that OXPHOSi could exert untoward toxicities on nonmalignant tissues that are unusually acidotic as a result of nonmalignant pathologic conditions.
As an acidic pHe promotes aggressive phenotypes in cancers (15), manipulation of intratumoral pHe has been considered as a potential cancer therapy (32, 50, 51). Systemic buffering by oral administration of NaHCO3 increased intratumoral pHe and inhibited local invasion and tumor growth in breast and in colon cancer cells (20). Moreover, treatment with oral bicarbonate increased the average tumor pHe from 7.0 to 7.4 and inhibited the spontaneous metastasis of metastatic breast and prostate cancer cells in immunodeficient mice (31). Our data show that the acidic microenvironment that promotes tumor progression also renders cancer cells susceptible to therapies that target mitochondrial metabolism, and that systemic alkalinization ameliorates this susceptibility.
It is notable that the vulnerability of melanoma cells to OXPHOSi in acidic pHe was independent of P53 mutation status, BRAF (V600E) mutation status, and resistance against vemurafenib. This indicates potential application of OXPHOSi across a range of clinical contexts, such as a second-line treatment against drug-resistant melanomas or combined with first-line therapies (12), to reduce primary resistance. Regardless, identifying biomarkers of acidic pHe, perhaps by tumor sampling (52, 53), will be important to identify subgroups of patients likely to respond to OXPHOSi, as recent clinical trials in unselected patients failed to show improved survival in cancer patients treated with OXPHOSi (54, 55). Clearly, concurrent use of alkalizer therapy and OXPHOSi should not be applied to melanoma patients.
In examining mechanisms of action of the increased susceptibility of melanoma cells to OXPHOSi in acidic pHe, we unexpectedly found that this vulnerability is not consistently dependent on PGC1alpha (6, 7), induction of senescence (12), or JARID1B expression (5, 8). Rather, we found that caspase- and pHe-dependent apoptotic effects of OXPHOSi were closely linked to increased intracellular calcium levels. The role of calcium in regulating cell metabolism in some pathophysiologic states is well understood. For example, myocardial cells undergo intracellular Ca2+ overload upon myocardial ischemia-reperfusion injury (56). In this context, a decrease in extracellular as well as intracellular pH during ischemia leads to Na+/H+ exchange and Na+ overload in myocardial cells. Subsequently, Ca2+ uptake occurs through reverse-mode Na+/Ca2+ exchange, resulting in intracellular Ca2+ overload. Concurrently impaired ATP production in mitochondria inhibits Ca2+ extrusion by plasmalemmal Ca2+-ATPase, compounding the intracellular Ca2+ increase, which results in mitochondrial-driven myocardial cell apoptosis (40, 41, 56, 57).
We speculate that the acidification of melanoma cells mimics this by inducing compensatory ion fluxes that increase intracellular calcium to synergize with reduced ATP production conferred by OXPHOSi. On the other hand, in A375 and SK-MEL-28 cells, mitochondrial activities are already decreased at acidic pHe without OXPHOSi (Fig. 6B and C). Therefore, it is possible that inhibition of OXPHOS at acidic pH might induce intracellular Ca2+ increase and activation of cell death signaling pathway in a manner independent from a decrease in ATP production (58, 59), although further studies are needed to address this mechanism. Our demonstration of intracellular Ca2+ overload and activation of apoptosis in acidic melanoma cells after treatment with OXPHOSi (Figs. 3 and 4; Supplementary Figs. S3 and Fig. S4) suggests that the combination of OXPHOSi and drugs that increase cytoplasmic Ca2+ level via manipulating Ca2+ channels, pumps, or storage (57) might have synergistic cytotoxicity against melanoma cells. This highlights the importance of understanding complex interactions between cell metabolism, tumoral pHe, and ionic homeostasis in developing novel therapy combinations that target specific susceptibilities in cancer cells.
Disclosure of Potential Conflicts of Interest
No potential conflicts of interest were disclosed.
Authors' Contributions
Conception and design: F. Noguchi, S. Inui, M. Shackleton, S. Itami
Development of methodology: F. Noguchi, S. Inui
Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): F. Noguchi, M. Shackleton
Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): F. Noguchi, S. Inui, C. Fedele, M. Shackleton, S. Itami
Writing, review, and/or revision of the manuscript: F. Noguchi, S. Inui, C. Fedele, M. Shackleton, S. Itami
Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): F. Noguchi, S. Itami
Study supervision: F. Noguchi, M. Shackleton, S. Itami
Grant Support
F. Noguchi was supported by Uehara Memorial Foundation. M. Shackleton was supported by Pfizer Australia, the Victorian Cancer Agency, the Australian National Health and Medical Research Council (NHMRC), and veski.
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
Acknowledgments
We appreciate Professor Seiji Takashima and Dr. Yasunori Shintani, Department of Medical Biochemistry, Osaka University Graduate School of Medicine, for provision of convenience in measuring oxygen consumption rates. We also greatly appreciate Ayako Sato, Yuko Yoshikawa, and Miyuki Nakamura for their technical assistance.
Footnotes
Note: Supplementary data for this article are available at Molecular Cancer Therapeutics Online (http://mct.aacrjournals.org/).
- Received March 24, 2015.
- Revision received April 30, 2015.
- Accepted February 8, 2017.
- ©2017 American Association for Cancer Research.