PK-M2-mediated metabolic changes in breast cancer cells induced by ionizing radiation

Abstract

Purpose

Radiotherapy (RT) constitutes an important part of breast cancer treatment. However, triple negative breast cancers (TNBC) exhibit remarkable resistance to most therapies, including RT. Developing new ways to radiosensitize TNBC cells could result in improved patient outcomes. The M2 isoform of pyruvate kinase (PK-M2) is believed to be responsible for the re-wiring of cancer cell metabolism after oxidative stress. The aim of the study was to determine the effect of ionizing radiation (IR) on PK-M2-mediated metabolic changes in TNBC cells, and their survival. In addition, we determine the effect of PK-M2 activators on breast cancer stem cells, a radioresistant subpopulation of breast cancer stem cells.

Methods

Glucose uptake, lactate production and glutamine consumption were assessed. The cellular localization of PK-M2 was evaluated by western blot and confocal microscopy. The small molecule activator of PK-M2, TEPP46, was used to promote its pyruvate kinase function. Finally, effects on cancer stem cell were evaluated via sphere forming capacity.

Results

Exposure of TNBC cells to IR increased their glucose uptake and lactate production. As expected, PK-M2 expression levels also increased, especially in the nucleus, although overall pyruvate kinase activity was decreased. PK-M2 nuclear localization was shown to be associated with breast cancer stem cells, and activation of PK-M2 by TEPP46 depleted this population.

Conclusions

Radiotherapy can induce metabolic changes in TNBC cells, and these changes seem to be mediated, at least in part by PK-M2. Importantly, our results show that activators of PKM2 can deplete breast cancer stem cells in vitro. This study supports the idea of combining PK-M2 activators with radiation to enhance the effect of radiotherapy in resistant cancers, such as TNBC.

1. Introduction

Radiation therapy (RT) is often part of a successful treatment regimen for breast cancer (BC) [1]. However, triple negative breast cancer (TNBC) has a poor prognosis and resists most treatment approaches, including RT [2]. Therapeutic approaches that radiosensitize this subtype of breast cancer could substantially improve the outcome. One reason for resistance to RT [3, 4] is the presence of cells with “stem cell” characteristics (cancer stem cells, CSCs) that escape cytotoxic therapy and cause relapses [5–8].

It is well recognized that ionizing radiation (IR), such as used in RT, exerts its cytotoxic effect mainly through the generation of highly reactive oxygen species (ROS) that damage macromolecules and critical cellular structures [9]. Often cancer cells adapt to oxidative stress as a way of surviving in a harsh tumor microenvironment [10], and metabolic plasticity can facilitate such adaptation. For example, yeast and normal mammalian cells are capable of re-wiring their glucose metabolism in response to oxidative stress and this process involves redox-sensitive enzymes in lower glycolysis, such as glyceraldehyde 3-phosphate dehydrogenase (GAPDH) [11–13]. Cancer cells however, generally use the last rate-limiting enzyme in glycolysis, the M2 isoform of pyruvate kinase (PK-M2) to reprogram their metabolism during oxidative stress [14].

PK-M2 is overexpressed in a number of cancers, including BC [15–18], where it facilitates the “Warburg effect” by increasing basal levels of glycolysis [17, 19, 20] and its expression correlates with patient survival [18]. PK-M2 is unique from all the other isoforms of pyruvate kinase due to its conformational flexibility. PK-M2 tetramers efficiently convert phosphoenol pyruvate to pyruvate in glycolysis, while PK-M2 dimers are enzymatically inactive - this creates a “bottleneck” in glycolytic flux resulting in rapid accumulation of intermediate metabolites that become available for the upstream pentose phosphate pathway (PPP). Growth factors can inactivate PK-M2 resulting in metabolic reprogramming to meet the anabolic demands of rapidly proliferating cancer cells [21, 22]. However, oxidative stress also inactivates PK-M2, re-directing glucose metabolites into the PPP resulting in increased production of the antioxidant NADPH [14] and restoration of the redox state of the cell. The conformational flexibility of PKM2 therefore, endows cancer cells with metabolic flexibility to meet context-specific metabolic needs [17]. In addition, PK-M2 dimers are often found in the nucleus moonlighting as transcriptional regulators of a number of genes ranging from stem cell pathways to DNA damage repair [23–26]. Importantly, small molecule activators of PKM2 boost oxidative stress and have anti-proliferative activity in cancer cells [14, 27, 28].

Most studies investigating the effect of oxidative stress on cellular metabolism have utilized either chemically induced oxidative stress (i.e. via H₂O₂ or diamide) or hypoxia [11, 13, 14, 29, 30]. There are multiple reasons why the cellular responses to chemically induced oxidative stress differs from the response to IR and are not directly relevant to RT. Therefore, in the study presented here we have investigated the effect of IR on the metabolism of TNBC, and more specifically on PK-M2 in unselected cells, as well as in radiation-resistant BC stem cells which are metabolically distinct from their more differentiated progeny [31, 32].

2. Materials and Methods

2.1. Cell culture

Human SUM159PT and MDA-MB-231 breast cancer cell lines were purchased from Asterandt (Detroit, MI) and ATCC respectively. SUM159PT cells were cultured in F12 medium (Thermofisher) supplemented with 5% fetal bovine serum (FBS), penicillin and streptomycin cocktail (Thermofisher), 5 μg/mL insulin (humulin R, Lilly), 10 mM HEPES (Thermofisher) and 1 μg/mL hydrocortisone (Solu-Cortef, Pfizer). SUM159PT-cODC-ZsGreen cells were obtained as previously described [33]. MDA-MB-231 cells were cultured in Dulbecco’s Modified Eagle Medium (DMEM) (Thermofisher) supplemented with 20% FBS, and penicillin and streptomycin cocktail. All cells were grown at 37°C, in a humidified incubator with 5% CO₂.

2.2. Irradiation and treatments

Cells were irradiated at room temperature using an experimental X-ray irradiator (Gulmay Medical Inc, Suwanee, GA) at a dose rate of 7.1702 Gy/min. The X-ray beam was operated at 300 kV and hardened using 4-mm Be, 3-mm Al, and 1.5-mm Cu filters. Corresponding controls were sham irradiated. Dosimetry was standardized to NIST using ion chambers and film. Cells were treated with 10 μM TEPP46 (CCKinase, Inc.) [27] one hour prior to irradiation.

2.3. Mammosphere assay

Mammosphere-forming capacity was performed by plating cells in DMEM/F-12, 0.4% bovine serum albumin (BSA) (VWR), 10 mL/500 mL B27 (Thermofisher) 5 μg/mL insulin, 4 μg/mL heparin (Sagent, Schaumburg, IL), 20 ng/mL fibroblast growth factor 2 (bFGF) (StemCell Technologies), and 20 ng/mL epidermal growth factor (EGF) (StemCell Technologies) into 96-well ultra-low adhesion plates, at a range of cell densities, with EGF, bFGF, and heparin added every 3 days. The number of spheres formed per well was counted and expressed as a percentage of the initial number of cells plated.

2.4. Clonogenic colony formation assays

SUM159PT and MDA-MB-231 cells were plated as monolayers in serum supplemented media and treated with vehicle or TEPP46 (10 μM) 24 hours after plating. A second treatment was administered to the cells 24 hours after the first treatment. Three hours later, cells were detached and counted. Cell suspensions were prepared at the appropriate cell concentration to account for cell toxicity in each irradiation dose (200 cells per well for 0 and 2 Gy for both cell lines; 1600 cells per well for 8 Gy for SUM159PT and 9600 cells for 8 Gy for MDA-MB-231). Cells were irradiated in conical 15 mL tubes right before plating in 6-well tissue culture-treated plates. Triplicate technical repeats were performed for each condition. Cells were incubated at 37°C with 5% CO₂ in a humidified atmosphere. After 10 to 15 days, cells were fixed with 70% ethanol and stained with crystal violet. Colonies with at least 50 cells were counted. The survival fractions were determined by normalizing each irradiated condition to its non-irradiated control.

2.5. Glucose uptake, glutamine consumption and lactate production assays

2×10⁵ cells/well were plated in 6-well plates, allowed to adhere overnight and irradiated the next day. For measuring glucose uptake, the fluorescent glucose analogue 2-deoxy-2-[(7-nitro-2,1,3-benzoxadiazol-4-yl)amino]-D-glucose (2NBDG) was used (Thermofisher). 1, 3, 24, 48 and 72 hours after irradiation exposure, cells were removed by trypsinization. Cells were then incubated for 1 hour with 200 μM 2NBDG in glucose-free media at 37°C. After incubation with 2-NBDG, cells were re-suspended in PBS, and analyzed by flow cytometry and the geometrical mean fluorescence was determined. Cells were also sorted based on 2NBDG uptake. 2NBDG^low and 2NBDG^high populations were collected and plated either for immunofluorescence or mammosphere forming assay.

For the [¹⁸F]fluorodeoxyglucose uptake assay, breast cancer cells were plated at 5 × 10⁴ cells/well on tissue-culture treated 6-well plates and allowed to adhere overnight. The next day cells were exposed to 0 or 8 Gy. After the indicated time points, cells were collected, counted and cell numbers were adjusted so the same number of cells for each treatment group was resuspended in glucose-free DMEM/F12 media containing [¹⁸F]FDG (radioactivity 1 μCi/mL). Cells were incubated at 37°C for 1 hour and then washed 3X with ice-cold PBS. The radioactivity of each sample was then measured with a gamma counter and the resulting counts-per-minute (CPM) values were normalized to the appropriate controls (non-irradiated samples).

For determining glutamine consumption and lactate production, media from wells with cells or empty controls was collected at the indicated time points after irradiation, centrifuged at 10,000g for 10 min at 4°C to remove insoluble components. The media was used to determine glutamine consumption (Glutamine Assay Kit, Abcam, #ab197011) and the amount of lactate produced (Lactate Assay Kit, Bio Vision, #K607–100). The results were normalized to the cell numbers of each group.

2.6. Protein extraction

Whole cell proteins were extracted from cells at various time points after treatments. Briefly, cells were scraped, washed with PBS, and incubated for 30 minutes under agitation at 4°C in RIPA buffer (10 mM Tris, 1 mM EDTA, 1% Triton X-100, 0.1% sodium deoxycholate, 0.1% SDS, 140 mM NaCl, 1 mM PMSF). Cell lysates were then centrifuged 20 min at 12,000 rpm at 4°C and supernatants were collected. Protein concentrations were determined by Pierce BCA protein assay. Laemmli buffer (Biorad) and 2-mercaptoethanol (Sigma) (350 mM) were added to the protein samples. Finally, proteins were denatured by heating the samples at 100°C for 5 min.

Nuclear fractions were obtained by scraping the cells and centrifuging the cells at 5000 rpm for 5 min. The cell pellet was then resuspended in cytoplasmic lysis buffer (10 mM HEPES, 0.1 mM EDTA, 10 mM KCl, NP40, 0.5 mM PMSF, 1 mM DTT, Protease inhibitor cocktail) and incubated on ice for 20 min with regular mixing, and then centrifuged at 12,000 rpm at 4°C for 10 min. The supernatant was collected which consisted of the cytoplasmic extract. The nuclei pellet was washed with the cytoplasmic lysis buffer twice and resuspended in nuclear lysis buffer (20 mM HEPES, 125 mM NaCl, 1 mM EDTA, 1 mM PMSF, 1 mM DTT, Protease inhibitor cocktail). Lysates were sonicated for 5 seconds, placed on ice for 1 min and sonicated again for 5 seconds. Samples were incubated on ice for 30 min, centrifuged at 12,000g for 15 min at 4°C. The supernatant was collected which consisted of the nuclear extract. As above, protein concentrations were determined by Pierce BCA protein assay.

2.7. Western Blotting

The extracted proteins were separated on 10% SDS-PAGE gels and transferred to PVDF membrane (Thermofisher). Blots were incubated overnight with the primary antibody against PK-M2 (#21578, 1:1000, Signalway Antibody), Lamin A/C (#ab8984, 1:2000, Abcam) and GAPDH (#AM4300, 1:5000, Ambion). The membranes were then incubated with secondary antibodies conjugated to horseradish peroxidase (HRP) (Santa Cruz Biotechnology) for 1 hour at room temperature. The protein band detection was performed with ECL (solution A: 2.5 mM luminol, 0.4 mM p-coumaric acid, 0.1 M Tris-HCl; solution B: 0.02% hydrogen peroxide, 0.1 M Tris-HCl).

2.8. PK-M2 flow cytometry

For PK-M2 flow cytometry study, cells were first fixed for 15 min at room temperature with 4% paraformaldehyde. Cells were then permeabilized with ice cold methanol, for 10 min on ice. After several washes with PBS, fixed cells were incubated with primary antibody against PK-M2 (#21578, 1:100, Signalway Antibody) for 1 hour at room temperature. Cells were washed before incubation with secondary antibody (Goat Anti-Rabbit IgG H&L, Alexa Fluor® 647, #ab150083, 1:500, Abcam) for 30 min at room temperature. Analysis was performed through flow cytometry.

2.9. PK-M2 Immunofluorescence

For PK-M2 immunofluorescence, 25,000 cells were seeded in Lab-Tek chamber slides (Thermofisher). Chambers were washed with PBS, and cells were incubated with 4% paraformaldehyde at 37°C for 10 min and chilled on ice for 1 min. Cells were permeabilized by adding 90% methanol for 30 min on ice. Blocking was performed with 10% goat serum, for 30 min at room temperature under agitation. PK-M2 (#21578, rabbit, 1:100, Signalway Antibody) and Vimentin (#ab8069, mouse, 1:100, Abcam) primary antibodies were incubated for 30 min at room temperature, under agitation. Secondary antibodies (anti-rabbit IgG-TRITC, #T5268, 1:32, Sigma; anti-mouse IgG (H+L) Alexa Fluor 488, #62–6511, 1:100, Invitrogen) were added and incubated in the dark at room temperature for 30 min. Lastly, cells were counterstained with DAPI. Images were taken with Keyence BZ-9000 microscope (Keyence Corporation of America, Itasca, IL).

2.10. RNA extraction and RT-qPCR

RNA was isolated from cells with TRIzol reagent (Ambion) according to the manufacturer’s recommendations. RNA levels were measured with Nanodrop spectrophotometer (Thermofisher). Complementary DNA synthesis was performed on 1 μg of isolated RNA by using cDNA synthesis kit (Genesee Scientific). Quantitative PCR was done with qPCRBIO SyGreen Mix lo-ROX (Genesee Scientific) according to manufacturer’s instructions. Primers were purchased from Thermofisher. PK-M2: Fw. 5’-ATTATTTGAGGAACTCCGCCGCCT-3’; Rev, 5’-ATTCCGGGTCACAGCAATGATGG-3’, RPLP0: Fw, 5’-TGATGTGCAGCTGATCAAGACT-3’, Rev, 5’-GATGACCAGCCCAAAGGAGA-3’).

2.11. Pyruvate kinase assay

Single cell suspensions of SUM159PT monolayers were seeded on 24-well tissue culture-treated plates, in duplicates. The cells were allowed to attach overnight and the next day cells were treated with either 1 mM H₂O₂, 10 μM TEPP46 or a single dose of 8 Gy. Cells were removed from one set of the wells from each treatment group and the number of live cells was determined by staining with trypan blue and counting with a hemocytometer. The other set of wells was used for determination of pyruvate kinase activity in each group via the Pyruvate Kinase Assay kit (Abcam, #ab83432) according to the manufacturer’s instructions. Briefly, cells were washed once with PBS, followed by addition of pyruvate kinase assay buffer to each well for cell lysis. The cell lysate mixture was diluted to contain 2000 cells / 50μL (based on the cell numbers obtained with the trypan blue counting) with assay buffer and 50μL of the cell lysate from each group was transferred to 96-well plates. Then, 50μL of the master reaction mix was added to each sample. The optical density (OD570nm) was measured as a function of time using a plate reader (SpectraMax M5, Molecular Devices, Sunnyvale, CA) at room temperature every minute for a total of 15 min. The signal increase per unit of time correlates with the amount of pyruvate generated by pyruvate kinase. Two OD570nm values (A₁ and A₂) within the linear range of the reaction for each sample were identified. The ΔA = A₂ – A₁ was determined and plotted for each group. Three biologically independent experiments were performed for each experimental group.

3. Results

3.1. Exposure to radiation induces metabolic changes in TNBC cells

To determine overall metabolic changes in TNBC cells induced by radiation, we first evaluated radiation-induced changes in glucose uptake and lactate production over 3 days in SUM159PT and MDA-MB-231 TNBC cells. Glucose uptake was measured via flow cytometry using the fluorescent analogue of glucose 2NBDG or radiolabeled glucose (¹⁸FDG). Both cell lines showed significantly more glucose uptake in the irradiated (8 Gy) compared to the non-irradiated controls by 24 hours (Figure 1A), and this continued for up to 3 days, which is the latest time point measured (Figure 1A and Supplementary Figure 1A). These data suggest that surviving cells significantly re-wire glucose metabolism for an extended period of time, relative to the non-irradiated cells. Interestingly, the increase in glucose uptake 24 hours following irradiation was partially prevented when cells were irradiated in the presence of a potent antioxidant, the reduced form of glutathione (GSH), which scavenges IR-induced reactive oxygen species (ROS) (Supplementary Figure 1B–C). This suggested that the metabolic reprogramming induced by radiation could be due, at least in part, to the ROS generated by radiation.

Figure 1. Exposure to IR modifies breast cancer cells metabolism.

SUM159PT and MDA-MB-231 cells were seeded as monolayers and irradiated with a single dose of 8 Gy. At the indicated time points after irradiation cells were incubated with a fluorescent glucose analogue (2NBDG, A) or a radiolabeled glucose analogue (¹⁸FDG, B) and glucose uptake was determined. (C) SUM159PT cells were grown as mammospheres and irradiated with a single dose of 8 Gy. At different time points after irradiation, cells were dissociated and incubated with 2NBDG and glucose uptake was determined via flow cytometry. (D) Lactate levels were measured in the media of irradiated SUM159PT and MDA-MB-231 monolayers at 24, 48 and 72 hours after irradiation and normalized to the non-irradiated cells. (E) SUM159PT monolayer cells were irradiated with 8 Gy and 24 hours later the media was collected and the concentration of glutamine in the media was determined via a Glutamine detection assay kit. (A-D) All results shown are normalized to 0 Gy controls (100%, dotted line). Paired, 2-tailed t test: *P < .05, **P < .01, ***P < .001

The IR-induced increase in glucose uptake in both TNBC lines was further confirmed using radiolabeled glucose (¹⁸FDG) (Figure 1B). Due to cell death caused by IR, irradiated cell cultures are less dense than the unirradiated ones at all the time points. Therefore, in order to exclude culture density as an artifact, we determined glucose uptake in non-irradiated cell cultures as a function of time and cell density. As shown in Supplementary Figure 2A–C, glucose uptake increases, rather than decreases with higher cell density. In addition, cell lines are often propagated in culture media containing supraphysiological levels of glucose, which could confound glucose uptake measurements. Therefore, we also tested the effect of IR on glucose uptake in cells propagated in media containing physiological levels of glucose (~5.6 mM). Interestingly, glucose uptake at basal levels increased when cells were propagated under physiological levels of glucose, however similar radiation-induced increases in glucose uptake were observed (Supplementary Figure 2D). Finally, TNBC cells propagated as mammospheres in serum-free media enriched for breast cancer stem cells (BCSCs) [34] also increased glucose uptake after irradiation (Figure 1C). These results suggested that the increase in glucose uptake by TNBC cells is a general phenomenon in different sub-types of breast cancer cells.

We next determined the effect of radiation on lactate production and observed that the increase in glucose uptake after radiation was accompanied by an increase in lactate production (Figure 1D). Although 48 hours after radiation no further increase was observed in SUM159PT cells, lactate production was still accumulating at 48 and 72 hours in MDA-MB-231 cells, with a 4-fold increase 72 hours post irradiation (Figure 1D). Since some of the lactate can also be derived from glutamine [35], we also measured glutamine consumption in irradiated cells. Indeed, irradiated cells consumed significantly more glutamine than non-irradiated controls 24 hours after irradiation (Figure 1E). Together, these results led us to conclude that TNBC cells that survive radiation robustly reprogram their metabolism independently of the cell culture system, and that this metabolic response must be important for cell survival and radioresistance.

3.2. Radiation affects expression, localization and enzyme activity of PK-M2

The increase in glucose uptake, accompanied by an increase in lactate production (Figure 1) suggested an enhanced glycolytic phenotype (Warburg effect [19, 36]) in irradiated TNBC cells. One of the main glycolytic enzymes that is often re-expressed in cancer cells and regulates flux through glycolysis is the M2 isoform of pyruvate kinase (PK-M2) [15, 22]. PK-M2 catalyzes the conversion of phosphoenol pyruvate (PEP) to pyruvate, which, in turn is converted to lactate. Therefore, we decided to determine the effect of radiation on the gene and protein expression levels of PK-M2. Interestingly, PKM2 mRNA and protein expression levels were significantly upregulated post irradiation (Figure 2A–B and Supplementary Figure 3A–C).

Figure 2. Radiation modulates PK-M2 expression levels and its enzymatic activity.

(A) PK-M2 gene expression levels were determined in SUM159PT monolayers at 1, 5, 24 and 72 hours after a single dose of 8 Gy. Results shown are normalized to 0 Gy control (dotted line). Paired, 2-tailed t test: **P < .01, ***P < .001. (B) SUM159PT monolayers were irradiated with 0 or 8 Gy. Cells were fixed, permeabilized and immunostained for the PK-M2 protein. PK-M2 protein expression levels were determined by flow cytometry at 24 and 72 hours after irradiation. PK-M2 mean fluorescence intensity was measured and normalized to 0 Gy controls (100%, dotted line). Paired, 2-tailed t test: **P < .01. (C) SUM159PT cells were irradiated with a single dose of 8 Gy. Pyruvate kinase activity was then measured at the indicated time points. Results shown are normalized to 0 Gy non-treated control (Cnt). Paired, 2-tailed t test: *P < .05. (D) MDA-MB-231 cells were treated with 10μM TEPP46 3 hours prior to irradiation followed by another treatment 24 hours later. Media was collected 3 hours after the second TEPP46 treatment and lactate levels were measured. Paired, 2-tailed t test: *P < .05, **P < .01

We next determined the effect of IR on the pyruvate kinase enzymatic activity of PK-M2. Although IR modestly increased PK-M2 enzymatic activity at 15 min and 3 hours after irradiation (Figure 2C), the PK activity was significantly inhibited 24 hours after irradiation (Figure 2C). As expected, and in agreement with a previous report [14] H₂O₂ significantly inhibited PK-M2 activity within 15 minutes of exposure (Supplementary Figure 3D). In contrast, a small molecule activator of PK-M2 (TEPP46) [28, 37] significantly increased PK-M2 enzymatic activity only 1 hour after treatment (Supplementary Figure 3D). Importantly, preventing the inhibition of PK-M2 enzymatic activity by radiation via the small molecule activator TEPP46, partly reversed the IR-induced increase in lactate production by the irradiated cells (Figure 2D). These findings suggested that the radiation-induced changes in PK-M2 levels and enzyme activity are regulating, at least in part, the metabolic changes occurring in irradiated TNBC cells. Interestingly however, although PK-M2 activation (via TEPP46) had a tendency to increase glucose uptake levels in non-irradiated and irradiated cells, it did not significantly alter them (Supplementary Figure 3E).

Unlike PK-M1, which is a constitutively active tetramer, PK-M2 has conformational flexibility and can readily switch between enzymatically-inactive dimers and active tetramers. Therefore, one interpretation of the observation that, in irradiated cells, PK-M2 enzymatic activity is reduced despite the increase in its expression and protein levels, is that after irradiation most of the PK-M2 protein exists in the enzymatically-inactive dimeric form. PK-M2 dimers are often found in the nucleus and have been implicated in gene regulation [23, 25, 38]. We therefore investigated whether IR affected the cellular localization of PK-M2 using immunofluorescence and western blotting of nuclear fractions. As expected, immunofluorescence studies showed that while non-irradiated cells showed weak nuclear staining for PK-M2, upon irradiation the fraction of cells with nuclear PK-M2 was increased (Figure 3A–B). This observation was confirmed by western blotting of nuclear fractions of irradiated and non-irradiated SUM159PT cells (Supplementary Figure 4A). Interestingly, irradiating TNBC cells in the presence of GSH partially prevented the nuclear translocation of PK-M2 (Figure 3C) suggesting that the ROS generated by radiation is contributing to the nuclear translocation of PK-M2. In addition, although cancer cells in general exhibit high glucose uptake (Warburg effect [19, 36]), heterogeneity exists (Supplementary Figure 4B). In light of our findings that radiation induces an increase in glucose uptake (Figure 1A–C) it seems that high glucose consumption may be an important factor in surviving oxidative stress, such as induced by radiation. Therefore, we decided to determine the relationship between glucose uptake levels and nuclear PK-M2. For this, SUM159PT cells were labeled with 2NBDG and sorted into populations with low (2NBDG^low) or high (2NBDG^high) glucose uptake (Supplementary Figure 4B). The cells were plated on microscope chamber slides and the next day immunostained for PK-M2 (Supplementary Figure 4C). Interestingly, a higher fraction of 2NBDG^low cells contained nuclear PK-M2 compared to 2NBDG^high (Figure 3D and Supplementary Figure 4C).

Figure 3. Radiation induces nuclear translocation of PK-M2 and PK-M2 activators enhance the effect of radiation on differentiated BC cells.

(A) SUM159PT cells were seeded as monolayers in microscope chamber slides and irradiated with a single dose of 8 Gy. PK-M2 (red) and vimentin (green) were immunostained and nuclei were counterstained with DAPI (blue). Cells were visualized via confocal microscopy (60X). (B) SUM159PT cells were immunostained as in A, and cells were visualized via fluorescence microscopy. The fraction of cells with nuclear PK-M2 was determined by counting the number of cells with PK-M2 staining in the nucleus and dividing by the total number of cells in the field of view (FOV). >30 FOVs were used to determine the average fraction. (C) SUM159PT and MDA-MB-231 cells were seeded and irradiated with a single dose of 8 Gy, with or without GSH (2.5 mM) (n = 3–4). As in A, cells were immunostained for PK-M2 and cells with nuclear PK-M2 were counted and normalized to the 8 Gy group (dotted line). (D) Sorted 2NBDG-low and 2NBDG-high cells were seeded on microscope chamber slides and allowed to adhere overnight. The next day cells were fixed, permeabilized and immunostained for the PK-M2 protein and the nuclei were counterstained with DAPI. The fraction of cells with nuclear PK-M2 was determined by counting the number of cells with PK-M2 staining in the nucleus and dividing by the total number of cells in the field of view (FOV). >30 FOVs were used to determine the average fraction. (E-F) SUM159PT and MDA-MB-231 cells were treated with DMSO or TEPP46 (10 μM) as monolayers. Cells were then detached, irradiated with 2 or 8 Gy, and plated at clonogenic density in 6-well plates. After 10 to 15 days, cells were fixed and stained, and colonies were then counted. The results were normalized to 0 Gy controls. Paired, 2-tailed t test: *P < .05.

The above data strongly suggested that the enzymatically-inactive, dimeric form of PK-M2 is mediating metabolic changes that might be important for cancer cell survival under oxidative stress, such as during radiation therapy. Therefore, we reasoned that PKM2 activators might sensitize TNBC cells to irradiation. To test this, we irradiated TNBC cells derived from either un-selected cultures propagated as adherent monolayers in serum-supplemented conditions or as BCSC-enriched mammosphere cultures in serum-free conditions, and performed either classical colony formation assays or extreme limiting dilution (mammosphere) assays (ELDA) [39]. Activating PK-M2 (via TEPP46) in adherent cultures had no effect on their plating efficiency (Supplementary Figure 4D) or the surviving fraction at 2 Gy for both TNBC lines (Figure 3E–F, light gray bars). However, PK-M2 activation did sensitize adherent cells to 8 Gy (Figure 3E–F, dark gray bars). Importantly, the sensitizing effect of activating PK-M2 was greatly enhanced in the BCSC-enriched cell populations as demonstrated by the cancer stem cell frequencies determined by ELDA [40] (Figure 4A–B). It is worth noting here that PK-M2 activation also significantly reduced intrinsic BCSC frequencies for both TNBC lines (Figure 4A–B, SUM159PT, DMSO vs. TEPP, p = 0.00316; MDA-MB-231, DMSO vs. TEPP, p = 0.0103).

Figure 4. PK-M2 activators enhance the effect of radiation on the radioresistant BCSC population.

(A-B) SUM159PT and MDA-MB-231 were seeded into low adhesion 96-well plates, irradiated with a single dose of 2 Gy or 8 Gy, and treated every day for 5 days with TEPP46 (10 μM). The number of spheres formed was counted 7 days after seeding. Left panels show the percentage of sphere forming capacity defined by the number of spheres formed relative to the number of cells plated. Right panels show the analysis through ELDA (Extreme Limiting Dilution Analysis) software and represent the CSC frequency as 1/CSC frequency. (C) SUM159PT-ZsGreen cells were plated and irradiated with a single dose of 8 Gy. Cells were fixed and immunostained for PKM2 expression 24 hours later, and then analysed by flow cytometry. Geometric mean fluorescence were normalized to 0 Gy controls. (D) SUM159PT-ZsGreen cells were sorted and PKM2 kinase activity was assessed in the ZsG^−high population as well as in ZsG^−neg cells. Paired, 2-tailed t test: *P < .05. (E) Sorted 2NBDG-low and 2NBDG-high SUM159PT cells were seeded in mammosphere-forming conditions in serum-free media on 96-well plates. Cells were allowed to form mammospheres for ~ 10 days and the % of cells capable of forming mammospheres was determined by counting the number of spheres per well.

3.3. PK-M2 and breast cancer stem cells

The remarkable effect of PK-M2 activation on the frequency of BCSCs either at basal levels or after irradiation (Figure 4A–B) led us to further investigate the status of PK-M2 in BCSCs. For these experiments we used a fluorescent reporter (ZsGreen-cODC) for BCSCs that is based on low proteasome activity of CSCs, a characteristic of BCSCs that we, and others have previously reported [32, 33, 41–51]. Differentiated cells have high proteasome activity that leads to immediate degradation of the fluorescent protein reporter (ZsGreen-negative), while BCSCs with low proteasome activity accumulate the fluorescent protein (ZsGreen-positive) and can thus readily be identified and isolated based on cell-associated fluorescence. SUM159PT cells stably expressing the reporter (SUM159PT-ZsGreen) were irradiated and 24 hours later immunostained for PK-M2. The PK-M2 protein levels in the non-stem, ZsGreen-neg cells and ZsGreen-pos BCSCs were assessed via flow cytometry. Interestingly, the protein levels of PK-M2 were similar in the two cell populations at basal level (Supplementary Figure 4E) and both the ZsGreen-neg and the ZsGreen-pos populations increased PKM2 levels after irradiation (Figure 4C), similar to the unselected population of cells (Figure 2B). However, the increase in PKM2 levels was greater in the BCSC population (Figure 4C), suggesting an enhanced role for PKM2 in the BCSC that could explain in part the sensitivity to PK-M2 activators (Figure 4A–B). To further explore this possibility we also compared the pyruvate kinase activity of the ZsGreen-neg non-stem cell population with the ZsGreenpos BCSCs. Surprisingly, the PK activity of BCSCs was significantly higher compared to that of non-stem cells (Figure 4D). These results suggested that the relative abundance of PK-M2 tetramers with respect to dimers is greater in the BCSC population compared to the non-stem (ZsGreen-neg) cells. Interestingly, and in agreement with a quiescent phenotype of BCSCs, cells with low glucose uptake (2NBDG^low) formed significantly more mammospheres compared to the population of cells with high glucose uptake (2NBDG^high) (Figure 4E). Together with the observation that 2NBDG^low cells are also enriched for cells with nuclear PK-M2, these data suggest that nuclear PK-M2 may maintain a stemness phenotype in BCSCs that have low, basal glucose uptake levels and are less glycolytically active.

4. Discussion

Studies with normal cells show that metabolic reprogramming is required for cells to survive lethal oxidative stress [12, 52]. Normal cells exposed to chemically-induced oxidative stress can immediately (seconds to minutes) shuttle glucose into the pentose phosphate pathway (PPP) [12, 29, 52]. Such acute metabolic reprogramming, which precedes transcriptional responses, generates reducing equivalents in the form of nicotinamide adenine dinucleotide phosphate (NADPH) that help neutralize potentially lethal levels of ROS and maintain cellular redox balance [29]. The study put forth here demonstrates that TNBC cells exposed to oxidative stress through IR reprogram their metabolism by increasing the long-term utilization of glucose, consuming more glutamine and producing more lactate (Figure 1). Oxidative stress induced by chemical means (H₂O₂) has been shown to rapidly re-wire metabolic flux in human lung cancer cells ultimately facilitating an anti-oxidant response which has been ascribed to the inhibition of the redox-sensitive pyruvate kinase activity of PK-M2 within 15 minutes of exposure as a result of the oxidation of a cysteine residue (Cys³⁵⁸) [14]. In contrast, our results show that inhibition of PK-M2 activity via oxidative stress induced by radiation only became apparent and significant 24 hours after radiation (Figure 2C), suggesting that at least the initial radiation-induced ROS are insufficient to inactivate the enzyme. This observation is important in that it demonstrates that oxidative stress induced by chemical means, as a model for oxidative stress does not necessarily reflect the cellular response during radiation therapy used to treat most cancers.

The appreciable inhibition of the pyruvate kinase activity of PK-M2 by irradiation (Figure 2C) seems contradictory with the observation that irradiated cells increase lactate production at the same time point (24 hours post irradiation) (Figure 1D). Theoretically, reduced activity of PK-M2 after radiation should result in less pyruvate production, which should in turn result in less pyruvate available for lactate formation. However, other groups have shown that the conversion of pyruvate to either lactate or acetyl-CoA, which then undergoes further oxidation by the mitochondria, is determined by the relative enzymatic activity of PKM2. Low enzymatic activity of PK-M2 promotes the conversion of pyruvate to lactate giving rise to the ‘Warburg effect’ observed in most cancers [19,53], whereas high activity of PK-M2, similar to the constitutively active PK-M1, promotes conversion of pyruvate to acetyl-CoA [20–22, 54]. Therefore, our results suggest that the inhibition of PK-M2 enzymatic activity by radiation contributes to the increased levels of lactate production in irradiated TNBC cells. This conclusion is reenforced by the observation that preventing the radiation-induced inhibition of the enzymatic activity of PK-M2 via the small molecule activator, TEPP46 partially reverses the increase in lactate production after radiation (Figure 2D). However, this does not exclude other possibilities for lactate production, such as the conversion of glutamine to lactate via glutaminolysis that could further add to the pool of lactate produced by irradiated cells. Indeed, this possibility is supported by the increased levels of glutamine consumption in irradiated TNBC cells (Figure 1E). Another contributing factor could be the changes in endogenous metabolite levels upstream of PK-M2 that are known allosteric activators of PK-M2 enzymatic activity, such as fructose-1,6-bisphosphate (FBP) or serine [55, 56]. Although we did not determine whether radiation induces changes in the levels of FBP or serine, others have shown that under (chemically induced) oxidative stress glucose carbons are diverted away from glycolysis and shunted through the PPP [14] suggesting a decreased availability of glucose carbons for the generation of FBP or serine. This scenario would further promote an overall inactivation of the cellular PK-M2 pool, and as a result contribute to the increased lactate production after irradiation.

Intriguingly, the radiation-induced decrease in pyruvate kinase activity of PK-M2 was accompanied by increases in mRNA and protein expression levels, suggesting that most of the PK-M2 protein existed in the enzymatically-inactive conformation (either monomers or dimers) after radiation. A number of studies have shown that PK-M2 dimers lack pyruvate kinase activity and translocate to the nucleus to regulate gene transcription [18, 23–26, 57, 58]. Some studies suggest that PK-M2 regulates transcription by acting as a protein kinase [23, 38, 59, 60], although the protein kinase activity of nuclear PK-M2 remains controversial [61]. In our study in TNBC cells, radiation induced translocation of PK-M2 to the nucleus and this appeared to be ROS dependent (Figure 3A–C). This finding agrees with a recent study showing that IR induces nuclear translocation of PK-M2 in glioblastoma multiforme cells and that nuclear PK-M2 participates in DNA repair after radiation by facilitating homologous recombination of DNA double strand breaks [62]. Interestingly, in our study, TNBC cells with low basal glucose uptake (2NBDG^low) were enriched for breast CSCs (Figure 4E) and had a higher proportion of cells with nuclear PK-M2 (Figure 3D) compared to those with high basal glucose uptake, suggesting that PK-M2 may play a role in maintaining “stemness” in TNBC cells. These results are consistent with the findings that PK-M2 plays a role in maintaining a “stem-cell” phenotype in breast cancer cells via the Wnt/β-catenin pathway [18].

We had anticipated that the PK activity of BCSCs would be lower relative to the non-stem cell population as this would mean that BCSCs depend more on the enzymatically-inactive PK-M2 dimers and therefore would be more susceptible to PKM2 activator that promote a tetrameric conformation, as shown in Figure 4A–B. However, the surprising finding that BCSCs have significantly higher PK-M2 activity relative to the non-stem cell population agree with a quiescent phenotype for BCSCs and suggest that the depletion of BCSCs by PK-M2 activators is not mainly due to the effect on the pyruvate kinase activity, but that other factors are in play, such as the possibility of sequestering PK-M2 into the cytoplasm and out of the nucleus.

Finally, small molecule compounds that promote the tetrameric, enzymatically active conformation of PK-M2 have been developed and have been shown to prevent the nuclear translocation of PK-M2 [63], and have anti-proliferative effects on lung cancer cells [27, 37]. Importantly, our results show that these small molecule activators of PKM2 can deplete breast CSCs in vitro (Figure 4A–B), thus strongly suggesting that in addition to having anti-proliferative effects [27, 37] these activators could also impact the therapy-resistant stem cell compartment of tumors. These studies support the idea of combining PK-M2 activators with radiation to enhance the effect of radiation therapy in resistant cancers, such as TNBC.

Abbreviations

¹⁸FDG

[¹⁸F]fluorodeoxyglucose

2NBDG

2-deoxy-2-[(7-nitro-2,1,3-benzoxadiazol-4-yl)amino]-D-glucose

BC

breast cancer

bFGF

fibroblast growth factor 2

CSC

cancer stem cell

EGF

epidermal growth factor

GAPDH

glyceraldehyde 3-phosphate dehydrogenase

GSH

glutathione

H2DCFA

2’,7’-dichlorodihydrofluorescein diacetate

HIF1α

hypoxia inducible factor

IR

ionizing radiation

NADPH

nicotinamide adenine dinucleotide phosphate

PK-M2

M2 isoform of pyruvate kinase

PPP

pentose phosphate pathway

ROS

reactive oxygen species

RT

radiation therapy

TNBC

triple negative breast cancer