Breast cancer-associated skeletal muscle mitochondrial dysfunction and lipid accumulation is reversed by PPARG

Abstract

The peroxisome proliferator-activated receptors (PPARs) have been previously implicated in the pathophysiology of skeletal muscle dysfunction in women with breast cancer (BC) and animal models of BC. This study investigated alterations induced in skeletal muscle by BC-derived factors in an in vitro conditioned media (CM) system and tested the hypothesis that BC cells secrete a factor that represses PPAR-γ (PPARG) expression and its transcriptional activity, leading to downregulation of PPARG target genes involved in mitochondrial function and other metabolic pathways. We found that BC-derived factors repress PPAR-mediated transcriptional activity without altering protein expression of PPARG. Furthermore, we show that BC-derived factors induce significant alterations in skeletal muscle mitochondrial function and lipid accumulation, which are rescued with exogenous expression of PPARG. The PPARG agonist drug rosiglitazone was able to rescue BC-induced lipid accumulation but did not rescue effects of BC-derived factors on PPAR-mediated transcription or mitochondrial function. These data suggest that BC-derived factors alter lipid accumulation and mitochondrial function via different mechanisms that are both related to PPARG signaling, with mitochondrial dysfunction likely being altered via repression of PPAR-mediated transcription, and lipid accumulation being altered via transcription-independent functions of PPARG.

INTRODUCTION

Breast cancer (BC)-associated skeletal muscle fatigue is a chronic problem among BC survivors, being reported by a majority of patients both before and after receiving anticancer therapies (1–5). Recent studies show that deficits in muscle function predict shorter survival in cancer, perhaps due to the fact that fatigue is known to reduce a patient’s tolerance to anticancer therapies (3, 6–10). Therefore, improving muscle function in patients with BC has the potential to improve both quality of life and survival in the most commonly diagnosed cancer type in women.

The syndrome of cancer-associated cachexia, where individuals with cancer lose a significant portion of their bodyweight and also experience a decline in physical function (11, 12), has gained increasing attention in recent years (13). Although most agree that for a person to be cachectic, they must exhibit a significant degree of weight loss (11, 12, 14), it is now generally accepted that cachexia is a process that progresses through distinct stages, including 1) precachexia, where individuals have not lost significant bodyweight but experience other cachexia-related symptoms; 2) cachexia, where patients have lost 5% or more of their prediagnosis body weight; and 3) refractory cachexia (12). Improvements in muscle function could be of substantial benefit to patients at any stage along this continuum. In precachexia and cachexia, improving muscle function could allow patients to complete anticancer treatment regimens and/or physical therapies aimed at preventing cachexia progression, and in refractory cachexia, improvements in muscle function would represent a welcomed quality of life benefit to many patients.

The peroxisome proliferator-activated receptors (PPARs) are lipid sensing, ligand-activated transcription factors previously implicated in a variety of pathologies, including cancer, atherosclerosis, Alzheimer’s disease, type 2 diabetes mellitus, and others (15–18). These transcription factors aid in regulating whole body energy homeostasis via regulation of genes involved in lipid metabolism and mitochondrial functions (15, 19–23). Recently, downregulation of PPAR-γ’s (PPARG) transcriptional activity in skeletal muscle has been identified as a potential central regulator of the increased muscle fatigue experienced by women with BC and is recapitulated in the patient-derived orthotopic xenograft model of BC (24, 25). However, the mechanism by which PPARG downregulation induces muscle dysfunction has not yet been explored. In the current study, we used conditioned media (CM) from murine breast tumor cells lines representing the luminal and Her2/neu⁺ BC subtypes to test the hypothesis BC cells secrete a factor that represses PPARG expression and its transcriptional activity, leading to downregulation of PPARG target genes involved in mitochondrial function and other metabolic pathways. The experimental data we present support a mechanism whereby breast tumor-secreted factors directly interact with skeletal muscle cells to induce mitochondrial dysfunction and lipid accumulation through both transcriptional-dependent and -independent activities of PPARG.

EXPERIMENTAL PROCEDURES

Cell Culture

Cell lines used include EpH4-EV (immortalized normal murine mammary epithelium), EO771 (murine luminal BC), NF639 (murine HER2/neu-overexpressing BC), HEK293 (human embryonic kidney), and C2C12 (murine myoblasts). All cell lines were obtained from ATCC (Virginia), with the exception of EO771, which were obtained from Dr. Metheny-Barlow at the Wake Forest University. All cell lines were cultured in DMEM (ThermoFisher, Massachusetts) supplemented with 10% heat-inactivated fetal bovine serum (Atlanta Biologicals, Georgia) and penicillin-streptomycin (ThermoFisher) at 37°C with 6% CO₂.

Exogenous PPARG Myogenic Cell Line

Lentiviral particles containing pLenti-C-PPARG2-mGFP-P2A-Puro or pLenti-mGFP-P2A-Puro were purchased from Origene (Maryland). Titers were provided by the manufacturer. C2C12 cells were plated at 125,000 cells per well in 24-well plates and infected with a multiplicity of infection of 75 transforming units per cell with 8 µg·mL⁻¹ polybrene in antibiotic-free DMEM. Media were changed 20 h after infection. Starting 48 h after infection, cells were cultured in 2 µg·mL⁻¹ puromycin for 10 days, after which cells were maintained in 0.5 µg·mL⁻¹ puromycin indefinitely. Uninfected control cells exhibited 100% death within 3 days of puromycin selection. Second passage cells were used in Western blotting analysis to assess expression of PPARG2 protein. All experiments with these cell lines were conducted within five passages of lentiviral infection as higher passage cells lost differentiation competence.

Conditioned Media Collection

CM donor cells were plated at ∼15% confluence in separate 10 cm dishes for 48 h. The 48-h CM was then removed from all cell lines using a serological pipette, centrifuged at 1,500 rpm for 10 min, and the supernatants were collected via decanting into new centrifuge tubes. The collected CM was then diluted 1:3 in fresh growth media before application to recipient cells. CM was always applied to recipient cells within 2 h of collection, in most cases within 15 min of collection (Fig. 1A). In some experiments, rosiglitazone (MilliporeSigma, Massachusetts) in dimethyl sulfoxide (DMSO) was added to the diluted conditioned media to final concentrations of 10 µM rosiglitazone and 0.1% DMSO before CM application to donor cells. In experiments where rosiglitazone was used, DMSO was added to control CM to 0.1% final concentration as a vehicle control.

Western Blotting

Differentiated C2C12s were lysed in 50 mM Tris-HCl (pH 6.8) with 2% sodium dodecyl sulfate and heated to 100 °C for three cycles of 3 min each with vortexing and brief centrifugation between heating cycles. Lysates were diluted to a final concentration of 1 µg·µL⁻¹ in NuPAGE LDS Sample Buffer (ThermoFisher Scientific) and 5% β-mercaptoethanol. Twelve micrograms of total protein was loaded per well and resolved in NuPAGE Novex 4%–12% Bis-Tris Gels (ThermoFisher Scientific). Proteins were transferred to nitrocellulose membrane, blocked for 1 h in 1× tris-buffered saline (TBS), 0.1% Tween-20, 5% milk followed by incubation with primary antibody overnight at 4 °C in TBS-Tween-20 with 5% milk. Membranes were then washed three times in TBS + 0.1% Tween-20 before application of appropriate secondary antibody (ThermoFisher Scientific) for 90 min at room temperature, and again before application of Pierce ECL Western Blotting Substrate (ThermoFisher Scientific). Relative band intensity was quantified using the GE Amersham Imager 600 (GE Healthcare Life Sciences, Marlborough, Massachusetts) and normalized to GAPDH. Primary antibodies included PPARγ (No. PA3-821A) and GAPDH (No. 2118S), and a goat anti-rabbit HRP secondary antibody (ThermoFisher, No. 32460) was used for all experiments.

BC-CM Metabolic Analyses in C2C12 Myotubes

C2C12 cells were plated into Agilent Seahorse XF96 or XF24 (Agilent Technologies, California) plates at 10,000 cells well⁻¹ and differentiated in 2% horse serum (Atlanta Biologicals) in DMEM with antibiotics for 48 h. Meanwhile, EpH4-EV, EO771, NF639, and C2C12 cells were plated at ∼15% confluence in separate 10 cm dishes for 48 h. Forty-eight-hour CM, prepared as described above, was then applied to the 2-day differentiated C2C12 cells in Seahorse assay plates for 48 h (n = 6–12 wells per treatment condition, noted in figure legends) before conducting the Agilent Seahorse XF Cell Mito Stress Test, Agilent Seahorse XF Glycolysis Stress Test, or Agilent XF Long-Chain Fatty Acid Stress Test protocols according to manufacturer’s instructions (sample sizes for each experiment are identified in figure legends). For the Mito Stress Test, substrate (Seahorse XF DMEM with 1 mM pyruvate, 2 mM glutamine, and 10 mM glucose) was added before running the assay. Injections in final concentrations were as follows: 1) oligomycin (1 µM), 2) FCCP (0.5 µM), and 3) rotenone/antimycin-A (1 µM). For the Long-Chain Fatty Acid Stress Test, substrate was prepared by adding BSA-Conjugated Palmitate (5 mM) to Seahorse XF Substrate Limited Media. Injections were performed as in the Mito Stress Test. For the Glycolysis Stress Test, substrate was prepared by adding 2 mM glutamine to Seahorse XF Base Medium. Injections were as follows: 1) glucose (10 mM), 2) oligomycin (1 µM), and 3) 2DG (50 mM). For the ATP Coupling assay, substrate (final concentration of 10 mM pyruvate + 1 mM malate) was added and incubated in a non-CO₂ incubator at 37°C for 10 min before being placed in the instrument. Injections in final concentrations were as follows: 1) ADP (4 µM), 2) oligomycin (4 mM), 3) FCCP (4 µM), and 4) rotenone (4 µM) + antimycin-A (4 µM). For the Electron Flow assay, substrate (final concentration of 10 mM pyruvate + 1 mM malate) was added and incubated in a non-CO₂ incubator at 37 °C for 10 min before being placed in the instrument. Injections in final concentrations were as follows: 1) rotenone (2 µM), 2) succinate (10 mM), 3) antimycin-A (4 µM), and 4) ascorbate (10 mM) + TMPD (1 mM). The mix, wait, and measure cycle was 30 s, 30 s, and 2 min. Two cycles were performed before the first injection and two cycles were performed following each injection. Data were normalized as described previously. Oxygen consumption and/or extracellular acidification rates were normalized to account for inter- and intraplate variability, resulting in each individual well’s first measurement representing 100%. Subsequent to this normalization, individual parameters of aerobic and glycolytic metabolism were calculated according to manufacturer’s instructions.

Analyses of BC-CM on Isolated Mitochondria from C2C12 Myotubes

C2C12 myoblasts were plated at ∼15% confluence in 10 cm² culture plates and induced to differentiate using 2% horse serum (HS). Forty-eight-hour CM from EO771 and NF639 cells was applied to 2-day differentiated C2C12 myotubes for 48 h. Following CM exposure, C2C12 myotubes were collected, and mitochondria were then isolated using the Mammalian Mitochondria Isolation Kit for Tissue & Cultured Cells (BioVision, Cat. No. K288), according to the manufacturer’s protocol. Isolated mitochondria were suspended in mitochondrial assay solution (MAS), and the Bradford protein assay was performed to quantify protein content. Mitochondrial function and health were subsequently assessed with ATP Coupling and Electron Flow assays, which were performed in the Agilent Seahorse Bioanalyzer per the manufacturer’s instructions using 10 µg/well and the substrates malate (1 mM) and pyruvate (10 mM). Three biological replicates were run for each cell line being examined, and three technical replicates were run for each biological replicate. This experiment was performed in duplicate plates to account for inter- and intraplate variability. The oxygen consumption rate (OCR) values for each time point of the technical replicates for each biological replicate were averaged and used to derive the mitochondrial respiratory parameter data for each biological replicate. Raw OCR data were normalized to offset instrument error by setting the most negative value from all successful wells in the data set to zero. State 2 respiration was taken by the highest OCR value of the two time points before the first injection, State 3 respiration was taken by the highest OCR value of the two time points immediately after the injection of ADP, State 4o respiration was taken by the lowest OCR value of the two time points immediately after the injection of oligomycin, State 3u respiration was taken by the highest OCR value of the two time points immediately after the injection of FCCP, and nonmitochondrial respiration was taken by the lowest OCR value of the two time points immediately after the injection of rotenone + antimycin-A. For the ATP Coupling Assay, injections in final concentrations were as follows: 1) ADP (4 µM), 2) oligomycin (4 mM), 3) FCCP (4 µM), and 4) rotenone (4 µM) + antimycin-A (4 µM). For the Electron Flow assay, injections in final concentrations were as follows: 1) rotenone (2 µM), 2) succinate (10 mM), 3) antimycin-A (4 µM), and 4) ascorbate (10 mM) + TMPD (1 mM). The mix, wait, and measure cycle was 30 s, 30 s, and 2 min. Two cycles were performed before the first injection and two cycles were performed following each injection. Data were normalized as described previously.

PPAR-Reporter Assays

HEK293 cells were transfected with PPRE-H2b-eGFP (26) (Addgene No. 84393) using Invitrogen Lipofectamine 3000 (ThermoFisher), selected with 500 ng·µL⁻¹ Gibco geneticin (ThermoFisher) for 20 days, and flow-sorted to select the cells expressing GFP. The resulting HEK293-PPRE-H2b-eGFP cell line was plated at ∼15% confluence in a 24-well plate. The following day, cells were imaged using the BioTek Cytation 5 Cell Imaging Multi-Mode Reader (Agilent Technologies) to collect baseline GFP intensity, and 72-h conditioned media from HEK293, EpH4-EV, EO771, and NF639 cell lines were applied to the 24-well plate, using individual wells as biological replicates (n = 6 wells per treatment condition). CM donor cells were plated on day 0 (D₀) as described above, and HEK-PGFP cells were plated at ∼15% confluence on day 1 (D₁). On D₂, baseline GFP intensity was collected using the BioTek Cytation 5 Cell Imaging Multi-Mode Reader (Agilent Technologies), and subsequently, 48-h conditioned media from HEK293, EpH4-EV, EO771, and NF639 cell lines were applied to the 24-well plate (n = 3–6 wells per treatment condition, noted in figure legends). HEK-PGFP cells were then incubated in CM under normal culture conditions and GFP intensity was quantified 24 h later using identical imaging settings as the baseline measurements. Mean GFP intensity per well was calculated using Gen5 Microplate Reader and Imager Software (Agilent Technologies), obtaining a single value per well that represented the mean GFP intensity of all cells in the imaging field for that well.

qRT-PCR for PPAR Target Genes

C2C12 cells were plated at ∼90% confluence in 24-well plates (150,000 cells/well) on D₀ and differentiated in 2% horse serum (Atlanta Biologicals) in DMEM with antibiotics. On D₃, differentiation media were refreshed on the C2C12s, and CM donor cells were plated as described in CM Collection. On D₄, media on the differentiating C2C12s were changed back to normal growth media (10% FBS in DMEM with antibiotics). On D₅, CM was collected and applied to the differentiated C2C12 cells for 2 h or 24 h, with n = 3–8 wells per condition, as noted in figure legends. Total RNA was isolated from CM- and/or drug-treated cells using Trizol (ThermoFisher Scientific, Waltham, Massachusetts) and established methods (27). Thousand two-hundred microgram of cDNA was produced using Invitrogen SuperScript III First-Strand Synthesis System (ThermoFisher Scientific) according to manufacturer’s protocol, and relative expression of selected genes was analyzed using SYBR Green PCR Master Mix (ThermoFisher Scientific) with the Applied Biosystems QuantStudio 6 Flex (ThermoFisher Scientific), using 60 ng cDNA per reaction. Primer efficiencies were determined to be between 85% and 115%, and relative mRNA expression was calculated using the Pfaffl method (28) normalized to 18 s rRNA. Primer3 was used to design qRT-PCR primers (Table 1) (29).

Table 1.

Primers used for qRT-PCR

	Forward Primer	Reverse Primer
18s	5′-AATGCTTTCGCTCTGGTCCG	5′-CCTGGATACCGCAGCTAGGA
Cidec	5′-AGCTAGCCCTTTCCCAGAAG	5′-CCTTGTAGCAGTGCAGGTCA
Fabp4	5′-CATCAGCGTAAATGGGGATT	5′-TCGACTTTCCATCCCACTTC
Pparg	5′-TGTGGGGATAAAGCATCAGGC	5′-CCGGCAGTTAAGATCACACCTAT
Slc1a5	5′-CATCACCATCCTGGTCACAG	5′-CCTTCCACGTTGAGGACAGT

qRT-PCR for Mitochondrial DNA

C2C12 cells were plated at ∼90% confluence in 24-well plates (100,000 cells well⁻¹) on D₀ and differentiated in 2% horse serum in Gibco DMEM with antibiotics. Differentiation and CM application were conducted as described in qRT-PCR for PPAR Target Genes, with CM applied for 24 h and n = 8 wells per condition. Total DNA was isolated from CM-treated cells using the DNeasy Blood & Tissue Kit (Qiagen, Hilden, Germany) according to manufacturer’s instructions. Forty to sixty nanogram of DNA in 1 µL eluting buffer was used for qRT-PCR using Applied Biosystems TaqMan Universal PCR Master Mix (ThermoFisher Scientific) with the Applied Biosystems QuantStudio 6 Flex (ThermoFisher Scientific). Primer-probe sets were purchased from ThermoFisher Scientific (Applied Biosystems Gene Expression Assays, COX2 Mm03294838_g1 and 18S Mm03928990_g1). Relative COX2 mtDNA was quantified using the ΔΔCT method normalized to 18S gDNA.

Oil Red O

C2C12 cells were plated at ∼90% confluence in 24-well plates (100,000 cells·well ^− 1) on D₀ and differentiated in 2% horse serum in Gibco DMEM with antibiotics. On D₂, differentiation media were refreshed on the C2C12s and CM donor plates were set up as described under CM Collection. On D₄, CM was collected and applied to differentiated C2C12s and CM donor cells were split to allow CM to be collected once more after 48 h incubation. On D₆, CM was again collected and was used to refresh the CM on the recipient cells. On D₇, cells were fixed in 4% formaldehyde for 30 min, washed three times in distilled water, and made permeable with 60% isopropanol for 10 min. Oil red o stock solution had been previously suspended in 100% isopropanol and filtered using a 0.2-µm bottle-top filter. Oil red o stock solution was diluted to 60% in distilled water to create the working solution. Working solution was applied to the fixed and permeable cells for 10 min. Cells were washed three times with distilled water and stored at 4°C until imaging, typically within 24 h of staining. Images were taken using a Zeiss Axiovert 40 CFL microscope with a Zeiss Axiocam 105 color camera, running ZEN 2.3 software. Effort was made to take images at the center of each well to produce even illumination across images. All images within each experiment were taken using identical imaging settings. Images were analyzed using the countcolors package v. 0.9.1 (30) in R v. 3.6.1 (31), quantifying the area of each image within a 50% radius from true red.

Assessment of Myotube Diameter

To assess myotube diameter, the diameter of every myotube per field was manually measured at its widest width using the measure function of the ImageJ ROI manager (National Institutes of Health, Bethesda, Maryland), and myotube diameter distributions for each experimental condition were constructed. These measurements were performed on the images used for the oil red o staining described previously, resulting in measurements being performed at day 7 of myotube differentiation.

Statistical Analyses, Reproducibility, and Rigor

Data analyses were conducted in R v. 3.6.1 (31) using ggpubr v. 0.2.4 (32) for data visualization and statistical tests. Unless otherwise stated, all statistical tests used were unpaired, two-tailed t tests for two-group comparisons, one-way ANOVA for multigroup comparisons, and two-way ANOVA for comparisons involving two variables. When multiple comparisons were used within an experiment, Holm–Bonferroni correction was applied to adjust P values. Graphs represent data from a single, independent experiment. All experiments have been repeated at least twice with similar results, with most experiments being repeated more than three times. All cases where inconsistent results were obtained across repeated experiments are explicitly stated in the text. For all box and whisker plots, the total height of the box represents the interquartile range (IQR), the thick center line represents the median value, and the whiskers extend to the most extreme data point that is not an outlier, with an outlier being defined as a value more extreme than 1.5 × IQR. Outlier values are represented by a single dot beyond the whiskers.

RESULTS

BC-CM Represses PPARG Expression and PPAR-Mediated Transcription

Because PPARG is downregulated in the skeletal muscle of women with BC and in mouse models of BC (24), we first investigated the capacity of BC-derived factors to downregulate Pparg mRNA expression in an in vitro model system. To test this, we collected CM from BC or control cell lines and applied this CM to differentiated skeletal muscle cells in culture (Fig. 1A). Cells and large debris were removed from the media via gentle centrifugation and decanting, and the conditioned media were diluted 1:3 in fresh growth media to correct pH and overall nutrient content. Importantly, all CM was applied to myotubes within 2 h of collection to ensure that biologically active components would not degrade or be damaged by freezing before application. Using this model system, we found that CM from BC cell lines significantly downregulated Pparg mRNA expression (Fig. 1B) and several of its purported transcriptional targets (33) in differentiated skeletal muscle after as little as 2-h of CM exposure, compared with media conditioned by skeletal muscle myoblasts (Fig. 1C). However, no significant change in PPARG protein expression was detected (Fig. 1D), suggesting that the downregulation of PPARG target genes’ mRNA is due to a repression of PPAR-mediated transcriptional activity rather than a downregulation of PPARG protein abundance. To directly address this discrepancy, we generated an HEK reporter cell line stably expressing a PPAR-responsive reporter construct driving expression of GFP. Using this system, we have consistently observed a significant reduction in GFP intensity in response to BC-CM. In contrast, CM from a nontumorigenic mammary epithelial cell line does not repress GFP expression (Fig. 1E).

Figure 1.

BC-CM represses PPARG expression and PPAR-mediated transcription. Forty-eight-hour conditioned media were collected from BC (EO771 or NF639) or control cell lines (EpH4-EV, C2C12, or HEK293), centrifuged to pellet large debris, decanted, and diluted in fresh complete growth media before application on recipient cells. Figure created using BioRender (A). Pparg mRNA expression in 5-day differentiated C2C12s exposed to 48-h CM for 24 h, quantified using qRT-PCR, n_CON = 12, n_EO771 = 6, n_NF639 = 5 (B). mRNA expression of PPARG target genes in 5-day differentiated C2C12s exposed to 48-h CM for 2 h, quantified using qRT-PCR, n_CON = 3, n_EO771 = 3 (C). Western blotting analysis of PPARG protein expression in 5-day differentiated C2C12s exposed to 48-h CM for 48 h, normalized to GAPDH. Error bars represent means ± standard deviation (D). Quantification of GFP intensity in HEK-PGFP reporter cells exposed to 48-h CM, quantified at 0 and 24 h of CM exposure. Adjusted P values represent Holm–Bonferroni adjusted P values of paired t tests, n = 8 per group (E). *P < 0.05, **P < 0.01. BC, breast cancer; CM, conditioned media; PPAR, peroxisome proliferator-activated receptors; PPARG, PPAR-γ.

BC-CM Induces Mitochondrial Dysfunction and Lipid Accumulation in Myotubes

We next asked whether BC-CM induced functional alterations in differentiated muscle cells that could be reflective of the increased muscle fatigability seen in women with BC and BC animal models. We found that BC-CM induced substantial changes in aerobic metabolism, specifically leading to a repeatable reduction in oxygen consumed in ATP generation, whereas oxygen consumed in ATP generation was unchanged by the normal mammary epithelial cell line EpH4-EV (Fig. 2A). Other parameters of mitochondrial function, including alterations of basal respiration, maximal respiration, and proton leak, were not consistently replicated across five individual experiments, as assessed in myotubes. CM from the luminal EO771 BC cell line repressed aerobic metabolism without altering glycolytic function, whereas the HER2-overexpressing NF639 line repressed both aerobic metabolism and glycolytic functions (Fig. 2B). This reduction in aerobic metabolism does not appear to be related to mitochondrial biogenesis, as suggested by the lack of alteration in mitochondrial DNA copy number in CM-treated myotubes (Fig. 2C).

Figure 2.

BC-CM induces mitochondrial dysfunction and lipid accumulation in myotubes. Assessment of mitochondrial function affected by BC-CM in 2-day differentiated C2C12 myotubes treated with 48-h CM for 48 h, quantified using the Seahorse XF Mito Stress Test. OCR is calculated as the percent change from the control values. Error bars represent means ± standard deviation, n = 10 per group. One experiment is presented, with consistent findings across 4/5 experiments for repression of aerobic ATP production, but inconsistent findings for other presented parameters across five experiments. Tables represent results of two-way ANOVA (A). Extracellular acidification rate of 2-day differentiated C2C12 myotubes treated with 48-h CM for 48 h, quantified using the Seahorse XF Glycolysis Stress Test. Error bars represent means ± standard deviation, n = 12 per group, except n_NF639 = 10. Extracellular acidification rate (ECAR) is calculated as the percent change from control (B). Relative quantitation of mtCOX2 DNA in 5-day differentiated C2C12s exposed to 48-h CM for 24 h, normalized to 18S gDNA, n = 8 per group (C). Oil red o image quantification from 4-day differentiated C2C12s treated with 48-h CM for 72 h, represented as the fraction of each image stained red with oil red o, n = 8 per group (D). Representative images from an independent oil red o experiment conducted under identical conditions to E, ×40 magnification (E). Assessment of mitochondrial function in myotubes following exposure to BC-CM, with the long chain fatty acid palmitate as a substrate (F). Myotube diameter was determined in fully differentiated myotubes exposed to CM from control and E0771 breast tumor cells (G). *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001. BC, breast cancer; CM, conditioned media; PPAR, peroxisome proliferator-activated receptors.

Because lipotoxicity-induced mitochondrial dysfunction has been reported in various clinical contexts (34–37) and because individuals with cancer have been shown to have increased lipid content in muscle (38, 39), we assessed intramyocellular lipid concentration in response to BC-CM using oil red o staining and automated image quantification. Using this methodology, we repeatedly observed a significant increase in intramyocellular lipid staining with EO771-CM, but in no experiment did we observe an increase in response to NF639-CM (Fig. 2, D and E). In addition to the quantifiable increase in lipid content, EO771 CM appeared to alter lipid localization. Rather than being relatively evenly distributed across the cell, the lipid in C2C12s treated with EO771 CM shifts to a more punctate distribution, perhaps due to concentration in lipid droplets. The potential contribution of greater lipid accumulation on mitochondrial function was analyzed by in myotubes after exposure to BC-CM and performing a mitochondrial stress test using palmitate as the substrate to determine fatty acid oxidation. Several functional parameters were assessed; however, there were no significant differences observed in any of the groups that were assessed (Fig. 2F). Diameter of C2C12 myotubes exposed to either EO771 or C2C12 CM was not different, suggesting the observed alterations in mitochondrial function are independent of myotube atrophy (Fig. 2G).

BC-CM Decreases Mitochondrial Respiration and Electron Flow through Mitochondrial Electron Transport Chain Complexes

To evaluate overall mitochondrial function and health in this in vitro system, we performed mitochondrial stress assays in isolated mitochondria from C2C12 myotubes exposed to BC-CM to assess States 3 and 4 respiration as well as Electron Flow through the mitochondrial ETC complexes. State 3 respiration was significantly lower in isolated mitochondria following exposure to both E0771 and NF639 CM, whereas State 4 respiration was only lower following exposure to E0771 CM; nonmitochondrial respiration was not affected (Fig. 3A). Calculation of the respiratory control ratio (RCR), using absolute values for State 3 and State 4 respiration, revealed a lower value following exposure of NF639 CM (C2C12: 8.2 ± 3.4; E0771: 6.2 ± 2.4, P = 0.054; and NF639: 3.7 ± 0.9, P = 0.023). Both E0771 and NF639 CM also negatively affected Electron Flow through the mitochondrial ETC complex I, whereas NF639 CM negatively affected Electron Flow through ETC complex 2 and 4 (Fig. 3B). These data in isolated mitochondria provide support of our data in skeletal myotubes exposed to BC-CM and suggest a direct effect of a factor or factors in BC-CM on mitochondrial function in skeletal muscle.

Figure 3.

BC-CM decreases mitochondrial respiration and Electron Flow but does not decrease lipid oxidation. ATP Coupling (A) and Electron Flow (B) analysis on isolated mitochondria from C2C12 cells treated with 48-h CM, quantified using the Seahorse XF bioanalyzer. OCR is calculated as the percent change from the control values. Error bars represent means ± standard error, n = 6 per group collected from two duplicate 96-well assay plates (n = 3 per plate). Respiration state data for State 3 (ADP-stimulated respiration of isolated coupled mitochondria with malate and pyruvate), State 3u (uncoupled respiration via FCCP), State 4o (oligomycin inhibited respiration), and injection responses are presented for the ATP Coupling Assay. Electron transport complex activity for complexes 1, 2, and 4 is presented, as well as the injection responses. BC, breast cancer; CM, conditioned media; PPAR, peroxisome proliferator-activated receptors.

Exogenous PPARG Expression Rescues BC-CM-Induced Mitochondrial Dysfunction and Lipid Accumulation

To determine whether the alterations induced in skeletal muscle by BC-derived factors are regulated by PPARG, we first attempted to generate stable myogenic cell lines expressing exogenous PPARG constructs or shRNA against PPARG using lentiviral infection and subsequent selection. Consistent with previous reports (40), genetic modification of PPARG in C2C12 myoblasts resulted in cell lines that rapidly lost differentiation competence, within five passages from lentiviral infection. Despite these limitations, early passage cells expressing exogenous PPARG (Fig. 4A) were used to ascertain whether PPARG is involved in the response of myotubes to BC-derived factors. In these experiments, it was found that cells expressing exogenous PPARG were resistant to both BC-CM-induced repression of aerobic ATP generation (Fig. 4B) and BC-CM-induced lipid accumulation (Fig. 4, C and D). Cells stably expressing shRNA against PPARG lost differentiation capacity immediately after lentiviral infection and could therefore not be used to determine whether PPARG ablation phenocopies the effect of BC-CM in the context of differentiated muscle cells.

Figure 4.

Exogenous PPARG expression rescues BC-CM-induced mitochondrial dysfunction and lipid accumulation. Western blotting analysis of PPARG expression in C2C12s infected with lentiviral particles carrying pLenti-C-PPARG2-mGFP-P2A-Puro (+PPARG2) or pLenti-mGFP-P2A-Puro (CON) after 5 days of differentiation (A). Assessment of mitochondrial function in 2-day differentiated CON-C2C12 or +PPARG2 cells exposed to 48-h CM for 48 h, quantified using the Seahorse XF Mito Stress Test, n = 10 per group. C2C12 cell lines are noted on the x-axis, and CM treatment conditions are listed above the bar graphs. Tables represent results of two-way ANOVA. *P < 0.05 for +PPARG2 cell line compared with CON-C2C12 cell line under same CM condition; #P < 0.05 for CON cell line treated with BC-CM compared with CON cell line treated with CON-CM (B). Oil red o image quantification from 4-day differentiated CON or +PPARG2 cells treated with 48-h CM for 72 h, represented as the fraction of each image stained red with oil red o, n = 6 per group (C). Representative images from the experiment quantified in C, ×20 magnification (D). *P < 0.05, **P < 0.001; ***P < 0.0001. BC, breast cancer; CM, conditioned media; PPAR, peroxisome proliferator-activated receptors; PPARG, PPAR-γ.

PPARG Agonist Rosiglitazone Rescues BC-CM-Induced Lipid Accumulation but Fails to Rescue BC-CM-Induced Mitochondrial Dysfunction and PPAR Repression

Finally, we pharmacologically targeted PPARG to overcome BC-induced mitochondrial dysfunction and lipid accumulation, using the potent PPARG agonist rosiglitazone (rosi) of the thiazolidinedione (TZD) drug class. Rosiglitazone did induce PPAR-mediated transcription in the absence of BC-CM (Fig. 5A), but it was unable to rescue BC-CM’s repression of PPAR-mediated transcription (Fig. 5B). Unsurprisingly then, this agonist did not rescue BC-induced repression of aerobic ATP production (Fig. 5C). However, rosiglitazone did potently repress BC-induced lipid accumulation (Fig. 5D). These data indicate that the mechanism by which rosiglitazone prevents BC-induced lipid accumulation is independent of PPARG’s transcriptional activity and that BC-induced lipid accumulation is not the cause of BC-induced mitochondrial dysfunction in this model system.

Figure 5.

PPARG agonist rosiglitazone fails to rescue BC-CM-induced mitochondrial dysfunction and PPAR repression but rescues BC-CM-induced lipid accumulation. Quantification of GFP intensity in HEK-PGFP reporter cells exposed to 10 µM rosiglitazone (rosi) or DMSO (control), quantified at 0 and 24 h of drug exposure, n = 3 per group (A). Quantification of GFP intensity in HEK-PGFP reporter cells exposed to 48-h CM in the presence of 10 µM rosiglitazone, quantified at 0 and 24 h of CM exposure. Adjusted P values represent Holm–Bonferroni adjusted P values of paired t tests, n = 4 per group (B). Assessment of mitochondrial function in 2-day differentiated C2C12s cells exposed to 48-hr CM ± rosiglitazone for 48 h, quantified using the Seahorse XF Mito Stress Test, n = 10 per group, except n_NF639+rosi = 7. Drug treatment conditions are noted on the x-axis, and CM treatment conditions are listed above the bar graphs. Tables represent results of two-way ANOVA. #P < 0.05 for DMSO control with BC-CM compared with DMSO control treated with CON-CM (C). Oil red o image quantification from 4-day differentiated C2C12 cells treated with 48-h CM for 72 h, represented as the fraction of each image stained red with oil red o, n = 8 per group (D). BC, breast cancer; CM, conditioned media; PPAR, peroxisome proliferator-activated receptors; PPARG, PPAR-γ. **P < 0.001.

DISCUSSION

In the present study, we have identified that BC-derived factors can significantly alter mitochondrial function and PPAR activity in skeletal muscle cells without the involvement of immune or stromal cell mediators, indicating that BC cells secrete a factor that induces these effects by interacting directly with skeletal muscle. Numerous publications have characterized changes in serum contents of women with BC including changes in miRNAs, lipids, and proteins (41–45), though the exact sources of many of these factors remain unknown. To our knowledge, this report is the first showing that BC-CM induces alterations in skeletal muscle mitochondrial function, with previous studies focusing on BC-CM’s effect on immune cells, fibroblasts, endothelial cells, or normal mammary epithelium (46–51), or on transcriptional changes induced in skeletal muscle (52). Although speculative, these in vitro data suggest that molecular and functional alterations in skeletal muscles are the result of tumor-specific factors that are secreted into the circulation and affect skeletal muscles and are not necessarily linked to systemic changes in inflammation, nutritional or metabolic disturbances, or side effects of treatments.

It is well established that BC cells can regulate both local and systemic immune functions (53, 54) and that inflammatory signaling is intrinsically linked to cancer-associated skeletal muscle wasting (12, 55). Thus, a reasonable assumption is that BC’s influence on immune cells is the cause of BC-induced skeletal muscle fatigue. This study contradicts this assumption by showing that no immunological mediators are required for BC cells to induce significant alterations in skeletal muscle metabolic function. This study also contradicts recent reports of negative results in other cancer types, where CM from cancer cells was unable to directly induce mitochondrial deficits in differentiated skeletal muscle (56), indicating that different cancer types induce skeletal muscle dysfunction via different mechanisms. While we cannot conclusively state that mitochondrial dysfunction or lipotoxicity are the root causes of BC-induced skeletal muscle dysfunction, it is reasonable to hypothesize a causal link that should be further investigated. Most likely, there are numerous mechanisms contributing to skeletal muscle dysfunction that would need to be simultaneously targeted to provide symptom control.

The mechanism by which BC cells induce mitochondrial dysfunction in this model system appears to be related to signaling by the PPAR proteins, as evidenced by the rescue of mitochondrial dysfunction and lipid content provided by exogenous expression of PPARG. This finding provides support for the translational relevance of this in vitro model system, as PPARG has been previously identified as a potential key regulator of BC-induced skeletal muscle fatigue using both human data and data generated in the patient-derived orthotopic xenograft model of BC (24, 25, 57). In addition, this finding illuminates potential therapeutic modalities using the numerous FDA-approved agents that target the PPAR proteins, including drugs with various specificities for the three PPAR isoforms. In support of this possibility, we show here that the PPARG-specific agonist rosiglitazone completely reversed BC-CM’s effect on skeletal muscle lipid accumulation. Interestingly, this result does not appear to be due to rosiglitazone activating PPAR-mediated transcription, suggesting that the effect is mediated by noncanonical functions of PPARG.

Although we were unable to rescue BC-induced mitochondrial dysfunction using rosiglitazone, it is possible that pharmacological agents targeting PPAR-α, PPAR-δ, or a combination of the three isoforms could overcome these effects of BC-derived factors (58). Alternatively, it is possible that the mitochondrial dysfunction induced by BC-CM is due to alteration of ligand-independent functions of PPARG, which would be overcome by the addition of PPARG protein but not by exogenous ligand. Numerous reports confirm the ability of PPARG to function as a transcription factor in the absence of ligand binding (59–62), and there is evidence of PPARG having ligand-independent functions that are also independent of its transactivation domain (63). Pharmacological strategies to target these unique functions could include non-TZD insulin-sensitizing agents or epigenetic modifiers.

Mechanistically, muscle fatigue can result from a myriad of factors, but the underlying ability of the contractile elements to maintain force production relies upon the muscle’s ability to resynthesize ATP following stimulus. We have observed dysregulation of metabolically significant PPAR transcriptional networks in the skeletal muscle of BC-PDOX mice (24), as well as a significant reduction in ATP from mitochondria isolated from skeletal muscle of these mice (24). In support of these murine observations, proteomics analysis in the skeletal muscle of HER2⁺ BC patients revealed lower protein expression of nearly every component of the mitochondrial ETC and a correspondingly lower ATP content in muscle biopsies from BC patients of multiple tumor subtypes (25). These data clearly demonstrate mitochondrial dysfunction in muscle of patients with BC, which will directly impact the fatigability of skeletal muscle. The results presented herein expand on our previous studies and suggest this mitochondrial dysfunction is a direct result of BC-derived factors. Intriguingly, we demonstrate that this mitochondrial dysfunction is accompanied by increased intracellular lipid deposition in myotubes exposed to BC-CM. Intramuscular lipid deposition has also been observed in type 2 diabetes (64) and smoking-induced insulin resistance (65). As the downregulation of PPARG expression and transcriptional activity is associated with the development of insulin resistance (66) and many PPAR transcriptional targets are mitochondrial lipid transport proteins (33), we hypothesized that a decreased capacity for mitochondrial lipid import may contribute to the observed phenotype of BC-induced lipid accumulation, which may then contribute to insulin resistance. However, this lipid accumulation does not appear to be due to decreased rates of fatty acid oxidation in cells exposed to BC-CM, as assessed by palmitate stimulated mitochondrial stress assays. This result was somewhat unexpected, and future studies should aim to further elucidate the mechanisms of this lipid accumulation. Finally, aberrant intramyocellular lipid deposition can in many cases contribute to lipotoxicity characterized by the accumulation of toxic lipid intermediates such as diacylglycerols and ceramides (67), increased production of ROS (68), and is associated with changes in autophagic dynamics (69, 70), all of which are implicated in reduced mitochondrial function within skeletal muscle. The sum of these observations suggests that skeletal muscle lipotoxicity in breast cancer could play a major role in the phenotype of BC-induced skeletal muscle fatigue independent of mitochondrial dysfunction, and future studies should investigate this potential link in greater detail.

In summary, our results confirm the ability of BC-derived factors to directly alter metabolic function in skeletal muscle cells without the involvement of other cell types. These BC-CM-induced phenotypes in skeletal muscle could be enacted by a variety of secreted factors, including miRNAs, exosomes, or proteins, and appears to be mediated via signaling of the PPAR proteins, though perhaps not through their canonical functions as ligand-activated transcription factors. Specifically, BC-derived factors have the ability to repress the transcriptional activity of PPARG, and this effect is not due to downregulation of PPARG protein abundance. In addition, repression of PPARG transcriptional activity induces mitochondrial dysfunction in muscle cells and a deficiency in ATP supply while also being associated with greater lipid accumulation. Although PPARG protein levels are not affected in our model, basal protein expression of PPARG is relatively low in skeletal muscle, providing an explanation for the ability of exogenous PPARG to rescue these phenotypes whereas PPARG agonists do not (Fig. 6). These results provide support for previous publications implicating the PPAR proteins in BC-induced skeletal muscle fatigue and provide rationale for investigating PPAR agonists to improve quality of life in survivors of BC (24, 57).

Figure 6.

Summary of findings. BC-derived factors induce several physiological changes in target cells in vitro, including mitochondrial dysfunction, lipid accumulation, and repression of PPAR-mediated transcription (left). The PPARG agonist rosiglitazone reverses BC-induced lipid accumulation but does not reverse mitochondrial dysfunction or the observed repression of PPAR-mediated transcription (center). Expression of exogenous PPARG2 in myotubes reverses both BC-induced mitochondrial dysfunction and lipid accumulation (right). Figure created using BioRender. BC, breast cancer; PPAR, peroxisome proliferator-activated receptors; PPARG, PPAR-γ.

GRANTS

This research was supported by the National Institute of General Medical Sciences of the National Institutes of Health (NIH) under Award Number P20GM121322 (Lockman), American Cancer Society Institutional Research Grant 09-061-04 (E. E. Pistilli), and the WVCTSI U54GM104942 (Hodder).

DISCLAIMERS

The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health or other funding agencies.

DISCLOSURES

No conflicts of interest, financial or otherwise, are declared by the authors.

AUTHOR CONTRIBUTIONS

E.E.P. conceived and designed research; H.E.W. and D.A.S. performed experiments; H.E.W., D.A.S., S.R., and W.G. analyzed data; H.E.W., D.A.S., S.R., W.G., and E.E.P. interpreted results of experiments; H.E.W. prepared figures; H.E.W. and E.E.P. drafted manuscript; H.E.W., D.A.S., and E.E.P. edited and revised manuscript; H.E.W., D.A.S., S.R., W.G., and E.E.P. approved final version of manuscript.

DATA AVAILABILITY

The data described herein are available upon request. Requests should be directed to Dr. Emidio E. Pistilli (epistilli2@hsc.wvu.edu).