¹³C Stable Isotope Tracing Reveals Distinct Fatty Acid Oxidation Pathways in Proliferative vs. Oxidative Cells

Abstract

The TCA cycle serves as a central hub to balance catabolic and anabolic needs of the cell, where carbon moieties can either contribute to oxidative metabolism or support biosynthetic reactions. This differential TCA cycle engagement for glucose-derived carbon has been extensively studied in cultured cells, but the fate of fatty acid (FA)-derived carbons is poorly understood. To fill the knowledge gap, we have developed a strategy to culture cells with long-chain FAs without altering cell viability. By tracing ¹³C-FA we show that FA oxidation (FAO) is robust in both proliferating and oxidative cells while the metabolic pathway after citrate formation is distinct. In proliferating cells, a significant portion of carbon derived from FAO exits canonical TCA cycle as citrate and converts to unlabeled malate in cytosol. Increasing FA supply or β-oxidation does not change the partition of FA-derived carbon between cytosol and mitochondria. Oxidation of glucose competes with FA derived carbon for the canonical TCA pathway thus promoting FA carbon flowing into the alternative TCA pathway. Moreover, the coupling between FAO and the canonical TCA pathway changes with the state of oxidative energy metabolism.

Graphical Abstract

New & Noteworthy

By using ¹³C stable isotope resolved metabolomics and FA-driven oxygen consumption rate analysis, our study provides novel insights into the fate of FA carbon through β-oxidation and downstream TCA cycle in proliferative and oxidative cells. While both proliferative and oxidative cells demonstrate robust β-oxidation, they demonstrate distinct metabolic carbon fate downstream of citrate during TCA cycle oxidation. This differential TCA cycle engagement is likely to be important to balance catabolic and anabolic demands of the cell.

Introduction

Fatty acids (FAs) play multi-faceted roles in cellular health. They are key components of the structural scaffolds in biological membranes, modulate signaling pathways, facilitate molecular trafficking, and serve as both fuel storage and energy supply of the cells (1). FAs are important fuel for high-energy consuming organs with robust oxidative metabolism - such as the heart, skeletal muscle, and kidney (2). Upon cytoplasmic import, FAs are activated to form long-chain acyl-CoAs. Acyl-moieties are subsequently shuttled into peroxisomal and mitochondrial compartments for further oxidation. In the mitochondria, catabolic fatty acid β-oxidation (FAO) converts acyl-CoAs into acetyl-CoA (3). The resultant acetyl-CoA enters the tricarboxylic acid (TCA) cycle by combining with oxaloacetate to form citrate. Complete combustion of the acetyl-CoA in TCA cycle produces CO₂ and H₂O. Electrons extracted during oxidation of FA are carried by NADH to the electron transport chain (ETC) in the mitochondria to produce ATP (4).

Besides providing cellular energy currency, carbon substrates fuel other pathways to meet the metabolic demand of cell growth and function (5). The TCA cycle, with robust anaplerosis and cataplerosis serves as a central hub to balance catabolic metabolism with the biosynthetic needs of the cell (6, 7). Recently, a sub-compartmentalized extension of the TCA cycle has been linked to cell state (8). This non-canonical TCA cycle exports citrate from the mitochondria and reimports malate from the cytosol. This exchange provides citrate and, via ATP-citrate lyase (ACLY), generate acetyl-CoA in the cytosol while recycling TCA cycle intermediates and regenerating cytosolic NAD⁺ (9-11). This observation is, however, made in cell lines cultured in high glucose and amino acid enriched media. While the role of FAO in cell growth and survival has been increasingly recognized in proliferating cells (12, 13), the intracellular fate of FAO-derived carbon is under investigated. FAO flux is commonly assessed for ATP production in highly oxidative and energy consuming cells such as cardiac and skeletal muscle. A direct comparison of carbon flux downstream of fatty acid β-oxidation in proliferative and glycolytic cells versus terminally differentiated oxidative cells is lacking.

To address this gap in knowledge, we have developed strategies to introduce exogenous FA into cell culture media for cells with a wide range of mitochondrial oxidative capacity. Using ¹³C-labelled mixed long-chain fatty acid we were able to trace the fate of FA carbon through β-oxidation and downstream TCA cycle in these cells under different conditions. We show that FA β-oxidation is robust in both proliferating and oxidative cells while the metabolic pathway downstream of citrate is distinct for the two cell types. Mature cardiomyocytes show a near complete oxidation of FA-derived acetyl-CoA in the TCA cycle while a significant portion of FA-derived citrate is not further oxidized in proliferating cells.

Materials & Methods

Additional data that support the findings of this study are available from the corresponding author upon request.

Long-chain fatty acid mix preparation

U-¹³C fatty acid mix (13C Algal Lipid Mixture, Sigma # 487937, LOT #MBBC0375V) was prepared at a final concentration of 5 mM fatty acids (FA) with 12% fatty acid-free BSA (molecular ratio 2.2:1) (all purchased from Sigma Aldrich) in glucose/pyruvate-free Krebs-Henseleit Buffer (KHB) as described previously (14). The following long chain fatty acids (in % from molecular weight) were used for the regular FA mix: palmitic acid 50.9, palmitoleic acid 10.8, linoleic acid 13.2, linolenic acid 0.3, oleic acid 23.6, stearic acid 1.2. FAs were dissolved in 38% ethanol in 0.5mM Na₂CO₃ at 60°C under constant nitrogen gas. Upon dissolving, ethanol was boiled off by increasing the temperature to 80-90°C. BSA was dissolved in KHB at 37°C. BSA- and FA-solution were mixed together under constant stirring at 37°C. Successful complexing of FA to BSA was monitored when no clouding of the solution occurred. The FA-BSA solution was stirred for an additional 5 minutes before being dialyzed overnight and sterile filtered using 0.45 μm pore size membrane filter. Aliquots were stored at −20°C. Fatty acid concentration of the stock was determined using the HR Series NEFA-HR kit from Wako Life Sciences according to the manufacturer’s instruction. The final concentration of the FA mix used in experiments is indicated in the figure legends. An unlabeled fatty acid mix was prepared accordingly.

Cell culture

HEK293, MCF7 and HeLa cells were routinely cultured in high glucose DMEM (4.5g/L) containing 10% FBS and 1% Penicillin/Streptomycin. Human-derived fibroblasts were purchased from Coriell Institute (GM07843, GM20266 and GM02037). All fibroblast lines were cultured in low glucose DMEM (1g/L) containing 20% FBS and 1% Penicillin/Streptomycin. Proliferating C2C12 cells were cultured in glucose DMEM (4.5g/L) containing 10% FBS and 1% Penicillin/Streptomycin. For differentiation, upon confluency medium was switched to low glucose DMEM (1g/L) containing 2% HS (horse serum) and 1% Penicillin/Streptomycin for 5 days. Proliferating H9C2 cells were cultured in glucose DMEM (4.5g/L) containing 10% FBS and 1% Penicillin/Streptomycin. Upon confluency, H9C2 cells were differentiated with 1 μM retinoic acid in DMEM high glucose, 1% Penicillin/Streptomycin, 1% FBS for 5 days. Mycoplasma infection was tested regularly with a PCR-based kit (Sigma-Aldrich).

Induced pluripotent stem cell culture and cardiac directed differentiation

The human induced pluripotent stem (hiPS) cell line WTC (Bruce R. Conklin, M.D., The J. David Gladstone Institutes) was used for this study. The undifferentiated hiPS cells were maintained in mTeSR1 medium. The cardiac-directed differentiation protocol was adapted and modified from previous reports. In brief, the hiPS cells were seeded on matrigel-coated 6-well plates, and treated with the Wnt/β-catenin agonist CHIR-99021 (1 μM) for 24 hours upon reaching appropriate confluency. Cells were then exposed to CHIR-99021 (6 μM) in RPMI/B27 (minus insulin) medium for 48 hours (day 0). On day 2, media was changed to RPMI/B27 (minus insulin) medium containing the Wnt/β-catenin antagonist Wnt-C59 (2 μM). After 48 hours, media was changed to RPMI/B27 (minus insulin) medium. On day 6, media was changed to RPMI/B27 medium and replaced every other day. Spontaneous beating is usually observed between day 6 and day 10. On day 15, the cells were replated in PRMI/B27 medium containing 5% FBS and Rock inhibitor. Media was changed to DMEM medium (no glucose or sodium pyruvate) supplemented with 4mM sodium lactate on the next day and changed daily until day 19. On day 20, media was changed to PRMI/B27 medium with penicillin-streptomycin and replaced every other day until day 30. The purity of iPSC-derived cardiomyocytes (iPSC-CMs) was determined by the percentage of cTnT-positive cells via flow cytometry. Only iPSC-CMs with a purity greater than 80% were used in this study.

Cell death and toxicity assay

Cells were plated overnight and treated with control BSA solution or unlabeled mixed FA in low glucose DMEM (1g/L) with or without 1mM carnitine and indicated FA concentrations (see figure legends).

To determine cell death after 24h incubation with BSA/ FA, ReadyProbes^™ Cell Viability Imaging Kit, Blue/ Green (Thermo Fisher Scientific) was used. This two-color assay allows detection of dead and total cells using a plate reader.

To determine toxicity of FA treatment after short time (up to 24h), ToxiLight Assay Kit (Lonza) was used according to the manufacturer’s instructions.

Stable isotope tracing

For isotope labeling experiments, cells were plated overnight, washed 2 times with PBS and switched to low glucose DMEM (5.5mM) without FBS containing 1mM carnitine and indicated FA concentrations (see figure legends) for 3 hours. Regular FA mix was used in the same way. No FBS was present in culture media to which FA mix was added.

Steady state metabolite level

Steady state metabolite levels were determined in patient fibroblasts cultured in regular growth medium (low glucose DMEM (1g/L) containing 20% FBS and 1% Penicillin/Streptomycin).

Metabolite extraction

After indicated time points, cells were washed with 0.9% NaCl and harvested in 80% methanol: 20% H2O. Internal standard (methyl succinate, Sigma Aldrich) was added and samples were homogenized. After centrifugation, aqueous extracts were dried at 30°C in a Speed-Vac. The protein pellet was dissolved in 0.1M NaOH and used to determine protein concentration.

Mass spectrometry

Derivatization for GC/MS:

Dried aqueous extracts were incubated with 20mg/ml methoxyamine hydrochloride in pyridine (Sigma Aldrich) for 90 min at 37°C followed by MSTFA+1%TCMS (Thermo Scientific) for 30 min. For isotope tracing, dried aqueous extracts were incubated with 20mg/ml methoxyamine hydrochloride in pyridine (Sigma Aldrich) for 90 min at 70°C followed by MTBSTFA (Sigma Aldrich) for 30 min. FAME standard (Sigma Aldrich) was spiked into samples before loading samples into GC autosampler vials.

For steady state metabolite levels, dried aqueous extracts were incubated with 20mg/ml methoxyamine hydrochloride in pyridine (Sigma Aldrich) for 90 min at 70°C followed by MSTFA at 37°C for 30 min. FAME standard (Sigma Aldrich) was spiked into samples before loading samples into GC autosampler vials.

GC/MS conditions:

All samples were analyzed using an Agilent 7890 GC instrument equipped with a 5977 mass selective detector (MSD), employing an HP-5MS UI GC column. GCMS conditions were used according to the Fiehn library instructions from Agilent as described previously (14-16). Retention time of individual metabolites were annotated according to known standards. Peak intensities were normalized to total ion count (TIC) and protein concentration and depicted as fold change from control. IsoCor software was used to correct the fractional labeling for natural isotopic abundance and to quantify isotopologue distribution of labeled metabolites (15, 17).

Determination of L-Carnitine metabolites

Plated cell cultures were processed following an adjusted extraction protocol targeting L-carnitine-related metabolites (18, 19). Briefly, metabolites were extracted with 50% acetonitrile: 50% methanol on ice, sonicated for 10 min and centrifuged at 20800 g for 15 min at 4°C. Supernatants were dried and resuspended in 50% acetonitrile: 50% methanol. A Carnitine mix standard sample containing all the analytes in Table 1. was prepared for retention time validation of the analytes. The concentration range of the standard mix was 20-200 μM.

Table 1.

Mass transitions (Q1/Q3), retention time (RT), compound ID, declustering potential (DP), entrance potential (EP), collision energy (CE) and collision cell exit potential (CxP). The first MRM transition was always used as quantifier, and the second MRM transition as qualifier.

Q1 Mass	Q3 Mass	RT	ID	DP	EP	CE	CxP
400.395	84.897	5.51	Palmitoyl-L-carnitine-1	90	10	45	11
400.395	341.3	5.51	Palmitoyl-L-carnitine-2	90	10	15	11
370.3	85	4.76	trans-2-Tetradecenoyl-L-carnitine-1	120	10	30	11
370.3	311.3	4.76	trans-2-Tetradecenoyl-L-carnitine-2	120	10	30	11
370.3	144.1	4.76	trans-2-Tetradecenoyl-L-carnitine-3	120	10	30	11
398.345	84.882	5.36	trans-2-Hexadecenoyl-L-carnitine-1	100	10	45	11
398.345	339.193	5.36	trans-2-Hexadecenoyl-L-carnitine-2	100	10	20	11
426.4	85	5.72	Oleoyl-L-carnitine-1	100	10	50	11
426.4	367.4	5.72	Oleoyl-L-carnitine-2	100	10	15	11
424.339	84.873	5.23	Z,Z-9,12-Octadecadienoyl-L-carnitine-1	75	10	50	11
424.339	365.154	5.23	Z,Z-9,12-Octadecadienoyl-L-carnitine-2	75	10	20	11
372.3	85	4.86	Myristoyl-L-carnitine-1	90	10	30	11
372.3	60.1	4.86	Myristoyl-L-carnitine-2	90	10	46	11
428.4	85	6.4	Stearoyl-L-carnitine-1	90	10	30	11
428.4	60.1	6.4	Stearoyl-L-carnitine-2	90	10	46	11
368.3	85	4.27	Z,Z-5.8-Tetradecadienoyl-L-carnitine-1	75	10	30	11
368.3	60.1	4.27	Z,Z-5.8-Tetradecadienoyl-L-carnitine-2	75	10	46	11

For metabolite separation and detection, an Acquity I-class Plus UPLC system (Waters) coupled to an QTRAP 6500+ (SCIEX) mass spectrometer with electrospray ionization (ESI) source was used. Metabolites were separated by reversed-phase chromatography on a CSH C18 column at 45°C. An overview of multiple reaction monitoring (MRM) transitions that were used can be found in Table 1. Clear separation of L-carnitine-derived compounds was achieved by the chromatographic gradient applied according to Table 2. Data acquisition and processing was performed with the Sciex OS software suite (SCIEX).

Table 2.

Chromatographic conditions for separation of L-Carnitine analytes on a CSH C18 column (150 mm x 2.1 mm, 1.7 μm, Waters) kept at 45 °C. Eluent A: deionized water + 0.1% formic acid, Eluent B: Acetonitrile + 0.1% formic acid.

Time (min.)	Flow rate (mL/min)	% of Eluent A	% of Eluent B
Initial	0.300	95	5
0.5	0.300	95	5
4.0	0.400	40	60
8.0	0.400	20	80
10.0	0.400	1	99
11.0	0.400	1	99
11.1	0.400	95	5
12.5	0.300	95	5

Measurement of Oxygen Consumption Rate (OCR) with Seahorse XFe Analyzer

Cells were seeded in a Seahorse XFe96 cell culture plate 16 hrs prior to experiment and subjected to DMEM based conditioning media containing substrates as shown in Fig. 5B. The conditioning medium was replaced with assay medium (Seahorse DMEM supplemented with 2.5 mM glucose, 1 mM carnitine, 0.1 mM mixed FA, pH to 7.4) with CPT inhibitor, etomoxir (5 μM), or vehicle control, DMSO, and incubated for 60 min in a non-CO2 incubator at 37C. The plate was assayed with the Seahorse XFe96 Analyzer and a pseudo-baseline OCR was measured. Sequential injections of inhibitors (2.5 ug/mL oligomycin, 3 μM carbonyl cyanide-4-(trifluoromethoxy)phenylhydrazone, and 1 μM rotenone/antimycin A) were added into each well and changes in OCR were measured.

Figure 5: Assessment of FAO by oxygen consumption rate.

A: Schematic depiction of TCA cycle fluxes and alternative pathways and substrates and their contribution to either 13C labeling or oxygen consumption rates.

B: Culture conditions to assess FA-sensitive respiration in HEK293 cells.

C-E: Representative OCR tracings to assess FA-driven respiration under full substrate conditions (C), glucose starvation (B) or combined glucose and glutamine starvation (C). n=4-5.

F: Percent change of OCR after etomoxir inhibition to assess FAO-sensitive respiration under different substrate conditions in HEK293 cells. Statistical comparison is between control vs. etomoxir. n = 4-5.

All data are mean ± SEM. For statistical considerations, a two sided t-test (F) was performed. * = p < 0.05, ** = p < 0.005.

Western Blot

For western blot analysis, protein samples were loaded onto a 4%-12% gradient gel for 10% SDS-PAGE, transferred to PVDF membrane, and blocked with 5% BSA in TSBT. Proteins were detected by specific antibodies and corresponding secondary antibodies (anti-HADHA, LSBio #LS-C482426-50; anti-HADHB, LSBio #LS-C482486; anti-vinculin, Cell Signaling Technologies #4650 or #13901; Anti-Total OXPHOS Cocktail, abcam #ab110413, anti-MT-ND1, abcam #ab181848, anti-VDAC, Cell Signaling Technologies #4661). Signals were visualized using a Bio-Rad or Licor imager.

Statistics

The numbers of independent experiments are specified in the relevant figure legends. Data are expressed as mean + standard error of the mean (SEM). Statistical analysis was performed with Prism 9.0 or 10.0 software (GraphPad). Normal distribution of data was confirmed by Shapiro-Wilk test. Statistical comparisons between 2 groups were conducted by unpaired two-tailed t-test. Statistical comparisons between 3 or more groups were conducted by 1-way or 2-way ANOVA followed by a posthoc analysis to determine statistical significance, as described in the figure legends. The value of p < 0.05 was considered statistically significant. Heatmaps were generated using GraphPad Prism 10.0. Peak intensities from GC/MS experiments were normalized and log transformed.

Results

1. Development of a cell culture system with mixed fatty acids.

Most cell culture studies are performed in media containing glucose and glutamine as the primary energy substrates. A limited number of studies assessed the role of FAO in cultured cells by adding one single species of FA, e.g. palmitate, a saturated and largely lipotoxic FA (20). These preparations are not ideal as FA composition in vivo is more complex, consisting of a mix of saturated and unsaturated FA of various chain lengths and stoichiometries (21, 22).

We previously used a long-chain FA mix (‘mixed FA’) that contains a physiological combination of saturated and unsaturated long-chain FAs (Fig. 1A), which we have used in both isolated perfused hearts and primary adult cardiomyocytes culture (14, 23). To expand this model to non-cardiac cells we first tested if mixed FA induces cell death in HEK293 cells cultured for 24 hours. Cell death was observed when cells were treated with palmitate or mixed FA in the absence of carnitine, which was completely prevented by including carnitine in the medium (Fig. 1B). Using a more sensitive readout for cellular cytotoxicity, we also measured at multiple time points (1-24h) and used different doses of mixed FA (5-600 μM) (Fig. 1C, D). No cytotoxicity was measured at any time point or dose. Together, these experiments demonstrate that mixed long-chain FA do not induce cell death in non-cardiac cells when mitochondrial uptake is stimulated simultaneously with carnitine. This culture condition allows us to trace FAO in cells with a range of FAO capacity and oxidative energy production.

Figure 1. FA enriched culture for proliferating cells.

A: Composition of long-chain FA mix. Fatty acids were conjugated to BSA in the molar ratio FA:BSA 2.2:1.

B: Quantification of cell death after FA treatment. HEK293 cells were treated with control BSA, mixed FA or palmitate (100 μM, 24h) in the presence of absence of carnitine (1 mM). n= 3-4.

C: Cell toxicity in HEK293 after stimulation with control BSA or 100 μM mixed FA for indicated time points. n = 4.

D: Cell toxicity in HEK293 after stimulation with control BSA or different concentrations of mixed FA for 3h. n = 2.

All data are mean ± SEM. For statistical considerations, a 2-way ANOVA with Tukey posthoc (B-D) was performed. *** = p < 0.0005, ns = p > 0.05.

2. Assessment of FAO by ¹³C stable isotope resolved metabolomics

To investigate how imported FAs are further metabolized by cultured cells, ¹³C stable isotope resolved metabolomics (SIRM) was performed in HEK293 cells. Stable isotope labeling of the TCA intermediates is commonly conducted with ¹³C-glucose or ¹³C-glutamine in cultured cells (24), but not with FA tracers. FA incorporation into the TCA cycle gives rise to M+2 isotopologues in the first spin of the TCA cycle (Fig. 2A), which increases in subsequent spins by two (M+4, M+6). By increasing fatty acid concentrations in the culture media, we were able to dose-dependently increase M+2 ¹³C-enrichment of citrate (Fig. 2B). Minimal enrichment was observed at 5 μM FA, which grew exponentially and plateaued at 100 μM (Fig. 2B). As previously demonstrated for glucose, ¹³C FA incorporation into succinate was relatively low compared with other TCA cycle intermediates, suggesting that only a minor pool of succinate is used for fumarate production (25). Increasing FA concentrations did not perturb pool size of TCA cycle intermediates, suggesting that this intervention does not change the balance of production and consumption of individual TCA cycle metabolites (Suppl. Fig. 1A). Thus, a 100 μM mixed FA dose was employed for all subsequent experiments.

Figure 2. ¹³C labeling traces mitochondrial beta oxidation in proliferating cells.

A: FA tracing principle depicting incorporation of 13C-labeled carbons (dark circle) into TCA cycle intermediates. Only the first spin of the TCA cycle is depicted.

B: Dose-dependency of FA enrichment into TCA cycle intermediates (3h). Concentrations are 5, 10, 25, 50, 100, 200, 400, 600 μM. n= 3.

C: Schematic of mitochondrial beta oxidation and highlighting of site of action of etoxomir (eto), ACC2i and of enzymatic steps performed by trifunctional protein (HADHA and HAHB).

D: Fraction of M+2 TCA cycle intermediates in HEK293 cells with etomoxir (5 μM). n=3.

E: Fraction of M+2 TCA cycle intermediates in HEK293 cells with ACC2i (10 μ M). n = 3

F: Quantification of long chain acylcarntines in HEK293 cells with etomoxir. n=4.

G: Quantification of long chain acylcarntines in HEK293 cells with ACC2i. n=4.

H: Fraction of M+2 citrate in primary patient fibroblasts with 13C FA. n=5.

All data are mean ± SEM. For statistical considerations, a two sided t-test (D-G) or 1-way ANOVA with Tukey posthoc (H) was performed. * = p , 0.05, ** = p < 0.005, *** = p < 0.0005, **** p < 0.0001.

To test whether ¹³C-enrichment of citrate specifically reports FAO in the cell, we manipulated mitochondrial FAO with pharmacological and genetic approaches. First, cells were treated with etomoxir (5μM), an inhibitor of carnitine palmitoyltransferase 1/2 (CPT1/2), which mediates FA uptake into the mitochondrial compartment (Fig. 2C). Etomoxir markedly reduced ¹³C- enrichment into citrate (Fig. 2D, Suppl. Fig. 1B). Conversely, activation of CPT1/2 by inhibiting acetyl-CoA carboxylase (ACC2) using CD-017-0191 led to increased ¹³C- enrichment of citrate (Fig. 2E, Suppl. Fig. 1C). Furthermore, changes of ¹³C-enrichment of citrate under these conditions were closely mirrored in changes of long-chain acyl-carnitines (AC) levels, which are clinically used as the proxy for mitochondrial β-oxidation (Fig. 2F, G) (26). The close correlation among mitochondrial FA uptake, acyl-carnitine levels, and ¹³C-enrichment of citrate suggested that ¹³C-citrate is a reliable indicator of mitochondrial β-oxidation.

We also tested the sensitivity of this approach in fibroblasts, a cell type with lower mitochondrial oxidative capacity. We used human primary fibroblasts (control) and two lines of patient fibroblasts carrying mutations in trifunctional protein subunits HADHA and HADHB which catalyze multiple steps in FA β-oxidation (Suppl. Fig. 2A) (26, 27). The mutant primary fibroblasts did not demonstrate compensatory alterations in metabolite abundance of TCA cycle intermediates or protein abundance in HADHA, HADHB or ETC protein subunits (Suppl. Fig. 2B, C). The ¹³C-labelling of citrate was robust in control fibroblasts but was substantially reduced in both patient cell lines compared to controls (Fig. 2H, Suppl. Fig. 2D). Together, these data demonstrate that M+2 enrichment of citrate by ¹³C-FA is specific and sensitive reporter for mitochondrial β-oxidation.

3. A significant fraction of carbons derived from FAO enters alternative TCA cycle in proliferating cells.

We observed a significant disconnect between the M+2 ¹³C-enrichment in citrate compared to the enrichment of TCA intermediates downstream of citrate in HEK293 cells, the primary fibroblasts and several other cancer cell lines (Fig. 3A, B, Suppl Fig. 3A), which suggests a partial loss of FA-derived carbons downstream of citrate from mitochondria. A lower ¹³C-enrichment of αKG than citrate could be due to dilution by unlabeled carbon from glutamate via glutaminolysis (Fig. 3A). To test this possibility, we treated cells with an inhibitor for GLS (glutaminase) (CB-839, GLSi) to prevent glutaminolysis. GLSi had a moderate effect on ¹³C-citrate enrichment in the tested cell lines, but markedly increased ¹³C-enrichment of downstream metabolites in both HEK293 and Hela cells (Fig. 3C, D). Furthermore, GLSi decreased total amount of metabolites in the second half of the TCA cycle (Fig. 3E, F) indicating a significant contribution of glutamine in maintaining the TCA cycle pools. However, the lower ¹³C-enrichment in αKG vs. citrate remained prominent after inhibition of glutaminolysis, suggesting that label dilution via glutamine anaplerosis could not accounted for majority of ¹³C lost in the TCA cycle (Fig. 3C, D).

Figure 3. FAO feeds into two TCA cycle pathways.

A: Schematic depiction of TCA cycle fluxes and alternative pathways/ substrates and their contribution to 13C labeling patterns of indicated TCA cycle intermediates.

B: Fraction of M+2 TCA cycle intermediates in indicated cell lines. n = 3

C+D: Fraction of M+2 TCA cycle intermediates in HEK293 cells (C) or Hela cells (D) with GLSi (CB-839, 1 μM). . n = 3-5.

E+F: Heatmap of TCA cycle intermediates with GLSi in HEK293 (E) and Hela (F) cells. n = 3-4.

G: Fractional enrichment of FA-derived M+2 malate relative to M+2 citrate in indicated cell lines. = 3.

H: Fractional enrichment of FA-derived M+2 malate relative to M+2 citrate in HEK293 and Hela cells with GLSi. n = 3.

All data are mean ± SEM. For statistical considerations, a 1-way ANOVA (H) or 2-way ANOVA (B-D) with Tukey posthoc was performed. * = p , 0.05, ** = p < 0.005, *** = p < 0.0005, **** p < 0.0001.

Alternatively, citrate can exit mitochondria and be converted by ATP citrate lyase (ACLY) into unlabeled oxalacetate and subsequently malate in the cytosol (10, 11). Malate can re-enter at the second half of the TCA cycle (Fig. 3A) (10, 11). The degree to which malate is derived from the alternative pathway can be represented as the ratio of M+2 ¹³C-malate relative to M+2 ¹³C-citrate (M+2 Mal/ Cit). This non-canonical TCA flux has recently been experimentally validated by ¹³C-glucose tracing in cells cultured in the absence of FA (8, 28). Results here showed that the non-canonical TCA flux was not unique to glucose metabolism. Loss of FA-derived carbons from the canonical TCA cycle was evident in all proliferating cell lines, with a M+2 Mal/Cit ratio of 0.4 or lower (Fig. 3G). GLSi also increased the M+2 Mal/ Cit ratio in HEK293 and Hela cells, indicating that glutaminolysis, in part, affects the contribution of FA-derived carbons for TCA cycling (Fig. 3H).

4. Regulation of TCA cycle choice by substrate availability.

To further investigate mechanisms regulating the FA-carbon partitioning in canonical vs. non-canonical TCA cycle, we manipulated FAO rate in HEK293 cells. Increasing FA concentrations led to a dose-dependent increase in the ¹³C-enrichment of TCA cycle intermediates and acyl-carnitines but no change of M+2 Mal/ Cit ratio (Fig. 4A, Suppl. Fig. 4A, B). Moreover, etomoxir and ACC2i led to opposite changes of ¹³C-citrate enrichment as expected but did not change M+2 Mal/ Cit ratio (Suppl. Fig. 4C-E). Thus, increasing the input of FA-carbon did not shift their partition between the two TCA pathways.

Figure 4: Effects of substrate availability on FA-driven TCA cycle flux.

A: Fractional enrichment of FA-derived M+2 malate relative to M+2 citrate in HEK293 cells cultured with increasing FA concentration. n = 3.

B: Fractional enrichment of FA-derived M+2 malate relative to M+2 citrate in HEK293 cells after glucose removal. n = 3.

C: Fractional enrichment of FA-derived M+2 malate relative to M+2 citrate in HEK293 cells after glucose removal. n = 3.

All data are mean ± SEM. For statistical considerations, a two sided t-test (C) or 2-way ANOVA (B) with Tukey posthoc was performed. ** = p < 0.005, *** = p < 0.0005.

To investigate the involvement of glucose as a competing substrate in regulating the fate of FA-derived carbon in TCA cycle, we removed glucose from the culture medium. This resulted in a moderate increase in ¹³C-enrichment of TCA cycle intermediates by ¹³C-FA (Fig. 4B). In contrast to the results of manipulating FA input described above, glucose removal significantly increased the M+2 Mal/ Cit ratio for FA-derived carbon (Fig. 4C).

These results suggest that TCA cycle choice for FA derived carbon is not determined by the amount of acetyl-CoA generated by β-oxidation but could be affected by the competition of other substrates. When multiple substrates are present, oxidation of FA-derived acetyl-CoA in canonical TCA cycle can be suppressed by oxidation of glucose. The presence of competing substrates drives FA derived carbon into the non-canonical pathway. This is different from cells cultured in glucose media, where glucose oxidation promotes canonical TCA flux. Under high glucose conditions, stimulating glucose oxidation in the mitochondria reduces non-canonical TCA cycle flux, and inhibition of mitochondrial pyruvate uptake increases non-canonical TCA flux (8).

5. Assessment of FAO by oxygen consumption rate vs. by ¹³C-FA tracing

FAO has also been assessed by measuring the oxygen consumption rate (OCR) of cells exposed to FA in the presence or absence of etomoxir (29). The etomoxir sensitive portion of the OCR is used as an indicator of FAO. This assay measures mitochondrial respiration using NADH or FADH₂ produced by β-oxidation and TCA cycle (Fig. 5A). Since other carbon substrates can also generate NADH or FADH₂, the results will be affected by the presence of other substrates during the assay. To determine the influence of other substrates on etomoxir sensitive OCR, we subjected HEK293 cells to the same substrates as in ¹³C-FA labelling experiments (Condition A in Fig. 5B) or conditions of either glucose starvation or glucose and glutamine starvation (Condition B and C in Fig. 5B). Basal respiration was strongly reduced when cells were starved of glucose and/ or glutamine (Fig. 5C). Etomoxir did not change OCR when cells were exposed to all substrates (Fig. 5C) although the ¹³C-Citrate enrichment by ¹³C-FA could be inhibited by ~90% with the same dose of Etomoxir (Fig. 1D). The results indicated that oxidation of glucose and glutamine can fully support mitochondrial respiration of these cells when FAO is inhibited. In contrast, we found that etomoxir could suppress OCR in HEK293 cells only under conditions of combined glucose and glutamine starvation (Fig. 5C), a similar condition as reported in the literature.

Taken together, OCR could be used as an indirect assessment of FAO when very limited or no other substrates were provided to the cells. Compared to the fractional M+2 ¹³C-enrichment of citrate, OCR measurement has the advantage of reporting a metabolic rate. However, it requires removal of other substrates which inevitably disturbs normal cellular metabolism. Therefore, etomoxir sensitive OCR can be used to assess the ability of the cells to oxidize FA but does not measure FAO under normal culture conditions. Moreover, OCR does not account for the fraction of carbon fluxed through the non-canonical TCA pathway, thus, not a true rate of FA β-oxidation.

6. Oxidative energy production drives the coupling of FAO with canonical TCA cycle in muscle cells.

FAO is a major energy source for muscle cells. To determine whether oxidative ATP production drives FAO-derived carbons into canonical TCA cycle, we compared proliferating C2C12 myoblasts that have higher demand of biosynthesis with differentiated C2C12 myotubes that have greater oxidative energy metabolism (8, 30-32). Indeed, differentiated myotubes demonstrate a 2-fold higher M+2 Mal/ Cit ratio than myoblasts (Fig. 6A, Suppl. Fig. 5A).

Figure 6: Coupling of FAO with canonical TCA cycle flux is increased in differentiated muscle cells.

A: Fractional enrichment of FA-derived M+2 malate relative to M+2 citrate in myoblasts and myotubes. n = 3.

B: Fraction of M+2 TCA cycle intermediates in indicated cardiac cells. n = 2-5.

C: RNA-seq analysis of indicated genes in myoblasts (blasts) and myotubes (tubes) that were differentiated for 5 days. Levels are represented as the log2-transformed fold change relative to the row mean. n = 3 independently derived samples. From Arnold et al.

D: Representative western blot for ETC subunit protein expression in myoblasts and myotubes. n = 2.

E: Representative western blot for mt-ND1 expression in indicated cardiac cells. n = 2.

F: Correlation between M+2 Mal/ Cit and mt-ND1 expression. iPS-CM n = 4, H9C2_diff = 7, ACM = 6.

All data are mean ± SEM. For statistical considerations, a two sided t-test (A), 1-way ANOVA (D) with Tukey posthoc or simple linear regression analysis was performed. * = p < 0.05, **** p < 0.0001.

In a recent study, we showed that primary adult cardiomyocytes (ACM) had a M+2 Mal/ Cit of ~ 1 for glucose-derived carbon when cultured with mixed substrates (14). Here we found that under similar culture condition, these cells demonstrated a comparable M+2 Mal/ Cit of ~0.9 for FA derived carbon (Fig. 6B). Thus, regardless of carbon source the alternative TCA cycle flux is minimal in terminally differentiated and highly oxidative adult cardiomyocytes. In contrast, other cardiac cell lines such as H9C2 cells (derived from embryonic rat heart) and induced pluripotent stem cell derived cardiomyocytes (iPS-CM) demonstrated much lower coupling of citrate and the canonical TCA cycle (Fig. 6B, Suppl. Fig. 5B).

In order to understand the molecular basis of these changes, we explored previous published RNA sequencing datasets (8). We found that the increase of M+2 Mal/Cit ratio was positively correlated with upregulation of gene and protein expressions for mitochondrial electron transport chain enzymes in the C2C12 myoblasts and myotubes (Fig. 6C, D). Similarly, M+2 Mal/Cit ratio in the cardiac cells was positively correlated with the Mt-ND1 levels, a complex I subunit component (Fig. 6E, F).

Together, these data show that the coupling between citrate and canonical TCA cycle is positively correlated with oxidative metabolism for energy production. Furthermore, although all cardiac cell lines can oxidize FA, the metabolite fate of citrate in those cells resembles proliferating cells more than adult primary cardiomyocytes.

Discussion

In this study, we have developed a strategy to culture cells with increasing concentrations of long chain FA without altering cell viability. This method allows us to trace FA β-oxidation and subsequent TCA cycle metabolism in a variety of cell types using ¹³C-labelled FAs. Our results show that a significant portion of carbon derived from FA β-oxidation exit mitochondria as citrate in proliferating cells. Citrate produces acetyl-CoA in the cytosol and the remaining carbons are returned to mitochondria as malate through the so called non-canonical TCA pathway. The partition of FAO flux between the cytosol and mitochondria changes with oxidative energy production in cells during differentiation. The non-canonical TCA pathway is substantial in glycolytic and proliferating cell lines but is markedly reduced in differentiated and oxidative myotubes. It becomes minimal in adult cardiomyocytes in which more than 90% ATP is produced through oxidative phosphorylation.

Compared to glucose metabolism, knowledge of FAO in cultured cells is limited partly due to technical challenges. The majority of cell culture studies do not include FA and carnitine in culture medium. Oxidation of FA in fetal serum or from endogenous lipids cannot be distinguished from the oxidation of glucose or glutamine and is therefore poorly understood. Recently, a method has been developed to assess FAO by measuring oxygen consumption rate (OCR) of cells with and without etomoxir, an inhibitor of long-chain FA entry into mitochondria (29). As shown in our results, this method requires removal of non-FA carbon substrates thus cannot measure FAO under normal culture conditions. Furthermore, the alternative TCA pathway contributes significantly less to the OCR than the canonical pathway thus is underrepresented in the OCR measurement. As a caution, etomoxir at higher dose (>5μM), which is commonly used in the assay, has been shown to inhibit OCR through off-target effects which could generate mis-leading results (33).

Using ¹³C-FA tracing, we show that mitochondrial β-oxidation is robust in all cell types when provided with multiple carbon substrates. Instead of being used for oxidative ATP production, a substantial portion of FA-derived carbons are routed through the cytosol via the citrate-malate shuttle in proliferating cells.

We also find that oxidation of glucose competes with FA carbon thus promoting FA carbon flowing to the non-canonical TCA pathway. This is corroborated by the prior finding that stimulating glucose oxidation leads to a great canonical TCA cycle flux in cancer cells (8). Although a significant portion of glucose-derived carbon flows through the non-canonical TCA cycle in cancer cells cultured in the absence of FA (8) ¹³C-glucose tracing in these cells growing in vivo find no loss of ¹³C-label between citrate and malate (34, 35). These observations suggest that the metabolic fate of glucose derived carbon may be altered by the presence of FA in vivo. It raises the possibility that FA derived carbon is a major supplier of acetyl-CoA in the cytosol of cancer cells in vivo. Future studies comparing ¹³C-FA and ¹³C-glucose tracing in vivo are thus warranted.

One caution is that glutamine can lead to both label dilution of FA-derived carbons in the TCA cycle and change the fate of citrate, but our fractional analysis cannot distinguish between these two possibilities.

The partitioning of glucose carbon between canonical and alternative TCA cycle has been shown dependent on cell state. Changes in the gene expression for enzymes involved in canonical vs. non-canonical pathways at different cell states closely correlates with the carbon partition between the two pathways (8). Moreover, low expression of ACO2 and IDH2 in iPS cells limits carbon flux from citrate to α-KG (canonical TCA) and favors citrate exiting the mitochondria (alternative TCA). The bottleneck is eliminated by increased expression of these enzymes as the iPS cells differentiate into cardiomyocytes (36). In the present study, we show that the partition of FA derived carbon between canonical and alternative TCA pathway is correlated with oxidative ATP production. In terminally differentiated adult cardiomyocytes, over 90% of FA derived carbon enters the canonical TCA cycle. However, this characteristic is not captured by any of the cardiac cell lines tested in the study including iPS cell derived cardiomyocytes. Our data raises the possibility that increases in oxidative energy metabolism shifts the TCA cycle choice towards the canonical pathway during cell differentiation and maturation. However, future studies are required to determine whether oxidative ATP production is a driver for canonical TCA cycle flux.

In summary, the study presents a method to trace substrate metabolism in cells cultured in FA enriched medium. Our results show distinct fates for FA derived carbon in proliferating and non-proliferating cells. Furthermore, this approach enables future studies of cell metabolism in a more physiologically relevant environment.

Supplementary Material

Supplemental Figs. S1-S5: https://doi.org/10.6084/m9.figshare.27852165.v1.

figshare

Supplemental Figures Ritterhoff et al FA-driven TCA cycle flux in proliferative cells

Supplemental Material for Article “13C Stable Isotope Tracing Reveals Distinct Fatty Acid Oxidation Pathways in Proliferative vs. Oxidative Cells”

Abbreviations

ACLY

ATP-citrate lyase

CPT 1/2

carnitine palmitoyltransferase 1/2

ETC

electron transport chain

FAO

fatty acid oxidation

GLS

glutaminase

OCR

oxygen consumption rate

TCA

tricarboxylic acid