Standardized bioenergetic profiling of adult mouse cardiomyocytes

Abstract

Mitochondria are at the crux of life and death and as such have become ideal targets of intervention in cardiovascular disease. Generally, current methods to measure mitochondrial dysfunction rely on working with the isolated organelle and fail to incorporate mitochondrial function in a cellular context. Extracellular flux methodology has been particularly advantageous in this respect; however, certain primary cell types, such as adult cardiac myocytes, have been difficult to standardize with this technology. Here, we describe methods for using extracellular flux (XF) analysis to measure mitochondrial bioenergetics in isolated, intact, adult mouse cardiomyocytes (ACMs). Following isolation, ACMs were seeded overnight onto laminin-coated (20 μg/ml) microplates, which resulted in high attachment efficiency. After establishing seeding density, we found that a commonly used assay medium (containing a supraphysiological concentration of pyruvate at 1 mmol/l) produced a maximal bioenergetic response. After performing a pyruvate dose-response, we determined that pyruvate titrated to 0.1 mmol/l was optimal for examining alternative substrate oxidation. Methods for measuring fatty acid oxidation were established. These methods lay the framework using XF analysis to profile metabolism of ACMs and will likely augment our ability to understand mitochondrial dysfunction in heart failure and acute myocardial ischemia. This platform could easily be extended to models of diabetes or other metabolic defects.

the high cardiac output of the mammalian heart creates tremendous energetic demand, which is met largely through mitochondrial metabolism. The adult heart relies predominately on fatty acid oxidation (FAO) for energy production; however, under conditions that increase the demand for ATP or during periods of fatty acid starvation, the heart is capable of utilizing other substrates (i.e., glucose) (27). During heart failure, cardiac metabolism reverts to a fetal metabolic pattern of increased reliance on glucose, which may eventually lead to insulin resistance and metabolic failure; this area remains contested.

To evaluate in vivo cardiac metabolism, several methods have been employed in the last few decades. ³¹P NMR provides a ratio of phosphocreatine to ATP (8). A significant limitation of ³¹P NMR is the low signal-to-noise ratio. Another method to evaluate cardiac metabolism is the ex vivo perfusion of the intact heart with radioisotopes (23). Advantages of this technique include high sensitivity and the ability to control determinants of cardiac function (i.e., afterload, perfusate). The relatively low-throughput nature, requirement for radioactivity, and the unknown cell type(s) responsible for changes in metabolism limit the usefulness of this technique.

Classically, mitochondrial function has been determined by using isolated organelles and measuring maximal respiration rates by polarographic methods (4, 18, 22) or by measuring activity of the individual complexes (19, 22). Although such techniques represent powerful reductionist approaches for assessing and making assumptions about mitochondrial “health,” they cannot account for extramitochondrial regulation of metabolism (i.e., substrate uptake). Therefore, studying bioenergetics in intact cells may provide a more physiological view of mitochondrial function. Recently, bioenergetic profiling of intact cells has been described in multiple cell lines (1, 9, 14, 16) and primary cells (6, 7, 10, 11, 20, 21, 28) using a microplate-based system that measures extracellular flux (XF). This automated system measures the two major cellular energy-producing pathways, glycolysis and mitochondrial respiration. XF measures the extracellular acidification rate (ECAR) of the assay medium (lactate, a byproduct of glycolysis, is the major contributor of protons) as a surrogate of glycolytic flux. A second measurement, the oxygen consumption rate (OCR), represents mitochondrial respiration. XF technology uses fluorophores that transmit a fluorescent signal to sensitive photodetectors to measure specific analytes (H⁺ or O₂) present in the assay medium. Continuous real-time measurement of pH and oxygen concentration during XF assays allows for the determination of ECAR and OCR.

There are several considerations for developing XF assays including cell attachment, protocol timing, seeding density, and substrate selection. Because a prerequisite for accurate measurements with XF analysis is the ability of cells to adhere to the microplate, the success of investigators to study nonadherent cells with XF has been limited (i.e., isolated adult cardiomyocytes, or ACMs). In the present experiments, we overcame this limitation by optimizing an attachment protocol for ACMs to XF microplates. We then profiled bioenergetics of ACMs with XF analysis using various substrates under basal conditions. After titrating the pyruvate concentration, we performed a dose-response for palmitate. The methods developed here will undoubtedly aid in our understanding of how mitochondrial function in ACMs is altered in genetic or surgical models of cardiac dysfunction and diabetes.

METHODS AND MATERIALS

ACM isolation and attachment.

Primary ACMs were isolated from 8 to 12 wk old male, wild-type C57BL/6 mice (n = 15) by the Langendorff perfusion method as previously described (12, 13, 17, 29). In brief, mice were anesthetized with 5% isoflurane, and the heart was rapidly excised and placed in ice-cold PBS for 2 min. Next, the aorta was cannulated with a blunted 23-gauge needle and secured with a silk suture under a dissecting microscope. The cannulated heart was then attached to the Langendorff apparatus and perfused with Tyrode's buffer (18 mmol/l sodium bicarbonate, 126 mmol/l sodium chloride, 4.4 mmol/l potassium chloride, 1 mmol/l magnesium chloride, 4 mmol/l HEPES, 11 mmol/l glucose, 10 mmol/l 2,3-butanedione monoxime, and 30 mmol/l taurine) at 37°C for 3–5 min. After an air bubble formed in the perfusion tubing, collagenase solution [Tyrode's + 0.1% bovine serum albumin (BSA) + 0.025 mmol/l calcium chloride + 0.1% type II collagenase (Worthington, CLS-2)] was added to the top reservoir. Importantly, formation of an air bubble prevented dilution of the collagenase solution and allowed for accurate perfusion timing of the enzyme solution. The heart was perfused with collagenase solution for 10–13 min. After digestion, atria were carefully dissected from the heart and discarded. Next, the left and right ventricles were gently pulled apart with fine forceps, triturated with a 10 ml pipette at slow speed, and filtered through a 140 μm nylon net filter (Millipore, NY4H). Lastly, cells were gravity-sedimented through five calcium gradients in a 15 ml conical tube at room temperature (0.05 mmol/l CaCl₂, 0.075 mmol/l CaCl₂, 0.125 mmol/l CaCl₂, 0.275 mmol/l CaCl₂, 0.525 mmol/l CaCl₂). After 15 min of sedimentation, the calcium solution was aspirated, and the pelleted cells were gently resuspended in the subsequent calcium gradient. This method produced highly viable (∼85–90%) rod-shaped cells (6–10 × 10⁵ cells/heart). Prior to plating, Seahorse V7 tissue culture plates were coated with 20 μg/ml laminin (Invitrogen mouse laminin, 23017) for 2 h at 37°C. For this, 40 μl of stock laminin (1 mg/ml) was diluted in 1.96 ml of PBS to yield a final concentration of 20 μg/ml laminin. Each well in the V7 tissue culture plate was coated with 75 μl of the diluted laminin solution. Immediately before plating, the laminin solution was aspirated from the well. Following isolation, cells were seeded overnight onto laminin-coated Seahorse V7 tissue culture plates in plating media [MEM media (Invitrogen, 11575032) supplemented with 1 μg/ml insulin, 0.55 μg/ml transferrin, 0.5 ng/ml selenium (GIBCO, 41400), 10 mmol/L HEPES (Sigma, H4034), 10 mmol/l 2,3-butanedione monoxime (BDM) (Sigma, B0753), 0.2% BSA (Sigma, A3059), and 5% fetal calf serum (Hyclone, SH30070)] at 37°C in the presence of 5% CO₂ in a humidified incubator. For seeding density experiments 6,000–25,000 cells were seeded per well. For the remaining experiments, 16,000 cells were seeded in each well. Following cell counting, cells were constituted such that the desired cell number per well was suspended in 75 μl of plating media. As such, 75 μl of the cell suspension was added to each well. Each microplate contained four empty wells that served as temperature controls.

XF24 bioenergetic profiling.

The bioenergetic response of ACMs was measured with the Seahorse Bioscience XF24 Flux Analyzer (5, 10, 20, 21). For seeding density experiments, the plating media were changed to 675 μl unbuffered DMEM supplemented with 4 mmol/l glutamine and 1 mmol/l pyruvate 1 h before assay. For FAO experiments, the medium was changed to 675 μl unbuffered DMEM supplemented with 4 mmol/l glutamine, 5 mmol/l glucose, and 0.1 mmol/l pyruvate 1 h prior to assay. Assay medium for FAO experiments also contained the corresponding concentration of BSA-conjugated palmitate and 0.2 mmol/l l-carnitine. The XF24 automated protocol consisted of 10 min delay following microplate insertion, baseline OCR/ECAR measurements [3 × (1.5 min mix, 2 min wait, 1.5 min measure)], followed by injection of port A (75 μl) and OCR/ECAR measurement [2 × (1.5 min mix, 2 min wait, 1.5 min measure)], injection of port B (83.3 μl) and OCR/ECAR measurement [2 × (1.5 min mix, 2 min wait, 1.5 min measure)], injection of port C (92.6 μl) and OCR/ECAR measurement [2 × (1.5 min mix, 2 min wait, 1.5 min measure)], and (if applicable) injection of port D (100 μl) and OCR/ECAR measurement [2 × (1.5 min mix, 2 min wait, 1.5 min measure)]. This modified protocol is important to avoid hypoxic conditions encountered with the standard XF measurement protocol. Optimal concentrations of oligomycin (Sigma, 75351), carbonyl cyanide-p-trifluoromethoxyphenylhydrazone (FCCP; Sigma, C2920), and antimycin A (Sigma, A6874) were diluted in DMSO (Sigma, 154938).

Conjugation of palmitate to BSA.

Sodium palmitate (1 mmol/l) (Sigma, P9767) was conjugated to ultrafatty acid-free BSA (0.17 mmol/l) (Roche, 0311740501) in a 6:1 molar ratio. Briefly, 2.267 g of BSA was dissolved in 100 ml of 150 mmol/l sodium chloride at 37°C while being stirred. After BSA was completely dissolved, the BSA was filtered with a 150 ml filter unit (22 μm). We diluted 50 ml of the BSA solution in 50 ml of 150 mmol/l sodium chloride, then aliquoted it into 4 ml vials, and stored it at −20°C. Next, 30.6 mg of sodium palmitate was dissolved in 44 ml of 150 mmol/l sodium chloride and heated to 70°C. Following heating, 40 ml of the palmitate solution was added to 50 ml of the BSA solution and heated for 1 h at 37°C (while being stirred). After 1 h the volume of the BSA-palmitate solution was adjusted to 100 ml. The pH of the BSA-palmitate solution was adjusted to 7.4 with NaOH. Lastly, the BSA-palmitate solution was aliquoted into 4 ml glass vials and stored at −20°C. This resulted in a 1 mmol/l sodium palmitate / 0.17 mmol/l BSA solution. Prior to use the BSA and BSA-palmitate was thawed for 10 min at 37°C.

Statistical analysis.

Results are shown as means ± SE. The statistical analysis (GraphPad 5.0) was conducted by Student's t-test or by one-way ANOVA followed by Newman-Keuls multiple comparison test, as appropriate. Each n represents a separate isolation. Differences were considered statistically significant if P < 0.05.

RESULTS

Seeding density optimization.

Cell densities of 6,000–25,000 ACMs were seeded overnight to determine the optimal seeding density for ACMs (isolated from left and right ventricles) in laminin-coated XF microplates. We determined the optimal attachment conditions by evaluating the efficiency of varying concentrations of laminin (0–40 μg/ml) on ACM attachment. ACMs did display characteristic rod-shape morphology immediately before assay (Fig. 1A). Because the OCR values were linear between 12,000 and 20,000 cells, we decided to use 16,000 cells/well (i.e., subsaturating density) for the remainder of our experiments (Fig. 1B). Building upon traditional bioenergetic experiments, we automatically injected oligomycin, FCCP, and antimycin A into wells containing isolated ACMs (Fig. 1C). Blockade of the ATP synthase by oligomycin (inducing state IV₀ respiration) allows for the determination of respiration due to ATP synthesis and proton leak. FCCP addition uncouples the transport of electrons from ATP production and results in maximal respiration. Injection of antimycin A inhibits the transfer of electrons from complex III to cytochrome C, thus blocking electrons from reducing oxygen at complex IV, which allows for the determination of nonmitochondrial oxygen consumption. Following three baseline measurements, we sequentially injected 50 mmol/l glucose, 10 μg/ml oligomycin, 0.01 mmol/l FCCP, and 0.1 mmol/l antimycin A through ports A–D, respectively, to yield final concentrations of 5 mmol/l glucose, 1 μg/ml oligomycin, 0.001 mmol/l FCCP, and 0.01 mmol/l antimycin A. Importantly, 25,000 cells resulted in a decrease in sensitivity of the instrument to measure OCR/ECAR (Fig. 1, D and E). Oxygen tension for 12,000 and 20,000 cells indicates that the XF system is capable of replenishing oxygen concentration following FCCP measurements. It is important to note that XF protocols (i.e., mixing durations, measurement durations) may need to be modified to prevent hypoxic conditions.

Fig. 1.

Standardization of adult cardiomyocyte (ACM) seeding density for extracellular flux (XF) analysis. ACMs were seeded overnight onto laminin-coated Seahorse V7 tissue culture plates (6,000–25,000 cells/well). A: representative photomicrograph of ACMs immediately prior to assay. B: basal oxygen consumption rate (OCR, state III respiration) plotted as a function of cell number. C: idealized metabolic profiling trace demonstrating basal OCR, proton leak, maximum OCR, and nonmitochondrial oxygen consumption. OCR (D) and extracellular acidification rate (ECAR, E) of ACMs seeded at different densities. Data points represent group means ± SE, n = 3–5/group.

Effects of substrates on mitochondrial function.

To determine the effects of various substrates on ACM bioenergetics, we profiled ACMs in DMEM assay medium with 4 mmol/l glutamine containing either 1 mmol/l pyruvate or 5 mmol/l glucose or 1 mmol/l pyruvate + 5 mmol/l glucose. In the presence of pyruvate, ACMs displayed similar bioenergetic profiles, regardless of the presence of glucose (Fig. 2A). This may indicate that when excess levels of pyruvate are present, isolated ACMs rely almost entirely on pyruvate oxidation for energy production; however, when glucose alone was used, cells experienced bioenergetic collapse. Because ACMs express predominately GLUT4 glucose transporters (2) it is likely that isolated ACMs (in the absence of insulin) are largely nonresponsive to glucose. In contrast to the reported response of neonatal rat cardiomyocytes (20), oligomycin addition did not decrease the OCR in ACMs (as would be expected by inhibition of the ATP synthase). This may be explained by decreased ATP demand of noncontracting ACMs and a relatively high rate of proton leak (Fig. 2B). To ensure that the proper dose of oligomycin was being utilized in these experiments, we performed an oligomycin dose-response (1–10 μg/ml oligomycin), which revealed that all assayed concentrations of oligomycin produced similar inhibition of the ATP synthase (data not shown); however, higher doses of oligomycin tended to decrease the maximal OCR. In cells assayed with pyruvate or pyruvate + glucose there was an increase in ECAR following oligomycin and FCCP addition (Fig. 2C). This represents a compensatory action of cells to increase ATP production by glycolysis when oxidative phosphorylation is inhibited/altered.

Fig. 2.

Effects of substrate on ACM bioenergetics. A: XF analysis of ACM in assay media containing pyruvate (1 mmol/l), glucose (5 mmol/l), or glucose (5 mmol/l) + pyruvate (1 mmol/l). B: quantification of basal OCR, proton leak (oligomycin OCR − antimycin OCR), maximal rate (FCCP OCR − antimycin OCR), and nonmitochondrial OCR (antimycin OCR). C: effect of substrate on ECAR. D: quantification of basal ECAR, oligomycin ECAR, and maximal ECAR. Data points represent group means ± SE, n = 3–4/group.

Pyruvate titration.

We observed that ACMs assayed in medium containing 1 mmol/l pyruvate displayed a maximal bioenergetic response, which was not altered by fatty acid and/or glucose content. These observations warranted a pyruvate dose-response experiment so that the concentration of pyruvate could be titrated such that the cellular reliance on other substrates (i.e., glucose or fatty acid) would be reflective of the in vivo setting. We assayed ACMs in DMEM assay medium with 4 mmol/l glutamine containing 0.01, 0.1, or 1 mmol/l pyruvate (Fig. 3A). We found that both 1 and 0.1 mmol/l pyruvate resulted in favorable bioenergetic profiles, with 0.01 mmol/l pyruvate significantly reducing the maximal OCR compared with 1.0 mmol/l pyruvate (Fig. 3B). Withdrawal of pyruvate resulted in bioenergetic collapse.

Fig. 3.

Pyruvate dose-response for ACM XF analysis. A: metabolic profiling of ACMs assayed in media containing various concentrations of pyruvate (0.01–1 mmol/l) and 4 mmol/l glutamine. B: quantification of maximal OCR (% baseline). Data points represent group means ± SE, n = 3/group. *P < 0.05 vs. 1.0 mmol/l pyruvate.

ACM bioenergetic profiling with palmitate.

ACMs were profiled in the presence of various concentrations of BSA-conjugated palmitate + 0.1 mmol/l pyruvate (Fig. 4A). ACMs assayed in BSA + 0.1 mmol/l pyruvate served as controls. Basal OCRs were higher in the presence of physiologically relevant levels of palmitate compared with palmitate-free (Fig. 4B). Maximal rates in groups containing palmitate were significantly increased (∼2-fold) compared with the BSA control group (Fig. 4C).

Fig. 4.

Palmitate dose-response for ACM metabolic profiling. A: XF analysis of ACM in various concentrations of palmitate. Effects of palmitate concentration on basal OCR (B) and maximal OCR (C). Data points represent group means ± SE, n = 3/group. *P < 0.05 vs. BSA.

DISCUSSION

In this study, we standardized the XF methodology for bioenergetic profiling of isolated intact-ACMs utilizing a high-throughput XF analyzer. Here, we have overcome the major obstacle inhibiting the use of ACMs in the XF24 by optimizing an attachment protocol using laminin (20 μg/ml) that results in high ACM attachment efficiency. Furthermore, optimal assay conditions for measuring palmitate oxidation were established.

The first step for developing an XF experiment is to determine empirically the optimal number of cells to plate in each well. Too many cells will result in a decrease in sensitivity; likewise, too few cells will not result in sufficient OCR/ECAR values. ACMs display a linear basal OCR response with respect to cell number between 12,000 and 20,000 cells. In contrast to spontaneously beating isolated neonatal rat cardiomyocytes (NRCMs), isolated ACMs appear to be largely oligomycin-insensitive. This indicates that proton leak is a major oxygen- and substrate-consuming process in isolated, nonbeating ACMs and they, once isolated, remain in a state 4-like state (21).

To date, several studies have evaluated the effects of alternative substrates on NRCM bioenergetic function. NRCMs display maximal reserve capacity (difference between basal OCR and uncoupled OCR) when pyruvate alone (1 mmol/l) is present (5, 20). NRCMs with glucose (5 mmol/l) + pyruvate had a lower reserve capacity than cells respiring on pyruvate alone. Cells respiring on glucose alone displayed a threefold lower reserve capacity than cells respiring on pyruvate as the sole substrate. In the current study, ACMs had similar maximal OCRs in the presence of pyruvate alone or pyruvate + glucose. ACMs respiring on glucose alone experienced bioenergetic failure. This has important implications for measuring glucose oxidation in cells that require insulin for glucose uptake (i.e., ACMs) and/or rely more heavily on alternative substrates such as fatty acids.

Typically, XF assay medium consists of supraphysiological concentrations of pyruvate (1 mmol/l), which produces a maximal bioenergetic response in most cells (16); however, if the goal of the study is to examine the role of substrates on mitochondrial function (i.e., fatty acid, glucose), the concentration of pyruvate must be titrated so that the utilization of other substrates is not masked. When supraphysiological levels of pyruvate (1 mmol/l) were included in the assay medium, the ACM bioenergetic response was independent of the presence of other substrates. Yet, when pyruvate was titrated to more physiological levels (0.1 mmol/l), a robust increase in the degree of reliance on other substrates emerged. In context with metabolic flux, a change in the environment (i.e., changing substrate concentrations) results in the adjustment of cardiac metabolism (26). Occurring concordantly with increases in workload, the activities of the pyruvate dehydrogenase complex (25) and carnitine palmitoyltransferase I (15) increase. This ability to switch from one substrate (metabolic flexibility) to another is lost in the failing heart. This metabolic inflexibility is predominately controlled at the transcriptional level (i.e., FAO genes transcription decrease in heart failure). The development of the present technique enables more rigorous investigation into such experimental questions.

Because ∼60–80% of cardiac metabolism is due to fatty acid oxidation, it was particularly relevant to develop a protocol for measuring FAO by XF analysis. Recently, bioenergetic profiling of ACMs following right ventricular hypertrophy with XF analysis was reported (7); however, details concerning the analyses were lacking and there were several major limitations, including the absence of an attachment method for ACMs to XF24 plates (ACMs do not readily attach to uncoated plastics), unclear assay media contents, and the use of toxic concentrations of fatty acid (1.2 mmol/l palmitate) (24). Others have described XF methods for measuring FAO in noncardiac cells; however, this is the first study to define the optimal conditions for measuring FAO in isolated ACMs using the XF. Physiological concentrations of palmitate (0.05–0.20 mmol/l) + pyruvate (0.1 mmol/l) produced similar bioenergetic profiles as when supraphysiological levels of pyruvate (1 mmol/l) were used alone. This method could be used to assess switches in cardiac metabolism following heart failure and diabetes.

Although measuring mitochondrial function with merely the isolated organelle has its advantages (well-established methods, ease of isolation, etc.), there are several major drawbacks. First, mitochondria isolated from organs originate from multiple cell types (i.e., cardiomyocyte, endothelium, fibroblasts). Second, there is a lack of cellular context in terms of substrate availability and mitochondrial localization. Third, a classical argument refers to the loss of damaged/injured mitochondria during isolation. By establishing a relatively high-throughput method to measure mitochondrial function in intact ACMs, we overcame many of these disadvantages associated with assaying the isolated organelle. This improved method has drawbacks for those wishing to initiate this type of model system, including the difficulty of isolating ACMs from mice, the lack of context at the organismal level, and the complexity of data interpretation. Of note, XF analysis of ACMs will provide a global picture of cardiac bioenergetics in the diseased state and as such XF data may include myocytes whose contribution to maintain overall cardiac performance is minimal (i.e., myocytes from infarct zone). However, this method likely allows for more accurate interpretation of bioenergetics of the heterogeneous populations of mitochondria in the diseased heart than working with isolated mitochondria, as the most dysfunctional mitochondria are lost in the mitochondrial isolation. Nonetheless, this important point should be carefully taken into consideration when interpreting data using this XF method.

In conclusion, mitochondrial function in intact ACMs can be readily evaluated with the XF assay. Isolated ACMs attach efficiently to laminin-coated V7 microplates. Depending on the purpose of the experiment, substrate(s) present in the assay media must be carefully chosen. For ACMs, supraphysiological levels of pyruvate (1 mmol/l) produce a maximal bioenergetic response. To assay alternative substrates, pyruvate must be titrated such that the need for other substrates increases. This method represents a powerful approach to interrogate metabolic (dys)function in the intact adult mouse myocyte, even with underlying pathology.

GRANTS

This work was supported by National Institutes of Health Grants R01 HL-083320, R01 HL-094419, P20 RR-024489, P01 HL-078825. R. E. Brainard was an American Heart Association Predoctoral Fellow, Great Rivers Affiliate.

DISCLOSURES

B. G. Hill trainees (none of whom are coauthors on this paper) in Dr. Hill's laboratory have been supported by travel grants from Seahorse Biosciences to present their data (unrelated to the present paper) at national conferences.

AUTHOR CONTRIBUTIONS

Author contributions: R.D.R. and S.P.J. conception and design of research; R.D.R. and R.E.B. performed experiments; R.D.R. analyzed data; R.D.R., B.G.H., and S.P.J. interpreted results of experiments; R.D.R. prepared figures; R.D.R. drafted manuscript; R.D.R., R.E.B., B.G.H., and S.P.J. approved final version of manuscript; R.E.B., B.G.H., and S.P.J. edited and revised manuscript.