|Home | About | Journals | Submit | Contact Us | Français|
Measurements of glycolysis and mitochondrial function are required to quantify energy metabolism in a wide variety of cellular contexts. In human pluripotent stem cells (hPSCs) and their differentiated progeny, this analysis can be challenging because of the unique cell properties, growth conditions and expense required to maintain these cell types. Here we provide protocols for analyzing energy metabolism in hPSCs and their early differentiated progenies that are generally applicable to mature cell types as well. Our approach has revealed distinct energy metabolism profiles used by hPSCs, differentiated cells, a variety of cancer cells and Rho-null cells. The protocols measure or estimate glycolysis on the basis of the extracellular acidification rate, and they measure or estimate oxidative phosphorylation on the basis of the oxygen consumption rate. Assays typically require 3 h after overnight sample preparation. Companion methods are also discussed and provided to aid researchers in developing more sophisticated experimental regimens for extended analyses of cellular bioenergetics.
Assessments of energy metabolism are essential for interrogating cell functions and also for the diagnosis and tracking of certain human diseases. Cells require energy in the form of ATP to support the biological processes of life, including growth, division, differentiation and many physiological activities. Studies in cellular energy metabolism encompass the biochemical pathways that generate and consume ATP, as well as carbon sources, signaling networks, intermediate metabolites and regulatory mechanisms that control these interconnected processes. Mitochondria, as the central organelle in a variety of essential cell functions including metabolism, have been the focus of many studies over many years. Protocols have been reported for studying the function of mitochondria isolated from yeast1,2, from mouse tissues and cultured cells3, and from permeabilized fibers, tissues and cells4. However, measuring mitochondrial functions in hPSCs provides challenges not encountered in these other experimental systems.
In mammalian cell and developmental biology, considerable interest has emerged for understanding the functions and manipulating the activities of PSCs, especially those generated from human materials. hPSCs distinguish themselves from differentiated cells through the capacity to self-renew and thereby maintain the pluripotent state. With directed differentiation, hPSCs are at least theoretically capable of forming any of the 200+ cell types present in a child or adult, and therefore, beyond the innate curiosity associated with these cells, they hold tremendous potential for organ repair or replacement in the burgeoning field of regenerative medicine. One type of hPSC, human embryonic stem cells (hESCs), was first established by isolating and culturing cells obtained from the inner cell mass of human blastocysts over a decade ago5. Within the past 5 or so years, the exciting discovery of defined transcription factor reprogramming, and derivative techniques based on this process, has also lead to the generation of another type of hPSC, so-called induced pluripotent stem cells (iPSCs), from a variety of differentiated cell types6–9. Currently, both hESCs and human iPSCs (hereafter referred to as hPSCs) are routinely grown in Petri dishes, typically in small adherent clumps or colonies. Compared with most differentiated mammalian cell types, hPSCs also proliferate rapidly with shortened cell cycle times and a higher proportion of cells in S phase of the cell cycle10,11. Given the enormous interest in these cells, much effort has been expended in unraveling the genetic and epigenetic control mechanisms of the self-renewing pluripotent state and many protocols have developed for early non-specific and lineage-specific directed differentiation with variable degrees of efficiency12. However, much less attention has thus far been focused on studies of the unique energy and biosynthetic requirements either for maintaining the pluripotent state or for differentiation or reprogramming processes.
Recent studies have shown that PSC energy metabolism contrasts sharply with energy metabolism in most terminally differentiated cell types13–16. Glycolysis and oxidative phosphorylation (OXPHOS) are the two major mechanisms that mammalian cells use to produce ATP (Supplementary Fig. 1)17. It is well-known that most cancer cell types with intact mitochondria nevertheless convert to energetically less favorable glycolysis from OXPHOS in a metabolic transition termed the Warburg effect18,19, facilitating a shift to anabolic pathways. Several adult stem cell types with functional mitochondria also have been reported to favor glycolysis over OXPHOS for energy production, in contrast to their differentiated progeny cells. For example, hematopoietic stem cells predominantly use glycolysis rather than OXPHOS, possibly because they reside in a hypoxic bone marrow niche, which limits available oxygen as a terminal electron acceptor20,21. A similarly hypoxic microenvironment may also stimulate the glycolytic preference of tumor cell energy metabolism, at least in some pathologic settings. hPSCs with a relatively fast proliferation rate also rely more heavily on glycolysis than OXPHOS for energy production, probably because of an associated increased flux through the pentose phosphate pathway, which is required to generate ribose and NADPH for nucleotide and lipid for biosynthesis to support rapid cell proliferation14–16.
Unlike most mammalian cell lines, unique challenges for measuring mitochondrial metabolism in hPSCs include the growth of cells in small adherent clumps or colonies and the requirement for generally expensive and specialized culture medium that can practically limit the number of cells available for analyses. Therefore, to assess hPSC bioenergetics, we established an approach that keeps the mitochondria in their native intracellular environment, requires smaller samples than conventional methods of analysis, and minimizes manipulation of the experimental system to reduce material losses. Specifically, the extracellular acidification rate (ECAR), which approximates glycolytic activity under certain conditions, and the mitochondrial oxygen consumption rate (OCR), which is a key metric of mitochondrial function, can be determined simultaneously within the same small population of hPSCs using a commercially available XF24 Extracellular Flux Analyzer (Seahorse Bioscience). Accurate XF24 measurements require the analyzed cells to be grown in a uniform monolayer configuration, which is not how hPSCs grow under standard culture conditions. This requirement must be met because the XF24 pH and oxygen sensors require a uniform detection environment, whereas nonuniform local cell densities result in nonreproducible error-prone measurements. To overcome this shortcoming, hPSCs and other cell types used in this protocol are treated with the Rho-associated kinase (ROCK) inhibitor, which maintains hPSCs as single cells after disaggregating hPSC colonies with retained viability and function; this is followed by plating in a homogenous monolayer before XF24 measurements22. Our protocol reproducibly assesses glycolytic and OXPHOS activities and can be used to establish basal and maximal respiratory capacities, evaluate mitochondrial uncoupling phenomena and can also be used to examine nonglucose carbon sources for bioenergetic studies in hPSCs and differentiated cell types14.
Several protocols have been established for studying mitochondrial physiology and energetics including those that use purified mitochondria, detergent-permeabilized cells and intact cells. Because hPSCs are expensive to maintain and grow in small clumps at comparatively low cell numbers, these prior approaches have several shortcomings, which motivated development of the protocol we provide here.
Studies of mitochondrial metabolic function with detergent-permeabilized cells allow researchers to manipulate the experimental system but still keep mitochondria in their native intracellular environment4. Detergents such as digitonin and saponin, which have a high affinity for cholesterol23, can be added to cells to selectively permeabilize the plasma membrane while leaving the outer and inner mitochondrial membranes intact. A crucial parameter for this approach is to titrate the appropriate detergent for the specific cells being used in order to establish an optimal concentration that opens the plasma membrane but not other organellar membranes. As a control, a mitochondrial enzyme such as matrix-localized citrate synthase is typically used to assess whether the mitochondrial membranes have been inadvertently opened24. Permeabilization of the plasma membrane will cause the release of soluble cytosolic components, although this approach is still advantageous because mitochondria remain in contact with cytoskeletal and other native elements25. Also, mitochondrial metabolites, such as succinate, glutamate and malate that cannot cross the plasma membrane in intact cells can now be added to mitochondria in permeabilized cells, facilitating OXPHOS studies of individual respiratory complexes. The permeabilized cell approach is also advantageous when cell numbers are limiting, because these studies can be performed with as few as 105–106 cells.
Metabolic studies with purified mitochondria provide the greatest flexibility but remove mitochondria from their native intracellular environment. Standard procedures for mitochondrial isolation based on differential centrifugation of cell or tissue homogenates have been reported previously3,26. Isolated mitochondria are useful for a broad range of experiments including assays of the enzymatic function of individual respiratory chain complexes as well as the entire OXPHOS system, protein and RNA import studies, and studies on the transport of ions and metabolites across the mitochondrial membranes. However, this methodology has several disadvantages, particularly with regard to the purification of mitochondria from hPSCs. Mitochondrial isolation methods require extensive Dounce homogenization that can negatively affect the function of the mitochondria being collected. Typically, 30–40 Dounce strokes are required, and cell breakage can be difficult in cultured cells requiring even harsher conditions to achieve mitochondrial isolation. In addition, relatively large numbers of cells (e.g., > 2 × 108 cells) are needed for optimal yield and quality3. For hPSCs, isolated mitochondria are frequently obtained uncoupled, regardless of whether they are uncoupled within cells or not, and the yield is characteristically low, mainly because the cytoplasm is scant and therefore the numbers of mitochondria per cell are fewer than in larger, differentiated cells with abundant cytoplasm, or than in cells specialized for OXPHOS, such as hepatocytes or cardiomyocytes. As a result, isolated hPSC mitochondria are often, but not always, unsuitable in quality or quantity for large-scale studies, such as traditional respiration measurements. However, when extensive biochemical analysis is not required, mitochondrial preparations from hPSCs can be useful for other crucial studies, such as the analysis of the assembly of respiratory complexes using blue-native gel electrophoresis with or without in-gel activity assays27 and for mitochondrial transcription and translation studies. Researchers should be cautious, however, to gauge the extent of mitochondrial damage with isolation by using standard assays, such as citrate synthase activity or respiratory control ratios, whenever possible.
Methodologies for assessing mitochondrial function typically rely on respiration studies and other approaches. To facilitate this need, we developed an intact cell approach because this is currently the best method given the technical hurdles in working with small numbers of clumped hPSCs. This approach requires relatively few cells for assaying each condition of interest and the measurements are noninvasive and can be repeated multiple times in multiple assay wells. In addition, both glycolysis and respiration are measured simultaneously. The main disadvantage of this approach is that the method is limited to respiratory uncouplers and inhibitors (Table 1) or other reagents that readily cross an intact plasma membrane. Accordingly, mitochondrial metabolites such as glutamate, malate and succinate cannot be directly added to intact cells. Additional methods have also been developed for measuring glycolysis or OXPHOS in vivo, such as fluorescence- or luminescence-based methods that measure oxygen consumption (e.g., Becton Dickinson Biosensor plates)28 and probes that quantify ECARs, such as a microphysiometer that measures acidification based on proton excretion29. In contrast to the XF24 analyzer, however, a major disadvantage is that both of these later approaches cannot be done simultaneously on the same platform on the same small population of cells.
The traditional workhorse for measuring mitochondrial respiration has been the Clark-type oxygen electrode30, which can be used for intact and permeabilized cells as well as on isolated mitochondria. The major benefit of the oxygen electrode is that a large number of inhibitors and substrates can be added with isolated mitochondria or permeabilized cells, so that the respiratory activity of the individual electron transport chain (ETC) complexes can be assessed in addition to the respiratory control ratio (RCR), which is an indication of coupling efficiency between substrate oxidation and phosphorylation, and the ADP/O ratio, which assesses the efficiency of mitochondrial OXPHOS31. The drawbacks of the oxygen electrode are that large numbers of mitochondria or cells are typically required and that the chamber is stirred so cells cannot be adherent or in large clumps. Below we provide an XF24 analyzer protocol with glucose as the carbon source and directly compare this to measurements made with a conventional Clark-type oxygen electrode (Box 1; see also Tables 2 and and3)3) to highlight the utility of the Clark-type oxygen electrode for measuring all ETC complex activities in one run, even though a large number of cells are required. We also provide an XF24 analyzer protocol with a carbon source alternative to glucose, in this case free fatty acid (Box 2). All these protocols can be conveniently applied to multiple cell types as well. Figure 1 presents a flow chart of how these protocols relate if you are using hPSCs.
The assembly, calibration and maintenance of the oxygen-sensing electrode and the operation of the Oxygraph software for the electrode have been described previously3 and researchers should use the instruction manual provided by the manufacturer (Hansatech Instruments is an affordable example that we have illustrated). Here we present an optimized protocol for assaying hPSC oxygen consumption using a traditional Clark-type oxygen electrode for bioenergetic profiling. As explained previously, experiments with a Clark-type oxygen electrode are challenging to perform because the number of hPSCs is often a limiting factor. In theory, the same approach with inhibitors or uncouplers as described with the XF24 analyzer could be used, but the oxygen electrode can also be advantageous for measuring the activities of individual respiratory complexes using complex-specific substrate feeding in semi-permeabilized cells.
The XF24 and associated experimental systems must be optimized for each cell type being assessed. Essential considerations at the beginning of an evaluation include whether or not the cells under study can be grown in large quantities at reasonable cost and effort and also what the physical growth characteristics of the cells are, such as adherent or suspension expansion and growth as single cells or clumps. For reproducible XF24 measurements, cells that grow in clumps require dispersion into monolayers without affecting their viability or function, which is the situation for hPSCs. Also, cells that grow in suspension need to be attached as a uniform monolayer to the bottom of an XF24 analyzer plate well, which can be the situation for white blood cells. In addition, a cell number titration to determine the optimal cell seeding density so that measurements are within the linear range of the XF24 analyzer is required. Titration of permeable substrates, such as respiratory uncouplers and inhibitors, also must be performed to find acceptable nontoxic dosing regiments over the time frame used for each experimental condition tested.
OCR and ECAR measurements on the same small population of cells can be determined during culture in defined medium conditions. The ratio of OCR to ECAR can indicate cellular preference for OXPHOS versus glycolysis when mitochondria are coupled for oxygen consumption and energy generation through complex V (F1F0 ATP synthase) activity. Studies with a variety of cell types have indicated that ECAR values determined by the XF24 analyzer are a reliable measure of glycolytic rate, even though, in principle, extracellular acidification can be contributed by respiratory CO2 and by monocarboxylates other than lactate32–36. The addition or removal of nutrients in the culture medium can be informative for determining cellular energetic compensation or responses across extremes of abundance or starvation, especially when linked with additional assessments of cell growth, proliferation, division and death. Similarly, the use of respiratory inhibitors can be an informative method for interrogating mitochondrial functions in respiration and energy production. Judicious use of respiratory inhibitors singly or in combinations can indicate (i) mitochondrial and nonmitochondrial components of cellular oxygen consumption, (ii) oxygen consumed for ATP generation through the F1F0 ATP synthase versus oxygen consumed with passive proton leakage across the mitochondrial inner membrane, which reduces the mitochondrial membrane potential, and (iii) maximal respiratory capacity in coupled and completely uncoupled conditions. The utilization of alternative carbon fuels, such as free fatty acids, can be assessed by changing the cell growth medium to the carbon source of interest and repeating the same basic measurements using the XF24 analyzer and ancillary studies (Box 2). These basic measurements of cellular energy metabolism can be obtained in a single XF24 run of 2–3 h in duration, which requires appropriate repetition and controls to achieve statistical significance.
For each experimental plan, control XF24 analyzer wells are needed to establish background measurements. The control wells typically have no cells, but contain the same quantities and types of assay media and undergo the same agent injection procedures and measurements. Agent vehicle control injections, such as DMSO or ethanol, should also be included as separate assay plate wells to compensate for their influence, if any, in the experimental outcomes. For experiments measuring agent responses, basal level OCR or ECAR quantities can be used as no-treatment controls.
We provide the following protocols for researchers working on cellular energy metabolism mainly in the stem cell field but with concepts and approaches that generalize to studies of energy metabolism in all mammalian cell systems. This protocol serves as a starting point for those not familiar with mitochondrial manipulations and addresses how to measure basic glycolytic and OXPHOS metabolic pathway activities. Metabolism research is highly topical and represented with increasing frequency in all fields of mammalian cell biology, including topics related to PSC self-renewal and differentiation, reprogramming to pluripotency, mechanisms in metabolism, and basic and clinical cancer biology37. As a researcher becomes more practiced, experiments can be refined to include permeabilized cells and isolated mitochondria.
According to instructions provided with the StemPro hESC SFM kit, to 454 ml of DMEM/F-12 medium add 36 ml of BSA, 10 ml of supplement, 5 ml of 100× penicillin-streptomycin (optional), 3.6 μl of 14.3 M 2-mercaptoethanol (final concentration: 0.1 M). Store at 4 °C for up to 1 month.
To 870 ml of DMEM or RPMI 1640 plain medium, add 100 ml of FBS, 10 ml of NEAA, 10 ml of l-glutamine and 10 ml of penicillin-streptomycin; mix well and store at 4 °C for up to 3 months.
To reconstitute bFGF, dissolve the lyophilized powder with sterile 1× PBS to 10 μg ml−1 and add BSA to a final concentration of 0.1% (vol/vol). It can be stored at − 80 °C for 3 months without detectable loss of activity.
Dilute 10× trypsin (0.5% (wt/vol)) to 1× (0.05% (wt/vol)) in sterile 1× PBS. Store at − 20 °C in 5-ml aliquots. Thaw a small aliquot freshly every time; do not reuse.
Dissolve 5× dispase (5 mg ml−1) fivefold with DMEM/F-12 plain medium to a concentration of 1 mg ml−1. Filter through a 0.22-μm filter unit. Prepare 1-ml aliquots and freeze them at − 20 °C for longer-term storage, or refrigerate them at 4 °C for up to 2 weeks.
Dissolve nonsterile 50 mg of trypsin inhibitor powder into 50 ml of 1× PBS and vortex vigorously. After complete dissolution, filter the solution with a 0.22-μm filter unit. Preserve at 4 °C for up to 2 weeks.
Dissolve ROCK inhibitor in water to make a 10 mM stock solution. Prepare 10-μl aliquots and freeze at − 20 °C for up to 1 month.
Dilute Matrigel 60-fold into DMEM/F-12 plain medium before making Matrigel plates. A volume of 100 μl of Matrigel in 6 ml of DMEM/F12 is enough for one six-well plate.
Add GlutaMAX to a final concentration of 2 mM. Add sodium pyruvate to a final concentration of 1 mM. Add glucose to a final concentration of 25 mM. (Note that these setups are assay dependent, especially when the assay is related to substrate utilization.) Freshly prepare this medium with each use.
Dissolve oligomycin A in ethanol to make a stock solution of 1 mM; dissolve FCCP in DMSO to make a stock solution of 1 mM; dissolve rotenone in DMSO to make a stock solution of 10 mM; dissolve antimycin A in ethanol to make a stock solution of 10 mM. Store all inhibitors at − 20 °C for up to 1 month. ! CAUTION These agents are ETC inhibitors and can be acutely toxic. Wear gloves, protective clothing, face/eye shields and respiratory protection during preparation. Wash skin thoroughly after handling. Avoid release into the environment. CRITICAL Inhibitors may be light sensitive; prepare them in dark tubes or store them in the dark.
Add 1.75 g of KCl, 32.45 g of NaCl, 1.2 g of MgSO4 and 0.85 g of Na2HPO4 into 1 liter of water to make 5× KHB assay medium. Store at 4 °C for up to 3 months. On the day of the experiment, dilute 5× KHB assay medium with water to a 1× final concentration and add 4.2 mM MgCl2, 0.4 mM glucose, 0.5 mM l-carnitine and 0.1 μM insulin (final concentration). Adjust the pH to 7.4 and filter sterilize.
Make 1 mM sodium palmitate in 150 mM NaCl and 0.17 mM BSA in 150 mM NaCl. Heat the sodium palmitate solution to 70 °C on a heated stir plate until it becomes clear, and then add an equal volume of hot sodium palmitate to the BSA solution (the molar ratio between palmitate and BSA is 6:1). Stir at 37 °C for 1 h. Adjust the pH to 7.4. Aliquot into 1-ml glass vials and store at − 20 °C for up to 1 month.
Dissolve ( + )-etomoxir sodium salt hydrate in water to make a stock solution of 100 mM. Store the stock solution at 4 °C for up to 1 month.
Dissolve retinoic acid into DMSO to make a 100 mM 10,000× stock solution, aliquot and store at − 20 °C for up to 3 months. To make the medium, add 1 μl of stock solution into 10 ml of StemPro medium without bFGF. Vortex to mix. Store at − 4 °C for no more than 2 weeks.
Prepare a 0.137 M NaCl, 5 mM KCl, 0.7 mM NaH2PO4, 25 mM Tris-HCl solution with water; adjust the pH to 7.4 and filter sterilize. Store at −4 °C for up to 3 months.
This buffer is composed of 50 mM Tris-HCl (pH 7.4), 150 mM NaCl, 1 mM EDTA and 1% (vol/vol) Triton X-100. Store it at −4 °C for up to 6 months.
hPSCs are initially grown on a MEF feeder layer. To eliminate MEF contamination before experiments, split hPSCs from MEF feeder layer plates to Matrigel-coated plates, as reported in detail previously39 and described below. hPSCs can be maintained on Matrigel-coated plates for approximately ten passages.
Here we summarize the experimental procedures and the time needed for each step (Fig. 1).
Step 1, growing sufficient numbers of hPSCs on feeder-free Matrigel plates: ~ 4–5 d
Steps 2–5, seeding hPSCs in a XF24 V7 cell culture microplate: 2 h
Steps 6 and 7, incubating hPSCs at 37 °C in a CO2 incubator and hydrating a sensor cartridge at 37 °C in a non-CO2 incubator: 16 h (overnight)
Steps 8–11, exchanging XF24 assay medium and loading the sensor cartridge with injection solutions: 1 h
Steps 12–16, loading the sensor cartridge and assay microplate and running the XF24 Extracellular Flux Analyzer detection program: 3 h
Steps 17–21, determining protein and lactate concentrations and collecting data for analyses: 1 h
Box 1, measuring oxygen consumption with a Clark-type oxygen electrode: 3 h
Box 2, measuring extracellular flux with free fatty acid as the carbon source: 3 h
Box 3, maintenance and splitting of hPSCS on Matrigel plates: 1 h
For each type of cell, a prerequisite control experiment is the titration of the number of cells seeded into XF24 V7 cell culture microplates. When 20,000–75,000 hPSCs or human fibroblasts (NHDFs) are seeded, OCR values are within the linear detection range of the instrument; however, ECAR values for 75,000 human fibroblasts become inaccurate and plateau (Fig. 4a). On the basis of our analysis, a range of 20,000–50,000 hPSCs or human fibroblasts provides robust detection signals over the linear range of the instrument for both OCR and ECAR measurements. When there are less cells available, a XF96 analyzer with a 96-well assay plate can be used to scale down to the fewer number of cells. In that event, careful adjustment of the mixing times before each measurement is required to ensure sufficient oxygen equilibration in a smaller assay volume. When the number of cells for different cell types can be kept in the same instrument linear range, this facilitates direct OCR and ECAR comparisons between different cell types. With a constant number of cells, either the absolute OCR or ECAR levels, normalized to protein concentration, or the ratio of OCR/ECAR can be graphically plotted. Figure 4b shows OCR/ECAR ratios for two hESCs (HSF1 and H1), two hiPSCs (HIPS2 and HIPS18) and human fibroblasts (NHDFs). Notably, the OCR/ECAR ratio is typically not affected with or without the addition of ROCK inhibitor, as shown for NHDFs (Supplementary Fig. 2). Low OCR/ECAR ratios for the hPSCs indicate their relatively higher reliance on glycolysis compared with fibroblasts14. For all experiments presented herein, data are presented as the mean ± s.d. A two-tailed test was used with P < 0.05 considered statistically significant. Using this method of analysis, we have also recently reported a progressive shift from a low to a high OCR/ECAR ratio during the process of hPSC differentiation14.
In addition to determining basal OCR and ECAR measurements, a key protocol capability is to analyze the hPSC bioenergetic profile with a series of mitochondria-specific inhibitors or respiration uncouplers. For example, oligomycin A inhibits the mitochondrial F1F0 ATP synthase, FCCP is a mitochondrial proton ionophore (uncoupler) and rotenone and antimycin A are ETC complex I and complex III inhibitors, respectively (Fig. 5a). Because they are membrane permeable, these inhibitors/uncouplers work well with intact cell studies in the XF24 Extracellular Flux Analyzer. To identify an OCR value at which coupled respiration is inhibited, cells are exposed to 1 μM oligomycin A to prevent proton movement through the F1F0 ATP synthase, which has the effect of increasing the proton gradient across the mitochondrial inner membrane. The addition of 0.3 μM FCCP re-establishes proton movement nonspecifically across the mitochondrial inner membrane and results in complete uncoupling of electron transport from ATP generation by the F1F0 ATP synthase, yielding an OCR value for maximal O2 consumption for the cell type being examined. Subsequent addition of rotenone + antimycin A blocks proton pumping by the ETC and rapidly cripples mitochondrial O2 consumption completely. This sequence of added inhibitors/uncouplers provides the following energy-profiling data: (i) the maximal respiration capacity, which is the difference between FCCP-induced OCR and rotenone + antimycin A-blocked OCR; (ii) the component of OCR used to generate ATP, which is the difference between the basal and oligomycin A-repressed OCR; and (iii) the component of OCR representing passive proton leakage across the mitochondrial inner membrane, which is the difference between oligomycin A and rotenone + antimycin-A inhibited OCR (Fig. 5a). The OCR ‘drift’ over time after oligomycin addition for HSF1 cells and after FCCP addition for NHDF cells is not due to changes in apoptosis or cell cycle parameters14, has also been reported for studies in additional cellular contexts by other investigators41,42, and may reflect distinct drug response kinetics in different cell types. Notably, our sample data show that hPSCs (HSF1 cells) consume O2 at maximal capacity in the basal state and that only 60–70% of O2 consumption is used to generate ATP (Fig. 5b). By contrast, human fibroblasts (NHDFs) respire at about half of their maximal capacity in the basal state, and almost all O2 consumed is used to synthesize ATP in the mitochondrial F1F0 ATP synthase (Fig. 5b). These results are reproducible with additional hPSCs and with double the concentration of pyruvate in the growth medium, which eliminates the concern that limited carbon substrate is responsible for a reduced maximal respiratory capacity14.
To illustrate that the XF24 Analyzer can generate comparable results as the conventional method, we performed the two methods in parallel. The measured OCR for basal-state, intact hPSCs (HSF1) using the XF24 Analyzer is 73.40 ± 8.57 pmol min−1 per 5 × 104 cells (Fig. 6a)14. The OCR measured for basal-state, intact HSF1 cells from the slope of the polarographic tracing using a Clark-type oxygen electrode (after ‘add cells’ at 2 min 44 s and before ‘digitonin’ permeabilization at 4 min 21 s in Fig. 6b) is 7.95 ± 1.40 nmol min1 per 5 × 106 cells14. From these previously published data, the calculated OCR per individual HSF1 cell is similar between the two methods, although the conventional oxygen electrode measurements require 100-fold more cells to obtain reproducible results (Fig. 6a).
Despite the XF24 Extracellular Flux Analyzer’s requirement for fewer cells than a Clark-type oxygen electrode, the electrode approach maintains an advantage by being able to determine the activities of individual ETC complexes in one experimental run. This is because permeabilized cells are continuously stirred in suspension with an electrode system, and manual injections with a Hamilton syringe theoretically permit an unlimited number of substrate and inhibitor deliveries until all of the oxygen in the assay buffer is consumed. This advantage is demonstrated in a polarographic tracing that determined the individual ETC complex activities for hPSCs (HSF1; Fig. 6b). The slope between the addition of a substrate and an inhibitor of the ETC complex being examined indicates the activity of that complex. For example, the slope between succinate (substrate) and malonate (inhibitor) additions indicates ETC complex II activity. As anticipated from the overall oxygen consumption data, ETC complexes I to IV in HSF1 cells are functional, as previously reported14. For ECAR, the difference between basal-state HSF1 cells and retinoic acid–induced differentiated progeny cells is an approximately 75% reduction when measured using the XF24 Extracellular Flux Analyzer and an approximately 80% reduction when measured using a commercial lactate assay kit (Fig. 6c). Both results support a similarly sized shift from glycolysis to respiration during hPSC differentiation, as reported previously14,16.
Cancer cells and mouse ESCs use fatty acid oxidation for ATP production, especially when these cell types are under metabolic stress, such as in conditions of glucose deprivation or hypoxia43. Our protocol can be modified to interrogate alternative carbon sources, such as free fatty acids, as energy sources in hPSCs and their differentiated progeny. In Figure 7a, the exogenous free fatty acid palmitate, which can be internalized by intact cells, was injected into the XF24 assay medium, followed by the addition of etomoxir and detection with the XF24 Extracellular Flux Analyzer. The component of OCR attributed to palmitate oxidation is determined by the addition of etomoxir, which freely diffuses into cells and is an inhibitor of carnitine palmitoyltransferase 1 (CPT1), an enzyme that transports long-chain fatty acids across the mitochondrial inner membrane, thereby effectively inhibiting fatty acid oxidation. When 200 μM palmitate is added to human fibroblasts (NHDFs) over 30 min, the OCR level increased by ~20%, which was subsequently inhibited by the addition of 100 μM etomoxir (Fig. 7a). These data support the increase in OCR as being specifically from the oxidation of the added free-fatty acid palmitate (Fig. 7a). By contrast, palmitate addition does not increase the OCR in hPSCs (H1), suggesting that hPSCs do not use exogenous fatty acids as a carbon source for OXPHOS (Fig. 7a). Similar results were obtained with HSF1 cells using a dose escalation of added palmitate. The OCR increase in HSF1 cells is < 5%, whereas the OCR increase in human fibroblasts (NHDFs) is proportional to the added palmitate and can be as high as ~30% (Fig. 7b). The data also show that the OCR in HSF1 cells cannot be increased by increasing doses of added palmitate.
This energy-profiling protocol can be used for hPSCs and additional mammalian cell types, including those that grow in suspension, such as human blood lineage cells. In Figure 8, a wide range of cell types was profiled for OCR/ECAR ratios and absolute OCR and ECAR values. Rho0 143B TK− is a human osteosarcoma cell line depleted of its mitochondrial DNA. Therefore, this line cannot respire and, as anticipated, these cells have the lowest OCR and high ECAR activities because they rely exclusively on glycolysis for energy production. hPSCs (HSF1) are also highly glycolytic, even though hPSCs do respire at their maximal capacity (Figs. 5a and and8a8a)14. HEK293T and HeLa cancer cells have OCR/ECAR ratios that show robust glyolytic and respiratory activity, which is perhaps an indication of high metabolic activity to support a high rate of cell proliferation. Of note, MCF7 human breast cancer cells appear to respire well, with an OCR > 200 pmol min−1 per 50,000 cells. This occurrence might not be anticipated from the Warburg effect in cancer cells18–19,44, although recent studies show that different cancer cell types or lines of the same type of cancer do not all have the same metabolic profile and some cancers are more oxidative than glycolytic45,46. The Ramos, Raji and Nalm6 lines, representing B lymphocyte lineage cancers, all show absolute OCR and ECAR values that are much lower compared with other cell types. The size of these cancer B cells ( <10 μm in diameter) is much smaller than that of other cells in the comparison (~20–30 μm in diameter), suggesting that cell size and volume may be important considerations because of varying protein/mitochondrial content per cell when comparing the absolute level of OCR or ECAR across different cell types14. Last, mitomycin-C-treated MEFs show the highest respiratory activity among all cell types examined. Their OCR is ~4.5-fold higher than HSF1 cells, whereas their ECAR is about half that of HSF1 cells. This large difference in energy profiles strengthens the importance of preparing hPSCs under feeder-free conditions to avoid MEF contamination and inaccurate OCR and ECAR measurements. On the basis of our data, even a contamination of 5% MEFs in a XF24 study on hPSCs could result in an ~17.5% overestimation of OCR and an ~20% overestimation of the OCR/ECAR ratio. Thus, growing hPSCs in feeder-free conditions is important for eliminating this potential error in energy-profiling measurements.
We thank J. Tang for hESCs. Supported by CIRM grants RS1-00313, RB1-01397, TB1-01183, TG2-01169, a training grant from the Broad Stem Cell Research Center at the University of California Los Angeles, and US National Institutes of Health grants GM061721, GM073981, PNEY018228, P01GM081621, CA156674 and CA90571. C.M.K. is an Established Investigator of the American Heart Association and M.A.T. was a Scholar of the Leukemia and Lymphoma Society. We thank D. Wallace (University of Pennsylvania) for Rho0 143B TK− human osteosarcoma cells.
Note: Supplementary information is available in the online version of the paper.
AUTHOR CONTRIBUTIONS J.Z.: conception and design, collection and assembly of data, data analysis and interpretation, manuscript writing; E.N.: conception and design, data collection and analysis, manuscript writing; D.R.R.W.: collection and assembly of data, data analysis and interpretation; K.S., J.S.H., C.M.V.H., S.S.I., L.V.: technical support and assistance; C.S.M.: conception and design, data analysis and interpretation, manuscript writing; C.M.K.: conception and design, data analysis and interpretation, manuscript writing; M.A.T.: conception and design, data analysis and interpretation, manuscript writing, final approval of the manuscript and financial support.
COMPETING FINANCIAL INTERESTS The authors declare no competing financial interests.
Reprints and permissions information is available online at http://www.nature.com/reprints/index.html.