How to Test Mitochondrial Function: Methods & Analysis

Testing mitochondrial function ranges from measuring how fast cells consume oxygen on a benchtop respirometer to tracking phosphocreatine recovery inside a living person’s muscle with an MRI scanner. No single assay captures the full picture, because mitochondria do many things at once: they burn fuel, maintain an electrical gradient across their inner membrane, produce reactive oxygen species, and constantly reshape themselves through fusion and fission. The method you choose depends on whether you are working with cells in a dish, a frozen tissue biopsy, circulating blood cells, or a patient who cannot undergo an invasive procedure at all.

Respirometry as the Workhorse

The most widely used laboratory approach measures how quickly a biological sample consumes oxygen. Oxygen consumption is a direct readout of the electron transport chain doing its job, so it serves as a reliable proxy for overall mitochondrial output. Two platforms dominate the field. The Agilent Seahorse XF Analyzer works with cells growing in multiwell plates, measuring oxygen consumption and acid production simultaneously in small volumes. The Oroboros O2k is a high-resolution respirometer that uses a sealed chamber and a polarographic oxygen sensor, giving continuous real-time traces with fine sensitivity.

A head-to-head comparison of these two platforms using human fibroblasts found no meaningful difference in baseline (“routine”) respiration or in the residual oxygen consumption that persists after blocking the cell’s ATP-producing machinery, regardless of whether the cells were attached to a plate or suspended in solution. Where they did diverge was in measuring maximum electron transfer capacity: the O2k gave higher values than the Seahorse, likely because the Seahorse limits you to two injections of the chemical uncoupler that pushes mitochondria to their top speed, while the O2k lets you titrate in as many small doses as needed to find the true peak.¹ That practical limitation matters: if you underestimate spare respiratory capacity, you might miss a deficit that only shows up when the cell is pushed hard.

A newer frozen-tissue method designed for cardiac samples reinforced this point. The Seahorse’s fixed four-injection format limits what you can verify mid-experiment, whereas the O2k’s flexibility allows a separate injection of rotenone (a blocker of the first enzyme complex) to confirm that the signal you just measured actually came from that complex.² For routine screening work, the Seahorse’s high throughput is hard to beat. For detailed mechanistic studies, the O2k’s open-ended protocol design gives researchers more room to troubleshoot and explore.

Stepwise Chemical Titrations That Isolate Each Complex

Simply measuring total oxygen consumption tells you that the engine is running, but not which cylinder is misfiring. That is where substrate-uncoupler-inhibitor titration protocols come in. The idea is straightforward: you add specific fuels and poisons in a deliberate sequence, and each step reveals the contribution of a particular enzyme complex or pathway. One step might feed electrons exclusively into complex I, the next adds fuel for complex II, and a later step blocks complex III entirely to establish a background correction.³

Because mitochondria operate as an interconnected network rather than a set of independent units, these protocols have evolved to include many steps that modulate both coupling control (how tightly oxygen consumption is linked to ATP production) and electron transfer pathway states. Multiple protocols can be run on the same sample to build a comprehensive profile of respiratory control.⁴ The output is a series of respiratory states, each with a defined functional meaning, rather than a single number. This layered information is what makes respirometry so powerful for diagnosing where in the chain something has gone wrong.

Why Sample Preparation Can Make or Break Results

The way you prepare tissue for respirometry shapes the answer you get, sometimes dramatically. The traditional approach isolates mitochondria by grinding up tissue and spinning it in a centrifuge to separate the organelles. This yields a clean preparation, but it tears mitochondria out of their natural networks and strips away the structural interactions with the rest of the cell.

Permeabilized muscle fibers offer an alternative. A mild detergent pokes holes in the outer cell membrane so that substrates and inhibitors can reach the mitochondria, while the organelles themselves stay embedded in their normal architecture. This approach requires only small amounts of tissue, as little as two to five milligrams, making it suitable for human needle biopsies as well as mouse muscle.⁵ The practical payoff is not just convenience. A study comparing isolated mitochondria with permeabilized fibers in aging rat muscle found that the isolation procedure itself exaggerated the apparent decline in mitochondrial function. Respiration, reactive oxygen species emission, and sensitivity to calcium-triggered damage all looked significantly worse in isolated organelles than in fibers measured in their intact cellular context.⁶ The implication is sobering: years of research on aging mitochondria may have overstated the problem by relying on a preparation method that introduces its own damage.

Fresh, Frozen, or Cryopreserved

Not every lab can run respirometry the moment a sample arrives. Freezing and thawing tissue destroys intact-cell respiration because ice crystals rupture membranes, but the individual enzyme complexes embedded in those membranes can survive if the sample is handled carefully. A protocol for frozen colon tissue demonstrated that complex I and II activity, as well as maximal respiration driven by specific substrates, remained intact after snap-freezing.⁷ When liquid nitrogen is not immediately available, keeping samples on ice-cold buffer preserves the integrity of the mitochondrial complexes.⁸

Cryopreservation, where tissue is frozen with a protective agent rather than simply plunged into liquid nitrogen, offers a middle ground. Placental biopsies cryopreserved using a controlled protocol produced oxygen consumption rates comparable to fresh tissue, whereas snap-freezing impaired mitochondrial activity.⁹ The takeaway for anyone designing a study: the freezing method you choose determines which aspects of mitochondrial function you can still measure later.

Membrane Potential Assays

Mitochondria maintain a steep voltage difference across their inner membrane, and this electrical gradient is what actually drives ATP production. Measuring that membrane potential offers a window into mitochondrial health that is complementary to oxygen consumption. The classic approach uses fluorescent dyes that accumulate inside mitochondria in proportion to the voltage. One widely used dye, TMRM, can be loaded into living cells and imaged on a confocal microscope either for a snapshot comparison between cell types or for tracking changes over time in response to a drug or stress.¹⁰

A limitation of standard intensity-based dye imaging is that brightness can be influenced by factors besides voltage, such as dye loading efficiency or focus drift. A newer technique sidesteps this by measuring the fluorescence lifetime of a dye rather than its brightness. When mitochondria were depolarized with known inhibitors, the fluorescence lifetime of the probe shifted in the predicted direction, rising when the membrane potential dropped and falling when it increased.¹¹ Lifetime-based readouts are less susceptible to artifacts from dye concentration or photobleaching, making them attractive for quantitative comparisons.

Noninvasive In Vivo Methods

All of the techniques above require a tissue or blood sample. For situations where you want to assess a patient’s mitochondrial function without a biopsy, two spectroscopy approaches stand out.

Phosphorus Magnetic Resonance Spectroscopy

Phosphorus-31 MRS works inside an MRI scanner. After a brief bout of exercise, the scanner tracks how quickly phosphocreatine rebuilds in the muscle. Phosphocreatine recovery depends almost entirely on mitochondrial ATP production, so the speed of that recovery serves as a noninvasive index of oxidative capacity.¹² The technique also captures changes in pH and inorganic phosphate, which add context about how the muscle manages its acid-base balance during and after effort.¹³ An expert consensus statement has highlighted pitfalls in interpreting these kinetics, since factors like blood flow and muscle activation patterns can confound the measurement if not carefully controlled.¹⁴

Near-Infrared Spectroscopy

NIRS is cheaper, portable, and far more accessible than MRS. A small sensor placed on the skin shines near-infrared light into the muscle and measures how much is absorbed by oxygenated versus deoxygenated hemoglobin and myoglobin. By briefly cuffing off arterial blood flow after a short contraction, researchers can watch how fast the muscle uses up its local oxygen supply, which reflects mitochondrial oxygen consumption.¹⁵ A validation study in healthy adults found that NIRS recovery kinetics from the thigh muscle after a roughly 15-second contraction correlated well with respirometry data from biopsied tissue, supporting NIRS as a legitimate field-ready tool.¹⁶ The tradeoff is that NIRS gives you a single composite number for the muscle region under the sensor, without the complex-by-complex detail that respirometry provides.

Blood-Based Approaches

Drawing blood is far less invasive than taking a muscle biopsy, and two complementary strategies make blood samples useful for mitochondrial testing.

Circulating Biomarkers

Two proteins, GDF15 and FGF21, have emerged as the most promising blood biomarkers for mitochondrial disease. Both are secreted into the bloodstream in response to mitochondrial stress, and both clearly outperform the traditional marker, lactate, which is elevated in only about a third to two-thirds of children with primary mitochondrial disorders.¹⁷ In adults, GDF15 showed a stronger diagnostic odds ratio than FGF21, along with higher sensitivity and better performance on receiver operating characteristic analysis.¹⁸ A pediatric study suggested a slightly different picture, with GDF15 serving best as a screening tool for diagnosis and FGF21 better at tracking disease severity over time.¹⁹ Neither marker is perfectly specific to mitochondrial disease. GDF15 in particular rises with many forms of cellular stress, which is why it tends to be more broadly sensitive but also more likely to flag non-mitochondrial conditions.

Platelet and Immune Cell Respirometry

Platelets and peripheral blood mononuclear cells are packed with mitochondria and can be isolated from a standard blood draw. Platelet respirometry has been validated as a reliable way to assess systemic mitochondrial function in both laboratory and field settings, including after physical stress.²⁰ A large study spanning ages three months to 86 years used high-resolution respirometry on both platelets and PBMCs to profile how mitochondrial respiration changes across the human lifespan, demonstrating that blood-cell respirometry can capture age-related trends without a single tissue biopsy.²¹ The catch is that blood cells do not always mirror what is happening in a specific organ. A person with a tissue-specific mitochondrial defect in the heart or brain might show normal platelet respiration.

Spectrophotometric Enzyme Assays

Before high-throughput respirometers existed, the standard clinical approach to diagnosing mitochondrial disease involved measuring the activity of each electron transport chain complex individually using a spectrophotometer. These assays track the rate at which a substrate changes color as it is oxidized or reduced by a specific complex. Protocols exist for complexes I through IV on minute quantities of muscle, cultured cells, or isolated mitochondria.²² Automated versions of these assays have improved throughput and reproducibility for clinical laboratories.²³

The strength of spectrophotometric assays is their specificity: they give you a number for each complex in isolation. Their weakness is that they measure maximal enzyme velocity in a test tube, which does not always predict how the system behaves when all the complexes are working together in an intact membrane. Refinements continue; an improved assay for complex III, for example, used optimized detergent conditions to achieve greater sensitivity and a higher fraction of inhibitor-sensitive activity, making it more reliable across liver, heart, skeletal muscle, and brain homogenates.²⁴

Reactive Oxygen Species Measurement

Mitochondria are a major source of reactive oxygen species, and measuring ROS output tells you something different from measuring oxygen consumption. A cell can consume plenty of oxygen yet leak an abnormally large fraction of electrons before they reach their intended destination, generating superoxide in the process. The most popular cell-based method uses MitoSOX, a fluorescent probe targeted to mitochondria. Flow cytometry with MitoSOX is straightforward, but the commonly used dye concentration may not be optimal: testing at different concentrations showed that a lower dose produced more reliable discrimination between normal cells and cells with depleted mitochondrial DNA.²⁵

For isolated mitochondria or submitochondrial preparations, the Amplex Red assay detects hydrogen peroxide released into the surrounding medium. One important caveat is that the rate of ROS production measured this way depends on the oxygen concentration in the experiment, which is often much higher than what mitochondria experience inside a living cell.²⁶ Running assays at atmospheric oxygen levels can overestimate ROS generation compared to physiological conditions.

Normalizing the Numbers

A common pitfall in mitochondrial testing is comparing raw oxygen consumption or enzyme activity between samples without accounting for how many mitochondria are present. A sample with twice as many mitochondria will naturally consume more oxygen, which does not mean each organelle is working harder. Citrate synthase activity is the most frequently used normalization marker because it correlates well with mitochondrial content in skeletal muscle.²⁷

Mitochondrial DNA copy number is another option, but it does not always track with actual mitochondrial mass. In young healthy subjects, there was no significant correlation between mtDNA content and the physical area occupied by mitochondria in muscle sections. Furthermore, both mtDNA copy number and citrate synthase activity change with age, with mtDNA rising steadily from birth and citrate synthase jumping after age 20.²⁸ A study measuring multiple markers in healthy young muscle found that complex I, II, and IV activities and complex II and V protein content all showed substantial concordance with transmission electron microscopy measurements of mitochondrial area, while mtDNA did not.²⁷ The choice of normalization marker can change your conclusions, especially in disease states where mitochondrial proliferation may compensate for a per-organelle defect.

Clinical Context and Lactate’s Limitations

In a clinical setting, the first hint of mitochondrial dysfunction often comes from a simple blood lactate measurement. A plasma lactate above about 3 mmol/L, properly collected, raises suspicion. But elevated lactate has many causes beyond primary mitochondrial disease, and its sensitivity is modest: studies report true elevated lactate in roughly a third to two-thirds of confirmed patients. Specificity is better, ranging from about 83 to 100 percent, meaning a normal lactate does not rule out mitochondrial disease, but a genuinely high one is a meaningful signal.²⁹ The ratio of lactate to pyruvate in blood can further narrow the differential, with a specificity of 100 percent for electron transport chain disease when lactate is elevated, though its sensitivity is only around 31 percent.²⁹ Because no single test is definitive, clinical diagnosis typically layers several methods: biomarkers, spectrophotometric enzyme assays, respirometry on biopsied tissue, and increasingly, genetic sequencing.

Single-Cell Profiling and Mitochondrial Heterogeneity

One of the reasons mitochondrial disease is so variable, even within a single patient, is heteroplasmy: different cells carry different proportions of mutant and normal mitochondrial DNA. Bulk assays average over thousands or millions of cells, potentially masking a subpopulation that is severely affected. New single-cell platforms aim to resolve that hidden diversity. One approach pairs mitochondrial DNA sequencing with simultaneous measurement of membrane potential and ROS in individual cells, revealing subpopulations with distinct mutation burdens and altered bioenergetic profiles within the same tissue sample.³⁰

A complementary strategy skips the genetic side and instead uses high-content imaging to quantify mitochondrial shape at the single-cell level. By applying dimensionality reduction to the morphological data, researchers can identify clusters of cells that respond differently to the same mitochondrial inhibitor, picking out subpopulations that are more severely affected even before any genetic testing is done.³¹ These tools are still largely research-grade, but they point toward a future where mitochondrial testing is not just about average function in a sample but about the distribution of function across individual cells.

Dynamics, Mitophagy, and Shape as Functional Readouts

Mitochondria are not static beans. They constantly fuse into networks and split apart, and damaged ones are tagged for recycling through a quality-control process called mitophagy. Measuring the proteins that drive these processes, such as OPA1, MFN1, and DRP1 for fusion and fission, and PINK1 and Parkin for mitophagy, gives a picture of mitochondrial turnover and quality control that respirometry alone cannot.³² A system where fusion dominates tends to have elongated, interconnected mitochondrial networks, while fragmentation often signals stress or preparation for disposal. These markers are typically measured through gene expression or protein blotting from tissue biopsies, and they add a layer of information about how the mitochondrial population is being maintained over time rather than just how it is functioning at a single moment.

Computational Modeling

As datasets from genomics, proteomics, and metabolomics grow, computational models are increasingly used to integrate these measurements into a coherent picture of bioenergetic flux. These models are built on fundamental physicochemical principles and simulate how changes in one part of the system propagate through the rest. Analytical tools borrowed from engineering, such as flux analysis and control analysis, help identify which steps in the metabolic network exert the most control over overall output.³³ For the individual patient or research subject, this is not yet a bedside tool. But for drug development and for interpreting complex multi-omic datasets from mitochondrial disease cohorts, computational modeling fills the gap between measuring individual components and understanding the system as a whole.