Skip to main content
Browse Subject Areas

Click through the PLOS taxonomy to find articles in your field.

For more information about PLOS Subject Areas, click here.

  • Loading metrics

Quantitative in vivo mapping of myocardial mitochondrial membrane potential

  • Nathaniel M. Alpert ,

    Roles Conceptualization, Formal analysis, Investigation, Methodology, Writing – original draft, Writing – review & editing

    Affiliation Gordon Center for Medical Imaging, Massachusetts General Hospital, Harvard Medical School, Boston, Massachusetts, United States of America

  • Nicolas Guehl,

    Roles Data curation, Formal analysis, Methodology, Writing – review & editing

    Affiliation Gordon Center for Medical Imaging, Massachusetts General Hospital, Harvard Medical School, Boston, Massachusetts, United States of America

  • Leon Ptaszek,

    Roles Investigation, Methodology, Writing – review & editing

    Affiliation Cardiac Arrhythmia Service, Massachusetts General Hospital, Harvard Medical School, Boston, Massachusetts, United States of America

  • Matthieu Pelletier-Galarneau,

    Roles Investigation, Writing – review & editing

    Affiliation Gordon Center for Medical Imaging, Massachusetts General Hospital, Harvard Medical School, Boston, Massachusetts, United States of America

  • Jeremy Ruskin,

    Roles Methodology, Resources

    Affiliation Cardiac Arrhythmia Service, Massachusetts General Hospital, Harvard Medical School, Boston, Massachusetts, United States of America

  • Moussa C. Mansour,

    Roles Conceptualization, Methodology, Resources

    Affiliation Cardiac Arrhythmia Service, Massachusetts General Hospital, Harvard Medical School, Boston, Massachusetts, United States of America

  • Dustin Wooten,

    Roles Investigation, Methodology

    Affiliation Gordon Center for Medical Imaging, Massachusetts General Hospital, Harvard Medical School, Boston, Massachusetts, United States of America

  • Chao Ma,

    Roles Investigation, Methodology

    Affiliation Gordon Center for Medical Imaging, Massachusetts General Hospital, Harvard Medical School, Boston, Massachusetts, United States of America

  • Kazue Takahashi,

    Roles Investigation, Methodology

    Affiliation Gordon Center for Medical Imaging, Massachusetts General Hospital, Harvard Medical School, Boston, Massachusetts, United States of America

  • Yun Zhou,

    Roles Conceptualization, Methodology, Writing – review & editing

    Affiliation The Russell H. Morgan Department of Radiology and Radiological Science, School of Medicine Johns Hopkins University, Baltimore, Maryland, United States of America

  • Timothy M. Shoup,

    Roles Conceptualization, Methodology

    Affiliation Gordon Center for Medical Imaging, Massachusetts General Hospital, Harvard Medical School, Boston, Massachusetts, United States of America

  • Marc D. Normandin,

    Roles Methodology, Writing – review & editing

    Affiliation Gordon Center for Medical Imaging, Massachusetts General Hospital, Harvard Medical School, Boston, Massachusetts, United States of America

  • Georges El Fakhri

    Roles Funding acquisition, Investigation, Methodology, Project administration, Writing – review & editing

    Affiliation Gordon Center for Medical Imaging, Massachusetts General Hospital, Harvard Medical School, Boston, Massachusetts, United States of America


8 Feb 2018: The PLOS ONE Staff (2018) Correction: Quantitative in vivo mapping of myocardial mitochondrial membrane potential. PLOS ONE 13(2): e0192876. View correction



Mitochondrial membrane potential (ΔΨm) arises from normal function of the electron transport chain. Maintenance of ΔΨm within a narrow range is essential for mitochondrial function. Methods for in vivo measurement of ΔΨm do not exist. We use 18F-labeled tetraphenylphosphonium (18F-TPP+) to measure and map the total membrane potential, ΔΨT, as the sum of ΔΨm and cellular (ΔΨc) electrical potentials.


Eight pigs, five controls and three with a scar-like injury, were studied. Pigs were studied with a dynamic PET scanning protocol to measure 18F-TPP+ volume of distribution, VT. Fractional extracellular space (fECS) was measured in 3 pigs. We derived equations expressing ΔΨT as a function of VT and the volume-fractions of mitochondria and fECS. Seventeen segment polar maps and parametric images of ΔΨT were calculated in millivolts (mV).


In controls, mean segmental ΔΨT = -129.4±1.4 mV (SEM). In pigs with segmental tissue injury, ΔΨT was clearly separated from control segments but variable, in the range -100 to 0 mV. The quality of ΔΨT maps was excellent, with low noise and good resolution. Measurements of ΔΨT in the left ventricle of pigs agree with previous in in-vitro measurements.


We have analyzed the factors affecting the uptake of voltage sensing tracers and developed a minimally invasive method for mapping ΔΨT in left ventricular myocardium of pigs. ΔΨT is computed in absolute units, allowing for visual and statistical comparison of individual values with normative data. These studies demonstrate the first in vivo application of quantitative mapping of total tissue membrane potential, ΔΨT.


Mitochondria produce approximately 90% of cellular adenosine triphosphate (ATP) through oxidative phosphorylation [1]. The electron transport chain (ETC) of the mitochondrion is ultimately responsible for converting the foods we eat into electrical and chemical energy gradients by pumping protons across the inner membrane in the mitochondrial intermembrane space. The energy stored in the electric field, referred to as mitochondrial membrane potential (ΔΨm), is then used to power the conversion of ADP to ATP. In a typical cell, the ΔΨm remains constant with time and is about -140 mV [2]. Table 1 lists ΔΨm for mitochondria of different cell types.

Table 1. Measured values of mitochondrial membrane potential (milliVolts).

If ΔΨm remains within the physiological range, a small amount of reactive oxygen species (ROS) is produced. However, in mitochondrial dysfunction, ΔΨm falls outside the normal range, with concomitant increase in ROS release, and impairment of ATP production [10]. And because mitochondria are the most important source of energy and ROS in the cell, mitochondrial dysfunction is at the core of many diseases, including myopathies [11], diabetes [12], degenerative diseases [13], inflammation [14], cancer [15], and cardiac arrhythmias [16, 17].

Despite continuing scientific interest in voltage sensitive probes, a noninvasive method for measuring ΔΨm in living animals does not currently exist. The basic physiological studies conducted several decades ago are highly relevant but not always mentioned: Historically, fluorescent dyes [18] and lipophilic cationic tracers have been developed for quantitative assay of ΔΨm in isolated mitochondria [4], cells [19], and isolated heart preparations [6]. Electrodes sensitive to tetraphenylphosphonium (TPP) have also been developed and used to evaluate ΔΨm in isolated mitochondrial fractions [9, 20]. [14C or 3H]-labeled lipophilic cations were used to study the electrical properties of membranes and mitochondria long before modern imaging methods were imagined [3, 2123]. More recently, Logan et al. [24] reported a "click" method for in vivo measurement of ΔΨ in the cells of mouse hearts, but their method requires the excision of the heart and hence is not suitable for translation to human studies.

The work cited above established the use of 3H-TPP+ as a reference tracer for measuring mitochondrial membrane potential (ΔΨm). Investigators have shown that 3H-TPP+ distributes slowly in accord with the electrochemical gradient [4, 22]. Min et al suggested that TPP might be an important imaging agent [25].By replacing the tritium label with 18F[26], it is possible to adapt the methodology to PET, making it feasible to extend these measurements to intact animals and to human studies. In this paper, we report work that adapts the methods used in the earlier bench top measurements to in vivo imaging using positron emission tomography (PET/CT) to measure and map total membrane potential (ΔΨT) We define ΔΨT as the sum of ΔΨm and cellular (ΔΨc) electrical potentials. The time scale of our measurements are tens of minutes and thus ΔΨc is represented by its time average. ΔΨT is chosen as a practical surrogate for ΔΨm, keeping in mind that in most situations ΔΨm ≈ 10*ΔΨc and thus ΔΨT ≈ ΔΨm. In our methodology a cationic lipophilic tracer TPP+, labeled with 18F, is used to quantitatively map myocardial ΔΨT. 18F-TPP+ was initially developed as a myocardial flow imaging agent under the trade name BFPET [26] However, 18F-TPP+ enters the tissue with a low first-pass extraction fraction and does not respond to pharmacological challenge with a stressor and hence cannot be considered a flow tracer [7]. Nonetheless, its electrochemical properties make it a tracer of interest for quantitative imaging of ΔΨT.

Previous work, attempting to detect changes in concentration due to alteration of the ΔΨm used tracers such as 99mTc-sestamibi [27], tetraphenyl phosphonium[25], (18F-fluoropentyl) triphenylphosphonium[28], and 18F-fluorobenzyl triphenyl phosphonium [29] [30], with semi-quantitative endpoints such as SUV [31]. The results of these studies are empirical and descriptive. In the sense that a binary decision threshold is sought; the electrical properties of transmembrane kinetics are not exploited for quantitative purposes. Changes indicative of graded mitochondrial dysfunction cannot be detected. Gurm et al reported the first attempt to quantitatively measure ΔΨm with PET and 18F-TPP+ [7]. Their analysis simply applied the Nernst equation to the PET and plasma concentrations measured 30 minutes after bolus injection of 18F-TPP+. But terminating the study at 30 minutes was arbitrary and did not consider that the plasma level falls monotonically for at least 120 minutes, meaning that had they used the data at, say, 45 minutes after injection, they would have obtained a different result. Thus, their assumption that tracer was in steady state 30 minutes after bolus injection is incorrect and leads to biased results. In addition, their analysis did not consider the effect of tracer in the extracellular space. Because of these errors, their method violates basic tracer kinetic principles and their results significantly underestimate ΔΨm and are not in good agreement with work from in vitro studies (Table 1).

General design of the studies

This investigation provides an initial assessment of a method for quantitative mapping of ΔΨT. Because the mammalian heart has the highest concentration of mitochondria, we chose myocardial imaging as the first application to facilitate optimization of the scanning conditions. The subjects of our study were domestic swine (Sus domesticus) imaged in one of two conditions: healthy controls or pigs with chronic injury to the left anterior descending arterial (LAD) territory. We used bolus injection of 18F-TPP+ in control and injured pigs to measure the kinetics of TPP+ for at least two hours, determining the volume of distribution by a regression model.


A new method for quantification of membrane potential

We partition the tissue distribution of 18F-TPP+ into several components: ECS, consisting of interstitial space and plasma, mitochondria (mito) and cytosolic (cyto) volume fractions (Fig 1). The Nernst equation [32, 33] equates transmembrane electric potential to the ratio of ion concentrations on either side of the membrane. Thus, the Nernst equation allows us to derive an equation relating the PET measurements of 18F-TPP+ concentration, to ΔΨT, ECS fraction (fECS), mitochondrial volume fraction (fmito) and the electric potential across the cellular membrane, ΔΨc. At steady state, the concentration of 18F-TPP+ in a PET voxel can be written as (1) where, , and are the steady state concentrations of TPP+ in the mitochondria, cytosol and ECS, respectively. At steady state plasma and ECS are equal and and are related through the Nernst equations: (2)

Fig 1. Volume of distribution model for18F-TPP+ in a PET image voxel.

The outer black line represents the voxel boundary. Cp, Cinter, Ccyto, and Cmito represent the concentrations of the plasma, interstitial space, cytosol, and mitochondria respectively. The arrows represent 18F-TPP+ transport between the different compartments. fECS represents the voxel volume fraction occupied by ECS and fmito represents the cellular volume fraction occupied by mitochondria.

We divide Eq 1 by and use Eq 2 to express the tracer volume of distribution, VT, as (3) where is a ratio of known physical parameters: F denotes Faraday's constant, z is the valence, R is the universal gas constant and T is the temperature in degrees Kelvin. In our calculations z = 1, F = 96485.3 [Coulombs per mole], R = 8.314472, and T = 310.2 [degrees Kelvin]. is the steady-state ratio of the tissue to plasma concentrations. This equation predicts that VT, a kinetically determined tracer quantity, is equal to the steady state concentration ratio . Thus, when the tracer concentrations in tissue and plasma are time-invariant, VT is, by definition, the tissue-to-plasma concentration ratio; whereas, after bolus injection the tissue-to-plasma concentration ratio varies with time and VT must be determined kinetically.

Unlike tracers whose distribution is governed by passive transport, 18F-TPP+ will have a very high (>>1) volume of distribution when the membrane potential is in the normal range. A reasonable approximation to the volume of distribution is given by (4)

Barth et al. [34] have studied 10 different mammalian species, including pigs and man, finding that the myocardial fmito is a specific and constant value for any particular species. Therefore, VT is sensitive to two independent variables; ΔΨT and fECS. The fundamental mathematical relationships expressed in Eqs 1, 2 and 3 are depicted in Fig 2, illustrating that the systematic error in ΔΨT due to neglecting fECS for normal values of VT is about 20 mV. But note that Fig 2 also shows that at low membrane potential, the effect of the ECS is negligible, meaning that knowledge of fECS is most important for detecting normal versus mildly dysfunctional mitochondria.

Fig 2. Dependence of VT on mitochondrial membrane potential and fractional ECS computed with Eq 3.

ΔΨT is more negative inside membranes. Black lines indicate variation (systematic error) in ΔΨT range at constant volume of distribution due to neglecting fractional ECS volume.

Study cohort

18F-TPP+ scans were performed on eight Yorkshire swine (Pigs were all American Yorkshire male ordered from Animal BiowareTM Series II Sofware Suite, vendor: Tufts). There were 5 animals without tissue injury, referred to as control pigs, and 3 animals with a left anterior descending artery (LAD) infarction, referred to as injury pigs. Pigs shared a room with other animals of the same species, individually housed in their own cage, free to turn and make normal movements and postural adjustments. The animals were given play toys to enrich their environment and fed adequately, with ready access to water, to ensure normal growth. An acclimation period greater than 4 days was observed upon the animal’s arrival before conducting the first imaging procedure.

Animals used in this study were housed and maintained under the supervision of the Massachusetts General Hospital Animal Care and Use Committee and our study was conducted under a protocol approved by the Institutional Animal Care and Use Committee of the Massachusetts General Hospital.


Following a 12 hour fast, pigs were sedated with 4.4 mg/kg Telazol. Anesthesia was induced with isoflurane 5% and maintained with 1.5% isoflurane. During anesthesia, the animals were mechanically ventilated. Vital signs and depth of anesthesia were assessed once every 5 minutes and this assessment was documented every 15 minutes.

Tissue injury

Percutaneous central access was achieved via the Seldinger technique [35] under anesthesia. Femoral artery access was obtained and a guide wire was advanced into the LAD coronary artery. A balloon catheter was fed over the guidewire, placed in the mid LAD, and inflated to 6–8 atm for 80 minutes. An infarct was confirmed by the appearance of large ST elevation on ECG. At the end of the surgical procedure, subcutaneous Carprofen was administered and the animal was returned to housing for recuperation. After the immediate postoperative period, the animals were observed at least twice daily by study staff. Post-procedural Carprofen was administered orally for 3 days at a dose of 150 mg/day and then as needed if stiffness or swelling continued after that point. In order to prevent arrhythmias, 200mg Amiodarone and 50 mg Atenolol were also given orally every day during 2 weeks.

During the study, animal health and well-being, as well as the adequacy of anesthesia, were monitored by checking respiration rate, ECG, blood gas, corneal or palpebral reflex, blood pressure, heart rate, pulse oximetry.

Blood sampling

Radiotracer injection, iodinated contrast injection, gadolinium injection, and venous blood sampling were performed through femoral vein catheters. An arteriovenous shunt was placed in the left femoral artery for arterial blood sampling. After tracer injection, arterial blood samples were obtained every 10 seconds for the first 3 minutes, then at 1 minute intervals for 5 minutes, and at increasing intervals until 120 minutes post injection. Venous blood samples were obtained at 5, 10, 15, 30, 60 and 90 minutes after tracer injection. All blood samples were centrifuged to determine plasma and red cell concentration histories.


Eight scans, 5 control and 3 injury pigs, were performed using a Siemens Biograph 64 PET/CT. Following administration of 185 MBq of 18F-TPP+ as a single intravenous bolus, scanning was performed over 120 minutes in list mode. CT angiography was performed for anatomic reference. List mode data were framed as a dynamic series of 12x3, 9x5, 7x10, 15x30 second frames. PET/CT data were reconstructed using a filtered back projection algorithm with CT-based attenuation correction to yield a radioactivity concentration map in units of Bq/cc with 83 slices and a voxel size of 2.14x2.14x3 mm3. fECS was measured in 3 injury pigs with CT scanning using a bolus plus infusion iodinated contrast protocol [36].

Data analysis

Dynamic PET data were analyzed using a Logan regression method [37] to produce quantitative parametric images of the VT of 18F-TPP+. Images of VT were reoriented into the short axis projection and cropped so that the cardiac chambers occupied nearly all of the image space. Eq 3 was used to convert the parametric maps of VT to maps of ΔΨT, assuming ΔΨc = -15 mV [7] and fmito = 0.26 [34]. Segmental values are reported as a grand mean ΔΨT and its standard error of the mean (SEM). No background subtraction or thresholding was applied to the parametric images of VT.


In vivo mapping of ΔΨT

Fig 2 was computed using Eq 3 to show the predicted behavior of mitochondrial membrane potential as a function of total volume of distribution, and size of the extracellular space. ΔΨc was assumed to be -15 mV for these calculations. Fig 2 shows a nearly exponential behavior and that ΔΨT is nearly independent of fECS when membrane potential drops below about -100 mV.

As shown in Fig 3, after bolus injection, the plasma concentration history decreased rapidly during the "equilibration" phase and slowly thereafter; whereas, TPP demonstrated an extended myocardial residence time characterized by nearly constant or slowly decreasing tissue concentration. The concentration ratio for whole blood versus plasma became constant, with a mean value of 0.98±0.02 (SEM), about 15 minutes after injection of TPP+ (Fig 4). Venous and arterial samples obtained later than 10 minutes after injection of TPP+ were in good agreement.

Fig 3. Variation of plasma and tissue concentration following intravenous bolus injection of 18F-TPP+.

Plasma concentration decreases monotonically over the first two hours; whereas, myocardial concentration is nearly constant.

Fig 4. Whole blood-to-plasma concentration ratio measured as a function of time after bolus injection of 18F-TPP+.

After about 15 minutes, whole blood and plasma concentrations equilibrate with equal concentration.

18F-TPP+ concentration was highest in heart and liver, reaching a plateau in normal myocardium after about 10 minutes and declining very slowly thereafter. The uptake of 18F-TPP+ in scar was variable, reflecting the variation in degree and extent of injury. Tissue concentration in the injured area peaked about 10 minutes after injection, followed by a slow biphasic clearance, with the plateau level about 40% as high as in the normal myocardium. In normal left ventricular myocardium, average fECS = 0.20 and varied less than 10%. Local values of fECS could not be obtained in the tissue injury, due to poor signal-to-noise ratio in the CT-studies and average values were used in computation of ΔΨT.

TPP+ is avidly taken up by normal LV myocardium, as shown in Fig 5. The distribution of 18F-TPP+ is shown as SUV integrated from 60–120 min post tracer injection in the three oblique projections obtained directly from the PET image volume. No further processing was done to these images.

Fig 5. Typical 18F-TPP+ SUV image, integrated from 60–120 minutes after IV bolus injection.

SUV is highest in liver, followed by LV myocardium, with lower activity in the visible in bone marrow.

Representative parametric images of VT and ΔΨT derived from kinetic data obtained from a control pig and an injury pig were reoriented into the standard cardiac coordinate system and presented in Fig 6. Images of VT have units of cc-tissue/g-plasma; whereas, images of ΔΨT are in units of negative millivolts (-mV). The apparent lower volume of distribution in right ventricle and atria is artifactual, due to the thinner walls of those structures in combination with the effects of finite spatial resolution and cardiac motion blurring.

Fig 6. Parametric images of VT and ΔΨT.

Each panel of three x two images shows short axis, vertical and horizontal slices. Images of a representative control pig are shown in the left panel. Images of a representative scar pig are shown in the right panel. The top row of each panel depicts the TPP+ volume of distribution and bottom row the membrane potential.

Segmental ΔΨT is tightly grouped for the five control pigs with a grand mean ± SEM over 17 segments of -129.4±1.4 mV (Fig 7). Values of ΔΨT are lower in injured segments, particularly in the apical-septal segments corresponding to the injury in the territory of the LAD artery.

Fig 7. Comparison of ΔΨT in 17 "bull’s eye" segments Results shown for five control (blue squares) and three pigs with injury to the LAD territory (red circles).


In this paper, we introduce the concept of quantitative mapping of ΔΨT for monitoring mitochondrial status. Our development emphasizes measurement of the total membrane potential, ΔΨT, while making clear that ΔΨT is a proxy for and tightly correlated to ΔΨm. Direct measurements of ΔΨm require a separate measurement of the cellular membrane potential that is currently not possible in vivo. Just as in the early studies conducted with 3H-TPP+, our method relies on measuring the total concentration of a lipophilic cationic tracer which is then analyzed by using a compartment model of the tissue and the steady state formulation of the Nernst equation, the same equation that is fundamental to electrochemistry [38] and cardiac electrophysiology [39, 40].

The relation of tracer concentration, which varies, and the physiological steady state can seem confusing. In the physiological state all the biological concentrations, transport rates and electropotentials are assumed to be fixed; whereas, the tracer concentrations evolve over time during the experiment (Fig 3). This complicates the application of the Nernst equation, whose use assumes the concentrations determining the membrane potential are invariant with time. We addressed that issue by using a kinetic model with extracellular and intracellular pools measured with dynamic PET to estimate the total volume of tracer distribution, a time invariant quantity characteristic of the tracer and the animal under study. We explicitly considered the effect of the electric field across the inner membrane of the mitochondrion on the kinetics of TPP+ by using a kinetic model to express the relation between the total volume of distribution of the tracer and the electrical properties of the membrane. The structure of this model is identical to that used to analyze the classic 3H-TPP+ experiments conducted many years ago.

While qualitative imaging may serve many purposes, the quantitative aspects of our method may provide additional important information about mitochondrial status not available from visual inspection of images, true because mitochondrial dysfunction may, in some tissues, be associated with uniformly reduced ΔΨm and such conditions cannot be detected by qualitative imaging methods.

Another potentially important result of this work is the more analytic understanding of the factors affecting the distribution of TPP+, in particular, and voltage sensing tracers in general. We used basic principles of physics and physiology to show how the volume of distribution of lipophilic cations depends on the volume fractions of the tissue occupied by the ECSand the mitochondria as well as on the magnitude of ΔΨT. VT is approximately equal to . Keeping in mind that for normal tissue the mitochondrial membrane potential is about -140 mV [2] for all mitochondria regardless of tissue type; whereas, the fractional mitochondrial volume varies by more than a factor 10, Eqs 3 and 4 imply that the intensity of the VT image will reflect the mitochondrial volume fraction of the tissue, thereby explaining the intensity variations depicted in Fig 5. Eqs 3 and 4 also make clear that the size of the ECS is an important factor when interpreting TPP+ images because this quantity may vary with age and disease. Fig 2 shows the model-prediction for the volume of distribution when we fix the mitochondrial volume fraction while varying the membrane potential and the fractional volume of the ECS. This result shows that increases in the size of the ECShave decreasing effect on the total volume of distribution as the membrane potential decreases and becomes depolarized. But if the goal is to detect more subtle changes from normal ΔΨT it is important to include the effect of variation in ECS. Failure to account for changes in ECS may lead to unexplained variability in the qualitative and quantitative assessments of voltage-sensing tracer distribution.

ΔΨm is sustained by the electron transport chain of the mitochondria, by which a balance is struck between protons pumped across the inner mitochondrial membrane and those pumped back to power the synthesis of ATP. ΔΨm is also affected by changes in the level of ROS and various mitochondrial ion channels [41]. Interestingly, increased ROS levels and modulation of mitochondrial ion channel function are seen early in numerous pathologies. Thus, the ability to quantitatively map ΔΨT may be useful for diagnosing or managing a number of such conditions. For example, reduction in ΔΨm has been implicated as a mechanism underlying ventricular arrhythmogenesis [42] and so might be used to improve the detection of arrhythmogenic foci.

We noted in Introduction that there are currently no methods for measuring ΔΨm in animal or human. This means that there is no independent method against which we can compare our PET methods. In this regard, it is important to emphasize that our noninvasive PET method is strongly related to the highly invasive and validated bench-top methods which preceded it. All prior methods employing in vitro application of cationic tracers were based on the Nernst equation to relate steady state concentrations and membrane potential in a compartment model of the mitochondrion, cell or tissue. Similarities to PET can be seen with 3H-TPP+ studies conducted in populations of cells and isolated mitochondria, where concentrations of TPP+ in the medium are related to the concentration in ensembles of cells or mitochondria by the Nernst equation. Measurements with TPP+ electrodes are also based on the same principles [9]. The similarity is most apparent in the work of Wan et al. [6] who studied ΔΨm in isolated rat hearts by using the arterio-venous difference in concentration of 3H-TPP+ to infer the tissue-to-perfusate concentration ratio. In essence, measuring the evolution of 18F-TPP+ concentration in a PET ROI is completely analogous to measurements with 3H-TPP+ in the isolated perfused rat heart studies of Wan et al. [6]. Hence, the PET studies are the natural extension of the classical bench top reference methods.

The fact that our PET measurements of ΔΨT are in accord with measurements in isolated rat hearts is an important observation, especially since direct validation of our method is impossible with existing methodologies. As illustrated in Table 1, the PET method yields results that are close to those found by the majority of prior studies, thereby providing additional support of our method.

With high specific activity, the mass of 18F-TPP+ is 10−10 to 10−7 lower than the intracellular potassium concentration, meaning its effect on membrane potential is negligible. However, it is worth mentioning that prior studies with 3H-TPP+ have sometimes made corrections for non-specific binding of TPP+ [4] but the literature is not concordant on the necessity of correction [4, 9, 4345]. Furthermore, previous bench top studies used very low specific activity TPP+, complicating an evaluation of its necessity in tracer measurements. Nevertheless, similar corrections could be applied to the PET analyses, but were found to be unnecessary to establish initial validity.

Close inspection of Fig 5 shows there is a high value of VT in normal myocardium, with mitochondrial concentrations nearly 30 times the plasma level. At secular equilibrium, the inward and outward fluxes across the mitochondrial membrane have to balance, meaning that outward clearance must be about 30 times lower than the inward rate, thereby explaining the slow clearance observed experimentally after bolus injection of 18F-TPP+. Fig 5 also shows that the volume of distribution in injured myocardium is much lower, with tissue concentrations less than 10 times the plasma concentration. Our kinetic model shows that VT depends linearly on fECS and exponentially on ΔΨT. Accordingly, we have converted VT to an estimate of ΔΨT by accounting for the effects of fECS, a quantity known to vary in age and disease [4649]. Thus, we have shown it will be necessary to account for changes in extracellular volume fraction in clinical studies to obtain the full benefit of such studies.

A marked difference in contrast between VT and ΔΨT was observed between normal and injured tissue (Fig 5). Both VT and ΔΨT are quantitative measures emphasizing different aspects of 18F-TPP+ distribution. On one hand, VT is the total volume of 18F-TPP+ distribution, including effects due to ΔΨT, fECS, and fmito. Therefore, mitochondrial density (fmito) will affect the relative uptake of a voltage-sensing tracer. This finding is interesting given that different pathologies are associated with reduced mitochondrial density. For example, a reduction in mitochondrial density is seen in the skeletal muscles of patients with chronic obstructive pulmonary disease [50]. In these conditions, VT measurements would provide an overall quantitative measure of mitochondrial status in tissue. On the other hand, ΔΨT focuses predominantly on the electrochemical conditions that prevail at the inner mitochondrial membrane. In the large tissue injury, we see that there is an area of profound reduction in ΔΨT, approaching total depolarization. Other parts of the injured region show lesser depolarization in the range of -60 mV, but still very different than ΔΨT in normal myocardium. This can also be appreciated from the data in Fig 7, where the blue circles indicate major reductions in the average ΔΨT for bull’s eye segment 8, 9, 14, 15 and 17. We can also see the variability of tissue injury expressed in Fig 7, which demonstrate a patchy nature to those injuries that might be better appreciated by direct examination of the parametric images.

As also shown in Fig 7, pigs in the control group exhibited segmental values of ΔΨT that averaged about -129.4 ± 1.4 mV (SEM). The tight grouping of ΔΨT measurements over all normal segments and pigs is remarkable. The kinetic approach, used in this study, requires a long measurement period for accurate estimation of the total volume of distribution. A protocol using primed constant infusion may be preferred for human investigation in order to restrict actual scan time to the equilibrium period, 95–120 min.

In this work we reported the results of kinetic analysis using the Logan graphical method, which is known to underestimate VT with increasing statistical noise in the measurements[51]. We mitigated the effect of statistical noise by averaging ΔΨT maps for 17 polar segments but the reader should be aware that the spatial averaging also causes some underestimation of ΔΨT. We also formed parametric images using the reGP method of Zhou et al [52] and the total least squares approach of Varga and Szabo,[53] but these methods yielded noisy parametric images; overall, they were not an improvement over the Logan plot.


This study is the first to demonstrate the feasibility of quantitative in vivo mapping of total membrane potential, ΔΨT, a proxy of ΔΨm. In vivo measurements of ΔΨT obtained with our new method yielded values remarkably constant within and across the hearts of domestic swine that are comparable to results from in vitro bench top experiments. We have derived a theory explaining, for the first time, the major factors affecting the transport and residence time of lipophilic cations in tissue, including ΔΨT and fECS. The fact that we can measure ΔΨT in mV suggests that it may be possible to compare individual's studies with normative data. Given the critical role of mitochondrial function in numerous pathologies, the potential applications of this new imaging method are immense. This novel technique could eventually be proved useful in numerous clinical and research scenarios.


The authors thank Kevin Cordaro, Victoria Douglas and Julia Scotton for animal preparation, handling and monitoring. We are also grateful to Dr. Moses Wilks and Dr. Eline Verwer for their help during experimental measurements. We also thank Henry Gewirtz, M.D. for helpful discussions.


  1. 1. Rich P. Chemiosmotic coupling: the cost of living. Nature. 2003;421(6923):583–. pmid:12571574
  2. 2. Alberts B, Johnson A, Lewis J, Walter P, Raff M, Roberts K. Molecular Biology of the Cell 4th Edition: International Student Edition. Routledge; 2002.
  3. 3. Kauppinen RA HI. Monitoring of mitochondrial membrane potential in isolated perfused rat heart. Am J Physiol. 1984; 247(508–516).
  4. 4. Rottenberg H. Membrane potential and surface potential in mitochondria: uptake and binding of lipophilic cations. The Journal of membrane biology. 1984;81(2):127–38. pmid:6492133
  5. 5. LaNoue KF, Strzelecki T, Strzelecka D, Koch C. Regulation of the uncoupling protein in brown adipose tissue. Journal of Biological Chemistry. 1986;261(1):298–305. pmid:3941078
  6. 6. Wan B, Doumen C, Duszynski J, Salama G, Vary TC, LaNoue KF. Effects of cardiac work on electrical potential gradient across mitochondrial membrane in perfused rat hearts. American Journal of Physiology. 1993;265:H453–H. pmid:8368348
  7. 7. Gurm GS, Danik SB, Shoup TM, Weise S, Takahashi K, Laferrier S, et al. 4-[18F]-tetraphenylphosphonium as a PET tracer for myocardial mitochondrial membrane potential. JACC Cardiovascular imaging. 2012;5(3):285–92. pmid:22421174
  8. 8. Gerencser AA, Chinopoulos C, Birket MJ, Jastroch M, Vitelli C, Nicholls DG, et al. Quantitative measurement of mitochondrial membrane potential in cultured cells: calcium†induced de†and hyperpolarization of neuronal mitochondria. The Journal of physiology. 2012;590(12):2845–71. pmid:22495585
  9. 9. Kamo N, Muratsugu M, Hongoh R, Kobatake Y. Membrane potential of mitochondria measured with an electrode sensitive to tetraphenyl phosphonium and relationship between proton electrochemical potential and phosphorylation potential in steady state. The Journal of membrane biology. 1979;49(2):105–21. pmid:490631
  10. 10. Huttemann M, Lee I, Pecinova A, Pecina P, Przyklenk K, Doan JW. Regulation of oxidative phosphorylation, the mitochondrial membrane potential, and their role in human disease. Journal of bioenergetics and biomembranes. 2008;40(5):445–56. pmid:18843528
  11. 11. Pfeffer G, Chinnery PF. Diagnosis and treatment of mitochondrial myopathies. Annals of medicine. 2013;45(1):4–16. pmid:21867371
  12. 12. De Felice FG, Ferreira ST. Inflammation, defective insulin signaling, and mitochondrial dysfunction as common molecular denominators connecting type 2 diabetes to Alzheimer disease. Diabetes. 2014;63(7):2262–72. pmid:24931033
  13. 13. Pohanka M. Alzheimer s disease and oxidative stress: a review. Current medicinal chemistry. 2014;21(3):356–64. pmid:24059239
  14. 14. O'rourke B. Metabolism: Beyond the power of mitochondria. Nature Reviews Cardiology. 2016;13(7):386–8. pmid:27303815
  15. 15. Summerhayes IC, Lampidis TJ, Bernal SD, Nadakavukaren JJ, Nadakavukaren KK, Shepherd EL, et al. Unusual retention of rhodamine 123 by mitochondria in muscle and carcinoma cells. Proc Natl Acad Sci U S A. 1982;79(17):5292–6. pmid:6752944
  16. 16. Nunnari J, Suomalainen A. Mitochondria: in sickness and in health. Cell. 2012;148(6):1145–59. pmid:22424226
  17. 17. Rutledge C, Dudley S. Mitochondria and arrhythmias. Expert Rev Cardiovasc Ther. 2013;11(7):799–801. pmid:23895021
  18. 18. Chen LB. Mitochondrial membrane potential in living cells. Annual review of cell biology. 1988;4(1):155–81.
  19. 19. Johnson LV, Walsh ML, Bockus BJ, Chen LB. Monitoring of relative mitochondrial membrane potential in living cells by fluorescence microscopy. The Journal of Cell Biology. 1981;88(3):526–35. pmid:6783667
  20. 20. Moreno AnJ Santos DL, Magalhaes Novais S Oliveira PJ. Measuring Mitochondrial Membrane Potential with a Tetraphenylphosphonium Selective Electrode. Current Protocols in Toxicology. 2015:25.5. 1-.5. 16.
  21. 21. J Ritchie R. A critical assessment of the use of lipophilic cations as membrane potential probes. Progress in biophysics and molecular biology. 1984;43(1):1–32. pmid:6374760
  22. 22. Kauppinen R. Proton electrochemical potential of the inner mitochondrial membrane in isolated perfused rat hearts, as measured by exogenous probes. Biochimica et Biophysica Acta (BBA)-Bioenergetics. 1983;725(1):131–7.
  23. 23. Lichtshtein D, Kaback HR, Blume AJ. Use of a lipophilic cation for determination of membrane potential in neuroblastoma-glioma hybrid cell suspensions. Proceedings of the National Academy of Sciences of the United States of America. 1979;76(2):650–4. pmid:284390
  24. 24. Logan A, Pell VR, Shaffer KJ, Evans C, Stanley NJ, Robb EL, et al. Assessing the mitochondrial membrane potential in cells and in vivo using targeted click chemistry and mass spectrometry. Cell metabolism. 2016;23(2):379–85. pmid:26712463
  25. 25. Min J-J, Biswal S, Deroose C, Gambhir SS. Tetraphenylphosphonium as a novel molecular probe for imaging tumors. Journal of Nuclear Medicine. 2004;45(4):636–43. pmid:15073261
  26. 26. Shoup TM, Elmaleh DR, Brownell AL, Zhu A, Guerrero JL, Fischman AJ. Evaluation of (4-[18F]Fluorophenyl)triphenylphosphonium ion. A potential myocardial blood flow agent for PET. Mol Imaging Biol. 2011;13(3):511–7. pmid:20563755
  27. 27. Kawamoto A, Kato T, Shioi T, Okuda J, Kawashima T, Tamaki Y, et al. Measurement of technetium-99m sestamibi signals in rats administered a mitochondrial uncoupler and in a rat model of heart failure. PloS one. 2015;10(1):e0117091. pmid:25594546
  28. 28. Kim D-Y, Kim H-S, Le UN, Jiang SN, Kim H-J, Lee K-C, et al. Evaluation of a mitochondrial voltage sensor,(18F-fluoropentyl) triphenylphosphonium cation, in a rat myocardial infarction model. Journal of Nuclear Medicine. 2012;53(11):1779–85. pmid:23038748
  29. 29. Madar I, Huang Y, Ravert H, Dalrymple SL, Davidson NE, Isaacs JT, et al. Detection and quantification of the evolution dynamics of apoptosis using the PET voltage sensor 18F-fluorobenzyl triphenyl phosphonium. Journal of Nuclear Medicine. 2009;50(5):774–80. pmid:19372481
  30. 30. Madar I, Ravert H, Nelkin B, Abro M, Pomper M, Dannals R, et al. Characterization of membrane potential-dependent uptake of the novel PET tracer 18F-fluorobenzyl triphenylphosphonium cation. Eur J Nucl Med Mol Imaging. 2007;34(12):2057–65. pmid:17786439
  31. 31. Kim D-Y, Kim HS, Reder S, Zheng JH, Herz M, Higuchi T, et al. Comparison of 18F-Labeled Fluoroalkylphosphonium Cations with 13N-NH3 for PET Myocardial Perfusion Imaging. Journal of Nuclear Medicine. 2015;56(10):1581–6. pmid:26069304
  32. 32. Nernst W. Die elektromotorische wirksamkeit der jonen. Zeitschrift für physikalische Chemie. 1889;4(1):129–81.
  33. 33. Nernst W. Experimental and theoretical applications of thermodynamics to chemistry: C. Scribner's sons; 1907.
  34. 34. Barth E, Stämmler G, Speiser B, Schaper J. Ultrastructural quantitation of mitochondria and myofilaments in cardiac muscle from 10 different animal species including man. Journal of molecular and cellular cardiology. 1992;24(7):669–81. pmid:1404407
  35. 35. Seldinger SI. Catheter replacement of the needle in percutaneous arteriography: a new technique. Acta radiologica. 1953;39(5):368–76. pmid:13057644
  36. 36. Bandula S, White SK, Flett AS, Lawrence D, Pugliese F, Ashworth MT, et al. Measurement of myocardial extracellular volume fraction by using equilibrium contrast-enhanced CT: validation against histologic findings. Radiology. 2013;269(2):396–403. pmid:23878282
  37. 37. Logan J, Fowler JS, Volkow ND, Wolf AP, Dewey SL, Schlyer DJ, et al. Graphical analysis of reversible radioligand binding from time—activity measurements applied to [N-11C-methyl]-(−)-cocaine PET studies in human subjects. Journal of Cerebral Blood Flow & Metabolism. 1990;10(5):740–7.
  38. 38. Bockris JOM, Reddy AK. Modern electrochemistry: an introduction to an interdisciplinary area: Springer Science & Business Media; 2012.
  39. 39. Whalley DW, Wendt DJ, Grant AO. Basic concepts in cellular cardiac electrophysiology: Part I: Ion channels, membrane currents, and the action potential. Pacing and Clinical Electrophysiology. 1995;18(8):1556–74. pmid:7479177
  40. 40. Iwai S, Markowitz SM, Lerman BB. Electrophysiology of Cardiac Arrhythmias. Essential Cardiology: Springer; 2013. p. 261–75.
  41. 41. O'Rourke B, Cortassa S, Aon MA. Mitochondrial ion channels: gatekeepers of life and death. Physiology. 2005;20(5):303–15.
  42. 42. Brown DA, O'rourke B. Cardiac mitochondria and arrhythmias. Cardiovascular research. 2010;88(2):241–9. pmid:20621924
  43. 43. Aiuchi T, Matsunaga M, Nakaya K, Nakamura Y. Calculation of membrane potential in synaptosomes with use of a lipophilic cation (tetraphenylphosphonium). Chemical and Pharmaceutical Bulletin. 1989;37(12):3333–7. pmid:2632080
  44. 44. Demura M, Kamo N, Kobatake Y. Mitochondrial membrane potential estimated with the correction of probe binding. Biochimica et Biophysica Acta (BBA)-Bioenergetics. 1987;894(3):355–64.
  45. 45. Labajova A, Vojtiskova A, Krivakova P, Kofranek J, Drahota Z, Houstek J. Evaluation of mitochondrial membrane potential using a computerized device with a tetraphenylphosphonium-selective electrode. Analytical biochemistry. 2006;353(1):37–42. pmid:16643832
  46. 46. Wong TC, Piehler KM, Kang IA, Kadakkal A, Kellman P, Schwartzman DS, et al. Myocardial extracellular volume fraction quantified by cardiovascular magnetic resonance is increased in diabetes and associated with mortality and incident heart failure admission. European heart journal. 2012:eht193.
  47. 47. Ho CY, Abbasi SA, Neilan TG, Shah RV, Chen Y, Heydari B, et al. T1 measurements identify extracellular volume expansion in hypertrophic cardiomyopathy sarcomere mutation carriers with and without left ventricular hypertrophy. Circulation: Cardiovascular Imaging. 2012;6(3):415–22.
  48. 48. Sado DM, Flett AS, Banypersad SM, White SK, Maestrini V, Quarta G, et al. Cardiovascular magnetic resonance measurement of myocardial extracellular volume in health and disease. Heart. 2012:heartjnl-2012-302346.
  49. 49. Schelbert EB, Testa SM, Meier CG, Ceyrolles WJ, Levenson JE, Blair AJ, et al. Myocardial extravascular extracellular volume fraction measurement by gadolinium cardiovascular magnetic resonance in humans: slow infusion versus bolus. Journal of Cardiovascular Magnetic Resonance.13(1):16.
  50. 50. Liu S-F, Kuo H-C, Tseng C-W, Huang H-T, Chen Y-C, Tseng C-C, et al. Leukocyte mitochondrial DNA copy number is associated with chronic obstructive pulmonary disease. PloS one. 2015;10(9):e0138716. pmid:26394041
  51. 51. Slifstein M, Laruelle M. Effects of statistical noise on graphic analysis of PET neuroreceptor studies. Journal of Nuclear Medicine. 2000;41(12):2083–8. pmid:11138696
  52. 52. Zhou Y, Ye W, Brasic JR, Crabb AH, Hilton J, Wong DF. A consistent and efficient graphical analysis method to improve the quantification of reversible tracer binding in radioligand receptor dynamic PET studies. Neuroimage. 2009;44(3):661–70. pmid:18930830
  53. 53. Jz Varga, Szabo Z. Modified regression model for the Logan plot. Journal of Cerebral Blood Flow & Metabolism. 2002;22(2):240–4.