Lipophilic monocations can pass through phospholipid bilayers and accumulate in negatively-charged compartments such as the mitochondrial matrix, driven by the membrane potential. This property is used to visualize mitochondria, to deliver therapeutic molecules to mitochondria and to measure the membrane potential. In theory, lipophilic dications have a number of advantages over monocations for these tasks, as the double charge should lead to a far greater and more selective uptake by mitochondria, increasing their therapeutic potential. However, the double charge might also limit the movement of lipophilic dications through phospholipid bilayers and little is known about their interaction with mitochondria. To see whether lipophilic dications could be taken up by mitochondria and cells, we made a series of bistriphenylphosphonium cations comprising two triphenylphosphonium moieties linked by a 2-, 4-, 5-, 6- or 10-carbon methylene bridge. The 5-, 6- and 10-carbon dications were taken up by energized mitochondria, whereas the 2- and 4-carbon dications were not. The accumulation of the dication was greater than that of the monocation methyltriphenylphosphonium. However, the uptake of dications was only described by the Nernst equation at low levels of accumulation, and beyond a threshold membrane potential of 90–100 mV there was negligible increase in dication uptake. Interestingly, the 5- and 6-carbon dications were not accumulated by cells, due to lack of permeation through the plasma membrane. These findings indicate that conjugating compounds to dications offers only a minor increase over monocations in delivery to mitochondria. Instead, this suggests that it may be possible to form dications within mitochondria that then remain within the cell.
Lipophilic cations, such as TPMP (methyltriphenylphosphonium), easily cross phospholipid bilayers and accumulate within the negatively-charged mitochondrial matrix [1,2]. This ability to pass through biological membranes is due to the distribution of charge over a large, hydrophobic surface area . The extent of uptake of these cations in response to the membrane potential (ΔΨ) is described by the Nernst equation, where z is the charge on the cation .
At 37 °C the factor before the logarithmic term for monocations is 61 mV, therefore as the mitochondrial membrane potential in vivo is approx. 120–180 mV , lipophilic monocations accumulate 102–103-fold within mitochondria . The mitochondrial accumulation of lipophilic monocations is used extensively to measure membrane potential and to visualize mitochondria in cells [5,6].
We have conjugated TPP (triphenylphosphonium) cations with antioxidants and other probes to direct them selectively to mitochondria within cells and in vivo, and one such compound is now in phase II trials for Parkinson's disease [4,7–11]. As well as promising potential therapies, these compounds are useful tools to explore mitochondrial oxidative damage [12–14]. We therefore sought to improve the extent and selectivity of mitochondrial delivery by using lipophilic dications, which offer the theoretical prospect of being dramatically more effective at delivering antioxidants to mitochondria in vivo. This is because the Nernst equation indicates that a dication should accumulate 100-fold within mitochondria for every 60 mV increase in membrane potential (in contrast with 10-fold for a monocation ), giving a 102–103-fold increase in uptake over a monocation. Such an increase would be a significant improvement in designing mitochondria-targeted therapies.
Little is known about the interaction of lipophilic dications with mitochondria, and therefore there is a need to explore their mitochondrial uptake. Furthermore, the limits to the Nernstian accumulation of these cations by mitochondria have not been tested and dications offer the possibility of exploring these limits. Only a few studies have been performed on the interaction of lipophilic dications with mitochondria [16–20], and in none of these investigations was the dependence of dication uptake on membrane potential determined, so it is unclear whether dication accumulation is described by the Nernst equation.
It also remains unclear whether the rate of uptake of a dication across the mitochondrial inner membrane will be limiting. Lipophilic cations pass readily across membranes because their interaction with the hydrophobic core of the membrane has a favourably high neutral energy (proportional to the hydrophobic surface area of the cation) and a favourably low Born energy (proportional to the square of the dication charge, and inversely proportional to the cation radius [3,21,22]). However, the Born energy for a dication will be four times greater than that of a monocation of the same radius, so a dication will have to be more hydrophobic than a related monocation to ensure a comparable rate of uptake.
To determine whether lipophilic dications were accumulated to a greater extent than monocations in the present study we used bisTPP dications comprising two TPP cations separated by alkyl linkers of various lengths. We hypothesized that uptake would depend on the length of the alkyl link because separating the cations sufficiently might facilitate their independent interaction with the membrane, so that the Born energy would be two, rather than four, times greater than that of a comparable monocation. Furthermore, the greater hydrophobicity as the alkyl chain is lengthened should lower the activation energy for transport. Therefore a dication series of two TPP moieties attached by 2-, 4-, 5-, 6- and 10-carbon methylene bridges was constructed. We then determined whether these molecules were taken up by energized mitochondria in accordance with the Nernst equation, and whether they were accumulated by mitochondria within cells.
Known bisTPP alkyl dications were prepared by refluxing dihalides with triphenylphosphine in acetonitrile for 24 h followed by evaporation and trituration with diethyl ether unless otherwise stated. Detailed descriptions of the syntheses are given in the online Supplementary Material that can be found at http://www.BiochemJ.org/bj/400/bj4000199add.htm.
Partition coefficients and solubility
Octan-1-ol/PBS partition coefficients were measured as described previously . The partition coefficient of [3H]PC5P (where PCnP is a bistriphenylphosphonium cation linked by an n carbon methylene bridge) was determined by adding 1 μM of compound to 1 ml of PBS and 1 ml of octan-1-ol, followed by incubation at room temperature for 30 min. After centrifugation (1000 g for 4 min), the amount of compound in each phase was determined by scintillation counting. To assess solubility, the PC5P dication (20–50 mg) was added to PBS or double-distilled water (50–400 μl) at 22 °C, dissolved by extensive vortex-mixing, and any undissolved compound pelleted for 5 min at 12000 g in a benchtop microcentrifuge. A sample of supernatant was diluted in PBS or double-distilled water, as appropriate, to a final A268 of 0.1–0.5, and quantified spectrophotometrically.
Preparation and incubation of mitochondria
Rat liver mitochondria were prepared at 4 °C in 250 mM sucrose, 5 mM Tris/HCl and 1 mM EGTA (pH 7.4) , or in 250 mM sucrose, 5 mM Hepes and 1 mM EGTA (pH 7.2), neutralized with TMA-OH (tetramethylammonium hydroxide) for the 86Rb+ (rubidium-86) uptake experiments. The protein concentration was determined by the biuret assay using BSA as a standard . To establish the concentrations at which bisTPP dications disrupted mitochondria, we measured their effects on respiration rate and membrane potential. To measure respiration, mitochondria (2 mg protein·ml−1) were suspended in 120 mM KCl, 10 mM Hepes (pH 7.2), 1 mM EGTA and 10 mM potassium phosphate supplemented with rotenone (4 μg·ml−1) at 25 °C in the 3 ml chamber of an oxygen electrode (Rank Brothers, Bottisham, Cambridge, U.K.). BisTPP dications or DMSO carrier were added and coupled respiration on succinate (10 mM) was measured followed by respiration in the presence of ADP (200 μM). There was negligible effect of the dications on respiration at concentrations below 1 μM. Slight inhibition of State III respiration began at 2–5 μM for the PC6P and PC10P dications, and at 5–10 μM for the other dications. To measure membrane potential, mitochondria (2 mg protein·ml−1) were suspended in 250 μl of KCl medium (as above) supplemented with 500 nM TPMP and [3H]TPMP (50 nCi·ml−1). After 5 min respiring on succinate, mitochondria were pelleted, and 200 μl of supernatant was removed to a scintillation vial. The remaining medium was aspirated off and the pellet was dried with a tissue and resuspended in 40 μl of 20% (v/v) Triton X-100 and transferred to a scintillation vial. Scintillant (3 ml) was added and the radioactivity was quantified. The membrane potential was estimated using the Nernst equation, assuming a mitochondrial volume of 0.5 μl·mg protein−1 and that 60% of intramitochondrial TPMP is membrane-bound . Minor effects of PC6P and PC10P were noticeable at 2.5–5 μM, while concentrations of 5–10 μM were required for disruption by the other compounds. Yeast mitochondria were prepared from lactate-grown yeast as described in  and stored at −80 °C in 0.6 M sorbitol, 20 mM Hepes/KOH (pH 7.4), supplemented with 10 mg·ml−1 BSA. On thawing, these mitochondria maintain the same membrane potential as freshly isolated mitochondria, as measured by TPMP accumulation by mitochondria respiring on ethanol.
Measurement of mitochondrial uptake of bisTPP dications using an ion-selective electrode
A series of TPP ion-selective electrodes was constructed from a PVC membrane incorporating a TPB ion-exchanger and filled with an aqueous solution of the cation (10 mM for TPMP and PC2P, 1 mM for other dications) [27,28]. The electrode and a Ag/AgCl reference electrode were inserted through the air-tight Perspex lid of a stirred incubation chamber thermostatically controlled at 30 °C and containing 3 ml of buffer comprising 120 mM KCl, 1 mM EGTA and 10 mM Hepes (pH 7.2) which was supplemented with nigericin (500 nM) and rotenone (4 μg·ml−1). Log[dication or cation] versus voltage was linear (R2 0.992–0.999). For TPMP, the electrode response gave a slope of 62.9 mV/decade. For the PC4P, PC5P and PC6P dications the slopes were 27.6–34.9 mV/decade, while PC2P and PC10P gave slopes of 22.2 mV/decade and 37.6 mV/decade respectively.
Uptake of [3H]TPMP, [3H]PC5P and [3H]PC6P by isolated mitochondria
To investigate mitochondrial dication uptake, mitochondria (2 mg protein·ml−1) were added to 250 μl of 120 mM KCl, 10 mM Hepes, 1 mM EGTA and 100 μM diethylenetriaminepentaacetic acid (pH 7.2) with 4 μg·ml−1 rotenone, 300 nM nigericin and 1 μM [3H]PC5P, [3H]PC6P or [3H]TPMP (7, 48 and 100 nCi·ml−1 respectively) at 30 °C. After 1 min, 10 mM succinate was added and at various times mitochondria were pelleted by centrifugation (1 min at 16000 g) and the [3H] content of the supernatant (200 μl) and the mitochondrial pellet determined by scintillation counting. The accumulation of cation within mitochondria was expressed as the apparent pellet space (μl.mg protein−1) using eqn 2:
These values were not corrected for non-mitochondrial pellet [3H] which is negligible at high cation accumulation . To measure the simultaneous accumulation of 86Rb+ and [3H]cations, mitochondria (2 mg protein·ml−1) were incubated in 250 μl of 250 mM sucrose, 5 mM Hepes and 1 mM EGTA (pH 7.2) supplemented with 4 μg·ml−1 rotenone, 0.2 nmol valinomycin·mg protein−1, 10 mM succinate (TMA salt), 86RbCl (5 nCi·ml−1) and 1 μM [3H]PC5P, [3H]PC6P or [3H]TPMP (16, 16 and 20 nCi·ml−1 respectively). Experiments using 200 nM and 40 nM [3H]PC5P were carried out at 3.2 and 0.64 nCi·ml−1 respectively. The membrane potential was varied by KCl (0–20 mM) or FCCP (carbonylcyanide-p-trifluoromethoxyphenylhydrazone; 0.5 μM) ±20 mM KCl. After 5 min mitochondria were pelleted and processed for scintillation counting as above. The 86Rb+ and [3H] DPM contents of the supernatant and pellets were then determined by dual label scintillation counting. The amount of [3H] compound taken up by mitochondria was calculated from the specific activity of the incubation mixture. In parallel experiments the nonmitochondrial space in the pellets was determined by incubation with [14C]sucrose (50 nCi·ml−1) over the range of KCl concentrations . The sucrose pellet space was 8.1±0.9 μl·mg protein−1 (means±S.D.), did not vary significantly with [KCl], and was used to correct the apparent pellet spaces for non-mitochondrial 86Rb+ or [3H]. These values were then divided by the mitochondrial volume under these conditions (0.6 μl·mg protein−1; M. P. Murphy and M. D. Brand, unpublished work) to give the ACR (accumulation ratio) :
Yeast mitochondria were washed in buffer [0.6 M mannitol, 0.5 mM EDTA, 10 mM Tris/maleate, 5 mM inorganic phosphate (pH 6.8)] and suspended at 1 mg protein·ml−1 at 30 °C in 250 μl of the same buffer supplemented with ethanol (5.7 mM), valinomycin (0.2 nmol·mg protein−1), [3H]PC5P (1 μM; 16 nCi·ml−1) and 86Rb+ (5 nCi·ml−1) and after 3 min incubation the pellet and supernatant were assessed as above. ACRs were estimated by correcting for the non-mitochondrial pellet space using [14C]sucrose data from liver mitochondria and assuming a yeast mitochondrial volume of 1.8 μl·mg protein−1 (referred to in ). The data were qualitatively unchanged if the correction for non-mitochondrial pellet volume was omitted.
Cell culture and incubations
To measure dication toxicity, Jurkat cells were seeded in duplicate at 1×105 cells per well in 2 ml of RPMI medium containing 10% FCS (foetal calf serum) and compounds were added from DMSO stock solutions and incubated for 24 h. Cell death was assessed by propidium iodide uptake by flow cytometry using the Calbiochem Annexin V-FITC Apoptosis Kit. Neither PC5P, PC6P nor TPMP up to 5 μM were toxic and showed similar toxicity up to 100 μM. To measure the uptake of [3H]cations, Jurkat cells (106 or 107 cells·ml−1) were incubated as above in medium supplemented with 1 μM [3H]TPMP, [3H]PC5P or [3H]PC6P (15, 6 or 10 nCi·ml−1 respectively). At various times 1 ml samples were taken, cells were pelleted by centrifugation (16000 g for 1 min) and the amount of [3H] compound in the pellet measured by scintillation counting as described above. To correct for cell growth, the cell densities of parallel incubations with unlabelled cations were determined at each time point. Over 23 h the cell density increased by approx. 100% for 1×106 cells·ml−1 or approx. 40% for 1×107 cells·ml−1. To measure PC10P uptake, cell-associated dication was extracted into 90% acetonitrile/0.1% trifluoroacetic acid, analysed by reverse phase HPLC (C18) on an acetonitrile gradient and detected at 220 nm. PC10P was quantified by comparison with standards of known concentration.
Human fibroblasts were grown overnight in DMEM (Dulbecco's modified Eagle's medium) containing 10% FCS, then incubated with fresh medium supplemented with 1 μM TPMP (20 nCi·ml−1) or 1 μM PC5P (16 nCi·ml−1) for 5 h at 37 °C. The medium was removed, cells were rapidly washed with 500 μl PBS and then resuspended by scraping in 200 μl of 1% Triton X-100. The [3H] content of the cells was determined by scintillation counting.
Preparation of BLM (bilayer planar phospholipid membranes)
BLM were formed on the 1 mm aperture of a Teflon septum bisecting the experimental chamber. To form BLM, Escherichia coli phospholipids (57% phosphatidylethanolamine, 15% phosphatidylglycerol, 10% cardiolipin and 18% others) were dissolved in decane (20 mg·ml−1). Measurements were carried out using AgCl electrodes and a VA-J-5 electrometer, and analysed using a personal computer with the help of L-card L-154 and homemade software . To measure the generation of a membrane potential by lipophilic cations, both compartments contained 100 nM NaTPB in 50 mM Tris/HCl (pH 7.4) at 22 °C. The concentration of cation in one compartment was fixed at 1 μM while that in the other compartment was varied from 1 μM to 100 μM and the voltage determined. Under these conditions the tetraphenylphosphonium cation gave a slope of 60 mV/decade when the voltage was plotted against the log[cation], as predicted by the Nernst equation.
RESULTS AND DISCUSSION
Synthesis and partition coefficients of bisTPP dications
The series of bisTPP dications synthesized and their octan-1-ol/PBS partition coefficients are shown in Figure 1. The partition coefficients for PC2P–PC6P ranged from 0.41 to 0.84, marginally more hydrophobic than the simple TPP monocation, TPMP (partition coefficient=0.35; ). Even the most hydrophobic dication, PC10P, has a partition coefficient of only 12.8, despite the 10-carbon chain linking the two cations. This contrasts with the TPP monocation, decylTPP, which has a partition coefficient of 5000 . Therefore bisTPP dications are significantly less hydrophobic than related TPP monocations.
Structures and octan-1-ol/PBS partition coefficients of bisTPP dications
Chain length dependence of bisTPP dication uptake by energized mitochondria
To see if bisTPP dications were accumulated by energized mitochondria we used ion-selective electrodes specific for the TPP moiety [27,31]. In the first set of experiments, mitochondria were added to the incubation chamber in the presence of the respiratory inhibitor rotenone to prevent membrane potential formation, as shown by no uptake of TPMP in its presence (results not shown). The electrode response was then calibrated by 5×1 μM additions of dication (Figures 2A–2E). The respiratory substrate succinate was then added to generate a membrane potential, with a decrease in dication concentration indicating uptake into mitochondria. Energization led to rapid mitochondrial accumulation of PC5P, PC6P and PC10P (Figures 2C–2E). The subsequent addition of the uncoupler FCCP abolished the membrane potential, releasing the dications from the mitochondria. In contrast, neither PC2P nor PC4P (Figures 2A and 2B) were accumulated by energized mitochondria.
Mitochondrial uptake of bisTPP dications measured using an ion-selective electrode
In the next set of experiments, dications were incubated with the lipophilic anion TPB, which facilitates the movement of lipophilic cations through biological membranes [31,32]. The PC4P dication was internalized rapidly by energized mitochondria when substoichiometric amounts of TPB were added, whereas the PC2P dication was not (Figures 2F and 2G). This suggests that TPB lowers the activation energy sufficiently to allow passage of the PC4P dication through the membrane. In summary, there is a clear dependence of uptake by energized mitochondria on the length of the alkyl chain connecting the two TPP moieties.
Quantification of TPMP, PC5P and PC6P accumulation by energized mitochondria
To quantify uptake, we prepared [3H]PC5P and [3H]PC6P and compared their uptake by energized mitochondria with that of the monocation [3H]TPMP (Figures 3A and 3B). This was performed by measuring the apparent pellet space (the amount of compound per mg of protein in the mitochondrial pellet divided by the amount of compound per μl of supernatant; expressed in μl·mg protein−1). Dividing the apparent pellet space by the mitochondrial volume per mg of protein gives the ACR, the ratio of the compound concentrations in the mitochondria and the supernatant, which is used in the Nernst equation. Incubating [3H]PC5P or [3H]PC6P with energized mitochondria led to their accumulation over 2–3 min, at which point they came to similar steady-state apparent pellet spaces of approx. 8000–9000 μl·mg protein−1. The uptake of the bisTPP dications was prevented, and reversed, by FCCP (Figures 3A and 3C), confirming the extensive membrane potential-dependent dication uptake seen in electrode traces (Figure 2). The uptake of the monocation [3H]TPMP was also determined and in Figure 3(A) this is shown on the same scale as the dications, and on an expanded scale in Figure 3(B). The effect of FCCP added 5 min after TPMP addition is also shown (Figure 3D). From these experiments it is clear that the uptake of the bisTPP dications is significantly greater than that of TPMP: the maximum apparent pellet space for TPMP was approx. 600 μl·mg protein−1, while the maximum apparent pellet space for the dications was 15-fold greater at approx. 9000 μl·mg protein−1.
Accumulation of [3H]TPMP, [3H]PC5P and [3H]PC6P by energized mitochondria
The increase in apparent pellet spaces for the dications over that for the monocation is far less than expected from the Nernst equation, which predicts that the dication ACR should be the square of that for the monocation, giving an apparent pellet space of approx. 360000 μl·mg protein−1. The measured apparent pellet spaces for the dications are approx. 40-fold less than this. However, comparisons of TPMP and the bisTPP dication uptake from separate experiments are not definitive as there may be differences in membrane potential and intramitochondrial binding. To overcome these difficulties we determined the mitochondrial ACR of [3H]PC5P, [3H]PC6P or [3H]TPMP simultaneously with that of 86Rb+ (Figure 4). This procedure was chosen because 86Rb+ is not thought to bind within mitochondria and therefore its uptake in the presence of the ionophore valinomycin is solely determined by the membrane potential . Dual isotope counting enables the apparent pellet space of 86Rb+ and of [3H]cations to be determined in the same incubation and this measurement can be repeated over a range of membrane potentials set by various amounts of KCl [25,33,34]. Parallel measurement of the non-mitochondrial space in the pellet from the distribution of [14C]sucrose enables non-mitochondrial 3H and 86Rb+ in the pellet to be corrected for, and knowledge of the mitochondrial volume enables the true ACRs for [3H]cations to be calculated over a range of known membrane potentials.
Simultaneous accumulation of 86Rb+ and [3H]TPMP, [3H]PC5P or [3H]PC6P by energized mitochondria
In Figure 4(A) the ACR of [3H]TPMP is plotted against that of 86Rb+ over a range of membrane potentials, giving the expected straight line . The uptake of TPMP is Nernstian up to the highest 8Rb+ ACR of approx. 700, which corresponds to a membrane potential of approx. 170 mV. The slope of 2.34 indicates that about 57% of intramitochondrial TPMP is membrane-bound in agreement with previous reports . These experiments were repeated for [3H]PC5P and [3H]PC6P, and the dication ACRs were plotted against [86Rb+ ACR]2, which should give a straight line if the dications accumulate in accordance with the Nernst equation (Figures 4B and 4C). For both [3H]PC5P and [3H]PC6P the dication ACR rises steeply at first with [86Rb+ ACR]2, but there is a clear discontinuity at about 95–100 mV (corresponding to a [Rb+ ACR]2 of 1500–2200), above which the relative increases in the dication ACRs are far less. The insets to Figures 4(B) and 4(C) show that at membrane potentials below about 100 mV, the dication ACRs are linear functions of the square of the [86Rb+ ACR]2, therefore below the discontinuity dication uptake is Nernstian. The slopes of the Nernstian regions of uptake indicate that within mitochondria about 80% of the accumulated dications are membrane bound, compared with approx. 57% bound for TPMP, which is consistent with their relative hydrophobicities.
The uptake of [3H]TPMP increases exponentially with membrane potential, in accordance with its Nernstian uptake characteristics (Figure 4D). In contrast, [3H]PC5P uptake increases dramatically at low values of membrane potential, but above about 90–100 mV the further increase in uptake with membrane potential is small. At low membrane potentials dication accumulation is about 5-fold greater than that of the monocation, but this drops to less than 2-fold greater at higher membrane potential.
Why is dication accumulation non-Nernstian at high membrane potentials?
These unexpected findings raise the question of why dication accumulation becomes non-Nernstian above about 100 mV. This discontinuity is not due to any effect of dications on the membrane potential, as the simultaneous measurement of the 86Rb+ ACR shows that the membrane potential is unaffected. One possible limitation to dication uptake is that their extensive adsorption to the inner membrane may saturate all available binding sites so that further dications taken up cannot be membrane-bound and remain free in solution. If this were the case then the slope of the dication ACR against the [86Rb+ ACR]2 would become 1 upon saturation. However, this does not occur and instead above the discontinuity the slopes are 0.01–0.02 (Figure 4).
The high dication concentration within mitochondria may inhibit subsequent uptake, perhaps due to excessive dication adsorption on the membrane affecting further uptake, or to the limited dication solubility in the matrix. The former possibility was appealing as uptake requires the cation to jump from adsorption sites on the outside of the membrane to similar sites on the inside of the membrane, which if saturated inhibit transport . If either of these possibilities led to the discontinuity then lowering or raising the dication concentration should shift the discontinuity to a higher or lower membrane potential. However, when the [3H]PC5P ACR against [86Rb+ ACR]2 was measured over a range of [3H]PC5P concentrations from 40 nM to 5 μM, all gave similar plots and for all concentrations the discontinuity occurred between 96 and 103 mV (Figure 5A and 5B). Furthermore, below the discontinuity the plots were linear and had similar slopes for all concentrations, indicating that the proportion of accumulated dication bound to the inner membrane was concentration independent (Figure 5B). In addition, we measured the solubility of the PC5P dication in PBS and water, giving values of approx. 100 mM and approx. 500 mM respectively at room temperature (22 °C). The shift to non-Nernstian behaviour is occurring well below these concentrations, since at the highest concentration of PC5P used (5 μM) the maximum possible matrix concentration was 4.2 mM under our experimental conditions. This makes it unlikely that insolubility of the dications in the matrix is contributing to the observed discontinuity.
Effect of lipophilic dication concentration and TPB on [3H]PC5P ACR
To explore further the possibility that dication binding to the membrane affected subsequent cation uptake, we measured the effect of 5 μM unlabelled PC5P on [3H]TPMP uptake (Figure 5C). The effect of PC5P on the relationship between the 86Rb+ and [3H]TPMP ACRs was negligible. As both TPMP and PC5P are expected to bind to the same region of the membrane, this suggests that PC5P does not saturate lipophilic cation binding sites on the matrix surface of the mitochondrial inner membrane. We next determined whether dication accumulation somehow increased the activation energy sufficiently so that above the discontinuity further uptake was inhibited. TPB increases lipophilic cation uptake by lowering the activation energy for transport. However, the presence of TPB slightly lowered [3H]PC5P uptake and the discontinuity occurred at the same membrane potential (Figure 5D). An alternative explanation for the discontinuity is that high dication accumulation induces a specific efflux pathway, so that uptake beyond a certain level is prevented. Energetic considerations suggest that such a process would have to be either driven directly by ATP hydrolysis, or occur electroneutrally. Furthermore, this putative pathway would have to be activated at the same membrane potential for a wide range of dication concentrations, so it is unlikely to simply be a carrier with a high Km for the dication. Even so, we investigated whether a putative dication efflux pathway might exist by measuring the effect of a series of mitochondrial transport inhibitors on the relationship between the [3H]PC5P ACR and [86Rb+ ACR]2 (Figure 6A). Inhibitors were chosen against a range of mitochondrial transport processes: calcium uniporter (Ruthenium Red), Na+/H+ exchanger (amiloride), ornithine carrier (ornithine), Ca2+/Na+ exchanger (CGP-37157), permeability transition (cyclosporin A) and the adenine nucleotide carrier (bongkrekic acid and carboxyatractyloside); N-ethylmaleimide (NEM) inhibits many mitochondrial transport activities and, finally, inhibition of intramitochondrial ATP synthesis with oligomycin was used to decrease ATP-dependent export. If an efflux pathway existed then its inhibition should increase the [3H]PC5P ACR dramatically at high membrane potentials. None of the inhibitors increased the [3H]PC5P ACR (Figure 6A), and the data conformed to the line used to describe the dependence of the [3H]PC5P ACR on [86Rb+ ACR]2 from Figure 4(B), indicating that any inhibitor effects were solely due to changes in the membrane potential. This finding makes it unlikely that the discontinuity is due to induction of a catalytic dication efflux pathway.
Effect of transport inhibitors and different types of mitochondria on [3H]PC5P uptake
Finally, to see if the discontinuity in dication uptake was a general property of mitochondria, we repeated this experiment in yeast mitochondria. Yeast mitochondria accumulated [3H]PC5P in a similar fashion to liver mitochondria, with uptake increasing sharply with membrane potential at low potentials, before plateauing at about 100 mV (Figure 6B). Although there were quantitative differences between dication uptake by yeast and mammalian mitochondria, this result suggests that the discontinuity in lipophilic dication uptake is a general property of mitochondria.
In summary, we have eliminated a number of possible explanations for the non-Nernstian accumulation of lipophilic dications at high membrane potentials. To see whether this discontinuity was a general property of phospholipid bilayers, we measured the diffusion potential established by varying the dication concentration gradient across a BLM. The diffusion potential was Nernstian (approx. 30 mV for every 10-fold transmembrane concentration difference) up to a concentration gradient of approx. 50–60-fold, but beyond this point there was only a limited increase in diffusion potential as the concentration gradient increased up to 400-fold (results not shown). This suggests that non-Nernstian distribution of dications at higher membrane potentials may be a general property of phospholipid bilayers.
Uptake of TPMP, PC5P and PC6P by cells
We next investigated whether there was greater uptake of PC5P and PC6P over TPMP by mitochondria within Jurkat cells . [3H]TPMP was taken up into cells and this uptake was largely abolished by FCCP, myxothiazol or gramicidin (results not shown). We next compared the uptake of [3H]TPMP with that of [3H]PC5P and [3H]PC6P over 23 h, and found that there was extensive uptake of [3H]TPMP, but not of [3H]PC5P or [3H]PC6P (Figures 7A and 7B). The lack of even minor dication uptake was confirmed by addition of FCCP, which did not affect the amount of dication associated with the cell pellets (Figures 7A and 7B). [3H]TPMP was also taken up by human fibroblasts in an FCCP-dependent fashion, but [3H]PC5P was not (results not shown). Therefore neither [3H]PC5P nor [3H]PC6P are taken up into cells.
Accumulation of [3H]TPMP, [3H]PC5P and [3H]PC6P by cells
To see whether TPB could facilitate dication uptake into cells, we incubated cells with 500 nM TPB for 5 h and found that TPB did not increase uptake of [3H]PC5P or [3H]PC6P (Figure 7C). However, a large excess of TPB (10 μM) did lead to the FCCP-sensitive uptake of [3H]PC5P and [3H]PC6P by cells, indicating some uptake into mitochondria. This suggests that the PC5P and PC6P dications are not accumulated into cells because they cannot cross the plasma membrane. To test the possibility that the hydrophobicity of the PC5P and PC6P dications was too low to enable cellular uptake, we also compared the uptake of PC5P and the PC10P dication into Jurkat cells. In accordance with previous data (Figures 7A and 7B), only a very small amount of PC5P (approx. 3 pmol/106 cells) was cell-associated, and this association was not FCCP-sensitive. In contrast, PC10P was taken up into cells (12±2 pmol/106 cells; means±range) over 4 h, with little further increase over 24 h, and about 50% of this uptake was ΔΨm-dependent. These findings imply that the plasma membrane is a greater activation energy barrier to dication transport than the mitochondrial inner membrane, possibly because the dielectric constant is higher within the mitochondrial inner membrane due to its greater protein content [3,36]. However, cellular uptake of PC10P, and that of PC5P and PC6P in the presence of TPB, only reached levels similar to that of TPMP, implying that even when the plasma membrane barrier was overcome, dication uptake by cells was not greater than that of the monocation. This lack of enhancement of cellular uptake suggests that lipophilic dications are unlikely to be more useful than monocations as mitochondrial delivery vectors in vivo.
BisTPP dication uptake into energized mitochondria showed clear chain-length dependence, with only dications linked by a 5-carbon chain or longer being taken up. A plausible explanation for this is the differing surface electrostatic potentials of the dications, which are shown in Supplementary Figure 1 at http://www.BiochemJ.org/bj/400/bj4000199add.htm. For PC2P the double charge is distributed over the whole surface of the molecule, while as the chain length increases, the surface potentials separate into two discrete areas, resulting in a lower Born energy. In addition, as the chain length increases, the overall dication surface potential decreases, increasing hydrophobicity (Figure 1). Together these will decrease the activation energy for transport. Therefore lipophilic dications can be accumulated by energized mitochondria; this is favoured by separating the charges and the dications need to be more hydrophobic than a comparable monocation.
Dication uptake was only adequately modelled by the Nernst equation at membrane potentials up to approx. 90–100 mV. This finding limits the utility of lipophilic dications as vectors for the delivery of compounds such as antioxidants to mitochondria. Even so, dications are accumulated to a significantly greater extent than monocations and as this differential is particularly large at low membrane potentials, lipophilic dications could be used to deliver compounds to, and measure membrane potential in, isolated mitochondria at low membrane potentials.
There was no uptake of the PC5P and PC6P dications into cells, and even when the plasma membrane barrier was overcome by addition of TPB, or by increasing dication hydrophobicity (PC10P), the uptake was no greater than that of TPMP. These findings suggest that it may be possible to form dications inside mitochondria in vivo that then remain within the cell. Since mitochondria in vivo rapidly accumulate orally-administered lipophilic monocations , simultaneous administration of two monocations containing functionalities that enable them to link together within mitochondria should lead to formation of a dicationic compound that would then be locked within the cell. This approach may prove to be a useful way to make probes for mitochondrial function in vivo or to prevent efflux of drugs or toxins from cells.
This work was supported by a Postgraduate Scholarship from the University of Otago, Dunedin, New Zealand and the European Community's Sixth Framework Programme for Research, Priority 1, ‘Life sciences, genomics and biotechnology for health’ contract number LSHM-CT-2004-503116. We thank Dr Martin D. Brand (MRC Dunn Human Nutrition Unit, Cambridge, U.K.) for helpful comments and suggestions on the manuscript and Helena Cochemé (MRC Dunn Human Nutrition Unit, Cambridge, U.K.) for providing yeast mitochondria.
bilayer planar phospholipid membrane
mitochondrial membrane potential
foetal calf serum
bistriphenylphosphonium cation linked by an n carbon methylene bridge
tetraphenyl borate anion
These authors contributed equally to this work.