Dual Effect of Phosphate Transport on Mitochondrial Ca²⁺ Dynamics

Background: Inorganic phosphate (P_i) buffers matrix Ca²⁺, but its impact on mitochondrial Ca²⁺ handling is often overlooked.

Results: Mitochondrial Ca²⁺ uptake and buffering strictly depend on anion transport rates; ATP accelerates P_i-independent Ca²⁺ uptake.

Conclusion: Maximal Ca²⁺ uniporter rate and Ca²⁺ buffering are anion transport limited; ATP alters influx without being hydrolyzed.

Significance: P_i transport is fundamentally important in controlling mitochondrial Ca²⁺ signals.

Abstract

The large inner membrane electrochemical driving force and restricted volume of the matrix confer unique constraints on mitochondrial ion transport. Cation uptake along with anion and water movement induces swelling if not compensated by other processes. For mitochondrial Ca²⁺ uptake, these include activation of countertransporters (Na⁺/Ca²⁺ exchanger and Na⁺/H⁺ exchanger) coupled to the proton gradient, ultimately maintained by the proton pumps of the respiratory chain, and Ca²⁺ binding to matrix buffers. Inorganic phosphate (P_i) is known to affect both the Ca²⁺ uptake rate and the buffering reaction, but the role of anion transport in determining mitochondrial Ca²⁺ dynamics is poorly understood. Here we simultaneously monitor extra- and intra-mitochondrial Ca²⁺ and mitochondrial membrane potential (ΔΨ_m) to examine the effects of anion transport on mitochondrial Ca²⁺ flux and buffering in P_i-depleted guinea pig cardiac mitochondria. Mitochondrial Ca²⁺ uptake proceeded slowly in the absence of P_i but matrix free Ca²⁺ ([Ca²⁺]_mito) still rose to ∼50 μm. P_i (0.001–1 mm) accelerated Ca²⁺ uptake but decreased [Ca²⁺]_mito by almost 50% while restoring ΔΨ_m. P_i-dependent effects on Ca²⁺ were blocked by inhibiting the phosphate carrier. Mitochondrial Ca²⁺ uptake rate was also increased by vanadate (V_i), acetate, ATP, or a non-hydrolyzable ATP analog (AMP-PNP), with differential effects on matrix Ca²⁺ buffering and ΔΨ_m recovery. Interestingly, ATP or AMP-PNP prevented the effects of P_i on Ca²⁺ uptake. The results show that anion transport imposes an upper limit on mitochondrial Ca²⁺ uptake and modifies the [Ca²⁺]_mito response in a complex manner.

Introduction

Mitochondrial Ca²⁺ plays a central role in regulating cellular energy supply, redox balance and cell death (1, 2) yet many questions remain about the determinants of mitochondrial Ca²⁺ flux, including the full complement of molecular components (2), signaling pathways affecting influx or efflux (2), and mitochondrial free Ca²⁺ concentrations during low and high work conditions. Accurate determination of matrix Ca²⁺ is particularly important in the heart, where workload continuously varies, accompanied by dynamic changes in ATP hydrolysis rate, Ca²⁺, and phosphate (P_i).²

P_i transport through the phosphate carrier (PiC) is fundamental for oxidative phosphorylation and affects the pH gradient across the inner membrane with the effective co-transport of protons. A less appreciated function of P_i is its obligatory role in modulating mitochondrial Ca²⁺ fluxes. The earliest descriptions of mitochondrial Ca²⁺ uptake/binding (3–5) noted that ATP, P_i, and Mg²⁺ could potentiate Ca²⁺ accumulation by mitochondria, but the mechanism by which these modulators acted was unclear. P_i has been reported to affect mitochondrial Ca²⁺ in several ways, including modulation of Ca²⁺ influx, efflux, and buffering. The conventional view is that P_i transport facilitates Ca²⁺ influx by dissipating the pH gradient that builds up during Ca²⁺ entry as the protonmotive force redistributes from ΔΨ_m to ΔpH upon stimulation of proton pumping (6). After entry, P_i then participates in dynamically buffering matrix Ca²⁺ by the formation of Ca-P_i complexes (7), to clamp mitochondrial Ca²⁺ ([Ca²⁺]_mito) at a fixed level (8). This not only permits further Ca²⁺ uptake but also renders efflux through the mitochondrial Na⁺/Ca²⁺ exchanger (mNCE) independent of total matrix Ca²⁺ load (7). The complex nature and impact of P_i, nucleotides, or other anions on mitochondrial Ca²⁺ dynamics is incompletely understood. Because of the restricted volume of the mitochondrial matrix, compensation for the electrophoretic entry of cations is a thermodynamic requirement. The PiC is regarded as an antiporter of H₂PO₄⁻/OH⁻, equivalent to a symport of H₂PO₄⁻/H⁺ (9, 10) and co-transport of protons has been noted to be an important feature of anions that are most effective at facilitating mitochondrial Ca²⁺ uptake (11, 12).

In a recent study in which PiC (SLC25A3) was knocked out in a cardiac-specific manner (13), mitochondrial permeability transition pore (mPTP) activation by Ca²⁺ was decreased and Ca²⁺-induced cell death was blunted. However, in that study, some residual P_i uptake was present, and it was not determined if the maximal mitochondrial Ca²⁺ uniporter (MCU) rate in the absence of mPTP opening was altered. In contrast, another recent study showed no effect of knockdown of SLC25A3 on mitochondrial Ca²⁺ loading capacity or mPTP sensitivity but an increased cardiac hypertrophic response to pressure overload (14). Other P_i transporters also exist on the mitochondrial inner membrane. The Ca²⁺-activated ATP-Mg/P_i carrier, SLC25A23, exchanges MgATP for P_i (15), and, in a recent study, it was proposed that SLC25A23 interacts with MCU and another accessory protein, MICU1, to regulate mitochondrial Ca²⁺ uptake, as evidenced by reduced mitochondrial Ca²⁺ uptake after knockdown of SLC25A23 (16).

Several modes of mitochondrial Ca²⁺ uptake have been reported in previous work. In addition to the main MCU-mediated uptake, a rapid uptake mode (RaM), activating and inactivating during rapid increases in Ca²⁺ near the mitochondria has been reported (17, 18). In addition, we have noted two slow MCU modes of uptake in cardiac mitochondria that displayed different Ca²⁺ affinities and Ru360 sensitivities (8), which were also modulated differentially by P_i. Mode 1 had a high Ca²⁺ affinity and low Ru360 sensitivity, while mode 2, comprising the bulk of mitochondrial Ca²⁺ uptake, was inhibited by low Ru360 concentrations (8). Extramitochondrial P_i enhanced mitochondrial Ca²⁺ accumulation through mode 1 but resulted in almost complete buffering of Ca²⁺ entry through mode 2. Interestingly, we found that mPTP opening was independent of [Ca²⁺]_mito but was dependent on total mitochondrial Ca²⁺ load, leading us to postulate that Ca-P_i complex accumulation was the trigger for activation of the pore.

To better understand the role of P_i in modulating mitochondrial Ca²⁺ dynamics, here we utilize P_i-depleted, isolated cardiac mitochondria to directly study the role of anions in mitochondrial Ca²⁺ uptake and buffering while monitoring changes in ΔΨ_m. We show that mitochondrial Ca²⁺ uptake occurs through both P_i-independent and P_i-dependent pathways, the latter of which could be blocked by inhibition of the PiC. Substitution of other anions led to differential effects on Ca²⁺ uptake and buffering, and revealed a novel action of adenine nucleoside trisphosphates in a non-hydrolytic role.

Experimental Procedures

Guinea pig heart mitochondria were isolated using a homogenization protocol on ice, as previously described (19). The homogenized heart suspension was centrifuged in the isolation solution (75 mm sucrose, 225 mm mannitol, 1 mm HEPES, and 1 mm EGTA, pH 7.4) at 480 g for 10 min at 4 °C, and the resulting supernatant was centrifuged at 7,700 × g for 10 min. The mitochondrial pellet was washed at 7,700 × g for 6 min and re-suspended in isolation solution at a concentration of ∼20 mg/ml. The protein concentration of isolated mitochondria was determined using the BCA (bicinchoninic acid assay) method.

Endogenous phosphate in the isolated cardiac mitochondria was depleted by utilizing the hexokinase reaction with 0.75 units/ml hexokinase, 1 mm glucose, 0.5 mm ADP, 1 mm MgCl₂, and 5 mm glutamate/malate in 37 °C for 5 min, as previously described (20).

Mitochondria (0.5 mg) were suspended in 2 ml of a potassium-based buffer solution without phosphate (137 mm KCl, 20 μm EGTA, 20 mm HEPES, 5 mm NaCl, 1 μm CsA, and 5 mm glutamate/malate at pH 7.2). Multiple mitochondrial parameters were simultaneously monitored using a wavelength-scanning fluorometer (QuantaMaster; Photon Technology International) at room temperature. The extra-mitochondrial calcium (Ca²⁺_out) was measured with 50 μm CaGreen-5N, hexapotasssium salt (505 nm ex.: 535 nm em.). Fura-FF-AM (20 μm fura-FF was incubated for 30 min at room temperature beforehand) was used to monitor intra-mitochondrial calcium (Ca²⁺_mito) (340 nm/380 nm ex.: 510 nm em.). Mitochondrial membrane potential (ΔΨ_m) was monitored by the ratiometric method with 300 nm tetramethylrhodamine methyl ester (TMRM) (546 nm/573 nm ex.: 590 nm em.), and mitochondria swelling was monitored by 90 degree light scattering (540 nm ex.), as previously described (8, 21).

To measure mitochondrial pH, the succinimidyl ester form of carboxy-SNARF-1 (Life Technologies, Inc.) was loaded into Pi-depleted isolated mitochondria (incubating 30 μm for 30 min at room temperature), then washed twice and resuspended in sucrose-based isolation solution. During the measurement (520 nm ex.: 580/680 nm em.), 0.5 mg of mitochondria were suspended in 2 ml of KCl-based buffer solution without phosphate. The SNARF-1 signal was calibrated in buffer solutions with different pH values (pH 6–8) in the presence of 1 μm FCCP, 20 μg oligomycin, and 10 μm nigericin.

Results

Effects of Acute P_i Addition on Mitochondrial Ca²⁺ Dynamics

In zero P_i buffer solution, mitochondrial Ca²⁺ uptake from the cuvette upon addition of 35 μm Ca²⁺ (which increases free Ca²⁺ in the cuvette to 7 μm after binding to Ca²⁺ buffers present in the solution; Fig. 1A; 120–240 s) was slow and incomplete; nevertheless, [Ca²⁺]_mito increased monotonically to 50 μm. Acute addition of 1 mm phosphate to the buffer (at 240 s; black trace) resulted in an immediate acceleration of Ca²⁺ uptake from the cuvette, but, paradoxically, a decrease in [Ca²⁺]_mito to ∼27 μm (Fig. 1B; black trace). The P_i effect on Ca²⁺ uptake was concentration-dependent and approached a maximum rate at 0.1 mm phosphate. The EC₅₀ for the P_i effect on the initial linear Ca²⁺ uptake rate was 0.087 mm (Fig. 1E, blue trace). The P_i effect on buffering, reflected in an increase in the ratio of bound/free matrix Ca²⁺ (Fig. 1D), had a similar P_i dependence, as did the rate of formation of the Ca²⁺-P_i complex (EC₅₀ = 0.096 mm; Fig. 1E, green trace). Mitochondrial membrane potential (ΔΨ_m) was depolarized from ∼215 mV to 170 mV in the presence of Ca²⁺ without P_i (Fig. 1C), but when phosphate was added, the decline in [Ca²⁺]_out was paralleled by restoration of ΔΨ_m to 210mV. The rate of recovery of ΔΨ_m was linearly related to mitochondrial Ca²⁺ uptake rate (Fig. 1F). Mitochondrial light scattering decreased slightly upon P_i addition, indicating low amplitude mitochondrial swelling but no activation of the mitochondrial permeability transition pore (mPTP) during the experiments (data not shown).

FIGURE 1.

Concentration-dependent effects of P_i on mitochondrial Ca²⁺ uptake and buffering in cardiac mitochondria. A single Ca²⁺ addition (35 μm total CaCl₂ or ∼7 μm free calcium) to a suspension of P_i-depleted guinea pig cardiac mitochondria results in a slow and limited uptake from the cuvette (A:[Ca²⁺]_out), while matrix Ca²⁺ (B:[Ca²⁺]_mito) rises with fast and slow components. Addition of various P_i concentrations (0.001–1 mm NaH₂PO₄) accelerated Ca²⁺ uptake (A) but decreased [Ca²⁺]_mito (B), reflecting an increase in matrix bound:free Ca²⁺ (D). Mitochondrial inner membrane potential (ΔΨ_m) depolarizes and recovers in parallel with the Ca²⁺ and P_i additions, respectively (C). E, concentration dependence of the P_i effect on Ca²⁺ uptake (blue) and buffering (green) rates, with saturation at ∼0.1 mm P_i. F, linear relationship between the rate of change of [Ca²⁺]_out and ΔΨ_m measured as the first derivative of the signals for first 15 s after the P_i addition. (Data in panels A-E represent mean ± S.E. for n = 3 experiments; panel F, n = 5).

Next, we examined the P_i effect on Ca²⁺ uptake for different Ca²⁺ additions (20–40 μm) and after adding a fixed concentration of P_i (0.1 mm; Fig. 2). Recordings of [Ca²⁺]_out (Fig. 2A), [Ca²⁺]_mito (Fig. 2B), and ΔΨ_m (Fig. 2C) reveal that the P_i-dependent buffering effect takes effect above a certain threshold of matrix Ca²⁺ is reached. In the absence of P_i, [Ca²⁺]_mito rises to a range of 20–80 μm for total Ca²⁺ additions between 20 and 40 μm (extramitochondrial Ca²⁺ range of 1–7 μm). When phosphate is added, [Ca²⁺]_mito is clamped to the same steady state level (∼30 μm) for any [Ca²⁺]_out larger than 2 μm (Fig. 2B). The degree of membrane potential depolarization was proportional to the amount of Ca²⁺ added to the cuvette, in accord with the size of the electrochemical gradient for Ca²⁺ across the inner membrane, the major determinant of Ca²⁺ current (Fig. 2C). Phase plane plots of [Ca²⁺]_mito versus[Ca²⁺]_out illustrates the effect of P_i to lower matrix Ca²⁺ to a similar level, despite markedly different loading histories (Fig. 2D, upper panel). In parallel, P_i transport permitted full ΔΨ_m repolarization after various Ca²⁺ loads were accommodated (Fig. 2D, lower panel).

FIGURE 2.

P_i uptake is required to clamp [Ca²⁺]_mito to a setpoint. A, Ca²⁺ additions to P_i-depleted mitochondria in the range of 20–40 μm (at 150 s) were followed by 0.1 mm P_i (at 300 s) to initiate Ca²⁺ uptake and buffering. B, increases in [Ca²⁺]_mito exceeding 30 μm in the absence of P_i were quickly clamped back to 30 μm upon P_i addition, regardless of total mitochondrial Ca²⁺ load. C, ΔΨ_m changes during the protocol. D, phase-plane analysis of [Ca²⁺]_out versus [Ca²⁺]_mito (upper) and [Ca²⁺]_out versus ΔΨ_m (lower) before and after P_i addition.

Mersalyl, a thiol reagent that inhibits the mitochondrial phosphate carrier (22, 23), was used to test whether phosphate regulation of mitochondrial Ca²⁺ uptake required transport through PiC (Fig. 3). Mersalyl (10 μm; blue line) partially inhibited Ca²⁺ uptake in the absence of P_i, possibly due to a partial depolarization of ΔΨ_m, yet [Ca²⁺]_mito still increased to the same plateau, but at a slightly slower rate. With PiC blocked, when 1 mm phosphate was added, depletion of cuvette [Ca²⁺] was not accelerated, and [Ca²⁺]_mito buffering did not occur, confirming that PiC-mediated P_i entry into the matrix was required for both processes.

FIGURE 3.

PiC inhibition blocks acceleration of Ca²⁺ uptake and matrix Ca²⁺ buffering. 10 μm (blue), but not 1 μm (red), mersalyl inhibited the P_i effect on mitochondrial Ca²⁺ uptake (A) and buffering (B). C, ΔΨ_m was partially depolarized in the presence of 10 μm mersalyl, which may account for the somewhat slower initial P_i-independent Ca²⁺ uptake kinetics.

As we previously reported (8), high capacity (mode 2) Ca²⁺ uptake is blocked by a low concentration (50 nm) of Ru360 (Fig. 4A), leaving only mode 1 uptake, detected by the rise in matrix Ca²⁺ (Fig. 4B). Inhibiting PiC with mersalyl (10 μm) had little effect on the rate of mode 1 Ca²⁺ uptake, suggesting that mode 1 uptake is P_i-independent; however, mersalyl still inhibited the effect of P_i on matrix Ca²⁺ buffering and ΔΨ_m recovery (Fig. 4C).

FIGURE 4.

Effects of P_i and mersalyl on mitochondrial Ca²⁺ transport in the presence of Ru360. A, with bulk Ca²⁺ uptake (mode 2) inhibited with Ru360 (50 nm), remaining [Ca²⁺]_mito increase (mode 1) was not inhibited by mersalyl, although the P_i effect on buffering was still blocked (B). C, ΔΨ_m responses.

Lehninger (11) previously reported that respiration-dependent Ca²⁺ uptake by mitochondria depends on anion transport across the inner mitochondrial membrane, supported only by permeant anions that were also capable of transporting a proton. Acetate anion is transported as the weak acid by passive diffusion with the concomitant release of a proton in the matrix. Acetate (1 mm) was capable of increasing the Ca²⁺ uptake rate in P_i-depleted mitochondria (Fig. 5, blue traces), albeit at a slower rate than a similar concentration of P_i, most likely due to the slow rate of acetate diffusion across the inner membrane. Increasing the acetate concentration to 30 mm accelerated the initial Ca²⁺ uptake rate to a level close to that of P_i (Fig. 5, red traces), with much smaller effects on matrix Ca²⁺ buffering. The acetate effect on Ca²⁺ uptake was not sustained as well as that of P_i, and was associated with slow depolarization following the initial repolarization phase. The acetate effect was not inhibited by mersalyl treatment (data not shown). Vanadate, a P_i analog, accelerated Ca²⁺ uptake in a manner similar to phosphate, and also strongly supported matrix Ca²⁺ buffering to decrease [Ca²⁺]_mito (Fig. 5, green traces). Notably, mitochondrial Ca²⁺ uptake in the presence of vanadate was not accompanied by restoration of ΔΨ_m, probably due to inhibition of OxPhos by vanadate (24). This finding argues against the idea that the anion-dependent acceleration of Ca²⁺ uptake is secondary to an increase in electrochemical driving as ΔpH is converted to ΔΨ_m (20).

FIGURE 5.

Anions acetate, vanadate, and phosphate show different effects on mitochondrial Ca²⁺ dynamics. Phosphate (0.5 mm Pi, black), acetate (1, blue or 30 mm Acet., red), or vanadate (1 mm V_i, green) were added at 240 s to phosphate-depleted mitochondria after a 35 μm Ca²⁺ addition. [Ca²⁺]_out (A), [Ca²⁺]_mito (B), ΔΨ_m (C), Ca²⁺ bound-to-free ratio (D).

To explore this in more detail, we measured mitochondrial matrix pH ratiometrically (SNARF-1) using the same protocol of Ca²⁺ addition followed by anion addition in P_i-depleted mitochondria (Fig. 6). Prior to Ca²⁺ addition, baseline matrix pH was ∼7.75 with extramitochondrial pH buffered at 7.2. Ca²⁺ addition in the absence of P_i resulted in matrix acidification to ∼7.6. Therefore, ΔpH decreased, rather than increased, during P_i-independent Ca²⁺ uptake, which could have been due to displacement of protons from matrix Ca²⁺ binding sites or the coupling of Ca²⁺ entry to Na⁺/Ca²⁺ and Na⁺/H⁺ exchange activity. The addition of P_i, V_i, or acetate led to a further decrease in pH to 7.35 during the phase of ΔΨ_m recovery (Fig. 6). The lack of increase in ΔpH in P_i-depleted mitochondria during Ca²⁺ uptake, and the similar effects of the 3 anions on matrix pH despite differences in the extent of ΔΨ_m recovery (Fig. 5C), provides further evidence against interconversion of ΔpH to ΔΨ_m as the primary explanation for the acceleration of MCU rate. It should also be noted that there was still ample membrane potential to drive MCU current both in the absence and presence of P_i (−170 mV versus −190 mV). Hence, anion entry, not matrix pH change, was the primary facilitator of Ca²⁺ uptake.

FIGURE 6.

Mitochondrial pH measurement in P_i-depleted mitochondria. Mitochondrial pH was measured using the ratiometric pH indicator carboxy-SNARF-1. Mitochondrial pH decreased to a similar extent after the Ca²⁺ addition and after anion addition regardless which anion was used. 40 μm Ca²⁺ was first added, followed by 1 mm P_i (or 1 mm V_i or 30 mm acetate), as indicated.

Nucleotides represent another major physiological anion, hence we investigated their possible effects on mitochondrial Ca²⁺ uptake. MgATP (1 mm) accelerated mitochondrial Ca²⁺ uptake in the absence of P_i when added after (Fig. 7A) or prior to (Fig. 7B) the Ca²⁺ addition. [Ca²⁺]_mito level was also decreased, indicating that MgATP could also buffer matrix Ca²⁺. Additional experiments were carried out to further investigate the mechanism of the MgATP effect on P_i-independent mitochondrial Ca²⁺ uptake. The addition of MgCl₂ (1 mm) alone showed only a partial inhibitory effect on mitochondrial Ca²⁺ uptake, either in the absence or presence of P_i, and no effect on matrix buffering (Fig. 7B). Treating the mitochondria with oligomycin (Fig. 8A) did not prevent the MgATP effect, ruling out the possibility that ATP hydrolysis by the F_OF₁ ATPase could be providing matrix P_i. The ATP effect was also not impacted by blocking the adenine nucleotide translocase with bongkrekic acid (Fig. 8B) or carboxyatractyloside (data not shown), and ADP was incapable of accelerating the Ca²⁺ uptake rate (Fig. 8B). PiC inhibition also had no effect on the MgATP facilitation of Ca²⁺ uptake (Fig. 8C).

FIGURE 7.

MgATP accelerates mitochondrial Ca²⁺ uptake in P_i-depleted mitochondria. A, acute addition of either 0.5 mm P_i (black), 1 mm MgATP(red), or 1 mm Li⁺-AMP-PNP/MgCl₂ (blue), added after Ca²⁺, accelerates mitochondrial Ca²⁺ uptake. Ca²⁺ (35 μm) was added at 100 s to initiate P_i-independent Ca²⁺ uptake (i); 1 mm P_i was added at 270 s to initiate the P_i-dependent phase (ii). B, Ca²⁺ uptake rate after pre-incubation with MgATP, AMP-PNP or MgCl₂, added 120 s before Ca²⁺. Notably, P_i (1 μm) did not accelerate Ca²⁺ uptake when mitochondria were preincubated with MgATP or MgAMP-PNP. C, summary of the effects of MgATP, AMP-PNP or MgCl₂ on Ca²⁺ uptake. Data in C are presented as mean ± S.E. n = 10, 8, 4, and 4 for the 4 groups shown. *: p < 0.05. The kinetics of extra-mitochondrial Ca²⁺ ([Ca²⁺]_out) from panel B were fitted using exponential functions to obtain time constants during the P_i-independent and P_i-dependent Ca²⁺ uptake phases. MgATP and AMP-PNP accelerated P_i-independent Ca²⁺ uptake rates, while MgCl₂ slightly prolonged Ca²⁺ uptake only in the presence of P_i. For phase (i), P_i-independent uptake, a double exponential was employed: [Ca²⁺]_out(t) = A_fast[exp(-t/τ_fast] + A_slow[exp(-t/τ_slow)]. A_fast and A_slow are the maximum amplitude; τ_fast and τ_slow are the time constants of the fast and slow components of exponential decay, respectively. Only the slow decay time constants are plotted for different treatments. In phase (ii), the [Ca²⁺]_out was fitted by a single exponential function [Ca²⁺]_out(t) = A[exp(−t/τ] + C.

FIGURE 8.

ATP effects were not altered by inhibition of ANT or the mitochondrial ATP synthase. A, oligomycin (ATP synthase inhibitor) pretreatment did not alter MgATP facilitation of P_i-independent Ca²⁺ uptake. B, ADP did not accelerate P_i-independent Ca²⁺ uptake. C, bongkrekic acid (BKA; ANT inhibitor) did not alter the ATP response.

Finally, the non-hydrolyzable ATP analog, AMP-PNP, was shown to have a similar effect on mitochondrial Ca²⁺ dynamics as MgATP (Fig. 7), i.e. increased Ca²⁺ uptake rate and decreased [Ca²⁺]_mito. These data indicate that ATP facilitates mitochondrial Ca²⁺ uptake through a novel direct effect of the molecule itself, presumably requiring nucleoside triphosphate uptake by a non-canonical transporter, since matrix Ca²⁺ buffering was also increased by ATP.

Discussion

The present study quantitatively investigated the fundamental role of P_i transport on mitochondrial Ca²⁺ uptake and buffering by monitoring extra- and intra-mitochondrial Ca²⁺ and ΔΨ_m. The major findings were that P_i directly facilitates mitochondrial Ca²⁺ uptake in a concentration-dependent manner while simultaneously buffering [Ca²⁺]_mito to a fixed setpoint. Using P_i-depleted mitochondria, a novel regulatory mechanism involving MgATP was also revealed, whereby it increased mitochondrial Ca²⁺ uptake. The MgATP effect did not require ATP hydrolysis. The results demonstrate that there is a strict dependence of maximal MCU, as well as [Ca²⁺]_mito, on anion flux rate, highlighting the regulatory complexity of cardiac mitochondrial Ca²⁺ dynamics.

The ability of mitochondria to accumulate massive amount of Ca²⁺ has been known for more than fifty years (3, 14, 25). While previous work established that P_i increases mitochondrial Ca²⁺ loading (11, 26) and participates in mitochondrial Ca²⁺ buffering to clamp matrix Ca²⁺ at a limiting setpoint (21, 27), the dual role of P_i has not been studied quantitatively in energized cardiac mitochondria while measuring changes in total Ca²⁺ uptake, matrix Ca²⁺, and ΔΨ_m in parallel. The protocol used here uniquely permitted the study of both the P_i-dependent and P_i-independent components of mitochondrial Ca²⁺uptake and defined the efficacy of the anion for activation of uptake versus buffering. P_i and V_i were approximately equally effective in supporting both functions, while acetate facilitated uptake but had little effect on matrix buffering.

ATP has been shown to participate in regulating mitochondrial Ca²⁺ accumulation. It was first reported by Vasington and Murphy (3), and DeLuca and Engstrom (25) that ATP was necessary for the uptake of Ca²⁺. In the former study, P_i or ATP enhanced, and respiratory inhibitors or uncouplers inhibited, “active” Ca²⁺ uptake, while in the latter study, it was concluded that neither P_i nor coupled respiration was required for Ca²⁺ uptake/binding (25). However, these studies did not take into account possible effects of Mg²⁺, ATP, P_i, or even substrate availability on the electrochemical driving force for Ca²⁺ uptake (indeed, this work took place prior to wide acceptance of the chemiosmotic hypothesis). Later, Carafoli and Lehninger showed that ATP (or ADP) is taken up along with Ca²⁺ and P_i to form hydroxyapatite-like precipitates in the mitochondrial matrix. Without the uptake of ATP (or ADP), no granules were formed (5). Weinbach and Von Brandt then found that the granules contained adenine nucleotides (28). The possible co-precipitation of Ca-P_i and nucleotide was, therefore, a potential explanation of why ATP accelerated mitochondrial Ca²⁺ uptake. Our findings show that, independent of P_i, ATP, but not ADP, can regulate mitochondrial Ca²⁺ uptake and buffering, and that hydrolysis is not required. The transporter mediating this effect, and whether ATP buffers by forming a high affinity complex with Ca²⁺, remains to be determined.

Recent studies have confirmed that MCU is a selective Ca²⁺ channel (29) comprised of a defined pore subunit (MCUa) (30, 31). In most studies, it is assumed that maximal MCU rates are a function of the pore's properties, regulated by accessory proteins in the Ca²⁺ uptake complex (e.g. MCUb (32), MICU1 and MICU2 (33–36), MCUR (37), EMRE (38)). However, our findings in intact mitochondria demonstrate that the maximal Ca²⁺ uptake rate is limited by the anion transport rate. This effect is masked under physiological conditions because P_i is usually at saturating concentrations with respect to effect on Ca²⁺ uptake (EC₅₀ of P_i = 87.5 μm). Interestingly, the [Ca²⁺]_mito level was buffered and reached a steady state near 20–30 μm when the phosphate concentration was > 0.01 mm (Fig. 1 B) or when [Ca²⁺]_out level was > 2 μm (Fig. 2). This calcium-phosphate buffering effect allows estimation of the buffering threshold; however, even though the extramitochondrial phosphate concentration required for calcium buffering was ∼0.01 mm, [P_i]_mito may not necessarily be equal to [P_i]_out. In this regard, employing phosphorous NMR spectroscopy, Garlick et al. (39, 40) reported that mitochondrial P_i concentration was 2.5× the cytoplasmic level in intact hearts (∼6 mm). In contrast, in solution, the calcium-phosphate complex only forms when calcium is in the millimolar range (Ksp = 3 × 10⁻³⁰ for Ca₃(PO₄)_2,) (20), a Ca²⁺ concentration at least an order of magnitude higher than we measured either outside or inside the mitochondria.

The method described here also provides an indirect way of measuring phosphate influx since the mitochondrial Ca²⁺ uptake rate is proportional to the phosphate concentration and sensitive to the phosphate carrier inhibitor. By assuming 1:1 or 1:1.5 P_i:Ca transport stoichiometry (27), we could approximate the phosphate flux to be around 0.5–1 nmol/mg/sec at 1 mm [P_i], and this rate is close to the range of the reported phosphate transport rate in the rat liver mitochondria at 2 mm, which was determined by the inhibitor stop method with ³²P at 0 °C (41). In keeping with the overall hypothesis that anion flux limits Ca²⁺ uptake, this maximum P_i transport rate corresponded to the maximum MCU rates previously reported for guinea pig cardiac mitochondria (21).

Phosphate is known to play a role in regulating mitochondrial Ca²⁺ and function. It has previously been reported that P_i is required for bulk mitochondrial Ca²⁺ uptake (42). Possible mechanisms of phosphate facilitation of Ca²⁺ uptake could be explained by (i) precipitation of mitochondrial Ca²⁺ by forming a Ca-P_i complex to maintain the electrochemical driving force for Ca²⁺, (ii) the requirement for co-transport of an anion via PiC with Ca²⁺ to maintain charge balance (supported by the acetate result), or (iii) direct interaction with Ca²⁺ transporters. In the current study, we confirmed that phosphate participates in clamping mitochondrial [Ca²⁺]_mito level to maintain a Ca²⁺ gradient for Ca²⁺ uptake. Dissipation of the proton gradient by phosphate influx was unlikely to be the main reason for enhancing Ca²⁺ uptake, since V_i could equally increase uptake without restoring ΔΨ_m (Fig. 5C).

Mitochondrial matrix or extramitochondrial P_i, by themselves, were probably not direct allosteric regulators of the mitochondrial calcium uniporter because concomitant P_i influx was absolutely required to maintain fast Ca²⁺ uptake. The PiC has also been implicated in the regulation of mPTP opening to modulate cell death (13, 43); however, whether the mechanism is through a direct PiC-mPTP interaction or merely the result of a loss of P_i influx to regulate matrix Ca²⁺, or the P_i-Ca²⁺ complex, is unclear. To verify the importance of P_i in regulating mitochondrial Ca²⁺ dynamics, further testing on PiC knockdown cells or animals (13, 16) would be helpful to understand the molecular basis of the P_i-dependent process described herein.

Nucleotide effects on mitochondrial Ca²⁺ uptake have been reported previously (44); however, ATP or AMP-PNP were shown to inhibit the uniporter from a site on the outside of the inner membrane. Thus, the effect of ATP to increase P_i-independent Ca²⁺ uptake shown here has not been observed before. The nature of the ATP transporter involved remains to be determined, as we ruled out adenine nucleotide translocase or the ATP synthase as contributors to the mechanism. An ATP-Mg/P_i carrier (SLC25A23) has also been described in cardiac mitochondria, representing another possible anion influx pathway. Knockdown of this exchanger was recently shown to impact mitochondrial Ca²⁺ dynamics (16), but it is unclear how this ATP/P_i antiporter would be able to function in the P_i-depleted mitochondria used in the present study.

In conclusion, these findings support a major role for anion influx in determining the maximal Ca²⁺ influx rate and matrix Ca²⁺ concentration in cardiac mitochondria. P_i uptake is required for maximal enhancement of MCU rate, as well as maximal buffering capacity. ATP is not strictly required for Ca²⁺ uptake, but facilitates P_i-independent uptake (ADP does not substitute). The Ca-P_i buffer system clamps the free Ca²⁺ at an upper limit, allowing massive accumulation of Ca²⁺ without increasing the efflux rate or diminishing the electrochemical driving force for entry. The mitochondria are therefore able to regulate metabolism via Ca²⁺ sensing and also can act as a cellular Ca²⁺ sink when there is a chronic, potentially pathological, increase in extramitochondrial Ca²⁺. Notably, we show that the first role is modulated by the P_i effect on Ca²⁺ uptake, effectively incorporating P_i, which also rises with energy demand, into the sensing process. Considering that P_i, ATP, and Ca²⁺ dynamics are altered under metabolic stress, it will be important to define how the two effects of P_i contribute to the pathophysiology of cardiac disease.