The Ψ_m depolarization that accompanies mitochondrial Ca²⁺ uptake is greater in mutant SOD1 than in wild-type mouse motor terminals

Abstract

The electrical gradient across the mitochondrial inner membrane (Ψ_m) is established by electron transport chain (ETC) activity and permits mitochondrial Ca²⁺ sequestration. Using rhodamine-123, we determined how repetitive nerve stimulation (100 Hz) affects Ψ_m in motor terminals innervating mouse levator auris muscles. Stimulation-induced Ψ_m depolarizations in wild-type (WT) terminals were small (<5 mV at 30 °C) and reversible. These depolarizations depended on Ca²⁺ influx into motor terminals, as they were inhibited when P/Q-type Ca²⁺ channels were blocked with ω-agatoxin. Stimulation-induced Ψ_m depolarization and elevation of cytosolic [Ca²⁺] both increased when complex I of the ETC was partially inhibited by low concentrations of rotenone (25–50 nmol/l). This finding is consistent with the hypothesis that acceleration of ETC proton extrusion normally limits the magnitude of Ψ_m depolarization during mitochondrial Ca²⁺ uptake, thereby permitting continued Ca²⁺ uptake. Compared with WT, stimulation-induced increases in rhodamine-123 fluorescence were ≈5 times larger in motor terminals from presymptomatic mice expressing mutations of human superoxide dismutase I (SOD1) that cause familial amyotrophic lateral sclerosis (SOD1-G85R, which lacks dismutase activity; SOD1-G93A, which retains dismutase activity). Ψ_m depolarizations were not significantly altered by expression of WT human SOD1 or knockout of SOD1 or by inhibiting opening of the mitochondrial permeability transition pore with cyclosporin A. We suggest that an early functional consequence of the association of SOD1-G85R or SOD1-G93A with motoneuronal mitochondria is reduced capacity of the ETC to limit Ca²⁺-induced Ψ_m depolarization, and that this impairment contributes to disease progression in mutant SOD1 motor terminals.

Mitochondria sequester significant amounts of stimulation-induced Ca²⁺ loads in many cell types (1–9). This mitochondrial Ca²⁺ uptake occurs via a Ca²⁺ uniporter/channel (10) down a potential gradient (Ψ_m, 150–200 mV, matrix negative) established by ETC activity (11, 12). Entry of Ca²⁺ into mitochondria depolarizes Ψ_m, which would be expected to reduce the gradient driving Ca²⁺ uptake. However, in motor nerve terminals, mitochondrial Ca²⁺ uptake continues throughout prolonged trains of action potentials (13). One possible explanation for this apparent paradox is that the Ψ_m depolarization produced by Ca²⁺ entry reduces the gradient against which ETC complexes I, III and IV extrude protons, thus accelerating ETC proton extrusion (14). This acceleration would limit the net Ψ_m depolarization, thereby allowing mitochondria to continue taking up Ca²⁺ even during prolonged stimulation (15).

We tested this hypothesis in mouse motor terminals, and found that the Ψ_m depolarizations produced by repetitive stimulation at 50–100 Hz were Ca²⁺ dependent and reversible, and were small (or undetectable) at near-physiological temperatures (30 °C). Partially inhibiting ETC activity with low concentrations of rotenone or low temperature increased the amplitude of these depolarizations.

We also tested motor terminals in transgenic mice expressing mutant versions of human SOD1 (G85R, G93A) that produce familial amyotrophic lateral sclerosis (fALS) (16–18). In these and other transgenic mice expressing fALS-inducing SOD1 mutations, motor terminals begin to degenerate well before motor neuron somata in the spinal cord begin to die, and some of the earliest reported morphological changes occur in motor terminal mitochondria (19–24). Misfolded fALS-linked mutant SOD1s associate with mitochondria (cytoplasmic face, intermembrane space, inner membrane, matrix) (25–34), and impaired ETC activity has been reported in the spinal cord of mice expressing SOD1-G93A (25, 30, 35–37). Damiano et al. (38) demonstrated a reduction in Ca²⁺ loading capacity in spinal cord mitochondria from both SOD1-G93A and SOD1-G85R mice. We hypothesized that this reduced ability to sequester Ca²⁺ loads might be caused by reduced ability to accelerate mitochondrial proton extrusion in response to the Ψ_m depolarization produced by Ca²⁺ entry. Consistent with this hypothesis, we demonstrate that at 30 °C stimulation-induced Ψ_m depolarizations are greater in SOD1-G85R and SOD1-G93A mice than in WT mice, and that this increase does not depend on dismutase activity. This is the first demonstration in presymptomatic fALS mice of an altered response to Ca²⁺ loads in mitochondria located exclusively within motor neurons (as distinct from surrounding glial cells).

Results

During repetitive stimulation at near-physiological temperatures, Ψ_m depolarization is small.

Fig. 1 shows representative changes in cytosolic [Ca²⁺], mitochondrial [Ca²⁺], and Ψ_m recorded in WT motor nerve terminals stimulated with three trains of action potentials (100 Hz, 5 seconds). The topmost trace shows the elevation of cytosolic [Ca²⁺], measured as the normalized increase in fluorescence (F/F_rest) of Oregon Green 488 BAPTA-5N (OG-5N) that had been ionophoretically injected into the axoplasm. After an initial rapid increase at the onset of stimulation, the rate of rise slows, due mainly to Ca²⁺ sequestration by mitochondria (5, 9). When stimulation stops, cytosolic [Ca²⁺] shows a rapid initial decrease. Measurements of cytosolic [Ca²⁺] made with a higher affinity Ca²⁺ indicator than that used here reveal an additional slowly-decaying phase caused in part by Ca²⁺ extrusion from mitochondria via a Na⁺/Ca²⁺ exchanger (39).

Fig. 1.

Stimulation at 100 Hz increases cytosolic and mitochondrial [Ca²⁺] and depolarizes Ψ_m in WT mouse motor terminals at 30 °C. (A) Upper trace: Elevation of cytosolic [Ca²⁺] in response to three stimulus trains at 100 Hz, monitored as normalized increases (F/F_rest) in the fluorescence of intra-axonally injected OG-5N (vertical lines indicate duration of stimulation). The average increase in cytosolic [Ca²⁺] is 1.0 μmol/l above an assumed resting value of 0.1 μmol/l (64). Second trace: Elevations of mitochondrial matrix [Ca²⁺], monitored as increases in the fluorescence of mitochondrially-loaded X-rhod-1 (mean of five traces). The average increase in mitochondrial [Ca²⁺] is ≈1–2 μmol/l above an assumed resting value of 0.05–0.1 μmol/l (13, 40, 41). Third trace: Depolarization of Ψ_m, monitored as increases in the fluorescence of Rh-123. The line is a smoothed moving bin average of three neighboring points. Lower trace shows (using a different vertical scale) these same data, along with the much larger Ψ_m depolarization produced in this terminal by the proton carrier CCCP (2 μmol/l). Cytosolic [Ca²⁺], mitochondrial [Ca²⁺], and Ψ_m were recorded from different preparations, aged 64–88 days.

The second trace in Fig. 1 shows the stimulation-induced increase in mitochondrial [Ca²⁺], measured as the fluorescence of mitochondrially-loaded X-rhod-1. Mitochondrial [Ca²⁺] exhibits a slower rate of rise and a slower decay than cytosolic [Ca²⁺] (5, 39). Mitochondrial [Ca²⁺] responses evoked by the second and third stimulus trains begin during the decaying phase of the response to previous trains, but the peak fluorescence evoked by subsequent trains does not exceed that evoked by the first train. This “cap” on the maximal increase in mitochondrial [Ca²⁺] is not due to saturation of the indicator dye (40), and is unlikely to be caused by decreased mitochondrial Ca²⁺ entry during subsequent trains, because both the sustained elevation of cytosolic [Ca²⁺] and recordings of transmitter release indicate that action potentials continue to admit Ca²⁺ into the cytosol throughout the train (9, 39), and the increase in cytosolic [Ca²⁺] produced by each train is similar. Also, if mitochondrial Ca²⁺ sequestration is blocked, cytosolic [Ca²⁺] rises progressively during stimulation (5, 9). Rather, the maximal increase in matrix [Ca²⁺] is limited by formation of an insoluble complex containing Ca and phosphate that reversibly sequesters matrix Ca²⁺ (41). The maximal increase in mitochondrial [Ca²⁺] is only ≈1–2 μmol/l (13, 40, 41).

The third trace in Fig. 1 shows stimulation-induced depolarizations of Ψ_m, recorded as increases in rhodamine 123 (Rh-123) fluorescence. In this motor terminal, each stimulus train reversibly increased fluorescence by ≈2%. The bottom trace shows these same data on a different vertical scale, along with the much larger increase in Rh-123 fluorescence (≈140%) measured during the near-complete Ψ_m depolarization produced by the proton carrier carbonyl cyanide m-chloro phenyl hydrazone (CCCP, 2 μmol/l). At 28–30 °C the mean increase in Rh-123 fluorescence after 500 stimuli at 100 Hz was only 0.92% ± 0.13% (SEM, range 0–4.5%, n = 47 terminals). These measurements are described further in SI Text Item #1 and Fig. S1), including analysis suggesting that this fluorescence increase corresponds to a Ψ_m depolarization of 3–5 mV.

Stimulation-induced Ψ_m depolarizations are Ca²⁺ dependent.

To study mechanisms underlying stimulation-induced Ψ_m depolarizations, we needed to increase the magnitude of the recorded Rh-123 signal. Fig. 2A demonstrates that signal magnitude increased when the temperature was reduced, or when action potential duration (and thus Ca²⁺ entry) was prolonged using 3,4-diaminopyridine (3,4-DAP, 20 μmol/l), which blocks certain depolarization-activated K⁺ channels in motor axons and terminals (42–44).

Fig. 2.

The stimulation-induced Ψ_m depolarization increases with cooling or addition of 3,4-diaminopyridine (3,4-DAP) and requires Ca²⁺ entry through plasma membrane Ca²⁺ channels. (A) The Ψ_m depolarization produced by 100 Hz stimulation at 30 °C (left) was increased by cooling to 18 °C (upper right) or by prolonging the action potential with 20 μmol/l 3,4-DAP (lower right). (B) Ψ_m depolarizations (the magnitudes of which were enhanced by both cooling to 20 °C and 3,4-DAP, open circles) were reduced by omitting Ca²⁺ from the bath (filled circles, left) or (in a different terminal) by adding 0.6 μmol/l ω-agatoxin-TK (filled circles, right). The effects of low bath [Ca²⁺] were readily reversible; reversal of agatoxin effects was slow and incomplete. Each record in (A) and (B) is the mean of two to nine traces. The effects of cooling, 3,4-DAP, and removal of bath Ca²⁺ were observed in 10, 5, and 4 additional terminals, respectively. Exposures to 3,4-DAP and agatoxin were 7–62 minutes and 60 minutes, respectively.

Fig. 2B shows that stimulation-induced Ψ_m depolarizations were inhibited both by replacing bath Ca²⁺ with Mg²⁺ and by ω-agatoxin TK (0.6 μmol/l), which blocks the P/Q-type (Cav2.1) Ca²⁺ channels that mediate most Ca²⁺ entry into motor terminals in mice (45, 46). Thus the Na⁺ influx associated with axonal action potential propagation, which continues in low [Ca²⁺] and ω-agatoxin, is not by itself sufficient to produce stimulation-induced Ψ_m depolarizations. Rather, these depolarizations require Ca²⁺ influx into motor terminals.

The elevation of cytosolic [Ca²⁺] increases with increasing frequencies of stimulation (5) (Fig. S2A). In isolated mitochondria, Ca²⁺ influx through the uniporter/channel exhibits a greater-than-linear dependence on external [Ca²⁺] (11). Thus one would predict faster Ψ_m depolarization with higher stimulation frequencies. As predicted, Ψ_m depolarized more rapidly during 100 Hz than during 25 Hz stimulation (Fig. S2B).

Partial inhibition of ETC complex I increases the stimulation-induced Ψ_m depolarization.

Results in Fig. 2 link stimulation-induced Ψ_m depolarizations to Ca²⁺ entry into motor terminals but do not prove a linkage to Ca²⁺ influx into mitochondria. It is difficult to block mitochondrial Ca²⁺ influx selectively, because Ru360, which blocks the uniporter/channel (47), has limited permeability across plasma membranes and appears to reduce Ca²⁺ influx across motor terminal membranes (13) (Fig. S3). Fig. 3B and 3C link stimulation-induced Ψ_m depolarizations to mitochondrial function by demonstrating that partial inhibition of complex I of the ETC with low concentrations of rotenone (25–50 nmol/l) produced a 4-fold increase in the Rh-123 signal evoked by 100 Hz stimulation. This result is consistent with the prediction that reducing mitochondrial ability to accelerate ETC activity in response to stimulation-induced Ca²⁺ influx will increase the magnitude of the resulting Ψ_m depolarization. In cell lines, neurons, and isolated nerve terminals, these low rotenone concentrations have been reported to decrease complex I activity by 20–85% but to have little effect on resting Ψ_m (48–53). Likewise, we found that these low rotenone concentrations had no significant effect on resting Rh-123 fluorescence, suggesting little or no effect on resting Ψ_m (SI Text Item #2).

Fig. 3.

A low concentration of rotenone increases stimulation-induced elevations of cytosolic [Ca²⁺] and Ψ_m depolarizations. Cytosolic [Ca²⁺] (A) and Ψ_m depolarizations (B) produced by three trains at 100 Hz before and after addition of rotenone (50 nM). [Ca²⁺] and Ψ_m traces came from different terminals. (C) Paired data from eight terminals studied before and after rotenone exposure show that in mice expressing normal SOD1 (WT or human) rotenone increased the average change in Rh-123 fluorescence from 0.96% ± 0.22% (SEM) in control medium to 4.13% ± 1.18% in rotenone (* P < 0.05). Only measurements from the initial 100 Hz train were included in the averages. The duration of rotenone exposure was 17–30 minutes. Mice were 50–375 days old.

Greater Ψ_m depolarization during stimulation would be predicted to reduce the electrical gradient that permits Ca²⁺ influx into mitochondria and thus increase the elevation of cytosolic [Ca²⁺]. Fig. 3A demonstrates the predicted increase of cytosolic [Ca²⁺] in rotenone.

Stimulation-induced Ψ_m depolarizations are enhanced in mice expressing mutant human superoxide dismutase 1.

If motor terminal mitochondria in presymptomatic SOD1-G85R and SOD1-G93A mice have a reduced ability to accelerate ETC activity (see Introduction), then one would predict a greater stimulation-induced Ψ_m depolarization. Fig. 4A presents phase and resting Rh-123 fluorescence images of a terminal in a presymptomatic SOD1-G85R mouse. The difference image highlights those regions in which Rh-123 fluorescence increased during stimulation, showing signals specific to motor terminal mitochondria. Fig. 4B shows Rh-123 recordings from this SOD1-G85R mouse, as well as representative traces from presymptomatic SOD1-G93A and WT terminals. The bar graph in Fig. 4B shows that the average stimulation-induced increase in Rh-123 fluorescence in SOD1-G85R and SOD1-G93A motor terminals was ≈5-fold greater than that in WT terminals.

Fig. 4.

Stimulation-induced Ψ_m depolarizations are increased in presymptomatic SOD1-G85R and SOD1-G93A mice. (A) Phase (left) and Rh-123 fluorescence (middle) images show a resting motor terminal in a 121-day-old SOD1-G85R mouse. At right is a difference image of the same region, calculated by subtracting the resting fluorescence from the fluorescence during 100 Hz stimulation. (B) Representative stimulation-induced Ψ_m depolarizations evoked by repeated brief 100 Hz trains in a WT mouse (mean of two traces), a mouse lacking SOD1 (SOD1-KO), an “overexpressor” mouse with both normal mouse and normal human SOD1 (SOD1-OX), the SOD1-G85R terminal in (A), and an SOD1-G93A terminal. Bars shows mean peak amplitude (±SEM) for the first 100-Hz train for each of these groups. The Ψ_m depolarizations in SOD1-G85R (P < 0.01) and SOD1-G93A (P < 0.05) mice were significantly different from WT at 30 °C (analysis of variance followed by Dunnett's multiple comparison test). These differences were not significant at lower temperatures, where mitochondrial Ca²⁺ uptake may be reduced (64). Data were averaged from 47 terminals in 16 WT mice 50–145 days old; comparable values for SOD-KO mice were 19, 3, and 253–355; for SOD1-OX 5, 4, and 89–375; for SOD1-G85R mice 60, 6, and 121–161; for SOD1-G93A 13, 6, and 43–85. At these ages, all endplates were fully innervated. Additional details in SI Text, Item #3.

SOD1 helps to defend against oxidative damage by catalyzing the conversion of superoxide to hydrogen peroxide. SOD1-G93A retains SOD1 enzymatic activity, but SOD1-G85R lacks this activity (54), although heterodimers of G85R and wt SOD1 show activity (55). The fact that both types of mutant SOD1 terminals showed increased Ψ_m depolarizations suggests that this increase did not depend on the level of SOD1 activity. This suggestion was tested further by conducting similar experiments on mice lacking SOD1 activity altogether, and on mice with excess SOD1 activity due to expression of normal human SOD1. Fig. 4B shows that stimulation-induced Ψ_m depolarizations in these knockout and overexpressing mice were not significantly different from those recorded in WT terminals.

Another possible mechanism underlying the increased stimulation-induced Ψ_m depolarizations measured in mutant SOD1 mice is transient opening of the mitochondrial permeability transition pore (MPTP). However, Fig. 5 shows that cyclosporin A, which inhibits pore opening, did not reduce the magnitude of these depolarizations in SOD1-G85R mice.

Fig. 5.

Stimulation-induced Ψ_m depolarizations in SOD1-G85R mice are not reduced by cyclosporin A. (A) Representative traces show the Rh-123 fluorescence response evoked by a 100-Hz stimulus train before and 56 minutes after exposure to 5 μmol/l cyclosporin. (B) Graph plots peak amplitudes (normalized to resting fluorescence) before and during drug exposure for this terminal. Similar experiments on three additional SOD1-G85R mice (130–150 days) also showed no significant change in Ψ_m depolarizations in cyclosporin A (paired Wilcoxon signed rank test, P > 0.50). Similar experiments in additional mouse types (e.g., SOD1 knockout treated with 50 μmol/l 3,4-diaminopyridine) confirmed the lack of effect of 8–10 μmol/l cyclosporin A.

Discussion

Determinants of the stimulation-evoked Ψ_m depolarization in mouse motor terminals.

Data presented here demonstrate that the Ψ_m depolarization produced by repetitive stimulation depends on Ca²⁺ influx into motor terminals. In WT mice this Ψ_m depolarization is small, reversible, and repeatable, in contrast to the large Ψ_m depolarizations measured after applying large Ca²⁺ loads to isolated mitochondria (56). Some of the difference in the magnitude and reversibility of the Ca²⁺-induced Ψ_m depolarization may be due to the fact that the Ca²⁺ load was delivered physiologically via repetitive nerve stimulation to mitochondria in situ. Isolated mitochondria can take up larger amounts of Ca²⁺ with minimal Ψ_m depolarization when the Ca²⁺ load is presented in increments rather than as a bolus (41).

To determine whether the magnitude of the stimulation-induced Ψ_m depolarization is consistent with electrical models of mitochondrial function, we estimated the current entering mitochondria using the initial slope of the mitochondrial [Ca²⁺] response to stimulation (as in Fig. 1). The depolarization expected from this current was then estimated using an electrical model of the mitochondrial inner membrane consisting of the resistance of the ETC (in series with a battery representing the electron motive force) in parallel with the resistance of the mitochondrial membrane (SI Text Item #4). These calculations yielded a predicted Ψ_m depolarization of ≈2 mV, similar to the value estimated independently from the magnitude of the stimulation-induced change in Rh-123 fluorescence (SI Text Item #1). Ψ_m depolarizations of this magnitude are sufficient to accelerate ETC activity significantly (57).

According to this electrical model of the mitochondrial membrane, the magnitude of the stimulation-induced Ψ_m depolarization can be varied in at least two ways. The first is to alter the magnitude of the Ca²⁺ current that enters mitochondria: this is the likely explanation for the changes in the magnitude of the Ψ_m depolarization produced by reducing (with ω-agatoxin) or increasing (with 3,4-DAP) the magnitude of the stimulation-induced Ca²⁺ influx into the motor terminal. The second way is to change ETC activity; this is the likely mechanism underlying the increase in the stimulation-induced Ψ_m depolarization produced by partial blockade of complex I with rotenone. Partial ETC inhibition produces a greater Ψ_m depolarization in stimulated than in resting terminals (15). Synergistic damaging effects of partial complex I inhibition and Ca²⁺ challenges have also been documented in other preparations (58, 59).

The larger stimulation-induced Ψ_m depolarizations in SOD1-G85R and SOD1-G93A motor terminals likely reflect reduced ability to accelerate ETC activity after Ca²⁺ entry.

The Introduction section cited morphological evidence for early mitochondrial damage in motor terminals and motor neurons of mutant SOD1 models of fALS, as well as functional evidence for reduced activity of certain ETC complexes and reduced capacity for Ca²⁺ uptake in spinal cord mitochondria. Work presented here demonstrates that rotenone (low concentration) mimics the effect of fALS mutations on stimulation-induced Ψ_m depolarizations, and that these depolarizations are not reduced by cyclosporin A. Given these past and present findings, a logical explanation for the larger Ψ_m depolarizations recorded in motor terminals of SOD1-G85R and SOD1-G93A mice is that their mitochondria have a defect in the ETC (or its regulation) that reduces their ability to accelerate ETC activity in response to the depolarization produced by Ca²⁺ entry. Synaptic mitochondria are more sensitive than somatic mitochondria to both complex I inhibition and Ca²⁺ overload (60, 61). This might explain the finding that, in fALS mice, motor terminal damage precedes damage to the motor neuron cell body.

Deficits in mitochondrial Ca²⁺ handling would be expected to be most evident in motor terminals innervating fast muscles, whose tendency to discharge in high frequency bursts (62) would expose them to larger Ca²⁺ loads and Ψ_m depolarizations than motor terminals innervating slow muscle. Consistent with this idea, motor terminals innervating fast limb muscles are the earliest to degenerate in fALS mice (20, 24, 63).

In summary, mitochondria have multiple features that permit them to sequester the Ca²⁺ loads introduced by repetitive stimulation, including powerful Ca²⁺ buffering within the matrix, and acceleration of ETC activity in response to the depolarization produced by Ca²⁺ entry, hence preserving the electrical gradient that favors Ca²⁺ entry. Evidence presented here suggests that the Ca²⁺-dependent Ψ_m depolarization produced by repetitive stimulation is <5 mV in WT motor terminals, but increases in mitochondria of presymptomatic SOD1-G85R and SOD1-G93A terminals. These Ψ_m depolarizations would be expected to increase in terminals of older mutant SOD1 mice as ETC function deteriorates, impairing both mitochondrial Ca²⁺ sequestration and ATP synthesis.

Materials and Methods

Experiments used neuromuscular junctions from the levator auris longus muscle of WT, hSOD1-G85R, and hSOD1-G93A mice, as well as mice that express normal human SOD1 and mice lacking SOD1. Most experiments were conducted at 30 °C, the warmest temperature at which the dissected preparation could remain functional for several hours. Action potentials were evoked in the motor nerve by applying suprathreshold 0.3 ms depolarizing pulses (500–2000 at 50–100 Hz) via a suction electrode. Muscle contractions were blocked with d-tubocurarine (7–15 μmol/l).

Stimulation-induced changes in cytosolic [Ca²⁺] were monitored using OG-5N (K_d 60 μmol/l) injected ionophoretically into an internodal axon (64). Stimulation-induced changes in mitochondrial [Ca²⁺] were measured using the indicator X-rhod-1, loaded as described (39). Changes in Ψ_m were measured using Rh-123, as described (15). The preparation was imaged using a spinning disk confocal microscope system designed to use low excitation light intensities. Spatially averaged changes in fluorescence were plotted as F/F_rest, where F_rest is the average net, background-subtracted fluorescence calculated from 20–30 images obtained before stimulation began. SI Text Item #5 contains additional methodological details.