The Journal of Neurophysiology Vol. 80 No. 1 July 1998, pp. 186-195
Copyright ©1998 by the American Physiological Society

Effects of Mitochondrion on Calcium Transients at Intact Presynaptic Terminals Depend on Frequency of Nerve Firing

Yan-Yi Peng

Department of Pharmacological and Physiological Sciences, Committees on Neurobiology and Cell Physiology, University of Chicago, Chicago, Illinois 60637

	ABSTRACT

Abstract Introduction Methods Results Discussion References

Peng, Yan-Yi. Effects of mitochondrion on calcium transients at intact presynaptic terminals depend on frequency of nerve firing. J Neurophysiol. 80: 186-195, 1998. The rate and the total amount of Ca²⁺ elevation in the presynaptic terminals of bullfrog sympathetic ganglia depend on the firing frequency of the terminals. Carbonyl cyanide m-chlorophenylhydrazone (CCCP), a mitochondrial uncoupler, was used for testing whether mitochondrial Ca²⁺ uptake is one of the mechanisms that underlie this frequency dependence. Fura-2 fluorimetry was used for measurement of intraterminal Ca²⁺. When stimulations of different durations (30 and 1.5 s) and frequencies (4 and 20 Hz) evoked Ca²⁺ transients with similar peak amplitudes (264 ± 22 nM vs. 251 ± 18 nM, means ± SE), CCCP augmented the responses to the 4-Hz stimulation 8.9 times more strongly than it did the responses to the 20-Hz stimulation (249.7 ± 81.5% vs. 25.3 ± 10.2%). When stimulations delivered at the two frequencies had the same durations (1.5, 3, 6, 10, 20, and 30 s), CCCP enlarged the responses to the 4-Hz stimulations up to 4.2 times more than it did the responses to the 20-Hz stimulations. When the same number of stimuli (120) was delivered at the two frequencies, the effects of CCCP on the responses evoked by the 4-Hz train were again 6.8 times stronger than its effects on the responses to the 20-Hz stimulation. Therefore neither the peak amplitudes of the responses nor the durations of the stimulations dictated the extent to which the mitochondria modulated the peak [Ca²⁺]_i. Instead, the extent of the modulation was governed by the frequency of stimulation. Specifically, the less frequent the Ca²⁺ influx, the stronger the mitochondrial modulation_. Also, during nerve firing Ca²⁺ release from the ryanodine-sensitive store had a higher potential to influence the [Ca²⁺]_i transients than did Ca²⁺ removal by the mitochondria for the first 6 s of the responses. On cessation of stimulation, CCCP reduced the initial rapid rate of Ca²⁺ decay. Thus uptake by the mitochondria was an important mechanism for Ca²⁺ removal after repetitive firing at the presynaptic terminals.

	INTRODUCTION

Abstract Introduction Methods Results Discussion References

Effects of uncoupling of the mitochondria on intracellular Ca²⁺ transients have been studied in endocrine and neuronal cells and in synaptosomes. The Ca²⁺ transients were evoked by voltage steps in the somata of adrenal chromaffin cells (Herrington et al. 1996; Park et al. 1996) and in synaptosomes (Stuenkel 1994), by application of a high-potassium bathing solution or glutamate, and by repetitive action potentials evoked by electrical field stimulation in the somata of cultured neuronal cells (Bleakman et al. 1993; Thayer and Miller 1990; Werth and Thayer 1994; White and Reynolds 1995). The frequency of action-potential firing is a fundamental means by which a neuron processes information. How the mitochondrial Ca²⁺ removal process is related to the firing frequencies of intact nerve terminals has not been investigated.

The presynaptic nerve terminals at the bullfrog sympathetic ganglia release neuropeptide luteinizing hormone-releasing hormone (LHRH) (Jan and Jan 1982). The small diameters of these nerve terminals (0.5-4.4 µm) allow Ca²⁺ entering through the plasma membrane channels during nerve firing to equilibrate throughout the terminals within 10 ms (Peng and Zucker 1993). This time lapse is much shorter than the synaptic delay for LHRH release, which is hundreds of milliseconds. Because both the neuropeptide-containing, dense-cored vesicles and the mitochondria are typically located away from the active zones of the terminals (Taxi 1967) and because LHRH release typically requires seconds of repetitive nerve firing, the Ca²⁺ transients evoked at the terminals are likely to be affected by mitochondrial Ca²⁺ uptake during exocytosis of the dense-cored vesicles. Furthermore, the rate and the total amount of both LHRH release and the presynaptic Ca²⁺ elevation steeply depend on the firing frequency, with 2 Hz as the minimal and 20 Hz as the optimal frequency (Peng and Horn 1991; Peng and Zucker 1993). In this work I first tested whether mitochondria played a role in modulating the presynaptic Ca²⁺ transients that were permissive for neuropeptide transmission. Then I investigated whether the mitochondrial effects on Ca²⁺ transients depend on the stimulation frequency. The relative potentials of the ryanodine-sensitive store and of the mitochondria to affect the Ca²⁺ transients were also studied.

This work was reported previously in abstract form (Peng 1996a, 1997).

	METHODS

Abstract Introduction Methods Results Discussion References

Preparation of isolated bullfrog sympathetic ganglia, electrical stimulation of the presynaptic nerve, selective filling of preganglionic nerve terminals with membrane-impermeant fura-2 pentapotassium salt, and fura-2 fluorimetric measurements of [Ca²⁺]_i in these terminals were carried out as described previously (Dodd and Horn 1983; Peng and Horn 1991; Peng and Zucker 1993). Briefly, preparations containing paravertebral ganglia 8-10 were isolated from 12- to 18-cm bullfrogs (Rana catesbiana). The sympathetic chain was cut ~4 mm rostral to ganglion 9. A grain of fura-2 pentapotassium was placed at the cut end of the sympathetic chain, which was placed on a small platform. The presynaptic axons were filled with the dye molecules within ~2 h after cutting, and after refrigerating an additional 2-10 h at ~4°C the terminals were filled. The presynaptic nerves were stimulated electrically via a suction electrode, which was fitted tightly to the cut end of the sympathetic chain. Fura-2 fluorescence emission from a group of terminals apposed to individual C neurons was measured by a photomultiplier tube (Thorn EMI, Middlesex, UK). A group of such terminals will be called a unit. The Ca²⁺ concentration was calculated as described previously (Peng and Zucker 1993). The fluorimetric data were digitized at 0.1-1 kHz. Fura-2 pentapotassium salt was obtained from Molecular Probes (Eugene, OR).

Normal Ringer solution contained (in mM) 115 NaCl, 2 KCl, 1.1 CaCl₂, and 2 N-2-hydroxyethylpiperazine-N'-2-ethanesulfonic acid (HEPES), pH 7.25-7.26. The bicarbonate sucrose Ringer solution contained (in mM) 92 NaCl, 2 KCl, 1.8 CaCl₂, 5 NaHCO₃, 20 NaHEPES, and 5 sucrose, pH 7.25-7.26.

Carbonyl cyanide m-chlorophenylhydrazone (CCCP; 10 µM), ryanodine (10 µM), and oligomycin (10 µM) were applied through the bathing solution. CCCP and oligomycin were made by a 1:1000 dilution of 10 mM stock solution in ethanol. CCCP and ryanodine were obtained from Calbiochem (La Jolla, CA). Salts and oligomycin were from Sigma (St. Louis, MO).

Because the effects of CCCP, oligomycin, and ryanodine were not readily reversible, data from a single unit were obtained from each preparation. For each unit, one to no more than three different stimulations were used to evoke intraterminal Ca²⁺ transients, first in normal Ringer solution and then in CCCP. When more than one stimulation train was used, these were typically separated by 2 min unless otherwise specified. The first response in CCCP was taken after 5 min of CCCP application. Because responses in CCCP for each unit took a total of 2-8 min, most experiments were completed between 7 and 13 min of CCCP application. For the experiments in which additional effects of ryanodine were investigated, another two or three responses were evoked. All measurements for various drugs' effects were made between responses from the same unit that were evoked by the same stimulation pattern in control and in drug. Results for group data are presented as means ± SE.

	RESULTS

Abstract Introduction Methods Results Discussion References

Ca²⁺ transients in the presynaptic terminals of the bullfrog sympathetic ganglia were induced by electrical shocks to the cut end of the nerve fibers and were monitored by fura-2 photometry. The effect of mitochondria on the intraterminal Ca²⁺ transients was studied by a protonophore CCCP, which collapses the mitochondrial membrane potential.

Effects of CCCP on intraterminal Ca²⁺ transients

Collapsing of the mitochondrial membrane potential could block the Ca²⁺ uniporters and reduce the mitochondrial ATP production because both processes are driven by this potential. However, the CCCP effects described below were unlikely to be caused by a reduction of the cytosolic [ATP]/[ADP] [P_i] ratio or an altered intraterminal pH. The three reasons for this conclusion are discussed in the following sections.

AMPLITUDES OF ACTION POTENTIALS IN C NEURONS WERE NOT CHANGED BY CCCP TREATMENT PROTOCOL. A single brief electrical shock to presynaptic C fibers generates action potentials in these fibers that propagate down to their terminals to evoke acetylcholine (ACh) release. ACh released elicits excitatory postsynaptic potentials (EPSPs), which result in orthodromic action potentials in the postsynaptic C neurons. Electrical shocks to the sialic nerve evoked action potentials in the axons of the C neurons, and the propagation of these potentials back into their somata can be recorded as antidromic action potentials. The presynaptic C fibers and the axons of the C neurons are both unmyelinated fibers that conduct slowly; the former conduct slightly faster than the latter at mean velocities of 0.41 and 0.32 m/s, respectively (Dodd and Horn 1983). The generation and the amplitudes of these action potentials depend on the Na⁺ electrochemical gradient in the presynaptic C fibers and their terminals in the somata and the axons of the C neurons. The gradient is maintained by the Na⁺/K⁺ adenosinetriphosphatases (ATPases).

Ouabain (1 mM) was used to block the Na⁺/K⁺ ATPase directly while both the antidromic and the orthodromic action potentials in C neurons were recorded. Because the effect of ouabain was only partially recovered after >40 min wash, three cells in three different preparations were studied. As shown in Fig. 1, the amplitudes of both action potentials were greatly reduced within 1 min of drug application (Fig. 1, B and E), and both failed after 10 min in ouabain (Fig. 1, C and F). However, stimulation to the presynaptic C fibers still elicited EPSP due to ACh release from the presynaptic terminals (Fig. 1F). Therefore antidromic action potential is more susceptible to the blockade of the Na⁺/K⁺ ATPase than the potential propagation into the presynaptic C terminals. These results demonstrated that the generation and the amplitudes of the antidromic as well as of the orthodromic action potentials in the C neurons were indeed sensitive indicators for the intracellular [ATP]/[ADP] [P_i] ratio in the pre- and the postsynaptic C fibers.

View larger version (29K): [in this window] [in a new window]   FIG. 1. Effects of CCCP (10 µM) and ouabain (1 mM) on action potentials in C neurons. A-F: antidromic (A-C) and orthodromic (D-F) action potentials were recorded from a single C neuron. G-L: antidromic (G-I) and orthodromic (J-L) action potentials were recorded from another C neuron. For each cell the orthodromic action potentials were recorded in the same bathing solutions as those labeled for the antidromic action potentials.

In contrast to the effects of ouabain, 10 µM CCCP did not alter either the antidromic or the orthodromic action potentials for 23 min (Fig. 1, H and K). Both action potentials failed after 42 min of CCCP treatment, whereas the graded potentials persisted (Fig. 1, I and L). Similar results were obtained in another three C neurons. These data suggest that the intraterminal [ATP]/[ADP] [P_i] ratio was sustained in CCCP for >23 min.

An alternative ATP synthesis pathway is glycolysis in the cytosol. The C terminals are rich in glycogen bodies, which are condensed glycogen. In fact the volume of glycogen bodies in a given terminal is comparable to the volume of the mitochondria (A. Lysakowski, H. T. Figueras, S. D. Price, and Y.-Y. Peng, unpublished data). This abundant supply of substrate appeared to have enabled glycolysis to compensate for a reduction in mitochondrial ATP synthesis for >20 min.

INHIBITION OF MITOCHONDRIAL ATP SYNTHASE DID NOT AFFECT INTRATERMINAL Ca²⁺ TRANSIENTS. Oligomycin (10 µM) was used to inhibit the ATP synthase on the inner mitochondrial membrane. Intraterminal Ca²⁺ transients evoked by 300 stimuli delivered at 20 Hz were recorded before and after application of oligomycin. To normalize for drug-induced changes in resting [Ca²⁺]_i, net peak [Ca²⁺]_i, defined as the difference between the peak and the resting [Ca²⁺]_i, was used to measure the drugs' effects. As shown in Fig. 2A, oligomycin decreased the net peak [Ca²⁺]_i by 12% without altering any other characteristics of the Ca²⁺ transients. Similarly, oligomycin caused only 11.5 and 20% reductions in the net peak [Ca²⁺]_i in another two units. Oligomycin had no effects on the rise and the decay phases of the Ca²⁺ transients.

View larger version (16K): [in this window] [in a new window]   FIG. 2. Effects of oligomycin (Olig, 10 µM), bicarbonate Ringer solution (5 mM), 0.1% ethanol, and CCCP (10 µM) on intraterminal Ca2+ transients evoked by 20-Hz stimulation to the presynaptic nerves. A-C: records in each panel were obtained from a single unit. Dashed line (CCCP trace) and solid lines (normal Ringer traces) in A and B are the extrapolated [Ca2+]i levels for periods when illumination was turned off to avoid unnecessary photobleaching of fura-2. Dotted vertical lines, the end of stimulation. A: Ca2+ transients were evoked by 300 stimuli. Response in oligomycin was recorded 12 min after its addition. Response in CCCP was recorded 8 min after the addition of CCCP, when the preparation was bathed in oligomycin for 24 min. Dotted horizontal line, resting [Ca2+]i level. B: Ca2+ transients were evoked by 600 stimuli in a single unit when the intracellular pH was buffered using a bicarbonate Ringer solution. C: Ca2+ transients were evoked by 600 stimuli in a single unit. Response in ethanol was taken 7 min after the bathing solution was changed to normal Ringer + 0.1% ethanol. Response in CCCP was taken 7 min after the addition of CCCP. Nine minutes in ethanol did not alter the [Ca2+]i responses, whereas CCCP did so during the same period.

DIFFERENCE IN EFFECTS OF CCCP ON INTRATERMINAL Ca²⁺ TRANSIENTS FROM EFFECTS OF INHIBITION OF THE MITOCHONDRIA ATP SYNTHASE. CCCP was added to the bathing solution after the tissue was exposed to oligomycin for 8-17 min. The CCCP effects were recorded after 15-24 min in oligomycin. In sharp contrast to the effect of oligomycin on the Ca²⁺ transients (i.e., a small decrease in their peak amplitudes), addition of CCCP after oligomycin produced four pronounced changes in Ca²⁺ transients. First, the net peak [Ca²⁺]_i was increased by 56 ± 20.3% (means ± SE, n = 6). Second, the rates of the initial decay were decreased. Third, the final decays to the resting levels were sped up. Finally, the rate of rise during the later part of the stimulation was increased when compared with the response in normal Ringer solution (Fig. 2A). Data presented below will show that CCCP without oligomycin also produced the same set of effects on the amplitudes and the dynamics of the intraterminal Ca²⁺ transients.

These results first suggest that within 24 min the blockade of mitochondrial ATP synthase did not significantly reduce the [ATP]/[ADP] [P_i] ratio in the cytosol of these terminals. Second, effects of CCCP described in the following sections were not due to reduction of the [ATP]/[ADP] [P_i] ratio.

Collapsing of the mitochondrial membrane potential might alter intraterminal pH. As shown in Fig. 2B, when intraterminal pH was buffered by addition of 5 mM NaHCO₃ in the Ringer solution (n = 6 units), the effects of CCCP were similar to those studied in normal Ringer solution. Thus possible changes of intraterminal pH cannot account for the CCCP effects on the Ca²⁺ transients.

The effects of CCCP were not caused by its ethanol content either, as shown in Fig. 2C, where 0.1% ethanol alone did not alter the intraterminal Ca²⁺ responses measurably, whereas CCCP did so. For 14 responses in 5 units, 0.1% ethanol and 10 µM CCCP caused an average of 0.6 and 122% increases in net peak [Ca²⁺]_i, respectively.

Effects of CCCP on resting [Ca²⁺]_i and the plateau in the decay phase of intraterminal Ca²⁺ transients

Without any previous stimulation, CCCP (10 µM) did not alter the resting [Ca²⁺]_i appreciably (0 and 5 nM in 2 units). This result suggests that when there was no previous Ca²⁺ influx through the voltage-gated Ca²⁺ channels on the plasma membrane, the mitochondria at the nerve terminals contained very little Ca²⁺ that could be released by CCCP.

Seven to 10 min after the last stimulation in normal Ringer solution, the resting [Ca²⁺]_i was largely restored to its prestimulation level, suggesting that most of the Ca²⁺ accumulated in the terminals during nerve firing had left the cytosol of the terminals. The slight or nonexistent increase (53 ± 14 nM) in resting Ca²⁺ (86 ± 13 nM, n = 19 units) by subsequent CCCP application indicated that within 7-10 min most of the Ca ions accumulated by the mitochondria during the stimulations had left as well.

In normal Ringer solution the [Ca²⁺]_i reached a plateau after a rapid decay on cessation of the stimulation (Fig. 2). The plateau lasted for 4-7 min at [Ca²⁺]_i levels that ranged from 50 to 300 nM, with a mean of 128 ± 9 nM (n = 48 responses in 19 units). The plateau was 42 nM above the resting [Ca²⁺]_i level. The plateau in the decay phase was largely or totally (Fig. 2, A and B) abolished by CCCP.

CCCP increased peak amplitude and net peak elevation of [Ca²⁺]_i

To normalize the increments in resting [Ca²⁺]_i caused by CCCP, its effects on the net peak elevation of [Ca²⁺]_i were reported. Intraterminal [Ca²⁺]_i responses were evoked by both a brief (30 stimuli) and a long (600 or 800 stimuli) 20-Hz stimulation train separated by 2 or 4 min. These stimulations were selected because 30 stimuli evoked little LHRH release, whereas 600 and 800 stimuli evoked a large LHRH release (unpublished data). As shown in Fig. 3, CCCP produced a small effect on the peak amplitude of [Ca²⁺]_i evoked by the brief trains (Fig. 3A) and a larger effect on the peak amplitude of [Ca²⁺]_i evoked by the long trains (Fig. 3B). On average, CCCP increased the net peak elevation of [Ca²⁺]_i produced by the brief and the long stimulations by 24.2 ± 10.7% and 100.2 ± 27.7% (n = 19 units), respectively (Table 1).

View larger version (18K): [in this window] [in a new window]   FIG. 3. Effects of CCCP (10 µM) on intraterminal Ca2+ transients evoked by 20- and 4-Hz nerve stimulations in a single unit. A-C: stimulations used to elicit responses are indicated at the top; ---, the beginning and the end of the stimulation; - - -, the resting [Ca2+]i level, correspondingly. B: - - -, the extrapolated [Ca2+]i level for the period when the illumination was turned off to avoid unnecessary photobleaching of fura-2 during the [Ca2+]i plateau. , [Ca2+]i level where the response in CCCP began to be larger than the response in normal Ringer solution. Inset: responses in normal Ringer and in CCCP to 120 stimuli delivered at 20 Hz, which corresponded to the 1st 6 s of the responses in B. , point where the 2 responses begin to diverge; its time of occurrence and [Ca2+]i level are given in parentheses. C:  and - - -, are the same as described for B.

Frequency dependence of CCCP augmentation on intraterminal [Ca²⁺] transients during nerve firing

The small effect of CCCP on the intraterminal Ca²⁺ elevation evoked by the brief train might be due to either the lower peak amplitude of [Ca²⁺]_i (343 ± 37 nM) of these transients or to the brevity of the rising phase of the response when the mitochondrial Ca²⁺ removal process had only a short time to operate. To distinguish between these two possibilities, I added another stimulation train of 120 shocks delivered at 4 Hz either before or after the long (600 stimuli) 20-Hz train in 10 of the 19 units. The order of delivery of the 4 Hz and the long 20-Hz stimulations did not affect the relative amounts of CCCP facilitation for the responses to these stimulations. The 4-Hz train produced a similar peak [Ca²⁺]_i as did the brief 20-Hz train (264 vs. 251 nM, Table 2) but had the same duration as did the prolonged 20 Hz train.

DIFFERENT AMPLITUDES OF CCCP EFFECT. Neither peak amplitude of [Ca²⁺]_i elevation in normal Ringer solution nor the duration of stimulation could account for the different amplitudes of the CCCP effect on the net peak [Ca²⁺] elevation.

As illustrated in Fig. 3 and summarized in Table 2, CCCP produced a much larger proportional increment for the net peak [Ca²⁺]_i elevation evoked by the 4-Hz train than it did for the net peak [Ca²⁺]_i elevation evoked by the short 20-Hz train. This suggests that the main cause for the small effect of CCCP on the peak [Ca²⁺]_i elevations evoked by the brief 20-Hz trains was not their low amplitude.

CCCP also affected the responses to 4-Hz stimulation much more strongly than it influenced the responses to the long 20-Hz stimulation, even though the two stimulations had the same duration.

To understand better how peak amplitude of [Ca²⁺] in normal Ringer solution and the duration of stimulation related to the amplitude of CCCP effect without prolonging tissue exposure to CCCP, I measured both parameters at various time points of the responses during the 4-Hz and the long 20-Hz stimulations. For every time point, including the earliest one at 1.5 s, 20-Hz stimulation evoked a higher peak [Ca²⁺]_i elevation (Fig. 4A), which was affected less by CCCP (Fig. 4B). For responses to 4-Hz stimulation, when the peak amplitude of [Ca²⁺]_i was >200 nM, the effects of CCCP became much larger (Fig. 4C). Again, this effect was stronger for the responses to 4-Hz stimulation than for the responses to 20-Hz stimulation (Fig. 4C).

View larger version (31K): [in this window] [in a new window]   FIG. 4. Averaged peak amplitude of [Ca2+]i in normal Ringer (A) and the averaged CCCP increments of the net peak [Ca2+]i elevations (B) were plotted against the duration of 4- and 20-Hz stimulations. C: mean CCCP increments of the net peak [Ca2+]i elevations were plotted against their mean peak amplitudes of [Ca2+]i in normal Ringer. D: mean values of peak [Ca2+]i in normal Ringer in response to both 4- and 20-Hz stimulation and the mean CCCP effects on the net peak [Ca2+]i elevations were normalized to their corresponding values measured 30 s after the stimulations started. Values were obtained by measurements at 1.5, 3, 6, 10, 20, and 30 s after the stimulation began in 19 units where responses were evoked by 20-Hz stimulations and in 10 of these 19 units where responses were evoked by 4-Hz stimulations. Error bars in A and B indicate SE.

The above results showed that neither the peak amplitude of the response nor the duration of the stimulation could explain the differential effects of CCCP on responses evoked by the 4- and 20-Hz stimulations. The other difference between these stimulations was their frequencies. Previous studies showed that, for a given number of stimuli, 20-Hz stimulation evoked both greater LHRH release and higher presynaptic [Ca²⁺] elevation at faster rates than did 4-Hz stimulation (Peng and Horn 1991; Peng and Zucker 1993). Direct comparisons were therefore made between CCCP effects on responses to 120 stimuli delivered at 20 and 4 Hz. As illustrated in Fig. 3C and in Fig. 3B, inset, the CCCP increments were 57 and 191%, respectively, for responses to the 20- and 4-Hz stimulations. On average, the CCCP increment of the net peak [Ca²⁺]_i elevation evoked by the 20-Hz stimulation was only 13% of that for the responses to the 4-Hz train (Table 2).

This result, together with the differential CCCP effects on responses evoked by stimulations of the same durations but different frequencies, suggests that the higher the rate of Ca²⁺ influx into the cytosol the less is the mitochondrial modulation of the intraterminal [Ca²⁺].

[Ca²⁺]_i increased almost monotonically in CCCP

When the responses in CCCP are superimposed on those recorded in normal Ringer solution, the effect of CCCP on the rate of [Ca²⁺]_i rise can be seen starting at the point where the response in CCCP was higher than that in normal Ringer solution (Figs. 2 and 3; see also Figs. 5 and 6). Clearly, CCCP had no or only a slight effect on the rate of rise of [Ca²⁺]_i within the first 1 to 3 s of stimulation (Fig. 3; also see Figs. 5A and 6A). The rate of rise was maintained for the latter part of the stimulation in CCCP whereas it was decreased in normal Ringer solution (Figs. 2 and 3, B and C; see also Figs. 5B and 6B). Moreover, for a given unit the diverging points were at higher [Ca²⁺]_i levels for responses evoked by the 20-Hz stimuli than for those evoked by the 4-Hz stimuli, i.e., 383 versus 120 nM for the diverging points (Fig. 3, B and C, ).

View larger version (20K): [in this window] [in a new window]   FIG. 5. Effects of CCCP (10 µM) and ryanodine (10 µM) on intraterminal Ca2+ transients evoked by 20-Hz nerve stimulations in a single unit (A and B). The dotted vertical lines delimit the period of stimulation. A and B insets: 1st 1.5 (A) and 30 (B) s. C: difference traces were calculated by subtraction of the responses in normal Ringer solution from those in CCCP plus ryanodine solution in A and B. The dotted horizontal line indicates the level of changes in resting [Ca2+]i; the dashed vertical line indicates the 600th stimuli. The difference response evoked by 600 stimuli reversed its polarity 6 s after the stimulation began.

View larger version (17K): [in this window] [in a new window]   FIG. 6. Effects of CCCP (10 µM) and ryanodine (10 µM) on intraterminal Ca2+ transients evoked by 20-Hz stimulations in a single unit. A inset: 1st 6 s of the responses evoked by 30 stimuli. A and B, ···: resting [Ca2+]i level. (for normal Ringer traces) and - - -, (for the CCCP traces) are the extrapolated [Ca2+]i levels for the period when the illumination was turned off to avoid unnecessary photobleaching of fura-2. C: difference traces were calculated from responses in A and B. - - -, level of changes in [Ca2+]i; , the 600th stimuli.

In normal Ringer solution, many Ca²⁺ responses evoked by prolonged stimulations (200 stimuli) had a plateau and some even had a decay phase during the stimulation (Fig. 5B) (Fig. 3 in Peng 1994). In contrast, [Ca²⁺]_i increased almost monotonically in CCCP. When the normalized means of both the peak amplitude of [Ca²⁺]_i in normal Ringer solution and the CCCP augmentation of the net peak [Ca²⁺] elevations were plotted against the durations of stimulations, the time courses for these two parameters to reach their peak values were quite different (Fig. 4D). During the first 6 s, responses to stimulations of 4 and 20 Hz increased rapidly in normal Ringer solution to reach 0.65 and 0.78 of their respective plateau levels, whereas CCCP effects remained <0.33 of their peak values. CCCP effects accelerated thereafter, whereas the peak amplitude of [Ca²⁺]_i in normal Ringer solution reached their plateau levels. This inverse relationship suggests that the slow activation and the subsequent steady increase of mitochondria Ca²⁺ removal might be a mechanism that allowed the fast initial rise of intraterminal [Ca²⁺]_i and underlay its subsequent plateau during stimulation in normal Ringer solution.

CCCP slowed down [Ca²⁺]_i decay after cessation of stimulation

Similar to the measurements of CCCP effects on net peak elevation of [Ca²⁺]_i, effects of CCCP on [Ca²⁺]_i decay were measured by comparing responses evoked by the same stimulation recorded from the same unit in normal Ringer solution and in CCCP. CCCP slowed down the [Ca²⁺]_i decay after cessation of the stimulation for all responses (n = 48) in the 19 units, regardless of the duration and the frequency of the stimulation and of the peak amplitude of [Ca²⁺]_i reached in normal Ringer solution (Figs. 2 and 3; see also Figs. 5 and 6). Thus Ca²⁺ removal by the mitochondria under control conditions sped up the Ca²⁺ decay for all of the responses.

Dynamic interaction between Ca²⁺ uptake by the mitochondria and Ca²⁺ release from the ryanodine-sensitive store

Previous work showed that Ca²⁺ released from the smooth endoplasmic reticulum via the ryanodine-sensitive channels, on average, accounts for 46% of the peak [Ca²⁺]_i elevation evoked by 20-Hz stimulation in normal Ringer solution (Peng 1996b). Because CCCP increased the amplitude of [Ca²⁺]_i transients by disabling Ca²⁺ removal by the mitochondria, it is possible for the accumulated intraterminal Ca²⁺ to enhance the Ca-induced Ca release (CICR) through the ryanodine-sensitive channels. This process should further amplify the Ca²⁺ signal in CCCP. This possibility was tested in 11 units by addition of ryanodine (10 µM), a blocker of the CICR process, after CCCP. As illustrated in Fig. 5B, 6 s after the stimulation began the response in CCCP plus ryanodine was greater than the response in normal Ringer solution but much reduced as compared with that in CCCP alone. This was the case for responses evoked by 600 stimuli (20 Hz) in another two units. Ryanodine has been shown to inhibit the peak Ca²⁺ elevation evoked by 20-Hz stimulation (Peng 1996b). For these three units, after CCCP treatment the peak Ca²⁺ elevation was higher in ryanodine than in normal Ringer solution. This supported the postulated secondary effects of CCCP on CICR.

In contrast, in the rest of the eight units, responses to 600 stimuli in CCCP plus ryanodine reached a peak amplitude lower than that in normal Ringer solution (Fig. 6B). This is not surprising, given the large effects of ryanodine (up to 82% inhibition) (Peng 1996b) on [Ca²⁺]_i elevation for many terminals. Interestingly, in all 11 units response to 30 stimuli delivered at 20 Hz in CCCP plus ryanodine was smaller than that in normal Ringer solution (Figs. 5A and 6A).

Because fluxes caused by ryanodine-sensitive release and CCCP-sensitive removal flow in opposite directions, their relative absolute values in a given unit should determine the net effect of CCCP plus ryanodine for the unit. In other words, in the presence of both ryanodine and CCCP the response will be smaller than in normal Ringer solution if CICR is greater than mitochondrial Ca²⁺ removal. Conversely, [Ca²⁺]_i elevation in ryanodine and CCCP will be larger than in normal Ringer solution if CICR is less than mitochondrial Ca²⁺ removal. The ryanodine-sensitive release includes both the release induced by Ca²⁺ influx through the plasma membrane and the release secondary to uncoupling of the mitochondria.

When responses in normal Ringer solution were subtracted from responses in CCCP plus ryanodine, the difference traces had different waveforms for different units (compare Fig. 5C with Fig. 6C). However, for the first 6 s of the responses, the difference traces were always negative relative to the change in resting [Ca²⁺]_i. Therefore the influx through the ryanodine channels could be larger than the efflux caused by the mitochondria at this early part of all of the responses. This was the situation for the first 1.5 s of 20-Hz stimulation because the response to 30 stimuli in CCCP plus ryanodine was smaller than that in normal Ringer solution for all the 11 units. A possible mechanism for this phenomenon is that the ryanodine-sensitive store could affect the [Ca²⁺]_i at a faster rate than did the removal by the mitochondrial Ca²⁺ uniporters.

On cessation of stimulation, mitochondrial Ca²⁺ uptake overrode release through the ryanodine-sensitive channels.

	DISCUSSION

Abstract Introduction Methods Results Discussion References

The nerve-evoked Ca²⁺ transients at the bullfrog sympathetic presynaptic nerve terminals had a similar range of Ca²⁺ elevation (200-2,000 nM) and a plateau phase in their decay, as did the transients reported for many other systems cited in the INTRODUCTION.

In four units, responses in CCCP not only had reduced peak [Ca²⁺]_i elevation but also failed to decay to their prestimulation levels over 10 min. Data from these units were discarded. These results are consistent with significantly reduced [ATP]/[ADP] [P_i] ratios. Similar results were reported by Budd and Nicholls (1996) when the mitochondria Ca²⁺ uniporters were inhibited after oligomycin treatment. In sharp contrast to results from these four units, in all the other units studied, CCCP, with or without previous exposure to oligomycin, not only increased the net peak [Ca²⁺] elevation but also sped up the final decay of intraterminal [Ca²⁺] (Figs. 2, 3, and 6). Both effects were opposite to those predicted by a reduction of the intraterminal [ATP]/[ADP] [P_i] ratio.

Because oligomycin did not change most of the rise phase and the entire decay phase of the Ca²⁺ transients, its small effect on the peak amplitude might be nonspecific. The slight effects of oligomycin suggest that the terminals maintained their [ATP]/[ADP] [P_i] ratio via nonmitochondrial means. One such means is to increase glycolysis in the cytosol. Indeed, the [ATP]/[ADP] [P_i] ratio in synaptosomes and in the soma and neurites of cultured cerebellar granule cells was maintained for up to 15 min in 10 µM oligomycin because of increased glycolysis (Budd and Nicholls 1996; Kauppinen and Nicholls 1986).

The results of the oligomycin experiments are consistent with the conclusion that the CCCP effects found in the present work are unlikely to be the result of a reduction of the [ATP]/[ADP] [P_i] ratio at the terminals. Although this conclusion might be further supported by applying CCCP before oligomycin, the sensitivity of the ouabain experiments rendered this alternative unnecessary.

The increased peak elevation, the sustained rate of rise, together with a slower initial decay were all consistent with CCCP blockade of the mitochondrial Ca²⁺ uniporters.

In these terminals, CCCP not only increased the peak elevation of [Ca²⁺]_i evoked by the brief 20-Hz trains but also slowed down their rapid decay phase (panel A in Figs. 3, 5, and 6). This is different from the results in the somata of adrenal chromaffin cells, where CCCP did not affect responses with peak amplitudes 400 nM (Herrington et al. 1996), and in rat isolated neurohypophysial nerve endings, where mitochondria were not involved in intracellular Ca²⁺ buffering when [Ca²⁺]_i was 600 nM (Stuenkel 1994).

Instead of the peak amplitude of [Ca²⁺]_i being the apparent cause for mitochondrial involvement in intracellular Ca²⁺ dynamics (Herrington et al. 1996; Stuenkel 1994), in the present study I found that the amplitudes of the effect that mitochondrial Ca²⁺ removal had on the intraterminal [Ca²⁺] depend on the frequency of nerve firing. The effect was much larger for responses to 4-Hz stimulation than for responses to 20-Hz stimulation. This increase was the case when responses evoked by the two frequencies had similar peak amplitudes, when the two stimulations lasted for the same duration, and when they contained the same number of stimuli, even though in the last two situations the peak amplitudes of [Ca²⁺]_i evoked by the 4-Hz stimulation were only a fraction of the peak amplitudes of [Ca²⁺]_i produced by the 20-Hz stimulation.

The frequency dependence of mitochondrial effects might reflect a mismatch between the rates of mitochondrial Ca²⁺ uptake and Ca²⁺ influx during neural activity. The mismatch appeared larger when the nerve terminals were activated at 20 Hz than when they were activated at 4 Hz. Specifically, the rate of Ca²⁺ influx evoked by 20-Hz stimulation was much higher than the rate of mitochondrial Ca²⁺ removal, such that elimination of the latter caused a moderate change. On the other hand, the rate of Ca²⁺ influx produced by 4-Hz stimulation was better counterbalanced by the rate of mitochondrial removal; therefore elimination of the latter caused a large increase in the peak amplitude of the responses.

Effects of CCCP on the Ca²⁺ transients evoked by 20-Hz stimulation had a slower onset than the delay allowed by diffusion of the Ca²⁺ ions entering through the plasma membrane to activate the Ca²⁺ uniporter on the mitochondria (~10 ms). Meanwhile, during this delay Ca²⁺ rose at its greatest rate. The influx through the ryanodine-sensitive channels also reached its highest rate 0.5-3 s after the beginning of a 20-Hz stimulation (Peng 1996b).

In summary, uncoupling of mitochondria from the Ca²⁺ dynamics of the nerve terminals by CCCP increased both the amplitude of Ca²⁺ elevation evoked by nerve firing and its rate of rise and decreased its rate of decay. These effects add up to increase the time integral of the Ca²⁺ transients. Thus, when their function was intact, mitochondria limited the time integral of Ca²⁺ elevation. Presynaptic Ca²⁺ responses evoked by 4-Hz firing were affected much more strongly by CCCP than were responses evoked by 20-Hz firing. Also, 4-Hz stimulation produces less LHRH release at a lower rate than does 20-Hz stimulation (Peng and Horn 1991). Therefore besides the obvious difference in the rate of Ca²⁺ influx, mitochondrial Ca²⁺ removal is likely to be the other major mechanism that underlies the frequency dependence of intraterminal [Ca²⁺] dynamics as well as the frequency dependence of LHRH release.

	ACKNOWLEDGEMENTS

  The author thanks H. T. Figueras for assistance in setting up the preparations and E. Lanzl for editing the manuscript.

  This work was supported by the Alfred P. Sloan Foundation and by National Institute of Neurological Disorders and Stroke Grant NS-32429.

	FOOTNOTES

  Received 3 December 1997; accepted in final form 27 March 1998.

	REFERENCES

Abstract Introduction Methods Results Discussion References

• BLEAKMAN, D., ROBACK, J. D., WAINER, B. H., MILLER, R. J., HARRISON, N. L. Calcium homeostasis in rat septal neurons in tissue culture. Brain Res. 600: 257-267, 1993.[Medline]
• BUDD, S. L., NICHOLLS, D. G. A reevaluation of the role of mitochondria in neuronal Ca²⁺ homeostasis. J. Neurochem. 66: 403-411, 1996.[Medline]
• DODD, J., HORN, J. P. A reclassification of B and C neurones in the ninth and tenth paravertebral sympathetic ganglia of the bullfrog. J. Physiol. (Lond.) 334: 255-269, 1983.[Abstract/Free Full Text]
• FRIEL, D. D., TSIEN, R. W. An FCCP-sensitive Ca²⁺ store in bullfrog sympathetic neurons and its participation in stimulus-evoked changes in [Ca²⁺]_i. J. Neurosci. 14: 4007-4024, 1994.[Abstract]
• GUNTER, K. K., GUNTER, T. E. Transport of calcium by mitochondria. J. Bioenerg. Biomembr. 26: 471-485, 1994.[Medline]
• HERRINGTON, J., PARK, Y. B., BABCOCK, D. F., HILLE, B. Dominant role of mitochondria in clearance of large Ca²⁺ loads from rat adrenal chromaffin cells. Neuron 16: 219-228, 1996.[Medline]
• JAN, L. Y., JAN, Y. N. Peptidergic transmitters in sympathetic ganglia of the frog. J. Physiol. (Lond.) 327: 219-246, 1982.[Abstract/Free Full Text]
• KAUPPINEN, R. A., NICHOLLS, D. G. Failure to maintain glycolysis in anoxic nerve terminals. J. Neurochem. 47: 1864-1869, 1986.[Medline]
• MILLER, R. J. The control of neuronal Ca²⁺ homeostasis. Prog. Neurobiol. 37: 255-285, 1991.[Medline]
• PARK, Y. B., HERRINGTON, J., BABCOCK, D. F., HILLE, B. Ca²⁺ clearance mechanisms in isolated rat adrenal chromaffin cells. J. Physiol. (Lond.) 492: 320-346, 1996.
• PENG, Y. Y. Dynamics of presynaptic Ca²⁺ and peptide release in bullfrog sympathetic ganglia. Comput. Neurons Neural Syst. 39-44: 1994.
• PENG, Y. Y. The dynamics of mitochondria Ca²⁺ removal and its effect on neuropeptide transmission at intact nerve terminals. Biophys. Soc. Abstr. 70: A197, 1996a.
• PENG, Y. Y. Ryanodine-sensitive component of calcium transients evoked by nerve firing at presynaptic nerve terminals. J. Neurosci. 16: 6703-6712, 1996b.[Abstract/Free Full Text]
• PENG, Y. Y. Effects of mitochondrion on calcium transients at intact presynaptic terminals depend on frequency of nerve firing. Neurosci. Soc. Abstr. 23: 11.1, 1997.
• PENG, Y. Y., HORN, J. P. Continuous repetitive stimuli are more effective than bursts for evoking LHRH release in bullfrog sympathetic ganglia. J. Neurosci. 11: 85-95, 1991.[Abstract]
• PENG, Y. Y., ZUCKER, R. S. Release of LHRH is linearly related to the time integral of presynaptic Ca²⁺ elevation above a threshold level in bullfrog sympathetic ganglia. Neuron 10: 465-473, 1993.[Medline]
• STUENKEL, E. L. Regulation of intracellular calcium and calcium buffering properties of rat isolated neurohypophysial nerve endings. J. Physiol. (Lond.) 481: 251-271, 1994.[Abstract/Free Full Text]
• TAXI, J. Observations on the ultrastructure of the ganglionic neurons and synapses of the frog Rana esculenta L. In: The Neuron, edited by H. Hyden. Amsterdam: Elsevier, 1967, p. 221-254.
• THAYER, S. A., MILLER, R. J. Regulation of the intracellular free calcium concentration in single rat dorsal root ganglion neurones in vitro. J. Physiol. (Lond.) 425: 85-115, 1990.[Abstract/Free Full Text]
• WERTH, J. L., THAYER, S. A. Mitochondria buffer physiological calcium loads in cultured rat dorsal root ganglia neurons. J. Neurosci. 14: 348-356, 1994.[Abstract]
• WHITE, R. J., REYNOLDS, I. J. Mitochondria and Na⁺/Ca²⁺ exchange buffer glutamate-induced calcium loads in cultured cortical neurons. J. Neurosci. 15: 1318-1328, 1995.[Abstract]

0022-3077/98 $5.00 Copyright ©1998 The American Physiological Society