Mitochondria Shape Hormonally Induced Cytoplasmic Calcium Oscillations and Modulate Exocytosis*

Pituitary gonadotropes transduce hormonal input into cytoplasmic calcium ([Ca2+]cyt) oscillations that drive rhythmic exocytosis of gonadotropins. Using Calcium Green-1 and rhod-2 as optical measures of cytoplasmic and mitochondrial free Ca2+, we show that mitochondria sequester Ca2+ and tune the frequency of [Ca2+]cyt oscillations in rat gonadotropes. Mitochondria accumulated Ca2+ rapidly and in phase with elevations of [Ca2+]cyt after GnRH stimulation or membrane depolarization. Inhibiting mitochondrial Ca2+ uptake by the protonophore CCCP reduced the frequency of GnRH-induced [Ca2+]cyt oscillations or, occasionally, stopped them. Much of the Ca2+ that entered mitochondria is bound by intramitochondrial Ca2+ buffering systems. The mitochondrial Ca2+ binding ratio may be dynamic because [Ca2+]mit appeared to reach a plateau as mitochondrial Ca2+ accumulation continued. Entry of Ca2+ into mitochondria was associated with a small drop in the mitochondrial membrane potential. Ca2+ was extruded from mitochondria more slowly than it entered, and much of this efflux could be blocked by CGP-37157, a selective inhibitor of mitochondrial Na+-Ca2+ exchange. Plasma membrane capacitance changes in response to depolarizing voltage trains were increased when CCCP was added, showing that mitochondria lower the local [Ca2+]cyt near sites that trigger exocytosis. Thus, we demonstrate a central role for mitochondria in a significant physiological response.


INTRODUCTION
A mitochondrial contribution to intracellular Ca 2+ dynamics has been debated for several decades. For some time it was supposed that mitochondria do not accumulate significant Ca 2+ unless the surrounding free cytoplasmic Ca 2+ concentration ([Ca 2+ ] cyt ) 1 is dangerously high (1,2). This view seemed plausible since accumulating Ca 2+ would divert energy of the mitochondrial membrane potential from its normal function of producing ATP, and, as the endoplasmic reticulum is already specialized for sequestering and releasing Ca 2+ , regulation by yet another organelle seemed unnecessary. Nevertheless, in some cells, mitochondria do take up Ca 2+ at apparently physiological [Ca 2+ ] cyt levels (<2 µM) (3-11) (for reviews, see (12)(13)(14)).
Anterior pituitary gonadotropes secrete gonadotropins in response to gonadotropin-releasing hormone (GnRH). Binding of GnRH to its cell-surface receptors induces [Ca 2+ ] cyt oscillations, driven primarily by the release and re-uptake of Ca 2+ from inositol 1,4,5-trisphosphate-sensitive Ca 2+ stores (for reviews see (15)(16)(17)). These [Ca 2+ ] cyt oscillations trigger secretion of luteinizing hormone and follicle-stimulating hormone, which are released into the circulation as [Ca 2+ ] cyt rises transiently and repetitively above 1 µM. Local [Ca 2+ ] cyt may become even higher at sites of secretion (18). As this dynamic behavior is almost exclusively under control of intracellular organelles, gonadotropes are ideal for examining the relative contribution of intracellular organelles, including mitochondria, (19) to Ca 2+ balance during a physiological response.
6 rhod-2 images were collected sequentially with excitations of 488 and 568 nm respectively, and merged digitally.

Whole-cell recording and capacitance measurement. Solutions used for
whole-cell recordings were described earlier (26) except that tetraethyl ammonium chloride replaced NaCl in the bath solution. Membrane capacitance (C m ) measurements were performed as described (27,28). To evoke Ca 2+ entry and exocytosis, 100 ms depolarizations were given to +15 mV at 5 Hz. Currents were filtered at 2 kHz and sampled at 16.7 kHz.
Materials. The compound CGP 37157 was obtained from Tocris Cookson Inc., Ballwin, MO. Fluorescent probes were from Molecular Probes, Eugene, OR and culture media from Gibco, Grand Island, NY. All other reagents were obtained from Sigma, St.
Louis, MO or Fisher, Fair Lawn, NJ.

Protonophores alter [Ca 2+ ] cyt oscillations.
Our working hypothesis was that a significant fraction of the cellular Ca 2+ would be taken up into mitochondria during a period of GnRH-induced Ca 2+ oscillations. To block this hypothesized mitochondrial contribution to [Ca 2+ ] cyt balance, we added the protonophore CCCP, which should collapse the mitochondrial membrane potential and eliminate the electrical driving force for mitochondrial Ca 2+ uptake. Figure 1 shows that application of CCCP (2 µM) during periodic Ca 2+ oscillations dramatically slowed the rate of each downstroke of [Ca 2+ ] cyt and reduced the frequency of oscillation. In this experiment, the rate of fall was 781 ± 43 nM s -1 (n=7) in the absence of CCCP and 391 ± 47 nM s -1 (n=4) in its presence. In by guest on November 7, 2016 http://www.jbc.org/ Downloaded from 7 five fully analyzed experiments, the falling phase was slowed 40 ± 6% by adding CCCP.
A similar effect was observed in 12 out of 15 cells tested in this manner. In the remaining 3 cells, CCCP treatment stopped [Ca 2+ ] cyt oscillations completely. In either case, the effect of protonophore was reversible within 30-40 s of washout. The cell in Figure 1 was exposed to oligomycin-B (2.5 µM) throughout to inhibit the mitochondrial F 1 -F 0 ATPase and thus prevent mitochondrial ATP consumption during CCCP application. Similar results were obtained with or without oligomycin-B, suggesting that the effects of protonophore are not due to a reduction in cytoplasmic ATP. Apparently energized mitochondria mediate about 40% of the Ca 2+ clearance that terminates each cycle of physiological Ca 2+ oscillations.

Mitochondria accumulate Ca 2+ during cytoplasmic oscillations.
Accumulation of Ca 2+ in mitochondria can be measured with Ca 2+ -sensitive dyes trapped in the mitochondria. Membrane-permeant rhod-2-AM has a net positive charge and distributes preferentially into energized mitochondria where it is cleaved and made membrane-impermeant by endogenous esterases (8,29). Figure 2 shows a merged confocal image of a gonadotrope co-loaded with the AM forms of Calcium Green-1 and rhod-2. Calcium Green-1 was distributed diffusely in the cytosol whereas rhod-2 had a more punctate, compartmentalized distribution, as expected from mitochondria. With cells loaded in this way, we could monitor changes of [Ca 2+ ] in the cytosol and in the mitochondria simultaneously.
where CBR c is the cytoplasmic Ca 2+ binding ratio (taken as 100)(31) and F m is the mitochondrial volume as a fraction of cytoplasmic volume (assumed to be 0.02). ratio ranged widely and averaged 4027 ± 2031. In five of the six experiments, the maximum rate of Ca 2+ clearance from the cytoplasm occurred at the same time or a few seconds after the maximum rate of rise of free mitochondrial Ca 2+ ( Figure 6A, inset).

∆[Ca
One possible explanation is that the intramitochondrial buffer is increasing as Ca 2+ enters the mitochondria. To further test this possibility, we monitored the increase in [Ca 2+ ] mit following two successive applications of GnRH ( Figure 6B). The first application resulted in a rapid and robust increase in [Ca 2+ ] mit while the second application produced a more gradual increase even though the [Ca 2+ ] cyt oscillations were very similar in both cases. The second application of GnRH was intentionally given before the rhod-2 signal was fully recovered to baseline. If full recovery was allowed, the [Ca 2+ ] mit accumulation would appear similar in both cases as in Figure 5.

Mitochondria lower [Ca 2+ ] cyt near sites of secretion.
We have shown that mitochondria take up cytoplasmic Ca 2+ as measured by the spatially averaged fluorescence of dyes. Does this uptake have significant local physiological effects? We decided to test if mitochondria influence the Ca 2+ dynamics near sites of secretion by using simultaneous measurements of whole-cell Ca 2+ current, [Ca 2+ ] cyt , and membrane capacitance (C m ), an indicator of exocytosis. Trains of depolarizing pulses (100 ms, +15 mV) were applied to gonadotropes at 5 Hz for 2 s to elicit periodic Ca 2+ entry through voltage-gated Ca 2+ channels. They evoked exocytosis. During the train, the membrane capacitance seemed to rise in two components (Fig. 7A): the first few pulses mobilized a small (~20 fF) pool of "immediately-releasable" vesicles; then a much larger, second "readily-releasable" pool was recruited (32,33). CCCP dramatically enhanced  Figure   7B shows that CCCP modestly accelerated the progressive reduction in the Ca 2+ current, an effect that would have, by itself, tended to reduce, rather than increase, the amount of exocytosis during the voltage pulses in CCCP. These results are in general agreement with previous work on chromaffin cells where we measured clearance rates (7), mitochondrial Ca 2+ binding ratios, and the effect of CGP-37157 (8). In chromaffin cells, mitochondria are the dominant clearance mechanism (84% of clearance), but they too reduce [Ca 2+ ] cyt at rates near 400 nM s -1 when [Ca 2+ ] cyt is around 1 µM. Apparently, mitochondria of the two cell types are comparable in their uptake rates and Ca 2+ binding ratios, but in gonadotropes, uptake by the endoplasmic reticulum (22) is much faster, so that the mitochondrial component is only 40% of the total. In both cells, the mitochondrial Na + -Ca 2+ exchanger is the principal route for Ca 2+ extrusion, but the exchanger in gonadotropes must be 4-10-fold less active since their mitochondria retain Ca 2+ much longer. The present results also are in qualitative agreement with our early experiments on gonadotropes showing that CCCP can stop GnRH-induced oscillations and release Ca 2+ into the cytosol (19). Very roughly, our results also fit with classical measurements in isolated heart mitochondria.

DISCUSSION
There the maximum values of Ca 2+ and H + fluxes are given as 10 and 55 µmol g -1 s -1 , respectively (34,35), referred to dry protein. With conversion factors summarized by Scarpa (36), the Ca 2+ influx would become 7 mmol l -1 s -1 , referred to cardiac mitochondrial volume. If we take the gonadotrope's mitochondrial volume as 2% of the cell and the cytoplasmic Ca 2+ binding ratio as 100, this maximal flux of cardiac mitochondria could reduce [Ca 2+ ] cyt at a rate of 1.4 µM s -1 , quite compatible with rates we measure in gonadotropes. Now consider the mitochondrial depolarization. We view the large insidenegative mitochondrial membrane potential as the result of a steady electric current of proton export working across a very poor conductor, the mitochondrial inner membrane.
Calcium uptake is also a current and reduces the membrane potential by momentarily canceling some fraction of the oppositely directed proton current. (As we saw above, these two maximal values are similar.) When Ca 2+ influx stops, the proton current repolarizes the mitochondrial membrane potential without delay. The Ca 2+ uptakeinduced mitochondrial depolarization that we report (Fig. 4) gives only 14% of the large increase in TMRE fluorescence obtained with strong depolarization by CCCP.
Considering that by the Nernst equation a depolarization of only 18 mV would change the TMRE distribution ratio twofold, we can argue that this Ca 2+ uptake depolarizes by only few millivolts. Thus, a Ca 2+ uptake sufficient to have a significant impact on Ca 2+ oscillations and exocytosis and large enough to be detected by TMRE makes but a small reduction in the -150 to -180 mV membrane potential of energized mitochondria. 15 We have reported here spatially averaged optical measurements. Very likely there are gradients within the cytosol and heterogeneity among mitochondria (37,38), so some quantitative relationships are not well resolved. For example, the ER and mitochondria may have a special spatial relationship that allows more efficient mitochondrial removal of Ca 2+ released from the ER (39). Our lab has shown that the local [Ca 2+ ] near sites of exocytosis is higher than the spatially averaged value during GnRH-induced oscillations (18), and here we have shown that mitochondria take up Ca 2+ near sites of exocytosis. Hence, the mitochondrial uptake we measure may be rapid because it occurs in regions where the [Ca 2+ ] is considerably higher than the averages reported by our dyes. Using ideas described in other cells (33,40,41) for explaining our voltage-step experiments, the immediately-releasable pool of secretory ganules may be those docked vesicles that are near to voltage-gated Ca 2+ channels, and the readily-releasable pool may be additional vesicles that are docked farther from the Ca 2+ channels. When mitochondria are fully functional (no CCCP), they would limit the diffusional range of Ca 2+ , so little spreads beyond the immediately-releasable pool.
However, when CCCP is present, the Ca 2+ can spread to the readily-releasable pool and more vesicles are discharged over a longer period of time. It should be noted that voltage-step protocol used to induce exocytosis in experiments like those in Figure 7 might not accurately reflect hormonally-induced exocytosis, as the physiological source of Ca 2+ is from IP 3 -gated stores and not voltage-gated stores. Our attempts to measure the effects of CCCP on GnRH-induced exocytosis were prevented by the large, uncontrollable [Ca 2+ ] cyt oscillations that occur following stimulation. These oscillations result in a large mitochondrial Ca 2+ load that is partially released by the addition of CCCP which then raises [Ca 2+ ] cyt . The voltage protocol used in Figure 7 allows a short and relatively small addition of a known amount of Ca 2+ (from current records) to the system. Here the mitochondria do not take on a large Ca 2+ load and same-cell comparisons of changes in capacitance are possible.
One surprising observation was that in experiments like that of Figure 6A