Feedback Inhibition of Sodium/Calcium Exchange by Mitochondrial Calcium Accumulation*

Chinese hamster ovary cells expressing the bovine cardiac Na 1 /Ca 2 1 exchanger were subjected to two pe-riods of 5 and 3 min, respectively, during which the extracellular Na 1 concentration ([Na 1 ] o ) was reduced to 20 m M ; these intervals were separated by a 5-min recov- ery period at 140 m M Na 1 o . The cytosolic Ca 2 1 concentration ([Ca 2 1 ] i ) increased during both intervals due to Na 1 -dependent Ca 2 1 influx by the exchanger. However, the peak rise in [Ca 2 1 ] i during the second interval was only 26% of the first. The reduced rise in [Ca 2 1 ] i was due to an inhibition of Na 1 /Ca 2 1 exchange activity rather than increased Ca 2 1 sequestration since the influx of Ba 2 1 , which is not sequestered by internal organelles, was also inhibited by a prior interval of Ca 2 1 influx. Mitochondria accumulated Ca 2 1 during the first interval of reduced [Na 1 ] o , as determined by an increase in fluorescence of the Ca 2 1 -indicating dye rhod-2, which preferentially labels mitochondria. Agents that blocked mitochondrial Ca 2 1 accumulation (uncouplers, nocodazole) eliminated the observed inhibition of exchange activity during the second period of low [Na 1 ] o . Con-versely, diltiazem, an inhibitor of the mitochondrial Na 1 /Ca 2 1 exchanger, increased mitochondrial Ca 2 1 accumulation and also increased the inhibition of exchange activity. We conclude that Na 1 /Ca 2 1 exchange activity is regulated by a feedback inhibition process linked to mitochondrial Ca 2 1 accumulation. The Na 1 /Ca 2 1 exchanger is fluorescence at 520 nm with alter- nate excitation at 334 and 380 nm using an Attofluor Ratiovison digital fluorescence imaging system coupled to a Zeiss Axiovert 100 fluores- cence microscope. For measurement of Ba 2 1 influx, a second filter set was employed for alternate excitation at 350 and 390 nm (15). To initiate the Ca 2 1 influx mode of Na 1 /Ca 2 1 exchange, Na-PSS replaced with 20:120 Na/K-PSS for a period of 5 min. The then replaced with Na-PSS for an additional 5 min, followed by a second interval (3 min) in 20:120 Na/K-PSS. Fluorescence signals were cali- brated for [Ca 2 1 ] i by the procedure Grynkiewicz et al. 10 m M ionomycin, 1 m M thapsigargin, and 50 m M EGTA/AM were added to fura-2-loaded cells in Ca 2 1 -free Na-PSS containing 0.6 m M EGTA; R min was recorded after at least 5 min. CaCl 2 (10 m M ) was subsequently added to determine R max . Significance testing was done using Student’s t test. Rhod-2 Fluorescence Measurements— Cells were loaded in the dark with 2 m M rhod-2/AM plus 0.02% (w/v) pluronic F-127 in Na-PSS plus 1% bovine serum albumin for 10–12 min at room temperature. Fluorescence images were acquired using a cooled CCD camera under com- puter control. The appropriate dichroic

The Na ϩ /Ca 2ϩ exchanger is an electrogenic, high capacity transporter found in the plasma membrane of many cell types. The exchanger is the primary mechanism for Ca 2ϩ efflux in cardiac myocytes, and its activity therefore plays a central role in regulating the force of cardiac muscle contraction (reviewed in Ref. 1). The stoichiometry of the exchange process is generally considered to be three Na ϩ ions/Ca 2ϩ (2), although a higher stoichiometry has been proposed recently (3). The exchanger can transport Ca 2ϩ in either direction across the plasma membrane, and a reduction in the extracellular Na ϩ concentration ([Na ϩ ] o ) 1 leads to a net influx of Ca 2ϩ in cells expressing the exchanger. Regulation of exchange activity through positive modulation by phosphatidylinositol 4,5bisphosphate or cytosolic Ca 2ϩ has been characterized in electrophysiological experiments with excised membrane patches (4,5). Results with transfected cells, cardiac myocytes, and squid giant axons have suggested that exchange activity may also be regulated by protein kinases (6 -9) and by the actin cytoskeleton (10). The importance of these modes of regulation in intact cells under physiological conditions is uncertain, however (see Ref. 11).
Our laboratory has been investigating the regulation of Na ϩ / Ca 2ϩ exchange activity in stably transfected Chinese hamster ovary (CHO) cells expressing the bovine cardiac Na ϩ /Ca 2ϩ exchanger. The results of our studies suggest that the above processes are of limited importance in modulating exchange activity under physiological conditions. Thus, Na ϩ /Ca 2ϩ exchange activity in these cells is nearly fully activated by cytosolic Ca 2ϩ under resting conditions (11), and it is not affected by variations in phosphatidylinositol 4,5-bisphosphate that are likely to occur physiologically (12) or by activation or inhibition of protein kinases (10). Although exchange activity is inhibited by agents that disturb cytoskeletal integrity in these cells (10,13), the relevance of these findings to more physiological conditions is unclear.
In this report, we describe a two-pulse protocol in which Ca 2ϩ influx during a "conditioning" interval in a low [Na ϩ ] o medium leads to inhibition of exchange activity during a subsequent "test" interval. The data show that mitochondria accumulate Ca 2ϩ during the conditioning interval of Ca 2ϩ influx and that this process is essential for the subsequent inhibition of exchange activity. These findings reveal a previously unsuspected link between mitochondria and the regulation of Na ϩ / Ca 2ϩ exchange activity. 0.02% (w/v) pluronic F-127. Fura-2, sulfinpyrazone, and pluronic F-127 were added as 1000-fold concentrated stock solutions in dimethyl sulfoxide. The coverslips were then rinsed in Na-PSS and mounted in an open stainless steel viewing chamber (Molecular Probes, Inc.). The chamber was then inserted into a thermostatically regulated (37°C) PDMI-2 open perfusion micro-incubator (Medical Systems Corp., Greenvale, NY). Fura-2 fluorescence was measured at 520 nm with alternate excitation at 334 and 380 nm using an Attofluor Ratiovison digital fluorescence imaging system coupled to a Zeiss Axiovert 100 fluorescence microscope. For measurement of Ba 2ϩ influx, a second filter set was employed for alternate excitation at 350 and 390 nm (15). To initiate the Ca 2ϩ influx mode of Na ϩ /Ca 2ϩ exchange, Na-PSS was replaced with 20:120 Na/K-PSS for a period of 5 min. The medium was then replaced with Na-PSS for an additional 5 min, followed by a second interval (3 min) in 20:120 Na/K-PSS. Fluorescence signals were calibrated for [Ca 2ϩ ] i by the procedure of Grynkiewicz et al. (16). Briefly, 10 M ionomycin, 1 M thapsigargin, and 50 M EGTA/AM were added to fura-2-loaded cells in Ca 2ϩ -free Na-PSS containing 0.6 mM EGTA; R min was recorded after at least 5 min. CaCl 2 (10 mM) was subsequently added to determine R max . Significance testing was done using Student's t test.
Rhod-2 Fluorescence Measurements-Cells were loaded in the dark with 2 M rhod-2/AM plus 0.02% (w/v) pluronic F-127 in Na-PSS plus 1% bovine serum albumin for 10 -12 min at room temperature. Fluorescence images were acquired using a cooled CCD camera under computer control. The appropriate dichroic reflector and emission filter were selected. Excitation at 548 nm was used with a special long-pass fura-2/rhod-2 emission filter set designed by Chroma Technology Corp. (Brattleboro, VT). The mitochondrion-rich perinuclear regions of the cells were sampled for time-dependent changes in fluorescence during Ca 2ϩ uptake by Na ϩ /Ca 2ϩ exchange.
Confocal Microscopy-Cells were labeled with 100 nM MitoTracker Green FM (Molecular Probes, Inc.) and 0.02% (w/v) pluronic acid in Na-PSS plus 1% bovine serum albumin for 40 min at room temperature, followed by a rhod-2/AM loading process as described above. The coverslips were rinsed six times to remove excess dye. Cells were then imaged with a Nikon two-laser (argon and helium-neon) PCM 2000 laser scanning confocal microscope using a Nikon Eclipse TE300 inverted microscope and 60X Plan Apo (1.20 WI). Fluorescence images were collected simultaneously with the aid of C-Imaging Systems Sim-plePCI software (Compix Inc.) using the appropriate rhodamine or fluorescein isothiocyanate filters.

RESULTS
Two-pulse Protocol for Na ϩ /Ca 2ϩ Exchange-In cells expressing Na ϩ /Ca 2ϩ exchange activity, a reduction in [Na ϩ ] o leads to an increase in [Ca 2ϩ ] i due to Na ϩ i -dependent Ca 2ϩ influx. For the data shown in Fig. 1 A conditioning pulse of 1 min (trace a) had no effect on the Ca 2ϩ influx during the test pulse. After a 3-min conditioning pulse (trace b), the peak increase in the fura-2 ratio during the test pulse was reduced to 54 Ϯ 6% of the conditioning peak. For 5and 7-min conditioning pulses (traces c and d, respectively), the test peaks were 37 Ϯ 7 and 41 Ϯ 7% of the conditioning peak, respectively. Thus, a conditioning interval of at least 3 min was required to inhibit the rise in [Ca 2ϩ ] i during the test pulse. Note that for each time interval, the peak [Ca 2ϩ ] i attained during the test interval was nearly identical to the [Ca 2ϩ ] i at the end of the declining phase of the conditioning pulse. In other experiments (data not shown), we varied the duration of the recovery interval between the conditioning and test pulses. Exchange activity remained inhibited following a recovery period of 10 min, but after 30 min of recovery, the rise in [Ca 2ϩ ] i during the test interval returned to normal.
As shown in Fig. 3 (boldface trace), when Ca 2ϩ was omitted from the medium during the conditioning interval in 20 mM Na ϩ o , the rise in [Ca 2ϩ ] i during the subsequent test interval was not inhibited. The slight rise in [Ca 2ϩ ] i during the conditioning interval probably reflected small amounts of residual Ca 2ϩ in the medium. We conclude that Ca 2ϩ influx during the conditioning interval was required to inhibit the rise in [Ca 2ϩ ] i during the test interval and that the inhibition was not simply a consequence of lowering [Na ϩ ] o to 20 mM.
Inhibition of Na ϩ /Ca 2ϩ Exchange Activity-The decrease in the peak [Ca 2ϩ ] i during the test interval could be due to either a reduced influx of Ca 2ϩ , i.e. an inhibition of Na ϩ /Ca 2ϩ exchange activity, or an increase in the sequestration of cytosolic Ca 2ϩ by intracellular organelles. The data in Fig. 4 demonstrate that the first alternative is correct, i.e. exchange activity is inhibited during the test pulse. For this experiment, we used Ba 2ϩ as a Ca 2ϩ surrogate during the test pulse. Ba 2ϩ is transported by the Na ϩ /Ca 2ϩ exchanger, but it is not sequestered by Na/K-PSS) was applied to the cells for intervals of 1, 3, 5, and 7 min (traces a-d, respectively), followed by a 5-min application of Na-PSS and a subsequent interval in 20:120 Na/K-PSS. The ordinate scale (⌬R ϭ 1.0) refers to a change in the 334:380 nm fura-2 ratio of 1 unit; traces have been displaced vertically. Results are the means Ϯ S.E. (for every fourth data point) of six to seven coverslips. B, shown is the peak ratio during the test interval as a percentage of the peak ratio during the conditioning interval for the data in A. The peak ratios at 3, 5, and 7 min were significantly different from the peak ratio at 1 min (p Ͻ 0.001), but were not significantly different from each other (p Ͼ 0.09).
the endoplasmic reticulum and is only poorly, if at all, taken up by mitochondria in these cells (17). As shown in Fig. 4 (inset), Ba 2ϩ entered cells rapidly when [Na ϩ ] o was reduced to 20 mM (arrow) in the presence of 2 mM Ba 2ϩ o ; this Ba 2ϩ concentration was chosen to produce a rise in the fura-2 ratio that was roughly equivalent to that seen with 1 mM Ca 2ϩ . For o was applied to the cells. This was followed by a 5-min recovery period in normal Na-PSS before [Na ϩ ] o was again reduced to 20 mM, this time in the presence of 2 mM Ba 2ϩ . As shown, Ba 2ϩ influx during the test pulse was nearly completely blocked. We conclude that the reduced rise in [Ca 2ϩ ] i during the test pulse for the experiments shown in Figs. 1 and 2 was due to an inhibition of Na ϩ /Ca 2ϩ exchange activity rather than increased Ca 2ϩ sequestration.
Mitochondrial Ca 2ϩ Accumulation during the Ca 2ϩ Influx Mode of Na ϩ /Ca 2ϩ Exchange-We have shown previously that in transfected cells that had been loaded with high concentrations of cytosolic Na ϩ using gramicidin (a channel-forming antibiotic) or ouabain (a Na ϩ /K ϩ -ATPase inhibitor), reducing [Na ϩ ] o led to large increases in [Ca 2ϩ ] i and an extensive accumulation of Ca 2ϩ in the mitochondrial compartment (18). To examine whether mitochondria also accumulated Ca 2ϩ in the absence of cytosolic Na ϩ loading, we labeled cells with the Ca 2ϩ -sensitive dye rhod-2. This indicator preferentially accumulates within mitochondria because the positively charged, unhydrolyzed form of the dye equilibrates with the negative mitochondrial membrane potential during dye loading (19). Fig. 5A shows rhod-2-labeled cells following exposure to 20 mM Na ϩ o in the presence of 1 mM Ca 2ϩ o . Rhod-2 fluorescence was intense in the perinuclear region of the cell and showed a particulate pattern suggestive of mitochondria. Prior to reducing [Na ϩ ] o , rhod-2 fluorescence was much lower in intensity and more generally distributed throughout the cell (data not shown, but see below). To demonstrate that the increased rhod-2 fluorescence was in fact associated with mitochondria, the cells in this experiment were also labeled with the selective mitochondrial Ca 2ϩ indicator MitoTracker Green (Fig. 5B). Fig.  5C is the superposition of the two images in A and B. The yellow color denotes areas where rhod-2 and MitoTracker Green coincide and demonstrates a nearly complete correspondence between the rhod-2 and MitoTracker labeling. The results confirm the conclusion of Rutter et al. (20) that rhod-2 preferentially labels mitochondria in CHO cells and further demonstrate that the mitochondria accumulate Ca 2ϩ under low [Na ϩ ] o conditions in the transfected cells. No mitochondrial Ca 2ϩ accumulation was detected when the low [Na ϩ ] o medium was applied to vector-transfected CHO cells, which do not exhibit Na ϩ /Ca 2ϩ exchange activity (data not shown).
For Fig. 6, the standard two-pulse protocol was applied to rhod-2-labeled cells, and fluorescence was monitored in mitochondrion-rich areas of the cells (see "Experimental Procedures"). Rhod-2 fluorescence increased ϳ5-fold when [Na ϩ ] o was reduced to 20 mM in the presence of 1 mM Ca 2ϩ o . Restoration of 140 mM Na ϩ o evoked a fall in fluorescence intensity back to initial levels. The changes in rhod-2 fluorescence were much slower than the corresponding changes in the fura-2 signal (compare with Fig. 1), indicating that the rhod-2 measurements did not simply reflect changes in [Ca 2ϩ ] i . The slow decline in rhod-2 fluorescence upon restoration of 140 mM Na ϩ is consistent with the relatively slow rate of Ca 2ϩ efflux from mitochondria (19). Upon re-exposure to 20 mM Na ϩ o , rhod-2 fluorescence did not increase, consistent with a reduced Ca 2ϩ influx during the test pulse. When Ca 2ϩ influx was stimulated with ionomycin (2 M) during the test pulse, there was a sharp increase in mitochondrial rhod-2 fluorescence (data not shown), indicating that the mitochondria were still responsive to an increase in [Ca 2ϩ ] i . From these results, taken together with the Ba 2ϩ influx measurements (Fig. 4), we conclude that the reduced rise in [Ca 2ϩ ] i during the test interval was due to an inhibition of Na ϩ /Ca 2ϩ exchange activity and could not be attributed to an increase in organellar Ca 2ϩ sequestration.
Inhibition of Exchange Activity Is Dependent upon Mitochondrial Ca 2ϩ Accumulation-The results presented below indicate that the accumulation of Ca 2ϩ by mitochondria during the conditioning pulse is an essential factor in mediating the subsequent inhibition of exchange activity during the test pulse. Fig. 7A shows that the inhibition of Ca 2ϩ influx during the test pulse was eliminated when the mitochondrial uncoupler carbonyl cyanide m-chlorophenylhydrazone (CCCP; 2 M) and the F 0 F 1 -ATPase inhibitor oligomycin (3.2 M) were both present during the two-pulse protocol. (Oligomycin was added to reduce ATP hydrolysis by the F 0 F 1 -ATPase; similar results were obtained in the presence of uncoupler alone.) In the presence of CCCP (Fig. 7A, C/O trace), the peak ratio during the conditioning pulse was similar to that for untreated controls, but the subsequent decline in [Ca 2ϩ ] i following the peak of the conditioning pulse was eliminated. Following addition of 140 mM Na ϩ o , [Ca 2ϩ ] i fell to values that were somewhat higher than for untreated controls, perhaps reflecting the loss of mitochondrial Ca 2ϩ -sequestering activity. The peak increase in the fura-2 ratio during the test interval was 110 Ϯ 9% of that during the conditioning interval for the cells treated with CCCP/oligomycin and was much greater than that for the untreated cells. Essentially identical results were obtained when the K ϩ ionophore valinomycin (5 M) was used in place of CCCP to depolarize the mitochondrial membrane potential (data not shown). The absence of exchange inhibition in the presence of the uncouplers was not due to a reduction in mitochondrially produced ATP since the normal inhibition of Ca 2ϩ influx during the test pulse was not affected by oligomycin alone (data not shown).
The data in Fig. 7B show a similar experiment carried out in the presence of CCCP/oligomycin, but in this case, 2 mM Ba 2ϩ was substituted for Ca 2ϩ during the test interval (arrow). In contrast to the results for untreated cells shown previously (Fig. 4), a robust influx of Ba 2ϩ was observed in the inhibitortreated cells. The data in Fig. 7C (C/O trace) depict rhod-2 fluorescence in the presence of CCCP and oligomycin. The initial value of rhod-2 fluorescence appeared to be higher than that for the untreated cells, but this difference was not significant (p Ͼ 0.2) and reflected the large variations among different coverslips in cellular rhod-2 fluorescence. Presumably, this variability primarily reflects differences in the extent of dye loading among different coverslips. As expected, the inhibitors completely blocked mitochondrial Ca 2ϩ accumulation during the conditioning pulse. Oligomycin alone did not block mitochondrial Ca 2ϩ accumulation (data not shown). These data confirm that the rhod-2 fluorescence measurements reflect mitochondrial rather than cytosolic [Ca 2ϩ ] since the rise in [Ca 2ϩ ] i was unaffected by CCCP. The control traces represent rhod-2 fluorescence for cells from the same batch of coverslips in the absence of inhibitors (n ϭ 4).
The mitochondrial Ca 2ϩ content is regulated by the balance between Ca 2ϩ uptake and efflux mechanisms. Previous results with permeabilized CHO cells demonstrated that the mitochondria in these cells possess a Na ϩ /Ca 2ϩ exchange mechanism for Ca 2ϩ efflux (18). The mitochondrial Na ϩ /Ca 2ϩ exchanger is different from the plasma membrane exchanger and is inhibited by agents such as diltiazem, clonazapam, and CGP-37157 that do not affect the plasma membrane exchanger (21,22). When the two-pulse protocol was carried out in the presence of diltiazem (100 M), we observed a more pronounced inhibition of Ca 2ϩ influx during the test pulse than for control cells (Fig.  8A). Diltiazem had no significant effect on Ca 2ϩ influx during the conditioning pulse. Changes in fluorescence of rhod-2-labeled cells in the presence of diltiazem are shown in Fig. 8B. Both the initial level and the peak values of rhod-2 fluorescence during the conditioning interval were higher for diltiazem-

FIG. 5. Images of rhod-2-and Mito-Tracker Green-labeled cells.
Cells were labeled with MitoTracker Green for 30 min and with rhod-2 for 12 min and then exposed to 20:120 Na/K-PSS. Images of an optical section using filter settings for rhodamine and fluorescein were obtained simultaneously with a Nikon PCM 2000 laser scanning confocal microscope. The image labeled Composite is a superposition of the two individual wavelength images. treated cells than for control cells (p ϭ 0.05 and 0.005, respectively). Whether this reflects coverslip variability or a true increase in mitochondrial Ca 2ϩ content with diltiazem is difficult to assess with certainty. More important, the results clearly demonstrate that the efflux of Ca 2ϩ from mitochondria was reduced when 140 mM Na ϩ was restored in the presence of diltiazem compared with untreated controls. Mitochondrial [Ca 2ϩ ] remained elevated in diltiazem-treated cells throughout the recovery period in 140 mM Na ϩ o (Fig. 8), in contrast to untreated cells (Fig. 8, Control trace; and Fig. 6), where mitochondrial [Ca 2ϩ ] decreased to initial values. These results are consistent with inhibition of the mitochondrial Na ϩ /Ca 2ϩ exchanger by diltiazem. We attempted to use the more selective mitochondrial exchange inhibitor CGP-37157 for these studies, but observed no effect either on [Ca 2ϩ ] i during the two-pulse protocol or on mitochondrial Ca 2ϩ uptake or efflux (data not shown). The lack of effect of CGP-37157 in these cells might be due to poor cellular permeability or intracellular binding. In permeabilized cells, we found that this compound was highly effective in blocking Na ϩ -dependent mitochondrial Ca 2ϩ efflux (18).
An additional correlation between mitochondrial Ca 2ϩ accumulation and inhibition of exchange activity was provided, surprisingly, by nocodazole, an agent that depolymerizes microtubules. Fig. 9A compares the results of the two-pulse protocol for control cells and for cells that had been pretreated with 17 M nocodazole for 15-20 min prior to beginning the experiment; nocodazole was also included in the medium during the two-pulse protocol. Nocodazole had no effect on the peak [Ca 2ϩ ] i and slightly reduced the post-peak decline in [Ca 2ϩ ] i during the conditioning pulse. However, nocodazole practically eliminated the inhibition of Ca 2ϩ influx during the test pulse; the peak [Ca 2ϩ ] i during the test interval was 84 Ϯ 6% of the conditioning peak. Fig. 9B shows that the fluorescence intensity of rhod-2 did not increase when the cells were exposed to 20 mM Na ϩ o . Since the rise in [Ca 2ϩ ] i was only slightly affected (Fig. 9A), we conclude that the nocodazole treatment blocked the accumulation of Ca 2ϩ by mitochondria during the conditioning interval. As a positive control for these data, we found that the addition of ionomycin to the nocodazole-treated cells during the conditioning interval produced the expected rise in rhod-2 fluorescence (data not shown). This indicates that the nocodazole treatment did not interfere with rhod-2 localization or its ability to respond to increased [Ca 2ϩ ].
In summary, the results demonstrate that the inhibition of exchange activity during the test interval was dependent upon mitochondrial Ca 2ϩ accumulation during the conditioning interval. Agents that blocked mitochondrial Ca 2ϩ accumulation (CCCP, nocodazole) also blocked the inhibition of exchange activity, whereas an agent that blocked mitochondrial Ca 2ϩ efflux (diltiazem) enhanced the inhibition of exchange activity. DISCUSSION Phosphatidylinositol 4,5-bisphosphate, cytosolic Ca 2ϩ , protein phosphorylation, and the actin cytoskeleton have each been suggested as regulators of the cardiac Na ϩ /Ca 2ϩ exchanger (see the Introduction). The evidence supporting their involvement stemmed from studies that either utilized subcellular preparations, i.e. excised membrane patches, or involved the addition of exogenous inhibitors or other agents. Thus far, there has been no clear demonstration that any of these processes are important regulators of exchange activity under physiological conditions. The experimental design in the present studies involved a simple two-pulse protocol for activating exchange activity and did not require the addition of exogenous agents or the experimental alteration of normal cytosolic concentrations of Na ϩ or Ca 2ϩ . Our results show that Na ϩ /Ca 2ϩ exchange activity in transfected CHO cells is regulated by a previously unsuspected process linked to mitochondrial Ca 2ϩ accumulation. These findings provide the first evidence for a distinct regulatory pathway that modulates Na ϩ /Ca 2ϩ exchange activity in intact cells.
The central finding of these studies is that a conditioning interval of Ca 2ϩ influx by Na ϩ /Ca 2ϩ exchange produced an inhibition of exchange activity during a subsequent test interval ( Figs. 1 and 2). Ba 2ϩ influx was also inhibited during the test interval (Fig. 4). Since Ba 2ϩ is not efficiently transported by organellar or plasma membrane Ca 2ϩ -ATPases (17), we concluded that the reduced rise in [Ca 2ϩ ] i during the test interval reflected an inhibition of exchange activity itself rather than an increased sequestration or efflux of Ca 2ϩ . This conclusion is consistent with the absence of mitochondrial Ca 2ϩ accumulation during the test interval (Fig. 6).
The inhibition of exchange activity was eliminated (or potentiated) by agents that prevented (or promoted) mitochondrial Ca 2ϩ accumulation. Thus, CCCP (Fig. 7) and nocodazole ( Fig.  9) each blocked mitochondrial Ca 2ϩ accumulation during the conditioning interval and eliminated the inhibition of exchange activity during the test interval. On the other hand, diltiazem, a blocker of the mitochondrial Na ϩ /Ca 2ϩ exchanger (21,22), inhibited mitochondrial Ca 2ϩ efflux during the recovery interval and enhanced the inhibition of exchange activity during the test interval (Fig. 8). These data suggest that the increase in the mitochondrial Ca 2ϩ content led to the development of an inhibitory signal that reduced Na ϩ /Ca 2ϩ exchange activity.
It seems unlikely that these effects were simply due to a reduction in [Na ϩ ] i resulting from the exposure to 20 mM Na ϩ o during the conditioning pulse. We were unable to detect changes in [Na ϩ ] i during the two-pulse protocol in cells loaded with the Na ϩ indicator sodium-binding benzofuran iso- 2؉ efflux (B). A, the standard two-pulse protocol was applied to cells in the presence or absence of 100 M diltiazem in all solutions (n ϭ 6). B, shown is the rhod-2 fluorescence of cells in the presence of 100 M diltiazem (n ϭ 4). 20:120 Na/K-PSS was applied at times between 60 and 360 s; Na-PSS was present at other times. Control traces are identical to those shown in Fig. 7. a.u., arbitrary units. phthalate (data not shown). However, it is possible that this method is not sufficiently sensitive to detect small reductions in [Na ϩ ] i that could inhibit exchange activity. Reducing [Na ϩ ] o to 20 mM in the absence of Ca 2ϩ o did not reduce exchange activity during a subsequent test pulse (Fig. 3), indicating that Ca 2ϩ entry was required to inhibit exchange activity. Finally, nocodazole, CCCP, and diltiazem would not be expected to alter [Na ϩ ] i in ways that would be compatible with the results obtained. Thus, a trivial explanation in terms of reduced [Na ϩ ] i seems unlikely, although we cannot rule out a more complex mechanism in which alterations in [Na ϩ ] i are somehow linked to mitochondrial Ca 2ϩ accumulation.

FIG. 8. Diltiazem increases inhibition of Ca 2؉ influx during test interval (A) and inhibits mitochondrial Ca
The nature of the inhibitory signal generated during mitochondrial Ca 2ϩ accumulation is unknown. Since an increase in Ca 2ϩ within the mitochondrial matrix activates several different dehydrogenases (19,23), it is possible that increased production of an end product of mitochondrial respiration provides a signal for inhibiting exchange activity. However, the mitochondrial production of ATP seems unimportant since oligomycin, an inhibitor of the F 0 F 1 -ATPase, did not block inhibition of exchange activity. Mitochondrial production of reactive oxygen species (24), nitric oxide (25,26), and glutamate (27) has been invoked as regulatory effectors in some cells, and these possibilities deserve further investigation.
The gradual development of exchange inhibition over a 1-5min period (Fig. 2) suggests that time is required to generate the signal that inhibits exchange activity. This possibility is consistent with the time course of the changes in [Ca 2ϩ ] i that occurred during the initial conditioning interval in low [Na ϩ ] o . When 20 mM Na ϩ o was first applied to the cells, [Ca 2ϩ ] i increased to a peak value and subsequently declined. During the following test interval, the maximal rise in [Ca 2ϩ ] i was approximately equal to the final value attained during the declining phase of the conditioning interval ( Figs. 1 and 2). In the presence of CCCP ϩ oligomycin, the post-peak decline in [Ca 2ϩ ] i during the conditioning interval was absent, suggesting that the decline reflected the developing inhibition of exchange activity. For diltiazem-treated cells, the decline in [Ca 2ϩ ] i after the peak was similar to that for control cells (Fig. 8B), indicating that exchange inhibition did not develop more rapidly, despite a possible increase in mitochondrial Ca 2ϩ content. However, the inhibition of exchange activity during the test pulse was more pronounced than for the control cells (Fig. 8A). We suggest that this reflects the continued development of the inhibition process due to the retention of high levels of mitochondrial Ca 2ϩ during the recovery period in diltiazem-treated cells (Fig. 8B).
The results with nocodazole are complex and incompletely understood. The data suggest that microtubular structure exerts an important influence on mitochondrial function. Mitochondria are attached to microtubules and utilize these structures for intracellular movement (28 -30). Following treatment with nocodazole, the mitochondria stained intensely with the potential-sensitive dye tetramethylrhodamine ethyl ester (data not shown), indicating that the reduced Ca 2ϩ accumulation was not a consequence of a reduced membrane potential. Perhaps microtubular disruption altered the location or the local environment of the mitochondria in a way that interfered with their ability to accumulate Ca 2ϩ during the conditioning interval. This might reflect an alteration in the local [Ca 2ϩ ] i in the vicinity of the mitochondria or a regulatory alteration in the activity of the mitochondrial Ca 2ϩ uptake mechanism. This issue is currently under investigation.
What are the physiological implications of the link between Na ϩ /Ca 2ϩ exchange activity and mitochondrial Ca 2ϩ accumulation? In cardiac myocytes, the mitochondrial Ca 2ϩ concentration increases markedly with contractile frequency, particularly in the presence of ␤-adrenergic stimulation (31)(32)(33). An associated inhibition of Na ϩ /Ca 2ϩ exchange activity might therefore make a significant contribution to the positive forcefrequency relationship observed in many species. Under pathological conditions, the mitochondrial exchanger interaction could in some circumstances form a deleterious positive feedback system that contributes to Ca 2ϩ overload. Alternatively, in cells with elevated [Na ϩ ] i , Ca 2ϩ accumulation in the mitochondria might act to limit further Ca 2ϩ entry by inhibiting the Ca 2ϩ influx mode of the exchanger. These considerations are entirely speculative at present, and additional experimentation is clearly needed to assess the physiological importance of this novel regulatory pathway.