Privileged Access to Mitochondria of Calcium Influx throughN-Methyl-d-Aspartate Receptors

Results and Discussion

Concurrent imaging of cytosolic and mitochondrial Ca²⁺with calcium green and rhod-2.

Most of the published studies on the role of mitochondria in buffering cytosolic Ca²⁺ loads have employed a single fluorescent Ca²⁺ indicator, such as Fura-2 or Indo-1, to monitor changes in intracellular Ca²⁺concentration ([Ca²⁺]_i). These Ca²⁺ indicators can easily get into the cell and distribute not only in the cytosol, but also into subcellular organelles, depending on loading temperature and loading time. Changes in the fluorescence signal from these Ca²⁺indicators in response to Ca²⁺ increases reflect changes in both the cytosol and the organelles into which the indicators have partitioned. In this study, to reliably, independently, and simultaneously detect cytosolic and mitochondrial Ca²⁺ concentration changes in the same cell, we used two fluorescent Ca²⁺ indicators, calcium green and rhod-2. Unlike most fluorescent Ca²⁺indicators that carry negative charges, rhod-2/AM (cell membrane-permeable form) contains one net positive charge. This unique chemical feature allows it to accumulate selectively into the highly negatively charged mitochondrial matrix. Once inside the matrix, mitochondrial esterases cleave the AM ester to liberate rhod-2 (membrane-impermeable free acid form). However, when loaded at warm temperature (37°), cytosolic esterases are active enough that the AM esters are cleaved before rhod-2/AM can enter mitochondria. This can lead to substantial cytosolic loading (unpublished observations). At cooler loading temperatures, when enzymatic activity is slowed, the fluorophore esters can reach the mitochondria before being hydrolyzed. We loaded rhod-2/AM into striatal neurons at room temperature to favor mitochondrial loading, and used a low concentration of the indicator (1–2 μm), and a short loading time (5–10 min) to minimize cytosolic loading. With this loading technique, we consistently obtain strong mitochondrial rhod-2 signal and weak cytosolic rhod-2 signal (Fig. 1A). We have previously published a more detailed account of the characteristics and selectivity of rhod-2 in monitoring mitochondrial Ca²⁺ concentration changes (Peng et al. 1998).

Calcium green loading was performed after rhod-2 loading had completed. The loading condition for calcium green is the reverse of that for rhod-2, i.e., warm loading temperature (37°), high concentration (5 μm), and long loading time (40–50 min). Calcium green signal distributes throughout the cell when loaded alone. However, its signal is weaker in mitochondrial areas when loaded after rhod-2 (Fig.1b). The reasons for this decreased signal in mitochondrial areas are unknown, but unimportant for the present experiments. Instead of detecting the fluorescence signals nonspecifically from the entire cell, as would occur with a microfluorimeter or photomultiplier tube, we imaged the cell with confocal microscopy and measured the calcium green fluorescence intensity in predefined cytosolic areas and the rhod-2 fluorescence intensity in predefined mitochondrial areas. This method avoids unwanted crossover signals between calcium green and rhod-2 in mitochondrial and cytosolic areas, and increases the signal-to-noise ratio when independently monitoring cytosolic and mitochondrial Ca²⁺ concentration changes. Fluorescence images were obtained with brief exposure times (0.15–0.40 sec) at 2-sec intervals to avoid photobleaching and phototoxicity while allowing adequate temporal resolution. In this report, fluorescence intensity changes were not converted into Ca²⁺concentrations owing to the controversy in using high affinity Ca²⁺ indicators in estimating [Ca²⁺]_i. For example,Hyrc et al. (1997) suggested that high affinity Ca²⁺ indicators may underestimate [Ca²⁺]_i, even at levels that seem to be within their range of detection.

Source specificity for fast mitochondrial Ca²⁺uptake.

Previous studies on glutamate- (or NMDA-) induced Ca²⁺ load buffering by mitochondria have focused on changes in the magnitude of the peak [Ca²⁺]_i transient and the recovery time to base-line [Ca²⁺]_i levels that result from mitochondrial Ca²⁺ uptake inhibition (White and Reynold, 1995, 1997; Wang and Thayer, 1996; Khodorovet al., 1996). That there is a much greater Ca²⁺ load induced by glutamate compared with KCl-induced membrane depolarization was suggested by the observation that inhibition of mitochondrial Ca²⁺ uptake markedly potentiated the peak [Ca²⁺]_i and prolonged the recovery time after a glutamate challenge but not after depolarization. In these experiments, the effects of mitochondrial Ca²⁺ uptake inhibition were inferred from changes in [Ca²⁺]_i measured with a single fluorescence indicator, which primarily reflected cytosolic Ca²⁺ concentration ([Ca²⁺]_c) changes. With this method, [Ca²⁺]_mchanges could not be observed directly.

Using our method to directly, independently, and simultaneously detect mitochondrial and cytosolic Ca²⁺ changes in the same cell, we were interested in studying 1) the time courses of cytosolic and mitochondrial Ca²⁺ transients, and 2) whether mitochondrial Ca²⁺ uptake elicited by NMDA receptor activation is different from that induced by other routes of Ca²⁺ entry. Changes of [Ca²⁺]_c and [Ca²⁺]_m, as reflected by changes in calcium green and rhod-2 fluorescence from single striatal neurons treated with 100 μm NMDA, 100 μmkainate, or 50 mm KCl are shown in Fig.2. (Most non-NMDA glutamate receptors have limited Ca²⁺ permeability and we did not perform ion substitution experiments. Therefore, we cannot make firm conclusions as to what proportion of the Ca²⁺entry elicited by kainate in these experiments was because of secondary activation of voltage-dependent Ca²⁺ channels.) The calcium green and rhod-2 fluorescence intensity changes are overlaid to compare the time courses of the [Ca²⁺]_c and [Ca²⁺]_m responses. In the initial rising phase after all three stimuli, both [Ca²⁺]_m and [Ca²⁺]_c reached peak levels in less than 30 sec and remained at those levels in the continued presence of the stimulus. We intentionally scaled the peakF/F₀ values of calcium green and rhod-2 to the same height for better comparison of the time courses. This does not indicate that [Ca²⁺]_m and [Ca²⁺]_c are at the same absolute level. In fact, after stimulating chromaffin cells co-loaded with calcium green and rhod-2 with electrical pulses, Babcock et al. (1997) estimated that the peak levels of [Ca²⁺]_m and [Ca²⁺]_c were different.

In Fig. 2, the [Ca²⁺]_crecovery time for the NMDA-induced response is longer than those for kainate or KCl, an observation in accordance with what has already been reported by others (White and Reynolds, 1995; Wang and Thayer, 1996).White and Reynolds (1997) noted that the [Ca²⁺]_i recovery time is correlated with the magnitude of the Ca²⁺ load induced by glutamate, and that the recovery of [Ca²⁺]_i to base-line after glutamate removal is critically regulated by release of Ca²⁺ from mitochondrial stores. Hence, in the present experiments, the longer [Ca²⁺]_c recovery time after NMDA application most likely indicates a larger Ca²⁺ load imposed on mitochondria. Interestingly, [Ca²⁺]_m recovery followed a similar but more prolonged time course than [Ca²⁺]_c in all cases. This more protracted [Ca²⁺]_m recovery was also noted by Babcock et al. (1997) after electrical depolarization of chromaffin cells. The mechanism and significance of this phenomenon await further investigation.

Although it seems that NMDA receptor activation imposes a larger Ca²⁺ load on mitochondria than other stimuli, the question remains as to whether mitochondria take up Ca²⁺ more efficiently in response to NMDA stimulation relative to other Ca²⁺ influx mechanisms. To address this issue we focused more closely on the initial rising phase of the Ca²⁺ response elicited by the various agonists. Fig. 3shows the average changes in cytosolic and mitochondrial fluorescence levels during the first 16 sec after application of 100 μm NMDA, 100 μm kainate, 50 mmKCl, or 0.3 μm ionomycin. Because the maximalF/F₀ values of calcium green and rhod-2 varied from cell to cell across experiments (White and Reynolds, 1997; Peng et al.; 1998), fluorescence was normalized, expressed as a percent of the maximal increase, and fitted to exponential equations. The time needed to reach 50% of the maximal increase in fluorescence intensity (t_1/2) was used to compare the rate of increase in fluorescence intensity across different experiments.

As shown in Figs. 3B and 4, during stimulation with NMDA, the t_1/2 values of cytosolic and mitochondrial Ca²⁺ accumulation were virtually identical (2.71 ± 0.09 versus 2.89 ± 0.17 sec; n = 4; not significant). These data suggest that mitochondrial and cytosolic Ca²⁺ levels are time-locked in neurons treated with NMDA. In contrast, mitochondrial Ca²⁺ uptake lagged about 3 sec behind cytosolic Ca²⁺ after kainate stimulation (2.93 ± 0.11 sec versus 6.15 ± 0.25 sec; n = 4;p < 0.0001) or KCl stimulation (2.86 ± 0.15 sec versus 5.56 ± 0.29 sec; n = 4; p< 0.0001) (Figs. 3 and 4). The apparently slower mitochondrial Ca²⁺ uptake responses in the kainate and KCl treatment groups are probably not the result of a slower cytosolic Ca²⁺ increase, because 1) the calcium greent_1/2 values for NMDA, kainate, and KCl were very similar and did not differ statistically from one another (Figs. 3A and4), and 2) analysis of variance of the mitochondrialt_1/2 values showed that the mitochondrial response to NMDA was significantly faster than with other stimuli (p < 0.001). Note that we did not use lower concentrations of NMDA; therefore, we do not know whether there would be a lag in mitochondrial Ca²⁺ uptake with less intense receptor stimulation. On the other hand, 100 μmis not a saturating concentration of NMDA in terms of receptor activation or neurotoxicity (Verdoorn and Dingledine, 1988; McNamara and Dingledine, 1990), so it is unlikely that the fast mitochondrial Ca²⁺ uptake we observed is because of saturation of the NMDA receptors.

Is fast mitochondrial Ca²⁺ uptake specific to NMDA receptor stimulation? To further address this question, we treated neurons with another type of stimulus, the Ca²⁺ionophore ionomycin, and determined the kinetics of the mitochondrial Ca²⁺ uptake response. A low dose of ionomycin (0.3 μm) was used in an effort to let Ca²⁺ penetrate the plasma membrane selectively. This ionophore caused a somewhat slower increase in cytosolic Ca²⁺ than the other stimuli (4.27 ± 0.23 sec; n = 4), but this did not reach statistical significance. Unlike the NMDA receptor stimulation, the mitochondrial Ca²⁺ uptake response to ionomycin lagged substantially behind the cytosolic response (10.70 ± 0.52 sec;p < 0.0001) Together, these data clearly demonstrate that mitochondrial Ca²⁺ uptake induced by NMDA receptor activation is different from that induced by a non-NMDA agonist, voltage-dependent Ca²⁺ channel activation, or a Ca²⁺ ionophore.

Fast mitochondrial Ca²⁺ uptake in response to NMDA receptor activation may have both adaptive and detrimental consequences. Under physiological conditions, rapid mitochondrial Ca²⁺ uptake may act to stimulate mitochondrial respiration and energy production, and it is well documented that fast mitochondrial Ca²⁺ uptake occurs in other systems. Babcock et al. (1997) reported that chromaffin cell mitochondria can take up Ca²⁺ rapidly; more than half of mitochondrial Ca²⁺ uptake is accomplished while cytosolic Ca²⁺ entry elicited by electrical depolarization is still in progress. Trollinger et al.(1997) observed rapid mitochondrial free Ca²⁺transients during each contractile cycle in rhod-2-loaded rabbit cardiac myocytes. The averaged transient showed a half rise time of about 65 msec with a total of 300 msec to rise from base-line to peak fluorescence. These observations demonstrate that mitochondria can rapidly take up physiological concentrations of Ca²⁺. Moreover, fast mitochondrial Ca²⁺ uptake probably helps to prevent a large rise in cytosolic Ca²⁺ after NMDA receptor stimulation, and it helps to shape the [Ca²⁺]_i response. However, this accelerated mitochondrial Ca²⁺uptake induced by NMDA may permit excessive mitochondrial Ca²⁺ accumulation and render mitochondria more susceptible to Ca²⁺-mediated injury. Thus, what is probably a protective mechanism may become a toxic mechanism during excessive or prolonged NMDA receptor activation.

Potential mechanisms.

Sparagna et al. (1995)demonstrated that Ca²⁺ can be sequestered into mitochondria by two distinct modes, a rapid uptake (high conductivity) mode of very short duration, followed by a slower uptake (low conductivity) mode, which has the characteristics of the uniporter. Could this rapid mode of mitochondrial Ca²⁺uptake be responsible for the NMDA-induced fast mitochondrial Ca²⁺ uptake observed in this report? If so, why is this rapid uptake mode activated selectively by NMDA but not by kainate or membrane depolarization? Polyamines may provide the answer.Sparagna et al. (1995) showed that the rapid mode of mitochondrial Ca²⁺ uptake can be activated specifically by polyamines, especially spermine. Spermine and spermidine are selectively released from striatum by NMDA receptor activation, but not by kainate or quisqualate (Fage et al., 1992). Interestingly, Siddiqui and Iqbal (1994) observed that NMDA receptor-mediated ⁴⁵Ca²⁺fluxes can be blocked by a polyamine synthesis inhibitor. Porcellaet al. (1991) and Kish et al. (1991) showed that a polyamine synthesis inhibitor is neuroprotective against NMDA-induced brain damage in vivo. These observations suggest that the specificity of the NMDA-induced mitochondrial response may be dictated via second messenger activation. In this scenario, activation of NMDA receptors, but not other receptors, may increase local intracellular concentrations of polyamines that would then activate the rapid mode of mitochondrial Ca²⁺ uptake. This possibility is under investigation.

An alternative explanation to be considered is that the NMDA receptor simply allows more calcium to enter the neuron in a shorter period of time than other stimuli. Studies of⁴⁵Ca²⁺ uptake (Hartleyet al., 1993; Eimerl and Schramm, 1994) and experiments using low affinity Ca²⁺ indicators (Hyrc et al., 1997) support this idea. Because the mitochondrial uniporter has a relatively low affinity, it is more efficient at higher cytosolic Ca²⁺ levels. However, this would not seem to be an adequate explanation for the complete lack of a lag phase in NMDA-induced mitochondrial Ca²⁺ uptake. It is also necessary to consider the possibility that mitochondria are simply in closer proximity to NMDA receptors than other routes of Ca²⁺ entry.

Regardless of the mechanisms, it is clear that Ca²⁺ loads that enter the cytoplasm during NMDA receptor stimulation have more rapid, or “privileged” access to mitochondria. This likely accounts for the selectivity of NMDA receptor activation in producing mitochondrial damage via Ca²⁺-dependent generation of free radicals and collapse of mitochondrial membrane potential.