A high cytoplasmic Na+ concentration may contribute to N-methyl-d-aspartate (NMDA)-induced excitotoxicity by promoting Ca2+ influx via reverse operation of the Na+/Ca2+ exchanger (NaCaX), but may simultaneously decrease the electrochemical Ca2+driving force by depolarizing the plasma membrane (PM). Digital fluorescence microscopy was used to compare the effects of Na+ versus ions that do not support the NaCaX operation, i.e., N-methyl-d-glucamine+ or Li+, on: PM potential; cytoplasmic concentrations of Ca2+, H+, and K+; mitochondrial Ca2+ storage; and viability of primary cultures of cerebellar granule cells exposed to NMDA receptor agonists. In the presence of Na+ or Li+, NMDA depolarized the PM and decreased cytoplasmic pH (pHC); in the presence of Li+, Ca2+ influx was reduced, mitochondrial Ca2+ overload did not occur, and the cytoplasm became more acidified than in the presence of Na+. In the presence ofN-methyl-d-glucamine+, NMDA instantly hyperpolarized the PM, but further changes in PM potential and pHC were Ca-dependent. In the absence of Ca2+, hyperpolarization persisted, pHC was decreasing very slowly, K+ was retained in the cytoplasm, and cerebellar granule cells survived the challenge; in the presence of Ca2+, pHC dropped rapidly, the K+concentration gradient across the PM began to collapse as the PM began to depolarize, and Ca2+ influx and excitotoxicity greatly increased. These results indicate that the dominant, very likely excitotoxic, component of NMDA-induced Ca2+ influx is mediated by reverse NaCaX and that direct Ca2+ influx via NMDA channels is curtailed by Na-dependent PM depolarization.
Excitotoxicity has been causally linked with glutamate-elicited Ca2+ influx (Choi, 1987; Hartley et al., 1993;Eimerl and Schramm, 1994) and with Ca2+accumulation in mitochondria (Kiedrowski and Costa, 1995; White and Reynolds, 1996; Schinder et al., 1996; Stout et al., 1998). Although exposure to glutamate also greatly elevates cytoplasmic Na+ concentrations ([Na+]C) in cultured neurons (Kiedrowski et al., 1994; Pinelis et al., 1994), it is yet unclear what role the increase in [Na+]C plays in excitotoxicity. Although excessive swelling caused by the influx of Na+, Cl−, and water leads to neuronal death (Rothman, 1985; Choi, 1987), such a mechanism characterizes kainate- rather than glutamate- or NMDA-induced excitotoxicity (Kiedrowski, 1998). Earlier studies have shown that high [Na+]C contributes to the NMDA-induced Ca2+ influx by activating the reverse operation of the plasma membrane Na+/Ca2+ exchanger (NaCaX) (Kiedrowski et al., 1994; Hoyt et al., 1998). The activation of the reverse NaCaX also requires that cytoplasmic Ca2+be raised to micromolar levels (Hilgemann et al., 1992). This makes the channels that are permeable for both Na+ and Ca2+, such as NMDA receptor channels (Mayer and Westbrook, 1987), perfect triggers of the reverse NaCaX activation.
The role of the NaCaX in excitotoxicity was previously tested in vitro using Na-free media (Mattson et al., 1989; Storozhevykh et al., 1998) in which Na+ was substituted with a large cation,N-methyl-d-glucamine (NMG+), that very poorly, if at all, permeates NMDA receptor channels. These studies showed that in the presence of NMG+, the NMDA-induced Ca2+influx and excitotoxicity are greatly enhanced, and the data were interpreted as indicating that such an outcome results from an inhibition of Ca2+ extrusion by the NaCaX (Mattson et al., 1989), or from an alleged excessive release of endogenous glutamate (Storozhevykh et al., 1998). Because these interpretations failed to consider that the substitution of Na+ with NMG+ might prevent the Na-dependent plasma membrane (PM) depolarization that normally accompanies activation of NMDA receptors, the aim of the present study was to test what impact such depolarization has on the NMDA-induced Ca2+ influx and excitotoxicity.
To this end, using digital fluorescence microscopy and fluorophores sensitive to the plasma membrane potential (Em) and to cytoplasmic Ca2+ concentrations ([Ca2+]C), Em, and [Ca2+]C were simultaneously monitored in primary cultures of cerebellar granule cells (CGCs) exposed to NMDA. NMDA receptors were activated by an agonist, whereas extracellular Na+ was present or substituted either with NMG+ or with Li+. Because neither NMG+nor Li+ support Ca2+transport via the NaCaX (Blaustein, 1977; Hilgemann 1989), but Li+, in contrast to NMG+, permeates NMDA receptor channels (Tsuzuki et al., 1994) and should depolarize the PM (Hösli et al., 1973), this experimental design was expected to isolate the effects of PM depolarization from the effects of Na-dependent activation of the NaCaX. Because substitution of extracellular Na+ with Li+ or NMG+ is also expected to affect the operation of the plasma membrane Na+/H+ exchanger (NaHX), which controls cytoplasmic pH (pHC) (Aronson, 1985), and because the role of pHC in Ca2+ homeostasis is obscure, pHC was also monitored.
Materials and Methods
CGCs were prepared from 8-day-old Sprague-Dawley rats and suspended in culture medium consisting of basal Eagle’s medium supplemented with 25 mM KCl, 10% bovine fetal serum, 2 mM glutamine, and 50 μg/ml gentamycin, as described previously (Kiedrowski et al., 1994). The cells were plated on poly-d-lysine (10 μg/ml)-coated 35-mm dishes at a density of 2 × 106 cells/dish; for the imaging experiments, CGCs were plated on poly-d-lysine-coated 2.5-cm glass coverslips. Glial proliferation was curtailed by the addition of 10 μM cytosine arabinofuranoside 24 h after plating. Cultures at 8 to 11 days in vitro were used for the experiments.
Experimental solutions were based on Locke’s buffer (Na-Locke’s) containing 154 mM NaCl, 5.6 mM KCl, 3.6 mM NaHCO3, 1.3 mM CaCl2, 1 mM MgCl2, 5 mM glucose, and 10 mM HEPES, pH 7.4, adjusted with Tris. The Na-free Locke’s buffer in which Na+ was replaced with Li+(Li-Locke’s) contained 157.6 mM LiCl, 2 mM KCl, and 3.6 mM KHCO3, with the remaining ingredients the same as in Na-Locke’s. When Na+ was replaced with NMG+ or Cs+, the buffer was identical with Li-Locke’s except that Li+ was substituted with NMG+ or Cs+, respectively. Ca-free solutions contained 3 mM EGTA and no CaCl2, with other components of Locke’s buffer unchanged. Glutamate (100 μM) or NMDA (300 μM) were applied in Mg-free media containing 10 μM glycine. Glucose-free solutions contained 5 mM 2-deoxy-d-glucose instead ofd-glucose. The mitochondria-depolarizing cocktail (MDC) was composed of a Ca- and glucose-free Na-Locke’s supplemented with 10 μM cyanide m-chlorophenylhydrazone (CCCP), 3 μg/ml oligomycin, and 10 μM MK-801 [(+)5-methyl-10–11-dihydro-5H-dibenzocyclohepten-5,10-imine].
[Ca2+]C levels were monitored in CGCs exposed to glutamate as described in the text. Following glutamate withdrawal, by washing with Na-Locke’s, [Ca2+]C levels were monitored for an additional 20 to 30 min. Then the cells were returned to a conditioned medium (CM), i.e., the medium (containing 25 mM KCl) in which the cells were cultured. Viability of the same cells was assessed after 20 to 24 h by comparing oblique illumination images of the neurons occupying the same microscopic field before and 20 to 24 h after the excitotoxic challenge; neurons with a damaged PM were identified using propidium iodide (Jones and Senft, 1985). To this end, the cells were incubated for about 10 min at room temperature with 10 μM propidium iodide dissolved in Na-Locke’s, and the propidium iodide fluorescence emitted at over 520 nm after 488 nm excitation was imaged. Whether the positions of propidium iodide-positive spots corresponded to the positions of dead cells was examined by digitally subtracting, using the Attofluor software (Atto Instruments, Rockville, MD), the images of the propidium fluorescence from the oblique illumination images of the same microscopic field. This procedure allows one to distinguish false signals, i.e., the propidium iodide-positive spots representing cellular debris floating in the medium. In some experiments propidium iodide was added to the CM immediately after the challenge with glutamate, and the progression of the plasma membrane deterioration was followed for up to 70 h by monitoring the propidium iodide fluorescence.
Simultaneous Assay of [Ca2+]C and Em.
The coverslips plated with CGCs were transferred to custom-made recording chambers, in which the cells were loaded for 60 min at 37°C with 4 μM of fura-2 acetoxymethyl ester dissolved in CM. The stock concentration of fura-2 acetoxymethyl ester was 1 mM, in dimethyl sulfoxide (DMSO). Following the loading, the cells were washed using CM without fura-2. Fluorescence data were acquired from the cells that, based on their neuronal appearance and the small size of the cell bodies (about 10 μm in diameter), could be identified as CGCs. Occasionally, larger cells (about 20 μm in diameter), most likely astrocytes, were also observed and monitored. All imaging experiments were carried out at 37°C using the TC-102 temperature controller and the LU-CB1 Leiden culture system from Medical Systems Corp. (Greenvale, NY). Monitoring of the fura-2 fluorescence was begun while the cells were still incubated in CM; the pH of the CM was maintained at physiological levels by delivering 6% CO2 to the recording chamber. Then, the CM in the recording chamber was replaced with Na-Locke’s supplemented with 100 nM bis(1,3-dibutylbarbituric acid)trimethine oxonol [DiBAC4(3)] to monitor Em (Bräuner et al., 1984; Laskey et al., 1992).
The fura-2 and the DiBAC4(3) fluorescences were monitored using the Attofluor digital imaging system, Zeiss Axiovert 10 microscope (Carl Zeiss, Thornwood, NY), and Zeiss Achrostigmat 40×, NA 1.30, objective. The images of fluorescence emitted at over 520 nm after excitation at 334 nm (F334) and 380 nm (F380) for fura-2, and at 488 nm (F488) for DiBAC4(3) were saved every 10 to 20 s. The fluorescence intensities, measured in selected regions of interest, were analyzed retroactively. To display fura-2 and DiBAC4(3) distribution within single cells, using Adobe Photoshop 4.0, the superimposed images of fura-2 and DiBAC4(3) fluorescence were created by transferring 8-bit gray scale images of fura-2 fluorescence (F334 + F380) and DiBAC4(3) fluorescence (F488) to red and green channels, respectively, of a 24-bit RGB (red-green-blue) image.
The DiBAC4(3) fluorescence measured in regions of interest positioned in the peripheral parts of neuronal somata was used as a relative index of Em. To normalize this fluorescence among various neurons, at the end of the experiments, the cells were depolarized using the following solution in which chloride was partially replaced with gluconate to prevent cell swelling: 5 μM gramicidin D, 134.2 mM K-gluconate, 25.4 mM KCl, 1.3 mM CaCl2, 1 mM MgCl2, 3.6 mM KHCO3, and 10 mM HEPES, pH 7.2, adjusted with Tris; this exposure caused an increase in F488 to a value representing the maximal depolarization (maxF488). The F488 in the background (bckgF488) was measured in cell-free areas. The normalized F488(nF488) was calculated according to the formula: [Ca2+]Ccalibration was performed in situ. The minimal F334/F380 ratio (Rmin) was measured in CGCs incubated for up to 30 min in a buffer containing 10 μM ionomycin, 5 mM EGTA, 154 mM NaCl, 5.6 mM KCl, 1 mM MgCl2, 3.6 mM NaHCO3, 5 mM glucose, and 10 mM HEPES, pH 7.4. The maximal F334/F380 ratio (Rmax) was measured at the end of the experiments by exposing the cells to 10 μM ionomycin, 10 μM CCCP, 134.2 mM K-gluconate, 25.4 mM KCl, 5 mM CaCl2, 1 mM MgCl2, 3.6 mM KHCO3, 10 mM HEPES, pH 7.2, adjusted with Tris; after 5 to 10 min, a stable F334/F380 ratio was reached, which was assumed to represent Rmax. The in situ calibration procedure also determined the ratio (β) of the fluorescence emitted at 380 nm by fura-2-Ca-free versus fura-2-Ca-saturated. Rmax, Rmin, β, and the association constant of fura-2 and Ca2+, of 224 nM, were used to calculate [Ca2+]C as described byGrynkiewicz et al. (1985). The background F334and F380 values were measured in cell-free areas at all time points of the experiments. When the F334/F380 ratio data were calibrated for [Ca2+]Cwith background subtraction, the average [Ca2+]C values did not differ by more than 10% from the values calculated without background subtraction. The data noise during the Rmax data acquisition, was greatly increased, however, when the background subtraction was performed (data not shown). This increase in the Rmax data noise was due to the fact that the background subtraction procedure greatly increased the variability of the already very low F380 values recorded during the Rmax acquisition. Because the effect of the background subtraction procedure on the calculated [Ca2+]C values was small and might have resulted from the increased noise of the F380 data, the [Ca2+]C values presented in this report were calculated without background subtraction.
Simultaneous Assay of [Ca2+]C and pHC.
The coverslips with CGCs were handled the same way as described above for the assay of [Ca2+]C and Em except that the cells were loaded for 60 min at 37°C with 4 μM fura-2 acetoxymethyl ester plus 0.2 μM BCECF (2′,7′-bis-(2-carboxyethyl)-5-(and-6)-carboxyfluorescein) acetoxymethyl ester. The stock concentration of BCECF acetoxymethyl ester was 0.05 mM, in DMSO. Following the loading, the cells were washed using CM without fura-2 and BCECF. Images of the fluorescence emitted at over 520 nm at 334, 380, 440, and 488 nm excitations were saved every 10 to 20 s. [Ca2+]C was calculated from the F334/F380 ratio as described earlier. The F488/F440 ratio was used to determine pHC from the in situ calibrations performed after each experiment. The background F440 and F488 values were measured in cell-free areas at all time points and were subtracted from the fluorescence data before calculating the F488/F440 ratio. The pH calibrating solutions contained 10 μM nigericin, 10 μM CCCP, 134.2 mM K-gluconate, 25.4 mM KCl, 1.3 mM CaCl2, 1 mM MgCl2, 3.6 mM KHCO3, and 10 mM 2-[N-morpholino]ethanesulfonic acid or 10 mM HEPES. 2-[N-morpholino]ethanesulfonic acid was used to adjust the pH of the calibrating solution to a range of 5.5 to 6.5; HEPES was used in the pH range of 7.0 to 8.0. The data were fitted to the four-parameter equation of a sigmoidal curve: Representative experiments (n = 7) yielded the following parameters: y0 = 0.39 ± 0.01, a = 1.18 ± 0.06, x0 = 6.99 ± 0.03, and b = 0.41 ± 0.02.
Assay of Cytoplasmic K+ Concentration ([K+]C).
CGCs were loaded for 60 min at 37°C with 5 μM K+ binding benzofuran isophthalate (PBFI) acetoxymethyl ester dissolved in CM. The stock concentration of PBFI acetoxymethyl ester was 1.25 mM, in DMSO. PBFI fluorescence was monitored at 37°C using the same excitation and emission settings as described for fura-2. The F334/F380 ratio was calibrated for [K+]C in situ at the end of the experiments. The [K+]C calibrating buffers were prepared by appropriate mixing of two solutions containing high and low concentrations of K+. The high-concentration K+ solution contained 5 μM gramicidin D, 134.2 mM K-gluconate, 25.4 mM KCl, 3.6 mM KHCO3, 1 mM MgCl2, and 10 mM HEPES, pH 7.2, adjusted with Tris. In the low-concentration K+ solution, 134.2 mM Li-gluconate and 25.4 mM LiCl were used instead of the respective K+salts, and the remaining ingredients were unchanged. [K+]C values were calculated using a nonlinear least-squares fit of the data to the Michaelis-Menten equation as described by Kasner and Ganz (1992).
CGCs were incubated for 15 min at 37°C with 1 ml of experimental media containing 1 μCi of 45Ca2+. The extracellular 45Ca2+ was then removed by triple washing with 2 ml of an ice-cold buffer that contained 154 mM NaCl, 5.6 mM KCl, 3.6 mM NaHCO3, 1 mM MgCl2, 2 mM EGTA, and 10 mM HEPES, pH 7.4, adjusted with Tris. The cells were then dissolved in 1 ml of 0.5 M NaOH; neutralized aliquots of this solution were used for scintillation spectroscopy and for protein determination.
All averaged data are expressed as the means ± S.E.M. The statistical tests used are indicated in the text. All statistical analyses of the [Ca2+]C data were performed using the F334/F380 ratios. The [Ca2+]C data could not be used for statistical analysis because when [Ca2+]C approached the fura-2 saturating levels, the calculated [Ca2+]C values became very imprecise. In some experiments, the fura-2 fluorescence was monitored using different sets of neutral density filters and/or gains, which affected the absolute values of the F334/F380 ratios (for example, note the differences in F334/F380 ratios between Figs. 2 and 6). To normalize the data, the F334/F380 ratios were converted to [Ca2+]Cvalues using the calibrating parameters Rmax, Rmin, and β obtained in sister cultures in the same experimental sessions; then, the [Ca2+]C values from various experiments were converted to the normalized F334/F380 ratio using a single representative set of the calibrating parameters Rmax′, Rmin′, and β′, according to the formula:
Fura-2 acetoxymethyl ester, PBFI acetoxymethyl ester, BCECF acetoxymethyl ester, and DiBAC4(3) were obtained from Molecular Probes (Eugene, OR). MK-801 was purchased from Research Biochemicals Inc. (Natick, MA). The culture media and all other chemicals were from Sigma Chemical Co. (St. Louis, MO).
Results and Discussion
[Ca2+]C Levels in CGCs Incubated in CM.
Although the very low (about 20 nM) [Ca2+]C levels in CGCs incubated in Na-Locke’s are very homogeneous, a small number of CGCs show a marked heterogeneity in glutamate-induced [Ca2+]C transients and fail to efficiently buffer cytoplasmic Ca2+ (see Fig. 4C in Kiedrowski and Costa, 1995). In the present study, it was routinely observed that it is possible to predict which neurons will fail to buffer the glutamate-induced [Ca2+]C transients. From among 1713 CGCs (10 different platings) incubated in CM (containing 25 mM K+), a subpopulation of 72 neurons could be distinguished that maintained [Ca2+]C at much higher levels (>600 nM) than the majority of CGCs (300–400 nM), but that nevertheless promptly decreased [Ca2+]C to about 20 to 30 nM when CM was replaced with Locke’s buffer and the extracellular K+([K+]E) concentration was reduced from 25 to 5.6 mM (Fig. 1). Such neurons failed to buffer the [Ca2+]C transients elicited by NMDA (see representative data in Fig. 1) or glutamate (data not shown). Because this subpopulation of neurons appears to have distinctly different properties, in the present study these neurons were singled out, and their data were excluded when average responses, typical for the majority of CGCs, were calculated. This abnormal subpopulation of CGCs requires an additional, separate, characterization.
Substitution of Na+ with NMG+ Exacerbates Glutamate-Induced Destabilization of Ca2+ Homeostasis and Excitotoxicity.
When the Na+ in the extracellular medium of CGCs was replaced with NMG+(NMG-Locke’s), a 15-min exposure to glutamate (100 μM glutamate plus 10 μM glycine in the absence of Mg2+) at 37°C resulted in a permanent destabilization of Ca2+ homeostasis (Fig.2A). When the test of neuronal viability based on propidium iodide staining was performed 24 h after the neuronal challenge with glutamate, it was observed that many neurons were propidium iodide-negative in spite of evident morphological changes, such as a prominent shrinkage of the cell body (see white arrowheads in Fig. 2, A2 and E2). The lack of propidium iodide staining in such neurons complicated the quantitation of excitotoxicity. Obviously, 24 h after the challenge with glutamate the process of neurodegeneration was not yet accomplished. Therefore, additional experiments were performed in which neuronal viability was monitored for up to 70 h after the glutamate exposure (Fig. 2E). It was observed that of 182 neurons exposed to glutamate and NMG-Locke’s, 115 died within the first 24 h, and an additional 45 during the next 46 h (Fig. 2F).
An analysis of whether the time of neuronal death is related to [Ca2+]C changes during and after the glutamate exposure showed that the neurons that failed to decrease [Ca2+]C levels within 20 min after glutamate withdrawal died within 19 h (compare the positions of neurons with high [Ca2+]C, red-colored in Fig. 3A4, with the positions of dead neurons in Fig. 3A6); all neurons that decreased [Ca2+]C at least to the levels characteristic for CGCs incubated in CM were alive at that time (indicated with green arrowheads in Fig. 3A4). The neuron that buffered [Ca2+]C sluggishly (indicated with white arrowhead in Fig. 3A4), was still alive after 19 h (Fig. 3A6), but was dead at 29 h (Fig. 3A7). Interestingly, a subpopulation of neurons begun to die after an even longer delay postglutamate; note two neurons that restored low [Ca2+]C levels within 20 min after glutamate withdrawal (yellow arrowheads in Fig. 3A4), survived the first 29 h, but were dead at 44 h (Fig. 3A8). It appears that also in this case neuronal death was induced by the Ca2+ influx elicited by NMDA receptor activation because the CGCs that were exposed to glutamate in the presence of MK-801 survived for as long as 68 h (Fig. 3B). These results suggest that glutamate exposure may induce two Ca2+ influx-dependent but distinct mechanisms of neuronal death: the first one occurring within the first 24 h after the glutamate challenge and expressed in the impaired ability to restore basal Ca2+ levels in the cytoplasm, and the second one occurring during the next 24 h and triggered by the initial Ca2+ influx but not accompanied by a destabilization of Ca2+ homeostasis. Further work is necessary to characterize these two mechanisms on a molecular level.
In the overwhelming majority of CGCs, the glutamate-elicited [Ca2+]C transient and excitotoxicity were completely blocked by 10 μM MK-801 (Fig. 3B), which indicates that the [Ca2+]C transient and excitotoxicity result from the activation of the NMDA class of receptors. Only a single neuron was dead after 68 h (Fig. 3B7) but not 19 h (Fig. 3B6) after the exposure to glutamate plus MK-801. This happened to be the same neuron in which MK-801 failed to inhibit the glutamate-induced [Ca2+]C transient (see arrowhead in Fig. 3B3). For the statistical relevance of this finding, more data have to be accumulated.
The data shown in Fig. 3 indicate that by using propidium iodide fluorescence, one can monitor the progress of neurodegeneration in vitro for an extended time, and relate the time of neuronal death to the severity of the glutamate-induced destabilization of Ca2+ homeostasis. Apparently, for up to 68 h propidium iodide by itself was not toxic to the neurons (Fig. 3B7). When PM was permeabilized with 0.01% Triton X-100, propidium iodide fluorescence was promptly observed in all CGCs (Fig. 3B8), which indicates that even as long as 68 h after the propidium iodide addition neuronal death could be detected.
When CGCs were challenged with glutamate while physiological Na+ concentrations were present in the medium, following glutamate withdrawal, [Ca2+]C returned to basal levels (Fig. 2C) and neuronal survival was greatly improved compared with the challenge with glutamate and NMG-Locke’s: of 205 neurons exposed to glutamate and Na-Locke’s, only 27 neurons died within 24 h and no additional neuronal death was detected for up to 70 h (Fig. 2, D and F). Interestingly, when CGCs were incubated in Na-Locke’s and exposed to glutamate at room temperature but otherwise under identical experimental conditions, the glutamate-induced destabilization of Ca2+ homeostasis was more pronounced and the exposure was more excitotoxic (Kiedrowski, 1998). The temperature sensitive component(s) of the excitotoxic mechanism needs to be further characterized.
An enhancement of the NMDA-induced excitotoxicity by a Na-free buffer (Na+ substituted with NMG+) was reported earlier and interpreted to indicate that the increased excitotoxicity in the Na-free medium was due to a failure by the NaCaX to remove Ca2+ from the cytoplasm (Mattson et al., 1989), or due to an alleged glutamate release (Storozhevykh et al., 1998). An alternative and straightforward explanation needs to be considered, however: namely, that the replacement of Na+ with a large extracellular cation, which cannot permeate NMDA channels, prevents the PM depolarization elicited by Na+ influx (Hösli et al., 1973). As a result, the electrochemical driving force for Ca2+ influx (CaDF), defined as the difference between Em and the Ca2+equilibrium potential, does not decay, or decays at a lower rate, which has to enhance Ca2+ influx via any Ca-permeable channel. To test the latter hypothesis, Emand [Ca2+]C were monitored simultaneously in CGCs exposed to NMDA.
Simultaneous Monitoring of Em and [Ca2+]C.
We have previously observed that activation of NMDA or kainate/α-amino-3-hydroxy-5-methyl-4-isoxazole propionic acid receptors in CGCs results in an elevation of [Na+]C up to 60 mM or higher (Kiedrowski et al., 1994), which depolarizes the PM. While evaluating various means of monitoring Emchanges, I was concerned that a direct electrophysiological approach might not faithfully reflect the Emchanges characteristic for an intact neuron exposed to glutamate because the ionic content of the intracellular electrode, in particular high [K+], might obscure the glutamate-induced changes in [Na+]C and [K+]C characteristic for an intact neuron. Therefore, to monitor Em for the purpose of this study, a noninvasive method was chosen, which makes use of the fluorescence of an Em-sensitive dye, DiBAC4(3), an anionic fluorescent probe that accumulates in the cytoplasm of depolarized cells via a Nernst equilibrium-dependent uptake (Bräuner et al., 1984). [Ca2+]C and Em were monitored simultaneously according to the approach described by Laskey et al. (1992). Cells were first loaded with fura-2 and then exposed to 100 nM DiBAC4(3) (see Materials and Methods for details). After illumination at 334 and 380 nm to excite fura-2, there was no fluorescence emission by DiBAC4(3), and inversely, after 488 nm illumination to excite DiBAC4(3), there was no fluorescence emission by fura-2 (Fig. 4, B and C); therefore [Ca2+]C and Em could be monitored selectively in a single cell. PM depolarization was associated with a robust increase in DiBAC4(3) fluorescence, which was very intense in peripheral parts of cell bodies but virtually absent in the nuclear area (Fig. 4B). The apparent lack of DiBAC4(3) fluorescence in the nucleus of depolarized cells is consistent with earlier observations (Bräuner et al., 1984) and was confirmed using confocal microscopy (data not shown). Therefore, the peripheral parts of cell bodies were chosen to monitor DiBAC4(3) fluorescence as an index of Em. To test the reliability of these Em and [Ca2+]C measurements, the Em changes were imposed on CGCs by applying an unselective monovalent cation ionophore, gramicidin D (5 μM), and a Na- and Ca-free solution in which the K+concentration was changed from 163.2 to 3.6 mM by substituting K+ with NMG+. Under these conditions, 163.2 mM K+ is expected to completely depolarize the PM, whereas 3.6 mM K+ is expected to create a large, negative inside, K+ diffusion potential. Repetitive applications of depolarizing pulses of 163.2 mM K+ or hyperpolarizing pulses of 3.6 mM K+ were associated with respective increases or decreases in DiBAC4(3) fluorescence intensity, indicative of the expected Em changes (Fig. 4E). These changes in Em failed to affect [Ca2+]C in the majority of CGCs (Fig. 4E). In the above-described abnormal 4% subpopulation of CGCs with high [Ca2+]Cwhile incubated in the CM, however, the hyperpolarizing pulses of 3.6 mM K+ were associated with [Ca2+]C transients (data not shown), which, considering that the extracellular medium was Ca-free, represented Ca2+ release from intracellular stores. Similar Ca2+ transients were observed in cells that were about twice as large as CGCs and probably represented astrocytes (Fig. 4D). The lack of such intracellular Ca2+ release in the majority of CGCs confirms our earlier observations that intracellular Ca2+ stores in unstimulated CGCs are virtually depleted (Kiedrowski and Costa, 1995). The mechanism of the PM hyperpolarization-induced Ca2+ release from intracellular stores requires further investigation.
The Em-dependent changes in DiBAC4(3) fluorescence intensity induced by a change in Em were readily detectable; however, they occurred at very slow rates, which was expected (Bräuner et al., 1984). For example, application of 163.2 mM K+ for as long as 3 min was not long enough to bring the DiBAC4(3) fluorescence to a steady plateau (Fig. 4E). Therefore, attempts to calibrate the DiBAC4(3) fluorescence for millivolts were abandoned. Instead, the DiBAC4(3) fluorescence was used as a relative measure of Em: increase of the fluorescence meaning depolarization, decrease meaning hyperpolarization.
Substitution of Na+ with NMG+ Increases NMDA-Mediated Ca2+ Influx by Affecting Em.
The above-described approach of a simultaneous monitoring of [Ca2+]C and Em was used to test whether the NMG-induced destabilization of Ca2+ homeostasis (Figs. 2A and3A) was caused by a dysfunction of the NaCaX or by a difference in Em. Because in the overwhelming majority of CGCs incubated in NMG-Locke’s glutamate elicited Ca2+influx and excitotoxicity by activating NMDA receptors (Fig. 3 B), in subsequent experiments NMDA rather than glutamate was used as an excitatory stimulus. CGCs were exposed to NMDA (300 μM NMDA plus 10 μM glycine, in the absence of Mg2+); for the first 2 min, NMDA was applied in a standard Locke’s buffer containing Na+ (Na-Locke’s), and the expected [Ca2+]C transient and depolarization of the PM promptly occurred (Fig.5A). Then Na+ was replaced with Li+ or NMG+, neither of which supports the NaCaX, but the remaining components of the previous solution were not changed. If the destabilization of Ca2+ homeostasis observed in CGCs exposed to glutamate when Na+ is replaced with NMG+ is caused by an inhibition of Ca2+ extrusion via the NaCaX, a replacement of Na+ with Li+ should yield an effect similar to that obtained by replacement of Na+ with NMG+. The switch from Na+ to NMG+ affected [Ca2+]C and Em differently, however, than the switch from Na+ to Li+ (Fig. 5A). In the presence of Ca2+, NMG+elicited a robust increase in [Ca2+]C, and transiently hyperpolarized the PM; in the absence of Ca2+, the NMG-dependent hyperpolarization persisted for as long as it was studied (Fig. 5A, lower). By contrast, when Na+was replaced with Li+, NMDA depolarized the PM similarly as if Na+ were present (Fig. 5A, lower), but [Ca2+]Cdecreased, within 5 min, to significantly lower levels, i.e., 360 ± 8 nM in the presence of Li+ (72 cells in four experiments) versus 665 ± 22 nM in the presence of Na+ (74 cells, four experiments), respectively (P < .001, Mann–Whitney rank sum test; see Fig. 5A, upper for representative data).
Because the [Ca2+]C data may represent Ca2+ influx as well as Ca2+ redistribution between the cytoplasm and the organelles, such as mitochondria and/or endoplasmic reticulum, the NMDA-elicited 45Ca2+accumulation was also monitored under similar experimental conditions; the 45Ca2+ accumulation data provide complementary information that characterizes the amount of Ca2+ that enters the cells. As shown in Fig. 5B, in a dose-dependent manner NMG+ potentiated, whereas Li+ inhibited, the NMDA-elicited45Ca2+ accumulation. From these data, it can be inferred that NMG+ enhances and Li+ inhibits the NMDA-elicited Ca2+ influx.
These results can be interpreted to indicate that the mechanism of the NMDA plus NMG+-induced neuronal overload with Ca2+ involves PM hyperpolarization and a consequent increase in the CaDF. Although Em in CGCs exposed to NMDA in the presence of Li+changed similarly as in the presence of Na+, there was an additional inhibition of the NMDA-induced Ca2+ influx by Li+. This result may be related to the fact that Li+ does not support the reverse operation of the NaCaX (Hilgemann, 1989), which contributes significantly to the NMDA-elicited Ca2+ influx in cultured neurons (Kiedrowski et al., 1994; Hoyt et al., 1998). It has to be noted, however, that the Ca2+/Ca2+ exchange mode of the NaCaX, which is potently stimulated by extracellular Li+ (Blaustein, 1977; Slaughter et al., 1983;DiPolo and Beaugé, 1990), may also contribute to the inhibitory effect of Li+ on the NMDA-induced Ca2+ influx. Discrimination between Li+ effects on the Ca2+/Ca2+ exchange versus the Na+/Ca2+ exchange mode in CGCs exposed to NMDA requires further work.
Relationships among Em, [Ca2+]C, and pHC.
Because NMG+, in contrast to Li+, does not support the operation of the NaHX (Aronson, 1985), one might speculate that a cytoplasmic acidification, caused by a tonic inhibition of NaHX by NMG+ (Raley-Susman et al., 1991), might affect Ca2+ homeostasis and contribute to the observed effects of substituting of Na+ with NMG+ versus Li+. To test whether this is the case, the effects of Li+ or NMG+ on NMDA-induced changes in [Ca2+]C and pHC were measured simultaneously using the Ca2+- and the pH-sensitive fluorescent dyes fura-2 and BCECF, respectively (see Materials and Methodsfor details). Basal pHC in CGCs incubated in a standard Locke’s buffer was 7.04 ± 0.01 (n = 340), which is consistent with previous observations (Raley-Susman et al., 1991; Hartley and Dubinsky, 1993; Irwin et al., 1994). While performing a simultaneous calibration of fura-2 and BCECF data in situ, it was observed that when pHC dropped below 7.0, the maximal F334/F380 ratio of fura-2 (Rmax) progressively decreased (Fig.6 A). It appears that a drop of pHC below 7.0 affects the fura-2 fluorescence properties and, therefore, the in situ calibration of fura-2 performed at a pHC of about 7.2 to 7.4 is not valid when pHC drops below 7.0. Although in this report it was not attempted to correct for this artifact, the pHC and the [Ca2+]C data have been displayed simultaneously, and one can determine when the fura-2 F334/F380 ratio might have been affected by the excessive drop in pHC.
To test whether pHC may directly affect [Ca2+]C, CGCs were exposed to 10 μM CCCP plus oligomycin (3 μg/ml). It was expected that CCCP, a protonophore, would disturb H+equilibria between the cytosol and the basic and acidic cytoplasmic organelles, as well as between the cytosol and the extracellular medium. Oligomycin was included to prevent hydrolysis of cytoplasmic ATP by mitochondrial ATP-ase (Budd and Nicholls, 1996). As shown in Fig. 6B, an application of CCCP plus oligomycin within 2 min caused a drop in pHC by 0.72 ± 0.01 pH units (n = 47), whereas [Ca2+]C remained unchanged. One may interpret this result as an indication that a drop in pHC does not disturb Ca2+ homeostasis at least while [Ca2+]C is maintained at basal levels.
To further explore the possibility of a link between pHC and [Ca2+]C regulation, how pHC and [Ca2+]C are affected by NMDA receptor activation in the presence of NMG+or Li+ was tested. It is well established that activation of NMDA receptors in a Ca-dependent manner acidifies the neuronal cytoplasm (Hartley and Dubinsky 1993; Irwin et al., 1994), and that a substitution of extracellular Na+ with NMG+ brings about a rapid further drop in pHC, which has been interpreted as the result of a NaHX inhibition by NMG+ (Hartley and Dubinsky, 1993). One has to consider, however, that the PM hyperpolarization induced by NMG+ (Fig. 5A) may contribute to the acidification of the cytoplasm by two additional mechanisms: 1) an efflux of H+ from the cytoplasm, as also an efflux of any other cation, is expected to be directly inhibited by PM hyperpolarization; and 2) PM hyperpolarization, by increasing the CaDF, increases Ca2+ influx, and is expected to enhance the mechanism by which Ca2+ influx alone affects pHC (Irwin et al., 1994).
When extracellular Na+ was replaced with NMG+ under Ca-free conditions, pHC decreased by 0.13 ± 0.01 pH units (n = 79), a subsequent application of CCCP plus oligomycin caused, similarly as in CGCs incubated in Na-Locke’s, a rapid drop in pHC, by 1.00 ± 0.02 pH units (compare Fig. 6, B and C). When NMDA was then added, neither [Ca2+]C nor pHC changed unless Ca2+(1.3 mM) was added to the medium, which resulted in a rapid increase of pHC, by 0.63 ± 0.03 pH units within 10 s, followed by a further, slow increase in pHC, by 0.07 ± 0.02 pH units within the next 3 min; when, however, NMG+ was then replaced with Na+, a rapid restoration of basal pHC was observed and [Ca2+]C began to decrease (Fig. 6C). Based on the data shown in Fig. 5A, an obvious explanation of the Na- and also Ca-induced cytoplasmic alkalization is that the Na+ influx via the NMDA receptor channel, and to a lesser degree the Ca2+ influx, depolarize the PM and annihilate the force that keeps protons in the cytoplasm, i.e., the highly negative Em. One may insist however, that that Na-induced cytoplasmic alkalization is due to the activation of the NaHX rather than to the PM depolarization. To verify this claim, it was tested whether Cs+, a cation that does not support the NaHX (Aronson, 1985) but permeates the NMDA receptor channel (Tsuzuki et al., 1994) and depolarizes the PM, can mimic the alkalizing effect of Na+. As shown in Fig. 6C, Cs+ reproduced the effect of Na+, which confirms the dominant role of PM depolarization in the mechanism of the here observed cytoplasmic alkalization, as well as the role of PM hyperpolarization in enhancing the Ca-dependent acidification of the cytoplasm.
Figure 6C also shows that after NMG+ was substituted with Cs+ or Na+, [Ca2+]C began to decrease, although NMDA, CCCP, and oligomycin were still present in the Mg-free extracellular medium, and that the rate of this decrease was faster when NMG+ was substituted with Cs+ than with Na+. The mechanism of the decrease in [Ca2+]C might include Ca2+ extrusion by the plasma membrane Ca2+ ATPase, or Ca2+ uptake to an internal store other than mitochondria, because mitochondria have already been depolarized with CCCP while the [Ca2+]C decrease was observed. The faster rate of the [Ca2+]C decrease in the presence of Cs+ than in the presence of Na+ may reflect the fact that cytoplasmic Na+ but not Cs+ activates the reverse mode of NaCaX operation, which counteracts the [Ca2+]C drop.
When CGCs were exposed to NMG-Locke’s in the presence of Ca2+, pHC begun to slowly decrease, and as Ca2+ was removed the rate of pHC decrease remained unchanged (Fig. 6D). An application of NMDA under the Ca-free conditions also failed to affect the rate of the drop in pHC. Only when 1.3 mM Ca2+ was introduced into the medium, was there a rapid drop in pHC, by 0.97 ± 0.03 pH units (n = 38) within 2 min, which was accompanied by a steep rise in [Ca2+]C to fura-2-saturating levels (Fig. 6D). These data confirm that the NMDA-induced drop in pHC is a consequence of the NMDA-induced Ca2+ influx. The Ca2+ influx acidifies the cytoplasm via at least two concomitant mechanisms: 1) the activity of the plasma membrane Ca2+ pump, which exchanges intracellular Ca2+ for extracellular H+(Trapp et al., 1996); and 2) Ca2+ sequestration in mitochondria (discussed in the next section of this report).
Relationships among pHC, Mitochondrial Ca2+Overload, Plasma Membrane Na+/Ca2+ Exchange Operation, and Energy Metabolism in CGCs Exposed to NMDA.
Because glutamate excitotoxicity is associated with Ca2+sequestration in mitochondria (Kiedrowski and Costa, 1995; Schinder et al., 1996; White and Reynolds, 1996), which has been causally linked to neuronal death (Stout et al., 1998), it was necessary to determine how substitution of extracellular Na+ with NMG+ versus Li+ would affect the amounts of Ca2+ diverted to mitochondria. For this purpose, at the end of the 5-min period of monitoring [Ca2+]C and pHC in CGCs exposed to NMDA under various ionic conditions, the cells were treated with MDC, a Na-Locke’s buffer that was Ca-free and glucose-free and contained 10 μM CCCP, 3 μg/ml oligomycin, and 10 μM MK-801. Any increase in [Ca2+]C in response to the MDC application was expected to result from Ca2+ released from the depolarized cytoplasmic Ca2+ stores, including mitochondria.
In CGCs incubated in NMG-Locke’s and exposed to NMDA, MDC caused a rapid alkalization of the cytoplasm by 1.9 ± 0.05 pH units (n = 38) within 2 min, i.e., to pH levels exceeding the basal pHC; simultaneously, MDC caused an increase in the F334/F380 ratio of fura-2 fluorescence (Fig. 6D). Although it is reasonable to expect that MDC would release Ca2+ from mitochondria and therefore increase [Ca2+]C, the [Ca2+]C measured before the MDC additions was already at or close to the fura-2 saturating levels and it is unlikely that a further increase in [Ca2+]C could be detected. Because the MDC addition also caused a very marked increase in pHC (Fig. 6D), which affects fura-2 fluorescence (Fig. 6A), the MDC-induced increase in the F334/F380 ratio in these cells most likely represents a pH-dependent change in fura-2 fluorescent properties.
An application of NMDA to CGCs under control conditions, i.e., incubated in Na-Locke’s, caused a modest drop in pHC, by 0.2 ± 0.02 pH units (n = 84) within 2 min (Fig.7A), and a typical [Ca2+]C transient (compare the NMDA-induced [Ca2+]C changes shown in Figs. 5A and 6D). It has to be noted, however, that just before NMDA induced a drop in the F488/F440 ratio of BCECF fluorescence, a very short-lasting (less than 10 s) increase in this ratio was consistently observed (Fig. 7, A–C). To elucidate whether this increase in the F488/F440 ratio represents a transient increase in pHC and its mechanism, additional experiments are required.
MDC, when added to CGCs incubated in Na-Locke’s and exposed to NMDA for 5 min, caused a prompt decrease of [Ca2+]C and a slow increase of pHC to basal levels (Fig. 7A). When, however, the exposure to NMDA was prolonged to 15 min, among 86 neurons tested, three populations of CGCs could be distinguished based on the effects of MDC on [Ca2+]Cand pHC: 26% of neurons showed an instant increase in [Ca2+]C and a rapid alkalization of the cytoplasm, which overshot the basal pHC levels; in 40% of CGCs the MDC-induced increase in [Ca2+]Coccurred with a latency of about 1 min, and the MDC-elicited cytoplasmic alkalization occurred more slowly; in 35% of CGCs, following the MDC addition [Ca2+]C dropped rapidly and pHC began to slowly increase toward basal levels (Fig. 7C).
The latency in the MDC-induced increase in [Ca2+]C that was observed in as many as 40% of CGCs exposed to NMDA for 15 min can be interpreted to indicate that the Ca2+ accumulated in mitochondria of these neurons was not present as free, ionized Ca2+ but as calcium bound by a mitochondrial Ca2+ buffer. Only after the mitochondrial [Ca2+] dropped to levels at which dissociation of free Ca2+ from this buffer was favorable, could Ca2+ be released from the mitochondria. One may expect that the formation of calcium phosphate in the mitochondrial matrix (Carafoli, 1987) participates in the mitochondrial Ca2+ buffering.
The fact that all neurons exposed to NMDA for 5 min (Fig. 7A) and 35% of neurons exposed to NMDA for 15 min (Fig. 7B) failed to show any increase in [Ca2+]C after the MDC addition does not necessarily mean that the mitochondria of these neurons did not accumulate any Ca2+. It may well be that the amounts of Ca2+ accumulated in these mitochondria were small and could be retained in the bound form even after the mitochondria were depolarized with MDC.
The rapid cytoplasmic alkalization induced by MDC in CGCs that have been exposed to NMDA and showed the prominent increase in [Ca2+]C on the MDC addition (Figs. 6D and 7B) can be interpreted as follows. When [Ca2+]C reaches the levels at which the rate of Ca2+ influx into mitochondria exceeds the rate of Ca2+ extrusion from mitochondria, the mitochondria start to accumulate Ca2+, which depolarizes the inner mitochondrial membrane. To compensate for the mitochondrial membrane potential drop, regulatory mechanisms that extrude extra protons from the mitochondrial matrix are activated: for example, protons may be extruded by the mitochondrial ATPase at the expense of cytoplasmic ATP (Budd and Nicholls, 1996). The opposite scenario takes place when mitochondria are rapidly depolarized with MDC; the protons are then rapidly returned from the cytosol to the mitochondrial matrix whereas Ca2+ travels in the opposite direction. The overshooting of pHC above basal values probably occurs because a fraction of the protons that have been extruded from the mitochondrial matrix to the cytosol in response to the mitochondrial Ca2+ influx is lost due to H+ diffusion to the extracellular medium. As a result, at the time when MDC is applied, there is a deficit of protons in the cytosol, and when the protons return to the mitochondrial matrix, the cytosolic [H+] drops to lower than basal levels. Obviously, this interpretation of the data needs to be tested further.
When extracellular Na+ was replaced with Li+, the basal [Ca2+]C remained unaffected but pHC dropped within the first 2 min by 0.11 ± 0.01 pH units (n = 89) and then stabilized at this lower level. When NMDA was applied, a [Ca2+]C transient occurred, and within 2 min pHC dropped by 0.45 ± 0.02 pH units (n = 32); within the next 2 min of the exposure to NMDA, pHC spontaneously recovered by 0.10 pH units. An application of MDC 5 min after the NMDA addition resulted in a rapid drop in [Ca2+]C to basal levels and a slow increase of pHC toward basal levels (data not shown; see, however, Fig. 7D for similar data obtained in CGCs deprived of glucose). The Li-Locke’s-elicited drop in basal pHC, as well as the more pronounced NMDA-induced drop in pHC in CGCs incubated in Li-Locke’s than in Na-Locke’s, most likely reflects that Li+ is less efficient than Na+ in supporting the NaHX operation (Aronson, 1985).
Mitochondrial Ca2+ Overload in Energetically Challenged Neurons Results from Ca2+ Influx via Reverse Operation of Plasmalemmal NaCaX.
Because the principal goal of in vitro studies of excitotoxicity is an understanding of the mechanisms of neurodegeneration characteristic of an ischemic brain, in which glutamate receptors are activated while the system is deprived of energy, it was of interest to examine how mitochondria would contribute to Ca2+ homeostasis and how pHC would change under conditions of energy deprivation. To this end, [Ca2+]C and pHC were monitored in CGCs exposed to NMDA under glucose-free conditions. In such CGCs, following NMDA addition the [Ca2+]C transient did not stabilize at a plateau level as observed when glucose was present (Fig.7A) but after 2 to 3 min started to increase and during the 5th min of exposure [Ca2+]Capproached 1040 ± 57 nM, n = 46 (Fig. 7C); the steady increase in [Ca2+]C was more clearly visible when the exposure to NMDA was longer than 5 min (data not shown). The pHC changes in the majority of CGCs exposed to NMDA for 5 min were not affected by glucose deprivation (compare Fig. 7, A and C); however, glucose deprivation dramatically changed the effect of MDC addition on [Ca2+]C and pHC. When MDC was applied, [Ca2+]C abruptly increased to fura-2 saturating levels, and this was accompanied by a rapid cytoplasmic alkalization that overshot the basal pHC levels in 46 of 49 neurons (Fig. 7C). Note that the opposite result was observed in CGCs exposed to NMDA in the presence of glucose (Fig. 7A). In only three of 49 glucose-deprived neurons MDC caused a decrease in [Ca2+]C accompanied by only a slight increase in pHC (data not shown).
The effects of glucose deprivation on the NMDA-induced mitochondrial Ca2+ load and on pHCchanges were also studied in CGCs incubated in Li-Locke’s. Based on the data shown in Fig. 5, it was expected that Li+ would inhibit the NaCaX-dependent fraction of the NMDA-induced Ca2+ influx. Indeed, in CGCs incubated in a glucose-free Li-Locke’s, the NMDA-induced Ca2+ transient stabilized at a steady plateau level of 300 ± 7 nM (n = 57), which was three times lower than in CGCs incubated in Na-Locke (compare Fig. 7, C and D). When MDC was applied, [Ca2+]C promptly dropped to basal levels, and a decrease in the rate of the cytoplasmic alkalization was observed (Fig. 7D), which is in dramatic contrast to the result observed in CGCs incubated in Na-Locke’s under otherwise identical conditions (compare Fig. 7, C and D).
These results can be interpreted to indicate that when NMDA receptors are activated in the presence of Li+, the NMDA-induced Ca2+ influx is limited because the reverse mode of NaCaX operation is inactive. Because the Na- or Li-induced depolarization of the PM decreases the CaDF, the subsequent Ca2+ influx directly via NMDA receptor channels as well as voltage-gated Ca2+ channels is very limited, and Ca2+ homeostasis can be maintained, for at least 15 min, without overloading mitochondria with Ca2+. By contrast, the PM depolarization favors the reverse mode of operation of the NaCaX, which in CGCs incubated in Na-Locke’s and exposed to NMDA appears to be the major source of the Ca2+ that overloads mitochondria and leads to neuronal death (Kiedrowski, 1999).
Because in CGCs incubated in Li-Locke’s and exposed to NMDA Ca2+ did not overload mitochondria, one can envision that Li+ might have a direct inhibitory effect on the mitochondrial Ca2+ influx. This possibility cannot be reconciled, however, with the data shown in Fig.7D. If the mitochondrial Ca2+ uptake were indeed inhibited by Li+ during the NMDA-exposure, [Ca2+]C should be greatly elevated, which was not observed. Therefore, it appears that the proposed explanation, i.e., inhibition of the reverse NaCaX by Li+, which is sufficient to explain all the data, remains valid.
NMDA Elicits a Ca-Dependent K+ Efflux.
From Fig.5A (bottom) it is apparent that when all extracellular Na+ was replaced with NMG+, NMDA was still able to depolarize the PM, provided that Ca2+ was present in the medium. A possible mechanism might involve a Ca-dependent collapse of the K+ concentration gradient across the PM, because the NMDA receptor channels are permeable to K+(Mayer and Westbrook, 1987; Tsuzuki et al., 1994). For this to be considered a valid explanation, one has to demonstrate first that the K+ concentration gradient indeed collapses. To test this possibility, [K+]C were measured using a K+-sensitive fluorescent dye, PBFI (Minta and Tsien, 1989). The suitability of PBFI as a probe to measure [K+]C was tested by monitoring PBFI fluorescence in CGCs exposed to glutamate. When glutamate was applied to CGCs incubated in Na- Locke’s, no decrease in the F334/F380 ratio of PBFI fluorescence was observed (Fig. 8A). Such an outcome might be caused by the poor selectivity of PBFI for K+ over Na+. The in vitro determined dissociation constant (K d) of PBFI for K+ is 8 mM and for Na+ is 21 mM (Minta and Tsien, 1989), therefore, one may expect only a small change in PBFI fluorescent properties when intracellular K+ is replaced by Na+. Indeed, as soon as Na+was substituted with Li+, an ion of low affinity for PBFI, K d = 380 mM (Minta and Tsien, 1989), a prompt decrease was observed in the F334/F380 ratio (Fig. 8A), which indicates an efflux of K+ plus Na+ from the cells and a simultaneous influx of Li+. Using the in situ calibration as shown in Fig. 8A, the K d of PBFI for K+ was calculated as being 17.5 ± 0.8 mM (29 experiments on seven different preparations of CGCs). Due to this high affinity of PBFI for K+, when [K+]C exceeded 100 mM, PBFI was practically saturated with K+, and changes in [K+]C could not be detected unless [K+]C dropped to 80 mM or lower.
In CGCs incubated in a Na-free Locke’s buffer in which Na+ was substituted with Li+, NMDA caused a rapid and almost complete collapse of the K+ concentration gradient across the PM, which was blocked by 10 μM MK-801 (Fig. 8B). When the [K+]E was changed during the NMDA exposure, [K+]Cpromptly adjusted to the new [K+]E and remained at this level even after the Na+/K+ ATPase was inhibited with 1 mM ouabain (Fig. 8C). These data indicate that the opening of the NMDA receptor channels causes K+ efflux and Li+ influx, which continue until the concentration gradients of these ions across the PM completely collapse. Such dramatic loss of K+ from the cytoplasm probably does not occur when physiological Na+ concentrations are present in the medium, allowing normal operation of the Na+/K+ ATPase. When Na+ is substituted with Li+, the Na+/K+ ATPase cannot pump K+ and Li+ against their concentration gradients because Li+ binds to both Na+ and K+ recognition sites of the enzyme (Hemsworth et al., 1997). As a result, the complete collapse of the Li/K concentration gradient across the PM is just a matter of time.
Although physiological changes in [K+]C are not reflected in the PBFI fluorescence of cells incubated in Na-free media, this method can be used to explain the mechanism of the PM depolarization when Na+ is replaced with NMG+. As shown in Fig. 8D, under these conditions, the glutamate-elicited K+ efflux occurred at a much lower rate and only in the presence of Ca2+ (Fig. 8D). This Ca-dependent drop in [K+]C was slow and steady (Fig. 8D), and was blocked by MK-801 (not shown); by contrast, stepwise increases in extracellular Li+ concentrations caused rapid drops of [K+]C to lower and lower plateau levels (Fig. 8E).
These results can be interpreted to indicate that when all extracellular Na+ is replaced with NMG+, the activation of NMDA receptors in the absence of Ca2+ causes a K+efflux that is immediately curtailed by the PM hyperpolarization (Fig.5A lower). In the presence of Ca2+, this hyperpolarization is counteracted by a depolarizing Ca2+ influx, and therefore, [K+]C continues to decrease (Fig. 8D). It appears that the different effects of Ca2+ and Li+ on the kinetics of K+ efflux (Fig. 8, D and E) occur because Ca2+ is buffered in the cytoplasm (Carafoli, 1987) but Li+ is not. As a result, the Li+ influx leads to a rapid equilibration of extracellular and intracellular Li+concentrations, followed by an arrest of the K+efflux due to the stabilization of Em at a new K+ equilibrium potential. In contrast, [Ca2+]C does not equilibrate promptly with the extracellular Ca2+concentration because of the cytoplasmic Ca2+buffering. Although the extracellular Ca2+concentration is only 1.3 mM, the supply of Ca2+is virtually unlimited, because in vitro there is a vast excess of extracellular over intracellular volume. As a result, the Ca2+ influx via NMDA receptor channels compensates for the hyperpolarization caused by the K+ efflux, and K+ continues to flow away. Eventually, the K+ concentration gradient across the PM collapses (Fig. 8D), which is reflected by the PM depolarization (Fig. 5A, lower). It should be noted that the monitoring of such a depolarization in an electrophysiological approach might be problematic; in that case, the K+concentration gradient would not collapse as long as the large excess of K+ present in the intracellular electrode could prevent the drop in [K+]C.
The following concluding remarks can be made based on the data presented in this report:
1) The NMDA-induced and Na-dependent depolarization of the PM may be considered as a physiological regulatory mechanism via which the NMDA-elicited Ca2+ influx is restricted.
2) The NMDA-induced depolarization of the PM suppresses direct Ca2+ influx via the NMDA receptor channel but not the influx mediated by the reverse NaCaX. When cellular energy reserves are low and can no longer efficiently support Ca2+ extrusion, the reverse NaCaX is the dominant source of the Ca2+ that accumulates in mitochondria, and may play an important role in excitotoxicity.
3) The NMDA-elicited Na+ influx may have a dual role in excitotoxicity: a protective role due to the depolarization-dependent decrease in the CaDF, and a promoting role due to Ca2+ influx via the reverse operation of the NaCaX. The final Ca2+ load that determines whether neurons will live or die (Hartley et al., 1993; Eimerl and Schramm, 1994) may depend on the balance between these two opposite effects. This balance may vary depending on the duration of the neuronal exposure to NMDA: i.e., the Na-dependent depolarization may be protective if the exposure is short-lasting, but when the exposure is prolonged, the NaCaX reversal may enhance Ca2+influx to toxic levels. It has to be emphasized that these two opposite mechanisms may be differently balanced in various neuronal populations, which might explain why a 15-min exposure to 100 μM glutamate at 37°C under physiological Na+ concentrations is excitotoxic to cortical neurons (Hoyt et al., 1998), whereas CGCs tolerate such exposure well (Fig. 2C).
4) The enhancement of the NMDA-induced Ca2+influx observed in neurons incubated in Na-free media, where Na+ is substituted with large cations that poorly, if at all, permeate the NMDA channel, occurs due to an increase in the CaDF. The CaDF is elevated because K+efflux via the NMDA channel hyperpolarizes the PM. The same consideration applies not only to Ca2+ but also to other cations, such as for example Mg2+. Under physiological conditions, Mg2+ blocks the NMDA channel in a voltage-dependent manner (Mayer et al., 1984; Nowak et al., 1984), but when Na+ is substituted with NMG+, Mg2+ permeates the NMDA channel with abnormal ease (Stout et al., 1996).
Additionally, the NMG-induced enhancement of Ca2+influx may also result from a lack of competition between Na+ and Ca2+ for the NMDA channel. Further study is needed to evaluate the importance of such competition.
5) The NMDA-induced decrease in pHC results from Ca2+ influx, which activates two pHC decreasing mechanisms: a) Ca2+ extrusion by the plasmalemmal Ca2+/H+ pump, and b) Ca2+ sequestration in mitochondria. The NMDA-induced depolarization of the PM in the presence of Na+, Li+, or Cs+ facilitates H+ efflux to the extracellular medium, which counteracts the above-mentioned acidification. When the PM depolarization is prevented by NMG+, H+ is retained in the cytoplasm, and, as a result, the NMDA-induced acidification is enhanced. Of course, the fact that the NaHX is inhibited under such conditions also contributes to the drop in pHC.
6) In the absence of monovalent cations in the extracellular medium, NMDA elicits a Ca-dependent, slow depolarization of the PM, which results from the collapse of the K+ concentration gradient across the PM.
I am grateful to Drs. J.-M. Mienville and N. Smalheiser for thoughtful discussions and to N. Grazulis for help in preparing the manuscript.
- Received December 7, 1998.
- Accepted June 9, 1999.
Send reprint requests to: Lech Kiedrowski, Ph.D., The Psychiatric Institute, 1601 W. Taylor St., Chicago, IL 60612. E-mail:
This work was supported in part by National Institutes of Health Grant NS 37390 and presented in part in abstract form, Society of Neuroscience Abstracts 300.5, 1998.
- [K+]C, and [Ca2+]C, cytoplasmic concentrations of Na+, K+, and Ca2+, respectively
- electrochemical force for Ca2+ influx
- cerebellar granule cells
- conditioned medium
- bis(1,3-dibutylbarbituric acid)trimethine oxonol
- plasma membrane potential
- extracellular concentration of K+
- NMG-L, Cs-L, Na-free Locke’s buffers in which Na+ was substituted with Li+, NMG+, or Cs+, respectively
- mitochondria-depolarizing cocktail
- NaCaX and NaHX
- Na+/Ca2+ and Na+/H+exchanger, respectively
- standard Locke’s buffer containing physiological Na+ concentrations
- normalized fluorescence emitted by DiBAC4(3) after excitation at 488 nm
- K+-binding benzofuran isophthalate
- cytoplasmic pH
- plasma membrane
- The American Society for Pharmacology and Experimental Therapeutics