A steep inwardly directed Na+ gradient is essential for glial functions such as glutamate reuptake and regulation of intracellular ion concentrations. We investigated the effects of glucose deprivation, chemical hypoxia, and simulated ischemia on intracellular Na+ concentration ([Na+]i) in cultured spinal cord astrocytes using fluorescence ratio imaging with sodium-binding benzofuran isophthalate (SBFI) AM. Glucose removal or chemical hypoxia (induced by 10 mm NaN3) for 60 min increased [Na+]i from a baseline of 8.3 to 11 mm. Combined glycolytic and respiratory blockage by NaN3 and 0 glucose saline caused [Na+]i to increase by 20 mm, similar to the [Na+]iincreases elicited by blocking the Na+/K+-ATPase with ouabain. Recovery from large [Na+]i increases (>15 mm) induced by the glutamatergic agonist kainate was attenuated during glucose deprivation or NaN3 application and was blocked in NaN3 and 0 glucose. To mimic in vivo ischemia, we exposed astrocytes to NaN3 and 0 glucose saline containing l-lactate and glutamate with increased [K+] and decreased [Na+], [Ca2+], and pH. This induced an [Na+]i decrease followed by an [Na+]i rise and a further [Na+]i increase after reperfusion with standard saline. Similar multiphasic [Na+]i changes were observed after NaN3 and 0 glucose saline with only reduced [Na+]e. Our results suggest that the ability to maintain a low [Na+]ienables spinal cord astrocytes to continue uptake of K+ and/or glutamate at the onset of energy failure. With prolonged energy failure, however, astrocytic [Na+]i rises; with loss of their steep transmembrane Na+ gradient, astrocytes may aggravate metabolic insults by carrier reversal and release of acid, K+, and/or glutamate into the extracellular space.
The nervous system has one of the highest metabolic rates in the body and, because it stores very little fuel, depends on a continuous exogenous supply of metabolic substrate and O2. Approximately 50% of cellular energy is needed for ion transport and the maintenance of ion gradients, mainly by the Na+/K+-ATPase (Hansen, 1985; Ames et al., 1992; Erecinska and Silver, 1994). Na+/K+-ATPase extrudes Na+ ions in exchange for extracellular K+ and generates an inwardly directed gradient for Na+ ions, which in turn energizes other secondary ion transporters (e.g., Na+–Ca2+exchange). Pathological conditions that diminish or arrest energy production such as hypoglycemia or anoxia/ischemia, therefore, lead to rapid breakdown of cellular ion gradients (Hansen, 1985; Erecinska and Silver, 1994).
Breakdown of the Na+ gradient after inhibition of Na+/K+-ATPase is probably one of the key elements in promoting cellular damage during energy failure (Lees, 1991; Lucas et al., 1997), and blocking Na+influx can protect neurons from anoxic injury (Stys et al., 1992;Agrawal and Fehlings, 1996; Imaizumi et al., 1997). Astrocytes are much less susceptible to hypoglycemia or ischemia than neurons (Auer and Siesjö, 1988; Ransom and Fern, 1996), in part because they possess larger glycogen stores (Hamprecht and Dringen, 1995) and are less subject to glutamatergic excitotoxicity [Choi and Rothman (1990); see, however, David et al. (1996)].
In addition, astrocytes can increase survival of neurons during metabolic insults (Vibulsreth et al., 1987; Swanson and Choi, 1993). This protection may be attributable to the release of lactate by astrocytes (Walz and Mukerji, 1988; Tsacopoulos and Magistretti, 1996) or to their ability to regulate extracellular glutamate (Rosenberg, 1991) or extracellular ion concentrations (Kraig et al., 1995). Other studies suggest, however, that astrocytes may aggravate ischemic damage because of extensive swelling and/or release of glutamate (Siesjö, 1985; Billups and Attwell, 1996), both of which probably occur secondary to increases in intracellular Na+concentration ([Na+]i). To elucidate the mechanisms of hypoglycemic or ischemic damage, therefore, one must understand precisely ion movements not only in neurons but also in glial cells in response to these metabolic insults. Culture systems are well suited for investigating altered ion regulation, because, in contrast to the situation in vivo, the extracellular conditions can be manipulated and controlled rigorously.
In the present study, we measured changes in [Na+]i and the capacity for [Na+]i regulation in cultured spinal cord astrocytes during glucose deprivation, chemical hypoxia, and simulated ischemia. Similar metabolic insults can affect the in vivo spinal cord (e.g., after trauma) and play a significant role in the establishment of axonal injury (Tator and Fehlings, 1991). Our results indicate that spinal cord astrocytes are able to maintain a steep inwardly directed Na+ gradient for tens of minutes during energy disruption and could, therefore, play a protective role during this period. Because of diminished capacity for Na+ extrusion during prolonged energy failure, however, the astrocytic Na+ gradient is gradually reduced, promoting reversal of astrocytic carriers and cellular damage.
MATERIALS AND METHODS
Cell culture. Cell cultures of spinal cord astrocytes were prepared from neonatal rats (Sprague Dawley rats; postnatal day 0) as described previously (Black et al., 1993). Briefly, the pups were anesthetized by CO2 and decapitated, and their spinal cords were removed. The tissue was digested with papain (20 U/ml for 30 min; Worthington, Freehold, NJ) and triturated. The dissociated cells were plated onto poly-l-ornithine- and laminin-coated (Sigma, St. Louis, MO) coverslips in complete medium (Earle’s minimum essential medium; JRH Biologicals, Lenexa, KA) containing 10% fetal bovine serum (HyClone, Logan, UT), 6 mm glucose, and penicillin and streptomycin at 120 U/ml and 120 μg/ml, respectively (Sigma). Cells were grown in an incubator at 37°C and 95 %O2/5% CO2 and were fed every 3–4 d. Cells between 14 and 21 d in culture were used for experiments. At this time in vitro, cultures were confluent, and nonstellate astrocytes [cf. Sontheimer et al. (1992), their Fig. 1] represented >95% of the cells on a given coverslip, as revealed by immunostaining for glial fibrillary acidic protein.
Solutions. During experiments, cells were perfused with standard saline containing (in mm): 140.7 NaCl, 3 KCl, 1.2 MgSO4, 1 CaCl2, 2 NaH2PO4, 20 HEPES, and 10 glucose, titrated to a pH of 7.4 with NaOH. In salines with altered ionic composition (e.g., increased [K+]), the NaCl concentration was reciprocally changed to maintain constant osmolarity. The sensitivity of SBFI to Na+ was calibrated after each experiment with salines that contained (in mm): 150 [K+ + Na+], 30 Cl−, 120 gluconic acid, and 10 HEPES, titrated to pH 6.9 with KOH (cf. Rose and Ransom, 1996a). In addition, these calibration solutions contained 3 μm gramicidin and 10 μm monensin for equilibration of extra- and intracellular [Na+].
Drugs and chemicals were purchased from Sigma and added to the saline shortly before use. Gramicidin, monensin, and ouabain were prepared as 1 m stock solutions in dimethylsulfoxide (DMSO; Sigma) and stored in the freezer. Acetoxymethyl ester of sodium-binding benzofuran isophthalate (SBFI AM) was obtained from Teflabs.
Measurement of [Na+]i.[Na+]i was measured using fluorescence ratio imaging with SBFI AM as described previously (Rose and Ransom, 1996a; Rose et al., 1997). In summary, cells were loaded with SBFI AM (20 μm; 90 min) in standard saline containing 0.1% pluronic (BASF/Knoll, Ludwigshafen, Germany) at room temperature (20–22°C). After dye-loading, the coverslips were transferred to an experimental chamber (volume, ∼500 μl) and perfused with saline warmed to 37°C at a flow rate of ∼1 ml/min. The chamber was mounted on the stage of a Nikon-Diaphot TMD inverted microscope equipped with an oil-immersion objective (Nikon Fluor 40/1.30). The cells were excited every 10 sec at 345 and 390 nm, and emission fluorescence of single cells above 510 nm was collected, averaging four video frames. The collection was performed by a GenIISys image intensifier system connected to a videocamera (MTI CCD72; Dage-MTI, Michigan City, IN). Fluorescence was quantified using an image acquisition program from Georgia Instruments (Roswell, GA).
Absolute [Na+]i was determined for each individual cell by a two- or three-point calibration performed after each experiment (Rose and Ransom, 1996a; Rose et al., 1997). For calibration of the fluorescence of SBFI, the astrocytes were perfused with calibration solutions containing known Na+concentrations (0 and 30 mm or 0, 30, and 50 mm[Na+]) and gramicidin and monensin (see above) for equilibration of extra- and intracellular ion concentrations. The 345/390 nm fluorescence ratio of intracellular SBFI changed monotonically with changes in [Na+]i; stepwise changes in Na+ concentrations resulted in stepwise changes in the SBFI ratio (Fig.1 A). The slope of the ratio change was 0.16 per 10 mm Na+ from 0 to 30 mm [Na+]i(n = 47) and 0.07 per 10 mmNa+ for 30–60 mm[Na+]i (n = 40; Fig.1 B).
As reported earlier (Rose and Ransom, 1996a), the K+sensitivity of intracellular SBFI was negligible. Reduction of [K+]i by 40 mm resulted in an apparent decrease in [Na+]i of <2 mm (data not shown). Intracellular SBFI fluorescence ratio did, however, react slightly to changes in pHi. Switching pHi from the standard value of 7.1 to 6.7 and then to 7.5 resulted in ratio changes mimicking [Na+]i changes of ∼3 mm(n = 50; Fig. 1 C). Acidifications mimicked a decrease, and alkalinizations mimicked an apparent increase in [Na+]i. This weak pH sensitivity of SBFI might lead to an underestimation of [Na+]i increases by a few millimolar in the present study during manipulations that can be accompanied by large acidifications such as simulated ischemia (Kraig and Chesler, 1990). It is unlikely, however, that the pHi of astrocytes would change so markedly under our in vitro conditions because it is strongly influenced by pHo which was held constant in these experiments (Pappas et al., 1996)
Experiments were repeated on at least four different coverslips. Data are presented as mean ± SD and were statistically analyzed using a Student’s t test (significance level, p< 0.001 unless stated otherwise).
In agreement with a previous report [Rose et al. (1997); see alsoRose and Ransom (1996a)], baseline [Na+]i was 8.3 ± 4.0 mm (n = 872) in cultured spinal cord astrocytes in standard, HEPES-buffered saline, resulting in a calculated reversal potential for Na+(E Na) of +78 mV. [Na+]i was very stable (± 1 mm over periods of 100 min) when no other experimental manipulation was performed. To investigate the importance of cellular energy production for maintenance of this low baseline [Na+]i, we monitored [Na+]i before, during, and after metabolic inhibition by removal of glucose, inhibition of glycolysis, and/or chemical hypoxia for periods of 1 hr. We also tested the capacity of astrocytes to recover from imposed [Na+]i loads under these conditions. In addition, the [Na+]i changes induced by simulated ischemia were investigated.
Effects of 0 glucose and/or glycolytic inhibition on baseline [Na+]i
The energy metabolism of astrocytes was reduced by (1) removal of glucose from the saline, (2) replacement of glucose with 10 mm 2-deoxyglucose (2-DG), or (3) addition of the glycolytic inhibitor NaF with concomitant glucose removal. Glucose deprivation alone does not result in complete inhibition of glycolysis, because glucose-6-phosphate can be provided from cellular glycogen stores and by gluconeogenesis (Erecinska and Silver, 1994; Hamprecht and Dringen, 1995). 2-DG, on the other hand, is phosphorylated to 2-DG-6-phosphate but is not metabolized further and, therefore, accumulates intracellularly, progressively blocking glycolysis and reducing glycogenolysis by inhibition of glycogen phosphorylase (Sokoloff et al., 1977; Hamprecht and Dringen, 1995). NaF, finally, not only inhibits glycolysis but also the pyruvate dehydrogenase complex (Erecinska and Silver, 1994), blocking flux of pyruvate into the citric acid cycle and reducing substrate recruitment from glycogenolysis and gluconeogenesis for respiratory oxidation. Of course, none of these approaches inhibits glycolysis in isolation. They all reduce pyruvate, the primary substrate for oxidative metabolism, at the same time.
Both glucose removal and replacement of glucose by 2-DG for a 60 min period led to a slow increase in [Na+]i in spinal cord astrocytes that was visible 5–10 min after the onset of the manipulation (Fig.2 A,B). Glucose removal significantly increased [Na+]i to 11.6 ± 6.6 mm (n = 143; see Fig. 4); this increase was completely reversible within 5–10 min after reperfusion with standard saline (Fig. 2 A). Replacement of glucose with 2-DG caused a statistically significant higher [Na+]i increase to 14.9 ± 6.4 mm (n = 48; see Fig. 4), which was only partly reversed after a 20 min wash with standard saline (Fig.2 B).
Concomitant removal of glucose and addition of 10 mm NaF led to a transient increase in [Na+]ito 21.3 ± 13.2 mm (n = 42), which peaked within 30 min and was followed by a decline in [Na+]i. At the end of the 1 hr exposure to the glucose-free NaF-containing saline, [Na+]i had reached 15.2 ± 8.2 mm, a value significantly higher than the [Na+]i in standard saline or during glucose deprivation (see Figs. 2 C, 4). The NaF-induced [Na+]i changes were more rapid in onset than were those with glucose-free saline or application of 2-DG and reversed within 10–15 min after switching back to standard saline (Fig. 2 C).
[Na+]i changes induced by chemical hypoxia and ouabain
To investigate the role of aerobic metabolism in maintaining low baseline [Na+]i in cultured spinal cord astrocytes, we inhibited mitochondrial respiration by the addition of 10 mm sodium azide (NaN3) in the presence or absence of glucose in the saline (conditions corresponding to normal or reduced glycolytic capacity, respectively). NaN3 blocks the oxygen-requiring steps in energy metabolism by inhibition of cytochrome oxidase and, therefore, induces a “chemical hypoxia” (Swanson, 1992).
Addition of NaN3 for 1 hr, with glucose present, caused a small but significant increase in astrocytic [Na+]i to 10.6 ± 4.1 mm (n = 40; Figs.3 A,4). [Na+]i rose to 27.5 ± 11.3 mm when NaN3 was applied in a glucose-free saline (NaN3 and 0 glucose; n = 65; Figs.3 A, 4), a result significantly different than that seen with NaN3 in the presence of glucose. This [Na+]i increase started 3–5 min after the onset of the perfusion with NaN3 and 0 glucose saline. The cells recovered within 10–15 min from this Na+load (Fig. 3 A).
Voltage-gated Na+ channels have been proposed to mediate significant Na+ influx into neurons under conditions of metabolic failure (Stys et al., 1992). Because cultured spinal cord astrocytes are characterized by a high density of voltage-gated Na+ channels (Sontheimer et al., 1992), we investigated whether the [Na+]i increase during chemical hypoxia was attributable to channel-mediated Na+influx. Blocking Na+ channels with 10 μm tetrodotoxin, however, did not visibly influence the slope or the amplitude of the [Na+]iincrease induced by NaN3 and 0 glucose (n = 69; Figs. 3 B, 4).
Metabolic inhibition lowers intracellular ATP concentration, which leads to reduced Na+/K+-ATPase activity (Hansen, 1985; Lees, 1991). To test the effect of Na+/K+-ATPase inhibition on [Na+]i in cultured spinal cord astrocytes, we applied 2 mm ouabain for 1 hr. A concentration of 1 mm ouabain was reported to inhibit Na+/K+-ATPase activity in astrocyte cultures completely (Walz and Hertz, 1982).
Ouabain increased [Na+]i within 1–2 min after its application. The ouabain-induced [Na+]i increase (n = 42) was similar to that induced by metabolic inhibition using NaN3 and 0 glucose (Figs. 3 C, 4), indicating that metabolic inhibition probably affected [Na+]i by leading to failure of Na+/K+-ATPase activity. This was also supported by the observation that adding another glycolytic blocker, NaF, to the combined glycolytic and respiratory inhibition by NaN3 and 0 glucose did not augment the [Na+]i increase seen with NaN3 and 0 glucose alone (n = 42; Fig.4).
Figure 4 summarizes the effects of metabolic inhibition on baseline [Na+]i in spinal cord astrocytes by showing the mean values of [Na+]i 60 min after the onset of the manipulation. Inhibition of glycolysis by 2-DG or NaF added to glucose-free saline increased [Na+]i to ∼15 mm. Inhibition of oxidative phosphorylation alone by NaN3increased [Na+]i by only 3 mm to ∼11 mm, and this increase was significantly less than that seen with glycolytic inhibition. Removal of glucose also produced only a small increase in [Na+]i, presumably because astrocytes could breakdown glycogen for fuel under these conditions.
A greater reduction in calculated E Na (to +44 mV maximally) was observed with combined inhibition of glycolysis and oxidative phosphorylation for 60 min. A similar value forE Na was calculated with inhibition of Na+/K+-ATPase activity by ouabain (Fig. 4). Even assuming that the astrocytes were depolarized under these conditions (Walz et al., 1993), the positiveE Na still indicates a large inwardly directed Na+ gradient.
Recovery from imposed [Na+]i loads during metabolic inhibition
Because inhibition of glycolysis and/or oxidative phosphorylation caused relatively small reductions in transmembrane Na+ gradient after 60 min, we tested the ability of the astrocytes to recover from additional Na+ loads under these conditions. [Na+]i was increased by lowering extracellular [K+] ([K+]e), which inhibits Na+/K+-ATPase (cf. Rose and Ransom, 1996a), or by application of the glutamatergic agonist kainate (Kimelberg et al., 1989; Rose et al., 1997).
Removal of [K+]e for 10 min reversibly increased [Na+]i by 3.2 ± 2.5 mm (n = 55; Fig.5). Neither the average amplitude nor the recovery from this [Na+]i increase was significantly altered in glucose-free saline ([Na+]i increased by 4.0 ± 1.5 mm; n = 47; Fig. 5), indicating that glucose removal and the subsequent increase in baseline [Na+]i did not hamper the ability of the cell to recover from small [Na+]iloads.
Kainate (0.5 mm; 2 min) changed [Na+]i in 39% (n = 184 total) of spinal cord astrocytes by opening a cation channel coupled to a non-NMDA receptor (Rose and Ransom, 1996b; Rose et al., 1997). As reported earlier (Rose et al., 1997), kainate decreased [Na+]i by ∼2 mm in 21% of the responding cells and variably increased [Na+]i in the remainder (average increase, 13.3 ± 10.9 mm). Removal of glucose or application of NaN3 and 0 glucose did not change kainate-induced decreases in [Na+]i, nor did it change kainate-induced increases in [Na+]i, if they were small (i.e., <15 mm; data not shown).
Kainate-induced [Na+]i increases of >15 mm, in contrast, were augmented in amplitude (increase, 45% larger than control; n = 3; Fig.6 A) or slowed in recovery [average recovery time, 29.0 ± 7.2 min (n = 5) vs 7.6 ± 4.8 min under control conditions (n = 56)] in 0 glucose (Fig. 6 B). Recovery from kainate-induced [Na+]iincreases of >15 mm was also significantly prolonged when oxidative phosphorylation was inhibited by NaN3 (recovery time, 24.3 ± 6.4 min; n = 5; Fig.7 A). No recovery at all was observed during metabolic inhibition with NaN3 and 0 glucose (n = 26; Fig. 7 B).
These results allow several conclusions about [Na+]i homeostasis in spinal cord astrocytes during metabolic inhibition for 60 min. During inhibition of glycolysis or oxidative phosphorylation, [Na+]i regulation remained functional, and despite a slight increase in baseline [Na+]i, the cells were able to extrude additional, small Na+ loads. Recovery from [Na+]i increases larger than ∼15 mm, however, was significantly impaired under these conditions, suggesting that neither of the two affected pathways (glycolysis and oxidative phosphorylation) was sufficient alone to ensure optimal recovery from large [Na+]i loads. Astrocytes completely lost their ability to extrude Na+ during total energy metabolism blockade caused by both glucose removal and inhibition of oxidative phosphorylation.
[Na+]i changes caused by simulated ischemia
Pathological insults such as anoxia and ischemia are characterized by dramatic changes in the ion composition of the extracellular space in the brain (Hansen, 1985; Ransom et al., 1992; Erecinska and Silver, 1994; Kraig et al., 1995). These changes include an increase in [K+]e and a decrease in [Na+]e, [Ca2+]e, and pHe. In addition, extracellular lactate as well as extracellular glutamate concentrations rise strongly (Shimada et al., 1993). To mimic these ischemic conditions in vitro, we exposed astrocytes to a glucose-free saline containing 10 mmNaN3, 60 mm K+, 95 mm Na+, 0.2 mmCa2+, 20 mm l-lactate, 0.1 mm glutamate, and 20 mm PIPES (1,4-piperazinediethanesulfonic acid), titrated to a pH of 6.5, called “ischemia solution” (cf. Hansen, 1985; Ransom et al., 1992; Shimada et al., 1993; Erecinska and Silver, 1994; Kraig et al., 1995).
A 60 min perfusion with this solution resulted in multiphasic changes in [Na+]i (n = 81; see Figs. 8, 11). At the onset of the simulated ischemia, [Na+]i quickly dropped by 4.0 ± 4.9 mm. After 10–15 min, the [Na+]i drop was followed by a slow increase in [Na+]i to a value 2.9 ± 3.1 mm higher than the original baseline in standard saline. The most dramatic change in [Na+]i, however, took place shortly after reperfusion with standard saline, when [Na+]i rapidly increased by another 7.6 ± 7.1 mm. In contrast to the effects observed with NaN3 and 0 glucose, which were completely reversible within 10–15 min (compare Fig. 3 A), the astrocytes recovered more slowly from the ischemia solution-induced [Na+]i increase, and [Na+]i was still elevated by ∼4 mm after a 30 min reperfusion with standard saline (Fig.8).
To analyze the mechanisms underlying these multiphasic [Na+]i changes, we investigated the individual effects of each ionic alteration made in the ischemia solution. Perfusion with 60 mm K+ and 95 mm Na+ (n = 36), or with 95 mm Na+ (n = 53; Na+ replaced byN-methyl-d-glucamine) caused a decrease in [Na+]i by <3 mm (see Figs. 9, 11). Decreasing pHe to 6.5 caused a small, reversible fall in [Na+]i (<2 mm) in 9 out of 41 cells investigated (Fig.9). All other changes to the standard saline-induced elevations in [Na+]iare as follows (see Figs. 9, 11): 0.2 mm[Ca2+]e increased [Na+]i by ∼2 mm(n = 38), 20 mm lactate increased [Na+]i by ∼4 mm(n = 55), and 0.1 mm glutamate increased [Na+]i by ∼5 mm(n = 53). Removal of glucose or addition of NaN3 increased [Na+]i as described above (compare Figs. 2, 3).
Each of the [Na+]i changes seen with individual ionic or chemical alterations was monophasic and completely reversible within 10–15 min after switching back to standard saline. Multiphasic [Na+]i changes mimicking those observed with simulated ischemia, however, were obtained when the astrocytes were perfused with a saline containing NaN3, 0 glucose, and 95 mmNa+ (n = 61; Figs.10,11); an initial decrease in [Na+]i was followed by a slow [Na+]i increase that was followed by a rapid secondary increase in [Na+]iwhen switching back to normal saline. In contrast to the situation found with simulated ischemia, however, recovery from this secondary [Na+]i increase was usually complete within 10–15 min (Fig. 10). These results suggest that the reduction in [Na+]e that accompanies inhibition of metabolism in ischemia (Hansen, 1985) may be of special importance as an influence on [Na+]i in spinal cord astrocytes during ischemia.
Mechanisms of [Na+]i increases during inhibition of energy production
The effects of various forms of metabolic inhibition on [Na+]i regulation in cultured rat spinal cord astrocytes were analyzed in this study. Only small increases in [Na+]i (<10 mm) were observed when either glycolysis or oxidative respiration alone were inhibited for 60 min. This is consistent with the effects of such inhibition on astrocyte [ATP]. In general, blocking either aerobic or glycolytic metabolism has little effect on [ATP] (Pauwels et al., 1985; Gregory et al., 1990; Kelleher et al., 1993; Gemba et al., 1994; Hori et al., 1994; Longuemare et al., 1994; for review, see Silver and Erecinska, 1997). The fact that blocking oxidative metabolism alone, as seen with NaN3, does not have a strong effect on [ATP] and, for that reason, on [Na+]i is a consequence of the high glycolytic rate of astrocytes (Silver and Erecinska, 1997). The minimal effect of 0 glucose on [ATP] is easily explained on the basis of their stored glycogen that can generate glucose residues. The capacity of astrocytes to maintain near normal [ATP] when glycolysis is blocked is probably attributable to their ability to metabolize ketone bodies and Krebs cycle intermediates (Wiesinger et al., 1997). It has been postulated that the ATP produced by glycolysis selectively supports Na+/K+-ATPase (Lipton and Robacker, 1983; Kauppinen et al., 1988; Raffin et al., 1988). Our observations that glycolytic blockade minimally changed [Na+]i is not consistent with this assertion.
In contrast to the small changes in [ATP] that occur with isolated inhibition of glycolysis or oxidative respiration, [ATP] drops within 5–10 min to near 0 when both are inhibited (Longuemare et al., 1994;Silver et al., 1997). The 20 mm increase in astrocytic [Na+]i observed after 60 min of combined glycolytic and respiratory blockade most likely resulted from complete failure of Na+/K+-ATPase (Lees, 1991). Consistent with this, we observed a very similar [Na+]i increase when Na+/K+-ATPase was inhibited by ouabain. These results indicate a very low leakage of sodium ions of spinal cord astrocytes.
Relatively moderate [Na+]i increases during metabolic inhibition were also reported in mouse astrocytes (Silver et al., 1997) and rat Schwann cells (Lehning et al., 1995), whereas [Na+]i rose to near 140 mm in cultured rat cortical astrocytes after 45 min of combined glycolytic and respiratory blockage (M. C. Longuemare, C. R. Rose, B. R. Ransom, S. G. Waxman, and R. A. Swanson, unpublished observations). These differences point to differential susceptibility of astrocytes derived from different brain regions to metabolic failure, similar to the situation with neurons (Auer and Siesjö, 1988) and echo a common theme in astrocyte physiology (Ransom, 1991). Neuronal [Na+]i changes may be larger than those in glial cells, however, when studied in the same preparation (Lehning et al., 1995; LoPachin and Stys, 1995). In cultured CA1-hippocampal neurons, [Na+]iincreased by 27 mm after only 3–4 min of anoxia (Friedman and Haddad, 1994), emphasizing the high susceptibility of these neurons to hypoglycemic and anoxic damage (Auer and Siesjö, 1988).
Glia are less susceptible to injury caused by metabolic inhibition than are neurons (Goldberg and Choi, 1993). This might be related to less Na+ influx in glia. Our investigation shows that despite their relatively high density (Sontheimer et al., 1992), voltage-gated Na+ channels probably do not mediate Na+ influx into spinal cord astrocytes during glycolytic and respiratory inhibition. In neurons, in contrast, TTX can provide significant protection from anoxic damage (Stys et al., 1992;Lynch et al., 1995; Agrawal and Fehlings, 1996; Lehning et al., 1996;Imaizumi et al., 1997). Astrocytes, in addition, probably experience less glutamate-gated Na+ influx than do neurons [Auer and Siesjö (1988); Choi and Rothman (1990); see, however,David et al. (1996)]. Consequently, harmful Ca2+influx caused by reversal of Na+–Ca2+ exchange (Waxman et al., 1991; Ransom and Fern, 1996) might be smaller in astrocytes because of less Na+ entry.
Na+ regulatory capacity during metabolic inhibition
The small [Na+]iincreases observed in spinal cord astrocytes with prolonged inhibition of glycolysis or respiration did not greatly reduce the transmembrane Na+ gradient. The astrocytes were still able to recover from small additional Na+ loads, indicating that [ATP] was sufficient to allow an increase in Na+/K+-ATPase activity. Challenged with large [Na+]i loads (>15 mm), however, recovery was attenuated, indicating that maximal glial Na+/K+-ATPase activity was limited during metabolic inhibition. In the face of attenuated [Na+]i recovery, the capacity to recover from accompanying changes in other ions would also be reduced, resulting in a general disturbance of cellular ion regulation as described for neurons (Kiedrowski et al., 1994; Koch and Barish, 1994).
The maintenance of high E Na even after 1 hr of combined glycolytic and respiratory blockage suggests that other ion gradients may be maintained as well; consistent with this prediction, relatively small changes in [Ca2+]ihave been reported from acutely isolated hippocampal astrocytes during these insults (Duffy and MacVicar, 1996). Other changes observed in astrocytes during glycolytic and respiratory failure include slow membrane depolarization, K+ loss, acidification, and swelling (Kimelberg and Ransom, 1986; Kauppinen et al., 1988; Kraig and Chesler, 1990; Walz et al., 1993).
Despite the high E Na maintained during metabolic failure, active [Na+]i regulation failed in spinal cord astrocytes when they were challenged by large additional [Na+]i loads [see alsoLees (1991); Erecinska and Silver (1994)]. These results suggest that the transmembrane Na+ gradient and capacity for intracellular ion regulation would be lost rapidly during metabolic failure in intact tissue if astrocytes experience heavy Na+ loading. This, in fact, was reported by Duffy and MacVicar (1996), who showed that the ischemia-induced [Ca2+]i increase was larger in astrocytes within slices than in isolated cells.
Mechanisms of [Na+]i changes caused by simulated ischemia
Ischemia-induced energy failure results in marked alterations of extracellular ion concentrations in intact tissue. In gray matter, [K+]e and [Na+]e change to ∼60 mm, [Ca2+]e decreases to 0.2 mm, pHe decreases to 6.5, and extracellular lactate and glutamate concentrations increase (Hansen, 1985; Erecinska and Silver, 1994; Sykova et al., 1994; Kraig et al., 1995). In white matter, extracellular ion changes are qualitatively similar but smaller in amplitude (Ransom et al., 1992).
Our results show that no one of these extracellular changes caused the sequence of [Na+]i change in spinal cord astrocytes that was characteristic of metabolic inhibition. The small decrease in [Na+]i observed with elevated [K+]e was probably attributable to activation of Na+/K+-ATPase and cell swelling (Walz and Hertz, 1983; Rose and Ransom, 1996a), whereas the [Na+]i decrease during low [Na+]e was probably caused by reduction of the Na+ gradient (Rose and Ransom, 1996a). The reason for the small [Na+]i increase induced by [Ca2+]e reduction is unclear, but it could have been caused by altered Na+–Ca2+ exchange (Goldman et al., 1994). Acidification did not alter [Na+]i significantly, whereas lactate caused a sustained increase in [Na+]i, consistent with activation of a Na+-dependent uptake mechanism in astrocytes (Walz and Mukerji, 1988). Glutamate, applied at a concentration highly toxic for neurons (Choi and Rothman, 1990), caused a small (∼5 mm), sustained [Na+]i increase that was probably related to Na+-coupled uptake (Rose and Ransom, 1996b).
Applying all these ionic changes in conjunction with glycolytic and respiratory inhibition (simulated ischemia, Fig. 8) revealed that [Na+]i changes caused by single manipulations were not simply additive. Instead, simulated ischemia resulted in multiphasic [Na+]i changes that could only be mimicked by subjecting the astrocytes to metabolic failure and decreased [Na+]e. This combination led to a small decrease in [Na+]i at the beginning of the insult and limited the strong [Na+]i increase observed during metabolic failure alone, presumably because of the Na+ gradient reduction combined with cell swelling. The largest [Na+]i change, however, occurred after reperfusion with standard saline when the normal [Na+]e was restored. This phenomenon, which has not been described before, suggests that the reinstatement of [ATP] and Na+/K+-ATPase activity lagged behind the re-establishment of the Na+ gradient.
An analogous reperfusion-mediated phenomenon reported in astrocytes is a strong rise in [Ca2+]i subsequent to perfusion with Ca2+-free saline; this so-called “reperfusion paradox” is attributable to elevation of [Na+]i via Na+-permeable Ca2+ channels, followed by reverse Na+–Ca2+exchange (Kim-Lee et al., 1992). These events, the rebound elevation of [Na+]i and the calcium paradox, suggest that astrocytes might be most susceptible to ischemic damage during the reperfusion period.
Possible role of astrocytes during metabolic insult
Despite a reduction in their capacity for Na+extrusion, cultured spinal cord astrocytes maintain a steep inwardly directed Na+ gradient during a 60 min period of energy metabolism inhibition. This is true even during combined glycolytic and respiratory failure, although Na+/K+-ATPase-mediated Na+ export is entirely blocked. The Na+ gradient may be even better protected duringin vivo ischemia because of extracellular ion changes. With a concomitant reduction in [Na+]e, the astrocytes can maintain their Na+ gradient close to normal even during prolonged periods of complete energy failure, possibly enabling them to maintain Na+-dependent ion regulation. Because Na+/K+-ATPase activity is blocked, any additional Na+ loads, e.g., attributable to channel-mediated influx, will, however, decrease the Na+ gradient. This would be expected to promote reversal of Na+-dependent plasma membrane carriers. We predict that, in intact tissue, astrocytes may protect neurons at the onset of ischemic insults by virtue of their continued capacity for regulation of extracellular ion concentrations and glutamate uptake but may aggravate damage by release of acid, K+, and/or glutamate as a result of compromised [Na+]i extrusion because of energy failure during long-lasting ischemia.
This work was supported in part by National Institutes of Health Grant NS-15589 to B.R.R., National Multiple Sclerosis Society Grant RG 1912, and a grant from the Medical Research Service, Veterans Administration, to S.G.W. C.R.R. was supported in part by a fellowship from the Eastern Paralyzed Veterans Association. We thank Dr. Ray Swanson for helpful comments on an earlier version of this manuscript.
Correspondence should be addressed to Dr. Christine R. Rose, I. Physiologisches Institut, Universität des Saarlandes, D-66421 Homburg/Saar, Germany.